Speed up range-query intersections via seek_danger on RangeDocSet (up to ~50x faster)

A regular seek on RangeDocSet is costly: on a miss it fetches blocks and
scans the column forward to materialize the next matching doc. As a
non-leading docset in an intersection that work is wasted — the driver only
asks "does this candidate match?". seek_danger answers that with a cheap
point lookup via Column::values_for_doc, returning a lower bound on a miss
and leaving forward progress to the caller.

Forward seek_danger through ConstScorer.

Benchmarks (bool_queries_with_range, _all_results / DocSetCollector):

```
dense and 0.1% a
a_AND_num_rand:[0_TO_9]_all_results                                 Avg: 0.0827ms (-4.60%)     Median: 0.0825ms (-4.82%)     [0.0809ms .. 0.0891ms]    Output: 43
a_AND_num_asc:[0_TO_9]_all_results                                  Avg: 0.1937ms (-3.70%)     Median: 0.1930ms (-3.59%)     [0.1806ms .. 0.2044ms]    Output: 100
a_AND_num_rand_fast:[0_TO_9]_all_results                            Avg: 0.0367ms (-92.67%)    Median: 0.0365ms (-92.65%)    [0.0340ms .. 0.0398ms]    Output: 43
a_AND_num_asc_fast:[0_TO_9]_all_results                             Avg: 0.1052ms (-98.05%)    Median: 0.1050ms (-97.98%)    [0.1009ms .. 0.1117ms]    Output: 100
num_rand_fast:[0_TO_9]_AND_num_asc_fast:[0_TO_9]_all_results        Avg: 2.7147ms (-51.42%)    Median: 2.7075ms (-49.58%)    [2.6806ms .. 2.7799ms]    Output: 968
dense and 1% a
a_AND_num_rand:[0_TO_9]_all_results                                 Avg: 0.4373ms (-9.71%)     Median: 0.4357ms (-10.12%)    [0.4117ms .. 0.4711ms]    Output: 463
a_AND_num_asc:[0_TO_9]_all_results                                  Avg: 0.2342ms (-2.50%)     Median: 0.2338ms (-2.56%)     [0.2247ms .. 0.2452ms]    Output: 1_054
a_AND_num_rand_fast:[0_TO_9]_all_results                            Avg: 0.3956ms (-82.86%)    Median: 0.3943ms (-82.90%)    [0.3815ms .. 0.4119ms]    Output: 463
a_AND_num_asc_fast:[0_TO_9]_all_results                             Avg: 0.4896ms (-91.16%)    Median: 0.4862ms (-90.81%)    [0.4797ms .. 0.5084ms]    Output: 1_054
num_rand_fast:[0_TO_9]_AND_num_asc_fast:[0_TO_9]_all_results        Avg: 2.7108ms (-50.81%)    Median: 2.6925ms (-49.51%)    [2.6688ms .. 2.7868ms]    Output: 968
dense and 10% a
a_AND_num_rand:[0_TO_9]_all_results                                 Avg: 0.9869ms (-3.71%)     Median: 0.9833ms (-3.83%)     [0.9518ms .. 1.1218ms]    Output: 4_914
a_AND_num_asc:[0_TO_9]_all_results                                  Avg: 0.6352ms (-3.74%)     Median: 0.6363ms (-3.32%)     [0.6158ms .. 0.6488ms]    Output: 10_152
a_AND_num_rand_fast:[0_TO_9]_all_results                            Avg: 3.1264ms (+0.39%)     Median: 3.1466ms (+1.34%)     [3.0261ms .. 3.2051ms]    Output: 4_914
a_AND_num_asc_fast:[0_TO_9]_all_results                             Avg: 4.1547ms (-31.12%)    Median: 4.0933ms (-28.55%)    [3.7648ms .. 4.7600ms]    Output: 10_152
num_rand_fast:[0_TO_9]_AND_num_asc_fast:[0_TO_9]_all_results        Avg: 2.6973ms (-52.30%)    Median: 2.6901ms (-49.86%)    [2.6689ms .. 2.7677ms]    Output: 968
```

Gains are largest when the range query is the non-leading docset of a low-cardinality intersection.

.github/workflows/coverage.yml

@@ -28,7 +28,7 @@ jobs:
       - name: Generate code coverage
         run: cargo +nightly-2025-12-01 llvm-cov --all-features --workspace --doctests --lcov --output-path lcov.info
       - name: Upload coverage to Codecov
-        uses: codecov/codecov-action@57e3a136b779b570ffcdbf80b3bdc90e7fab3de2 # v6.0.0
+        uses: codecov/codecov-action@fb8b3582c8e4def4969c97caa2f19720cb33a72f # v7.0.0
         continue-on-error: true
         with:
           token: ${{ secrets.CODECOV_TOKEN }} # not required for public repos

benches/agg_bench.rs

@@ -66,6 +66,9 @@ fn bench_agg(mut group: InputGroup<Index>) {
     register!(group, terms_status_with_terms_zipf_1000_sub_agg);
     register!(group, terms_zipf_1000_with_terms_status_sub_agg);
     register!(group, terms_status_with_histogram);
+    register!(group, terms_status_with_date_histogram);
+    register!(group, terms_status_with_date_histogram_hard_bounds);
+    register!(group, terms_status_with_date_histogram_and_sibling_terms);
     register!(group, terms_zipf_1000);
     register!(group, terms_zipf_1000_with_histogram);
     register!(group, terms_zipf_1000_with_avg_sub_agg);
@@ -390,6 +393,57 @@ fn terms_status_with_histogram(index: &Index) {
     execute_agg(index, agg_req);
 }
 
+fn terms_status_with_date_histogram(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": { "field": "text_few_terms_status" },
+            "aggs": {
+                "over_time": { "date_histogram": { "field": "timestamp", "fixed_interval": "1h" } }
+            }
+        }
+    });
+    execute_agg(index, agg_req);
+}
+
+/// Same fused terms × date_histogram, but with `hard_bounds`. The timestamps span 0..120h; the
+/// bounds drop only the first and last hour (ms: 1h=3_600_000, 119h=428_400_000), so almost every
+/// doc is in-bounds. This exercises the collector's hard-bounds path: `bounds.contains` runs per
+/// doc (the `all_docs_in_bounds` short-circuit is off) and the rare out-of-bounds doc takes the
+/// `term_counts` branch.
+fn terms_status_with_date_histogram_hard_bounds(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": { "field": "text_few_terms_status" },
+            "aggs": {
+                "over_time": {
+                    "date_histogram": {
+                        "field": "timestamp",
+                        "fixed_interval": "1h",
+                        "hard_bounds": { "min": 3_600_000, "max": 428_400_000 }
+                    }
+                }
+            }
+        }
+    });
+    execute_agg(index, agg_req);
+}
+
+/// Same fused terms × date_histogram, but with a sibling terms aggregation next to it. The fused
+/// fast path should still trigger for `my_texts` (sibling aggregations are independent top-level
+/// aggregations, so they don't change its eligibility).
+fn terms_status_with_date_histogram_and_sibling_terms(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": { "field": "text_few_terms_status" },
+            "aggs": {
+                "over_time": { "date_histogram": { "field": "timestamp", "fixed_interval": "1h" } }
+            }
+        },
+        "other_texts": { "terms": { "field": "text_few_terms" } }
+    });
+    execute_agg(index, agg_req);
+}
+
 fn terms_zipf_1000_with_histogram(index: &Index) {
     let agg_req = json!({
         "my_texts": {
@@ -783,7 +837,9 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
             doc_with_value /= 20;
         }
         let _val_max = 1_000_000.0;
-        for _ in 0..doc_with_value {
+        const SPAN_MS: i64 = 120 * 3600 * 1000; // 120 hours in ms
+        const NOISE_MS: i64 = 2 * 3600 * 1000; // ±2h noise
+        for i in 0..doc_with_value {
             let val: f64 = rng.random_range(0.0..1_000_000.0);
             let json = if rng.random_bool(0.1) {
                 // 10% are numeric values
@@ -791,6 +847,9 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
             } else {
                 json!({"mixed_type": many_terms_data.choose(&mut rng).unwrap().to_string()})
             };
+            let base_ms = (i as i64 * SPAN_MS) / doc_with_value as i64;
+            let noise_ms = rng.random_range(-NOISE_MS..NOISE_MS);
+            let ts_ms = (base_ms + noise_ms).clamp(0, SPAN_MS);
             index_writer.add_document(doc!(
                 single_term => "single_term",
                 text_field => "cool",
@@ -803,7 +862,7 @@ fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
                 score_field => val as u64,
                 score_field_f64 => lg_norm.sample(&mut rng),
                 score_field_i64 => val as i64,
-                date_field => DateTime::from_timestamp_millis((val * 1_000_000.) as i64),
+                date_field => DateTime::from_timestamp_millis(ts_ms),
             ))?;
             if cardinality == Cardinality::OptionalSparse {
                 for _ in 0..20 {

benches/bool_queries_with_range.rs

@@ -110,43 +110,31 @@ fn main() {
     // Prepare corpora with varying scenarios
     let scenarios = vec![
         (
-            "dense and 99% a".to_string(),
-            10_000_000,
-            0.99,
+            "dense and 0.1% a".to_string(),
+            5_000_000,
+            0.001,
             "dense",
             0,
             9,
         ),
+        ("dense and 1% a".to_string(), 5_000_000, 0.01, "dense", 0, 9),
+        ("dense and 10% a".to_string(), 5_000_000, 0.1, "dense", 0, 9),
         (
-            "dense and 99% a".to_string(),
-            10_000_000,
-            0.99,
-            "dense",
-            990,
-            999,
-        ),
-        (
-            "sparse and 99% a".to_string(),
-            10_000_000,
+            "sparse and 50% a".to_string(),
+            5_000_000,
             0.99,
             "sparse",
             0,
             9,
         ),
-        (
-            "sparse and 99% a".to_string(),
-            10_000_000,
-            0.99,
-            "sparse",
-            9_999_990,
-            9_999_999,
-        ),
     ];
 
     let mut runner = BenchRunner::new();
-    for (scenario_id, n, p_title_a, num_rand_distribution, range_low, range_high) in scenarios {
+    for (scenario_id, num_docs, p_title_a, num_rand_distribution, range_low, range_high) in
+        scenarios
+    {
         // Build index for this scenario
-        let bench_index = build_shared_indices(n, p_title_a, num_rand_distribution);
+        let bench_index = build_shared_indices(num_docs, p_title_a, num_rand_distribution);
 
         // Create benchmark group
         let mut group = runner.new_group();
@@ -158,7 +146,7 @@ fn main() {
         let field_names = ["num_rand", "num_asc", "num_rand_fast", "num_asc_fast"];
 
         // Define the three terms we want to test with
-        let terms = ["a", "b", "z"];
+        let terms = ["a"];
 
         // Generate all combinations of terms and field names
         let mut queries = Vec::new();
@@ -203,7 +191,7 @@ fn run_benchmark_tasks(
         bench_index,
         query_str,
         DocSetCollector,
-        "all results",
+        "all_results",
     );
 
     // Test top 100 by the field (if it's a FAST field)

columnar/src/block_accessor.rs

@@ -15,9 +15,37 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
 {
     #[inline]
     pub fn fetch_block<'a>(&'a mut self, docs: &'a [u32], accessor: &Column<T>) {
-        if accessor.index.get_cardinality().is_full() {
-            self.val_cache.resize(docs.len(), T::default());
-            accessor.values.get_vals(docs, &mut self.val_cache);
+        self.fetch_block_with_is_full(docs, accessor, accessor.index.get_cardinality().is_full());
+    }
+
+    /// Like [`Self::fetch_block`] but takes the column's fullness instead of querying
+    /// `accessor.index.get_cardinality()` each call — for callers that know it up front (e.g.
+    /// checked once at construction). `is_full` must equal
+    /// `accessor.index.get_cardinality().is_full()`.
+    #[inline]
+    pub fn fetch_block_with_is_full<'a>(
+        &'a mut self,
+        docs: &'a [u32],
+        accessor: &Column<T>,
+        is_full: bool,
+    ) {
+        if is_full {
+            // Skip the resize when already the right length (common case: fixed-size blocks).
+            if self.val_cache.len() != docs.len() {
+                self.val_cache.resize(docs.len(), T::default());
+            }
+            // When the docs form a contiguous ascending run we can fetch the values
+            // as a single range. This lets codecs (e.g. bitpacked) bulk-decode the
+            // slice instead of gathering value-by-value, and avoids per-value dynamic
+            // dispatch. `docs` is always sorted ascending and free of duplicates here,
+            // so comparing the endpoints is enough to detect contiguity.
+            if is_contiguous(docs) {
+                accessor
+                    .values
+                    .get_range(docs[0] as u64, &mut self.val_cache);
+            } else {
+                accessor.values.get_vals(docs, &mut self.val_cache);
+            }
         } else {
             self.docid_cache.clear();
             self.row_id_cache.clear();
@@ -158,6 +186,22 @@ impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
     }
 }
 
+/// Returns true if `docs` is a contiguous ascending run `[d, d + 1, ..., d + n - 1]`.
+///
+/// Assumes `docs` is sorted ascending and free of duplicates (the invariant for the
+/// doc blocks passed to `fetch_block`), so comparing the endpoints is sufficient.
+#[inline]
+fn is_contiguous(docs: &[u32]) -> bool {
+    let (Some(&first), Some(&last)) = (docs.first(), docs.last()) else {
+        return false;
+    };
+    debug_assert!(
+        docs.windows(2).all(|w| w[0] < w[1]),
+        "fetch_block requires docs sorted ascending without duplicates"
+    );
+    (last - first) as usize + 1 == docs.len()
+}
+
 /// Given two sorted lists of docids `docs` and `hits`, hits is a subset of `docs`.
 /// Return all docs that are not in `hits`.
 fn find_missing_docs<F>(docs: &[u32], hits: &[u32], mut callback: F)
@@ -288,4 +332,46 @@ mod tests {
         assert_eq!(accessor.docid_cache, vec![0]);
         assert_eq!(accessor.val_cache, vec![1]);
     }
+
+    #[test]
+    fn test_is_contiguous() {
+        assert!(!is_contiguous(&[]));
+        assert!(is_contiguous(&[5]));
+        assert!(is_contiguous(&[5, 6, 7, 8]));
+        assert!(is_contiguous(&[0, 1, 2]));
+        assert!(!is_contiguous(&[5, 7, 8]));
+        assert!(!is_contiguous(&[0, 1, 3]));
+    }
+
+    #[test]
+    fn test_fetch_block_contiguous_and_gather_match() {
+        use crate::column_index::ColumnIndex;
+        use crate::column_values::{
+            ALL_U64_CODEC_TYPES, serialize_and_load_u64_based_column_values,
+        };
+
+        let vals: Vec<u64> = (0..200u64).map(|i| i * 7 + 3).collect();
+        let values =
+            serialize_and_load_u64_based_column_values::<u64>(&&vals[..], &ALL_U64_CODEC_TYPES);
+        let column = Column {
+            index: ColumnIndex::Full,
+            values,
+        };
+
+        let check = |accessor: &mut ColumnBlockAccessor<u64>, docs: &[u32]| {
+            accessor.fetch_block(docs, &column);
+            let got: Vec<(u32, u64)> = accessor.iter_docid_vals(docs, &column).collect();
+            let expected: Vec<(u32, u64)> = docs.iter().map(|&d| (d, vals[d as usize])).collect();
+            assert_eq!(got, expected);
+        };
+
+        let mut accessor = ColumnBlockAccessor::<u64>::default();
+        // Contiguous block -> get_range fast path.
+        check(&mut accessor, &(10..74).collect::<Vec<u32>>());
+        // Non-contiguous block -> get_vals gather path.
+        check(&mut accessor, &[0, 5, 9, 100, 199]);
+        // Single doc and full span.
+        check(&mut accessor, &[42]);
+        check(&mut accessor, &(0..200).collect::<Vec<u32>>());
+    }
 }

columnar/src/column_values/mod.rs

@@ -119,8 +119,18 @@ pub trait ColumnValues<T: PartialOrd = u64>: Send + Sync + DowncastSync {
     /// the segment's `maxdoc`.
     #[inline(always)]
     fn get_range(&self, start: u64, output: &mut [T]) {
-        for (out, idx) in output.iter_mut().zip(start..) {
+        let mut out_chunks = output.chunks_exact_mut(4);
+        let mut idx = start;
+        for out_x4 in out_chunks.by_ref() {
+            out_x4[0] = self.get_val(idx as u32);
+            out_x4[1] = self.get_val((idx + 1) as u32);
+            out_x4[2] = self.get_val((idx + 2) as u32);
+            out_x4[3] = self.get_val((idx + 3) as u32);
+            idx += 4;
+        }
+        for out in out_chunks.into_remainder() {
             *out = self.get_val(idx as u32);
+            idx += 1;
         }
     }
 

columnar/src/column_values/u64_based/tests.rs

@@ -121,6 +121,22 @@ pub(crate) fn create_and_validate<TColumnCodec: ColumnCodec>(
     reader.get_vals(&all_docs, &mut buffer);
     assert_eq!(vals, buffer);
 
+    // Validate `get_range` over the full column and a sub-range. The sub-range starts
+    // at a non-zero offset to exercise the entrance-ramp alignment of the batch decode.
+    buffer.resize(all_docs.len(), 0);
+    reader.get_range(0, &mut buffer);
+    assert_eq!(vals, buffer, "get_range (full) mismatch in data set {name}");
+    if vals.len() >= 2 {
+        let start = 1usize;
+        buffer.resize(vals.len() - start, 0);
+        reader.get_range(start as u64, &mut buffer);
+        assert_eq!(
+            &vals[start..],
+            &buffer[..],
+            "get_range (sub-range) mismatch in data set {name}"
+        );
+    }
+
     if !vals.is_empty() {
         let test_rand_idx = rand::rng().random_range(0..=vals.len() - 1);
         let expected_positions: Vec<u32> = vals

common/src/bitset.rs

@@ -196,11 +196,13 @@ impl TinySet {
 #[derive(Clone)]
 pub struct BitSet {
     tinysets: Box<[TinySet]>,
+    len: u64,
     max_value: u32,
 }
 impl std::fmt::Debug for BitSet {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         f.debug_struct("BitSet")
+            .field("len", &self.len)
             .field("max_value", &self.max_value)
             .finish()
     }
@@ -228,6 +230,7 @@ impl BitSet {
         let tinybitsets = vec![TinySet::empty(); num_buckets as usize].into_boxed_slice();
         BitSet {
             tinysets: tinybitsets,
+            len: 0,
             max_value,
         }
     }
@@ -245,6 +248,7 @@ impl BitSet {
         }
         BitSet {
             tinysets: tinybitsets,
+            len: max_value as u64,
             max_value,
         }
     }
@@ -263,19 +267,17 @@ impl BitSet {
 
     /// Intersect with tinysets
     fn intersect_update_with_iter(&mut self, other: impl Iterator<Item = TinySet>) {
+        self.len = 0;
         for (left, right) in self.tinysets.iter_mut().zip(other) {
             *left = left.intersect(right);
+            self.len += left.len() as u64;
         }
     }
 
     /// Returns the number of elements in the `BitSet`.
     #[inline]
     pub fn len(&self) -> usize {
-        self.tinysets
-            .iter()
-            .copied()
-            .map(|tinyset| tinyset.len())
-            .sum::<u32>() as usize
+        self.len as usize
     }
 
     /// Inserts an element in the `BitSet`
@@ -284,7 +286,7 @@ impl BitSet {
         // we do not check saturated els.
         let higher = el / 64u32;
         let lower = el % 64u32;
-        self.tinysets[higher as usize].insert_mut(lower);
+        self.len += u64::from(self.tinysets[higher as usize].insert_mut(lower));
     }
 
     /// Inserts an element in the `BitSet`
@@ -293,7 +295,7 @@ impl BitSet {
         // we do not check saturated els.
         let higher = el / 64u32;
         let lower = el % 64u32;
-        self.tinysets[higher as usize].remove_mut(lower);
+        self.len -= u64::from(self.tinysets[higher as usize].remove_mut(lower));
     }
 
     /// Returns true iff the elements is in the `BitSet`.
@@ -315,9 +317,6 @@ impl BitSet {
             .map(|delta_bucket| bucket + delta_bucket as u32)
     }
 
-    /// Returns the maximum number of elements in the bitset.
-    ///
-    /// Warning: The largest element the bitset can contain is `max_value - 1`.
     #[inline]
     pub fn max_value(&self) -> u32 {
         self.max_value

examples/iterating_docs_and_positions.rs

@@ -91,10 +91,46 @@ fn main() -> tantivy::Result<()> {
         }
     }
 
-    // Some other powerful operations (especially `.seek`) may be useful to consume these
+    // A `Term` is a text token associated with a field.
+    // Let's go through all docs containing the term `title:the` and access their position
+    let term_the = Term::from_field_text(title, "the");
+
+    // Some other powerful operations (especially `.skip_to`) may be useful to consume these
     // posting lists rapidly.
     // You can check for them in the [`DocSet`](https://docs.rs/tantivy/~0/tantivy/trait.DocSet.html) trait
     // and the [`Postings`](https://docs.rs/tantivy/~0/tantivy/trait.Postings.html) trait
 
+    // Also, for some VERY specific high performance use case like an OLAP analysis of logs,
+    // you can get better performance by accessing directly the blocks of doc ids.
+    for segment_reader in searcher.segment_readers() {
+        // A segment contains different data structure.
+        // Inverted index stands for the combination of
+        // - the term dictionary
+        // - the inverted lists associated with each terms and their positions
+        let inverted_index = segment_reader.inverted_index(title)?;
+
+        // This segment posting object is like a cursor over the documents matching the term.
+        // The `IndexRecordOption` arguments tells tantivy we will be interested in both term
+        // frequencies and positions.
+        //
+        // If you don't need all this information, you may get better performance by decompressing
+        // less information.
+        if let Some(mut block_segment_postings) =
+            inverted_index.read_block_postings(&term_the, IndexRecordOption::Basic)?
+        {
+            loop {
+                let docs = block_segment_postings.docs();
+                if docs.is_empty() {
+                    break;
+                }
+                // Once again these docs MAY contains deleted documents as well.
+                let docs = block_segment_postings.docs();
+                // Prints `Docs [0, 2].`
+                println!("Docs {docs:?}");
+                block_segment_postings.advance();
+            }
+        }
+    }
+
     Ok(())
 }

query-grammar/src/query_grammar.rs

@@ -327,7 +327,9 @@ fn exists(inp: &str) -> IResult<&str, UserInputLeaf> {
             peek(alt((
                 value(
                     "",
-                    satisfy(|c: char| c.is_whitespace() || ESCAPE_IN_WORD.contains(&c)),
+                    satisfy(|c: char| {
+                        c.is_whitespace() || (ESCAPE_IN_WORD.contains(&c) && c != '\\')
+                    }),
                 ),
                 eof,
             ))),
@@ -345,7 +347,9 @@ fn exists_precond(inp: &str) -> IResult<&str, (), ()> {
             peek(alt((
                 value(
                     "",
-                    satisfy(|c: char| c.is_whitespace() || ESCAPE_IN_WORD.contains(&c)),
+                    satisfy(|c: char| {
+                        c.is_whitespace() || (ESCAPE_IN_WORD.contains(&c) && c != '\\')
+                    }),
                 ),
                 eof,
             ))), // we need to check this isn't a wildcard query
@@ -707,6 +711,7 @@ fn regex(inp: &str) -> IResult<&str, UserInputLeaf> {
             peek(alt((
                 value((), multispace1),
                 value((), char(')')),
+                value((), char('^')),
                 value((), eof),
             ))),
         ),
@@ -728,9 +733,10 @@ fn regex_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
             peek(alt((
                 value((), multispace1),
                 value((), char(')')),
+                value((), char('^')),
                 value((), eof),
             ))),
-            "expected whitespace, closing parenthesis, or end of input",
+            "expected whitespace, closing parenthesis, boost, or end of input",
         ),
     )(inp)
     {
@@ -773,6 +779,10 @@ fn leaf(inp: &str) -> IResult<&str, UserInputAst> {
                     value((), multispace1),
                     value((), char(')')),
                     value((), eof),
+                    value(
+                        (),
+                        satisfy(|c: char| ESCAPE_IN_WORD.contains(&c) && c != '\\'),
+                    ),
                 ))),
             ),
             |_| UserInputAst::from(UserInputLeaf::All),
@@ -805,6 +815,10 @@ fn leaf_infallible(inp: &str) -> JResult<&str, Option<UserInputAst>> {
                             value((), multispace1),
                             value((), char(')')),
                             value((), eof),
+                            value(
+                                (),
+                                satisfy(|c: char| ESCAPE_IN_WORD.contains(&c) && c != '\\'),
+                            ),
                         ))),
                     ),
                 ),
@@ -1751,6 +1765,8 @@ mod test {
         test_parse_query_to_ast_helper("*", "*");
         test_parse_query_to_ast_helper("(*)", "*");
         test_parse_query_to_ast_helper("(* )", "*");
+        // All query with boost
+        test_parse_query_to_ast_helper("*^2", "(*)^2");
     }
 
     #[test]
@@ -1813,6 +1829,7 @@ mod test {
         test_parse_query_to_ast_helper("a:b*", "\"a\":b*");
         test_parse_query_to_ast_helper("a:*b", "\"a\":*b");
         test_parse_query_to_ast_helper(r#"a:*def*"#, "\"a\":*def*");
+        test_parse_query_to_ast_helper("a:*\\:foo", "\"a\":*:foo");
     }
 
     #[test]
@@ -1877,6 +1894,8 @@ mod test {
             },
             _ => panic!("Expected a leaf"),
         }
+        // Regex followed by `^boost`
+        test_parse_query_to_ast_helper(r#"foo:/bar/^2"#, r#"("foo":/bar/)^2"#);
     }
 
     #[test]

src/aggregation/agg_data.rs

@@ -10,11 +10,11 @@ use crate::aggregation::accessor_helpers::{
 };
 use crate::aggregation::agg_req::{Aggregation, AggregationVariants, Aggregations};
 use crate::aggregation::bucket::{
-    build_segment_filter_collector, build_segment_range_collector, CompositeAggReqData,
-    CompositeAggregation, CompositeSourceAccessors, FilterAggReqData, HistogramAggReqData,
-    HistogramBounds, IncludeExcludeParam, MissingTermAggReqData, RangeAggReqData,
-    SegmentHistogramCollector, TermMissingAgg, TermsAggReqData, TermsAggregation,
-    TermsAggregationInternal,
+    build_segment_filter_collector, build_segment_histogram_collector,
+    build_segment_range_collector, CompositeAggReqData, CompositeAggregation,
+    CompositeSourceAccessors, FilterAggReqData, HistogramAggReqData, HistogramBounds,
+    IncludeExcludeParam, MissingTermAggReqData, RangeAggReqData, TermMissingAgg, TermsAggReqData,
+    TermsAggregation, TermsAggregationInternal,
 };
 use crate::aggregation::metric::{
     build_segment_stats_collector, AverageAggregation, CardinalityAggReqData,
@@ -41,7 +41,7 @@ pub struct AggregationsSegmentCtx {
 
 impl AggregationsSegmentCtx {
     pub(crate) fn push_term_req_data(&mut self, data: TermsAggReqData) -> usize {
-        self.per_request.term_req_data.push(Some(Box::new(data)));
+        self.per_request.term_req_data.push(data);
         self.per_request.term_req_data.len() - 1
     }
     pub(crate) fn push_cardinality_req_data(&mut self, data: CardinalityAggReqData) -> usize {
@@ -61,31 +61,25 @@ impl AggregationsSegmentCtx {
         self.per_request.missing_term_req_data.len() - 1
     }
     pub(crate) fn push_histogram_req_data(&mut self, data: HistogramAggReqData) -> usize {
-        self.per_request
-            .histogram_req_data
-            .push(Some(Box::new(data)));
+        self.per_request.histogram_req_data.push(data);
         self.per_request.histogram_req_data.len() - 1
     }
     pub(crate) fn push_range_req_data(&mut self, data: RangeAggReqData) -> usize {
-        self.per_request.range_req_data.push(Some(Box::new(data)));
+        self.per_request.range_req_data.push(data);
         self.per_request.range_req_data.len() - 1
     }
     pub(crate) fn push_filter_req_data(&mut self, data: FilterAggReqData) -> usize {
-        self.per_request.filter_req_data.push(Some(Box::new(data)));
+        self.per_request.filter_req_data.push(data);
         self.per_request.filter_req_data.len() - 1
     }
     pub(crate) fn push_composite_req_data(&mut self, data: CompositeAggReqData) -> usize {
-        self.per_request
-            .composite_req_data
-            .push(Some(Box::new(data)));
+        self.per_request.composite_req_data.push(data);
         self.per_request.composite_req_data.len() - 1
     }
 
     #[inline]
     pub(crate) fn get_term_req_data(&self, idx: usize) -> &TermsAggReqData {
-        self.per_request.term_req_data[idx]
-            .as_deref()
-            .expect("term_req_data slot is empty (taken)")
+        &self.per_request.term_req_data[idx]
     }
     #[inline]
     pub(crate) fn get_cardinality_req_data(&self, idx: usize) -> &CardinalityAggReqData {
@@ -103,116 +97,6 @@ impl AggregationsSegmentCtx {
     pub(crate) fn get_missing_term_req_data(&self, idx: usize) -> &MissingTermAggReqData {
         &self.per_request.missing_term_req_data[idx]
     }
-    #[inline]
-    pub(crate) fn get_histogram_req_data(&self, idx: usize) -> &HistogramAggReqData {
-        self.per_request.histogram_req_data[idx]
-            .as_deref()
-            .expect("histogram_req_data slot is empty (taken)")
-    }
-    #[inline]
-    pub(crate) fn get_range_req_data(&self, idx: usize) -> &RangeAggReqData {
-        self.per_request.range_req_data[idx]
-            .as_deref()
-            .expect("range_req_data slot is empty (taken)")
-    }
-    #[inline]
-    pub(crate) fn get_composite_req_data(&self, idx: usize) -> &CompositeAggReqData {
-        self.per_request.composite_req_data[idx]
-            .as_deref()
-            .expect("composite_req_data slot is empty (taken)")
-    }
-
-    // ---------- mutable getters ----------
-
-    #[inline]
-    pub(crate) fn get_metric_req_data_mut(&mut self, idx: usize) -> &mut MetricAggReqData {
-        &mut self.per_request.stats_metric_req_data[idx]
-    }
-
-    #[inline]
-    pub(crate) fn get_cardinality_req_data_mut(
-        &mut self,
-        idx: usize,
-    ) -> &mut CardinalityAggReqData {
-        &mut self.per_request.cardinality_req_data[idx]
-    }
-
-    #[inline]
-    pub(crate) fn get_histogram_req_data_mut(&mut self, idx: usize) -> &mut HistogramAggReqData {
-        self.per_request.histogram_req_data[idx]
-            .as_deref_mut()
-            .expect("histogram_req_data slot is empty (taken)")
-    }
-
-    // ---------- take / put (terms, histogram, range) ----------
-
-    /// Move out the boxed Histogram request at `idx`, leaving `None`.
-    #[inline]
-    pub(crate) fn take_histogram_req_data(&mut self, idx: usize) -> Box<HistogramAggReqData> {
-        self.per_request.histogram_req_data[idx]
-            .take()
-            .expect("histogram_req_data slot is empty (taken)")
-    }
-
-    /// Put back a Histogram request into an empty slot at `idx`.
-    #[inline]
-    pub(crate) fn put_back_histogram_req_data(
-        &mut self,
-        idx: usize,
-        value: Box<HistogramAggReqData>,
-    ) {
-        debug_assert!(self.per_request.histogram_req_data[idx].is_none());
-        self.per_request.histogram_req_data[idx] = Some(value);
-    }
-
-    /// Move out the boxed Range request at `idx`, leaving `None`.
-    #[inline]
-    pub(crate) fn take_range_req_data(&mut self, idx: usize) -> Box<RangeAggReqData> {
-        self.per_request.range_req_data[idx]
-            .take()
-            .expect("range_req_data slot is empty (taken)")
-    }
-
-    /// Put back a Range request into an empty slot at `idx`.
-    #[inline]
-    pub(crate) fn put_back_range_req_data(&mut self, idx: usize, value: Box<RangeAggReqData>) {
-        debug_assert!(self.per_request.range_req_data[idx].is_none());
-        self.per_request.range_req_data[idx] = Some(value);
-    }
-
-    /// Move out the boxed Filter request at `idx`, leaving `None`.
-    #[inline]
-    pub(crate) fn take_filter_req_data(&mut self, idx: usize) -> Box<FilterAggReqData> {
-        self.per_request.filter_req_data[idx]
-            .take()
-            .expect("filter_req_data slot is empty (taken)")
-    }
-
-    /// Put back a Filter request into an empty slot at `idx`.
-    #[inline]
-    pub(crate) fn put_back_filter_req_data(&mut self, idx: usize, value: Box<FilterAggReqData>) {
-        debug_assert!(self.per_request.filter_req_data[idx].is_none());
-        self.per_request.filter_req_data[idx] = Some(value);
-    }
-
-    /// Move out the Composite request at `idx`.
-    #[inline]
-    pub(crate) fn take_composite_req_data(&mut self, idx: usize) -> Box<CompositeAggReqData> {
-        self.per_request.composite_req_data[idx]
-            .take()
-            .expect("composite_req_data slot is empty (taken)")
-    }
-
-    /// Put back a Composite request into an empty slot at `idx`.
-    #[inline]
-    pub(crate) fn put_back_composite_req_data(
-        &mut self,
-        idx: usize,
-        value: Box<CompositeAggReqData>,
-    ) {
-        debug_assert!(self.per_request.composite_req_data[idx].is_none());
-        self.per_request.composite_req_data[idx] = Some(value);
-    }
 }
 
 /// Each type of aggregation has its own request data struct. This struct holds
@@ -223,15 +107,14 @@ impl AggregationsSegmentCtx {
 /// for a node with [AggKind::Terms]).
 #[derive(Default)]
 pub struct PerRequestAggSegCtx {
-    // Box for cheap take/put - Only necessary for bucket aggs that have sub-aggregations
     /// TermsAggReqData contains the request data for a terms aggregation.
-    pub term_req_data: Vec<Option<Box<TermsAggReqData>>>,
+    pub term_req_data: Vec<TermsAggReqData>,
     /// HistogramAggReqData contains the request data for a histogram aggregation.
-    pub histogram_req_data: Vec<Option<Box<HistogramAggReqData>>>,
+    pub histogram_req_data: Vec<HistogramAggReqData>,
     /// RangeAggReqData contains the request data for a range aggregation.
-    pub range_req_data: Vec<Option<Box<RangeAggReqData>>>,
+    pub range_req_data: Vec<RangeAggReqData>,
     /// FilterAggReqData contains the request data for a filter aggregation.
-    pub filter_req_data: Vec<Option<Box<FilterAggReqData>>>,
+    pub filter_req_data: Vec<FilterAggReqData>,
     /// Shared by avg, min, max, sum, stats, extended_stats, count
     pub stats_metric_req_data: Vec<MetricAggReqData>,
     /// CardinalityAggReqData contains the request data for a cardinality aggregation.
@@ -241,7 +124,7 @@ pub struct PerRequestAggSegCtx {
     /// MissingTermAggReqData contains the request data for a missing term aggregation.
     pub missing_term_req_data: Vec<MissingTermAggReqData>,
     /// CompositeAggReqData contains the request data for a composite aggregation.
-    pub composite_req_data: Vec<Option<Box<CompositeAggReqData>>>,
+    pub composite_req_data: Vec<CompositeAggReqData>,
 
     /// Request tree used to build collectors.
     pub agg_tree: Vec<AggRefNode>,
@@ -252,22 +135,22 @@ impl PerRequestAggSegCtx {
     fn get_memory_consumption(&self) -> usize {
         self.term_req_data
             .iter()
-            .map(|b| b.as_ref().unwrap().get_memory_consumption())
+            .map(|t| t.get_memory_consumption())
             .sum::<usize>()
             + self
                 .histogram_req_data
                 .iter()
-                .map(|b| b.as_ref().unwrap().get_memory_consumption())
+                .map(|t| t.get_memory_consumption())
                 .sum::<usize>()
             + self
                 .range_req_data
                 .iter()
-                .map(|b| b.as_ref().unwrap().get_memory_consumption())
+                .map(|t| t.get_memory_consumption())
                 .sum::<usize>()
             + self
                 .filter_req_data
                 .iter()
-                .map(|b| b.as_ref().unwrap().get_memory_consumption())
+                .map(|t| t.get_memory_consumption())
                 .sum::<usize>()
             + self
                 .stats_metric_req_data
@@ -292,7 +175,7 @@ impl PerRequestAggSegCtx {
             + self
                 .composite_req_data
                 .iter()
-                .map(|b| b.as_ref().map(|d| d.get_memory_consumption()).unwrap_or(0))
+                .map(|t| t.get_memory_consumption())
                 .sum::<usize>()
             + self.agg_tree.len() * std::mem::size_of::<AggRefNode>()
     }
@@ -301,40 +184,16 @@ impl PerRequestAggSegCtx {
         let idx = node.idx_in_req_data;
         let kind = node.kind;
         match kind {
-            AggKind::Terms => self.term_req_data[idx]
-                .as_deref()
-                .expect("term_req_data slot is empty (taken)")
-                .name
-                .as_str(),
+            AggKind::Terms => self.term_req_data[idx].name.as_str(),
             AggKind::Cardinality => &self.cardinality_req_data[idx].name,
             AggKind::StatsKind(_) => &self.stats_metric_req_data[idx].name,
             AggKind::TopHits => &self.top_hits_req_data[idx].name,
             AggKind::MissingTerm => &self.missing_term_req_data[idx].name,
-            AggKind::Histogram => self.histogram_req_data[idx]
-                .as_deref()
-                .expect("histogram_req_data slot is empty (taken)")
-                .name
-                .as_str(),
-            AggKind::DateHistogram => self.histogram_req_data[idx]
-                .as_deref()
-                .expect("histogram_req_data slot is empty (taken)")
-                .name
-                .as_str(),
-            AggKind::Range => self.range_req_data[idx]
-                .as_deref()
-                .expect("range_req_data slot is empty (taken)")
-                .name
-                .as_str(),
-            AggKind::Filter => self.filter_req_data[idx]
-                .as_deref()
-                .expect("filter_req_data slot is empty (taken)")
-                .name
-                .as_str(),
-            AggKind::Composite => self.composite_req_data[idx]
-                .as_deref()
-                .expect("composite_req_data slot is empty (taken)")
-                .name
-                .as_str(),
+            AggKind::Histogram => self.histogram_req_data[idx].name.as_str(),
+            AggKind::DateHistogram => self.histogram_req_data[idx].name.as_str(),
+            AggKind::Range => self.range_req_data[idx].name.as_str(),
+            AggKind::Filter => self.filter_req_data[idx].name.as_str(),
+            AggKind::Composite => self.composite_req_data[idx].name.as_str(),
         }
     }
 
@@ -412,7 +271,7 @@ pub(crate) fn build_segment_agg_collector(
             Ok(Box::new(TermMissingAgg::new(req, node)?))
         }
         AggKind::Cardinality => {
-            let req_data = &mut req.get_cardinality_req_data_mut(node.idx_in_req_data);
+            let req_data = req.get_cardinality_req_data(node.idx_in_req_data);
             // For str columns, choose the per-bucket entries representation
             // based on the segment's column.max_value():
             //   * small (< BITSET_MAX_TERM_ORD): `BitSet`, pre-allocated, no promotion machinery.
@@ -459,7 +318,7 @@ pub(crate) fn build_segment_agg_collector(
                     SegmentExtendedStatsCollector::from_req(req_data, sigma),
                 )),
                 StatsType::Percentiles => {
-                    let req_data = req.get_metric_req_data_mut(node.idx_in_req_data);
+                    let req_data = req.get_metric_req_data(node.idx_in_req_data);
                     Ok(Box::new(
                         SegmentPercentilesCollector::from_req_and_validate(
                             req_data.field_type,
@@ -479,12 +338,8 @@ pub(crate) fn build_segment_agg_collector(
                 req_data.segment_ordinal,
             )))
         }
-        AggKind::Histogram => Ok(Box::new(SegmentHistogramCollector::from_req_and_validate(
-            req, node,
-        )?)),
-        AggKind::DateHistogram => Ok(Box::new(SegmentHistogramCollector::from_req_and_validate(
-            req, node,
-        )?)),
+        AggKind::Histogram => build_segment_histogram_collector(req, node),
+        AggKind::DateHistogram => build_segment_histogram_collector(req, node),
         AggKind::Range => Ok(build_segment_range_collector(req, node)?),
         AggKind::Filter => build_segment_filter_collector(req, node),
         AggKind::Composite => Ok(Box::new(
@@ -799,23 +654,18 @@ fn build_nodes(
             let schema = reader.schema();
             let tokenizers = &data.context.tokenizers;
             let query = filter_req.parse_query(schema, tokenizers)?;
-            let evaluator = crate::aggregation::bucket::DocumentQueryEvaluator::new(
-                query,
-                schema.clone(),
-                reader,
-            )?;
-
-            // Pre-allocate buffer for batch filtering
-            let max_doc = reader.max_doc();
-            let buffer_capacity = crate::docset::COLLECT_BLOCK_BUFFER_LEN.min(max_doc as usize);
-            let matching_docs_buffer = Vec::with_capacity(buffer_capacity);
+            let evaluator =
+                std::rc::Rc::new(crate::aggregation::bucket::DocumentQueryEvaluator::new(
+                    query,
+                    schema.clone(),
+                    reader,
+                )?);
 
             let idx_in_req_data = data.push_filter_req_data(FilterAggReqData {
                 name: agg_name.to_string(),
                 req: filter_req.clone(),
                 segment_reader: reader.clone(),
                 evaluator,
-                matching_docs_buffer,
                 is_top_level,
             });
             let children = build_children(&req.sub_aggregation, reader, segment_ordinal, data)?;

src/aggregation/agg_req.rs

@@ -299,6 +299,12 @@ impl AggregationVariants {
             _ => None,
         }
     }
+    pub(crate) fn as_sum(&self) -> Option<&SumAggregation> {
+        match &self {
+            AggregationVariants::Sum(sum) => Some(sum),
+            _ => None,
+        }
+    }
 }
 
 #[cfg(test)]

src/aggregation/bucket/composite/accessors.rs

@@ -16,6 +16,7 @@ use crate::{SegmentReader, TantivyError};
 
 /// Contains all information required by the SegmentCompositeCollector to perform the
 /// composite aggregation on a segment.
+#[derive(Debug, Clone)]
 pub struct CompositeAggReqData {
     /// The name of the aggregation.
     pub name: String,
@@ -34,6 +35,7 @@ impl CompositeAggReqData {
 }
 
 /// Accessors for a single column in a composite source.
+#[derive(Debug, Clone)]
 pub struct CompositeAccessor {
     /// The fast field column
     pub column: Column<u64>,
@@ -48,6 +50,7 @@ pub struct CompositeAccessor {
 }
 
 /// Accessors to all the columns that belong to the field of a composite source.
+#[derive(Debug, Clone)]
 pub struct CompositeSourceAccessors {
     /// The accessors for this source
     pub accessors: Vec<CompositeAccessor>,
@@ -358,7 +361,7 @@ impl PrecomputedDateInterval {
 ///
 /// Some column types (term, IP) might not have an exact representation of the
 /// specified after key
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub enum PrecomputedAfterKey {
     /// The after key could be exactly represented in the column space.
     Exact(u64),

src/aggregation/bucket/composite/collector.rs

@@ -118,7 +118,7 @@ impl InternalValueRepr {
 pub struct SegmentCompositeCollector {
     /// One DynArrayHeapMap per parent bucket.
     parent_buckets: Vec<DynArrayHeapMap<InternalValueRepr, CompositeBucketCollector>>,
-    accessor_idx: usize,
+    req_data: CompositeAggReqData,
     sub_agg: Option<BufferedSubAggs<HighCardSubAggBuffer>>,
     bucket_id_provider: BucketIdProvider,
     /// Number of sources, needed when creating new DynArrayHeapMaps.
@@ -132,10 +132,7 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
         results: &mut IntermediateAggregationResults,
         parent_bucket_id: BucketId,
     ) -> crate::Result<()> {
-        let name = agg_data
-            .get_composite_req_data(self.accessor_idx)
-            .name
-            .clone();
+        let name = self.req_data.name.clone();
 
         let buckets = self.add_intermediate_bucket_result(agg_data, parent_bucket_id)?;
         results.push(
@@ -153,12 +150,11 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
         agg_data: &mut AggregationsSegmentCtx,
     ) -> crate::Result<()> {
         let mem_pre = self.get_memory_consumption(parent_bucket_id);
-        let composite_agg_data = agg_data.take_composite_req_data(self.accessor_idx);
 
         for doc in docs {
             let mut visitor = CompositeKeyVisitor {
                 doc_id: *doc,
-                composite_agg_data: &composite_agg_data,
+                composite_agg_data: &self.req_data,
                 buckets: &mut self.parent_buckets[parent_bucket_id as usize],
                 sub_agg: &mut self.sub_agg,
                 bucket_id_provider: &mut self.bucket_id_provider,
@@ -166,7 +162,6 @@ impl SegmentAggregationCollector for SegmentCompositeCollector {
             };
             visitor.visit(0, true)?;
         }
-        agg_data.put_back_composite_req_data(self.accessor_idx, composite_agg_data);
 
         if let Some(sub_agg) = &mut self.sub_agg {
             sub_agg.check_flush_local(agg_data)?;
@@ -221,7 +216,13 @@ impl SegmentCompositeCollector {
         req_data: &mut AggregationsSegmentCtx,
         node: &AggRefNode,
     ) -> crate::Result<Self> {
-        validate_req(req_data, node.idx_in_req_data)?;
+        let composite_req_data =
+            req_data.per_request.composite_req_data[node.idx_in_req_data].clone();
+        validate_req(&composite_req_data)?;
+        req_data
+            .context
+            .limits
+            .add_memory_consumed(composite_req_data.get_memory_consumption() as u64)?;
 
         let has_sub_aggregations = !node.children.is_empty();
         let sub_agg = if has_sub_aggregations {
@@ -231,12 +232,11 @@ impl SegmentCompositeCollector {
             None
         };
 
-        let composite_req_data = req_data.get_composite_req_data(node.idx_in_req_data);
         let num_sources = composite_req_data.req.sources.len();
 
         Ok(SegmentCompositeCollector {
             parent_buckets: vec![DynArrayHeapMap::try_new(num_sources)?],
-            accessor_idx: node.idx_in_req_data,
+            req_data: composite_req_data,
             sub_agg,
             bucket_id_provider: BucketIdProvider::default(),
             num_sources,
@@ -258,7 +258,7 @@ impl SegmentCompositeCollector {
         let mut dict: FxHashMap<Vec<CompositeIntermediateKey>, IntermediateCompositeBucketEntry> =
             Default::default();
         dict.reserve(heap_map.size());
-        let composite_data = agg_data.get_composite_req_data(self.accessor_idx);
+        let composite_data = &self.req_data;
         for (key_internal_repr, agg) in heap_map.into_iter() {
             let key = resolve_key(&key_internal_repr, composite_data)?;
             let mut sub_aggregation_res = IntermediateAggregationResults::default();
@@ -298,8 +298,7 @@ impl SegmentCompositeCollector {
     }
 }
 
-fn validate_req(req_data: &mut AggregationsSegmentCtx, accessor_idx: usize) -> crate::Result<()> {
-    let composite_data = req_data.get_composite_req_data(accessor_idx);
+fn validate_req(composite_data: &CompositeAggReqData) -> crate::Result<()> {
     let req = &composite_data.req;
     if req.sources.is_empty() {
         return Err(TantivyError::InvalidArgument(

src/aggregation/bucket/filter.rs

@@ -1,4 +1,5 @@
 use std::fmt::Debug;
+use std::rc::Rc;
 
 use common::BitSet;
 use serde::{Deserialize, Deserializer, Serialize, Serializer};
@@ -396,6 +397,7 @@ impl PartialEq for FilterAggregation {
 
 /// Request data for filter aggregation
 /// This struct holds the per-segment data needed to execute a filter aggregation
+#[derive(Clone)]
 pub struct FilterAggReqData {
     /// The name of the filter aggregation
     pub name: String,
@@ -403,22 +405,20 @@ pub struct FilterAggReqData {
     pub req: FilterAggregation,
     /// The segment reader
     pub segment_reader: SegmentReader,
-    /// Document evaluator for the filter query (precomputed BitSet)
-    /// This is built once when the request data is created
-    pub evaluator: DocumentQueryEvaluator,
-    /// Reusable buffer for matching documents to minimize allocations during collection
-    pub matching_docs_buffer: Vec<DocId>,
+    /// Document evaluator for the filter query (precomputed BitSet).
+    /// Wrapped in `Rc` so cloning the request data does not duplicate the (potentially large)
+    /// underlying BitSet.
+    pub evaluator: Rc<DocumentQueryEvaluator>,
     /// True if this filter aggregation is at the top level of the aggregation tree (not nested).
     pub is_top_level: bool,
 }
 
 impl FilterAggReqData {
     pub(crate) fn get_memory_consumption(&self) -> usize {
-        // Estimate: name + segment reader reference + bitset + buffer capacity
+        // Estimate: name + segment reader reference + bitset
         self.name.len()
         + std::mem::size_of::<SegmentReader>()
         + self.evaluator.bitset.len() / 8 // BitSet memory (bits to bytes)
-        + self.matching_docs_buffer.capacity() * std::mem::size_of::<DocId>()
         + std::mem::size_of::<bool>()
     }
 }
@@ -509,8 +509,10 @@ pub struct SegmentFilterCollector<B: SubAggBuffer> {
     /// Sub-aggregation collectors
     sub_aggregations: Option<BufferedSubAggs<B>>,
     bucket_id_provider: BucketIdProvider,
-    /// Accessor index for this filter aggregation (to access FilterAggReqData)
-    accessor_idx: usize,
+    /// Per-segment filter request data, owned by this collector.
+    req_data: FilterAggReqData,
+    /// Reusable buffer for matching documents to minimize allocations during collection.
+    matching_docs_buffer: Vec<DocId>,
 }
 
 impl<B: SubAggBuffer> SegmentFilterCollector<B> {
@@ -518,6 +520,7 @@ impl<B: SubAggBuffer> SegmentFilterCollector<B> {
     pub(crate) fn from_req_and_validate(
         req: &mut AggregationsSegmentCtx,
         node: &AggRefNode,
+        req_data: FilterAggReqData,
     ) -> crate::Result<Self> {
         // Build sub-aggregation collectors if any
         let sub_agg_collector = if !node.children.is_empty() {
@@ -527,11 +530,15 @@ impl<B: SubAggBuffer> SegmentFilterCollector<B> {
         };
         let sub_agg_collector = sub_agg_collector.map(BufferedSubAggs::new);
 
+        let max_doc = req_data.segment_reader.max_doc();
+        let buffer_capacity = crate::docset::COLLECT_BLOCK_BUFFER_LEN.min(max_doc as usize);
+
         Ok(SegmentFilterCollector {
             parent_buckets: Vec::new(),
             sub_aggregations: sub_agg_collector,
-            accessor_idx: node.idx_in_req_data,
+            req_data,
             bucket_id_provider: BucketIdProvider::default(),
+            matching_docs_buffer: Vec::with_capacity(buffer_capacity),
         })
     }
 }
@@ -540,18 +547,23 @@ pub(crate) fn build_segment_filter_collector(
     req: &mut AggregationsSegmentCtx,
     node: &AggRefNode,
 ) -> crate::Result<Box<dyn SegmentAggregationCollector>> {
-    let is_top_level = req.per_request.filter_req_data[node.idx_in_req_data]
-        .as_ref()
-        .expect("filter_req_data slot is empty")
-        .is_top_level;
+    let req_data = req.per_request.filter_req_data[node.idx_in_req_data].clone();
+    req.context
+        .limits
+        .add_memory_consumed(req_data.get_memory_consumption() as u64)?;
+    let is_top_level = req_data.is_top_level;
 
     if is_top_level {
         Ok(Box::new(
-            SegmentFilterCollector::<LowCardSubAggBuffer>::from_req_and_validate(req, node)?,
+            SegmentFilterCollector::<LowCardSubAggBuffer>::from_req_and_validate(
+                req, node, req_data,
+            )?,
         ))
     } else {
         Ok(Box::new(
-            SegmentFilterCollector::<HighCardSubAggBuffer>::from_req_and_validate(req, node)?,
+            SegmentFilterCollector::<HighCardSubAggBuffer>::from_req_and_validate(
+                req, node, req_data,
+            )?,
         ))
     }
 }
@@ -561,7 +573,7 @@ impl<B: SubAggBuffer> Debug for SegmentFilterCollector<B> {
         f.debug_struct("SegmentFilterCollector")
             .field("buckets", &self.parent_buckets)
             .field("has_sub_aggs", &self.sub_aggregations.is_some())
-            .field("accessor_idx", &self.accessor_idx)
+            .field("name", &self.req_data.name)
             .finish()
     }
 }
@@ -598,11 +610,7 @@ impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentFilterCollector<B>
         };
 
         // Get the name of this filter aggregation
-        let name = agg_data.per_request.filter_req_data[self.accessor_idx]
-            .as_ref()
-            .expect("filter_req_data slot is empty")
-            .name
-            .clone();
+        let name = self.req_data.name.clone();
 
         results.push(
             name,
@@ -623,27 +631,24 @@ impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentFilterCollector<B>
         }
 
         let mut bucket = self.parent_buckets[parent_bucket_id as usize];
-        // Take the request data to avoid borrow checker issues with sub-aggregations
-        let mut req = agg_data.take_filter_req_data(self.accessor_idx);
 
         // Use batch filtering with O(1) BitSet lookups
-        req.matching_docs_buffer.clear();
-        req.evaluator
-            .filter_batch(docs, &mut req.matching_docs_buffer);
+        self.matching_docs_buffer.clear();
+        self.req_data
+            .evaluator
+            .filter_batch(docs, &mut self.matching_docs_buffer);
 
-        bucket.doc_count += req.matching_docs_buffer.len() as u64;
+        bucket.doc_count += self.matching_docs_buffer.len() as u64;
 
         // Batch process sub-aggregations if we have matches
-        if !req.matching_docs_buffer.is_empty() {
+        if !self.matching_docs_buffer.is_empty() {
             if let Some(sub_aggs) = &mut self.sub_aggregations {
-                for &doc_id in &req.matching_docs_buffer {
+                for &doc_id in &self.matching_docs_buffer {
                     sub_aggs.push(bucket.bucket_id, doc_id);
                 }
             }
         }
 
-        // Put the request data back
-        agg_data.put_back_filter_req_data(self.accessor_idx, req);
         if let Some(sub_aggs) = &mut self.sub_aggregations {
             sub_aggs.check_flush_local(agg_data)?;
         }

src/aggregation/bucket/histogram/histogram.rs

@@ -21,6 +21,7 @@ use crate::TantivyError;
 
 /// Contains all information required by the SegmentHistogramCollector to perform the
 /// histogram or date_histogram aggregation on a segment.
+#[derive(Debug, Clone)]
 pub struct HistogramAggReqData {
     /// The column accessor to access the fast field values.
     pub accessor: Column<u64>,
@@ -243,19 +244,52 @@ impl Display for HistogramBounds {
 }
 
 impl HistogramBounds {
-    fn contains(&self, val: f64) -> bool {
+    pub(crate) fn contains(&self, val: f64) -> bool {
         val >= self.min && val <= self.max
     }
 }
 
-#[derive(Default, Clone, Debug, PartialEq)]
-pub(crate) struct SegmentHistogramBucketEntry {
-    pub key: f64,
-    pub doc_count: u64,
-    pub bucket_id: BucketId,
+/// The per-bucket identifier stored in a [`SegmentHistogramBucketEntry`].
+///
+/// It is [`BucketId`] when the histogram has sub aggregations (which key their state by it), and
+/// the zero-sized `()` when it does not. Without sub aggregations the id is never read, so storing
+/// `()` drops 8 bytes per bucket (24 -> 16) and turns id assignment into a no-op.
+pub trait BucketIdSlot: Copy + Default + std::fmt::Debug + PartialEq {
+    /// Assigns the next id from the provider, called once when a bucket is first filled.
+    fn assign(provider: &mut BucketIdProvider) -> Self;
+    /// Resolves to the `BucketId` for sub-aggregation bookkeeping.
+    ///
+    /// Only ever called for the [`BucketId`] slot: the `()` slot is used exactly when there are no
+    /// sub aggregations, so every call site is guarded by `sub_agg.is_some()` and is dead for `()`.
+    fn to_bucket_id(self) -> BucketId;
+}
+impl BucketIdSlot for BucketId {
+    #[inline(always)]
+    fn assign(provider: &mut BucketIdProvider) -> Self {
+        provider.next_bucket_id()
+    }
+    #[inline(always)]
+    fn to_bucket_id(self) -> BucketId {
+        self
+    }
+}
+impl BucketIdSlot for () {
+    #[inline(always)]
+    fn assign(_provider: &mut BucketIdProvider) -> Self {}
+    #[inline(always)]
+    fn to_bucket_id(self) -> BucketId {
+        unreachable!("bucket ids are only resolved when sub aggregations are present")
+    }
 }
 
-impl SegmentHistogramBucketEntry {
+#[derive(Default, Clone, Debug, PartialEq)]
+pub(crate) struct SegmentHistogramBucketEntry<B> {
+    pub key: f64,
+    pub doc_count: u64,
+    pub bucket_id: B,
+}
+
+impl<B: BucketIdSlot> SegmentHistogramBucketEntry<B> {
     pub(crate) fn into_intermediate_bucket_entry(
         self,
         sub_aggregation: &mut Option<HighCardBufferedSubAggs>,
@@ -268,7 +302,7 @@ impl SegmentHistogramBucketEntry {
                 .add_intermediate_aggregation_result(
                     agg_data,
                     &mut sub_aggregation_res,
-                    self.bucket_id,
+                    self.bucket_id.to_bucket_id(),
                 )?;
         }
         Ok(IntermediateHistogramBucketEntry {
@@ -279,39 +313,147 @@ impl SegmentHistogramBucketEntry {
     }
 }
 
-#[derive(Clone, Debug, Default)]
-struct HistogramBuckets {
-    pub buckets: FxHashMap<i64, SegmentHistogramBucketEntry>,
+/// The contiguous bucket range a histogram can span, derived from the column min/max (clamped to
+/// the histogram bounds). Buckets in `[base_pos, base_pos + len)` can be stored in a flat `Vec`
+/// indexed by `bucket_pos - base_pos`, avoiding the hash map on the hot path.
+#[derive(Clone, Copy, Debug)]
+pub(crate) struct DenseRange {
+    /// `bucket_pos` mapped to index 0 of the dense `Vec`.
+    pub(crate) base_pos: i64,
+    /// Number of bucket positions in the range.
+    pub(crate) len: usize,
 }
-impl HistogramBuckets {
+
+/// Storage for the histogram buckets of a single parent bucket.
+///
+/// Starts out sparse (a hash map keyed by `bucket_pos`). Once enough distinct buckets have been
+/// filled that we are clearly going to cover most of the column's theoretical range, it switches
+/// to a dense `Vec` indexed by `bucket_pos - base_pos`, which removes hashing from the hot loop.
+#[derive(Clone, Debug)]
+enum HistogramBuckets<B> {
+    Sparse(FxHashMap<i64, SegmentHistogramBucketEntry<B>>),
+    Dense {
+        base_pos: i64,
+        /// One slot per bucket position; a slot with `doc_count == 0` has not been hit yet.
+        buckets: Vec<SegmentHistogramBucketEntry<B>>,
+    },
+}
+impl<B> Default for HistogramBuckets<B> {
+    fn default() -> Self {
+        HistogramBuckets::Sparse(FxHashMap::default())
+    }
+}
+impl<B: BucketIdSlot> HistogramBuckets<B> {
     fn memory_consumption(&self) -> u64 {
-        self.buckets.capacity() as u64 * std::mem::size_of::<SegmentHistogramBucketEntry>() as u64
+        let num_slots = match self {
+            HistogramBuckets::Sparse(map) => map.capacity(),
+            HistogramBuckets::Dense { buckets, .. } => buckets.capacity(),
+        };
+        num_slots as u64 * std::mem::size_of::<SegmentHistogramBucketEntry<B>>() as u64
+    }
+
+    /// Switches from sparse to dense storage once the dense `Vec` would use no more memory than the
+    /// hash map does now, so the switch never increases memory. Called at block boundaries.
+    ///
+    /// The `Vec` holds one `Entry` per bucket position in the range. The map additionally stores
+    /// the key and a control byte per slot, at a load factor of 7/16..7/8, so for a dense histogram
+    /// its footprint grows past the `Vec` well before full coverage. And since the `Vec` never
+    /// grows afterwards while the map would keep growing, dense only gets relatively cheaper — so
+    /// no upper bound on the range is needed: a large but sparse range simply never crosses over.
+    #[inline]
+    fn maybe_densify(&mut self, dense_range: Option<DenseRange>) {
+        let Some(range) = dense_range else { return };
+        let HistogramBuckets::Sparse(map) = self else {
+            return;
+        };
+        let dense_bytes = range
+            .len
+            .saturating_mul(std::mem::size_of::<SegmentHistogramBucketEntry<B>>());
+        let sparse_bytes = map
+            .capacity()
+            .saturating_mul(std::mem::size_of::<(i64, SegmentHistogramBucketEntry<B>)>() + 1);
+        if dense_bytes > sparse_bytes {
+            return;
+        }
+        let map = std::mem::take(map);
+        let mut buckets = vec![SegmentHistogramBucketEntry::<B>::default(); range.len];
+        for (bucket_pos, entry) in map {
+            buckets[(bucket_pos - range.base_pos) as usize] = entry;
+        }
+        *self = HistogramBuckets::Dense {
+            base_pos: range.base_pos,
+            buckets,
+        };
+    }
+
+    /// Returns the bucket entry for `bucket_pos`, setting its key (and `bucket_id`, when `B` is
+    /// [`BucketId`]) on first use.
+    ///
+    /// For the dense variant `bucket_pos` is guaranteed to be inside the range, since it is
+    /// derived from the column min/max that bounds every value (see [`compute_dense_range`]).
+    #[inline]
+    fn get_or_create(
+        &mut self,
+        bucket_pos: i64,
+        bucket_id_provider: &mut BucketIdProvider,
+        key_from_pos: impl FnOnce(i64) -> f64,
+    ) -> &mut SegmentHistogramBucketEntry<B> {
+        match self {
+            HistogramBuckets::Sparse(map) => {
+                map.entry(bucket_pos)
+                    .or_insert_with(|| SegmentHistogramBucketEntry {
+                        key: key_from_pos(bucket_pos),
+                        doc_count: 0,
+                        bucket_id: B::assign(bucket_id_provider),
+                    })
+            }
+            HistogramBuckets::Dense { base_pos, buckets } => {
+                let idx = (bucket_pos - *base_pos) as usize;
+                debug_assert!(idx < buckets.len(), "bucket_pos outside the dense range");
+                let entry = &mut buckets[idx];
+                if entry.doc_count == 0 {
+                    entry.key = key_from_pos(bucket_pos);
+                    entry.bucket_id = B::assign(bucket_id_provider);
+                }
+                entry
+            }
+        }
+    }
+
+    /// Consumes the storage, yielding all non-empty bucket entries.
+    fn into_filled_entries(self) -> Vec<SegmentHistogramBucketEntry<B>> {
+        match self {
+            HistogramBuckets::Sparse(map) => map.into_values().collect(),
+            HistogramBuckets::Dense { buckets, .. } => {
+                buckets.into_iter().filter(|b| b.doc_count > 0).collect()
+            }
+        }
     }
 }
 
 /// The collector puts values from the fast field into the correct buckets and does a conversion to
 /// the correct datatype.
 #[derive(Debug)]
-pub struct SegmentHistogramCollector {
+pub struct SegmentHistogramCollector<B> {
     /// The buckets containing the aggregation data.
     /// One Histogram bucket per parent bucket id.
-    parent_buckets: Vec<HistogramBuckets>,
+    parent_buckets: Vec<HistogramBuckets<B>>,
     sub_agg: Option<HighCardBufferedSubAggs>,
-    accessor_idx: usize,
+    req_data: HistogramAggReqData,
     bucket_id_provider: BucketIdProvider,
+    /// Theoretical bucket range derived from the column min/max, if dense `Vec` storage is
+    /// viable. `None` keeps every parent bucket in the sparse hash map.
+    dense_range: Option<DenseRange>,
 }
 
-impl SegmentAggregationCollector for SegmentHistogramCollector {
+impl<B: BucketIdSlot> SegmentAggregationCollector for SegmentHistogramCollector<B> {
     fn add_intermediate_aggregation_result(
         &mut self,
         agg_data: &AggregationsSegmentCtx,
         results: &mut IntermediateAggregationResults,
         parent_bucket_id: BucketId,
     ) -> crate::Result<()> {
-        let name = agg_data
-            .get_histogram_req_data(self.accessor_idx)
-            .name
-            .clone();
+        let name = self.req_data.name.clone();
         // TODO: avoid prepare_max_bucket here and handle empty buckets.
         self.prepare_max_bucket(parent_bucket_id, agg_data)?;
         let histogram = std::mem::take(&mut self.parent_buckets[parent_bucket_id as usize]);
@@ -328,10 +470,13 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
         docs: &[crate::DocId],
         agg_data: &mut AggregationsSegmentCtx,
     ) -> crate::Result<()> {
-        let req = agg_data.take_histogram_req_data(self.accessor_idx);
         let mem_pre = self.get_memory_consumption(parent_bucket_id);
-        let buckets = &mut self.parent_buckets[parent_bucket_id as usize].buckets;
+        let dense_range = self.dense_range;
+        let store = &mut self.parent_buckets[parent_bucket_id as usize];
+        // Upgrade to dense storage before processing the block if the buckets are dense enough.
+        store.maybe_densify(dense_range);
 
+        let req = &self.req_data;
         let bounds = req.bounds;
         let interval = req.req.interval;
         let offset = req.offset;
@@ -340,31 +485,42 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
         agg_data
             .column_block_accessor
             .fetch_block(docs, &req.accessor);
-        for (doc, val) in agg_data
-            .column_block_accessor
-            .iter_docid_vals(docs, &req.accessor)
-        {
-            let val = f64_from_fastfield_u64(val, req.field_type);
-            let bucket_pos = get_bucket_pos(val);
-            if bounds.contains(val) {
-                let bucket = buckets.entry(bucket_pos).or_insert_with(|| {
-                    let key = get_bucket_key_from_pos(bucket_pos as f64, interval, offset);
-                    SegmentHistogramBucketEntry {
-                        key,
-                        doc_count: 0,
-                        bucket_id: self.bucket_id_provider.next_bucket_id(),
-                    }
-                });
-                bucket.doc_count += 1;
-                if let Some(sub_agg) = &mut self.sub_agg {
-                    sub_agg.push(bucket.bucket_id, doc);
+        // special path for nested buckets
+        if let Some(sub_agg) = &mut self.sub_agg {
+            for (doc, val) in agg_data
+                .column_block_accessor
+                .iter_docid_vals(docs, &req.accessor)
+            {
+                let val = f64_from_fastfield_u64(val, req.field_type);
+                if bounds.contains(val) {
+                    let bucket = store.get_or_create(
+                        get_bucket_pos(val),
+                        &mut self.bucket_id_provider,
+                        |pos| get_bucket_key_from_pos(pos as f64, interval, offset),
+                    );
+                    bucket.doc_count += 1;
+                    sub_agg.push(bucket.bucket_id.to_bucket_id(), doc);
+                }
+            }
+        } else {
+            for val in agg_data.column_block_accessor.iter_vals() {
+                let val = f64_from_fastfield_u64(val, req.field_type);
+                if bounds.contains(val) {
+                    let bucket = store.get_or_create(
+                        get_bucket_pos(val),
+                        &mut self.bucket_id_provider,
+                        |pos| get_bucket_key_from_pos(pos as f64, interval, offset),
+                    );
+                    bucket.doc_count += 1;
                 }
             }
         }
-        agg_data.put_back_histogram_req_data(self.accessor_idx, req);
 
-        let mem_delta = self.get_memory_consumption(parent_bucket_id) - mem_pre;
-        if mem_delta > 0 {
+        // `checked_sub` is `None` when densifying shrank the accounted memory; only account growth.
+        if let Some(mem_delta) = self
+            .get_memory_consumption(parent_bucket_id)
+            .checked_sub(mem_pre)
+        {
             agg_data.context.limits.add_memory_consumed(mem_delta)?;
         }
 
@@ -388,9 +544,7 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
         _agg_data: &AggregationsSegmentCtx,
     ) -> crate::Result<()> {
         while self.parent_buckets.len() <= max_bucket as usize {
-            self.parent_buckets.push(HistogramBuckets {
-                buckets: FxHashMap::default(),
-            });
+            self.parent_buckets.push(HistogramBuckets::default());
         }
         Ok(())
     }
@@ -407,7 +561,7 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
     }
 }
 
-impl SegmentHistogramCollector {
+impl<B: BucketIdSlot> SegmentHistogramCollector<B> {
     fn get_memory_consumption(&self, parent_bucket_id: BucketId) -> u64 {
         self.parent_buckets[parent_bucket_id as usize].memory_consumption()
     }
@@ -416,21 +570,19 @@ impl SegmentHistogramCollector {
     fn add_intermediate_bucket_result(
         &mut self,
         agg_data: &AggregationsSegmentCtx,
-        histogram: HistogramBuckets,
+        histogram: HistogramBuckets<B>,
     ) -> crate::Result<IntermediateBucketResult> {
-        let mut buckets = Vec::with_capacity(histogram.buckets.len());
+        let filled = histogram.into_filled_entries();
+        let mut buckets = Vec::with_capacity(filled.len());
 
-        for bucket in histogram.buckets.into_values() {
+        for bucket in filled {
             let bucket_res = bucket.into_intermediate_bucket_entry(&mut self.sub_agg, agg_data);
 
             buckets.push(bucket_res?);
         }
         buckets.sort_unstable_by(|b1, b2| b1.key.total_cmp(&b2.key));
 
-        let is_date_agg = agg_data
-            .get_histogram_req_data(self.accessor_idx)
-            .field_type
-            == ColumnType::DateTime;
+        let is_date_agg = self.req_data.field_type == ColumnType::DateTime;
         Ok(IntermediateBucketResult::Histogram {
             buckets,
             is_date_agg,
@@ -446,32 +598,175 @@ impl SegmentHistogramCollector {
         } else {
             None
         };
-        let req_data = agg_data.get_histogram_req_data_mut(node.idx_in_req_data);
-        req_data.req.validate()?;
-        if req_data.field_type == ColumnType::DateTime && !req_data.is_date_histogram {
-            req_data.req.normalize_date_time();
-        }
-        req_data.bounds = req_data.req.hard_bounds.unwrap_or(HistogramBounds {
-            min: f64::MIN,
-            max: f64::MAX,
-        });
-        req_data.offset = req_data.req.offset.unwrap_or(0.0);
+        let mut req_data = agg_data.per_request.histogram_req_data[node.idx_in_req_data].clone();
+        normalize_histogram_req(&mut req_data)?;
+        agg_data
+            .context
+            .limits
+            .add_memory_consumed(req_data.get_memory_consumption() as u64)?;
+        let dense_range = compute_dense_range(
+            &req_data.accessor,
+            req_data.field_type,
+            req_data.req.interval,
+            req_data.offset,
+            req_data.bounds,
+        );
         let sub_agg = sub_agg.map(BufferedSubAggs::new);
 
         Ok(Self {
             parent_buckets: Default::default(),
             sub_agg,
-            accessor_idx: node.idx_in_req_data,
+            req_data,
             bucket_id_provider: BucketIdProvider::default(),
+            dense_range,
         })
     }
 }
 
+impl SegmentHistogramCollector<()> {
+    /// Builds a histogram collector whose parent `t` is a dense histogram filled from
+    /// `counts[t * num_time_buckets .. (t + 1) * num_time_buckets]` (row-major). Used by the fused
+    /// terms×histogram collector to turn its flat 2D counters into the regular intermediate result,
+    /// so cross-segment merging is shared with the general path.
+    pub(crate) fn from_dense_rows(
+        req_data: HistogramAggReqData,
+        base_pos: i64,
+        num_time_buckets: usize,
+        counts: &[u32],
+    ) -> Self {
+        let interval = req_data.req.interval;
+        let offset = req_data.offset;
+        let num_parents = counts.len().checked_div(num_time_buckets).unwrap_or(0);
+        let parent_buckets = (0..num_parents)
+            .map(|t| {
+                let row = &counts[t * num_time_buckets..(t + 1) * num_time_buckets];
+                let buckets = row
+                    .iter()
+                    .enumerate()
+                    .map(|(b, &doc_count)| SegmentHistogramBucketEntry {
+                        key: get_bucket_key_from_pos(
+                            (base_pos + b as i64) as f64,
+                            interval,
+                            offset,
+                        ),
+                        doc_count: doc_count as u64,
+                        bucket_id: (),
+                    })
+                    .collect();
+                HistogramBuckets::Dense { base_pos, buckets }
+            })
+            .collect();
+        Self {
+            parent_buckets,
+            sub_agg: None,
+            req_data,
+            bucket_id_provider: BucketIdProvider::default(),
+            dense_range: None,
+        }
+    }
+}
+
+/// Validates and normalizes a histogram request in place: applies date ns-normalization (for a
+/// `histogram` on a date column) and resolves `bounds`/`offset` from the request.
+fn normalize_histogram_req(req_data: &mut HistogramAggReqData) -> crate::Result<()> {
+    req_data.req.validate()?;
+    if req_data.field_type == ColumnType::DateTime && !req_data.is_date_histogram {
+        req_data.req.normalize_date_time();
+    }
+    req_data.bounds = req_data.req.hard_bounds.unwrap_or(HistogramBounds {
+        min: f64::MIN,
+        max: f64::MAX,
+    });
+    req_data.offset = req_data.req.offset.unwrap_or(0.0);
+    // Drop `hard_bounds` that can't exclude any value (the column's range already sits inside
+    // them): the per-doc `bounds.contains` check is then a no-op, so collapsing to the unbounded
+    // sentinel lets the histogram hot loop skip it and the fused term×histogram path derive
+    // per-term counts from the grid. Only this collect-time filter is touched — empty-bucket
+    // emission reads `req.hard_bounds` directly (see `get_req_min_max`), and `hard_bounds` only
+    // ever clips that range, so a wider-than-data bound leaves the result unchanged.
+    if req_data.req.hard_bounds.is_some() {
+        let col_min = f64_from_fastfield_u64(req_data.accessor.min_value(), req_data.field_type);
+        let col_max = f64_from_fastfield_u64(req_data.accessor.max_value(), req_data.field_type);
+        if col_min >= req_data.bounds.min && col_max <= req_data.bounds.max {
+            req_data.bounds = HistogramBounds {
+                min: f64::MIN,
+                max: f64::MAX,
+            };
+        }
+    }
+    Ok(())
+}
+
+/// Clones and normalizes (resolving interval/offset/bounds) the histogram request at `node`, and
+/// returns it together with its dense bucket range — or `None` if the column has no usable range.
+/// Used by the fused terms×histogram collector, which then owns the normalized request.
+pub(crate) fn prepare_histogram_dense_range(
+    agg_data: &AggregationsSegmentCtx,
+    node: &AggRefNode,
+) -> crate::Result<Option<(HistogramAggReqData, DenseRange)>> {
+    let mut req_data = agg_data.per_request.histogram_req_data[node.idx_in_req_data].clone();
+    normalize_histogram_req(&mut req_data)?;
+    let dense_range = compute_dense_range(
+        &req_data.accessor,
+        req_data.field_type,
+        req_data.req.interval,
+        req_data.offset,
+        req_data.bounds,
+    );
+    Ok(dense_range.map(|range| (req_data, range)))
+}
+
+/// Builds a boxed histogram (or date histogram) segment collector, picking the bucket-id storage
+/// based on whether there are sub aggregations: `()` (no id stored) when there are none, otherwise
+/// [`BucketId`].
+pub(crate) fn build_segment_histogram_collector(
+    agg_data: &mut AggregationsSegmentCtx,
+    node: &AggRefNode,
+) -> crate::Result<Box<dyn SegmentAggregationCollector>> {
+    if node.children.is_empty() {
+        Ok(Box::new(
+            SegmentHistogramCollector::<()>::from_req_and_validate(agg_data, node)?,
+        ))
+    } else {
+        Ok(Box::new(
+            SegmentHistogramCollector::<BucketId>::from_req_and_validate(agg_data, node)?,
+        ))
+    }
+}
+
 #[inline]
-fn get_bucket_pos_f64(val: f64, interval: f64, offset: f64) -> f64 {
+pub(crate) fn get_bucket_pos_f64(val: f64, interval: f64, offset: f64) -> f64 {
     ((val - offset) / interval).floor()
 }
 
+/// Computes the dense bucket range for a column from its min/max value (clamped to the histogram
+/// bounds), or `None` if there are no values within bounds (or the range overflows `usize`).
+///
+/// There is no upper bound on the range: whether dense storage is actually used is decided later,
+/// per parent bucket, by [`HistogramBuckets::maybe_densify`] based on the memory it would save.
+///
+/// The column min/max bound every value the collector can see, so a `Vec` sized to this range can
+/// be indexed by `bucket_pos - base_pos` without any out-of-bounds check on the hot path.
+fn compute_dense_range(
+    accessor: &Column<u64>,
+    field_type: ColumnType,
+    interval: f64,
+    offset: f64,
+    bounds: HistogramBounds,
+) -> Option<DenseRange> {
+    let col_min = f64_from_fastfield_u64(accessor.min_value(), field_type);
+    let col_max = f64_from_fastfield_u64(accessor.max_value(), field_type);
+    let lo = col_min.max(bounds.min);
+    let hi = col_max.min(bounds.max);
+    if lo > hi {
+        return None;
+    }
+    let base_pos = get_bucket_pos_f64(lo, interval, offset) as i64;
+    let top_pos = get_bucket_pos_f64(hi, interval, offset) as i64;
+    let len = usize::try_from(top_pos.checked_sub(base_pos)?.checked_add(1)?).ok()?;
+    (len > 0).then_some(DenseRange { base_pos, len })
+}
+
 #[inline]
 fn get_bucket_key_from_pos(bucket_pos: f64, interval: f64, offset: f64) -> f64 {
     bucket_pos * interval + offset
@@ -776,6 +1071,62 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn histogram_dense_storage_test() -> crate::Result<()> {
+        histogram_dense_storage_test_with_opt(false)?;
+        histogram_dense_storage_test_with_opt(true)?;
+        Ok(())
+    }
+
+    /// Exercises the switch from sparse hash map to dense `Vec` storage. The switch happens at a
+    /// block boundary (a block is `COLLECT_BLOCK_BUFFER_LEN` = 64 docs), so we need many docs in a
+    /// single segment, densely covering the bucket range. `with_sub_agg` toggles the `iter_vals`
+    /// fast path vs. the `iter_docid_vals` path used when there is a sub aggregation.
+    fn histogram_dense_storage_test_with_opt(with_sub_agg: bool) -> crate::Result<()> {
+        let num_buckets = 50usize;
+        let docs_per_bucket = 10usize;
+        // Value `k` repeated `docs_per_bucket` times for each bucket `k`, so every value in bucket
+        // `k` equals `k` and the per-bucket average is exactly `k`.
+        let values: Vec<f64> = (0..num_buckets * docs_per_bucket)
+            .map(|i| (i % num_buckets) as f64)
+            .collect();
+        // `merge_segments = true` collapses the per-value segments into a single segment with all
+        // the docs, which is collected in 64-doc blocks and therefore switches to dense storage.
+        let index = get_test_index_from_values(true, &values)?;
+
+        let agg_req: Aggregations = serde_json::from_value(if with_sub_agg {
+            json!({
+                "histogram": {
+                    "histogram": { "field": "score_f64", "interval": 1.0 },
+                    "aggs": { "avg": { "avg": { "field": "score_f64" } } }
+                }
+            })
+        } else {
+            json!({
+                "histogram": {
+                    "histogram": { "field": "score_f64", "interval": 1.0 }
+                }
+            })
+        })
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+
+        for k in 0..num_buckets {
+            assert_eq!(res["histogram"]["buckets"][k]["key"], k as f64);
+            assert_eq!(
+                res["histogram"]["buckets"][k]["doc_count"],
+                docs_per_bucket as u64
+            );
+            if with_sub_agg {
+                assert_eq!(res["histogram"]["buckets"][k]["avg"]["value"], k as f64);
+            }
+        }
+        assert_eq!(res["histogram"]["buckets"][num_buckets], Value::Null);
+
+        Ok(())
+    }
+
     #[test]
     fn histogram_memory_limit() -> crate::Result<()> {
         let index = get_test_index_with_num_docs(true, 100)?;
@@ -1070,6 +1421,55 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn histogram_non_binding_hard_bounds_test_multi_segment() -> crate::Result<()> {
+        histogram_non_binding_hard_bounds_test_with_opt(false)
+    }
+    #[test]
+    fn histogram_non_binding_hard_bounds_test_single_segment() -> crate::Result<()> {
+        histogram_non_binding_hard_bounds_test_with_opt(true)
+    }
+    /// `hard_bounds` wider than the data (here with mid-interval edges, to cover the "bound cuts a
+    /// bucket" case) can't exclude any value, so the result must be identical to the same request
+    /// without bounds. Guards the normalization that collapses such bounds to the unbounded
+    /// sentinel so the hot loop / fused path can skip the per-doc bounds check.
+    fn histogram_non_binding_hard_bounds_test_with_opt(merge_segments: bool) -> crate::Result<()> {
+        let values = vec![10.0, 12.0, 14.0, 16.0, 10.0, 13.0, 10.0, 12.0];
+        let index = get_test_index_from_values(merge_segments, &values)?;
+
+        // Mid-interval edges, but wider than the data range [10, 16] -> they exclude nothing.
+        let with_bounds: Aggregations = serde_json::from_value(json!({
+            "histogram": {
+                "histogram": {
+                    "field": "score_f64",
+                    "interval": 1.0,
+                    "hard_bounds": { "min": 9.5, "max": 16.5 }
+                }
+            }
+        }))
+        .unwrap();
+        let no_bounds: Aggregations = serde_json::from_value(json!({
+            "histogram": {
+                "histogram": { "field": "score_f64", "interval": 1.0 }
+            }
+        }))
+        .unwrap();
+
+        let res_bounds = exec_request(with_bounds, &index)?;
+        let res_plain = exec_request(no_bounds, &index)?;
+        // Dropping a non-binding bound must not change anything.
+        assert_eq!(res_bounds, res_plain);
+
+        // Sanity: buckets span the data range with gaps filled (min_doc_count defaults to 0).
+        assert_eq!(res_bounds["histogram"]["buckets"][0]["key"], 10.0);
+        assert_eq!(res_bounds["histogram"]["buckets"][0]["doc_count"], 3);
+        assert_eq!(res_bounds["histogram"]["buckets"][6]["key"], 16.0);
+        assert_eq!(res_bounds["histogram"]["buckets"][6]["doc_count"], 1);
+        assert_eq!(res_bounds["histogram"]["buckets"][7], Value::Null);
+
+        Ok(())
+    }
+
     #[test]
     fn histogram_empty_result_behaviour_test_single_segment() -> crate::Result<()> {
         histogram_empty_result_behaviour_test_with_opt(true)

src/aggregation/bucket/range.rs

@@ -23,6 +23,7 @@ use crate::TantivyError;
 
 /// Contains all information required by the SegmentRangeCollector to perform the
 /// range aggregation on a segment.
+#[derive(Debug, Clone)]
 pub struct RangeAggReqData {
     /// The column accessor to access the fast field values.
     pub accessor: Column<u64>,
@@ -161,7 +162,7 @@ pub struct SegmentRangeCollector<B: SubAggBuffer> {
     /// One for each ParentBucketId
     parent_buckets: Vec<Vec<SegmentRangeAndBucketEntry>>,
     column_type: ColumnType,
-    pub(crate) accessor_idx: usize,
+    pub(crate) req_data: RangeAggReqData,
     sub_agg: Option<BufferedSubAggs<B>>,
     /// Here things get a bit weird. We need to assign unique bucket ids across all
     /// parent buckets. So we keep track of the next available bucket id here.
@@ -184,7 +185,7 @@ impl<B: SubAggBuffer> Debug for SegmentRangeCollector<B> {
         f.debug_struct("SegmentRangeCollector")
             .field("parent_buckets_len", &self.parent_buckets.len())
             .field("column_type", &self.column_type)
-            .field("accessor_idx", &self.accessor_idx)
+            .field("name", &self.req_data.name)
             .field("has_sub_agg", &self.sub_agg.is_some())
             .finish()
     }
@@ -239,10 +240,7 @@ impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentRangeCollector<B> {
     ) -> crate::Result<()> {
         self.prepare_max_bucket(parent_bucket_id, agg_data)?;
         let field_type = self.column_type;
-        let name = agg_data
-            .get_range_req_data(self.accessor_idx)
-            .name
-            .to_string();
+        let name = self.req_data.name.to_string();
 
         let buckets = std::mem::take(&mut self.parent_buckets[parent_bucket_id as usize]);
 
@@ -281,17 +279,15 @@ impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentRangeCollector<B> {
         docs: &[crate::DocId],
         agg_data: &mut AggregationsSegmentCtx,
     ) -> crate::Result<()> {
-        let req = agg_data.take_range_req_data(self.accessor_idx);
-
         agg_data
             .column_block_accessor
-            .fetch_block(docs, &req.accessor);
+            .fetch_block(docs, &self.req_data.accessor);
 
         let buckets = &mut self.parent_buckets[parent_bucket_id as usize];
 
         for (doc, val) in agg_data
             .column_block_accessor
-            .iter_docid_vals(docs, &req.accessor)
+            .iter_docid_vals(docs, &self.req_data.accessor)
         {
             let bucket_pos = get_bucket_pos(val, buckets);
             let bucket = &mut buckets[bucket_pos];
@@ -301,7 +297,6 @@ impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentRangeCollector<B> {
             }
         }
 
-        agg_data.put_back_range_req_data(self.accessor_idx, req);
         if let Some(sub_agg) = self.sub_agg.as_mut() {
             sub_agg.check_flush_local(agg_data)?;
         }
@@ -319,10 +314,10 @@ impl<B: SubAggBuffer> SegmentAggregationCollector for SegmentRangeCollector<B> {
     fn prepare_max_bucket(
         &mut self,
         max_bucket: BucketId,
-        agg_data: &AggregationsSegmentCtx,
+        _agg_data: &AggregationsSegmentCtx,
     ) -> crate::Result<()> {
         while self.parent_buckets.len() <= max_bucket as usize {
-            let new_buckets = self.create_new_buckets(agg_data)?;
+            let new_buckets = self.create_new_buckets()?;
             self.parent_buckets.push(new_buckets);
         }
 
@@ -346,8 +341,11 @@ pub(crate) fn build_segment_range_collector(
     agg_data: &mut AggregationsSegmentCtx,
     node: &AggRefNode,
 ) -> crate::Result<Box<dyn SegmentAggregationCollector>> {
-    let accessor_idx = node.idx_in_req_data;
-    let req_data = agg_data.get_range_req_data(node.idx_in_req_data);
+    let req_data = agg_data.per_request.range_req_data[node.idx_in_req_data].clone();
+    agg_data
+        .context
+        .limits
+        .add_memory_consumed(req_data.get_memory_consumption() as u64)?;
     let field_type = req_data.field_type;
 
     // TODO: A better metric instead of is_top_level would be the number of buckets expected.
@@ -365,7 +363,7 @@ pub(crate) fn build_segment_range_collector(
         Ok(Box::new(SegmentRangeCollector::<LowCardSubAggBuffer> {
             sub_agg: sub_agg.map(LowCardBufferedSubAggs::new),
             column_type: field_type,
-            accessor_idx,
+            req_data,
             parent_buckets: Vec::new(),
             bucket_id_provider: BucketIdProvider::default(),
             limits: agg_data.context.limits.clone(),
@@ -374,7 +372,7 @@ pub(crate) fn build_segment_range_collector(
         Ok(Box::new(SegmentRangeCollector::<HighCardSubAggBuffer> {
             sub_agg: sub_agg.map(BufferedSubAggs::new),
             column_type: field_type,
-            accessor_idx,
+            req_data,
             parent_buckets: Vec::new(),
             bucket_id_provider: BucketIdProvider::default(),
             limits: agg_data.context.limits.clone(),
@@ -383,12 +381,9 @@ pub(crate) fn build_segment_range_collector(
 }
 
 impl<B: SubAggBuffer> SegmentRangeCollector<B> {
-    pub(crate) fn create_new_buckets(
-        &mut self,
-        agg_data: &AggregationsSegmentCtx,
-    ) -> crate::Result<Vec<SegmentRangeAndBucketEntry>> {
+    pub(crate) fn create_new_buckets(&mut self) -> crate::Result<Vec<SegmentRangeAndBucketEntry>> {
         let field_type = self.column_type;
-        let req_data = agg_data.get_range_req_data(self.accessor_idx);
+        let req_data = &self.req_data;
         // The range input on the request is f64.
         // We need to convert to u64 ranges, because we read the values as u64.
         // The mapping from the conversion is monotonic so ordering is preserved.
@@ -563,17 +558,16 @@ mod tests {
         get_test_index_with_num_docs,
     };
 
-    pub fn get_collector_from_ranges(
-        ranges: Vec<RangeAggregationRange>,
+    pub fn build_test_buckets(
+        ranges: &[RangeAggregationRange],
         field_type: ColumnType,
-    ) -> SegmentRangeCollector<HighCardSubAggBuffer> {
+    ) -> Vec<SegmentRangeAndBucketEntry> {
         let req = RangeAggregation {
             field: "dummy".to_string(),
-            ranges,
+            ranges: ranges.to_vec(),
             ..Default::default()
         };
-        // Build buckets directly as in from_req_and_validate without AggregationsData
-        let buckets: Vec<_> = extend_validate_ranges(&req.ranges, &field_type)
+        extend_validate_ranges(&req.ranges, &field_type)
             .expect("unexpected error in extend_validate_ranges")
             .iter()
             .map(|range| {
@@ -604,16 +598,7 @@ mod tests {
                     },
                 }
             })
-            .collect();
-
-        SegmentRangeCollector {
-            parent_buckets: vec![buckets],
-            column_type: field_type,
-            accessor_idx: 0,
-            sub_agg: None,
-            bucket_id_provider: Default::default(),
-            limits: AggregationLimitsGuard::default(),
-        }
+            .collect()
     }
 
     #[test]
@@ -856,10 +841,10 @@ mod tests {
 
     #[test]
     fn bucket_test_extend_range_hole() {
-        let buckets = vec![(10f64..20f64).into(), (30f64..40f64).into()];
-        let collector = get_collector_from_ranges(buckets, ColumnType::F64);
+        let buckets = [(10f64..20f64).into(), (30f64..40f64).into()];
+        let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
 
-        let buckets = collector.parent_buckets[0].clone();
+        let buckets = parent_buckets[0].clone();
         assert_eq!(buckets[0].range.start, u64::MIN);
         assert_eq!(buckets[0].range.end, 10f64.to_u64());
         assert_eq!(buckets[1].range.start, 10f64.to_u64());
@@ -875,14 +860,14 @@ mod tests {
     fn bucket_test_range_conversion_special_case() {
         // the monotonic conversion between f64 and u64, does not map f64::MIN.to_u64() ==
         // u64::MIN, but the into trait converts f64::MIN/MAX to None
-        let buckets = vec![
+        let buckets = [
             (f64::MIN..10f64).into(),
             (10f64..20f64).into(),
             (20f64..f64::MAX).into(),
         ];
-        let collector = get_collector_from_ranges(buckets, ColumnType::F64);
+        let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
 
-        let buckets = collector.parent_buckets[0].clone();
+        let buckets = parent_buckets[0].clone();
         assert_eq!(buckets[0].range.start, u64::MIN);
         assert_eq!(buckets[0].range.end, 10f64.to_u64());
         assert_eq!(buckets[1].range.start, 10f64.to_u64());
@@ -894,28 +879,28 @@ mod tests {
 
     #[test]
     fn bucket_range_test_negative_vals() {
-        let buckets = vec![(-10f64..-1f64).into()];
-        let collector = get_collector_from_ranges(buckets, ColumnType::F64);
+        let buckets = [(-10f64..-1f64).into()];
+        let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
 
-        let buckets = collector.parent_buckets[0].clone();
+        let buckets = parent_buckets[0].clone();
         assert_eq!(&buckets[0].bucket.key.to_string(), "*--10");
         assert_eq!(&buckets[buckets.len() - 1].bucket.key.to_string(), "-1-*");
     }
     #[test]
     fn bucket_range_test_positive_vals() {
-        let buckets = vec![(0f64..10f64).into()];
-        let collector = get_collector_from_ranges(buckets, ColumnType::F64);
+        let buckets = [(0f64..10f64).into()];
+        let parent_buckets = [build_test_buckets(&buckets, ColumnType::F64)];
 
-        let buckets = collector.parent_buckets[0].clone();
+        let buckets = parent_buckets[0].clone();
         assert_eq!(&buckets[0].bucket.key.to_string(), "*-0");
         assert_eq!(&buckets[buckets.len() - 1].bucket.key.to_string(), "10-*");
     }
 
     #[test]
     fn range_binary_search_test_u64() {
-        let check_ranges = |ranges: Vec<RangeAggregationRange>| {
-            let collector = get_collector_from_ranges(ranges, ColumnType::U64);
-            let search = |val: u64| get_bucket_pos(val, &collector.parent_buckets[0]);
+        let check_ranges = |ranges: &[RangeAggregationRange]| {
+            let parent_buckets = [build_test_buckets(ranges, ColumnType::U64)];
+            let search = |val: u64| get_bucket_pos(val, &parent_buckets[0]);
 
             assert_eq!(search(u64::MIN), 0);
             assert_eq!(search(9), 0);
@@ -928,7 +913,7 @@ mod tests {
         };
 
         let ranges = vec![(10.0..100.0).into()];
-        check_ranges(ranges);
+        check_ranges(&ranges);
 
         let ranges = vec![
             RangeAggregationRange {
@@ -938,7 +923,7 @@ mod tests {
             },
             (10.0..100.0).into(),
         ];
-        check_ranges(ranges);
+        check_ranges(&ranges);
 
         let ranges = vec![
             RangeAggregationRange {
@@ -953,15 +938,15 @@ mod tests {
                 from: Some(100.0),
             },
         ];
-        check_ranges(ranges);
+        check_ranges(&ranges);
     }
 
     #[test]
     fn range_binary_search_test_f64() {
-        let ranges = vec![(10.0..100.0).into()];
+        let ranges = [(10.0..100.0).into()];
 
-        let collector = get_collector_from_ranges(ranges, ColumnType::F64);
-        let search = |val: u64| get_bucket_pos(val, &collector.parent_buckets[0]);
+        let parent_buckets = [build_test_buckets(&ranges, ColumnType::F64)];
+        let search = |val: u64| get_bucket_pos(val, &parent_buckets[0]);
 
         assert_eq!(search(u64::MIN), 0);
         assert_eq!(search(9f64.to_u64()), 0);

src/aggregation/bucket/term_agg/mod.rs

@@ -29,6 +29,8 @@ use crate::aggregation::{format_date, BucketId, Key};
 use crate::error::DataCorruption;
 use crate::TantivyError;
 
+mod term_histogram;
+
 /// Contains all information required by the SegmentTermCollector to perform the
 /// terms aggregation on a segment.
 #[derive(Debug, Clone)]
@@ -374,9 +376,21 @@ pub(crate) fn build_segment_term_collector(
     // Let's see if we can use a vec to aggregate our data
     // instead of a hashmap.
     let col_max_value = terms_req_data.accessor.max_value();
-    let max_term_id: u64 =
+    let max_column_val: u64 =
         col_max_value.max(terms_req_data.missing_value_for_accessor.unwrap_or(0u64));
 
+    // Fused fast path: low-cardinality terms × a single `histogram`/`date_histogram` leaf over full
+    // columns with a small enough bucket grid. Anything else falls through to the general path.
+    if let Some(collector) = term_histogram::maybe_build_collector(
+        req_data,
+        node,
+        &terms_req_data,
+        max_column_val,
+        is_top_level,
+    )? {
+        return Ok(collector);
+    }
+
     let sub_agg_collector = if has_sub_aggregations {
         Some(build_segment_agg_collectors(req_data, &node.children)?)
     } else {
@@ -385,30 +399,30 @@ pub(crate) fn build_segment_term_collector(
 
     let mut bucket_id_provider = BucketIdProvider::default();
     // Decide which bucket storage is best suited for this aggregation.
-    if is_top_level && max_term_id < MAX_NUM_TERMS_FOR_VEC && !has_sub_aggregations {
-        let term_buckets = VecTermBucketsNoAgg::new(max_term_id + 1, &mut bucket_id_provider);
+    if is_top_level && max_column_val < MAX_NUM_TERMS_FOR_VEC && !has_sub_aggregations {
+        let term_buckets = VecTermBucketsNoAgg::new(max_column_val + 1, &mut bucket_id_provider);
         let collector: SegmentTermCollector<_, HighCardSubAggBuffer> = SegmentTermCollector {
             parent_buckets: vec![term_buckets],
             sub_agg: None,
             bucket_id_provider,
-            max_term_id,
+            max_term_id: max_column_val,
             terms_req_data,
         };
         Ok(Box::new(collector))
-    } else if is_top_level && max_term_id < MAX_NUM_TERMS_FOR_VEC {
-        let term_buckets = VecTermBuckets::new(max_term_id + 1, &mut bucket_id_provider);
+    } else if is_top_level && max_column_val < MAX_NUM_TERMS_FOR_VEC {
+        let term_buckets = VecTermBuckets::new(max_column_val + 1, &mut bucket_id_provider);
         let sub_agg = sub_agg_collector.map(LowCardBufferedSubAggs::new);
         let collector: SegmentTermCollector<_, LowCardSubAggBuffer> = SegmentTermCollector {
             parent_buckets: vec![term_buckets],
             sub_agg,
             bucket_id_provider,
-            max_term_id,
+            max_term_id: max_column_val,
             terms_req_data,
         };
         Ok(Box::new(collector))
-    } else if max_term_id < 8_000_000 && is_top_level {
+    } else if max_column_val < 8_000_000 && is_top_level {
         let term_buckets: PagedTermMap =
-            PagedTermMap::new(max_term_id + 1, &mut bucket_id_provider);
+            PagedTermMap::new(max_column_val + 1, &mut bucket_id_provider);
         // Build sub-aggregation blueprint (flat pairs)
         let sub_agg = sub_agg_collector.map(BufferedSubAggs::new);
         let collector: SegmentTermCollector<PagedTermMap, HighCardSubAggBuffer> =
@@ -416,7 +430,7 @@ pub(crate) fn build_segment_term_collector(
                 parent_buckets: vec![term_buckets],
                 sub_agg,
                 bucket_id_provider,
-                max_term_id,
+                max_term_id: max_column_val,
                 terms_req_data,
             };
         Ok(Box::new(collector))
@@ -429,7 +443,7 @@ pub(crate) fn build_segment_term_collector(
                 parent_buckets: vec![term_buckets],
                 sub_agg,
                 bucket_id_provider,
-                max_term_id,
+                max_term_id: max_column_val,
                 terms_req_data,
             };
         Ok(Box::new(collector))

src/aggregation/bucket/term_agg/term_histogram.rs

@@ -0,0 +1,585 @@
+//! Fused collector for the very common shape `terms` (low cardinality) × a single
+//! `histogram`/`date_histogram` sub-aggregation with nothing nested below it.
+//!
+//! See [`SegmentTermHistogramCollector`] for the approach and [`maybe_build_collector`] for the
+//! conditions under which it is used.
+
+use columnar::ColumnBlockAccessor;
+
+use super::{Bucket, SegmentTermCollector, TermsAggReqData, VecTermBuckets};
+use crate::aggregation::agg_data::{AggKind, AggRefNode, AggregationsSegmentCtx};
+use crate::aggregation::bucket::{
+    get_bucket_pos_f64, prepare_histogram_dense_range, HistogramAggReqData,
+    SegmentHistogramCollector,
+};
+use crate::aggregation::buffered_sub_aggs::LowCardSubAggBuffer;
+use crate::aggregation::intermediate_agg_result::{
+    IntermediateAggregationResult, IntermediateAggregationResults,
+};
+use crate::aggregation::segment_agg_result::{BucketIdProvider, SegmentAggregationCollector};
+use crate::aggregation::{f64_from_fastfield_u64, BucketId};
+
+/// Maximum number of cells (`num_terms × num_time_buckets`) in the fused flat 2D grid. Above this
+/// the grid would be too large/cache-unfriendly, so we fall back to the general buffered path.
+/// `1 << 14` cells = 128 KB of `u64` counters, comfortably L2-resident.
+///
+/// Since we are only at the top-level, this won't be multiplied by any parent buckets.
+const MAX_FUSED_GRID_BUCKETS: usize = 16384;
+
+/// Fused collector for `terms` (low cardinality) × a single `histogram`/`date_histogram` leaf with
+/// nothing nested below it, when the resulting `num_terms × num_time_buckets` grid is small (see
+/// [`MAX_FUSED_GRID_BUCKETS`]).
+///
+/// It keeps a flat, fully dense 2D counter grid (`counts[term * num_time_buckets + bucket]`) and a
+/// per-term total. A single pass reads both the term and histogram columns in document order and
+/// bumps the counters directly — no doc-id buffering, no per-term scattered re-fetch, no dynamic
+/// dispatch on flush, no per-bucket key/id storage during collection (keys are derived from the
+/// index at the end).
+///
+/// At result time the flat grid is expanded back into the regular term map + histogram storage and
+/// handed to the shared intermediate-result builders, so cross-segment merging is identical to the
+/// general path.
+#[derive(Debug)]
+pub(crate) struct SegmentTermHistogramCollector {
+    /// Per-term count of docs *outside* `hard_bounds` (still in `doc_count`, but in no bucket).
+    /// Per-term total = this + the term's `counts` row-sum; left empty when there are no hard
+    /// bounds (every doc is in-bounds, so there's no remainder to track).
+    term_counts: Vec<u32>,
+    /// Flattened `[num_terms * num_time_buckets]` histogram counters (`u32`, see
+    /// `term_counts`).
+    ///
+    /// Each term id get its own contiguous slice of `num_time_buckets` histogram counter.
+    /// When we count all docs (#nofilter), we can derive the per-term total as the sum over that
+    /// term's slice.
+    counts: Vec<u32>,
+    /// Histogram buckets per term (the dense time-range length).
+    num_time_buckets: usize,
+    /// `bucket_pos` mapped to time-bucket index 0.
+    base_pos: i64,
+    terms_req_data: TermsAggReqData,
+    /// The (cloned, normalized) histogram request: its column + interval/offset/bounds.
+    hist_req_data: HistogramAggReqData,
+    /// Private block accessors for both columns. We read them together, so each needs its own
+    /// (the shared `agg_data` scratch accessor only holds one block at a time). Owning them keeps
+    /// `collect` independent of `agg_data`.
+    term_block: ColumnBlockAccessor<u64>,
+    hist_block: ColumnBlockAccessor<u64>,
+    /// No hard bounds, so every doc is in-bounds.
+    all_docs_in_bounds: bool,
+    /// Both columns are full (fused-path precondition); cached so `collect` skips the per-block
+    /// cardinality lookup in `fetch_block`.
+    is_full: bool,
+}
+
+impl SegmentAggregationCollector for SegmentTermHistogramCollector {
+    fn add_intermediate_aggregation_result(
+        &mut self,
+        agg_data: &AggregationsSegmentCtx,
+        results: &mut IntermediateAggregationResults,
+        parent_bucket_id: BucketId,
+    ) -> crate::Result<()> {
+        debug_assert_eq!(
+            parent_bucket_id, 0,
+            "fused term-histogram collector is top-level only"
+        );
+        // Expand the flat grid back into the regular structures and reuse the shared builders, so
+        // ordering/cut-off/dict handling and cross-segment merging match the general path exactly.
+        let mut bucket_id_provider = BucketIdProvider::default();
+        // Per-term total = histogram row-sum (in-bounds) + `term_counts` (out-of-bounds remainder,
+        // empty when there are no hard bounds).
+        let term_buckets = VecTermBuckets {
+            buckets: self
+                .counts
+                .chunks_exact(self.num_time_buckets)
+                .enumerate()
+                .map(|(term_id, row)| {
+                    let in_bounds: u32 = row.iter().sum();
+                    let out_of_bounds = self.term_counts.get(term_id).copied().unwrap_or(0);
+                    Bucket {
+                        count: in_bounds + out_of_bounds,
+                        bucket_id: bucket_id_provider.next_bucket_id(),
+                    }
+                })
+                .collect(),
+        };
+        let mut histogram = SegmentHistogramCollector::<()>::from_dense_rows(
+            self.hist_req_data.clone(),
+            self.base_pos,
+            self.num_time_buckets,
+            &self.counts,
+        );
+        let name = self.terms_req_data.name.clone();
+        let bucket = SegmentTermCollector::<VecTermBuckets, LowCardSubAggBuffer>::into_intermediate_bucket_result(
+            &self.terms_req_data,
+            Some(&mut histogram as &mut dyn SegmentAggregationCollector),
+            term_buckets,
+            agg_data,
+        )?;
+        results.push(name, IntermediateAggregationResult::Bucket(bucket))?;
+        Ok(())
+    }
+
+    #[inline]
+    fn collect(
+        &mut self,
+        parent_bucket_id: BucketId,
+        docs: &[crate::DocId],
+        _agg_data: &mut AggregationsSegmentCtx,
+    ) -> crate::Result<()> {
+        debug_assert_eq!(
+            parent_bucket_id, 0,
+            "fused term-histogram collector is top-level only"
+        );
+
+        // Fetch both columns into our own accessors (we read them together, so they can't share the
+        // single `agg_data` scratch accessor). The collector owns all its inputs, so `collect`
+        // doesn't touch `agg_data`.
+        self.term_block
+            .fetch_block_with_is_full(docs, &self.terms_req_data.accessor, self.is_full);
+        self.hist_block
+            .fetch_block_with_is_full(docs, &self.hist_req_data.accessor, self.is_full);
+
+        // Hoist the loop-invariant fields into locals: the optimizer can't prove the
+        // `self.counts`/`self.term_counts` writes don't alias these `self` fields, so it can't keep
+        // them in registers and re-reads them from memory every iteration — ~15% slower on
+        // `terms_status_with_date_histogram` when read straight from `self`.
+        // Note: check which are actually relevant.
+        let field_type = self.hist_req_data.field_type;
+        let bounds = self.hist_req_data.bounds;
+        let interval = self.hist_req_data.req.interval;
+        let offset = self.hist_req_data.offset;
+        let base_pos = self.base_pos;
+        let num_time_buckets = self.num_time_buckets;
+        let all_docs_in_bounds = self.all_docs_in_bounds;
+        let term_counts = &mut self.term_counts;
+        let counts = &mut self.counts;
+
+        // Both columns are full (checked at construction), so values align with `docs` positionally
+        // and are read together in one pass.
+        // In-bounds docs bump the `counts` grid, out-of-bounds bump `term_counts`; deriving the
+        // total at flush avoids a per-doc `term_counts` RMW that serializes on
+        // store-to-load forwarding.
+        for (term_id, hist_raw) in self.term_block.iter_vals().zip(self.hist_block.iter_vals()) {
+            let term_id = term_id as usize;
+            let val = f64_from_fastfield_u64(hist_raw, field_type);
+            if all_docs_in_bounds || bounds.contains(val) {
+                let bucket = (get_bucket_pos_f64(val, interval, offset) as i64 - base_pos) as usize;
+                debug_assert!(
+                    bucket < num_time_buckets,
+                    "histogram bucket outside dense range"
+                );
+                counts[term_id * num_time_buckets + bucket] += 1;
+            } else {
+                term_counts[term_id] += 1;
+            }
+        }
+        Ok(())
+    }
+
+    fn flush(&mut self, _agg_data: &mut AggregationsSegmentCtx) -> crate::Result<()> {
+        // Nothing is buffered: `collect` writes the flat grid directly.
+        Ok(())
+    }
+
+    fn prepare_max_bucket(
+        &mut self,
+        _max_bucket: BucketId,
+        _agg_data: &AggregationsSegmentCtx,
+    ) -> crate::Result<()> {
+        // Top-level: the flat grid is allocated up front.
+        Ok(())
+    }
+
+    fn compute_metric_value(
+        &self,
+        _bucket_id: BucketId,
+        _sub_agg_name: &str,
+        _sub_agg_property: &str,
+        _agg_data: &AggregationsSegmentCtx,
+    ) -> Option<f64> {
+        None
+    }
+}
+
+/// Builds the fused terms×histogram collector for a single top-level parent, when the shape is
+/// eligible. Returns `Ok(None)` to fall back to the general buffered terms path.
+///
+/// Eligibility: top-level, low-cardinality terms over a full column with no missing/include-exclude
+/// handling; a single `histogram`/`date_histogram` leaf (no nesting below it) over a full column;
+/// and a `num_terms × num_time_buckets` grid no larger than [`MAX_FUSED_GRID_BUCKETS`].
+pub(super) fn maybe_build_collector(
+    agg_data: &mut AggregationsSegmentCtx,
+    node: &AggRefNode,
+    terms_req_data: &TermsAggReqData,
+    col_max_val: u64,
+    is_top_level: bool,
+) -> crate::Result<Option<Box<dyn SegmentAggregationCollector>>> {
+    // Both columns must be full (one value per doc) so their values align positionally with `docs`
+    // and we can zip them. Requiring full columns also makes the terms agg's `missing` config a
+    // no-op (`fetch_block_with_missing` early-returns on full columns), so we needn't check for it.
+    //
+    // We don't cap the term cardinality here: the flat grid is bounded by the total cell count
+    // (`num_terms * num_time_buckets <= MAX_FUSED_GRID_BUCKETS`) checked below, which subsumes it.
+    //
+    // We only allow this at the top-level, since we don't know how many buckets are created. We
+    // are less likely to get enough docs for the preallocation to be worth and there's a risk of
+    // using too much memory. We could check the maximum theoretical buckets up-front and pass
+    // them down.
+    let fuseable = is_top_level
+        // TODO: We can easily support this
+        && terms_req_data.allowed_term_ids.is_none()
+        && terms_req_data.accessor.get_cardinality().is_full()
+        // The flat counters are `u32`, bumped once per value, so no count can exceed the column's
+        // value count. (Essentially always true here: the column is full, so its value count
+        // equals the doc count, and `DocId` is `u32`.)
+        && terms_req_data.accessor.values.num_vals() < u32::MAX
+        && node.children.len() == 1
+        && matches!(
+            node.children[0].kind,
+            AggKind::Histogram | AggKind::DateHistogram
+        )
+        && node.children[0].children.is_empty()
+        && agg_data.per_request.histogram_req_data[node.children[0].idx_in_req_data]
+            .accessor
+            .get_cardinality()
+            .is_full();
+    if !fuseable {
+        return Ok(None);
+    }
+
+    // Clone + normalize the histogram request and get its dense bucket range; only take the fused
+    // path when the flat `num_terms × num_time_buckets` grid is small enough.
+    let Some((hist_req_data, range)) = prepare_histogram_dense_range(agg_data, &node.children[0])?
+    else {
+        return Ok(None);
+    };
+    let num_terms = col_max_val.saturating_add(1) as usize;
+    if num_terms.saturating_mul(range.len) > MAX_FUSED_GRID_BUCKETS {
+        return Ok(None);
+    }
+
+    // No hard bounds means every doc is in-bounds, letting `collect` short-circuit the bounds
+    // check — and leaving `term_counts` (the out-of-bounds remainder) unused, so we skip allocating
+    // it.
+    let all_docs_in_bounds =
+        hist_req_data.bounds.min == f64::MIN && hist_req_data.bounds.max == f64::MAX;
+    let counts = vec![0u32; num_terms * range.len];
+    let term_counts = if all_docs_in_bounds {
+        Vec::new()
+    } else {
+        vec![0u32; num_terms]
+    };
+    // Charge both grids to the aggregation memory limit.
+    agg_data.context.limits.add_memory_consumed(
+        ((counts.len() + term_counts.len()) * std::mem::size_of::<u32>()) as u64,
+    )?;
+    Ok(Some(Box::new(SegmentTermHistogramCollector {
+        term_counts,
+        counts,
+        num_time_buckets: range.len,
+        base_pos: range.base_pos,
+        terms_req_data: terms_req_data.clone(),
+        hist_req_data,
+        term_block: ColumnBlockAccessor::default(),
+        hist_block: ColumnBlockAccessor::default(),
+        all_docs_in_bounds,
+        is_full: terms_req_data.accessor.get_cardinality().is_full(),
+    })))
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::aggregation::agg_req::Aggregations;
+    use crate::aggregation::tests::{
+        exec_request, exec_request_with_query_and_memory_limit,
+        get_test_index_from_values_and_terms,
+    };
+    use crate::aggregation::AggregationLimitsGuard;
+
+    /// Hand-computed correctness check for the fused terms×histogram fast path
+    /// ([`super::SegmentTermHistogramCollector`]): low-cardinality terms × a histogram leaf over
+    /// full columns, exercised single- and multi-segment.
+    #[test]
+    fn fused_term_histogram_test() -> crate::Result<()> {
+        fused_term_histogram_with_opt(false)?;
+        fused_term_histogram_with_opt(true)?;
+        Ok(())
+    }
+
+    fn fused_term_histogram_with_opt(merge_segments: bool) -> crate::Result<()> {
+        // 300 docs: term = {a, b, c} by i % 3, histogram value = i % 20 (interval 1 => buckets
+        // 0..19). gcd(3, 20) = 1, so every (term, bucket) pair occurs exactly 300 / 60 = 5 times.
+        let docs: Vec<(f64, String)> = (0..300u64)
+            .map(|i| {
+                (
+                    (i % 20) as f64,
+                    ["a", "b", "c"][(i % 3) as usize].to_string(),
+                )
+            })
+            .collect();
+        // Two segments, to also exercise cross-segment merging of the fused per-term histograms.
+        let segments = vec![docs[..150].to_vec(), docs[150..].to_vec()];
+        let index = get_test_index_from_values_and_terms(merge_segments, &segments)?;
+
+        let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
+            "by_term": {
+                "terms": { "field": "string_id", "order": { "_key": "asc" } },
+                "aggs": {
+                    "histo": { "histogram": { "field": "score_f64", "interval": 1.0 } }
+                }
+            }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+
+        for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
+            assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
+            assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
+            let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
+            for b in 0..20usize {
+                assert_eq!(histo[b]["key"], b as f64, "term {term} bucket {b}");
+                assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {b}");
+            }
+            assert_eq!(histo[20], serde_json::Value::Null);
+        }
+        assert_eq!(res["by_term"]["buckets"][3], serde_json::Value::Null);
+
+        Ok(())
+    }
+
+    /// A `missing` config on a *full* term column still takes the fused path (the string sentinel
+    /// is just `col_max + 1`, so the column stays low-cardinality). Since no doc is missing, the
+    /// real term buckets must be exactly as without `missing`.
+    #[test]
+    fn fused_term_histogram_with_missing_on_full_column() -> crate::Result<()> {
+        let docs: Vec<(f64, String)> = (0..300u64)
+            .map(|i| {
+                (
+                    (i % 20) as f64,
+                    ["a", "b", "c"][(i % 3) as usize].to_string(),
+                )
+            })
+            .collect();
+        let index = get_test_index_from_values_and_terms(true, &[docs])?;
+
+        let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
+            "by_term": {
+                "terms": { "field": "string_id", "missing": "MISSING", "order": { "_key": "asc" } },
+                "aggs": {
+                    "histo": { "histogram": { "field": "score_f64", "interval": 1.0 } }
+                }
+            }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+
+        // Column is full, so "MISSING" never applies: a, b, c are unchanged (100 docs, 5 per
+        // bucket).
+        for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
+            assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
+            assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
+            let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
+            for b in 0..20usize {
+                assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {b}");
+            }
+        }
+
+        Ok(())
+    }
+
+    /// Term cardinality above the general path's `MAX_NUM_TERMS_FOR_VEC` (100) still fuses: the
+    /// flat grid is bounded by the total cell count (`num_terms * num_time_buckets`), not the
+    /// term count.
+    #[test]
+    fn fused_term_histogram_many_terms() -> crate::Result<()> {
+        let num_terms = 150usize;
+        let docs_per_term = 2usize;
+        // All docs share histogram value 0 (a single bucket), so the grid is 150 x 1 = 150 cells.
+        let docs: Vec<(f64, String)> = (0..num_terms * docs_per_term)
+            .map(|i| (0.0, format!("t{:03}", i % num_terms)))
+            .collect();
+        let index = get_test_index_from_values_and_terms(true, &[docs])?;
+
+        let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
+            "by_term": {
+                "terms": { "field": "string_id", "size": 1000, "order": { "_key": "asc" } },
+                "aggs": {
+                    "histo": { "histogram": { "field": "score_f64", "interval": 1.0 } }
+                }
+            }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+
+        let buckets = res["by_term"]["buckets"].as_array().unwrap();
+        assert_eq!(buckets.len(), num_terms);
+        for (i, bucket) in buckets.iter().enumerate() {
+            assert_eq!(bucket["key"], format!("t{i:03}"));
+            assert_eq!(bucket["doc_count"], docs_per_term as u64);
+            assert_eq!(bucket["histo"]["buckets"][0]["key"], 0.0);
+            assert_eq!(
+                bucket["histo"]["buckets"][0]["doc_count"],
+                docs_per_term as u64
+            );
+        }
+
+        Ok(())
+    }
+
+    /// `hard_bounds` exercises the non-derived `term_counts` branch: a term's `doc_count` must
+    /// count *every* doc with that term, including docs whose histogram value is outside the
+    /// bounds (those are excluded from the histogram buckets but still counted for the term). This
+    /// is the case where the per-doc `term_counts` increment cannot be replaced by the grid
+    /// row-sum.
+    #[test]
+    fn fused_term_histogram_with_hard_bounds() -> crate::Result<()> {
+        // 300 docs: term = {a, b, c} by i % 3, value = i % 20. Per term: 100 docs, each value in
+        // 0..=19 occurring 5 times.
+        let docs: Vec<(f64, String)> = (0..300u64)
+            .map(|i| {
+                (
+                    (i % 20) as f64,
+                    ["a", "b", "c"][(i % 3) as usize].to_string(),
+                )
+            })
+            .collect();
+        let index = get_test_index_from_values_and_terms(true, &[docs])?;
+
+        // hard_bounds [5, 14] (inclusive) keeps only values 5..=14 in the histogram (10 buckets);
+        // values 0..=4 and 15..=19 are out of bounds.
+        let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
+            "by_term": {
+                "terms": { "field": "string_id", "order": { "_key": "asc" } },
+                "aggs": {
+                    "histo": {
+                        "histogram": {
+                            "field": "score_f64",
+                            "interval": 1.0,
+                            "hard_bounds": { "min": 5.0, "max": 14.0 }
+                        }
+                    }
+                }
+            }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+
+        for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
+            assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
+            // doc_count includes the 50 per-term docs whose value is outside [5, 14].
+            assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
+            let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
+            for b in 0..10usize {
+                let key = 5 + b;
+                assert_eq!(histo[b]["key"], key as f64, "term {term} bucket key {key}");
+                assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {key}");
+            }
+            // Only the 10 in-bounds buckets exist.
+            assert_eq!(histo[10], serde_json::Value::Null);
+        }
+
+        Ok(())
+    }
+
+    /// Non-binding `hard_bounds` (wider than the data, with mid-interval edges) must still produce
+    /// exact results via the derive-from-grid path: since no doc is out of bounds, normalization
+    /// drops the bound, every doc lands in the dense range, and each term's total equals its
+    /// histogram row-sum. This is the case that previously fell back to the per-doc counter only
+    /// because `bounds != [MIN, MAX]`.
+    #[test]
+    fn fused_term_histogram_with_non_binding_hard_bounds() -> crate::Result<()> {
+        // 300 docs: term = {a, b, c} by i % 3, value = i % 20. Data values span [0, 19].
+        let docs: Vec<(f64, String)> = (0..300u64)
+            .map(|i| {
+                (
+                    (i % 20) as f64,
+                    ["a", "b", "c"][(i % 3) as usize].to_string(),
+                )
+            })
+            .collect();
+        let index = get_test_index_from_values_and_terms(true, &[docs])?;
+
+        // Bounds wider than [0, 19], with mid-interval edges -> they exclude nothing.
+        let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
+            "by_term": {
+                "terms": { "field": "string_id", "order": { "_key": "asc" } },
+                "aggs": {
+                    "histo": {
+                        "histogram": {
+                            "field": "score_f64",
+                            "interval": 1.0,
+                            "hard_bounds": { "min": -0.5, "max": 19.5 }
+                        }
+                    }
+                }
+            }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+
+        for (term_idx, term) in ["a", "b", "c"].iter().enumerate() {
+            assert_eq!(res["by_term"]["buckets"][term_idx]["key"], *term);
+            // Every doc is in-bounds, so the per-term total is the full 100 (as without bounds).
+            assert_eq!(res["by_term"]["buckets"][term_idx]["doc_count"], 100);
+            let histo = &res["by_term"]["buckets"][term_idx]["histo"]["buckets"];
+            for b in 0..20usize {
+                assert_eq!(histo[b]["key"], b as f64, "term {term} bucket {b}");
+                assert_eq!(histo[b]["doc_count"], 5, "term {term} bucket {b}");
+            }
+            assert_eq!(histo[20], serde_json::Value::Null);
+        }
+
+        Ok(())
+    }
+
+    /// Regression: with hard bounds the fused path allocates `term_counts` (one `u32`/term) on top
+    /// of the grid, and that allocation must be charged to the memory limit. With many terms and a
+    /// single time bucket the two are equal in size, so a limit admitting the grid alone but not
+    /// grid + `term_counts` must fail.
+    #[test]
+    fn fused_term_histogram_hard_bounds_charges_term_counts() -> crate::Result<()> {
+        // 16k distinct terms, one doc each; values alternate in/out of the single-bucket bounds
+        // [5, 5] so the bounds bind and `term_counts` is allocated. num_terms=16000,
+        // num_time_buckets=1 => `counts` and `term_counts` are ~64 KB each.
+        let docs: Vec<(f64, String)> = (0..16_000u64)
+            .map(|i| (if i % 2 == 0 { 5.0 } else { 10.0 }, format!("t{i:05}")))
+            .collect();
+        let index = get_test_index_from_values_and_terms(true, &[docs])?;
+
+        let agg_req: Aggregations = serde_json::from_value(serde_json::json!({
+            "by_term": {
+                "terms": { "field": "string_id" },
+                "aggs": {
+                    "histo": {
+                        "histogram": {
+                            "field": "score_f64",
+                            "interval": 1.0,
+                            "hard_bounds": { "min": 5.0, "max": 5.0 }
+                        }
+                    }
+                }
+            }
+        }))
+        .unwrap();
+
+        // ~96 KB admits the grid (~64 KB) but not grid + `term_counts` (~128 KB).
+        let err = exec_request_with_query_and_memory_limit(
+            agg_req,
+            &index,
+            None,
+            AggregationLimitsGuard::new(Some(96_000), None),
+        )
+        .unwrap_err();
+        assert!(
+            err.to_string().contains("memory limit was exceeded"),
+            "expected a memory-limit error, got: {err}"
+        );
+
+        Ok(())
+    }
+}

src/aggregation/buffered_sub_aggs.rs

@@ -138,6 +138,7 @@ impl SubAggBuffer for HighCardSubAggBuffer {
         }
     }
 
+    #[inline]
     fn push(&mut self, bucket_id: BucketId, doc_id: DocId) {
         let idx = bucket_id % NUM_PARTITIONS as u32;
         let slot = &mut self.partitions[idx as usize];
@@ -196,6 +197,7 @@ impl SubAggBuffer for LowCardSubAggBuffer {
         }
     }
 
+    #[inline]
     fn push(&mut self, bucket_id: BucketId, doc_id: DocId) {
         let idx = bucket_id as usize;
         if self.per_bucket_docs.len() <= idx {

src/aggregation/intermediate_agg_result.rs

@@ -377,7 +377,22 @@ impl IntermediateMetricResult {
                 MetricResult::ExtendedStats(intermediate_stats.finalize())
             }
             IntermediateMetricResult::Sum(intermediate_sum) => {
-                MetricResult::Sum(intermediate_sum.finalize().into())
+                // By default match Elasticsearch: empty / all-missing sum
+                // buckets serialize as `"value": 0`, not `"value": null`.
+                // The non-ES `none_if_no_match` flag on `SumAggregation`
+                // opts into SQL-style `null` for downstream consumers.
+                let none_if_no_match = req
+                    .agg
+                    .as_sum()
+                    .and_then(|sum| sum.none_if_no_match)
+                    .unwrap_or(false);
+                let value = intermediate_sum.finalize();
+                if none_if_no_match {
+                    MetricResult::Sum(value.into())
+                } else {
+                    let value = Some(value.unwrap_or(0.0));
+                    MetricResult::Sum(value.into())
+                }
             }
             IntermediateMetricResult::Percentiles(percentiles) => MetricResult::Percentiles(
                 percentiles

src/aggregation/metric/cardinality.rs

@@ -171,6 +171,7 @@ impl CouponCache {
             let uninitialized_coupon = Coupon::from_hash(0);
             let mut coupon_map: Vec<Coupon> =
                 vec![uninitialized_coupon; highest_term_ord as usize + 1];
+
             for (term_ord, coupon) in term_ords.into_iter().zip(coupons) {
                 coupon_map[term_ord as usize] = coupon;
             }

src/aggregation/metric/sum.rs

@@ -27,6 +27,16 @@ pub struct SumAggregation {
     /// { "field": "my_numbers", "missing": "10.0" }
     #[serde(default, deserialize_with = "deserialize_option_f64")]
     pub missing: Option<f64>,
+    /// Non-Elasticsearch extension. When `Some(true)`, the serialized result
+    /// returns `"value": null` if no values were collected (all documents had
+    /// missing/NULL values for the field), matching the behavior of `min`,
+    /// `max`, and `avg`. When `None` or `Some(false)` (the default) the
+    /// result returns `"value": 0`, matching Elasticsearch.
+    ///
+    /// Intended for SQL-style consumers where `SUM` of zero rows is `NULL`
+    /// and must be distinguishable from a bucket that genuinely sums to `0`.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub none_if_no_match: Option<bool>,
 }
 
 impl SumAggregation {
@@ -35,6 +45,7 @@ impl SumAggregation {
         Self {
             field: field_name,
             missing: None,
+            none_if_no_match: None,
         }
     }
     /// Returns the field name the aggregation is computed on.
@@ -59,8 +70,104 @@ impl IntermediateSum {
     pub fn merge_fruits(&mut self, other: IntermediateSum) {
         self.stats.merge_fruits(other.stats);
     }
-    /// Computes the final minimum value.
+    /// Computes the final sum value.
+    ///
+    /// Returns `None` when no values were collected, matching the Rust-side
+    /// behavior of `IntermediateMin`, `IntermediateMax`, and
+    /// `IntermediateAvg`. The Elasticsearch-vs-SQL choice for the
+    /// user-visible result is made at the boundary in
+    /// [`IntermediateMetricResult::into_final_metric_result`]: by default
+    /// `None` is coerced to `Some(0.0)` to match Elasticsearch
+    /// (`"value": 0`), and the [`SumAggregation::none_if_no_match`] flag
+    /// opts out of that coercion for SQL-style consumers.
     pub fn finalize(&self) -> Option<f64> {
-        Some(self.stats.finalize().sum)
+        let stats = self.stats.finalize();
+        if stats.count == 0 {
+            None
+        } else {
+            Some(stats.sum)
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_sum_finalize_returns_none_when_no_values() {
+        // Default IntermediateSum has count=0 — finalize should return None,
+        // matching MIN/MAX/AVG behavior for all-NULL groups.
+        let sum = IntermediateSum::default();
+        assert_eq!(sum.finalize(), None);
+    }
+
+    #[test]
+    fn test_sum_finalize_returns_value_when_has_values() {
+        let mut sum = IntermediateSum::default();
+        // Merge in a result that has actual values
+        let stats = IntermediateStats {
+            count: 3,
+            sum: 42.0,
+            min: 10.0,
+            max: 20.0,
+            ..Default::default()
+        };
+        let other = IntermediateSum::from_stats(stats);
+        sum.merge_fruits(other);
+        assert_eq!(sum.finalize(), Some(42.0));
+    }
+
+    #[test]
+    fn test_sum_merge_two_empty_still_none() {
+        let mut a = IntermediateSum::default();
+        let b = IntermediateSum::default();
+        a.merge_fruits(b);
+        assert_eq!(a.finalize(), None);
+    }
+
+    #[test]
+    fn test_sum_aggregation_empty_index_default_matches_es() -> crate::Result<()> {
+        use serde_json::json;
+
+        use crate::aggregation::agg_req::Aggregations;
+        use crate::aggregation::tests::{exec_request, get_test_index_from_terms};
+
+        // Empty index — sum has no values to collect.
+        let values: Vec<Vec<&str>> = vec![];
+        let index = get_test_index_from_terms(false, &values)?;
+        let agg_req: Aggregations = serde_json::from_value(json!({
+            "score_sum": { "sum": { "field": "score" } }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+        // Default: match Elasticsearch — empty sum serializes as 0, not null.
+        assert_eq!(res["score_sum"]["value"], 0.0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_sum_aggregation_empty_index_none_if_no_match_opt_in() -> crate::Result<()> {
+        use serde_json::json;
+
+        use crate::aggregation::agg_req::Aggregations;
+        use crate::aggregation::tests::{exec_request, get_test_index_from_terms};
+
+        let values: Vec<Vec<&str>> = vec![];
+        let index = get_test_index_from_terms(false, &values)?;
+        let agg_req: Aggregations = serde_json::from_value(json!({
+            "score_sum": { "sum": { "field": "score", "none_if_no_match": true } }
+        }))
+        .unwrap();
+
+        let res = exec_request(agg_req, &index)?;
+        // Opt-in non-ES extension — empty sum serializes as null.
+        assert!(
+            res["score_sum"]["value"].is_null(),
+            "expected null, got {:?}",
+            res["score_sum"]["value"]
+        );
+        Ok(())
     }
 }

src/codec/mod.rs

@@ -1,229 +0,0 @@
-/// Codec specific to postings data.
-pub mod postings;
-
-/// Standard tantivy codec. This is the codec you use by default.
-pub mod standard;
-
-use std::io;
-
-pub use standard::StandardCodec;
-
-use crate::codec::postings::PostingsCodec;
-use crate::fieldnorm::FieldNormReader;
-use crate::postings::{Postings, TermInfo};
-use crate::query::score_combiner::DoNothingCombiner;
-use crate::query::term_query::TermScorer;
-use crate::query::{box_scorer, Bm25Weight, BufferedUnionScorer, Scorer, SumCombiner};
-use crate::schema::IndexRecordOption;
-use crate::{DocId, InvertedIndexReader, Score};
-
-/// Codecs describes how data is layed out on disk.
-///
-/// For the moment, only postings codec can be custom.
-pub trait Codec: Clone + std::fmt::Debug + Send + Sync + 'static {
-    /// The specific postings type used by this codec.
-    type PostingsCodec: PostingsCodec;
-
-    /// ID of the codec. It should be unique to your codec.
-    /// Make it human-readable, descriptive, short and unique.
-    const ID: &'static str;
-
-    /// Load codec based on the codec configuration.
-    fn from_json_props(json_value: &serde_json::Value) -> crate::Result<Self>;
-
-    /// Get codec configuration.
-    fn to_json_props(&self) -> serde_json::Value;
-
-    /// Returns the postings codec.
-    fn postings_codec(&self) -> &Self::PostingsCodec;
-}
-
-/// Object-safe codec is a Codec that can be used in a trait object.
-///
-/// The point of it is to offer a way to use a codec without a proliferation of generics.
-pub trait ObjectSafeCodec: 'static + Send + Sync {
-    /// Loads a type-erased Postings object for the given term.
-    ///
-    /// If the schema used to build the index did not provide enough
-    /// information to match the requested `option`, a Postings is still
-    /// returned in a best-effort manner.
-    fn load_postings_type_erased(
-        &self,
-        term_info: &TermInfo,
-        option: IndexRecordOption,
-        inverted_index_reader: &InvertedIndexReader,
-    ) -> io::Result<Box<dyn Postings>>;
-
-    /// Loads a type-erased TermScorer object for the given term.
-    ///
-    /// If the schema used to build the index did not provide enough
-    /// information to match the requested `option`, a TermScorer is still
-    /// returned in a best-effort manner.
-    ///
-    /// The point of this contraption is that the return TermScorer is backed,
-    /// not by Box<dyn Postings> but by the codec's concrete Postings type.
-    fn load_term_scorer_type_erased(
-        &self,
-        term_info: &TermInfo,
-        option: IndexRecordOption,
-        inverted_index_reader: &InvertedIndexReader,
-        fieldnorm_reader: FieldNormReader,
-        similarity_weight: Bm25Weight,
-    ) -> io::Result<Box<dyn Scorer>>;
-
-    /// Loads a type-erased PhraseScorer object for the given term.
-    ///
-    /// If the schema used to build the index did not provide enough
-    /// information to match the requested `option`, a TermScorer is still
-    /// returned in a best-effort manner.
-    ///
-    /// The point of this contraption is that the return PhraseScorer is backed,
-    /// not by Box<dyn Postings> but by the codec's concrete Postings type.
-    fn new_phrase_scorer_type_erased(
-        &self,
-        term_infos: &[(usize, TermInfo)],
-        similarity_weight: Option<Bm25Weight>,
-        fieldnorm_reader: FieldNormReader,
-        slop: u32,
-        inverted_index_reader: &InvertedIndexReader,
-    ) -> io::Result<Box<dyn Scorer>>;
-
-    /// Performs a for_each_pruning operation on the given scorer.
-    ///
-    /// The function will go through matching documents and call the callback
-    /// function for all docs with a score exceeding the threshold.
-    ///
-    /// The function itself will return a larger threshold value,
-    /// meant to update the threshold value.
-    ///
-    /// If the codec and the scorer allow it, this function can rely on
-    /// optimizations like the block-max wand.
-    fn for_each_pruning(
-        &self,
-        threshold: Score,
-        scorer: Box<dyn Scorer>,
-        callback: &mut dyn FnMut(DocId, Score) -> Score,
-    );
-
-    /// Builds a union scorer possibly specialized if
-    /// all scorers are `Term<Self::Postings>`.
-    fn build_union_scorer_with_sum_combiner(
-        &self,
-        scorers: Vec<Box<dyn Scorer>>,
-        num_docs: DocId,
-        score_combiner_type: SumOrDoNothingCombiner,
-    ) -> Box<dyn Scorer>;
-}
-
-impl<TCodec: Codec> ObjectSafeCodec for TCodec {
-    fn load_postings_type_erased(
-        &self,
-        term_info: &TermInfo,
-        option: IndexRecordOption,
-        inverted_index_reader: &InvertedIndexReader,
-    ) -> io::Result<Box<dyn Postings>> {
-        let postings = inverted_index_reader
-            .read_postings_from_terminfo_specialized(term_info, option, self)?;
-        Ok(Box::new(postings))
-    }
-
-    fn load_term_scorer_type_erased(
-        &self,
-        term_info: &TermInfo,
-        option: IndexRecordOption,
-        inverted_index_reader: &InvertedIndexReader,
-        fieldnorm_reader: FieldNormReader,
-        similarity_weight: Bm25Weight,
-    ) -> io::Result<Box<dyn Scorer>> {
-        let scorer = inverted_index_reader.new_term_scorer_specialized(
-            term_info,
-            option,
-            fieldnorm_reader,
-            similarity_weight,
-            self,
-        )?;
-        Ok(box_scorer(scorer))
-    }
-
-    fn new_phrase_scorer_type_erased(
-        &self,
-        term_infos: &[(usize, TermInfo)],
-        similarity_weight: Option<Bm25Weight>,
-        fieldnorm_reader: FieldNormReader,
-        slop: u32,
-        inverted_index_reader: &InvertedIndexReader,
-    ) -> io::Result<Box<dyn Scorer>> {
-        let scorer = inverted_index_reader.new_phrase_scorer_type_specialized(
-            term_infos,
-            similarity_weight,
-            fieldnorm_reader,
-            slop,
-            self,
-        )?;
-        Ok(box_scorer(scorer))
-    }
-
-    fn build_union_scorer_with_sum_combiner(
-        &self,
-        scorers: Vec<Box<dyn Scorer>>,
-        num_docs: DocId,
-        sum_or_do_nothing_combiner: SumOrDoNothingCombiner,
-    ) -> Box<dyn Scorer> {
-        if !scorers.iter().all(|scorer| {
-            scorer.is::<TermScorer<<<Self as Codec>::PostingsCodec as PostingsCodec>::Postings>>()
-        }) {
-            return box_scorer(BufferedUnionScorer::build(
-                scorers,
-                SumCombiner::default,
-                num_docs,
-            ));
-        }
-        let specialized_scorers: Vec<
-            TermScorer<<<Self as Codec>::PostingsCodec as PostingsCodec>::Postings>,
-        > = scorers
-            .into_iter()
-            .map(|scorer| {
-                *scorer.downcast::<TermScorer<_>>().ok().expect(
-                    "Downcast failed despite the fact we already checked the type was correct",
-                )
-            })
-            .collect();
-        match sum_or_do_nothing_combiner {
-            SumOrDoNothingCombiner::Sum => box_scorer(BufferedUnionScorer::build(
-                specialized_scorers,
-                SumCombiner::default,
-                num_docs,
-            )),
-            SumOrDoNothingCombiner::DoNothing => box_scorer(BufferedUnionScorer::build(
-                specialized_scorers,
-                DoNothingCombiner::default,
-                num_docs,
-            )),
-        }
-    }
-
-    fn for_each_pruning(
-        &self,
-        threshold: Score,
-        scorer: Box<dyn Scorer>,
-        callback: &mut dyn FnMut(DocId, Score) -> Score,
-    ) {
-        let accerelerated_foreach_pruning_res =
-            <TCodec as Codec>::PostingsCodec::try_accelerated_for_each_pruning(
-                threshold, scorer, callback,
-            );
-        if let Err(mut scorer) = accerelerated_foreach_pruning_res {
-            // No acceleration available. We need to do things manually.
-            scorer.for_each_pruning(threshold, callback);
-        }
-    }
-}
-
-/// SumCombiner or DoNothingCombiner
-#[derive(Copy, Clone)]
-pub enum SumOrDoNothingCombiner {
-    /// Sum scores together
-    Sum,
-    /// Do not track any score.
-    DoNothing,
-}

src/codec/postings/mod.rs

@@ -1,122 +0,0 @@
-use std::io;
-
-/// Block-max WAND algorithm.
-pub mod block_wand;
-use common::OwnedBytes;
-
-use crate::fieldnorm::FieldNormReader;
-use crate::postings::Postings;
-use crate::query::{Bm25Weight, Scorer};
-use crate::schema::IndexRecordOption;
-use crate::{DocId, Score};
-
-/// Postings codec.
-pub trait PostingsCodec: Send + Sync + 'static {
-    /// Serializer type for the postings codec.
-    type PostingsSerializer: PostingsSerializer;
-    /// Postings type for the postings codec.
-    type Postings: Postings + Clone;
-    /// Creates a new postings serializer.
-    fn new_serializer(
-        &self,
-        avg_fieldnorm: Score,
-        mode: IndexRecordOption,
-        fieldnorm_reader: Option<FieldNormReader>,
-    ) -> Self::PostingsSerializer;
-
-    /// Loads postings
-    ///
-    /// Record option is the option that was passed at indexing time.
-    /// Requested option is the option that is requested.
-    ///
-    /// For instance, we may have term_freq in the posting list
-    /// but we can skip decompressing as we read the posting list.
-    ///
-    /// If record option does not support the requested option,
-    /// this method does NOT return an error and will in fact restrict
-    /// requested_option to what is available.
-    fn load_postings(
-        &self,
-        doc_freq: u32,
-        postings_data: OwnedBytes,
-        record_option: IndexRecordOption,
-        requested_option: IndexRecordOption,
-        positions_data: Option<OwnedBytes>,
-    ) -> io::Result<Self::Postings>;
-
-    /// If your codec supports different ways to accelerate `for_each_pruning` that's
-    /// where you should implement it.
-    ///
-    /// Returning `Err(scorer)` without mutating the scorer nor calling the callback function,
-    /// is never "wrong". It just leaves the responsability to the caller to call a fallback
-    /// implementation on the scorer.
-    ///
-    /// If your codec supports BlockMax-Wand, you just need to have your
-    /// postings implement `PostingsWithBlockMax` and copy what is done in the StandardPostings
-    /// codec to enable it.
-    fn try_accelerated_for_each_pruning(
-        _threshold: Score,
-        scorer: Box<dyn Scorer>,
-        _callback: &mut dyn FnMut(DocId, Score) -> Score,
-    ) -> Result<(), Box<dyn Scorer>> {
-        Err(scorer)
-    }
-}
-
-/// A postings serializer is a listener that is in charge of serializing postings
-///
-/// IO is done only once per postings, once all of the data has been received.
-/// A serializer will therefore contain internal buffers.
-///
-/// A serializer is created once and recycled for all postings.
-///
-/// Clients should use PostingsSerializer as follows.
-/// ```text
-/// // First postings list
-/// serializer.new_term(2, true);
-/// serializer.write_doc(2, 1);
-/// serializer.write_doc(6, 2);
-/// serializer.close_term(3, &mut wrt)?;
-/// // Second postings list
-/// serializer.new_term(1, true);
-/// serializer.write_doc(3, 1);
-/// serializer.close_term(1, &mut wrt)?;
-/// ```
-pub trait PostingsSerializer {
-    /// The term_doc_freq here is the number of documents
-    /// in the postings lists.
-    ///
-    /// It can be used to compute the idf that will be used for the
-    /// blockmax parameters.
-    ///
-    /// If not available (e.g. if we do not collect `term_frequencies`
-    /// blockwand is disabled), the term_doc_freq passed will be set 0.
-    fn new_term(&mut self, term_doc_freq: u32, record_term_freq: bool);
-
-    /// Records a new document id for the current term.
-    /// The serializer may ignore it.
-    fn write_doc(&mut self, doc_id: DocId, term_freq: u32);
-
-    /// Closes the current term and writes the postings list associated.
-    fn close_term(&mut self, doc_freq: u32, wrt: &mut impl io::Write) -> io::Result<()>;
-}
-
-/// A light complement interface to Postings to allow block-max wand acceleration.
-pub trait PostingsWithBlockMax: Postings {
-    /// Moves the postings to the block containign `target_doc` and returns
-    /// an upperbound of the score for documents in the block.
-    ///
-    /// `Warning`: Calling this method may leave the postings in an invalid state.
-    /// callers are required to call seek before calling any other of the
-    /// `Postings` method (like doc / advance etc.).
-    fn seek_block_max(
-        &mut self,
-        target_doc: crate::DocId,
-        fieldnorm_reader: &FieldNormReader,
-        similarity_weight: &Bm25Weight,
-    ) -> Score;
-
-    /// Returns the last document in the current block (or Terminated if this
-    /// is the last block).
-    fn last_doc_in_block(&self) -> crate::DocId;
-}

src/codec/standard/mod.rs

@@ -1,35 +0,0 @@
-use serde::{Deserialize, Serialize};
-
-use crate::codec::standard::postings::StandardPostingsCodec;
-use crate::codec::Codec;
-
-/// Tantivy's default postings codec.
-pub mod postings;
-
-/// Tantivy's default codec.
-#[derive(Debug, Default, Clone, Serialize, Deserialize)]
-pub struct StandardCodec;
-
-impl Codec for StandardCodec {
-    type PostingsCodec = StandardPostingsCodec;
-
-    const ID: &'static str = "tantivy-default";
-
-    fn from_json_props(json_value: &serde_json::Value) -> crate::Result<Self> {
-        if !json_value.is_null() {
-            return Err(crate::TantivyError::InvalidArgument(format!(
-                "Codec property for the StandardCodec are unexpected. expected null, got {}",
-                json_value.as_str().unwrap_or("null")
-            )));
-        }
-        Ok(StandardCodec)
-    }
-
-    fn to_json_props(&self) -> serde_json::Value {
-        serde_json::Value::Null
-    }
-
-    fn postings_codec(&self) -> &Self::PostingsCodec {
-        &StandardPostingsCodec
-    }
-}

src/codec/standard/postings/block.rs

@@ -1,50 +0,0 @@
-use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
-use crate::DocId;
-
-pub struct Block {
-    doc_ids: [DocId; COMPRESSION_BLOCK_SIZE],
-    term_freqs: [u32; COMPRESSION_BLOCK_SIZE],
-    len: usize,
-}
-
-impl Block {
-    pub fn new() -> Self {
-        Block {
-            doc_ids: [0u32; COMPRESSION_BLOCK_SIZE],
-            term_freqs: [0u32; COMPRESSION_BLOCK_SIZE],
-            len: 0,
-        }
-    }
-
-    pub fn doc_ids(&self) -> &[DocId] {
-        &self.doc_ids[..self.len]
-    }
-
-    pub fn term_freqs(&self) -> &[u32] {
-        &self.term_freqs[..self.len]
-    }
-
-    pub fn clear(&mut self) {
-        self.len = 0;
-    }
-
-    pub fn append_doc(&mut self, doc: DocId, term_freq: u32) {
-        let len = self.len;
-        self.doc_ids[len] = doc;
-        self.term_freqs[len] = term_freq;
-        self.len = len + 1;
-    }
-
-    pub fn is_full(&self) -> bool {
-        self.len == COMPRESSION_BLOCK_SIZE
-    }
-
-    pub fn is_empty(&self) -> bool {
-        self.len == 0
-    }
-
-    pub fn last_doc(&self) -> DocId {
-        assert_eq!(self.len, COMPRESSION_BLOCK_SIZE);
-        self.doc_ids[COMPRESSION_BLOCK_SIZE - 1]
-    }
-}

src/codec/standard/postings/mod.rs

@@ -1,164 +0,0 @@
-use std::io;
-
-use crate::codec::postings::block_wand::{block_wand, block_wand_single_scorer};
-use crate::codec::postings::PostingsCodec;
-use crate::codec::standard::postings::block_segment_postings::BlockSegmentPostings;
-pub use crate::codec::standard::postings::segment_postings::SegmentPostings;
-use crate::fieldnorm::FieldNormReader;
-use crate::positions::PositionReader;
-use crate::query::term_query::TermScorer;
-use crate::query::{BufferedUnionScorer, Scorer, SumCombiner};
-use crate::schema::IndexRecordOption;
-use crate::{DocSet as _, Score, TERMINATED};
-
-mod block;
-mod block_segment_postings;
-mod segment_postings;
-mod skip;
-mod standard_postings_serializer;
-
-pub use segment_postings::SegmentPostings as StandardPostings;
-pub use standard_postings_serializer::StandardPostingsSerializer;
-
-/// The default postings codec for tantivy.
-pub struct StandardPostingsCodec;
-
-#[expect(clippy::enum_variant_names)]
-#[derive(Debug, PartialEq, Clone, Copy, Eq)]
-pub(crate) enum FreqReadingOption {
-    NoFreq,
-    SkipFreq,
-    ReadFreq,
-}
-
-impl PostingsCodec for StandardPostingsCodec {
-    type PostingsSerializer = StandardPostingsSerializer;
-    type Postings = SegmentPostings;
-
-    fn new_serializer(
-        &self,
-        avg_fieldnorm: Score,
-        mode: IndexRecordOption,
-        fieldnorm_reader: Option<FieldNormReader>,
-    ) -> Self::PostingsSerializer {
-        StandardPostingsSerializer::new(avg_fieldnorm, mode, fieldnorm_reader)
-    }
-
-    fn load_postings(
-        &self,
-        doc_freq: u32,
-        postings_data: common::OwnedBytes,
-        record_option: IndexRecordOption,
-        requested_option: IndexRecordOption,
-        positions_data_opt: Option<common::OwnedBytes>,
-    ) -> io::Result<Self::Postings> {
-        // Rationalize record_option/requested_option.
-        let requested_option = requested_option.downgrade(record_option);
-        let block_segment_postings =
-            BlockSegmentPostings::open(doc_freq, postings_data, record_option, requested_option)?;
-        let position_reader = positions_data_opt.map(PositionReader::open).transpose()?;
-        Ok(SegmentPostings::from_block_postings(
-            block_segment_postings,
-            position_reader,
-        ))
-    }
-
-    fn try_accelerated_for_each_pruning(
-        mut threshold: Score,
-        mut scorer: Box<dyn Scorer>,
-        callback: &mut dyn FnMut(crate::DocId, Score) -> Score,
-    ) -> Result<(), Box<dyn Scorer>> {
-        scorer = match scorer.downcast::<TermScorer<Self::Postings>>() {
-            Ok(term_scorer) => {
-                block_wand_single_scorer(*term_scorer, threshold, callback);
-                return Ok(());
-            }
-            Err(scorer) => scorer,
-        };
-        let mut union_scorer =
-            scorer.downcast::<BufferedUnionScorer<Box<dyn Scorer>, SumCombiner>>()?;
-        if !union_scorer
-            .scorers()
-            .iter()
-            .all(|scorer| scorer.is::<TermScorer<Self::Postings>>())
-        {
-            return Err(union_scorer);
-        }
-        let doc = union_scorer.doc();
-        if doc == TERMINATED {
-            return Ok(());
-        }
-        let score = union_scorer.score();
-        if score > threshold {
-            threshold = callback(doc, score);
-        }
-        let boxed_scorers: Vec<Box<dyn Scorer>> = union_scorer.into_scorers();
-        let scorers: Vec<TermScorer<Self::Postings>> = boxed_scorers
-            .into_iter()
-            .map(|scorer| {
-                *scorer.downcast::<TermScorer<Self::Postings>>().ok().expect(
-                    "Downcast failed despite the fact we already checked the type was correct",
-                )
-            })
-            .collect();
-        block_wand(scorers, threshold, callback);
-        Ok(())
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use common::OwnedBytes;
-
-    use super::*;
-    use crate::codec::postings::PostingsSerializer as _;
-    use crate::postings::Postings as _;
-
-    fn test_segment_postings_tf_aux(num_docs: u32, include_term_freq: bool) -> SegmentPostings {
-        let mut postings_serializer =
-            StandardPostingsCodec.new_serializer(1.0f32, IndexRecordOption::WithFreqs, None);
-        let mut buffer = Vec::new();
-        postings_serializer.new_term(num_docs, include_term_freq);
-        for i in 0..num_docs {
-            postings_serializer.write_doc(i, 2);
-        }
-        postings_serializer
-            .close_term(num_docs, &mut buffer)
-            .unwrap();
-        StandardPostingsCodec
-            .load_postings(
-                num_docs,
-                OwnedBytes::new(buffer),
-                IndexRecordOption::WithFreqs,
-                IndexRecordOption::WithFreqs,
-                None,
-            )
-            .unwrap()
-    }
-
-    #[test]
-    fn test_segment_postings_small_block_with_and_without_freq() {
-        let small_block_without_term_freq = test_segment_postings_tf_aux(1, false);
-        assert!(!small_block_without_term_freq.has_freq());
-        assert_eq!(small_block_without_term_freq.doc(), 0);
-        assert_eq!(small_block_without_term_freq.term_freq(), 1);
-
-        let small_block_with_term_freq = test_segment_postings_tf_aux(1, true);
-        assert!(small_block_with_term_freq.has_freq());
-        assert_eq!(small_block_with_term_freq.doc(), 0);
-        assert_eq!(small_block_with_term_freq.term_freq(), 2);
-    }
-
-    #[test]
-    fn test_segment_postings_large_block_with_and_without_freq() {
-        let large_block_without_term_freq = test_segment_postings_tf_aux(128, false);
-        assert!(!large_block_without_term_freq.has_freq());
-        assert_eq!(large_block_without_term_freq.doc(), 0);
-        assert_eq!(large_block_without_term_freq.term_freq(), 1);
-
-        let large_block_with_term_freq = test_segment_postings_tf_aux(128, true);
-        assert!(large_block_with_term_freq.has_freq());
-        assert_eq!(large_block_with_term_freq.doc(), 0);
-        assert_eq!(large_block_with_term_freq.term_freq(), 2);
-    }
-}

src/codec/standard/postings/standard_postings_serializer.rs

@@ -1,184 +0,0 @@
-use std::cmp::Ordering;
-use std::io::{self, Write as _};
-
-use common::{BinarySerializable as _, VInt};
-
-use crate::codec::postings::PostingsSerializer;
-use crate::codec::standard::postings::block::Block;
-use crate::codec::standard::postings::skip::SkipSerializer;
-use crate::fieldnorm::FieldNormReader;
-use crate::postings::compression::{BlockEncoder, VIntEncoder as _, COMPRESSION_BLOCK_SIZE};
-use crate::query::Bm25Weight;
-use crate::schema::IndexRecordOption;
-use crate::{DocId, Score};
-
-/// Serializer object for tantivy's default postings format.
-pub struct StandardPostingsSerializer {
-    last_doc_id_encoded: u32,
-
-    block_encoder: BlockEncoder,
-    block: Box<Block>,
-
-    postings_write: Vec<u8>,
-    skip_write: SkipSerializer,
-
-    mode: IndexRecordOption,
-    fieldnorm_reader: Option<FieldNormReader>,
-
-    bm25_weight: Option<Bm25Weight>,
-    avg_fieldnorm: Score, /* Average number of term in the field for that segment.
-                           * this value is used to compute the block wand information. */
-    term_has_freq: bool,
-}
-
-impl StandardPostingsSerializer {
-    pub(crate) fn new(
-        avg_fieldnorm: Score,
-        mode: IndexRecordOption,
-        fieldnorm_reader: Option<FieldNormReader>,
-    ) -> StandardPostingsSerializer {
-        Self {
-            last_doc_id_encoded: 0,
-            block_encoder: BlockEncoder::new(),
-            block: Box::new(Block::new()),
-            postings_write: Vec::new(),
-            skip_write: SkipSerializer::new(),
-            mode,
-            fieldnorm_reader,
-            bm25_weight: None,
-            avg_fieldnorm,
-            term_has_freq: false,
-        }
-    }
-}
-
-impl PostingsSerializer for StandardPostingsSerializer {
-    fn new_term(&mut self, term_doc_freq: u32, record_term_freq: bool) {
-        self.clear();
-
-        self.term_has_freq = self.mode.has_freq() && record_term_freq;
-        if !self.term_has_freq {
-            return;
-        }
-
-        let num_docs_in_segment: u64 =
-            if let Some(fieldnorm_reader) = self.fieldnorm_reader.as_ref() {
-                fieldnorm_reader.num_docs() as u64
-            } else {
-                return;
-            };
-
-        if num_docs_in_segment == 0 {
-            return;
-        }
-
-        self.bm25_weight = Some(Bm25Weight::for_one_term_without_explain(
-            term_doc_freq as u64,
-            num_docs_in_segment,
-            self.avg_fieldnorm,
-        ));
-    }
-
-    fn write_doc(&mut self, doc_id: DocId, term_freq: u32) {
-        self.block.append_doc(doc_id, term_freq);
-        if self.block.is_full() {
-            self.write_block();
-        }
-    }
-
-    fn close_term(&mut self, doc_freq: u32, output_write: &mut impl io::Write) -> io::Result<()> {
-        if !self.block.is_empty() {
-            // we have doc ids waiting to be written
-            // this happens when the number of doc ids is
-            // not a perfect multiple of our block size.
-            //
-            // In that case, the remaining part is encoded
-            // using variable int encoding.
-            {
-                let block_encoded = self
-                    .block_encoder
-                    .compress_vint_sorted(self.block.doc_ids(), self.last_doc_id_encoded);
-                self.postings_write.write_all(block_encoded)?;
-            }
-            // ... Idem for term frequencies
-            if self.term_has_freq {
-                let block_encoded = self
-                    .block_encoder
-                    .compress_vint_unsorted(self.block.term_freqs());
-                self.postings_write.write_all(block_encoded)?;
-            }
-            self.block.clear();
-        }
-        if doc_freq >= COMPRESSION_BLOCK_SIZE as u32 {
-            let skip_data = self.skip_write.data();
-            VInt(skip_data.len() as u64).serialize(output_write)?;
-            output_write.write_all(skip_data)?;
-        }
-        output_write.write_all(&self.postings_write[..])?;
-        self.skip_write.clear();
-        self.postings_write.clear();
-        self.bm25_weight = None;
-        Ok(())
-    }
-}
-
-impl StandardPostingsSerializer {
-    fn clear(&mut self) {
-        self.bm25_weight = None;
-        self.block.clear();
-        self.last_doc_id_encoded = 0;
-    }
-
-    fn write_block(&mut self) {
-        {
-            // encode the doc ids
-            let (num_bits, block_encoded): (u8, &[u8]) = self
-                .block_encoder
-                .compress_block_sorted(self.block.doc_ids(), self.last_doc_id_encoded);
-            self.last_doc_id_encoded = self.block.last_doc();
-            self.skip_write
-                .write_doc(self.last_doc_id_encoded, num_bits);
-            // last el block 0, offset block 1,
-            self.postings_write.extend(block_encoded);
-        }
-        if self.term_has_freq {
-            let (num_bits, block_encoded): (u8, &[u8]) = self
-                .block_encoder
-                .compress_block_unsorted(self.block.term_freqs(), true);
-            self.postings_write.extend(block_encoded);
-            self.skip_write.write_term_freq(num_bits);
-            if self.mode.has_positions() {
-                // We serialize the sum of term freqs within the skip information
-                // in order to navigate through positions.
-                let sum_freq = self.block.term_freqs().iter().cloned().sum();
-                self.skip_write.write_total_term_freq(sum_freq);
-            }
-            let mut blockwand_params = (0u8, 0u32);
-            if let Some(bm25_weight) = self.bm25_weight.as_ref() {
-                if let Some(fieldnorm_reader) = self.fieldnorm_reader.as_ref() {
-                    let docs = self.block.doc_ids().iter().cloned();
-                    let term_freqs = self.block.term_freqs().iter().cloned();
-                    let fieldnorms = docs.map(|doc| fieldnorm_reader.fieldnorm_id(doc));
-                    blockwand_params = fieldnorms
-                        .zip(term_freqs)
-                        .max_by(
-                            |(left_fieldnorm_id, left_term_freq),
-                             (right_fieldnorm_id, right_term_freq)| {
-                                let left_score =
-                                    bm25_weight.tf_factor(*left_fieldnorm_id, *left_term_freq);
-                                let right_score =
-                                    bm25_weight.tf_factor(*right_fieldnorm_id, *right_term_freq);
-                                left_score
-                                    .partial_cmp(&right_score)
-                                    .unwrap_or(Ordering::Equal)
-                            },
-                        )
-                        .unwrap();
-                }
-            }
-            let (fieldnorm_id, term_freq) = blockwand_params;
-            self.skip_write.write_blockwand_max(fieldnorm_id, term_freq);
-        }
-        self.block.clear();
-    }
-}

src/core/json_utils.rs

@@ -4,7 +4,7 @@ use common::{replace_in_place, JsonPathWriter};
 use rustc_hash::FxHashMap;
 
 use crate::indexer::indexing_term::IndexingTerm;
-use crate::postings::{IndexingContext, IndexingPosition, PostingsWriter as _, PostingsWriterEnum};
+use crate::postings::{IndexingContext, IndexingPosition, PostingsWriter};
 use crate::schema::document::{ReferenceValue, ReferenceValueLeaf, Value};
 use crate::schema::{Type, DATE_TIME_PRECISION_INDEXED};
 use crate::time::format_description::well_known::Rfc3339;
@@ -80,7 +80,7 @@ fn index_json_object<'a, V: Value<'a>>(
     text_analyzer: &mut TextAnalyzer,
     term_buffer: &mut IndexingTerm,
     json_path_writer: &mut JsonPathWriter,
-    postings_writer: &mut PostingsWriterEnum,
+    postings_writer: &mut dyn PostingsWriter,
     ctx: &mut IndexingContext,
     positions_per_path: &mut IndexingPositionsPerPath,
 ) {
@@ -110,7 +110,7 @@ pub(crate) fn index_json_value<'a, V: Value<'a>>(
     text_analyzer: &mut TextAnalyzer,
     term_buffer: &mut IndexingTerm,
     json_path_writer: &mut JsonPathWriter,
-    postings_writer: &mut PostingsWriterEnum,
+    postings_writer: &mut dyn PostingsWriter,
     ctx: &mut IndexingContext,
     positions_per_path: &mut IndexingPositionsPerPath,
 ) {

src/docset.rs

@@ -1,6 +1,4 @@
-use std::ops::{Deref as _, DerefMut as _};
-
-use common::BitSet;
+use std::borrow::{Borrow, BorrowMut};
 
 use common::TinySet;
 
@@ -140,19 +138,6 @@ pub trait DocSet: Send {
         buffer.len()
     }
 
-    /// Fills the given bitset with the documents in the docset.
-    ///
-    /// If the docset max_doc is smaller than the largest doc, this function might not consume the
-    /// docset entirely.
-    fn fill_bitset(&mut self, bitset: &mut BitSet) {
-        let bitset_max_value: u32 = bitset.max_value();
-        let mut doc = self.doc();
-        while doc < bitset_max_value {
-            bitset.insert(doc);
-            doc = self.advance();
-        }
-    }
-
     /// Returns the current document
     /// Right after creating a new `DocSet`, the docset points to the first document.
     ///
@@ -293,30 +278,27 @@ impl DocSet for &mut dyn DocSet {
     fn count_including_deleted(&mut self) -> u32 {
         (**self).count_including_deleted()
     }
-
-    fn fill_bitset(&mut self, bitset: &mut BitSet) {
-        (**self).fill_bitset(bitset);
-    }
 }
 
 impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
-    #[inline]
     fn advance(&mut self) -> DocId {
-        self.deref_mut().advance()
+        let unboxed: &mut TDocSet = self.borrow_mut();
+        unboxed.advance()
     }
 
-    #[inline]
     fn seek(&mut self, target: DocId) -> DocId {
-        self.deref_mut().seek(target)
+        let unboxed: &mut TDocSet = self.borrow_mut();
+        unboxed.seek(target)
     }
 
     fn seek_danger(&mut self, target: DocId) -> SeekDangerResult {
-        self.deref_mut().seek_danger(target)
+        let unboxed: &mut TDocSet = self.borrow_mut();
+        unboxed.seek_danger(target)
     }
 
-    #[inline]
     fn fill_buffer(&mut self, buffer: &mut [DocId; COLLECT_BLOCK_BUFFER_LEN]) -> usize {
-        self.deref_mut().fill_buffer(buffer)
+        let unboxed: &mut TDocSet = self.borrow_mut();
+        unboxed.fill_buffer(buffer)
     }
 
     fn fill_bitset_block(
@@ -324,35 +306,32 @@ impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
         min_doc: DocId,
         mask: &mut [TinySet; BLOCK_NUM_TINYBITSETS],
     ) -> DocId {
-        let unboxed: &mut TDocSet = &mut **self;
+        let unboxed: &mut TDocSet = self.borrow_mut();
         unboxed.fill_bitset_block(min_doc, mask)
     }
 
-    #[inline]
     fn doc(&self) -> DocId {
-        self.deref().doc()
+        let unboxed: &TDocSet = self.borrow();
+        unboxed.doc()
     }
 
-    #[inline]
     fn size_hint(&self) -> u32 {
-        self.deref().size_hint()
+        let unboxed: &TDocSet = self.borrow();
+        unboxed.size_hint()
     }
 
-    #[inline]
     fn cost(&self) -> u64 {
-        self.deref().cost()
+        let unboxed: &TDocSet = self.borrow();
+        unboxed.cost()
     }
 
-    #[inline]
     fn count(&mut self, alive_bitset: &AliveBitSet) -> u32 {
-        self.deref_mut().count(alive_bitset)
+        let unboxed: &mut TDocSet = self.borrow_mut();
+        unboxed.count(alive_bitset)
     }
 
     fn count_including_deleted(&mut self) -> u32 {
-        self.deref_mut().count_including_deleted()
-    }
-
-    fn fill_bitset(&mut self, bitset: &mut BitSet) {
-        self.deref_mut().fill_bitset(bitset);
+        let unboxed: &mut TDocSet = self.borrow_mut();
+        unboxed.count_including_deleted()
     }
 }

src/index/codec_configuration.rs

@@ -1,49 +0,0 @@
-use std::borrow::Cow;
-
-use serde::{Deserialize, Serialize};
-
-use crate::codec::{Codec, StandardCodec};
-
-/// A Codec configuration is just a serializable object.
-#[derive(Serialize, Deserialize, Clone, Debug)]
-pub struct CodecConfiguration {
-    codec_id: Cow<'static, str>,
-    #[serde(default, skip_serializing_if = "serde_json::Value::is_null")]
-    props: serde_json::Value,
-}
-
-impl CodecConfiguration {
-    /// Returns true if the codec is the standard codec.
-    pub fn is_standard(&self) -> bool {
-        self.codec_id == StandardCodec::ID && self.props.is_null()
-    }
-
-    /// Creates a codec instance from the configuration.
-    ///
-    /// If the codec id does not match the code's name, an error is returned.
-    pub fn to_codec<C: Codec>(&self) -> crate::Result<C> {
-        if self.codec_id != C::ID {
-            return Err(crate::TantivyError::InvalidArgument(format!(
-                "Codec id mismatch: expected {}, got {}",
-                C::ID,
-                self.codec_id
-            )));
-        }
-        C::from_json_props(&self.props)
-    }
-}
-
-impl<'a, C: Codec> From<&'a C> for CodecConfiguration {
-    fn from(codec: &'a C) -> Self {
-        CodecConfiguration {
-            codec_id: Cow::Borrowed(C::ID),
-            props: codec.to_json_props(),
-        }
-    }
-}
-
-impl Default for CodecConfiguration {
-    fn default() -> Self {
-        CodecConfiguration::from(&StandardCodec)
-    }
-}

src/index/index.rs

@@ -8,14 +8,12 @@ use std::thread::available_parallelism;
 use super::segment::Segment;
 use super::segment_reader::merge_field_meta_data;
 use super::{FieldMetadata, IndexSettings};
-use crate::codec::StandardCodec;
 use crate::core::{Executor, META_FILEPATH};
 use crate::directory::error::OpenReadError;
 #[cfg(feature = "mmap")]
 use crate::directory::MmapDirectory;
 use crate::directory::{Directory, ManagedDirectory, RamDirectory, INDEX_WRITER_LOCK};
 use crate::error::{DataCorruption, TantivyError};
-use crate::index::codec_configuration::CodecConfiguration;
 use crate::index::{IndexMeta, SegmentId, SegmentMeta, SegmentMetaInventory};
 use crate::indexer::index_writer::{
     IndexWriterOptions, MAX_NUM_THREAD, MEMORY_BUDGET_NUM_BYTES_MIN,
@@ -61,7 +59,6 @@ fn save_new_metas(
     schema: Schema,
     index_settings: IndexSettings,
     directory: &dyn Directory,
-    codec: CodecConfiguration,
 ) -> crate::Result<()> {
     save_metas(
         &IndexMeta {
@@ -70,7 +67,6 @@ fn save_new_metas(
             schema,
             opstamp: 0u64,
             payload: None,
-            codec,
         },
         directory,
     )?;
@@ -105,21 +101,18 @@ fn save_new_metas(
 /// };
 /// let index = Index::builder().schema(schema).settings(settings).create_in_ram();
 /// ```
-pub struct IndexBuilder<Codec: crate::codec::Codec = StandardCodec> {
+pub struct IndexBuilder {
     schema: Option<Schema>,
     index_settings: IndexSettings,
     tokenizer_manager: TokenizerManager,
     fast_field_tokenizer_manager: TokenizerManager,
-    codec: Codec,
 }
-
-impl Default for IndexBuilder<StandardCodec> {
+impl Default for IndexBuilder {
     fn default() -> Self {
         IndexBuilder::new()
     }
 }
-
-impl IndexBuilder<StandardCodec> {
+impl IndexBuilder {
     /// Creates a new `IndexBuilder`
     pub fn new() -> Self {
         Self {
@@ -127,21 +120,6 @@ impl IndexBuilder<StandardCodec> {
             index_settings: IndexSettings::default(),
             tokenizer_manager: TokenizerManager::default(),
             fast_field_tokenizer_manager: TokenizerManager::default(),
-            codec: StandardCodec,
-        }
-    }
-}
-
-impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
-    /// Set the codec
-    #[must_use]
-    pub fn codec<NewCodec: crate::codec::Codec>(self, codec: NewCodec) -> IndexBuilder<NewCodec> {
-        IndexBuilder {
-            schema: self.schema,
-            index_settings: self.index_settings,
-            tokenizer_manager: self.tokenizer_manager,
-            fast_field_tokenizer_manager: self.fast_field_tokenizer_manager,
-            codec,
         }
     }
 
@@ -176,7 +154,7 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
     /// The index will be allocated in anonymous memory.
     /// This is useful for indexing small set of documents
     /// for instances like unit test or temporary in memory index.
-    pub fn create_in_ram(self) -> Result<Index<Codec>, TantivyError> {
+    pub fn create_in_ram(self) -> Result<Index, TantivyError> {
         let ram_directory = RamDirectory::create();
         self.create(ram_directory)
     }
@@ -187,7 +165,7 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
     /// If a previous index was in this directory, it returns an
     /// [`TantivyError::IndexAlreadyExists`] error.
     #[cfg(feature = "mmap")]
-    pub fn create_in_dir<P: AsRef<Path>>(self, directory_path: P) -> crate::Result<Index<Codec>> {
+    pub fn create_in_dir<P: AsRef<Path>>(self, directory_path: P) -> crate::Result<Index> {
         let mmap_directory: Box<dyn Directory> = Box::new(MmapDirectory::open(directory_path)?);
         if Index::exists(&*mmap_directory)? {
             return Err(TantivyError::IndexAlreadyExists);
@@ -208,7 +186,7 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
         self,
         dir: impl Into<Box<dyn Directory>>,
         mem_budget: usize,
-    ) -> crate::Result<SingleSegmentIndexWriter<Codec, D>> {
+    ) -> crate::Result<SingleSegmentIndexWriter<D>> {
         let index = self.create(dir)?;
         let index_simple_writer = SingleSegmentIndexWriter::new(index, mem_budget)?;
         Ok(index_simple_writer)
@@ -224,7 +202,7 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
     /// For other unit tests, prefer the [`RamDirectory`], see:
     /// [`IndexBuilder::create_in_ram()`].
     #[cfg(feature = "mmap")]
-    pub fn create_from_tempdir(self) -> crate::Result<Index<Codec>> {
+    pub fn create_from_tempdir(self) -> crate::Result<Index> {
         let mmap_directory: Box<dyn Directory> = Box::new(MmapDirectory::create_from_tempdir()?);
         self.create(mmap_directory)
     }
@@ -237,15 +215,12 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
     }
 
     /// Opens or creates a new index in the provided directory
-    pub fn open_or_create<T: Into<Box<dyn Directory>>>(
-        self,
-        dir: T,
-    ) -> crate::Result<Index<Codec>> {
+    pub fn open_or_create<T: Into<Box<dyn Directory>>>(self, dir: T) -> crate::Result<Index> {
         let dir: Box<dyn Directory> = dir.into();
         if !Index::exists(&*dir)? {
             return self.create(dir);
         }
-        let mut index: Index<Codec> = Index::<Codec>::open_with_codec(dir)?;
+        let mut index = Index::open(dir)?;
         index.set_tokenizers(self.tokenizer_manager.clone());
         if index.schema() == self.get_expect_schema()? {
             Ok(index)
@@ -269,25 +244,18 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
     /// Creates a new index given an implementation of the trait `Directory`.
     ///
     /// If a directory previously existed, it will be erased.
-    pub fn create<T: Into<Box<dyn Directory>>>(self, dir: T) -> crate::Result<Index<Codec>> {
-        self.create_avoid_monomorphization(dir.into())
-    }
-
-    fn create_avoid_monomorphization(self, dir: Box<dyn Directory>) -> crate::Result<Index<Codec>> {
+    fn create<T: Into<Box<dyn Directory>>>(self, dir: T) -> crate::Result<Index> {
         self.validate()?;
+        let dir = dir.into();
         let directory = ManagedDirectory::wrap(dir)?;
-        let codec: CodecConfiguration = CodecConfiguration::from(&self.codec);
         save_new_metas(
             self.get_expect_schema()?,
             self.index_settings.clone(),
             &directory,
-            codec,
         )?;
-        let schema = self.get_expect_schema()?;
-        let mut metas = IndexMeta::with_schema_and_codec(schema, &self.codec);
+        let mut metas = IndexMeta::with_schema(self.get_expect_schema()?);
         metas.index_settings = self.index_settings;
-        let mut index: Index<Codec> =
-            Index::<Codec>::open_from_metas(directory, &metas, SegmentMetaInventory::default())?;
+        let mut index = Index::open_from_metas(directory, &metas, SegmentMetaInventory::default());
         index.set_tokenizers(self.tokenizer_manager);
         index.set_fast_field_tokenizers(self.fast_field_tokenizer_manager);
         Ok(index)
@@ -296,7 +264,7 @@ impl<Codec: crate::codec::Codec> IndexBuilder<Codec> {
 
 /// Search Index
 #[derive(Clone)]
-pub struct Index<Codec: crate::codec::Codec = crate::codec::StandardCodec> {
+pub struct Index {
     directory: ManagedDirectory,
     schema: Schema,
     settings: IndexSettings,
@@ -304,7 +272,6 @@ pub struct Index<Codec: crate::codec::Codec = crate::codec::StandardCodec> {
     tokenizers: TokenizerManager,
     fast_field_tokenizers: TokenizerManager,
     inventory: SegmentMetaInventory,
-    codec: Codec,
 }
 
 impl Index {
@@ -312,6 +279,41 @@ impl Index {
     pub fn builder() -> IndexBuilder {
         IndexBuilder::new()
     }
+    /// Examines the directory to see if it contains an index.
+    ///
+    /// Effectively, it only checks for the presence of the `meta.json` file.
+    pub fn exists(dir: &dyn Directory) -> Result<bool, OpenReadError> {
+        dir.exists(&META_FILEPATH)
+    }
+
+    /// Accessor to the search executor.
+    ///
+    /// This pool is used by default when calling `searcher.search(...)`
+    /// to perform search on the individual segments.
+    ///
+    /// By default the executor is single thread, and simply runs in the calling thread.
+    pub fn search_executor(&self) -> &Executor {
+        &self.executor
+    }
+
+    /// Replace the default single thread search executor pool
+    /// by a thread pool with a given number of threads.
+    pub fn set_multithread_executor(&mut self, num_threads: usize) -> crate::Result<()> {
+        self.executor = Executor::multi_thread(num_threads, "tantivy-search-")?;
+        Ok(())
+    }
+
+    /// Custom thread pool by a outer thread pool.
+    pub fn set_executor(&mut self, executor: Executor) {
+        self.executor = executor;
+    }
+
+    /// Replace the default single thread search executor pool
+    /// by a thread pool with as many threads as there are CPUs on the system.
+    pub fn set_default_multithread_executor(&mut self) -> crate::Result<()> {
+        let default_num_threads = available_parallelism()?.get();
+        self.set_multithread_executor(default_num_threads)
+    }
 
     /// Creates a new index using the [`RamDirectory`].
     ///
@@ -322,13 +324,6 @@ impl Index {
         IndexBuilder::new().schema(schema).create_in_ram().unwrap()
     }
 
-    /// Examines the directory to see if it contains an index.
-    ///
-    /// Effectively, it only checks for the presence of the `meta.json` file.
-    pub fn exists(directory: &dyn Directory) -> Result<bool, OpenReadError> {
-        directory.exists(&META_FILEPATH)
-    }
-
     /// Creates a new index in a given filepath.
     /// The index will use the [`MmapDirectory`].
     ///
@@ -375,108 +370,20 @@ impl Index {
         schema: Schema,
         settings: IndexSettings,
     ) -> crate::Result<Index> {
-        Self::create_to_avoid_monomorphization(dir.into(), schema, settings)
-    }
-
-    fn create_to_avoid_monomorphization(
-        dir: Box<dyn Directory>,
-        schema: Schema,
-        settings: IndexSettings,
-    ) -> crate::Result<Index> {
+        let dir: Box<dyn Directory> = dir.into();
         let mut builder = IndexBuilder::new().schema(schema);
         builder = builder.settings(settings);
         builder.create(dir)
     }
 
-    /// Opens a new directory from an index path.
-    #[cfg(feature = "mmap")]
-    pub fn open_in_dir<P: AsRef<Path>>(directory_path: P) -> crate::Result<Index> {
-        Self::open_in_dir_to_avoid_monomorphization(directory_path.as_ref())
-    }
-
-    #[cfg(feature = "mmap")]
-    #[inline(never)]
-    fn open_in_dir_to_avoid_monomorphization(directory_path: &Path) -> crate::Result<Index> {
-        let mmap_directory = MmapDirectory::open(directory_path)?;
-        Index::open(mmap_directory)
-    }
-
-    /// Open the index using the provided directory
-    pub fn open<T: Into<Box<dyn Directory>>>(directory: T) -> crate::Result<Index> {
-        Index::<StandardCodec>::open_with_codec(directory.into())
-    }
-}
-
-impl<Codec: crate::codec::Codec> Index<Codec> {
-    /// Returns a version of this index with the standard codec.
-    /// This is useful when you need to pass the index to APIs that
-    /// don't care about the codec (e.g., for reading).
-    pub(crate) fn with_standard_codec(&self) -> Index<StandardCodec> {
-        Index {
-            directory: self.directory.clone(),
-            schema: self.schema.clone(),
-            settings: self.settings.clone(),
-            executor: self.executor.clone(),
-            tokenizers: self.tokenizers.clone(),
-            fast_field_tokenizers: self.fast_field_tokenizers.clone(),
-            inventory: self.inventory.clone(),
-            codec: StandardCodec,
-        }
-    }
-
-    /// Open the index using the provided directory
-    #[inline(never)]
-    pub fn open_with_codec(directory: Box<dyn Directory>) -> crate::Result<Index<Codec>> {
-        let directory = ManagedDirectory::wrap(directory)?;
-        let inventory = SegmentMetaInventory::default();
-        let metas = load_metas(&directory, &inventory)?;
-        let index: Index<Codec> = Index::<Codec>::open_from_metas(directory, &metas, inventory)?;
-        Ok(index)
-    }
-
-    /// Accessor to the codec.
-    pub fn codec(&self) -> &Codec {
-        &self.codec
-    }
-
-    /// Accessor to the search executor.
-    ///
-    /// This pool is used by default when calling `searcher.search(...)`
-    /// to perform search on the individual segments.
-    ///
-    /// By default the executor is single thread, and simply runs in the calling thread.
-    pub fn search_executor(&self) -> &Executor {
-        &self.executor
-    }
-
-    /// Replace the default single thread search executor pool
-    /// by a thread pool with a given number of threads.
-    pub fn set_multithread_executor(&mut self, num_threads: usize) -> crate::Result<()> {
-        self.executor = Executor::multi_thread(num_threads, "tantivy-search-")?;
-        Ok(())
-    }
-
-    /// Custom thread pool by a outer thread pool.
-    pub fn set_executor(&mut self, executor: Executor) {
-        self.executor = executor;
-    }
-
-    /// Replace the default single thread search executor pool
-    /// by a thread pool with as many threads as there are CPUs on the system.
-    pub fn set_default_multithread_executor(&mut self) -> crate::Result<()> {
-        let default_num_threads = available_parallelism()?.get();
-        self.set_multithread_executor(default_num_threads)
-    }
-
     /// Creates a new index given a directory and an [`IndexMeta`].
-    fn open_from_metas<C: crate::codec::Codec>(
+    fn open_from_metas(
         directory: ManagedDirectory,
         metas: &IndexMeta,
         inventory: SegmentMetaInventory,
-    ) -> crate::Result<Index<C>> {
+    ) -> Index {
         let schema = metas.schema.clone();
-        let codec = metas.codec.to_codec::<C>()?;
-        Ok(Index {
+        Index {
             settings: metas.index_settings.clone(),
             directory,
             schema,
@@ -484,8 +391,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
             fast_field_tokenizers: TokenizerManager::default(),
             executor: Executor::single_thread(),
             inventory,
-            codec,
-        })
+        }
     }
 
     /// Setter for the tokenizer manager.
@@ -541,7 +447,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     /// Create a default [`IndexReader`] for the given index.
     ///
     /// See [`Index.reader_builder()`].
-    pub fn reader(&self) -> crate::Result<IndexReader<Codec>> {
+    pub fn reader(&self) -> crate::Result<IndexReader> {
         self.reader_builder().try_into()
     }
 
@@ -549,10 +455,17 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     ///
     /// Most project should create at most one reader for a given index.
     /// This method is typically called only once per `Index` instance.
-    pub fn reader_builder(&self) -> IndexReaderBuilder<Codec> {
+    pub fn reader_builder(&self) -> IndexReaderBuilder {
         IndexReaderBuilder::new(self.clone())
     }
 
+    /// Opens a new directory from an index path.
+    #[cfg(feature = "mmap")]
+    pub fn open_in_dir<P: AsRef<Path>>(directory_path: P) -> crate::Result<Index> {
+        let mmap_directory = MmapDirectory::open(directory_path)?;
+        Index::open(mmap_directory)
+    }
+
     /// Returns the list of the segment metas tracked by the index.
     ///
     /// Such segments can of course be part of the index,
@@ -593,6 +506,16 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
         self.inventory.new_segment_meta(segment_id, max_doc)
     }
 
+    /// Open the index using the provided directory
+    pub fn open<T: Into<Box<dyn Directory>>>(directory: T) -> crate::Result<Index> {
+        let directory = directory.into();
+        let directory = ManagedDirectory::wrap(directory)?;
+        let inventory = SegmentMetaInventory::default();
+        let metas = load_metas(&directory, &inventory)?;
+        let index = Index::open_from_metas(directory, &metas, inventory);
+        Ok(index)
+    }
+
     /// Reads the index meta file from the directory.
     pub fn load_metas(&self) -> crate::Result<IndexMeta> {
         load_metas(self.directory(), &self.inventory)
@@ -616,7 +539,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     pub fn writer_with_options<D: Document>(
         &self,
         options: IndexWriterOptions,
-    ) -> crate::Result<IndexWriter<Codec, D>> {
+    ) -> crate::Result<IndexWriter<D>> {
         let directory_lock = self
             .directory
             .acquire_lock(&INDEX_WRITER_LOCK)
@@ -658,7 +581,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
         &self,
         num_threads: usize,
         overall_memory_budget_in_bytes: usize,
-    ) -> crate::Result<IndexWriter<Codec, D>> {
+    ) -> crate::Result<IndexWriter<D>> {
         let memory_arena_in_bytes_per_thread = overall_memory_budget_in_bytes / num_threads;
         let options = IndexWriterOptions::builder()
             .num_worker_threads(num_threads)
@@ -672,7 +595,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     /// That index writer only simply has a single thread and a memory budget of 15 MB.
     /// Using a single thread gives us a deterministic allocation of DocId.
     #[cfg(test)]
-    pub fn writer_for_tests<D: Document>(&self) -> crate::Result<IndexWriter<Codec, D>> {
+    pub fn writer_for_tests<D: Document>(&self) -> crate::Result<IndexWriter<D>> {
         self.writer_with_num_threads(1, MEMORY_BUDGET_NUM_BYTES_MIN)
     }
 
@@ -690,7 +613,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     pub fn writer<D: Document>(
         &self,
         memory_budget_in_bytes: usize,
-    ) -> crate::Result<IndexWriter<Codec, D>> {
+    ) -> crate::Result<IndexWriter<D>> {
         let mut num_threads = std::cmp::min(available_parallelism()?.get(), MAX_NUM_THREAD);
         let memory_budget_num_bytes_per_thread = memory_budget_in_bytes / num_threads;
         if memory_budget_num_bytes_per_thread < MEMORY_BUDGET_NUM_BYTES_MIN {
@@ -717,7 +640,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     }
 
     /// Returns the list of segments that are searchable
-    pub fn searchable_segments(&self) -> crate::Result<Vec<Segment<Codec>>> {
+    pub fn searchable_segments(&self) -> crate::Result<Vec<Segment>> {
         Ok(self
             .searchable_segment_metas()?
             .into_iter()
@@ -726,12 +649,12 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
     }
 
     #[doc(hidden)]
-    pub fn segment(&self, segment_meta: SegmentMeta) -> Segment<Codec> {
+    pub fn segment(&self, segment_meta: SegmentMeta) -> Segment {
         Segment::for_index(self.clone(), segment_meta)
     }
 
     /// Creates a new segment.
-    pub fn new_segment(&self) -> Segment<Codec> {
+    pub fn new_segment(&self) -> Segment {
         let segment_meta = self
             .inventory
             .new_segment_meta(SegmentId::generate_random(), 0);
@@ -785,7 +708,7 @@ impl<Codec: crate::codec::Codec> Index<Codec> {
 }
 
 impl fmt::Debug for Index {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         write!(f, "Index({:?})", self.directory)
     }
 }

src/index/index_meta.rs

@@ -5,8 +5,7 @@ use std::path::PathBuf;
 use serde::{Deserialize, Serialize};
 
 use super::SegmentComponent;
-use crate::codec::Codec;
-use crate::index::{CodecConfiguration, SegmentId};
+use crate::index::SegmentId;
 use crate::schema::Schema;
 use crate::store::Compressor;
 use crate::{Inventory, Opstamp, TrackedObject};
@@ -287,10 +286,8 @@ pub struct IndexMeta {
     /// This payload is entirely unused by tantivy.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub payload: Option<String>,
-    /// Codec configuration for the index.
-    #[serde(skip_serializing_if = "CodecConfiguration::is_standard")]
-    pub codec: CodecConfiguration,
 }
+
 #[derive(Deserialize, Debug)]
 struct UntrackedIndexMeta {
     pub segments: Vec<InnerSegmentMeta>,
@@ -300,8 +297,6 @@ struct UntrackedIndexMeta {
     pub opstamp: Opstamp,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub payload: Option<String>,
-    #[serde(default)]
-    pub codec: CodecConfiguration,
 }
 
 impl UntrackedIndexMeta {
@@ -316,7 +311,6 @@ impl UntrackedIndexMeta {
             schema: self.schema,
             opstamp: self.opstamp,
             payload: self.payload,
-            codec: self.codec,
         }
     }
 }
@@ -327,14 +321,13 @@ impl IndexMeta {
     ///
     /// This new index does not contains any segments.
     /// Opstamp will the value `0u64`.
-    pub fn with_schema_and_codec<C: Codec>(schema: Schema, codec: &C) -> IndexMeta {
+    pub fn with_schema(schema: Schema) -> IndexMeta {
         IndexMeta {
             index_settings: IndexSettings::default(),
             segments: vec![],
             schema,
             opstamp: 0u64,
             payload: None,
-            codec: CodecConfiguration::from(codec),
         }
     }
 
@@ -385,38 +378,14 @@ mod tests {
             schema,
             opstamp: 0u64,
             payload: None,
-            codec: Default::default(),
         };
-        let json_value: serde_json::Value =
-            serde_json::to_value(&index_metas).expect("serialization failed");
+        let json = serde_json::ser::to_string(&index_metas).expect("serialization failed");
         assert_eq!(
-            &json_value,
-            &serde_json::json!(
-            {
-              "index_settings": {
-                "docstore_compression": "none",
-                "docstore_blocksize": 16384
-              },
-              "segments": [],
-              "schema": [
-                {
-                  "name": "text",
-                  "type": "text",
-                  "options": {
-                    "indexing": {
-                      "record": "position",
-                      "fieldnorms": true,
-                      "tokenizer": "default"
-                    },
-                    "stored": false,
-                    "fast": false
-                  }
-                }
-              ],
-              "opstamp": 0
-            })
+            json,
+            r#"{"index_settings":{"docstore_compression":"none","docstore_blocksize":16384},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","fieldnorms":true,"tokenizer":"default"},"stored":false,"fast":false}}],"opstamp":0}"#
         );
-        let deser_meta: UntrackedIndexMeta = serde_json::from_value(json_value).unwrap();
+
+        let deser_meta: UntrackedIndexMeta = serde_json::from_str(&json).unwrap();
         assert_eq!(index_metas.index_settings, deser_meta.index_settings);
         assert_eq!(index_metas.schema, deser_meta.schema);
         assert_eq!(index_metas.opstamp, deser_meta.opstamp);
@@ -442,39 +411,14 @@ mod tests {
             schema,
             opstamp: 0u64,
             payload: None,
-            codec: Default::default(),
         };
-        let json_value = serde_json::to_value(&index_metas).expect("serialization failed");
+        let json = serde_json::ser::to_string(&index_metas).expect("serialization failed");
         assert_eq!(
-            &json_value,
-            &serde_json::json!(
-                {
-                  "index_settings": {
-                    "docstore_compression": "zstd(compression_level=4)",
-                    "docstore_blocksize": 1000000
-                  },
-                  "segments": [],
-                  "schema": [
-                    {
-                      "name": "text",
-                      "type": "text",
-                      "options": {
-                        "indexing": {
-                          "record": "position",
-                          "fieldnorms": true,
-                          "tokenizer": "default"
-                        },
-                        "stored": false,
-                        "fast": false
-                      }
-                    }
-                  ],
-                  "opstamp": 0
-                }
-            )
+            json,
+            r#"{"index_settings":{"docstore_compression":"zstd(compression_level=4)","docstore_blocksize":1000000},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","fieldnorms":true,"tokenizer":"default"},"stored":false,"fast":false}}],"opstamp":0}"#
         );
 
-        let deser_meta: UntrackedIndexMeta = serde_json::from_value(json_value).unwrap();
+        let deser_meta: UntrackedIndexMeta = serde_json::from_str(&json).unwrap();
         assert_eq!(index_metas.index_settings, deser_meta.index_settings);
         assert_eq!(index_metas.schema, deser_meta.schema);
         assert_eq!(index_metas.opstamp, deser_meta.opstamp);

src/index/inverted_index_reader.rs

@@ -1,8 +1,7 @@
 use std::io;
-use std::sync::Arc;
 
 use common::json_path_writer::JSON_END_OF_PATH;
-use common::{BinarySerializable, ByteCount, OwnedBytes};
+use common::{BinarySerializable, ByteCount};
 #[cfg(feature = "quickwit")]
 use futures_util::{FutureExt, StreamExt, TryStreamExt};
 #[cfg(feature = "quickwit")]
@@ -10,13 +9,9 @@ use itertools::Itertools;
 #[cfg(feature = "quickwit")]
 use tantivy_fst::automaton::{AlwaysMatch, Automaton};
 
-use crate::codec::postings::PostingsCodec;
-use crate::codec::{Codec, ObjectSafeCodec, StandardCodec};
 use crate::directory::FileSlice;
-use crate::fieldnorm::FieldNormReader;
-use crate::postings::{Postings, TermInfo};
-use crate::query::term_query::TermScorer;
-use crate::query::{Bm25Weight, PhraseScorer, Scorer};
+use crate::positions::PositionReader;
+use crate::postings::{BlockSegmentPostings, SegmentPostings, TermInfo};
 use crate::schema::{IndexRecordOption, Term, Type};
 use crate::termdict::TermDictionary;
 
@@ -38,7 +33,6 @@ pub struct InvertedIndexReader {
     positions_file_slice: FileSlice,
     record_option: IndexRecordOption,
     total_num_tokens: u64,
-    codec: Arc<dyn ObjectSafeCodec>,
 }
 
 /// Object that records the amount of space used by a field in an inverted index.
@@ -74,7 +68,6 @@ impl InvertedIndexReader {
         postings_file_slice: FileSlice,
         positions_file_slice: FileSlice,
         record_option: IndexRecordOption,
-        codec: Arc<dyn ObjectSafeCodec>,
     ) -> io::Result<InvertedIndexReader> {
         let (total_num_tokens_slice, postings_body) = postings_file_slice.split(8);
         let total_num_tokens = u64::deserialize(&mut total_num_tokens_slice.read_bytes()?)?;
@@ -84,7 +77,6 @@ impl InvertedIndexReader {
             positions_file_slice,
             record_option,
             total_num_tokens,
-            codec,
         })
     }
 
@@ -97,7 +89,6 @@ impl InvertedIndexReader {
             positions_file_slice: FileSlice::empty(),
             record_option,
             total_num_tokens: 0u64,
-            codec: Arc::new(StandardCodec),
         }
     }
 
@@ -169,98 +160,61 @@ impl InvertedIndexReader {
         Ok(fields)
     }
 
-    pub(crate) fn new_term_scorer_specialized<C: Codec>(
-        &self,
-        term_info: &TermInfo,
-        option: IndexRecordOption,
-        fieldnorm_reader: FieldNormReader,
-        similarity_weight: Bm25Weight,
-        codec: &C,
-    ) -> io::Result<TermScorer<<<C as Codec>::PostingsCodec as PostingsCodec>::Postings>> {
-        let postings = self.read_postings_from_terminfo_specialized(term_info, option, codec)?;
-        let term_scorer = TermScorer::new(postings, fieldnorm_reader, similarity_weight);
-        Ok(term_scorer)
-    }
-
-    pub(crate) fn new_phrase_scorer_type_specialized<C: Codec>(
-        &self,
-        term_infos: &[(usize, TermInfo)],
-        similarity_weight_opt: Option<Bm25Weight>,
-        fieldnorm_reader: FieldNormReader,
-        slop: u32,
-        codec: &C,
-    ) -> io::Result<PhraseScorer<<<C as Codec>::PostingsCodec as PostingsCodec>::Postings>> {
-        let mut offset_and_term_postings: Vec<(
-            usize,
-            <<C as Codec>::PostingsCodec as PostingsCodec>::Postings,
-        )> = Vec::with_capacity(term_infos.len());
-        for (offset, term_info) in term_infos {
-            let postings = self.read_postings_from_terminfo_specialized(
-                term_info,
-                IndexRecordOption::WithFreqsAndPositions,
-                codec,
-            )?;
-            offset_and_term_postings.push((*offset, postings));
-        }
-        let phrase_scorer = PhraseScorer::new(
-            offset_and_term_postings,
-            similarity_weight_opt,
-            fieldnorm_reader,
-            slop,
-        );
-        Ok(phrase_scorer)
-    }
-
-    /// Build a new term scorer.
-    pub fn new_term_scorer(
-        &self,
-        term_info: &TermInfo,
-        option: IndexRecordOption,
-        fieldnorm_reader: FieldNormReader,
-        similarity_weight: Bm25Weight,
-    ) -> io::Result<Box<dyn Scorer>> {
-        let term_scorer = self.codec.load_term_scorer_type_erased(
-            term_info,
-            option,
-            self,
-            fieldnorm_reader,
-            similarity_weight,
-        )?;
-        Ok(term_scorer)
-    }
-
-    /// Returns a postings object specific with a concrete type.
+    /// Resets the block segment to another position of the postings
+    /// file.
     ///
-    /// This requires you to provied the actual codec.
-    pub fn read_postings_from_terminfo_specialized<C: Codec>(
+    /// This is useful for enumerating through a list of terms,
+    /// and consuming the associated posting lists while avoiding
+    /// reallocating a [`BlockSegmentPostings`].
+    ///
+    /// # Warning
+    ///
+    /// This does not reset the positions list.
+    pub fn reset_block_postings_from_terminfo(
         &self,
         term_info: &TermInfo,
+        block_postings: &mut BlockSegmentPostings,
+    ) -> io::Result<()> {
+        let postings_slice = self
+            .postings_file_slice
+            .slice(term_info.postings_range.clone());
+        let postings_bytes = postings_slice.read_bytes()?;
+        block_postings.reset(term_info.doc_freq, postings_bytes)?;
+        Ok(())
+    }
+
+    /// Returns a block postings given a `Term`.
+    /// This method is for an advanced usage only.
+    ///
+    /// Most users should prefer using [`Self::read_postings()`] instead.
+    pub fn read_block_postings(
+        &self,
+        term: &Term,
         option: IndexRecordOption,
-        codec: &C,
-    ) -> io::Result<<<C as Codec>::PostingsCodec as PostingsCodec>::Postings> {
-        let option = option.downgrade(self.record_option);
+    ) -> io::Result<Option<BlockSegmentPostings>> {
+        self.get_term_info(term)?
+            .map(move |term_info| self.read_block_postings_from_terminfo(&term_info, option))
+            .transpose()
+    }
+
+    /// Returns a block postings given a `term_info`.
+    /// This method is for an advanced usage only.
+    ///
+    /// Most users should prefer using [`Self::read_postings()`] instead.
+    pub fn read_block_postings_from_terminfo(
+        &self,
+        term_info: &TermInfo,
+        requested_option: IndexRecordOption,
+    ) -> io::Result<BlockSegmentPostings> {
         let postings_data = self
             .postings_file_slice
-            .slice(term_info.postings_range.clone())
-            .read_bytes()?;
-        let positions_data: Option<OwnedBytes> = if option.has_positions() {
-            let positions_data = self
-                .positions_file_slice
-                .slice(term_info.positions_range.clone())
-                .read_bytes()?;
-            Some(positions_data)
-        } else {
-            None
-        };
-        let postings: <<C as Codec>::PostingsCodec as PostingsCodec>::Postings =
-            codec.postings_codec().load_postings(
-                term_info.doc_freq,
-                postings_data,
-                self.record_option,
-                option,
-                positions_data,
-            )?;
-        Ok(postings)
+            .slice(term_info.postings_range.clone());
+        BlockSegmentPostings::open(
+            term_info.doc_freq,
+            postings_data,
+            self.record_option,
+            requested_option,
+        )
     }
 
     /// Returns a posting object given a `term_info`.
@@ -271,9 +225,25 @@ impl InvertedIndexReader {
         &self,
         term_info: &TermInfo,
         option: IndexRecordOption,
-    ) -> io::Result<Box<dyn Postings>> {
-        self.codec
-            .load_postings_type_erased(term_info, option, self)
+    ) -> io::Result<SegmentPostings> {
+        let option = option.downgrade(self.record_option);
+
+        let block_postings = self.read_block_postings_from_terminfo(term_info, option)?;
+        let position_reader = {
+            if option.has_positions() {
+                let positions_data = self
+                    .positions_file_slice
+                    .read_bytes_slice(term_info.positions_range.clone())?;
+                let position_reader = PositionReader::open(positions_data)?;
+                Some(position_reader)
+            } else {
+                None
+            }
+        };
+        Ok(SegmentPostings::from_block_postings(
+            block_postings,
+            position_reader,
+        ))
     }
 
     /// Returns the total number of tokens recorded for all documents
@@ -296,7 +266,7 @@ impl InvertedIndexReader {
         &self,
         term: &Term,
         option: IndexRecordOption,
-    ) -> io::Result<Option<Box<dyn Postings>>> {
+    ) -> io::Result<Option<SegmentPostings>> {
         self.get_term_info(term)?
             .map(move |term_info| self.read_postings_from_terminfo(&term_info, option))
             .transpose()

src/index/mod.rs

@@ -2,7 +2,6 @@
 //!
 //! It contains `Index` and `Segment`, where a `Index` consists of one or more `Segment`s.
 
-mod codec_configuration;
 mod index;
 mod index_meta;
 mod inverted_index_reader;
@@ -11,7 +10,6 @@ mod segment_component;
 mod segment_id;
 mod segment_reader;
 
-pub use self::codec_configuration::CodecConfiguration;
 pub use self::index::{Index, IndexBuilder};
 pub(crate) use self::index_meta::SegmentMetaInventory;
 pub use self::index_meta::{IndexMeta, IndexSettings, Order, SegmentMeta};

src/index/segment.rs

@@ -2,7 +2,6 @@ use std::fmt;
 use std::path::PathBuf;
 
 use super::SegmentComponent;
-use crate::codec::StandardCodec;
 use crate::directory::error::{OpenReadError, OpenWriteError};
 use crate::directory::{Directory, FileSlice, WritePtr};
 use crate::index::{Index, SegmentId, SegmentMeta};
@@ -11,25 +10,25 @@ use crate::Opstamp;
 
 /// A segment is a piece of the index.
 #[derive(Clone)]
-pub struct Segment<C: crate::codec::Codec = StandardCodec> {
-    index: Index<C>,
+pub struct Segment {
+    index: Index,
     meta: SegmentMeta,
 }
 
-impl<C: crate::codec::Codec> fmt::Debug for Segment<C> {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+impl fmt::Debug for Segment {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         write!(f, "Segment({:?})", self.id().uuid_string())
     }
 }
 
-impl<C: crate::codec::Codec> Segment<C> {
+impl Segment {
     /// Creates a new segment given an `Index` and a `SegmentId`
-    pub(crate) fn for_index(index: Index<C>, meta: SegmentMeta) -> Segment<C> {
+    pub(crate) fn for_index(index: Index, meta: SegmentMeta) -> Segment {
         Segment { index, meta }
     }
 
     /// Returns the index the segment belongs to.
-    pub fn index(&self) -> &Index<C> {
+    pub fn index(&self) -> &Index {
         &self.index
     }
 
@@ -47,7 +46,7 @@ impl<C: crate::codec::Codec> Segment<C> {
     ///
     /// This method is only used when updating `max_doc` from 0
     /// as we finalize a fresh new segment.
-    pub fn with_max_doc(self, max_doc: u32) -> Segment<C> {
+    pub fn with_max_doc(self, max_doc: u32) -> Segment {
         Segment {
             index: self.index,
             meta: self.meta.with_max_doc(max_doc),
@@ -56,7 +55,7 @@ impl<C: crate::codec::Codec> Segment<C> {
 
     #[doc(hidden)]
     #[must_use]
-    pub fn with_delete_meta(self, num_deleted_docs: u32, opstamp: Opstamp) -> Segment<C> {
+    pub fn with_delete_meta(self, num_deleted_docs: u32, opstamp: Opstamp) -> Segment {
         Segment {
             index: self.index,
             meta: self.meta.with_delete_meta(num_deleted_docs, opstamp),

src/index/segment_reader.rs

@@ -6,7 +6,6 @@ use common::{ByteCount, HasLen};
 use fnv::FnvHashMap;
 use itertools::Itertools;
 
-use crate::codec::ObjectSafeCodec;
 use crate::directory::error::OpenReadError;
 use crate::directory::{CompositeFile, FileSlice};
 use crate::error::DataCorruption;
@@ -49,7 +48,6 @@ pub struct SegmentReader {
     store_file: FileSlice,
     alive_bitset_opt: Option<AliveBitSet>,
     schema: Schema,
-    codec: Arc<dyn ObjectSafeCodec>,
 }
 
 impl SegmentReader {
@@ -70,11 +68,6 @@ impl SegmentReader {
         &self.schema
     }
 
-    /// Returns the index codec.
-    pub fn codec(&self) -> &dyn ObjectSafeCodec {
-        &*self.codec
-    }
-
     /// Return the number of documents that have been
     /// deleted in the segment.
     pub fn num_deleted_docs(&self) -> DocId {
@@ -148,16 +141,15 @@ impl SegmentReader {
     }
 
     /// Open a new segment for reading.
-    pub fn open<C: crate::codec::Codec>(segment: &Segment<C>) -> crate::Result<SegmentReader> {
+    pub fn open(segment: &Segment) -> crate::Result<SegmentReader> {
         Self::open_with_custom_alive_set(segment, None)
     }
 
     /// Open a new segment for reading.
-    pub fn open_with_custom_alive_set<C: crate::codec::Codec>(
-        segment: &Segment<C>,
+    pub fn open_with_custom_alive_set(
+        segment: &Segment,
         custom_bitset: Option<AliveBitSet>,
     ) -> crate::Result<SegmentReader> {
-        let codec: Arc<dyn ObjectSafeCodec> = Arc::new(segment.index().codec().clone());
         let termdict_file = segment.open_read(SegmentComponent::Terms)?;
         let termdict_composite = CompositeFile::open(&termdict_file)?;
 
@@ -211,7 +203,6 @@ impl SegmentReader {
             alive_bitset_opt,
             positions_composite,
             schema,
-            codec,
         })
     }
 
@@ -281,7 +272,6 @@ impl SegmentReader {
             postings_file,
             positions_file,
             record_option,
-            self.codec.clone(),
         )?);
 
         // by releasing the lock in between, we may end up opening the inverting index

src/indexer/index_writer.rs

@@ -9,7 +9,6 @@ use smallvec::smallvec;
 use super::operation::{AddOperation, UserOperation};
 use super::segment_updater::SegmentUpdater;
 use super::{AddBatch, AddBatchReceiver, AddBatchSender, PreparedCommit};
-use crate::codec::{Codec, StandardCodec};
 use crate::directory::{DirectoryLock, GarbageCollectionResult, TerminatingWrite};
 use crate::error::TantivyError;
 use crate::fastfield::write_alive_bitset;
@@ -69,12 +68,12 @@ pub struct IndexWriterOptions {
 /// indexing queue.
 /// Each indexing thread builds its own independent [`Segment`], via
 /// a `SegmentWriter` object.
-pub struct IndexWriter<C: Codec = StandardCodec, D: Document = TantivyDocument> {
+pub struct IndexWriter<D: Document = TantivyDocument> {
     // the lock is just used to bind the
     // lifetime of the lock with that of the IndexWriter.
     _directory_lock: Option<DirectoryLock>,
 
-    index: Index<C>,
+    index: Index,
 
     options: IndexWriterOptions,
 
@@ -83,7 +82,7 @@ pub struct IndexWriter<C: Codec = StandardCodec, D: Document = TantivyDocument>
     index_writer_status: IndexWriterStatus<D>,
     operation_sender: AddBatchSender<D>,
 
-    segment_updater: SegmentUpdater<C>,
+    segment_updater: SegmentUpdater,
 
     worker_id: usize,
 
@@ -129,8 +128,8 @@ fn compute_deleted_bitset(
 /// is `==` target_opstamp.
 /// For instance, there was no delete operation between the state of the `segment_entry` and
 /// the `target_opstamp`, `segment_entry` is not updated.
-pub fn advance_deletes<C: Codec>(
-    mut segment: Segment<C>,
+pub fn advance_deletes(
+    mut segment: Segment,
     segment_entry: &mut SegmentEntry,
     target_opstamp: Opstamp,
 ) -> crate::Result<()> {
@@ -180,11 +179,11 @@ pub fn advance_deletes<C: Codec>(
     Ok(())
 }
 
-fn index_documents<C: crate::codec::Codec, D: Document>(
+fn index_documents<D: Document>(
     memory_budget: usize,
-    segment: Segment<C>,
+    segment: Segment,
     grouped_document_iterator: &mut dyn Iterator<Item = AddBatch<D>>,
-    segment_updater: &SegmentUpdater<C>,
+    segment_updater: &SegmentUpdater,
     mut delete_cursor: DeleteCursor,
 ) -> crate::Result<()> {
     let mut segment_writer = SegmentWriter::for_segment(memory_budget, segment.clone())?;
@@ -227,8 +226,8 @@ fn index_documents<C: crate::codec::Codec, D: Document>(
 }
 
 /// `doc_opstamps` is required to be non-empty.
-fn apply_deletes<C: crate::codec::Codec>(
-    segment: &Segment<C>,
+fn apply_deletes(
+    segment: &Segment,
     delete_cursor: &mut DeleteCursor,
     doc_opstamps: &[Opstamp],
 ) -> crate::Result<Option<BitSet>> {
@@ -263,7 +262,7 @@ fn apply_deletes<C: crate::codec::Codec>(
     })
 }
 
-impl<C: Codec, D: Document> IndexWriter<C, D> {
+impl<D: Document> IndexWriter<D> {
     /// Create a new index writer. Attempts to acquire a lockfile.
     ///
     /// The lockfile should be deleted on drop, but it is possible
@@ -279,7 +278,7 @@ impl<C: Codec, D: Document> IndexWriter<C, D> {
     /// If the memory arena per thread is too small or too big, returns
     /// `TantivyError::InvalidArgument`
     pub(crate) fn new(
-        index: &Index<C>,
+        index: &Index,
         options: IndexWriterOptions,
         directory_lock: DirectoryLock,
     ) -> crate::Result<Self> {
@@ -346,7 +345,7 @@ impl<C: Codec, D: Document> IndexWriter<C, D> {
     }
 
     /// Accessor to the index.
-    pub fn index(&self) -> &Index<C> {
+    pub fn index(&self) -> &Index {
         &self.index
     }
 
@@ -394,7 +393,7 @@ impl<C: Codec, D: Document> IndexWriter<C, D> {
     /// It is safe to start writing file associated with the new `Segment`.
     /// These will not be garbage collected as long as an instance object of
     /// `SegmentMeta` object associated with the new `Segment` is "alive".
-    pub fn new_segment(&self) -> Segment<C> {
+    pub fn new_segment(&self) -> Segment {
         self.index.new_segment()
     }
 
@@ -616,7 +615,7 @@ impl<C: Codec, D: Document> IndexWriter<C, D> {
     /// It is also possible to add a payload to the `commit`
     /// using this API.
     /// See [`PreparedCommit::set_payload()`].
-    pub fn prepare_commit(&mut self) -> crate::Result<PreparedCommit<'_, C, D>> {
+    pub fn prepare_commit(&mut self) -> crate::Result<PreparedCommit<'_, D>> {
         // Here, because we join all of the worker threads,
         // all of the segment update for this commit have been
         // sent.
@@ -666,7 +665,7 @@ impl<C: Codec, D: Document> IndexWriter<C, D> {
         self.prepare_commit()?.commit()
     }
 
-    pub(crate) fn segment_updater(&self) -> &SegmentUpdater<C> {
+    pub(crate) fn segment_updater(&self) -> &SegmentUpdater {
         &self.segment_updater
     }
 
@@ -805,7 +804,7 @@ impl<C: Codec, D: Document> IndexWriter<C, D> {
     }
 }
 
-impl<C: Codec, D: Document> Drop for IndexWriter<C, D> {
+impl<D: Document> Drop for IndexWriter<D> {
     fn drop(&mut self) {
         self.segment_updater.kill();
         self.drop_sender();

src/indexer/merge_index_test.rs

@@ -1,10 +1,9 @@
 #[cfg(test)]
 mod tests {
-    use crate::codec::StandardCodec;
     use crate::collector::TopDocs;
     use crate::fastfield::AliveBitSet;
     use crate::index::Index;
-    use crate::postings::{DocFreq, Postings};
+    use crate::postings::Postings;
     use crate::query::QueryParser;
     use crate::schema::{
         self, BytesOptions, Facet, FacetOptions, IndexRecordOption, NumericOptions,
@@ -122,26 +121,21 @@ mod tests {
             let my_text_field = index.schema().get_field("text_field").unwrap();
             let term_a = Term::from_field_text(my_text_field, "text");
             let inverted_index = segment_reader.inverted_index(my_text_field).unwrap();
-            let term_info = inverted_index.get_term_info(&term_a).unwrap().unwrap();
             let mut postings = inverted_index
-                .read_postings_from_terminfo_specialized(
-                    &term_info,
-                    IndexRecordOption::WithFreqsAndPositions,
-                    &StandardCodec,
-                )
+                .read_postings(&term_a, IndexRecordOption::WithFreqsAndPositions)
+                .unwrap()
                 .unwrap();
-            assert_eq!(postings.doc_freq(), DocFreq::Exact(2));
+            assert_eq!(postings.doc_freq(), 2);
             let fallback_bitset = AliveBitSet::for_test_from_deleted_docs(&[0], 100);
             assert_eq!(
-                crate::indexer::merger::doc_freq_given_deletes(
-                    &postings,
+                postings.doc_freq_given_deletes(
                     segment_reader.alive_bitset().unwrap_or(&fallback_bitset)
                 ),
                 2
             );
 
             assert_eq!(postings.term_freq(), 1);
-            let mut output = Vec::new();
+            let mut output = vec![];
             postings.positions(&mut output);
             assert_eq!(output, vec![1]);
             postings.advance();

src/indexer/merger.rs

@@ -7,8 +7,6 @@ use common::ReadOnlyBitSet;
 use itertools::Itertools;
 use measure_time::debug_time;
 
-use crate::codec::postings::PostingsCodec;
-use crate::codec::{Codec, StandardCodec};
 use crate::directory::WritePtr;
 use crate::docset::{DocSet, TERMINATED};
 use crate::error::DataCorruption;
@@ -17,7 +15,7 @@ use crate::fieldnorm::{FieldNormReader, FieldNormReaders, FieldNormsSerializer,
 use crate::index::{Segment, SegmentComponent, SegmentReader};
 use crate::indexer::doc_id_mapping::{MappingType, SegmentDocIdMapping};
 use crate::indexer::SegmentSerializer;
-use crate::postings::{InvertedIndexSerializer, Postings};
+use crate::postings::{InvertedIndexSerializer, Postings, SegmentPostings};
 use crate::schema::{value_type_to_column_type, Field, FieldType, Schema};
 use crate::store::StoreWriter;
 use crate::termdict::{TermMerger, TermOrdinal};
@@ -78,11 +76,10 @@ fn estimate_total_num_tokens(readers: &[SegmentReader], field: Field) -> crate::
     Ok(total_num_tokens)
 }
 
-pub struct IndexMerger<C: Codec = StandardCodec> {
+pub struct IndexMerger {
     schema: Schema,
     pub(crate) readers: Vec<SegmentReader>,
     max_doc: u32,
-    codec: C,
 }
 
 struct DeltaComputer {
@@ -147,8 +144,8 @@ fn extract_fast_field_required_columns(schema: &Schema) -> Vec<(String, ColumnTy
         .collect()
 }
 
-impl<C: Codec> IndexMerger<C> {
-    pub fn open(schema: Schema, segments: &[Segment<C>]) -> crate::Result<IndexMerger<C>> {
+impl IndexMerger {
+    pub fn open(schema: Schema, segments: &[Segment]) -> crate::Result<IndexMerger> {
         let alive_bitset = segments.iter().map(|_| None).collect_vec();
         Self::open_with_custom_alive_set(schema, segments, alive_bitset)
     }
@@ -165,15 +162,11 @@ impl<C: Codec> IndexMerger<C> {
     // This can be used to merge but also apply an additional filter.
     // One use case is demux, which is basically taking a list of
     // segments and partitions them e.g. by a value in a field.
-    //
-    // # Panics if segments is empty.
     pub fn open_with_custom_alive_set(
         schema: Schema,
-        segments: &[Segment<C>],
+        segments: &[Segment],
         alive_bitset_opt: Vec<Option<AliveBitSet>>,
-    ) -> crate::Result<IndexMerger<C>> {
-        assert!(!segments.is_empty());
-        let codec = segments[0].index().codec().clone();
+    ) -> crate::Result<IndexMerger> {
         let mut readers = vec![];
         for (segment, new_alive_bitset_opt) in segments.iter().zip(alive_bitset_opt) {
             if segment.meta().num_docs() > 0 {
@@ -196,7 +189,6 @@ impl<C: Codec> IndexMerger<C> {
             schema,
             readers,
             max_doc,
-            codec,
         })
     }
 
@@ -295,7 +287,7 @@ impl<C: Codec> IndexMerger<C> {
         &self,
         indexed_field: Field,
         _field_type: &FieldType,
-        serializer: &mut InvertedIndexSerializer<C>,
+        serializer: &mut InvertedIndexSerializer,
         fieldnorm_reader: Option<FieldNormReader>,
         doc_id_mapping: &SegmentDocIdMapping,
     ) -> crate::Result<()> {
@@ -363,10 +355,7 @@ impl<C: Codec> IndexMerger<C> {
                          indexed. Have you modified the schema?",
         );
 
-        let mut segment_postings_containing_the_term: Vec<(
-            usize,
-            <C::PostingsCodec as PostingsCodec>::Postings,
-        )> = Vec::with_capacity(self.readers.len());
+        let mut segment_postings_containing_the_term: Vec<(usize, SegmentPostings)> = vec![];
 
         while merged_terms.advance() {
             segment_postings_containing_the_term.clear();
@@ -378,24 +367,17 @@ impl<C: Codec> IndexMerger<C> {
             for (segment_ord, term_info) in merged_terms.current_segment_ords_and_term_infos() {
                 let segment_reader = &self.readers[segment_ord];
                 let inverted_index: &InvertedIndexReader = &field_readers[segment_ord];
-                let postings = inverted_index.read_postings_from_terminfo_specialized(
-                    &term_info,
-                    segment_postings_option,
-                    &self.codec,
-                )?;
+                let segment_postings = inverted_index
+                    .read_postings_from_terminfo(&term_info, segment_postings_option)?;
                 let alive_bitset_opt = segment_reader.alive_bitset();
                 let doc_freq = if let Some(alive_bitset) = alive_bitset_opt {
-                    doc_freq_given_deletes(&postings, alive_bitset)
+                    segment_postings.doc_freq_given_deletes(alive_bitset)
                 } else {
-                    // We do not an exact document frequency here.
-                    match postings.doc_freq() {
-                        crate::postings::DocFreq::Approximate(_) => exact_doc_freq(&postings),
-                        crate::postings::DocFreq::Exact(doc_freq) => doc_freq,
-                    }
+                    segment_postings.doc_freq()
                 };
                 if doc_freq > 0u32 {
                     total_doc_freq += doc_freq;
-                    segment_postings_containing_the_term.push((segment_ord, postings));
+                    segment_postings_containing_the_term.push((segment_ord, segment_postings));
                 }
             }
 
@@ -413,7 +395,11 @@ impl<C: Codec> IndexMerger<C> {
             assert!(!segment_postings_containing_the_term.is_empty());
 
             let has_term_freq = {
-                let has_term_freq = segment_postings_containing_the_term[0].1.has_freq();
+                let has_term_freq = !segment_postings_containing_the_term[0]
+                    .1
+                    .block_cursor
+                    .freqs()
+                    .is_empty();
                 for (_, postings) in &segment_postings_containing_the_term[1..] {
                     // This may look at a strange way to test whether we have term freq or not.
                     // With JSON object, the schema is not sufficient to know whether a term
@@ -429,7 +415,7 @@ impl<C: Codec> IndexMerger<C> {
                     //
                     // Overall the reliable way to know if we have actual frequencies loaded or not
                     // is to check whether the actual decoded array is empty or not.
-                    if postings.has_freq() != has_term_freq {
+                    if has_term_freq == postings.block_cursor.freqs().is_empty() {
                         return Err(DataCorruption::comment_only(
                             "Term freqs are inconsistent across segments",
                         )
@@ -481,7 +467,7 @@ impl<C: Codec> IndexMerger<C> {
 
     fn write_postings(
         &self,
-        serializer: &mut InvertedIndexSerializer<C>,
+        serializer: &mut InvertedIndexSerializer,
         fieldnorm_readers: FieldNormReaders,
         doc_id_mapping: &SegmentDocIdMapping,
     ) -> crate::Result<()> {
@@ -539,7 +525,7 @@ impl<C: Codec> IndexMerger<C> {
     ///
     /// # Returns
     /// The number of documents in the resulting segment.
-    pub fn write(&self, mut serializer: SegmentSerializer<C>) -> crate::Result<u32> {
+    pub fn write(&self, mut serializer: SegmentSerializer) -> crate::Result<u32> {
         let doc_id_mapping = self.get_doc_id_from_concatenated_data()?;
         debug!("write-fieldnorms");
         if let Some(fieldnorms_serializer) = serializer.extract_fieldnorms_serializer() {
@@ -567,43 +553,6 @@ impl<C: Codec> IndexMerger<C> {
     }
 }
 
-/// Compute the number of non-deleted documents.
-///
-/// This method will clone and scan through the posting lists.
-/// (this is a rather expensive operation).
-pub(crate) fn doc_freq_given_deletes<P: Postings + Clone>(
-    postings: &P,
-    alive_bitset: &AliveBitSet,
-) -> u32 {
-    let mut docset = postings.clone();
-    let mut doc_freq = 0;
-    loop {
-        let doc = docset.doc();
-        if doc == TERMINATED {
-            return doc_freq;
-        }
-        if alive_bitset.is_alive(doc) {
-            doc_freq += 1u32;
-        }
-        docset.advance();
-    }
-}
-
-/// If the postings is not able to inform us of the document frequency,
-/// we just scan through it.
-pub(crate) fn exact_doc_freq<P: Postings + Clone>(postings: &P) -> u32 {
-    let mut docset = postings.clone();
-    let mut doc_freq = 0;
-    loop {
-        let doc = docset.doc();
-        if doc == TERMINATED {
-            return doc_freq;
-        }
-        doc_freq += 1u32;
-        docset.advance();
-    }
-}
-
 #[cfg(test)]
 mod tests {
 
@@ -612,16 +561,12 @@ mod tests {
     use proptest::strategy::Strategy;
     use schema::FAST;
 
-    use crate::codec::postings::PostingsCodec;
-    use crate::codec::standard::postings::StandardPostingsCodec;
     use crate::collector::tests::{
         BytesFastFieldTestCollector, FastFieldTestCollector, TEST_COLLECTOR_WITH_SCORE,
     };
     use crate::collector::{Count, FacetCollector};
-    use crate::fastfield::AliveBitSet;
     use crate::index::{Index, SegmentId};
     use crate::indexer::NoMergePolicy;
-    use crate::postings::{DocFreq, Postings as _};
     use crate::query::{AllQuery, BooleanQuery, EnableScoring, Scorer, TermQuery};
     use crate::schema::{
         Facet, FacetOptions, IndexRecordOption, NumericOptions, TantivyDocument, Term,
@@ -1573,10 +1518,10 @@ mod tests {
         let searcher = reader.searcher();
         let mut term_scorer = term_query
             .specialized_weight(EnableScoring::enabled_from_searcher(&searcher))?
-            .term_scorer_for_test(searcher.segment_reader(0u32), 1.0)
+            .term_scorer_for_test(searcher.segment_reader(0u32), 1.0)?
             .unwrap();
         assert_eq!(term_scorer.doc(), 0);
-        assert_nearly_equals!(term_scorer.seek_block_max(0), 0.0079681855);
+        assert_nearly_equals!(term_scorer.block_max_score(), 0.0079681855);
         assert_nearly_equals!(term_scorer.score(), 0.0079681855);
         for _ in 0..81 {
             writer.add_document(doc!(text=>"hello happy tax payer"))?;
@@ -1589,13 +1534,13 @@ mod tests {
         for segment_reader in searcher.segment_readers() {
             let mut term_scorer = term_query
                 .specialized_weight(EnableScoring::enabled_from_searcher(&searcher))?
-                .term_scorer_for_test(segment_reader, 1.0)
+                .term_scorer_for_test(segment_reader, 1.0)?
                 .unwrap();
             // the difference compared to before is intrinsic to the bm25 formula. no worries
             // there.
             for doc in segment_reader.doc_ids_alive() {
                 assert_eq!(term_scorer.doc(), doc);
-                assert_nearly_equals!(term_scorer.seek_block_max(doc), 0.003478312);
+                assert_nearly_equals!(term_scorer.block_max_score(), 0.003478312);
                 assert_nearly_equals!(term_scorer.score(), 0.003478312);
                 term_scorer.advance();
             }
@@ -1615,12 +1560,12 @@ mod tests {
         let segment_reader = searcher.segment_reader(0u32);
         let mut term_scorer = term_query
             .specialized_weight(EnableScoring::enabled_from_searcher(&searcher))?
-            .term_scorer_for_test(segment_reader, 1.0)
+            .term_scorer_for_test(segment_reader, 1.0)?
             .unwrap();
         // the difference compared to before is intrinsic to the bm25 formula. no worries there.
         for doc in segment_reader.doc_ids_alive() {
             assert_eq!(term_scorer.doc(), doc);
-            assert_nearly_equals!(term_scorer.seek_block_max(doc), 0.003478312);
+            assert_nearly_equals!(term_scorer.block_max_score(), 0.003478312);
             assert_nearly_equals!(term_scorer.score(), 0.003478312);
             term_scorer.advance();
         }
@@ -1634,16 +1579,4 @@ mod tests {
         assert!(((super::MAX_DOC_LIMIT - 1) as i32) >= 0);
         assert!((super::MAX_DOC_LIMIT as i32) < 0);
     }
-
-    #[test]
-    fn test_doc_freq_given_delete() {
-        let docs =
-            <StandardPostingsCodec as PostingsCodec>::Postings::create_from_docs(&[0, 2, 10]);
-        assert_eq!(docs.doc_freq(), DocFreq::Exact(3));
-        let alive_bitset = AliveBitSet::for_test_from_deleted_docs(&[2], 12);
-        assert_eq!(super::doc_freq_given_deletes(&docs, &alive_bitset), 2);
-        let all_deleted =
-            AliveBitSet::for_test_from_deleted_docs(&[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], 12);
-        assert_eq!(super::doc_freq_given_deletes(&docs, &all_deleted), 0);
-    }
 }

src/indexer/prepared_commit.rs

@@ -1,17 +1,16 @@
 use super::IndexWriter;
-use crate::codec::Codec;
 use crate::schema::document::Document;
 use crate::{FutureResult, Opstamp, TantivyDocument};
 
 /// A prepared commit
-pub struct PreparedCommit<'a, C: Codec, D: Document = TantivyDocument> {
-    index_writer: &'a mut IndexWriter<C, D>,
+pub struct PreparedCommit<'a, D: Document = TantivyDocument> {
+    index_writer: &'a mut IndexWriter<D>,
     payload: Option<String>,
     opstamp: Opstamp,
 }
 
-impl<'a, C: Codec, D: Document> PreparedCommit<'a, C, D> {
-    pub(crate) fn new(index_writer: &'a mut IndexWriter<C, D>, opstamp: Opstamp) -> Self {
+impl<'a, D: Document> PreparedCommit<'a, D> {
+    pub(crate) fn new(index_writer: &'a mut IndexWriter<D>, opstamp: Opstamp) -> Self {
         Self {
             index_writer,
             payload: None,

src/indexer/segment_serializer.rs

@@ -8,17 +8,17 @@ use crate::store::StoreWriter;
 
 /// Segment serializer is in charge of laying out on disk
 /// the data accumulated and sorted by the `SegmentWriter`.
-pub struct SegmentSerializer<C: crate::codec::Codec> {
-    segment: Segment<C>,
+pub struct SegmentSerializer {
+    segment: Segment,
     pub(crate) store_writer: StoreWriter,
     fast_field_write: WritePtr,
     fieldnorms_serializer: Option<FieldNormsSerializer>,
-    postings_serializer: InvertedIndexSerializer<C>,
+    postings_serializer: InvertedIndexSerializer,
 }
 
-impl<C: crate::codec::Codec> SegmentSerializer<C> {
+impl SegmentSerializer {
     /// Creates a new `SegmentSerializer`.
-    pub fn for_segment(mut segment: Segment<C>) -> crate::Result<SegmentSerializer<C>> {
+    pub fn for_segment(mut segment: Segment) -> crate::Result<SegmentSerializer> {
         let settings = segment.index().settings().clone();
         let store_writer = {
             let store_write = segment.open_write(SegmentComponent::Store)?;
@@ -50,12 +50,12 @@ impl<C: crate::codec::Codec> SegmentSerializer<C> {
         self.store_writer.mem_usage()
     }
 
-    pub fn segment(&self) -> &Segment<C> {
+    pub fn segment(&self) -> &Segment {
         &self.segment
     }
 
     /// Accessor to the `PostingsSerializer`.
-    pub fn get_postings_serializer(&mut self) -> &mut InvertedIndexSerializer<C> {
+    pub fn get_postings_serializer(&mut self) -> &mut InvertedIndexSerializer {
         &mut self.postings_serializer
     }
 

src/indexer/segment_updater.rs

@@ -10,13 +10,10 @@ use std::sync::{Arc, RwLock};
 use rayon::{ThreadPool, ThreadPoolBuilder};
 
 use super::segment_manager::SegmentManager;
-use crate::codec::Codec;
 use crate::core::META_FILEPATH;
 use crate::directory::{Directory, DirectoryClone, GarbageCollectionResult};
 use crate::fastfield::AliveBitSet;
-use crate::index::{
-    CodecConfiguration, Index, IndexMeta, IndexSettings, Segment, SegmentId, SegmentMeta,
-};
+use crate::index::{Index, IndexMeta, IndexSettings, Segment, SegmentId, SegmentMeta};
 use crate::indexer::delete_queue::DeleteCursor;
 use crate::indexer::index_writer::advance_deletes;
 use crate::indexer::merge_operation::MergeOperationInventory;
@@ -64,10 +61,10 @@ pub(crate) fn save_metas(metas: &IndexMeta, directory: &dyn Directory) -> crate:
 // We voluntarily pass a merge_operation ref to guarantee that
 // the merge_operation is alive during the process
 #[derive(Clone)]
-pub(crate) struct SegmentUpdater<C: Codec>(Arc<InnerSegmentUpdater<C>>);
+pub(crate) struct SegmentUpdater(Arc<InnerSegmentUpdater>);
 
-impl<C: Codec> Deref for SegmentUpdater<C> {
-    type Target = InnerSegmentUpdater<C>;
+impl Deref for SegmentUpdater {
+    type Target = InnerSegmentUpdater;
 
     #[inline]
     fn deref(&self) -> &Self::Target {
@@ -75,8 +72,8 @@ impl<C: Codec> Deref for SegmentUpdater<C> {
     }
 }
 
-fn garbage_collect_files<C: Codec>(
-    segment_updater: SegmentUpdater<C>,
+fn garbage_collect_files(
+    segment_updater: SegmentUpdater,
 ) -> crate::Result<GarbageCollectionResult> {
     info!("Running garbage collection");
     let mut index = segment_updater.index.clone();
@@ -87,8 +84,8 @@ fn garbage_collect_files<C: Codec>(
 
 /// Merges a list of segments the list of segment givens in the `segment_entries`.
 /// This function happens in the calling thread and is computationally expensive.
-fn merge<Codec: crate::codec::Codec>(
-    index: &Index<Codec>,
+fn merge(
+    index: &Index,
     mut segment_entries: Vec<SegmentEntry>,
     target_opstamp: Opstamp,
 ) -> crate::Result<Option<SegmentEntry>> {
@@ -111,13 +108,13 @@ fn merge<Codec: crate::codec::Codec>(
 
     let delete_cursor = segment_entries[0].delete_cursor().clone();
 
-    let segments: Vec<Segment<Codec>> = segment_entries
+    let segments: Vec<Segment> = segment_entries
         .iter()
         .map(|segment_entry| index.segment(segment_entry.meta().clone()))
         .collect();
 
     // An IndexMerger is like a "view" of our merged segments.
-    let merger: IndexMerger<Codec> = IndexMerger::open(index.schema(), &segments[..])?;
+    let merger: IndexMerger = IndexMerger::open(index.schema(), &segments[..])?;
 
     // ... we just serialize this index merger in our new segment to merge the segments.
     let segment_serializer = SegmentSerializer::for_segment(merged_segment.clone())?;
@@ -142,10 +139,10 @@ fn merge<Codec: crate::codec::Codec>(
 /// meant to work if you have an `IndexWriter` running for the origin indices, or
 /// the destination `Index`.
 #[doc(hidden)]
-pub fn merge_indices<Codec: crate::codec::Codec>(
-    indices: &[Index<Codec>],
-    output_directory: Box<dyn Directory>,
-) -> crate::Result<Index<Codec>> {
+pub fn merge_indices<T: Into<Box<dyn Directory>>>(
+    indices: &[Index],
+    output_directory: T,
+) -> crate::Result<Index> {
     if indices.is_empty() {
         // If there are no indices to merge, there is no need to do anything.
         return Err(crate::TantivyError::InvalidArgument(
@@ -166,7 +163,7 @@ pub fn merge_indices<Codec: crate::codec::Codec>(
         ));
     }
 
-    let mut segments: Vec<Segment<Codec>> = Vec::new();
+    let mut segments: Vec<Segment> = Vec::new();
     for index in indices {
         segments.extend(index.searchable_segments()?);
     }
@@ -188,12 +185,12 @@ pub fn merge_indices<Codec: crate::codec::Codec>(
 /// meant to work if you have an `IndexWriter` running for the origin indices, or
 /// the destination `Index`.
 #[doc(hidden)]
-pub fn merge_filtered_segments<C: crate::codec::Codec, T: Into<Box<dyn Directory>>>(
-    segments: &[Segment<C>],
+pub fn merge_filtered_segments<T: Into<Box<dyn Directory>>>(
+    segments: &[Segment],
     target_settings: IndexSettings,
     filter_doc_ids: Vec<Option<AliveBitSet>>,
     output_directory: T,
-) -> crate::Result<Index<C>> {
+) -> crate::Result<Index> {
     if segments.is_empty() {
         // If there are no indices to merge, there is no need to do anything.
         return Err(crate::TantivyError::InvalidArgument(
@@ -214,15 +211,14 @@ pub fn merge_filtered_segments<C: crate::codec::Codec, T: Into<Box<dyn Directory
         ));
     }
 
-    let mut merged_index: Index<C> = Index::builder()
-        .schema(target_schema.clone())
-        .codec(segments[0].index().codec().clone())
-        .settings(target_settings.clone())
-        .create(output_directory.into())?;
-
+    let mut merged_index = Index::create(
+        output_directory,
+        target_schema.clone(),
+        target_settings.clone(),
+    )?;
     let merged_segment = merged_index.new_segment();
     let merged_segment_id = merged_segment.id();
-    let merger: IndexMerger<C> =
+    let merger: IndexMerger =
         IndexMerger::open_with_custom_alive_set(merged_index.schema(), segments, filter_doc_ids)?;
     let segment_serializer = SegmentSerializer::for_segment(merged_segment)?;
     let num_docs = merger.write(segment_serializer)?;
@@ -239,7 +235,6 @@ pub fn merge_filtered_segments<C: crate::codec::Codec, T: Into<Box<dyn Directory
             ))
             .trim_end()
     );
-    let codec_configuration = CodecConfiguration::from(segments[0].index().codec());
 
     let index_meta = IndexMeta {
         index_settings: target_settings, // index_settings of all segments should be the same
@@ -247,7 +242,6 @@ pub fn merge_filtered_segments<C: crate::codec::Codec, T: Into<Box<dyn Directory
         schema: target_schema,
         opstamp: 0u64,
         payload: Some(stats),
-        codec: codec_configuration,
     };
 
     // save the meta.json
@@ -256,7 +250,7 @@ pub fn merge_filtered_segments<C: crate::codec::Codec, T: Into<Box<dyn Directory
     Ok(merged_index)
 }
 
-pub(crate) struct InnerSegmentUpdater<C: Codec> {
+pub(crate) struct InnerSegmentUpdater {
     // we keep a copy of the current active IndexMeta to
     // avoid loading the file every time we need it in the
     // `SegmentUpdater`.
@@ -267,7 +261,7 @@ pub(crate) struct InnerSegmentUpdater<C: Codec> {
     pool: ThreadPool,
     merge_thread_pool: ThreadPool,
 
-    index: Index<C>,
+    index: Index,
     segment_manager: SegmentManager,
     merge_policy: RwLock<Arc<dyn MergePolicy>>,
     killed: AtomicBool,
@@ -275,13 +269,13 @@ pub(crate) struct InnerSegmentUpdater<C: Codec> {
     merge_operations: MergeOperationInventory,
 }
 
-impl<Codec: crate::codec::Codec> SegmentUpdater<Codec> {
+impl SegmentUpdater {
     pub fn create(
-        index: Index<Codec>,
+        index: Index,
         stamper: Stamper,
         delete_cursor: &DeleteCursor,
         num_merge_threads: usize,
-    ) -> crate::Result<Self> {
+    ) -> crate::Result<SegmentUpdater> {
         let segments = index.searchable_segment_metas()?;
         let segment_manager = SegmentManager::from_segments(segments, delete_cursor);
         let pool = ThreadPoolBuilder::new()
@@ -411,14 +405,12 @@ impl<Codec: crate::codec::Codec> SegmentUpdater<Codec> {
             // Segment 1 from disk 1, Segment 1 from disk 2, etc.
             committed_segment_metas
                 .sort_by_key(|segment_meta| std::cmp::Reverse(segment_meta.max_doc()));
-            let codec = CodecConfiguration::from(index.codec());
             let index_meta = IndexMeta {
                 index_settings: index.settings().clone(),
                 segments: committed_segment_metas,
                 schema: index.schema(),
                 opstamp,
                 payload: commit_message,
-                codec,
             };
             // TODO add context to the error.
             save_metas(&index_meta, directory.box_clone().borrow_mut())?;
@@ -452,7 +444,7 @@ impl<Codec: crate::codec::Codec> SegmentUpdater<Codec> {
         opstamp: Opstamp,
         payload: Option<String>,
     ) -> FutureResult<Opstamp> {
-        let segment_updater: SegmentUpdater<Codec> = self.clone();
+        let segment_updater: SegmentUpdater = self.clone();
         self.schedule_task(move || {
             let segment_entries = segment_updater.purge_deletes(opstamp)?;
             segment_updater.segment_manager.commit(segment_entries);
@@ -708,7 +700,6 @@ impl<Codec: crate::codec::Codec> SegmentUpdater<Codec> {
 #[cfg(test)]
 mod tests {
     use super::merge_indices;
-    use crate::codec::StandardCodec;
     use crate::collector::TopDocs;
     use crate::directory::RamDirectory;
     use crate::fastfield::AliveBitSet;
@@ -939,7 +930,7 @@ mod tests {
 
     #[test]
     fn test_merge_empty_indices_array() {
-        let merge_result = merge_indices::<StandardCodec>(&[], Box::new(RamDirectory::default()));
+        let merge_result = merge_indices(&[], RamDirectory::default());
         assert!(merge_result.is_err());
     }
 
@@ -966,10 +957,7 @@ mod tests {
         };
 
         // mismatched schema index list
-        let result = merge_indices(
-            &[first_index, second_index],
-            Box::new(RamDirectory::default()),
-        );
+        let result = merge_indices(&[first_index, second_index], RamDirectory::default());
         assert!(result.is_err());
 
         Ok(())

src/indexer/segment_writer.rs

@@ -4,7 +4,6 @@ use itertools::Itertools;
 use tokenizer_api::BoxTokenStream;
 
 use super::operation::AddOperation;
-use crate::codec::Codec;
 use crate::fastfield::FastFieldsWriter;
 use crate::fieldnorm::{FieldNormReaders, FieldNormsWriter};
 use crate::index::{Segment, SegmentComponent};
@@ -13,7 +12,7 @@ use crate::indexer::segment_serializer::SegmentSerializer;
 use crate::json_utils::{index_json_value, IndexingPositionsPerPath};
 use crate::postings::{
     compute_table_memory_size, serialize_postings, IndexingContext, IndexingPosition,
-    PerFieldPostingsWriter, PostingsWriter, PostingsWriterEnum,
+    PerFieldPostingsWriter, PostingsWriter,
 };
 use crate::schema::document::{Document, Value};
 use crate::schema::{FieldEntry, FieldType, Schema, DATE_TIME_PRECISION_INDEXED};
@@ -46,11 +45,11 @@ fn compute_initial_table_size(per_thread_memory_budget: usize) -> crate::Result<
 ///
 /// They creates the postings list in anonymous memory.
 /// The segment is laid on disk when the segment gets `finalized`.
-pub struct SegmentWriter<Codec: crate::codec::Codec> {
+pub struct SegmentWriter {
     pub(crate) max_doc: DocId,
     pub(crate) ctx: IndexingContext,
     pub(crate) per_field_postings_writers: PerFieldPostingsWriter,
-    pub(crate) segment_serializer: SegmentSerializer<Codec>,
+    pub(crate) segment_serializer: SegmentSerializer,
     pub(crate) fast_field_writers: FastFieldsWriter,
     pub(crate) fieldnorms_writer: FieldNormsWriter,
     pub(crate) json_path_writer: JsonPathWriter,
@@ -61,7 +60,7 @@ pub struct SegmentWriter<Codec: crate::codec::Codec> {
     schema: Schema,
 }
 
-impl<Codec: crate::codec::Codec> SegmentWriter<Codec> {
+impl SegmentWriter {
     /// Creates a new `SegmentWriter`
     ///
     /// The arguments are defined as follows
@@ -71,10 +70,7 @@ impl<Codec: crate::codec::Codec> SegmentWriter<Codec> {
     ///   behavior as a memory limit.
     /// - segment: The segment being written
     /// - schema
-    pub fn for_segment(
-        memory_budget_in_bytes: usize,
-        segment: Segment<Codec>,
-    ) -> crate::Result<Self> {
+    pub fn for_segment(memory_budget_in_bytes: usize, segment: Segment) -> crate::Result<Self> {
         let schema = segment.schema();
         let tokenizer_manager = segment.index().tokenizers().clone();
         let tokenizer_manager_fast_field = segment.index().fast_field_tokenizer().clone();
@@ -173,7 +169,7 @@ impl<Codec: crate::codec::Codec> SegmentWriter<Codec> {
             }
 
             let (term_buffer, ctx) = (&mut self.term_buffer, &mut self.ctx);
-            let postings_writer: &mut PostingsWriterEnum =
+            let postings_writer: &mut dyn PostingsWriter =
                 self.per_field_postings_writers.get_for_field_mut(field);
             term_buffer.clear_with_field(field);
 
@@ -390,13 +386,13 @@ impl<Codec: crate::codec::Codec> SegmentWriter<Codec> {
 /// to the `SegmentSerializer`.
 ///
 /// `doc_id_map` is used to map to the new doc_id order.
-fn remap_and_write<C: Codec>(
+fn remap_and_write(
     schema: Schema,
     per_field_postings_writers: &PerFieldPostingsWriter,
     ctx: IndexingContext,
     fast_field_writers: FastFieldsWriter,
     fieldnorms_writer: &FieldNormsWriter,
-    mut serializer: SegmentSerializer<C>,
+    mut serializer: SegmentSerializer,
 ) -> crate::Result<()> {
     debug!("remap-and-write");
     if let Some(fieldnorms_serializer) = serializer.extract_fieldnorms_serializer() {

src/indexer/single_segment_index_writer.rs

@@ -1,7 +1,5 @@
 use std::marker::PhantomData;
 
-use crate::codec::StandardCodec;
-use crate::index::CodecConfiguration;
 use crate::indexer::operation::AddOperation;
 use crate::indexer::segment_updater::save_metas;
 use crate::indexer::SegmentWriter;
@@ -9,25 +7,22 @@ use crate::schema::document::Document;
 use crate::{Directory, Index, IndexMeta, Opstamp, Segment, TantivyDocument};
 
 #[doc(hidden)]
-pub struct SingleSegmentIndexWriter<
-    Codec: crate::codec::Codec = StandardCodec,
-    D: Document = TantivyDocument,
-> {
-    pub segment_writer: SegmentWriter<Codec>,
-    segment: Segment<Codec>,
+pub struct SingleSegmentIndexWriter<D: Document = TantivyDocument> {
+    segment_writer: SegmentWriter,
+    segment: Segment,
     opstamp: Opstamp,
-    _doc: PhantomData<D>,
+    _phantom: PhantomData<D>,
 }
 
-impl<Codec: crate::codec::Codec, D: Document> SingleSegmentIndexWriter<Codec, D> {
-    pub fn new(index: Index<Codec>, mem_budget: usize) -> crate::Result<Self> {
+impl<D: Document> SingleSegmentIndexWriter<D> {
+    pub fn new(index: Index, mem_budget: usize) -> crate::Result<Self> {
         let segment = index.new_segment();
         let segment_writer = SegmentWriter::for_segment(mem_budget, segment.clone())?;
         Ok(Self {
             segment_writer,
             segment,
             opstamp: 0,
-            _doc: PhantomData,
+            _phantom: PhantomData,
         })
     }
 
@@ -42,10 +37,10 @@ impl<Codec: crate::codec::Codec, D: Document> SingleSegmentIndexWriter<Codec, D>
             .add_document(AddOperation { opstamp, document })
     }
 
-    pub fn finalize(self) -> crate::Result<Index<Codec>> {
+    pub fn finalize(self) -> crate::Result<Index> {
         let max_doc = self.segment_writer.max_doc();
         self.segment_writer.finalize()?;
-        let segment: Segment<Codec> = self.segment.with_max_doc(max_doc);
+        let segment: Segment = self.segment.with_max_doc(max_doc);
         let index = segment.index();
         let index_meta = IndexMeta {
             index_settings: index.settings().clone(),
@@ -53,7 +48,6 @@ impl<Codec: crate::codec::Codec, D: Document> SingleSegmentIndexWriter<Codec, D>
             schema: index.schema(),
             opstamp: 0,
             payload: None,
-            codec: CodecConfiguration::from(index.codec()),
         };
         save_metas(&index_meta, index.directory())?;
         index.directory().sync_directory()?;

src/lib.rs

@@ -166,9 +166,6 @@ mod functional_test;
 
 #[macro_use]
 mod macros;
-
-/// Tantivy codecs describes how data is layed out on disk.
-pub mod codec;
 mod future_result;
 
 // Re-exports

src/postings/block_segment_postings.rs

@@ -1,19 +1,28 @@
 use std::io;
 
-use common::{OwnedBytes, VInt};
+use common::VInt;
 
-use crate::codec::standard::postings::skip::{BlockInfo, SkipReader};
-use crate::codec::standard::postings::FreqReadingOption;
+use crate::directory::{FileSlice, OwnedBytes};
 use crate::fieldnorm::FieldNormReader;
-use crate::postings::compression::{BlockDecoder, VIntDecoder as _, COMPRESSION_BLOCK_SIZE};
+use crate::postings::compression::{BlockDecoder, VIntDecoder, COMPRESSION_BLOCK_SIZE};
+use crate::postings::{BlockInfo, FreqReadingOption, SkipReader};
 use crate::query::Bm25Weight;
 use crate::schema::IndexRecordOption;
 use crate::{DocId, Score, TERMINATED};
 
+fn max_score<I: Iterator<Item = Score>>(mut it: I) -> Option<Score> {
+    it.next().map(|first| it.fold(first, Score::max))
+}
+
 /// `BlockSegmentPostings` is a cursor iterating over blocks
 /// of documents.
+///
+/// # Warning
+///
+/// While it is useful for some very specific high-performance
+/// use cases, you should prefer using `SegmentPostings` for most usage.
 #[derive(Clone)]
-pub(crate) struct BlockSegmentPostings {
+pub struct BlockSegmentPostings {
     pub(crate) doc_decoder: BlockDecoder,
     block_loaded: bool,
     freq_decoder: BlockDecoder,
@@ -79,7 +88,7 @@ fn split_into_skips_and_postings(
 }
 
 impl BlockSegmentPostings {
-    /// Opens a `StandardPostingsReader`.
+    /// Opens a `BlockSegmentPostings`.
     /// `doc_freq` is the number of documents in the posting list.
     /// `record_option` represents the amount of data available according to the schema.
     /// `requested_option` is the amount of data requested by the user.
@@ -87,10 +96,11 @@ impl BlockSegmentPostings {
     /// term frequency blocks.
     pub(crate) fn open(
         doc_freq: u32,
-        bytes: OwnedBytes,
+        data: FileSlice,
         mut record_option: IndexRecordOption,
         requested_option: IndexRecordOption,
     ) -> io::Result<BlockSegmentPostings> {
+        let bytes = data.read_bytes()?;
         let (skip_data_opt, postings_data) = split_into_skips_and_postings(doc_freq, bytes)?;
         let skip_reader = match skip_data_opt {
             Some(skip_data) => {
@@ -128,86 +138,6 @@ impl BlockSegmentPostings {
         block_segment_postings.load_block();
         Ok(block_segment_postings)
     }
-}
-
-fn max_score<I: Iterator<Item = Score>>(mut it: I) -> Option<Score> {
-    it.next().map(|first| it.fold(first, Score::max))
-}
-
-impl BlockSegmentPostings {
-    /// Returns the overall number of documents in the block postings.
-    /// It does not take in account whether documents are deleted or not.
-    ///
-    /// This `doc_freq` is simply the sum of the length of all of the blocks
-    /// length, and it does not take in account deleted documents.
-    pub fn doc_freq(&self) -> u32 {
-        self.doc_freq
-    }
-
-    /// Returns the array of docs in the current block.
-    ///
-    /// Before the first call to `.advance()`, the block
-    /// returned by `.docs()` is empty.
-    #[inline]
-    pub fn docs(&self) -> &[DocId] {
-        debug_assert!(self.block_loaded);
-        self.doc_decoder.output_array()
-    }
-
-    /// Return the document at index `idx` of the block.
-    #[inline]
-    pub fn doc(&self, idx: usize) -> u32 {
-        self.doc_decoder.output(idx)
-    }
-
-    /// Return the array of `term freq` in the block.
-    #[inline]
-    pub fn freqs(&self) -> &[u32] {
-        debug_assert!(self.block_loaded);
-        self.freq_decoder.output_array()
-    }
-
-    /// Return the frequency at index `idx` of the block.
-    #[inline]
-    pub fn freq(&self, idx: usize) -> u32 {
-        debug_assert!(self.block_loaded);
-        self.freq_decoder.output(idx)
-    }
-
-    /// Position on a block that may contains `target_doc`.
-    ///
-    /// If all docs are smaller than target, the block loaded may be empty,
-    /// or be the last an incomplete VInt block.
-    pub fn seek(&mut self, target_doc: DocId) -> usize {
-        // Move to the block that might contain our document.
-        self.seek_block_without_loading(target_doc);
-        self.load_block();
-
-        // At this point we are on the block that might contain our document.
-        let doc = self.doc_decoder.seek_within_block(target_doc);
-
-        // The last block is not full and padded with TERMINATED,
-        // so we are guaranteed to have at least one value (real or padding)
-        // that is >= target_doc.
-        debug_assert!(doc < COMPRESSION_BLOCK_SIZE);
-
-        // `doc` is now the first element >= `target_doc`.
-        // If all docs are smaller than target, the current block is incomplete and padded
-        // with TERMINATED. After the search, the cursor points to the first TERMINATED.
-        doc
-    }
-
-    pub fn position_offset(&self) -> u64 {
-        self.skip_reader.position_offset()
-    }
-
-    /// Advance to the next block.
-    pub fn advance(&mut self) {
-        self.skip_reader.advance();
-        self.block_loaded = false;
-        self.block_max_score_cache = None;
-        self.load_block();
-    }
 
     /// Returns the block_max_score for the current block.
     /// It does not require the block to be loaded. For instance, it is ok to call this method
@@ -230,7 +160,7 @@ impl BlockSegmentPostings {
         }
         // this is the last block of the segment posting list.
         // If it is actually loaded, we can compute block max manually.
-        if self.block_loaded {
+        if self.block_is_loaded() {
             let docs = self.doc_decoder.output_array().iter().cloned();
             let freqs = self.freq_decoder.output_array().iter().cloned();
             let bm25_scores = docs.zip(freqs).map(|(doc, term_freq)| {
@@ -247,25 +177,145 @@ impl BlockSegmentPostings {
         // We do not cache it however, so that it gets computed when once block is loaded.
         bm25_weight.max_score()
     }
-}
 
-impl BlockSegmentPostings {
-    /// Returns an empty segment postings object
-    pub fn empty() -> BlockSegmentPostings {
-        BlockSegmentPostings {
-            doc_decoder: BlockDecoder::with_val(TERMINATED),
-            block_loaded: true,
-            freq_decoder: BlockDecoder::with_val(1),
-            freq_reading_option: FreqReadingOption::NoFreq,
-            block_max_score_cache: None,
-            doc_freq: 0,
-            data: OwnedBytes::empty(),
-            skip_reader: SkipReader::new(OwnedBytes::empty(), 0, IndexRecordOption::Basic),
-        }
+    pub(crate) fn freq_reading_option(&self) -> FreqReadingOption {
+        self.freq_reading_option
     }
 
-    pub(crate) fn skip_reader(&self) -> &SkipReader {
-        &self.skip_reader
+    // Resets the block segment postings on another position
+    // in the postings file.
+    //
+    // This is useful for enumerating through a list of terms,
+    // and consuming the associated posting lists while avoiding
+    // reallocating a `BlockSegmentPostings`.
+    //
+    // # Warning
+    //
+    // This does not reset the positions list.
+    pub(crate) fn reset(&mut self, doc_freq: u32, postings_data: OwnedBytes) -> io::Result<()> {
+        let (skip_data_opt, postings_data) =
+            split_into_skips_and_postings(doc_freq, postings_data)?;
+        self.data = postings_data;
+        self.block_max_score_cache = None;
+        self.block_loaded = false;
+        if let Some(skip_data) = skip_data_opt {
+            self.skip_reader.reset(skip_data, doc_freq);
+        } else {
+            self.skip_reader.reset(OwnedBytes::empty(), doc_freq);
+        }
+        self.doc_freq = doc_freq;
+        self.load_block();
+        Ok(())
+    }
+
+    /// Returns the overall number of documents in the block postings.
+    /// It does not take in account whether documents are deleted or not.
+    ///
+    /// This `doc_freq` is simply the sum of the length of all of the blocks
+    /// length, and it does not take in account deleted documents.
+    pub fn doc_freq(&self) -> u32 {
+        self.doc_freq
+    }
+
+    /// Returns the array of docs in the current block.
+    ///
+    /// Before the first call to `.advance()`, the block
+    /// returned by `.docs()` is empty.
+    #[inline]
+    pub fn docs(&self) -> &[DocId] {
+        debug_assert!(self.block_is_loaded());
+        self.doc_decoder.output_array()
+    }
+
+    /// Return the document at index `idx` of the block.
+    #[inline]
+    pub fn doc(&self, idx: usize) -> u32 {
+        self.doc_decoder.output(idx)
+    }
+
+    /// Return the array of `term freq` in the block.
+    #[inline]
+    pub fn freqs(&self) -> &[u32] {
+        debug_assert!(self.block_is_loaded());
+        self.freq_decoder.output_array()
+    }
+
+    /// Return the frequency at index `idx` of the block.
+    #[inline]
+    pub fn freq(&self, idx: usize) -> u32 {
+        debug_assert!(self.block_is_loaded());
+        self.freq_decoder.output(idx)
+    }
+
+    /// Returns the length of the current block.
+    ///
+    /// Returns the decoded term-frequency buffer for the current block.
+    #[inline]
+    pub(crate) fn freq_output_array(&self) -> &[u32] {
+        self.freq_decoder.output_array()
+    }
+
+    /// All blocks have a length of `NUM_DOCS_PER_BLOCK`,
+    /// except the last block that may have a length
+    /// of any number between 1 and `NUM_DOCS_PER_BLOCK - 1`
+    #[inline]
+    pub fn block_len(&self) -> usize {
+        debug_assert!(self.block_is_loaded());
+        self.doc_decoder.output_len
+    }
+
+    /// Position on a block that may contains `target_doc`.
+    ///
+    /// If all docs are smaller than target, the block loaded may be empty,
+    /// or be the last an incomplete VInt block.
+    pub fn seek(&mut self, target_doc: DocId) -> usize {
+        // Move to the block that might contain our document.
+        self.seek_block(target_doc);
+        self.load_block();
+
+        // At this point we are on the block that might contain our document.
+        let doc = self.doc_decoder.seek_within_block(target_doc);
+
+        // The last block is not full and padded with TERMINATED,
+        // so we are guaranteed to have at least one value (real or padding)
+        // that is >= target_doc.
+        debug_assert!(doc < COMPRESSION_BLOCK_SIZE);
+
+        // `doc` is now the first element >= `target_doc`.
+        // If all docs are smaller than target, the current block is incomplete and padded
+        // with TERMINATED. After the search, the cursor points to the first TERMINATED.
+        doc
+    }
+
+    /// Returns the number of documents with a doc id strictly smaller than `target`
+    /// (i.e. the *rank* of `target` in this posting list).
+    ///
+    /// This jumps to the block that may contain `target` through the skip list, so no
+    /// skipped block is decoded; a single block is then decoded to locate `target`
+    /// within it. The cost is therefore `O(number_of_skip_list_entries)` plus one block
+    /// decode, rather than `O(doc_freq)`.
+    ///
+    /// Like [`Self::seek`], the underlying cursor only ever moves forward. This method
+    /// must be called with **non-decreasing** `target` values (galloping); calling it
+    /// with a `target` smaller than a previous one yields an incorrect result. `target`
+    /// must be a valid doc id (i.e. `target <= TERMINATED`), exactly as for `seek`.
+    ///
+    /// Edge cases: returns `0` when `target` is smaller than every doc id, and
+    /// `doc_freq()` when `target` is larger than every doc id.
+    pub fn rank(&mut self, target: DocId) -> u32 {
+        if self.doc_freq == 0 {
+            return 0;
+        }
+        // `within` = number of docs in the landed block with a doc id < target.
+        let within = self.seek(target);
+        // `remaining_docs` counts the landed block and everything after it, so the
+        // difference is the number of docs in all blocks strictly before it.
+        let docs_before_block = self.doc_freq - self.skip_reader.remaining_docs();
+        docs_before_block + within as u32
+    }
+
+    pub(crate) fn position_offset(&self) -> u64 {
+        self.skip_reader.position_offset()
     }
 
     /// Dangerous API! This calls seeks the next block on the skip list,
@@ -274,15 +324,24 @@ impl BlockSegmentPostings {
     /// `.load_block()` needs to be called manually afterwards.
     /// If all docs are smaller than target, the block loaded may be empty,
     /// or be the last an incomplete VInt block.
-    pub(crate) fn seek_block_without_loading(&mut self, target_doc: DocId) {
+    pub(crate) fn seek_block(&mut self, target_doc: DocId) {
         if self.skip_reader.seek(target_doc) {
             self.block_max_score_cache = None;
             self.block_loaded = false;
         }
     }
 
+    #[inline]
+    pub(crate) fn has_remaining_docs(&self) -> bool {
+        self.skip_reader.has_remaining_docs()
+    }
+
+    pub(crate) fn block_is_loaded(&self) -> bool {
+        self.block_loaded
+    }
+
     pub(crate) fn load_block(&mut self) {
-        if self.block_loaded {
+        if self.block_is_loaded() {
             return;
         }
         let offset = self.skip_reader.byte_offset();
@@ -330,40 +389,68 @@ impl BlockSegmentPostings {
         }
         self.block_loaded = true;
     }
+
+    /// Advance to the next block.
+    pub fn advance(&mut self) {
+        self.skip_reader.advance();
+        self.block_loaded = false;
+        self.block_max_score_cache = None;
+        self.load_block();
+    }
+
+    /// Returns an empty segment postings object
+    pub fn empty() -> BlockSegmentPostings {
+        BlockSegmentPostings {
+            doc_decoder: BlockDecoder::with_val(TERMINATED),
+            block_loaded: true,
+            freq_decoder: BlockDecoder::with_val(1),
+            freq_reading_option: FreqReadingOption::NoFreq,
+            block_max_score_cache: None,
+            doc_freq: 0,
+            data: OwnedBytes::empty(),
+            skip_reader: SkipReader::new(OwnedBytes::empty(), 0, IndexRecordOption::Basic),
+        }
+    }
+
+    pub(crate) fn skip_reader(&self) -> &SkipReader {
+        &self.skip_reader
+    }
 }
 
 #[cfg(test)]
 mod tests {
-    use common::OwnedBytes;
+    use common::HasLen;
 
     use super::BlockSegmentPostings;
-    use crate::codec::postings::PostingsSerializer;
-    use crate::codec::standard::postings::segment_postings::SegmentPostings;
-    use crate::codec::standard::postings::StandardPostingsSerializer;
     use crate::docset::{DocSet, TERMINATED};
+    use crate::index::Index;
     use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
-    use crate::schema::IndexRecordOption;
+    use crate::postings::postings::Postings;
+    use crate::postings::SegmentPostings;
+    use crate::schema::{IndexRecordOption, Schema, Term, INDEXED};
+    use crate::DocId;
 
-    #[cfg(test)]
-    fn build_block_postings(docs: &[u32]) -> BlockSegmentPostings {
-        let doc_freq = docs.len() as u32;
-        let mut postings_serializer =
-            StandardPostingsSerializer::new(1.0f32, IndexRecordOption::Basic, None);
-        postings_serializer.new_term(docs.len() as u32, false);
-        for doc in docs {
-            postings_serializer.write_doc(*doc, 1u32);
-        }
-        let mut buffer: Vec<u8> = Vec::new();
-        postings_serializer
-            .close_term(doc_freq, &mut buffer)
-            .unwrap();
-        BlockSegmentPostings::open(
-            doc_freq,
-            OwnedBytes::new(buffer),
-            IndexRecordOption::Basic,
-            IndexRecordOption::Basic,
-        )
-        .unwrap()
+    #[test]
+    fn test_empty_segment_postings() {
+        let mut postings = SegmentPostings::empty();
+        assert_eq!(postings.doc(), TERMINATED);
+        assert_eq!(postings.advance(), TERMINATED);
+        assert_eq!(postings.advance(), TERMINATED);
+        assert_eq!(postings.doc_freq(), 0);
+        assert_eq!(postings.len(), 0);
+    }
+
+    #[test]
+    fn test_empty_postings_doc_returns_terminated() {
+        let mut postings = SegmentPostings::empty();
+        assert_eq!(postings.doc(), TERMINATED);
+        assert_eq!(postings.advance(), TERMINATED);
+    }
+
+    #[test]
+    fn test_empty_postings_doc_term_freq_returns_0() {
+        let postings = SegmentPostings::empty();
+        assert_eq!(postings.term_freq(), 1);
     }
 
     #[test]
@@ -378,7 +465,7 @@ mod tests {
 
     #[test]
     fn test_block_segment_postings() -> crate::Result<()> {
-        let mut block_segments = build_block_postings(&(0..100_000).collect::<Vec<u32>>());
+        let mut block_segments = build_block_postings(&(0..100_000).collect::<Vec<u32>>())?;
         let mut offset: u32 = 0u32;
         // checking that the `doc_freq` is correct
         assert_eq!(block_segments.doc_freq(), 100_000);
@@ -403,7 +490,7 @@ mod tests {
         doc_ids.push(129);
         doc_ids.push(130);
         {
-            let block_segments = build_block_postings(&doc_ids);
+            let block_segments = build_block_postings(&doc_ids)?;
             let mut docset = SegmentPostings::from_block_postings(block_segments, None);
             assert_eq!(docset.seek(128), 129);
             assert_eq!(docset.doc(), 129);
@@ -412,7 +499,7 @@ mod tests {
             assert_eq!(docset.advance(), TERMINATED);
         }
         {
-            let block_segments = build_block_postings(&doc_ids);
+            let block_segments = build_block_postings(&doc_ids).unwrap();
             let mut docset = SegmentPostings::from_block_postings(block_segments, None);
             assert_eq!(docset.seek(129), 129);
             assert_eq!(docset.doc(), 129);
@@ -421,7 +508,7 @@ mod tests {
             assert_eq!(docset.advance(), TERMINATED);
         }
         {
-            let block_segments = build_block_postings(&doc_ids);
+            let block_segments = build_block_postings(&doc_ids)?;
             let mut docset = SegmentPostings::from_block_postings(block_segments, None);
             assert_eq!(docset.doc(), 0);
             assert_eq!(docset.seek(131), TERMINATED);
@@ -430,13 +517,38 @@ mod tests {
         Ok(())
     }
 
+    fn build_block_postings(docs: &[DocId]) -> crate::Result<BlockSegmentPostings> {
+        let mut schema_builder = Schema::builder();
+        let int_field = schema_builder.add_u64_field("id", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        let mut last_doc = 0u32;
+        for &doc in docs {
+            for _ in last_doc..doc {
+                index_writer.add_document(doc!(int_field=>1u64))?;
+            }
+            index_writer.add_document(doc!(int_field=>0u64))?;
+            last_doc = doc + 1;
+        }
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let segment_reader = searcher.segment_reader(0);
+        let inverted_index = segment_reader.inverted_index(int_field).unwrap();
+        let term = Term::from_field_u64(int_field, 0u64);
+        let term_info = inverted_index.get_term_info(&term)?.unwrap();
+        let block_postings = inverted_index
+            .read_block_postings_from_terminfo(&term_info, IndexRecordOption::Basic)?;
+        Ok(block_postings)
+    }
+
     #[test]
     fn test_block_segment_postings_seek() -> crate::Result<()> {
-        let mut docs = Vec::new();
+        let mut docs = vec![0];
         for i in 0..1300 {
             docs.push((i * i / 100) + i);
         }
-        let mut block_postings = build_block_postings(&docs[..]);
+        let mut block_postings = build_block_postings(&docs[..])?;
         for i in &[0, 424, 10000] {
             block_postings.seek(*i);
             let docs = block_postings.docs();
@@ -447,4 +559,74 @@ mod tests {
         assert_eq!(block_postings.doc(COMPRESSION_BLOCK_SIZE - 1), TERMINATED);
         Ok(())
     }
+
+    #[test]
+    fn test_reset_block_segment_postings() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let int_field = schema_builder.add_u64_field("id", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests()?;
+        // create two postings list, one containing even number,
+        // the other containing odd numbers.
+        for i in 0..6 {
+            let doc = doc!(int_field=> (i % 2) as u64);
+            index_writer.add_document(doc)?;
+        }
+        index_writer.commit()?;
+        let searcher = index.reader()?.searcher();
+        let segment_reader = searcher.segment_reader(0);
+
+        let mut block_segments;
+        {
+            let term = Term::from_field_u64(int_field, 0u64);
+            let inverted_index = segment_reader.inverted_index(int_field)?;
+            let term_info = inverted_index.get_term_info(&term)?.unwrap();
+            block_segments = inverted_index
+                .read_block_postings_from_terminfo(&term_info, IndexRecordOption::Basic)?;
+        }
+        assert_eq!(block_segments.docs(), &[0, 2, 4]);
+        {
+            let term = Term::from_field_u64(int_field, 1u64);
+            let inverted_index = segment_reader.inverted_index(int_field)?;
+            let term_info = inverted_index.get_term_info(&term)?.unwrap();
+            inverted_index.reset_block_postings_from_terminfo(&term_info, &mut block_segments)?;
+        }
+        assert_eq!(block_segments.docs(), &[1, 3, 5]);
+        Ok(())
+    }
+
+    #[test]
+    fn test_block_segment_postings_rank() -> crate::Result<()> {
+        // ~8 blocks worth of docs so the skip list is actually exercised.
+        let docs: Vec<DocId> = (0..1000u32).map(|i| i * 3).collect();
+        let mut block_postings = build_block_postings(&docs[..])?;
+        let doc_freq = block_postings.doc_freq();
+
+        // rank(target) must equal the number of docs strictly below target.
+        // Targets are queried in non-decreasing order, as the API requires.
+        // `target` values must be a valid doc id (<= TERMINATED) and non-decreasing.
+        let targets = [
+            0u32, 1, 2, 3, 4, 299, 300, 301, 1500, 2996, 2997, 3000, 10_000,
+        ];
+        for &target in &targets {
+            let expected = docs.iter().filter(|&&d| d < target).count() as u32;
+            assert_eq!(
+                block_postings.rank(target),
+                expected,
+                "rank({target}) mismatch"
+            );
+        }
+
+        // Edge cases: below the first doc -> 0, above the last doc -> doc_freq.
+        let mut fresh = build_block_postings(&docs[..])?;
+        assert_eq!(fresh.rank(0), 0);
+        let mut fresh = build_block_postings(&docs[..])?;
+        assert_eq!(fresh.rank(1_000_000), doc_freq);
+
+        // Empty postings: rank is always 0.
+        let mut empty = BlockSegmentPostings::empty();
+        assert_eq!(empty.rank(42), 0);
+        Ok(())
+    }
 }

src/postings/json_postings_writer.rs

@@ -3,7 +3,6 @@ use std::io;
 use common::json_path_writer::JSON_END_OF_PATH;
 use stacker::Addr;
 
-use crate::codec::Codec;
 use crate::indexer::indexing_term::IndexingTerm;
 use crate::indexer::path_to_unordered_id::OrderedPathId;
 use crate::postings::postings_writer::SpecializedPostingsWriter;
@@ -23,6 +22,12 @@ pub(crate) struct JsonPostingsWriter<Rec: Recorder> {
     non_str_posting_writer: SpecializedPostingsWriter<DocIdRecorder>,
 }
 
+impl<Rec: Recorder> From<JsonPostingsWriter<Rec>> for Box<dyn PostingsWriter> {
+    fn from(json_postings_writer: JsonPostingsWriter<Rec>) -> Box<dyn PostingsWriter> {
+        Box::new(json_postings_writer)
+    }
+}
+
 impl<Rec: Recorder> PostingsWriter for JsonPostingsWriter<Rec> {
     #[inline]
     fn subscribe(
@@ -53,12 +58,12 @@ impl<Rec: Recorder> PostingsWriter for JsonPostingsWriter<Rec> {
     }
 
     /// The actual serialization format is handled by the `PostingsSerializer`.
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         ordered_term_addrs: &[(Field, OrderedPathId, &[u8], Addr)],
         ordered_id_to_path: &[&str],
         ctx: &IndexingContext,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer,
     ) -> io::Result<()> {
         let mut term_buffer = JsonTermSerializer(Vec::with_capacity(48));
         let mut buffer_lender = BufferLender::default();

src/postings/loaded_postings.rs

@@ -1,5 +1,5 @@
 use crate::docset::{DocSet, TERMINATED};
-use crate::postings::{DocFreq, Postings};
+use crate::postings::{Postings, SegmentPostings};
 use crate::DocId;
 
 /// `LoadedPostings` is a `DocSet` and `Postings` implementation.
@@ -25,16 +25,16 @@ impl LoadedPostings {
     /// Creates a new `LoadedPostings` from a `SegmentPostings`.
     ///
     /// It will also preload positions, if positions are available in the SegmentPostings.
-    pub fn load(postings: &mut Box<dyn Postings>) -> LoadedPostings {
-        let num_docs: usize = u32::from(postings.doc_freq()) as usize;
+    pub fn load(segment_postings: &mut SegmentPostings) -> LoadedPostings {
+        let num_docs = segment_postings.doc_freq() as usize;
         let mut doc_ids = Vec::with_capacity(num_docs);
         let mut positions = Vec::with_capacity(num_docs);
         let mut position_offsets = Vec::with_capacity(num_docs);
-        while postings.doc() != TERMINATED {
+        while segment_postings.doc() != TERMINATED {
             position_offsets.push(positions.len() as u32);
-            doc_ids.push(postings.doc());
-            postings.append_positions_with_offset(0, &mut positions);
-            postings.advance();
+            doc_ids.push(segment_postings.doc());
+            segment_postings.append_positions_with_offset(0, &mut positions);
+            segment_postings.advance();
         }
         position_offsets.push(positions.len() as u32);
         LoadedPostings {
@@ -101,14 +101,6 @@ impl Postings for LoadedPostings {
             output.push(*pos + offset);
         }
     }
-
-    fn has_freq(&self) -> bool {
-        true
-    }
-
-    fn doc_freq(&self) -> DocFreq {
-        DocFreq::Exact(self.doc_ids.len() as u32)
-    }
 }
 
 #[cfg(test)]

src/postings/mod.rs

@@ -4,6 +4,7 @@ mod block_search;
 
 pub(crate) use self::block_search::branchless_binary_search;
 
+mod block_segment_postings;
 pub(crate) mod compression;
 mod indexing_context;
 mod json_postings_writer;
@@ -12,23 +13,33 @@ mod per_field_postings_writer;
 mod postings;
 mod postings_writer;
 mod recorder;
+mod segment_postings;
 /// Serializer module for the inverted index
 pub mod serializer;
+mod skip;
 mod term_info;
 
 pub(crate) use loaded_postings::LoadedPostings;
-pub use postings::DocFreq;
 pub(crate) use stacker::compute_table_memory_size;
 
+pub use self::block_segment_postings::BlockSegmentPostings;
 pub(crate) use self::indexing_context::IndexingContext;
 pub(crate) use self::per_field_postings_writer::PerFieldPostingsWriter;
 pub use self::postings::Postings;
-pub(crate) use self::postings_writer::{
-    serialize_postings, IndexingPosition, PostingsWriter, PostingsWriterEnum,
-};
+pub(crate) use self::postings_writer::{serialize_postings, IndexingPosition, PostingsWriter};
+pub use self::segment_postings::SegmentPostings;
 pub use self::serializer::{FieldSerializer, InvertedIndexSerializer};
+pub(crate) use self::skip::{BlockInfo, SkipReader};
 pub use self::term_info::TermInfo;
 
+#[expect(clippy::enum_variant_names)]
+#[derive(Debug, PartialEq, Clone, Copy, Eq)]
+pub(crate) enum FreqReadingOption {
+    NoFreq,
+    SkipFreq,
+    ReadFreq,
+}
+
 #[cfg(test)]
 pub(crate) mod tests {
     use std::mem;
@@ -39,7 +50,6 @@ pub(crate) mod tests {
     use crate::index::{Index, SegmentComponent, SegmentReader};
     use crate::indexer::operation::AddOperation;
     use crate::indexer::SegmentWriter;
-    use crate::postings::DocFreq;
     use crate::query::Scorer;
     use crate::schema::{
         Field, IndexRecordOption, Schema, Term, TextFieldIndexing, TextOptions, INDEXED, TEXT,
@@ -270,11 +280,11 @@ pub(crate) mod tests {
             }
             {
                 let term_a = Term::from_field_text(text_field, "a");
-                let mut postings_a: Box<dyn Postings> = segment_reader
+                let mut postings_a = segment_reader
                     .inverted_index(term_a.field())?
                     .read_postings(&term_a, IndexRecordOption::WithFreqsAndPositions)?
                     .unwrap();
-                assert_eq!(postings_a.doc_freq(), DocFreq::Exact(1000));
+                assert_eq!(postings_a.len(), 1000);
                 assert_eq!(postings_a.doc(), 0);
                 assert_eq!(postings_a.term_freq(), 6);
                 postings_a.positions(&mut positions);
@@ -297,7 +307,7 @@ pub(crate) mod tests {
                     .inverted_index(term_e.field())?
                     .read_postings(&term_e, IndexRecordOption::WithFreqsAndPositions)?
                     .unwrap();
-                assert_eq!(postings_e.doc_freq(), DocFreq::Exact(1000 - 2));
+                assert_eq!(postings_e.len(), 1000 - 2);
                 for i in 2u32..1000u32 {
                     assert_eq!(postings_e.term_freq(), i);
                     postings_e.positions(&mut positions);

src/postings/per_field_postings_writer.rs

@@ -1,15 +1,16 @@
 use crate::postings::json_postings_writer::JsonPostingsWriter;
-use crate::postings::postings_writer::{PostingsWriterEnum, SpecializedPostingsWriter};
+use crate::postings::postings_writer::SpecializedPostingsWriter;
 use crate::postings::recorder::{DocIdRecorder, TermFrequencyRecorder, TfAndPositionRecorder};
+use crate::postings::PostingsWriter;
 use crate::schema::{Field, FieldEntry, FieldType, IndexRecordOption, Schema};
 
 pub(crate) struct PerFieldPostingsWriter {
-    per_field_postings_writers: Vec<PostingsWriterEnum>,
+    per_field_postings_writers: Vec<Box<dyn PostingsWriter>>,
 }
 
 impl PerFieldPostingsWriter {
     pub fn for_schema(schema: &Schema) -> Self {
-        let per_field_postings_writers: Vec<PostingsWriterEnum> = schema
+        let per_field_postings_writers = schema
             .fields()
             .map(|(_, field_entry)| posting_writer_from_field_entry(field_entry))
             .collect();
@@ -18,16 +19,16 @@ impl PerFieldPostingsWriter {
         }
     }
 
-    pub(crate) fn get_for_field(&self, field: Field) -> &PostingsWriterEnum {
-        &self.per_field_postings_writers[field.field_id() as usize]
+    pub(crate) fn get_for_field(&self, field: Field) -> &dyn PostingsWriter {
+        self.per_field_postings_writers[field.field_id() as usize].as_ref()
     }
 
-    pub(crate) fn get_for_field_mut(&mut self, field: Field) -> &mut PostingsWriterEnum {
-        &mut self.per_field_postings_writers[field.field_id() as usize]
+    pub(crate) fn get_for_field_mut(&mut self, field: Field) -> &mut dyn PostingsWriter {
+        self.per_field_postings_writers[field.field_id() as usize].as_mut()
     }
 }
 
-fn posting_writer_from_field_entry(field_entry: &FieldEntry) -> PostingsWriterEnum {
+fn posting_writer_from_field_entry(field_entry: &FieldEntry) -> Box<dyn PostingsWriter> {
     match *field_entry.field_type() {
         FieldType::Str(ref text_options) => text_options
             .get_indexing_options()
@@ -50,7 +51,7 @@ fn posting_writer_from_field_entry(field_entry: &FieldEntry) -> PostingsWriterEn
         | FieldType::Date(_)
         | FieldType::Bytes(_)
         | FieldType::IpAddr(_)
-        | FieldType::Facet(_) => <SpecializedPostingsWriter<DocIdRecorder>>::default().into(),
+        | FieldType::Facet(_) => Box::<SpecializedPostingsWriter<DocIdRecorder>>::default(),
         FieldType::JsonObject(ref json_object_options) => {
             if let Some(text_indexing_option) = json_object_options.get_text_indexing_options() {
                 match text_indexing_option.index_option() {

src/postings/postings.rs

@@ -1,25 +1,5 @@
 use crate::docset::DocSet;
 
-/// Result of the doc_freq method.
-///
-/// Postings can inform us that the document frequency is approximate.
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
-pub enum DocFreq {
-    /// The document frequency is approximate.
-    Approximate(u32),
-    /// The document frequency is exact.
-    Exact(u32),
-}
-
-impl From<DocFreq> for u32 {
-    fn from(doc_freq: DocFreq) -> Self {
-        match doc_freq {
-            DocFreq::Approximate(approximate_doc_freq) => approximate_doc_freq,
-            DocFreq::Exact(doc_freq) => doc_freq,
-        }
-    }
-}
-
 /// Postings (also called inverted list)
 ///
 /// For a given term, it is the list of doc ids of the doc
@@ -34,9 +14,6 @@ pub trait Postings: DocSet + 'static {
     /// The number of times the term appears in the document.
     fn term_freq(&self) -> u32;
 
-    /// Returns the number of documents containing the term in the segment.
-    fn doc_freq(&self) -> DocFreq;
-
     /// Returns the positions offsetted with a given value.
     /// It is not necessary to clear the `output` before calling this method.
     /// The output vector will be resized to the `term_freq`.
@@ -54,16 +31,6 @@ pub trait Postings: DocSet + 'static {
     fn positions(&mut self, output: &mut Vec<u32>) {
         self.positions_with_offset(0u32, output);
     }
-
-    /// Returns true if the term_frequency is available.
-    ///
-    /// This is a tricky question, because on JSON fields, it is possible
-    /// for a text term to have term freq, whereas a number term in the field has none.
-    ///
-    /// This function returns whether the actual term has term frequencies or not.
-    /// In this above JSON field example, `has_freq` should return true for the
-    /// earlier and false for the latter.
-    fn has_freq(&self) -> bool;
 }
 
 impl Postings for Box<dyn Postings> {
@@ -74,12 +41,4 @@ impl Postings for Box<dyn Postings> {
     fn append_positions_with_offset(&mut self, offset: u32, output: &mut Vec<u32>) {
         (**self).append_positions_with_offset(offset, output);
     }
-
-    fn has_freq(&self) -> bool {
-        (**self).has_freq()
-    }
-
-    fn doc_freq(&self) -> DocFreq {
-        (**self).doc_freq()
-    }
 }

src/postings/postings_writer.rs

@@ -4,14 +4,10 @@ use std::ops::Range;
 
 use stacker::Addr;
 
-use crate::codec::Codec;
 use crate::fieldnorm::FieldNormReaders;
 use crate::indexer::indexing_term::IndexingTerm;
 use crate::indexer::path_to_unordered_id::OrderedPathId;
-use crate::postings::json_postings_writer::JsonPostingsWriter;
-use crate::postings::recorder::{
-    BufferLender, DocIdRecorder, Recorder, TermFrequencyRecorder, TfAndPositionRecorder,
-};
+use crate::postings::recorder::{BufferLender, Recorder};
 use crate::postings::{
     FieldSerializer, IndexingContext, InvertedIndexSerializer, PerFieldPostingsWriter,
 };
@@ -49,12 +45,12 @@ fn make_field_partition(
 /// Serialize the inverted index.
 /// It pushes all term, one field at a time, towards the
 /// postings serializer.
-pub(crate) fn serialize_postings<C: Codec>(
+pub(crate) fn serialize_postings(
     ctx: IndexingContext,
     schema: Schema,
     per_field_postings_writers: &PerFieldPostingsWriter,
     fieldnorm_readers: FieldNormReaders,
-    serializer: &mut InvertedIndexSerializer<C>,
+    serializer: &mut InvertedIndexSerializer,
 ) -> crate::Result<()> {
     // Replace unordered ids by ordered ids to be able to sort
     let unordered_id_to_ordered_id: Vec<OrderedPathId> =
@@ -104,141 +100,6 @@ pub(crate) struct IndexingPosition {
     pub end_position: u32,
 }
 
-pub enum PostingsWriterEnum {
-    DocId(SpecializedPostingsWriter<DocIdRecorder>),
-    DocIdTf(SpecializedPostingsWriter<TermFrequencyRecorder>),
-    DocTfAndPosition(SpecializedPostingsWriter<TfAndPositionRecorder>),
-    JsonDocId(JsonPostingsWriter<DocIdRecorder>),
-    JsonDocIdTf(JsonPostingsWriter<TermFrequencyRecorder>),
-    JsonDocTfAndPosition(JsonPostingsWriter<TfAndPositionRecorder>),
-}
-
-impl From<SpecializedPostingsWriter<DocIdRecorder>> for PostingsWriterEnum {
-    fn from(doc_id_recorder_writer: SpecializedPostingsWriter<DocIdRecorder>) -> Self {
-        PostingsWriterEnum::DocId(doc_id_recorder_writer)
-    }
-}
-
-impl From<SpecializedPostingsWriter<TermFrequencyRecorder>> for PostingsWriterEnum {
-    fn from(doc_id_tf_recorder_writer: SpecializedPostingsWriter<TermFrequencyRecorder>) -> Self {
-        PostingsWriterEnum::DocIdTf(doc_id_tf_recorder_writer)
-    }
-}
-
-impl From<SpecializedPostingsWriter<TfAndPositionRecorder>> for PostingsWriterEnum {
-    fn from(
-        doc_id_tf_and_positions_recorder_writer: SpecializedPostingsWriter<TfAndPositionRecorder>,
-    ) -> Self {
-        PostingsWriterEnum::DocTfAndPosition(doc_id_tf_and_positions_recorder_writer)
-    }
-}
-
-impl From<JsonPostingsWriter<DocIdRecorder>> for PostingsWriterEnum {
-    fn from(doc_id_recorder_writer: JsonPostingsWriter<DocIdRecorder>) -> Self {
-        PostingsWriterEnum::JsonDocId(doc_id_recorder_writer)
-    }
-}
-
-impl From<JsonPostingsWriter<TermFrequencyRecorder>> for PostingsWriterEnum {
-    fn from(doc_id_tf_recorder_writer: JsonPostingsWriter<TermFrequencyRecorder>) -> Self {
-        PostingsWriterEnum::JsonDocIdTf(doc_id_tf_recorder_writer)
-    }
-}
-
-impl From<JsonPostingsWriter<TfAndPositionRecorder>> for PostingsWriterEnum {
-    fn from(
-        doc_id_tf_and_positions_recorder_writer: JsonPostingsWriter<TfAndPositionRecorder>,
-    ) -> Self {
-        PostingsWriterEnum::JsonDocTfAndPosition(doc_id_tf_and_positions_recorder_writer)
-    }
-}
-
-impl PostingsWriter for PostingsWriterEnum {
-    fn subscribe(&mut self, doc: DocId, pos: u32, term: &IndexingTerm, ctx: &mut IndexingContext) {
-        match self {
-            PostingsWriterEnum::DocId(writer) => writer.subscribe(doc, pos, term, ctx),
-            PostingsWriterEnum::DocIdTf(writer) => writer.subscribe(doc, pos, term, ctx),
-            PostingsWriterEnum::DocTfAndPosition(writer) => writer.subscribe(doc, pos, term, ctx),
-            PostingsWriterEnum::JsonDocId(writer) => writer.subscribe(doc, pos, term, ctx),
-            PostingsWriterEnum::JsonDocIdTf(writer) => writer.subscribe(doc, pos, term, ctx),
-            PostingsWriterEnum::JsonDocTfAndPosition(writer) => {
-                writer.subscribe(doc, pos, term, ctx)
-            }
-        }
-    }
-
-    fn serialize<C: Codec>(
-        &self,
-        term_addrs: &[(Field, OrderedPathId, &[u8], Addr)],
-        ordered_id_to_path: &[&str],
-        ctx: &IndexingContext,
-        serializer: &mut FieldSerializer<C>,
-    ) -> io::Result<()> {
-        match self {
-            PostingsWriterEnum::DocId(writer) => {
-                writer.serialize(term_addrs, ordered_id_to_path, ctx, serializer)
-            }
-            PostingsWriterEnum::DocIdTf(writer) => {
-                writer.serialize(term_addrs, ordered_id_to_path, ctx, serializer)
-            }
-            PostingsWriterEnum::DocTfAndPosition(writer) => {
-                writer.serialize(term_addrs, ordered_id_to_path, ctx, serializer)
-            }
-            PostingsWriterEnum::JsonDocId(writer) => {
-                writer.serialize(term_addrs, ordered_id_to_path, ctx, serializer)
-            }
-            PostingsWriterEnum::JsonDocIdTf(writer) => {
-                writer.serialize(term_addrs, ordered_id_to_path, ctx, serializer)
-            }
-            PostingsWriterEnum::JsonDocTfAndPosition(writer) => {
-                writer.serialize(term_addrs, ordered_id_to_path, ctx, serializer)
-            }
-        }
-    }
-
-    /// Tokenize a text and subscribe all of its token.
-    fn index_text(
-        &mut self,
-        doc_id: DocId,
-        token_stream: &mut dyn TokenStream,
-        term_buffer: &mut IndexingTerm,
-        ctx: &mut IndexingContext,
-        indexing_position: &mut IndexingPosition,
-    ) {
-        match self {
-            PostingsWriterEnum::DocId(writer) => {
-                writer.index_text(doc_id, token_stream, term_buffer, ctx, indexing_position)
-            }
-            PostingsWriterEnum::DocIdTf(writer) => {
-                writer.index_text(doc_id, token_stream, term_buffer, ctx, indexing_position)
-            }
-            PostingsWriterEnum::DocTfAndPosition(writer) => {
-                writer.index_text(doc_id, token_stream, term_buffer, ctx, indexing_position)
-            }
-            PostingsWriterEnum::JsonDocId(writer) => {
-                writer.index_text(doc_id, token_stream, term_buffer, ctx, indexing_position)
-            }
-            PostingsWriterEnum::JsonDocIdTf(writer) => {
-                writer.index_text(doc_id, token_stream, term_buffer, ctx, indexing_position)
-            }
-            PostingsWriterEnum::JsonDocTfAndPosition(writer) => {
-                writer.index_text(doc_id, token_stream, term_buffer, ctx, indexing_position)
-            }
-        }
-    }
-
-    fn total_num_tokens(&self) -> u64 {
-        match self {
-            PostingsWriterEnum::DocId(writer) => writer.total_num_tokens(),
-            PostingsWriterEnum::DocIdTf(writer) => writer.total_num_tokens(),
-            PostingsWriterEnum::DocTfAndPosition(writer) => writer.total_num_tokens(),
-            PostingsWriterEnum::JsonDocId(writer) => writer.total_num_tokens(),
-            PostingsWriterEnum::JsonDocIdTf(writer) => writer.total_num_tokens(),
-            PostingsWriterEnum::JsonDocTfAndPosition(writer) => writer.total_num_tokens(),
-        }
-    }
-}
-
 /// The `PostingsWriter` is in charge of receiving documenting
 /// and building a `Segment` in anonymous memory.
 ///
@@ -255,12 +116,12 @@ pub(crate) trait PostingsWriter: Send + Sync {
 
     /// Serializes the postings on disk.
     /// The actual serialization format is handled by the `PostingsSerializer`.
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         term_addrs: &[(Field, OrderedPathId, &[u8], Addr)],
         ordered_id_to_path: &[&str],
         ctx: &IndexingContext,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer,
     ) -> io::Result<()>;
 
     /// Tokenize a text and subscribe all of its token.
@@ -310,14 +171,22 @@ pub(crate) struct SpecializedPostingsWriter<Rec: Recorder> {
     _recorder_type: PhantomData<Rec>,
 }
 
+impl<Rec: Recorder> From<SpecializedPostingsWriter<Rec>> for Box<dyn PostingsWriter> {
+    fn from(
+        specialized_postings_writer: SpecializedPostingsWriter<Rec>,
+    ) -> Box<dyn PostingsWriter> {
+        Box::new(specialized_postings_writer)
+    }
+}
+
 impl<Rec: Recorder> SpecializedPostingsWriter<Rec> {
     #[inline]
-    pub(crate) fn serialize_one_term<C: Codec>(
+    pub(crate) fn serialize_one_term(
         term: &[u8],
         addr: Addr,
         buffer_lender: &mut BufferLender,
         ctx: &IndexingContext,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer,
     ) -> io::Result<()> {
         let recorder: Rec = ctx.term_index.read(addr);
         let term_doc_freq = recorder.term_doc_freq().unwrap_or(0u32);
@@ -358,12 +227,12 @@ impl<Rec: Recorder> PostingsWriter for SpecializedPostingsWriter<Rec> {
         });
     }
 
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         term_addrs: &[(Field, OrderedPathId, &[u8], Addr)],
         _ordered_id_to_path: &[&str],
         ctx: &IndexingContext,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer,
     ) -> io::Result<()> {
         let mut buffer_lender = BufferLender::default();
         for (_field, _path_id, term, addr) in term_addrs {

src/postings/recorder.rs

@@ -1,7 +1,6 @@
 use common::read_u32_vint;
 use stacker::{ExpUnrolledLinkedList, MemoryArena};
 
-use crate::codec::Codec;
 use crate::postings::FieldSerializer;
 use crate::DocId;
 
@@ -68,10 +67,10 @@ pub(crate) trait Recorder: Copy + Default + Send + Sync + 'static {
     /// Close the document. It will help record the term frequency.
     fn close_doc(&mut self, arena: &mut MemoryArena);
     /// Pushes the postings information to the serializer.
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         arena: &MemoryArena,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer<'_>,
         buffer_lender: &mut BufferLender,
     );
     /// Returns the number of document containing this term.
@@ -111,10 +110,10 @@ impl Recorder for DocIdRecorder {
     #[inline]
     fn close_doc(&mut self, _arena: &mut MemoryArena) {}
 
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         arena: &MemoryArena,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer<'_>,
         buffer_lender: &mut BufferLender,
     ) {
         let buffer = buffer_lender.lend_u8();
@@ -179,10 +178,10 @@ impl Recorder for TermFrequencyRecorder {
         self.current_tf = 0;
     }
 
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         arena: &MemoryArena,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer<'_>,
         buffer_lender: &mut BufferLender,
     ) {
         let buffer = buffer_lender.lend_u8();
@@ -236,10 +235,10 @@ impl Recorder for TfAndPositionRecorder {
         self.stack.writer(arena).write_u32_vint(POSITION_END);
     }
 
-    fn serialize<C: Codec>(
+    fn serialize(
         &self,
         arena: &MemoryArena,
-        serializer: &mut FieldSerializer<C>,
+        serializer: &mut FieldSerializer<'_>,
         buffer_lender: &mut BufferLender,
     ) {
         let (buffer_u8, buffer_positions) = buffer_lender.lend_all();

src/postings/segment_postings.rs

@@ -1,14 +1,11 @@
-use common::BitSet;
+use common::HasLen;
 
-use super::BlockSegmentPostings;
-use crate::codec::postings::PostingsWithBlockMax;
 use crate::docset::DocSet;
-use crate::fieldnorm::FieldNormReader;
+use crate::fastfield::AliveBitSet;
 use crate::positions::PositionReader;
 use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
-use crate::postings::{DocFreq, Postings};
-use crate::query::Bm25Weight;
-use crate::{DocId, Score};
+use crate::postings::{BlockSegmentPostings, Postings};
+use crate::{DocId, TERMINATED};
 
 /// `SegmentPostings` represents the inverted list or postings associated with
 /// a term in a `Segment`.
@@ -32,6 +29,31 @@ impl SegmentPostings {
         }
     }
 
+    /// Compute the number of non-deleted documents.
+    ///
+    /// This method will clone and scan through the posting lists.
+    /// (this is a rather expensive operation).
+    pub fn doc_freq_given_deletes(&self, alive_bitset: &AliveBitSet) -> u32 {
+        let mut docset = self.clone();
+        let mut doc_freq = 0;
+        loop {
+            let doc = docset.doc();
+            if doc == TERMINATED {
+                return doc_freq;
+            }
+            if alive_bitset.is_alive(doc) {
+                doc_freq += 1u32;
+            }
+            docset.advance();
+        }
+    }
+
+    /// Returns the overall number of documents in the block postings.
+    /// It does not take in account whether documents are deleted or not.
+    pub fn doc_freq(&self) -> u32 {
+        self.block_cursor.doc_freq()
+    }
+
     /// Creates a segment postings object with the given documents
     /// and no frequency encoded.
     ///
@@ -42,19 +64,13 @@ impl SegmentPostings {
     /// buffer with the serialized data.
     #[cfg(test)]
     pub fn create_from_docs(docs: &[u32]) -> SegmentPostings {
-        use common::OwnedBytes;
-
+        use crate::directory::FileSlice;
+        use crate::postings::serializer::PostingsSerializer;
         use crate::schema::IndexRecordOption;
         let mut buffer = Vec::new();
         {
-            use crate::codec::postings::PostingsSerializer;
-
             let mut postings_serializer =
-                crate::codec::standard::postings::StandardPostingsSerializer::new(
-                    0.0,
-                    IndexRecordOption::Basic,
-                    None,
-                );
+                PostingsSerializer::new(0.0, IndexRecordOption::Basic, None);
             postings_serializer.new_term(docs.len() as u32, false);
             for &doc in docs {
                 postings_serializer.write_doc(doc, 1u32);
@@ -65,7 +81,7 @@ impl SegmentPostings {
         }
         let block_segment_postings = BlockSegmentPostings::open(
             docs.len() as u32,
-            OwnedBytes::new(buffer),
+            FileSlice::from(buffer),
             IndexRecordOption::Basic,
             IndexRecordOption::Basic,
         )
@@ -79,11 +95,9 @@ impl SegmentPostings {
         doc_and_tfs: &[(u32, u32)],
         fieldnorms: Option<&[u32]>,
     ) -> SegmentPostings {
-        use common::OwnedBytes;
-
-        use crate::codec::postings::PostingsSerializer as _;
-        use crate::codec::standard::postings::StandardPostingsSerializer;
+        use crate::directory::FileSlice;
         use crate::fieldnorm::FieldNormReader;
+        use crate::postings::serializer::PostingsSerializer;
         use crate::schema::IndexRecordOption;
         use crate::Score;
         let mut buffer: Vec<u8> = Vec::new();
@@ -100,7 +114,7 @@ impl SegmentPostings {
                 total_num_tokens as Score / fieldnorms.len() as Score
             })
             .unwrap_or(0.0);
-        let mut postings_serializer = StandardPostingsSerializer::new(
+        let mut postings_serializer = PostingsSerializer::new(
             average_field_norm,
             IndexRecordOption::WithFreqs,
             fieldnorm_reader,
@@ -114,7 +128,7 @@ impl SegmentPostings {
             .unwrap();
         let block_segment_postings = BlockSegmentPostings::open(
             doc_and_tfs.len() as u32,
-            OwnedBytes::new(buffer),
+            FileSlice::from(buffer),
             IndexRecordOption::WithFreqs,
             IndexRecordOption::WithFreqs,
         )
@@ -144,6 +158,7 @@ impl DocSet for SegmentPostings {
     // next needs to be called a first time to point to the correct element.
     #[inline]
     fn advance(&mut self) -> DocId {
+        debug_assert!(self.block_cursor.block_is_loaded());
         if self.cur == COMPRESSION_BLOCK_SIZE - 1 {
             self.cur = 0;
             self.block_cursor.advance();
@@ -182,31 +197,13 @@ impl DocSet for SegmentPostings {
     }
 
     fn size_hint(&self) -> u32 {
-        self.doc_freq().into()
+        self.len() as u32
     }
+}
 
-    fn fill_bitset(&mut self, bitset: &mut BitSet) {
-        let bitset_max_value: DocId = bitset.max_value();
-        loop {
-            let docs = self.block_cursor.docs();
-            let Some(&last_doc) = docs.last() else {
-                break;
-            };
-            if last_doc < bitset_max_value {
-                // All docs are within the range of the bitset
-                for &doc in docs {
-                    bitset.insert(doc);
-                }
-            } else {
-                for &doc in docs {
-                    if doc < bitset_max_value {
-                        bitset.insert(doc);
-                    }
-                }
-                break;
-            }
-            self.block_cursor.advance();
-        }
+impl HasLen for SegmentPostings {
+    fn len(&self) -> usize {
+        self.block_cursor.doc_freq() as usize
     }
 }
 
@@ -232,13 +229,6 @@ impl Postings for SegmentPostings {
         self.block_cursor.freq(self.cur)
     }
 
-    /// Returns the overall number of documents in the block postings.
-    /// It does not take in account whether documents are deleted or not.
-    #[inline(always)]
-    fn doc_freq(&self) -> DocFreq {
-        DocFreq::Exact(self.block_cursor.doc_freq())
-    }
-
     fn append_positions_with_offset(&mut self, offset: u32, output: &mut Vec<u32>) {
         let term_freq = self.term_freq();
         let prev_len = output.len();
@@ -262,42 +252,24 @@ impl Postings for SegmentPostings {
             }
         }
     }
-
-    fn has_freq(&self) -> bool {
-        !self.block_cursor.freqs().is_empty()
-    }
-}
-
-impl PostingsWithBlockMax for SegmentPostings {
-    fn seek_block_max(
-        &mut self,
-        target_doc: crate::DocId,
-        fieldnorm_reader: &FieldNormReader,
-        similarity_weight: &Bm25Weight,
-    ) -> Score {
-        self.block_cursor.seek_block_without_loading(target_doc);
-        self.block_cursor
-            .block_max_score(fieldnorm_reader, similarity_weight)
-    }
-
-    fn last_doc_in_block(&self) -> crate::DocId {
-        self.block_cursor.skip_reader().last_doc_in_block()
-    }
 }
 
 #[cfg(test)]
 mod tests {
+
+    use common::HasLen;
+
     use super::SegmentPostings;
     use crate::docset::{DocSet, TERMINATED};
-    use crate::postings::Postings;
+    use crate::fastfield::AliveBitSet;
+    use crate::postings::postings::Postings;
 
     #[test]
     fn test_empty_segment_postings() {
         let mut postings = SegmentPostings::empty();
-        assert_eq!(postings.doc(), TERMINATED);
         assert_eq!(postings.advance(), TERMINATED);
         assert_eq!(postings.advance(), TERMINATED);
-        assert_eq!(postings.doc_freq(), crate::postings::DocFreq::Exact(0));
+        assert_eq!(postings.len(), 0);
     }
 
     #[test]
@@ -312,4 +284,15 @@ mod tests {
         let postings = SegmentPostings::empty();
         assert_eq!(postings.term_freq(), 1);
     }
+
+    #[test]
+    fn test_doc_freq() {
+        let docs = SegmentPostings::create_from_docs(&[0, 2, 10]);
+        assert_eq!(docs.doc_freq(), 3);
+        let alive_bitset = AliveBitSet::for_test_from_deleted_docs(&[2], 12);
+        assert_eq!(docs.doc_freq_given_deletes(&alive_bitset), 2);
+        let all_deleted =
+            AliveBitSet::for_test_from_deleted_docs(&[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], 12);
+        assert_eq!(docs.doc_freq_given_deletes(&all_deleted), 0);
+    }
 }

src/postings/serializer.rs

@@ -1,15 +1,17 @@
+use std::cmp::Ordering;
 use std::io::{self, Write};
 
-use common::{BinarySerializable, CountingWriter};
+use common::{BinarySerializable, CountingWriter, VInt};
 
 use super::TermInfo;
-use crate::codec::postings::PostingsSerializer;
-use crate::codec::Codec;
 use crate::directory::{CompositeWrite, WritePtr};
 use crate::fieldnorm::FieldNormReader;
 use crate::index::Segment;
 use crate::positions::PositionSerializer;
-use crate::schema::{Field, FieldEntry, FieldType, IndexRecordOption, Schema};
+use crate::postings::compression::{BlockEncoder, VIntEncoder, COMPRESSION_BLOCK_SIZE};
+use crate::postings::skip::SkipSerializer;
+use crate::query::Bm25Weight;
+use crate::schema::{Field, FieldEntry, IndexRecordOption, Schema};
 use crate::termdict::TermDictionaryBuilder;
 use crate::{DocId, Score};
 
@@ -44,27 +46,22 @@ use crate::{DocId, Score};
 ///
 /// A description of the serialization format is
 /// [available here](https://fulmicoton.gitbooks.io/tantivy-doc/content/inverted-index.html).
-pub struct InvertedIndexSerializer<C: Codec> {
+pub struct InvertedIndexSerializer {
     terms_write: CompositeWrite<WritePtr>,
     postings_write: CompositeWrite<WritePtr>,
     positions_write: CompositeWrite<WritePtr>,
     schema: Schema,
-    codec: C,
 }
 
-use crate::codec::postings::PostingsCodec;
-
-impl<C: Codec> InvertedIndexSerializer<C> {
+impl InvertedIndexSerializer {
     /// Open a new `InvertedIndexSerializer` for the given segment
-    pub fn open(segment: &mut Segment<C>) -> crate::Result<InvertedIndexSerializer<C>> {
+    pub fn open(segment: &mut Segment) -> crate::Result<InvertedIndexSerializer> {
         use crate::index::SegmentComponent::{Positions, Postings, Terms};
-        let codec = segment.index().codec().clone();
         let inv_index_serializer = InvertedIndexSerializer {
             terms_write: CompositeWrite::wrap(segment.open_write(Terms)?),
             postings_write: CompositeWrite::wrap(segment.open_write(Postings)?),
             positions_write: CompositeWrite::wrap(segment.open_write(Positions)?),
             schema: segment.schema(),
-            codec,
         };
         Ok(inv_index_serializer)
     }
@@ -78,19 +75,22 @@ impl<C: Codec> InvertedIndexSerializer<C> {
         field: Field,
         total_num_tokens: u64,
         fieldnorm_reader: Option<FieldNormReader>,
-    ) -> io::Result<FieldSerializer<'_, C>> {
+    ) -> io::Result<FieldSerializer<'_>> {
         let field_entry: &FieldEntry = self.schema.get_field_entry(field);
         let term_dictionary_write = self.terms_write.for_field(field);
         let postings_write = self.postings_write.for_field(field);
         let positions_write = self.positions_write.for_field(field);
+        let index_record_option = field_entry
+            .field_type()
+            .index_record_option()
+            .unwrap_or(IndexRecordOption::Basic);
         FieldSerializer::create(
-            field_entry.field_type(),
+            index_record_option,
             total_num_tokens,
             term_dictionary_write,
             postings_write,
             positions_write,
             fieldnorm_reader,
-            &self.codec,
         )
     }
 
@@ -105,40 +105,34 @@ impl<C: Codec> InvertedIndexSerializer<C> {
 
 /// The field serializer is in charge of
 /// the serialization of a specific field.
-pub struct FieldSerializer<'a, C: Codec> {
-    term_dictionary_builder: TermDictionaryBuilder<&'a mut CountingWriter<WritePtr>>,
-    postings_serializer: <C::PostingsCodec as PostingsCodec>::PostingsSerializer,
-    positions_serializer_opt: Option<PositionSerializer<&'a mut CountingWriter<WritePtr>>>,
+pub struct FieldSerializer<'a, W: Write = WritePtr> {
+    term_dictionary_builder: TermDictionaryBuilder<&'a mut CountingWriter<W>>,
+    postings_serializer: PostingsSerializer,
+    positions_serializer_opt: Option<PositionSerializer<&'a mut CountingWriter<W>>>,
     current_term_info: TermInfo,
     term_open: bool,
-    postings_write: &'a mut CountingWriter<WritePtr>,
+    postings_write: &'a mut CountingWriter<W>,
     postings_start_offset: u64,
 }
 
-impl<'a, C: Codec> FieldSerializer<'a, C> {
-    fn create(
-        field_type: &FieldType,
+impl<'a, W: Write> FieldSerializer<'a, W> {
+    /// Creates a new `FieldSerializer` for the given field type.
+    pub fn create(
+        index_record_option: IndexRecordOption,
         total_num_tokens: u64,
-        term_dictionary_write: &'a mut CountingWriter<WritePtr>,
-        postings_write: &'a mut CountingWriter<WritePtr>,
-        positions_write: &'a mut CountingWriter<WritePtr>,
+        term_dictionary_write: &'a mut CountingWriter<W>,
+        postings_write: &'a mut CountingWriter<W>,
+        positions_write: &'a mut CountingWriter<W>,
         fieldnorm_reader: Option<FieldNormReader>,
-        codec: &C,
-    ) -> io::Result<FieldSerializer<'a, C>> {
-        let index_record_option = field_type
-            .index_record_option()
-            .unwrap_or(IndexRecordOption::Basic);
+    ) -> io::Result<FieldSerializer<'a, W>> {
         total_num_tokens.serialize(postings_write)?;
         let term_dictionary_builder = TermDictionaryBuilder::create(term_dictionary_write)?;
         let average_fieldnorm = fieldnorm_reader
             .as_ref()
             .map(|ff_reader| total_num_tokens as Score / ff_reader.num_docs() as Score)
             .unwrap_or(0.0);
-        let postings_serializer = codec.postings_codec().new_serializer(
-            average_fieldnorm,
-            index_record_option,
-            fieldnorm_reader,
-        );
+        let postings_serializer =
+            PostingsSerializer::new(average_fieldnorm, index_record_option, fieldnorm_reader);
         let positions_serializer_opt = if index_record_option.has_positions() {
             Some(PositionSerializer::new(positions_write))
         } else {
@@ -191,6 +185,7 @@ impl<'a, C: Codec> FieldSerializer<'a, C> {
             "Called new_term, while the previous term was not closed."
         );
         self.term_open = true;
+        self.postings_serializer.clear();
         self.current_term_info = self.current_term_info();
         self.term_dictionary_builder.insert_key(term)?;
         self.postings_serializer
@@ -259,3 +254,234 @@ impl<'a, C: Codec> FieldSerializer<'a, C> {
         Ok(())
     }
 }
+
+struct Block {
+    doc_ids: [DocId; COMPRESSION_BLOCK_SIZE],
+    term_freqs: [u32; COMPRESSION_BLOCK_SIZE],
+    len: usize,
+}
+
+impl Block {
+    fn new() -> Self {
+        Block {
+            doc_ids: [0u32; COMPRESSION_BLOCK_SIZE],
+            term_freqs: [0u32; COMPRESSION_BLOCK_SIZE],
+            len: 0,
+        }
+    }
+
+    fn doc_ids(&self) -> &[DocId] {
+        &self.doc_ids[..self.len]
+    }
+
+    fn term_freqs(&self) -> &[u32] {
+        &self.term_freqs[..self.len]
+    }
+
+    fn clear(&mut self) {
+        self.len = 0;
+    }
+
+    fn append_doc(&mut self, doc: DocId, term_freq: u32) {
+        let len = self.len;
+        self.doc_ids[len] = doc;
+        self.term_freqs[len] = term_freq;
+        self.len = len + 1;
+    }
+
+    fn is_full(&self) -> bool {
+        self.len == COMPRESSION_BLOCK_SIZE
+    }
+
+    fn is_empty(&self) -> bool {
+        self.len == 0
+    }
+
+    fn last_doc(&self) -> DocId {
+        assert_eq!(self.len, COMPRESSION_BLOCK_SIZE);
+        self.doc_ids[COMPRESSION_BLOCK_SIZE - 1]
+    }
+}
+
+/// Serializer for postings lists.
+pub struct PostingsSerializer {
+    last_doc_id_encoded: u32,
+
+    block_encoder: BlockEncoder,
+    block: Box<Block>,
+
+    postings_write: Vec<u8>,
+    skip_write: SkipSerializer,
+
+    mode: IndexRecordOption,
+    fieldnorm_reader: Option<FieldNormReader>,
+
+    bm25_weight: Option<Bm25Weight>,
+    avg_fieldnorm: Score, /* Average number of term in the field for that segment.
+                           * this value is used to compute the block wand information. */
+    term_has_freq: bool,
+}
+
+impl PostingsSerializer {
+    /// Creates a new `PostingsSerializer`.
+    /// * avg_fieldnorm - average field norm for the field being serialized.
+    /// * mode - indexing options for the field being serialized.
+    pub fn new(
+        avg_fieldnorm: Score,
+        mode: IndexRecordOption,
+        fieldnorm_reader: Option<FieldNormReader>,
+    ) -> PostingsSerializer {
+        PostingsSerializer {
+            block_encoder: BlockEncoder::new(),
+            block: Box::new(Block::new()),
+
+            postings_write: Vec::new(),
+            skip_write: SkipSerializer::new(),
+
+            last_doc_id_encoded: 0u32,
+            mode,
+
+            fieldnorm_reader,
+            bm25_weight: None,
+            avg_fieldnorm,
+            term_has_freq: false,
+        }
+    }
+
+    /// Starts the serialization for a new term.
+    /// * term_doc_freq - the number of documents containing the term.
+    pub fn new_term(&mut self, term_doc_freq: u32, record_term_freq: bool) {
+        self.bm25_weight = None;
+
+        self.term_has_freq = self.mode.has_freq() && record_term_freq;
+        if !self.term_has_freq {
+            return;
+        }
+
+        let num_docs_in_segment: u64 =
+            if let Some(fieldnorm_reader) = self.fieldnorm_reader.as_ref() {
+                fieldnorm_reader.num_docs() as u64
+            } else {
+                return;
+            };
+
+        if num_docs_in_segment == 0 {
+            return;
+        }
+
+        self.bm25_weight = Some(Bm25Weight::for_one_term_without_explain(
+            term_doc_freq as u64,
+            num_docs_in_segment,
+            self.avg_fieldnorm,
+        ));
+    }
+
+    fn write_block(&mut self) {
+        {
+            // encode the doc ids
+            let (num_bits, block_encoded): (u8, &[u8]) = self
+                .block_encoder
+                .compress_block_sorted(self.block.doc_ids(), self.last_doc_id_encoded);
+            self.last_doc_id_encoded = self.block.last_doc();
+            self.skip_write
+                .write_doc(self.last_doc_id_encoded, num_bits);
+            // last el block 0, offset block 1,
+            self.postings_write.extend(block_encoded);
+        }
+        if self.term_has_freq {
+            // encode the term frequencies
+            let (num_bits, block_encoded): (u8, &[u8]) = self
+                .block_encoder
+                .compress_block_unsorted(self.block.term_freqs(), true);
+            self.postings_write.extend(block_encoded);
+            self.skip_write.write_term_freq(num_bits);
+            if self.mode.has_positions() {
+                // We serialize the sum of term freqs within the skip information
+                // in order to navigate through positions.
+                let sum_freq = self.block.term_freqs().iter().cloned().sum();
+                self.skip_write.write_total_term_freq(sum_freq);
+            }
+            let mut blockwand_params = (0u8, 0u32);
+            if let Some(bm25_weight) = self.bm25_weight.as_ref() {
+                if let Some(fieldnorm_reader) = self.fieldnorm_reader.as_ref() {
+                    let docs = self.block.doc_ids().iter().cloned();
+                    let term_freqs = self.block.term_freqs().iter().cloned();
+                    let fieldnorms = docs.map(|doc| fieldnorm_reader.fieldnorm_id(doc));
+                    blockwand_params = fieldnorms
+                        .zip(term_freqs)
+                        .max_by(
+                            |(left_fieldnorm_id, left_term_freq),
+                             (right_fieldnorm_id, right_term_freq)| {
+                                let left_score =
+                                    bm25_weight.tf_factor(*left_fieldnorm_id, *left_term_freq);
+                                let right_score =
+                                    bm25_weight.tf_factor(*right_fieldnorm_id, *right_term_freq);
+                                left_score
+                                    .partial_cmp(&right_score)
+                                    .unwrap_or(Ordering::Equal)
+                            },
+                        )
+                        .unwrap();
+                }
+            }
+            let (fieldnorm_id, term_freq) = blockwand_params;
+            self.skip_write.write_blockwand_max(fieldnorm_id, term_freq);
+        }
+        self.block.clear();
+    }
+
+    /// Register that the given document contains the current term.
+    /// * doc_id - the document id.
+    /// * term_freq - the term frequency within the document.
+    pub fn write_doc(&mut self, doc_id: DocId, term_freq: u32) {
+        self.block.append_doc(doc_id, term_freq);
+        if self.block.is_full() {
+            self.write_block();
+        }
+    }
+
+    /// Finish the serialization for this term.
+    pub fn close_term(
+        &mut self,
+        doc_freq: u32,
+        output_write: &mut impl std::io::Write,
+    ) -> io::Result<()> {
+        if !self.block.is_empty() {
+            // we have doc ids waiting to be written
+            // this happens when the number of doc ids is
+            // not a perfect multiple of our block size.
+            //
+            // In that case, the remaining part is encoded
+            // using variable int encoding.
+            {
+                let block_encoded = self
+                    .block_encoder
+                    .compress_vint_sorted(self.block.doc_ids(), self.last_doc_id_encoded);
+                self.postings_write.write_all(block_encoded)?;
+            }
+            // ... Idem for term frequencies
+            if self.term_has_freq {
+                let block_encoded = self
+                    .block_encoder
+                    .compress_vint_unsorted(self.block.term_freqs());
+                self.postings_write.write_all(block_encoded)?;
+            }
+            self.block.clear();
+        }
+        if doc_freq >= COMPRESSION_BLOCK_SIZE as u32 {
+            let skip_data = self.skip_write.data();
+            VInt(skip_data.len() as u64).serialize(output_write)?;
+            output_write.write_all(skip_data)?;
+        }
+        output_write.write_all(&self.postings_write[..])?;
+        self.skip_write.clear();
+        self.postings_write.clear();
+        self.bm25_weight = None;
+        Ok(())
+    }
+
+    fn clear(&mut self) {
+        self.block.clear();
+        self.last_doc_id_encoded = 0;
+    }
+}

src/postings/skip.rs

@@ -146,6 +146,28 @@ impl SkipReader {
         skip_reader
     }
 
+    #[inline(always)]
+    pub fn has_remaining_docs(&self) -> bool {
+        self.remaining_docs != 0
+    }
+
+    pub fn reset(&mut self, data: OwnedBytes, doc_freq: u32) {
+        self.last_doc_in_block = if doc_freq >= COMPRESSION_BLOCK_SIZE as u32 {
+            0
+        } else {
+            TERMINATED
+        };
+        self.last_doc_in_previous_block = 0u32;
+        self.owned_read = data;
+        self.block_info = BlockInfo::VInt { num_docs: doc_freq };
+        self.byte_offset = 0;
+        self.remaining_docs = doc_freq;
+        self.position_offset = 0u64;
+        if doc_freq >= COMPRESSION_BLOCK_SIZE as u32 {
+            self.read_block_info();
+        }
+    }
+
     // Returns the block max score for this block if available.
     //
     // The block max score is available for all full bitpacked block,
@@ -165,6 +187,12 @@ impl SkipReader {
         self.last_doc_in_block
     }
 
+    /// Number of docs from the start of the current block to the end of the postings
+    /// (i.e. the current block plus every block after it).
+    pub(crate) fn remaining_docs(&self) -> u32 {
+        self.remaining_docs
+    }
+
     pub fn position_offset(&self) -> u64 {
         self.position_offset
     }

src/query/all_query.rs

@@ -2,7 +2,7 @@ use crate::docset::{DocSet, COLLECT_BLOCK_BUFFER_LEN, TERMINATED};
 use crate::index::SegmentReader;
 use crate::query::boost_query::BoostScorer;
 use crate::query::explanation::does_not_match;
-use crate::query::{box_scorer, EnableScoring, Explanation, Query, Scorer, Weight};
+use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
 use crate::{DocId, Score};
 
 /// Query that matches all of the documents.
@@ -24,9 +24,9 @@ impl Weight for AllWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         let all_scorer = AllScorer::new(reader.max_doc());
         if boost != 1.0 {
-            Ok(box_scorer(BoostScorer::new(all_scorer, boost)))
+            Ok(Box::new(BoostScorer::new(all_scorer, boost)))
         } else {
-            Ok(box_scorer(all_scorer))
+            Ok(Box::new(all_scorer))
         }
     }
 

src/query/automaton_weight.rs

@@ -10,7 +10,7 @@ use crate::postings::TermInfo;
 use crate::query::{BitSetDocSet, ConstScorer, Explanation, Scorer, Weight};
 use crate::schema::{Field, IndexRecordOption};
 use crate::termdict::{TermDictionary, TermStreamer};
-use crate::{DocId, DocSet, Score, TantivyError};
+use crate::{DocId, Score, TantivyError};
 
 /// A weight struct for Fuzzy Term and Regex Queries
 pub struct AutomatonWeight<A> {
@@ -92,9 +92,18 @@ where
         let mut term_stream = self.automaton_stream(term_dict)?;
         while term_stream.advance() {
             let term_info = term_stream.value();
-            let mut block_segment_postings =
-                inverted_index.read_postings_from_terminfo(term_info, IndexRecordOption::Basic)?;
-            block_segment_postings.fill_bitset(&mut doc_bitset);
+            let mut block_segment_postings = inverted_index
+                .read_block_postings_from_terminfo(term_info, IndexRecordOption::Basic)?;
+            loop {
+                let docs = block_segment_postings.docs();
+                if docs.is_empty() {
+                    break;
+                }
+                for &doc in docs {
+                    doc_bitset.insert(doc);
+                }
+                block_segment_postings.advance();
+            }
         }
         let doc_bitset = BitSetDocSet::from(doc_bitset);
         let const_scorer = ConstScorer::new(doc_bitset, boost);

src/query/bitset/mod.rs

@@ -24,13 +24,6 @@ impl BitSetDocSet {
         self.cursor_bucket = bucket_addr;
         self.cursor_tinybitset = self.docs.tinyset(bucket_addr);
     }
-
-    /// Returns the number of documents in the bitset.
-    ///
-    /// This call is not free: it will bitcount the number of bits in the bitset.
-    pub fn doc_freq(&self) -> u32 {
-        self.docs.len() as u32
-    }
 }
 
 impl From<BitSet> for BitSetDocSet {

src/query/boolean_query/block_wand_union.rs

@@ -1,6 +1,5 @@
 use std::ops::{Deref, DerefMut};
 
-use crate::codec::postings::PostingsWithBlockMax;
 use crate::query::term_query::TermScorer;
 use crate::query::Scorer;
 use crate::{DocId, DocSet, Score, TERMINATED};
@@ -14,8 +13,8 @@ use crate::{DocId, DocSet, Score, TERMINATED};
 /// We always have `before_pivot_len` < `pivot_len`.
 ///
 /// `None` is returned if we establish that no document can exceed the threshold.
-fn find_pivot_doc<TPostings: PostingsWithBlockMax>(
-    term_scorers: &[TermScorerWithMaxScore<TPostings>],
+fn find_pivot_doc(
+    term_scorers: &[TermScorerWithMaxScore],
     threshold: Score,
 ) -> Option<(usize, usize, DocId)> {
     let mut max_score = 0.0;
@@ -47,8 +46,8 @@ fn find_pivot_doc<TPostings: PostingsWithBlockMax>(
 /// the next doc candidate defined by the min of `last_doc_in_block + 1` for
 /// scorer in scorers[..pivot_len] and `scorer.doc()` for scorer in scorers[pivot_len..].
 /// Note: before and after calling this method, scorers need to be sorted by their `.doc()`.
-fn block_max_was_too_low_advance_one_scorer<TPostings: PostingsWithBlockMax>(
-    scorers: &mut [TermScorerWithMaxScore<TPostings>],
+fn block_max_was_too_low_advance_one_scorer(
+    scorers: &mut [TermScorerWithMaxScore],
     pivot_len: usize,
 ) {
     debug_assert!(scorers.iter().map(|scorer| scorer.doc()).is_sorted());
@@ -83,10 +82,7 @@ fn block_max_was_too_low_advance_one_scorer<TPostings: PostingsWithBlockMax>(
 // Given a list of term_scorers and a `ord` and assuming that `term_scorers[ord]` is sorted
 // except term_scorers[ord] that might be in advance compared to its ranks,
 // bubble up term_scorers[ord] in order to restore the ordering.
-fn restore_ordering<TPostings: PostingsWithBlockMax>(
-    term_scorers: &mut [TermScorerWithMaxScore<TPostings>],
-    ord: usize,
-) {
+fn restore_ordering(term_scorers: &mut [TermScorerWithMaxScore], ord: usize) {
     let doc = term_scorers[ord].doc();
     for i in ord + 1..term_scorers.len() {
         if term_scorers[i].doc() >= doc {
@@ -101,10 +97,9 @@ fn restore_ordering<TPostings: PostingsWithBlockMax>(
 // If this works, return true.
 // If this fails (ie: one of the term_scorer does not contain `pivot_doc` and seek goes past the
 // pivot), reorder the term_scorers to ensure the list is still sorted and returns `false`.
-// If a term_scorer reach TERMINATED in the process return false remove the term_scorer and
-// return.
-fn align_scorers<TPostings: PostingsWithBlockMax>(
-    term_scorers: &mut Vec<TermScorerWithMaxScore<TPostings>>,
+// If a term_scorer reach TERMINATED in the process return false remove the term_scorer and return.
+fn align_scorers(
+    term_scorers: &mut Vec<TermScorerWithMaxScore>,
     pivot_doc: DocId,
     before_pivot_len: usize,
 ) -> bool {
@@ -131,10 +126,7 @@ fn align_scorers<TPostings: PostingsWithBlockMax>(
 // Assumes terms_scorers[..pivot_len] are positioned on the same doc (pivot_doc).
 // Advance term_scorers[..pivot_len] and out of these removes the terminated scores.
 // Restores the ordering of term_scorers.
-fn advance_all_scorers_on_pivot<TPostings: PostingsWithBlockMax>(
-    term_scorers: &mut Vec<TermScorerWithMaxScore<TPostings>>,
-    pivot_len: usize,
-) {
+fn advance_all_scorers_on_pivot(term_scorers: &mut Vec<TermScorerWithMaxScore>, pivot_len: usize) {
     for term_scorer in &mut term_scorers[..pivot_len] {
         term_scorer.advance();
     }
@@ -153,8 +145,8 @@ fn advance_all_scorers_on_pivot<TPostings: PostingsWithBlockMax>(
 /// Implements the WAND (Weak AND) algorithm for dynamic pruning
 /// described in the paper "Faster Top-k Document Retrieval Using Block-Max Indexes".
 /// Link: <http://engineering.nyu.edu/~suel/papers/bmw.pdf>
-pub fn block_wand<TPostings: PostingsWithBlockMax>(
-    mut scorers: Vec<TermScorer<TPostings>>,
+pub fn block_wand(
+    mut scorers: Vec<TermScorer>,
     mut threshold: Score,
     callback: &mut dyn FnMut(u32, Score) -> Score,
 ) {
@@ -163,7 +155,7 @@ pub fn block_wand<TPostings: PostingsWithBlockMax>(
         let scorer = scorers.pop().unwrap();
         return block_wand_single_scorer(scorer, threshold, callback);
     }
-    let mut scorers: Vec<TermScorerWithMaxScore<TPostings>> = scorers
+    let mut scorers: Vec<TermScorerWithMaxScore> = scorers
         .iter_mut()
         .map(TermScorerWithMaxScore::from)
         .collect();
@@ -178,7 +170,10 @@ pub fn block_wand<TPostings: PostingsWithBlockMax>(
 
         let block_max_score_upperbound: Score = scorers[..pivot_len]
             .iter_mut()
-            .map(|scorer| scorer.seek_block_max(pivot_doc))
+            .map(|scorer| {
+                scorer.seek_block(pivot_doc);
+                scorer.block_max_score()
+            })
             .sum();
 
         // Beware after shallow advance, skip readers can be in advance compared to
@@ -229,22 +224,21 @@ pub fn block_wand<TPostings: PostingsWithBlockMax>(
 ///   - On a block, advance until the end and execute `callback` when the doc score is greater or
 ///     equal to the `threshold`.
 pub fn block_wand_single_scorer(
-    mut scorer: TermScorer<impl PostingsWithBlockMax>,
+    mut scorer: TermScorer,
     mut threshold: Score,
     callback: &mut dyn FnMut(u32, Score) -> Score,
 ) {
     let mut doc = scorer.doc();
-    let mut block_max_score = scorer.seek_block_max(doc);
     loop {
         // We position the scorer on a block that can reach
         // the threshold.
-        while block_max_score < threshold {
+        while scorer.block_max_score() <= threshold {
             let last_doc_in_block = scorer.last_doc_in_block();
             if last_doc_in_block == TERMINATED {
                 return;
             }
             doc = last_doc_in_block + 1;
-            block_max_score = scorer.seek_block_max(doc);
+            scorer.seek_block(doc);
         }
         // Seek will effectively load that block.
         doc = scorer.seek(doc);
@@ -266,33 +260,31 @@ pub fn block_wand_single_scorer(
             }
         }
         doc += 1;
-        block_max_score = scorer.seek_block_max(doc);
+        scorer.seek_block(doc);
     }
 }
 
-struct TermScorerWithMaxScore<'a, TPostings: PostingsWithBlockMax> {
-    scorer: &'a mut TermScorer<TPostings>,
+struct TermScorerWithMaxScore<'a> {
+    scorer: &'a mut TermScorer,
     max_score: Score,
 }
 
-impl<'a, TPostings: PostingsWithBlockMax> From<&'a mut TermScorer<TPostings>>
-    for TermScorerWithMaxScore<'a, TPostings>
-{
-    fn from(scorer: &'a mut TermScorer<TPostings>) -> Self {
+impl<'a> From<&'a mut TermScorer> for TermScorerWithMaxScore<'a> {
+    fn from(scorer: &'a mut TermScorer) -> Self {
         let max_score = scorer.max_score();
         TermScorerWithMaxScore { scorer, max_score }
     }
 }
 
-impl<TPostings: PostingsWithBlockMax> Deref for TermScorerWithMaxScore<'_, TPostings> {
-    type Target = TermScorer<TPostings>;
+impl Deref for TermScorerWithMaxScore<'_> {
+    type Target = TermScorer;
 
     fn deref(&self) -> &Self::Target {
         self.scorer
     }
 }
 
-impl<TPostings: PostingsWithBlockMax> DerefMut for TermScorerWithMaxScore<'_, TPostings> {
+impl DerefMut for TermScorerWithMaxScore<'_> {
     fn deref_mut(&mut self) -> &mut Self::Target {
         self.scorer
     }

src/query/boolean_query/boolean_weight.rs

@@ -1,18 +1,25 @@
 use std::collections::HashMap;
 
-use crate::codec::{ObjectSafeCodec, SumOrDoNothingCombiner};
 use crate::docset::COLLECT_BLOCK_BUFFER_LEN;
 use crate::index::SegmentReader;
+use crate::postings::FreqReadingOption;
 use crate::query::disjunction::Disjunction;
 use crate::query::explanation::does_not_match;
 use crate::query::score_combiner::{DoNothingCombiner, ScoreCombiner};
-use crate::query::weight::for_each_docset_buffered;
+use crate::query::term_query::TermScorer;
+use crate::query::weight::{for_each_docset_buffered, for_each_pruning_scorer, for_each_scorer};
 use crate::query::{
-    box_scorer, intersect_scorers, AllScorer, BufferedUnionScorer, EmptyScorer, Exclude,
-    Explanation, Occur, RequiredOptionalScorer, Scorer, SumCombiner, Weight,
+    intersect_scorers, AllScorer, BufferedUnionScorer, EmptyScorer, Exclude, Explanation, Occur,
+    RequiredOptionalScorer, Scorer, Weight,
 };
 use crate::{DocId, Score};
 
+enum SpecializedScorer {
+    TermUnion(Vec<TermScorer>),
+    TermIntersection(Vec<TermScorer>),
+    Other(Box<dyn Scorer>),
+}
+
 fn scorer_disjunction<TScoreCombiner>(
     scorers: Vec<Box<dyn Scorer>>,
     score_combiner: TScoreCombiner,
@@ -26,7 +33,7 @@ where
     if scorers.len() == 1 {
         return scorers.into_iter().next().unwrap(); // Safe unwrap.
     }
-    box_scorer(Disjunction::new(
+    Box::new(Disjunction::new(
         scorers,
         score_combiner,
         minimum_match_required,
@@ -38,41 +45,66 @@ fn scorer_union<TScoreCombiner>(
     scorers: Vec<Box<dyn Scorer>>,
     score_combiner_fn: impl Fn() -> TScoreCombiner,
     num_docs: u32,
-    codec: &dyn ObjectSafeCodec,
-) -> Box<dyn Scorer>
+) -> SpecializedScorer
 where
     TScoreCombiner: ScoreCombiner,
 {
-    match scorers.len() {
-        0 => box_scorer(EmptyScorer),
-        1 => scorers.into_iter().next().unwrap(),
-        _ => {
-            let combiner_opt: Option<SumOrDoNothingCombiner> = if std::any::TypeId::of::<
-                TScoreCombiner,
-            >() == std::any::TypeId::of::<
-                SumCombiner,
-            >() {
-                Some(SumOrDoNothingCombiner::Sum)
-            } else if std::any::TypeId::of::<TScoreCombiner>()
-                == std::any::TypeId::of::<DoNothingCombiner>()
+    assert!(!scorers.is_empty());
+    if scorers.len() == 1 && !scorers[0].is::<TermScorer>() {
+        return SpecializedScorer::Other(scorers.into_iter().next().unwrap()); //< we checked the size beforehand
+    }
+    {
+        let is_all_term_queries = scorers.iter().all(|scorer| scorer.is::<TermScorer>());
+        if is_all_term_queries {
+            let scorers: Vec<TermScorer> = scorers
+                .into_iter()
+                .map(|scorer| *(scorer.downcast::<TermScorer>().map_err(|_| ()).unwrap()))
+                .collect();
+            if scorers
+                .iter()
+                .all(|scorer| scorer.freq_reading_option() == FreqReadingOption::ReadFreq)
             {
-                Some(SumOrDoNothingCombiner::DoNothing)
+                // Block wand is only available if we read frequencies.
+                return SpecializedScorer::TermUnion(scorers);
+            } else if scorers.len() == 1 {
+                // Single TermScorer without freq reading — unwrap directly.
+                return SpecializedScorer::Other(Box::new(scorers.into_iter().next().unwrap()));
             } else {
-                None
-            };
-            if let Some(combiner) = combiner_opt {
-                let scorer =
-                    codec.build_union_scorer_with_sum_combiner(scorers, num_docs, combiner);
-                scorer
-            } else {
-                box_scorer(BufferedUnionScorer::build(
+                return SpecializedScorer::Other(Box::new(BufferedUnionScorer::build(
                     scorers,
                     score_combiner_fn,
                     num_docs,
-                ))
+                )));
             }
         }
     }
+    SpecializedScorer::Other(Box::new(BufferedUnionScorer::build(
+        scorers,
+        score_combiner_fn,
+        num_docs,
+    )))
+}
+
+fn into_box_scorer<TScoreCombiner: ScoreCombiner>(
+    scorer: SpecializedScorer,
+    score_combiner_fn: impl Fn() -> TScoreCombiner,
+    num_docs: u32,
+) -> Box<dyn Scorer> {
+    match scorer {
+        SpecializedScorer::TermUnion(term_scorers) => {
+            let union_scorer =
+                BufferedUnionScorer::build(term_scorers, score_combiner_fn, num_docs);
+            Box::new(union_scorer)
+        }
+        SpecializedScorer::TermIntersection(term_scorers) => {
+            let boxed_scorers: Vec<Box<dyn Scorer>> = term_scorers
+                .into_iter()
+                .map(|s| Box::new(s) as Box<dyn Scorer>)
+                .collect();
+            intersect_scorers(boxed_scorers, num_docs)
+        }
+        SpecializedScorer::Other(scorer) => scorer,
+    }
 }
 
 /// Returns the effective MUST scorer, accounting for removed AllScorers.
@@ -88,7 +120,7 @@ fn effective_must_scorer(
     if must_scorers.is_empty() {
         if removed_all_scorer_count > 0 {
             // Had AllScorer(s) only - all docs match
-            Some(box_scorer(AllScorer::new(max_doc)))
+            Some(Box::new(AllScorer::new(max_doc)))
         } else {
             // No MUST constraint at all
             None
@@ -106,26 +138,28 @@ fn effective_must_scorer(
 /// When `scoring_enabled` is false, we can just return AllScorer alone since
 /// we don't need score contributions from the should_scorer.
 fn effective_should_scorer_for_union<TScoreCombiner: ScoreCombiner>(
-    should_scorer: Box<dyn Scorer>,
+    should_scorer: SpecializedScorer,
     removed_all_scorer_count: usize,
     max_doc: DocId,
     num_docs: u32,
     score_combiner_fn: impl Fn() -> TScoreCombiner,
     scoring_enabled: bool,
-) -> Box<dyn Scorer> {
+) -> SpecializedScorer {
     if removed_all_scorer_count > 0 {
         if scoring_enabled {
             // Need to union to get score contributions from both
-            let all_scorers: Vec<Box<dyn Scorer>> =
-                vec![should_scorer, box_scorer(AllScorer::new(max_doc))];
-            box_scorer(BufferedUnionScorer::build(
+            let all_scorers: Vec<Box<dyn Scorer>> = vec![
+                into_box_scorer(should_scorer, &score_combiner_fn, num_docs),
+                Box::new(AllScorer::new(max_doc)),
+            ];
+            SpecializedScorer::Other(Box::new(BufferedUnionScorer::build(
                 all_scorers,
                 score_combiner_fn,
                 num_docs,
-            ))
+            )))
         } else {
             // Scoring disabled - AllScorer alone is sufficient
-            box_scorer(AllScorer::new(max_doc))
+            SpecializedScorer::Other(Box::new(AllScorer::new(max_doc)))
         }
     } else {
         should_scorer
@@ -136,9 +170,9 @@ enum ShouldScorersCombinationMethod {
     // Should scorers are irrelevant.
     Ignored,
     // Only contributes to final score.
-    Optional(Box<dyn Scorer>),
+    Optional(SpecializedScorer),
     // Regardless of score, the should scorers may impact whether a document is matching or not.
-    Required(Box<dyn Scorer>),
+    Required(SpecializedScorer),
 }
 
 /// Weight associated to the `BoolQuery`.
@@ -200,7 +234,7 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
         reader: &SegmentReader,
         boost: Score,
         score_combiner_fn: impl Fn() -> TComplexScoreCombiner,
-    ) -> crate::Result<Box<dyn Scorer>> {
+    ) -> crate::Result<SpecializedScorer> {
         let num_docs = reader.num_docs();
         let mut per_occur_scorers = self.per_occur_scorers(reader, boost)?;
 
@@ -210,7 +244,7 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
         let must_special_scorer_counts = remove_and_count_all_and_empty_scorers(&mut must_scorers);
 
         if must_special_scorer_counts.num_empty_scorers > 0 {
-            return Ok(box_scorer(EmptyScorer));
+            return Ok(SpecializedScorer::Other(Box::new(EmptyScorer)));
         }
 
         let mut should_scorers = per_occur_scorers.remove(&Occur::Should).unwrap_or_default();
@@ -225,7 +259,7 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
 
         if exclude_special_scorer_counts.num_all_scorers > 0 {
             // We exclude all documents at one point.
-            return Ok(box_scorer(EmptyScorer));
+            return Ok(SpecializedScorer::Other(Box::new(EmptyScorer)));
         }
 
         let effective_minimum_number_should_match = self
@@ -237,7 +271,7 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
             if effective_minimum_number_should_match > num_of_should_scorers {
                 // We don't have enough scorers to satisfy the minimum number of should matches.
                 // The request will match no documents.
-                return Ok(box_scorer(EmptyScorer));
+                return Ok(SpecializedScorer::Other(Box::new(EmptyScorer)));
             }
             match effective_minimum_number_should_match {
                 0 if num_of_should_scorers == 0 => ShouldScorersCombinationMethod::Ignored,
@@ -245,13 +279,11 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
                     should_scorers,
                     &score_combiner_fn,
                     num_docs,
-                    reader.codec(),
                 )),
                 1 => ShouldScorersCombinationMethod::Required(scorer_union(
                     should_scorers,
                     &score_combiner_fn,
                     num_docs,
-                    reader.codec(),
                 )),
                 n if num_of_should_scorers == n => {
                     // When num_of_should_scorers equals the number of should clauses,
@@ -259,41 +291,59 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
                     must_scorers.append(&mut should_scorers);
                     ShouldScorersCombinationMethod::Ignored
                 }
-                _ => ShouldScorersCombinationMethod::Required(scorer_disjunction(
-                    should_scorers,
-                    score_combiner_fn(),
-                    effective_minimum_number_should_match,
+                _ => ShouldScorersCombinationMethod::Required(SpecializedScorer::Other(
+                    scorer_disjunction(
+                        should_scorers,
+                        score_combiner_fn(),
+                        effective_minimum_number_should_match,
+                    ),
                 )),
             }
         };
 
-        let exclude_scorer_opt: Option<Box<dyn Scorer>> = if exclude_scorers.is_empty() {
-            None
-        } else {
-            let exclude_scorers_union: Box<dyn Scorer> = scorer_union(
-                exclude_scorers,
-                DoNothingCombiner::default,
-                num_docs,
-                reader.codec(),
-            );
-            Some(exclude_scorers_union)
-        };
-
-
         let include_scorer = match (should_scorers, must_scorers) {
             (ShouldScorersCombinationMethod::Ignored, must_scorers) => {
                 // No SHOULD clauses (or they were absorbed into MUST).
                 // Result depends entirely on MUST + any removed AllScorers.
                 let combined_all_scorer_count = must_special_scorer_counts.num_all_scorers
                     + should_special_scorer_counts.num_all_scorers;
-                let boxed_scorer: Box<dyn Scorer> = effective_must_scorer(
-                    must_scorers,
-                    combined_all_scorer_count,
-                    reader.max_doc(),
-                    num_docs,
-                )
-                .unwrap_or_else(|| box_scorer(EmptyScorer));
-                boxed_scorer
+
+                // Try to detect a pure TermScorer intersection for block-max optimization.
+                // Preconditions: no removed AllScorers, at least 2 scorers, all TermScorer
+                // with frequency reading enabled.
+                if combined_all_scorer_count == 0
+                    && must_scorers.len() >= 2
+                    && must_scorers.iter().all(|s| s.is::<TermScorer>())
+                {
+                    let term_scorers: Vec<TermScorer> = must_scorers
+                        .into_iter()
+                        .map(|s| *(s.downcast::<TermScorer>().map_err(|_| ()).unwrap()))
+                        .collect();
+                    if term_scorers
+                        .iter()
+                        .all(|s| s.freq_reading_option() == FreqReadingOption::ReadFreq)
+                    {
+                        SpecializedScorer::TermIntersection(term_scorers)
+                    } else {
+                        let must_scorers: Vec<Box<dyn Scorer>> = term_scorers
+                            .into_iter()
+                            .map(|s| Box::new(s) as Box<dyn Scorer>)
+                            .collect();
+                        let boxed_scorer: Box<dyn Scorer> =
+                            effective_must_scorer(must_scorers, 0, reader.max_doc(), num_docs)
+                                .unwrap_or_else(|| Box::new(EmptyScorer));
+                        SpecializedScorer::Other(boxed_scorer)
+                    }
+                } else {
+                    let boxed_scorer: Box<dyn Scorer> = effective_must_scorer(
+                        must_scorers,
+                        combined_all_scorer_count,
+                        reader.max_doc(),
+                        num_docs,
+                    )
+                    .unwrap_or_else(|| Box::new(EmptyScorer));
+                    SpecializedScorer::Other(boxed_scorer)
+                }
             }
             (ShouldScorersCombinationMethod::Optional(should_scorer), must_scorers) => {
                 // Optional SHOULD: contributes to scoring but not required for matching.
@@ -318,12 +368,16 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
                     Some(must_scorer) => {
                         // Has MUST constraint: SHOULD only affects scoring.
                         if self.scoring_enabled {
-                            box_scorer(RequiredOptionalScorer::<_, _, TScoreCombiner>::new(
+                            SpecializedScorer::Other(Box::new(RequiredOptionalScorer::<
+                                _,
+                                _,
+                                TScoreCombiner,
+                            >::new(
                                 must_scorer,
-                                should_scorer,
-                            ))
+                                into_box_scorer(should_scorer, &score_combiner_fn, num_docs),
+                            )))
                         } else {
-                            must_scorer
+                            SpecializedScorer::Other(must_scorer)
                         }
                     }
                 }
@@ -343,16 +397,33 @@ impl<TScoreCombiner: ScoreCombiner> BooleanWeight<TScoreCombiner> {
                     }
                     Some(must_scorer) => {
                         // Has MUST constraint: intersect MUST with SHOULD.
-                        intersect_scorers(vec![must_scorer, should_scorer], num_docs)
+                        let should_boxed =
+                            into_box_scorer(should_scorer, &score_combiner_fn, num_docs);
+                        SpecializedScorer::Other(intersect_scorers(
+                            vec![must_scorer, should_boxed],
+                            num_docs,
+                        ))
                     }
                 }
             }
         };
-        if let Some(exclude_scorer) = exclude_scorer_opt {
-            Ok(box_scorer(Exclude::new(include_scorer, exclude_scorer)))
-        } else {
-            Ok(include_scorer)
+        if exclude_scorers.is_empty() {
+            return Ok(include_scorer);
         }
+
+        let include_scorer_boxed = into_box_scorer(include_scorer, &score_combiner_fn, num_docs);
+        let scorer: Box<dyn Scorer> = if exclude_scorers.len() == 1 {
+            let exclude_scorer = exclude_scorers.pop().unwrap();
+            match exclude_scorer.downcast::<TermScorer>() {
+                // Cast to TermScorer succeeded
+                Ok(exclude_scorer) => Box::new(Exclude::new(include_scorer_boxed, *exclude_scorer)),
+                // We get back the original Box<dyn Scorer>
+                Err(exclude_scorer) => Box::new(Exclude::new(include_scorer_boxed, exclude_scorer)),
+            }
+        } else {
+            Box::new(Exclude::new(include_scorer_boxed, exclude_scorers))
+        };
+        Ok(SpecializedScorer::Other(scorer))
     }
 }
 
@@ -382,6 +453,7 @@ fn remove_and_count_all_and_empty_scorers(
 
 impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombiner> {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
+        let num_docs = reader.num_docs();
         if self.weights.is_empty() {
             Ok(Box::new(EmptyScorer))
         } else if self.weights.len() == 1 {
@@ -393,8 +465,14 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
             }
         } else if self.scoring_enabled {
             self.complex_scorer(reader, boost, &self.score_combiner_fn)
+                .map(|specialized_scorer| {
+                    into_box_scorer(specialized_scorer, &self.score_combiner_fn, num_docs)
+                })
         } else {
             self.complex_scorer(reader, boost, DoNothingCombiner::default)
+                .map(|specialized_scorer| {
+                    into_box_scorer(specialized_scorer, DoNothingCombiner::default, num_docs)
+                })
         }
     }
 
@@ -423,8 +501,26 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score),
     ) -> crate::Result<()> {
-        let mut scorer = self.complex_scorer(reader, 1.0, &self.score_combiner_fn)?;
-        scorer.for_each(callback);
+        let scorer = self.complex_scorer(reader, 1.0, &self.score_combiner_fn)?;
+        let num_docs = reader.num_docs();
+        match scorer {
+            SpecializedScorer::TermUnion(term_scorers) => {
+                let mut union_scorer =
+                    BufferedUnionScorer::build(term_scorers, &self.score_combiner_fn, num_docs);
+                for_each_scorer(&mut union_scorer, callback);
+            }
+            SpecializedScorer::TermIntersection(term_scorers) => {
+                let boxed_scorers: Vec<Box<dyn Scorer>> = term_scorers
+                    .into_iter()
+                    .map(|term_scorer| Box::new(term_scorer) as Box<dyn Scorer>)
+                    .collect();
+                let mut intersection = intersect_scorers(boxed_scorers, num_docs);
+                for_each_scorer(intersection.as_mut(), callback);
+            }
+            SpecializedScorer::Other(mut scorer) => {
+                for_each_scorer(scorer.as_mut(), callback);
+            }
+        }
         Ok(())
     }
 
@@ -433,9 +529,28 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
         reader: &SegmentReader,
         callback: &mut dyn FnMut(&[DocId]),
     ) -> crate::Result<()> {
-        let mut scorer = self.complex_scorer(reader, 1.0, || DoNothingCombiner)?;
+        let scorer = self.complex_scorer(reader, 1.0, || DoNothingCombiner)?;
+        let num_docs = reader.num_docs();
         let mut buffer = [0u32; COLLECT_BLOCK_BUFFER_LEN];
-        for_each_docset_buffered(scorer.as_mut(), &mut buffer, callback);
+
+        match scorer {
+            SpecializedScorer::TermUnion(term_scorers) => {
+                let mut union_scorer =
+                    BufferedUnionScorer::build(term_scorers, &self.score_combiner_fn, num_docs);
+                for_each_docset_buffered(&mut union_scorer, &mut buffer, callback);
+            }
+            SpecializedScorer::TermIntersection(term_scorers) => {
+                let boxed_scorers: Vec<Box<dyn Scorer>> = term_scorers
+                    .into_iter()
+                    .map(|term_scorer| Box::new(term_scorer) as Box<dyn Scorer>)
+                    .collect();
+                let mut intersection = intersect_scorers(boxed_scorers, num_docs);
+                for_each_docset_buffered(intersection.as_mut(), &mut buffer, callback);
+            }
+            SpecializedScorer::Other(mut scorer) => {
+                for_each_docset_buffered(scorer.as_mut(), &mut buffer, callback);
+            }
+        }
         Ok(())
     }
 
@@ -456,7 +571,17 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
         callback: &mut dyn FnMut(DocId, Score) -> Score,
     ) -> crate::Result<()> {
         let scorer = self.complex_scorer(reader, 1.0, &self.score_combiner_fn)?;
-        reader.codec().for_each_pruning(threshold, scorer, callback);
+        match scorer {
+            SpecializedScorer::TermUnion(term_scorers) => {
+                super::block_wand(term_scorers, threshold, callback);
+            }
+            SpecializedScorer::TermIntersection(term_scorers) => {
+                super::block_wand_intersection(term_scorers, threshold, callback);
+            }
+            SpecializedScorer::Other(mut scorer) => {
+                for_each_pruning_scorer(scorer.as_mut(), threshold, callback);
+            }
+        }
         Ok(())
     }
 }

src/query/boolean_query/mod.rs

@@ -1,7 +1,10 @@
+mod block_wand_intersection;
+mod block_wand_union;
 mod boolean_query;
 mod boolean_weight;
 
-
+pub(crate) use self::block_wand_intersection::block_wand_intersection;
+pub(crate) use self::block_wand_union::{block_wand, block_wand_single_scorer};
 pub use self::boolean_query::BooleanQuery;
 pub use self::boolean_weight::BooleanWeight;
 

src/query/const_score_query.rs

@@ -1,7 +1,7 @@
 use std::fmt;
 
-use crate::docset::COLLECT_BLOCK_BUFFER_LEN;
-use crate::query::{box_scorer, EnableScoring, Explanation, Query, Scorer, Weight};
+use crate::docset::{SeekDangerResult, COLLECT_BLOCK_BUFFER_LEN};
+use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
 use crate::{DocId, DocSet, Score, SegmentReader, TantivyError, Term};
 
 /// `ConstScoreQuery` is a wrapper over a query to provide a constant score.
@@ -65,10 +65,7 @@ impl ConstWeight {
 impl Weight for ConstWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         let inner_scorer = self.weight.scorer(reader, boost)?;
-        Ok(box_scorer(ConstScorer::new(
-            inner_scorer,
-            boost * self.score,
-        )))
+        Ok(Box::new(ConstScorer::new(inner_scorer, boost * self.score)))
     }
 
     fn explain(&self, reader: &SegmentReader, doc: u32) -> crate::Result<Explanation> {
@@ -122,6 +119,10 @@ impl<TDocSet: DocSet> DocSet for ConstScorer<TDocSet> {
         self.docset.seek(target)
     }
 
+    fn seek_danger(&mut self, target: DocId) -> SeekDangerResult {
+        self.docset.seek_danger(target)
+    }
+
     fn fill_buffer(&mut self, buffer: &mut [DocId; COLLECT_BLOCK_BUFFER_LEN]) -> usize {
         self.docset.fill_buffer(buffer)
     }

src/query/empty_query.rs

@@ -2,7 +2,7 @@ use super::Scorer;
 use crate::docset::TERMINATED;
 use crate::index::SegmentReader;
 use crate::query::explanation::does_not_match;
-use crate::query::{box_scorer, EnableScoring, Explanation, Query, Weight};
+use crate::query::{EnableScoring, Explanation, Query, Weight};
 use crate::{DocId, DocSet, Score, Searcher};
 
 /// `EmptyQuery` is a dummy `Query` in which no document matches.
@@ -27,7 +27,7 @@ impl Query for EmptyQuery {
 pub struct EmptyWeight;
 impl Weight for EmptyWeight {
     fn scorer(&self, _reader: &SegmentReader, _boost: Score) -> crate::Result<Box<dyn Scorer>> {
-        Ok(box_scorer(EmptyScorer))
+        Ok(Box::new(EmptyScorer))
     }
 
     fn explain(&self, _reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {

src/query/exist_query.rs

@@ -3,7 +3,7 @@ use core::fmt::Debug;
 use columnar::{ColumnIndex, DynamicColumn};
 use common::BitSet;
 
-use super::{box_scorer, ConstScorer, EmptyScorer};
+use super::{ConstScorer, EmptyScorer};
 use crate::docset::{DocSet, TERMINATED};
 use crate::index::SegmentReader;
 use crate::query::all_query::AllScorer;
@@ -117,7 +117,7 @@ impl Weight for ExistsWeight {
             }
         }
         if non_empty_columns.is_empty() {
-            return Ok(box_scorer(EmptyScorer));
+            return Ok(Box::new(EmptyScorer));
         }
 
         // If any column is full, all docs match.
@@ -128,9 +128,9 @@ impl Weight for ExistsWeight {
         {
             let all_scorer = AllScorer::new(max_doc);
             if boost != 1.0f32 {
-                return Ok(box_scorer(BoostScorer::new(all_scorer, boost)));
+                return Ok(Box::new(BoostScorer::new(all_scorer, boost)));
             } else {
-                return Ok(box_scorer(all_scorer));
+                return Ok(Box::new(all_scorer));
             }
         }
 
@@ -138,7 +138,7 @@ impl Weight for ExistsWeight {
         // NOTE: A lower number may be better for very sparse columns
         if non_empty_columns.len() < 4 {
             let docset = ExistsDocSet::new(non_empty_columns, reader.max_doc());
-            return Ok(box_scorer(ConstScorer::new(docset, boost)));
+            return Ok(Box::new(ConstScorer::new(docset, boost)));
         }
 
         // If we have many dynamic columns, precompute a bitset of matching docs
@@ -162,7 +162,7 @@ impl Weight for ExistsWeight {
             }
         }
         let docset = BitSetDocSet::from(doc_bitset);
-        Ok(box_scorer(ConstScorer::new(docset, boost)))
+        Ok(Box::new(ConstScorer::new(docset, boost)))
     }
 
     fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {

src/query/intersection.rs

@@ -3,7 +3,7 @@ use common::TinySet;
 use super::size_hint::estimate_intersection;
 use crate::docset::{DocSet, SeekDangerResult, BLOCK_NUM_TINYBITSETS, TERMINATED};
 use crate::query::term_query::TermScorer;
-use crate::query::{box_scorer, EmptyScorer, Scorer};
+use crate::query::{EmptyScorer, Scorer};
 use crate::{DocId, Score};
 
 /// Returns the intersection scorer.
@@ -22,7 +22,7 @@ pub fn intersect_scorers(
     segment_num_docs: u32,
 ) -> Box<dyn Scorer> {
     if scorers.is_empty() {
-        return box_scorer(EmptyScorer);
+        return Box::new(EmptyScorer);
     }
     if scorers.len() == 1 {
         return scorers.pop().unwrap();
@@ -31,7 +31,7 @@ pub fn intersect_scorers(
     scorers.sort_by_key(|scorer| scorer.cost());
     let doc = go_to_first_doc(&mut scorers[..]);
     if doc == TERMINATED {
-        return box_scorer(EmptyScorer);
+        return Box::new(EmptyScorer);
     }
     // We know that we have at least 2 elements.
     let left = scorers.remove(0);
@@ -40,14 +40,14 @@ pub fn intersect_scorers(
         .iter()
         .all(|&scorer| scorer.is::<TermScorer>());
     if all_term_scorers {
-        return box_scorer(Intersection {
+        return Box::new(Intersection {
             left: *(left.downcast::<TermScorer>().map_err(|_| ()).unwrap()),
             right: *(right.downcast::<TermScorer>().map_err(|_| ()).unwrap()),
             others: scorers,
             segment_num_docs,
         });
     }
-    box_scorer(Intersection {
+    Box::new(Intersection {
         left,
         right,
         others: scorers,

src/query/mod.rs

@@ -24,7 +24,7 @@ mod reqopt_scorer;
 mod scorer;
 mod set_query;
 mod size_hint;
-pub(crate) mod term_query;
+mod term_query;
 mod union;
 mod weight;
 
@@ -54,14 +54,13 @@ pub use self::more_like_this::{MoreLikeThisQuery, MoreLikeThisQueryBuilder};
 pub use self::phrase_prefix_query::PhrasePrefixQuery;
 pub use self::phrase_query::regex_phrase_query::{wildcard_query_to_regex_str, RegexPhraseQuery};
 pub use self::phrase_query::PhraseQuery;
-pub(crate) use self::phrase_query::PhraseScorer;
 pub use self::query::{EnableScoring, Query, QueryClone};
 pub use self::query_parser::{QueryParser, QueryParserError};
 pub use self::range_query::*;
 pub use self::regex_query::RegexQuery;
 pub use self::reqopt_scorer::RequiredOptionalScorer;
 pub use self::score_combiner::{DisjunctionMaxCombiner, ScoreCombiner, SumCombiner};
-pub use self::scorer::{box_scorer, Scorer};
+pub use self::scorer::Scorer;
 pub use self::set_query::TermSetQuery;
 pub use self::term_query::TermQuery;
 pub use self::union::BufferedUnionScorer;

src/query/phrase_prefix_query/phrase_prefix_scorer.rs

@@ -2,7 +2,7 @@ use crate::docset::{DocSet, SeekDangerResult, TERMINATED};
 use crate::fieldnorm::FieldNormReader;
 use crate::postings::Postings;
 use crate::query::bm25::Bm25Weight;
-use crate::query::phrase_query::{intersection_exists, PhraseScorer};
+use crate::query::phrase_query::{intersection_count, PhraseScorer};
 use crate::query::Scorer;
 use crate::{DocId, Score};
 
@@ -100,6 +100,7 @@ pub struct PhrasePrefixScorer<TPostings: Postings> {
     phrase_scorer: PhraseKind<TPostings>,
     suffixes: Vec<TPostings>,
     suffix_offset: u32,
+    phrase_count: u32,
     suffix_position_buffer: Vec<u32>,
 }
 
@@ -143,6 +144,7 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
             phrase_scorer,
             suffixes,
             suffix_offset: (max_offset - suffix_pos) as u32,
+            phrase_count: 0,
             suffix_position_buffer: Vec::with_capacity(100),
         };
         if phrase_prefix_scorer.doc() != TERMINATED && !phrase_prefix_scorer.matches_prefix() {
@@ -151,7 +153,12 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
         phrase_prefix_scorer
     }
 
+    pub fn phrase_count(&self) -> u32 {
+        self.phrase_count
+    }
+
     fn matches_prefix(&mut self) -> bool {
+        let mut count = 0;
         let current_doc = self.doc();
         let pos_matching = self.phrase_scorer.get_intersection();
         for suffix in &mut self.suffixes {
@@ -161,12 +168,11 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
             let doc = suffix.seek(current_doc);
             if doc == current_doc {
                 suffix.positions_with_offset(self.suffix_offset, &mut self.suffix_position_buffer);
-                if intersection_exists(pos_matching, &self.suffix_position_buffer) {
-                    return true;
-                }
+                count += intersection_count(pos_matching, &self.suffix_position_buffer);
             }
         }
-        false
+        self.phrase_count = count as u32;
+        count != 0
     }
 }
 

src/query/phrase_prefix_query/phrase_prefix_weight.rs

@@ -1,11 +1,12 @@
 use super::{prefix_end, PhrasePrefixScorer};
 use crate::fieldnorm::FieldNormReader;
 use crate::index::SegmentReader;
-use crate::postings::Postings;
+use crate::postings::SegmentPostings;
 use crate::query::bm25::Bm25Weight;
-use crate::query::{box_scorer, EmptyScorer, Scorer, Weight};
+use crate::query::explanation::does_not_match;
+use crate::query::{EmptyScorer, Explanation, Scorer, Weight};
 use crate::schema::{IndexRecordOption, Term};
-use crate::Score;
+use crate::{DocId, DocSet, Score};
 
 pub struct PhrasePrefixWeight {
     phrase_terms: Vec<(usize, Term)>,
@@ -45,13 +46,13 @@ impl PhrasePrefixWeight {
         &self,
         reader: &SegmentReader,
         boost: Score,
-    ) -> crate::Result<Option<Box<dyn Scorer>>> {
+    ) -> crate::Result<Option<PhrasePrefixScorer<SegmentPostings>>> {
         let similarity_weight_opt = self
             .similarity_weight_opt
             .as_ref()
             .map(|similarity_weight| similarity_weight.boost_by(boost));
         let fieldnorm_reader = self.fieldnorm_reader(reader)?;
-        let mut term_postings_list: Vec<(usize, Box<dyn Postings>)> = Vec::new();
+        let mut term_postings_list = Vec::new();
         for &(offset, ref term) in &self.phrase_terms {
             if let Some(postings) = reader
                 .inverted_index(term.field())?
@@ -102,32 +103,49 @@ impl PhrasePrefixWeight {
             }
         }
 
-        Ok(Some(box_scorer(PhrasePrefixScorer::new(
+        Ok(Some(PhrasePrefixScorer::new(
             term_postings_list,
             similarity_weight_opt,
             fieldnorm_reader,
             suffixes,
             self.prefix.0,
-        ))))
+        )))
     }
 }
 
 impl Weight for PhrasePrefixWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         if let Some(scorer) = self.phrase_scorer(reader, boost)? {
-            Ok(scorer)
+            Ok(Box::new(scorer))
         } else {
-            Ok(box_scorer(EmptyScorer))
+            Ok(Box::new(EmptyScorer))
         }
     }
+
+    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
+        let scorer_opt = self.phrase_scorer(reader, 1.0)?;
+        if scorer_opt.is_none() {
+            return Err(does_not_match(doc));
+        }
+        let mut scorer = scorer_opt.unwrap();
+        if scorer.seek(doc) != doc {
+            return Err(does_not_match(doc));
+        }
+        let fieldnorm_reader = self.fieldnorm_reader(reader)?;
+        let fieldnorm_id = fieldnorm_reader.fieldnorm_id(doc);
+        let phrase_count = scorer.phrase_count();
+        let mut explanation = Explanation::new("Phrase Prefix Scorer", scorer.score());
+        if let Some(similarity_weight) = self.similarity_weight_opt.as_ref() {
+            explanation.add_detail(similarity_weight.explain(fieldnorm_id, phrase_count));
+        }
+        Ok(explanation)
+    }
 }
 
 #[cfg(test)]
 mod tests {
     use crate::docset::TERMINATED;
     use crate::index::Index;
-    use crate::postings::Postings;
-    use crate::query::phrase_prefix_query::PhrasePrefixScorer;
     use crate::query::{EnableScoring, PhrasePrefixQuery, Query};
     use crate::schema::{Schema, TEXT};
     use crate::{DocSet, IndexWriter, Term};
@@ -168,14 +186,14 @@ mod tests {
             .phrase_prefix_query_weight(enable_scoring)
             .unwrap()
             .unwrap();
-        let mut phrase_scorer_boxed = phrase_weight
+        let mut phrase_scorer = phrase_weight
             .phrase_scorer(searcher.segment_reader(0u32), 1.0)?
             .unwrap();
-        let phrase_scorer: &mut PhrasePrefixScorer<Box<dyn Postings>> =
-            phrase_scorer_boxed.as_any_mut().downcast_mut().unwrap();
         assert_eq!(phrase_scorer.doc(), 1);
+        assert_eq!(phrase_scorer.phrase_count(), 2);
         assert_eq!(phrase_scorer.advance(), 2);
         assert_eq!(phrase_scorer.doc(), 2);
+        assert_eq!(phrase_scorer.phrase_count(), 1);
         assert_eq!(phrase_scorer.advance(), TERMINATED);
         Ok(())
     }
@@ -195,15 +213,14 @@ mod tests {
             .phrase_prefix_query_weight(enable_scoring)
             .unwrap()
             .unwrap();
-        let mut phrase_scorer_boxed = phrase_weight
+        let mut phrase_scorer = phrase_weight
             .phrase_scorer(searcher.segment_reader(0u32), 1.0)?
             .unwrap();
-        let phrase_scorer = phrase_scorer_boxed
-            .downcast_mut::<PhrasePrefixScorer<Box<dyn Postings>>>()
-            .unwrap();
         assert_eq!(phrase_scorer.doc(), 1);
+        assert_eq!(phrase_scorer.phrase_count(), 2);
         assert_eq!(phrase_scorer.advance(), 2);
         assert_eq!(phrase_scorer.doc(), 2);
+        assert_eq!(phrase_scorer.phrase_count(), 1);
         assert_eq!(phrase_scorer.advance(), TERMINATED);
         Ok(())
     }

src/query/phrase_query/mod.rs

@@ -5,7 +5,7 @@ pub mod regex_phrase_query;
 mod regex_phrase_weight;
 
 pub use self::phrase_query::PhraseQuery;
-pub(crate) use self::phrase_scorer::intersection_exists;
+pub(crate) use self::phrase_scorer::intersection_count;
 pub use self::phrase_scorer::PhraseScorer;
 pub use self::phrase_weight::PhraseWeight;
 

src/query/phrase_query/phrase_query.rs

@@ -126,7 +126,7 @@ impl PhraseQuery {
         };
         let mut weight = PhraseWeight::new(self.phrase_terms.clone(), bm25_weight_opt);
         if self.slop > 0 {
-            weight.set_slop(self.slop);
+            weight.slop(self.slop);
         }
         Ok(weight)
     }

src/query/phrase_query/phrase_scorer.rs

@@ -1,11 +1,10 @@
 use std::cmp::Ordering;
 
-use crate::codec::standard::postings::StandardPostings;
 use crate::docset::{DocSet, SeekDangerResult, TERMINATED};
 use crate::fieldnorm::FieldNormReader;
 use crate::postings::Postings;
 use crate::query::bm25::Bm25Weight;
-use crate::query::{Explanation, Intersection, Scorer};
+use crate::query::{Intersection, Scorer};
 use crate::{DocId, Score};
 
 struct PostingsWithOffset<TPostings> {
@@ -44,7 +43,7 @@ impl<TPostings: Postings> DocSet for PostingsWithOffset<TPostings> {
     }
 }
 
-pub struct PhraseScorer<TPostings: Postings = StandardPostings> {
+pub struct PhraseScorer<TPostings: Postings> {
     intersection_docset: Intersection<PostingsWithOffset<TPostings>, PostingsWithOffset<TPostings>>,
     num_terms: usize,
     left_positions: Vec<u32>,
@@ -59,7 +58,7 @@ pub struct PhraseScorer<TPostings: Postings = StandardPostings> {
 }
 
 /// Returns true if and only if the two sorted arrays contain a common element
-pub(crate) fn intersection_exists(left: &[u32], right: &[u32]) -> bool {
+fn intersection_exists(left: &[u32], right: &[u32]) -> bool {
     let mut left_index = 0;
     let mut right_index = 0;
     while left_index < left.len() && right_index < right.len() {
@@ -80,7 +79,7 @@ pub(crate) fn intersection_exists(left: &[u32], right: &[u32]) -> bool {
     false
 }
 
-fn intersection_count(left: &[u32], right: &[u32]) -> usize {
+pub(crate) fn intersection_count(left: &[u32], right: &[u32]) -> usize {
     let mut left_index = 0;
     let mut right_index = 0;
     let mut count = 0;
@@ -403,7 +402,6 @@ impl<TPostings: Postings> PhraseScorer<TPostings> {
         scorer
     }
 
-    /// Returns the number of phrases identified in the current matching doc.
     pub fn phrase_count(&self) -> u32 {
         self.phrase_count
     }
@@ -586,17 +584,6 @@ impl<TPostings: Postings> Scorer for PhraseScorer<TPostings> {
             1.0f32
         }
     }
-
-    fn explain(&mut self) -> Explanation {
-        let doc = self.doc();
-        let phrase_count = self.phrase_count();
-        let fieldnorm_id = self.fieldnorm_reader.fieldnorm_id(doc);
-        let mut explanation = Explanation::new("Phrase Scorer", self.score());
-        if let Some(similarity_weight) = self.similarity_weight_opt.as_ref() {
-            explanation.add_detail(similarity_weight.explain(fieldnorm_id, phrase_count));
-        }
-        explanation
-    }
 }
 
 #[cfg(test)]

src/query/phrase_query/phrase_weight.rs

@@ -1,10 +1,11 @@
+use super::PhraseScorer;
 use crate::fieldnorm::FieldNormReader;
 use crate::index::SegmentReader;
-use crate::postings::TermInfo;
+use crate::postings::SegmentPostings;
 use crate::query::bm25::Bm25Weight;
 use crate::query::explanation::does_not_match;
-use crate::query::{box_scorer, EmptyScorer, Explanation, Scorer, Weight};
-use crate::schema::Term;
+use crate::query::{EmptyScorer, Explanation, Scorer, Weight};
+use crate::schema::{IndexRecordOption, Term};
 use crate::{DocId, DocSet, Score};
 
 pub struct PhraseWeight {
@@ -20,10 +21,11 @@ impl PhraseWeight {
         phrase_terms: Vec<(usize, Term)>,
         similarity_weight_opt: Option<Bm25Weight>,
     ) -> PhraseWeight {
+        let slop = 0;
         PhraseWeight {
             phrase_terms,
             similarity_weight_opt,
-            slop: 0,
+            slop,
         }
     }
 
@@ -41,52 +43,32 @@ impl PhraseWeight {
         &self,
         reader: &SegmentReader,
         boost: Score,
-    ) -> crate::Result<Option<Box<dyn Scorer>>> {
+    ) -> crate::Result<Option<PhraseScorer<SegmentPostings>>> {
         let similarity_weight_opt = self
             .similarity_weight_opt
             .as_ref()
             .map(|similarity_weight| similarity_weight.boost_by(boost));
         let fieldnorm_reader = self.fieldnorm_reader(reader)?;
-
-        if self.phrase_terms.is_empty() {
-            return Ok(None);
-        }
-        let field = self.phrase_terms[0].1.field();
-
-        if !self
-            .phrase_terms
-            .iter()
-            .all(|(_offset, term)| term.field() == field)
-        {
-            return Err(crate::TantivyError::InvalidArgument(
-                "All terms in a phrase query must belong to the same field".to_string(),
-            ));
-        }
-
-        let inverted_index_reader = reader.inverted_index(field)?;
-
-        let mut term_infos: Vec<(usize, TermInfo)> = Vec::with_capacity(self.phrase_terms.len());
-
+        let mut term_postings_list = Vec::new();
         for &(offset, ref term) in &self.phrase_terms {
-            let Some(term_info) = inverted_index_reader.get_term_info(term)? else {
+            if let Some(postings) = reader
+                .inverted_index(term.field())?
+                .read_postings(term, IndexRecordOption::WithFreqsAndPositions)?
+            {
+                term_postings_list.push((offset, postings));
+            } else {
                 return Ok(None);
-            };
-            term_infos.push((offset, term_info));
+            }
         }
-
-        let scorer = reader.codec().new_phrase_scorer_type_erased(
-            &term_infos[..],
+        Ok(Some(PhraseScorer::new(
+            term_postings_list,
             similarity_weight_opt,
             fieldnorm_reader,
             self.slop,
-            &inverted_index_reader,
-        )?;
-
-        Ok(Some(scorer))
+        )))
     }
 
-    /// Sets the slop for the given PhraseWeight.
-    pub fn set_slop(&mut self, slop: u32) {
+    pub fn slop(&mut self, slop: u32) {
         self.slop = slop;
     }
 }
@@ -94,9 +76,9 @@ impl PhraseWeight {
 impl Weight for PhraseWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         if let Some(scorer) = self.phrase_scorer(reader, boost)? {
-            Ok(scorer)
+            Ok(Box::new(scorer))
         } else {
-            Ok(box_scorer(EmptyScorer))
+            Ok(Box::new(EmptyScorer))
         }
     }
 
@@ -109,7 +91,14 @@ impl Weight for PhraseWeight {
         if scorer.seek(doc) != doc {
             return Err(does_not_match(doc));
         }
-        Ok(scorer.explain())
+        let fieldnorm_reader = self.fieldnorm_reader(reader)?;
+        let fieldnorm_id = fieldnorm_reader.fieldnorm_id(doc);
+        let phrase_count = scorer.phrase_count();
+        let mut explanation = Explanation::new("Phrase Scorer", scorer.score());
+        if let Some(similarity_weight) = self.similarity_weight_opt.as_ref() {
+            explanation.add_detail(similarity_weight.explain(fieldnorm_id, phrase_count));
+        }
+        Ok(explanation)
     }
 }
 
@@ -117,8 +106,7 @@ impl Weight for PhraseWeight {
 mod tests {
     use super::super::tests::create_index;
     use crate::docset::TERMINATED;
-    use crate::query::phrase_query::PhraseScorer;
-    use crate::query::{EnableScoring, PhraseQuery, Scorer};
+    use crate::query::{EnableScoring, PhraseQuery};
     use crate::{DocSet, Term};
 
     #[test]
@@ -133,11 +121,9 @@ mod tests {
         ]);
         let enable_scoring = EnableScoring::enabled_from_searcher(&searcher);
         let phrase_weight = phrase_query.phrase_weight(enable_scoring).unwrap();
-        let phrase_scorer_boxed: Box<dyn Scorer> = phrase_weight
+        let mut phrase_scorer = phrase_weight
             .phrase_scorer(searcher.segment_reader(0u32), 1.0)?
             .unwrap();
-        let mut phrase_scorer: Box<PhraseScorer> =
-            phrase_scorer_boxed.downcast::<PhraseScorer>().ok().unwrap();
         assert_eq!(phrase_scorer.doc(), 1);
         assert_eq!(phrase_scorer.phrase_count(), 2);
         assert_eq!(phrase_scorer.advance(), 2);

src/query/phrase_query/regex_phrase_weight.rs

@@ -6,13 +6,11 @@ use tantivy_fst::Regex;
 use super::PhraseScorer;
 use crate::fieldnorm::FieldNormReader;
 use crate::index::SegmentReader;
-use crate::postings::{LoadedPostings, Postings, TermInfo};
+use crate::postings::{LoadedPostings, Postings, SegmentPostings, TermInfo};
 use crate::query::bm25::Bm25Weight;
 use crate::query::explanation::does_not_match;
 use crate::query::union::{BitSetPostingUnion, SimpleUnion};
-use crate::query::{
-    box_scorer, AutomatonWeight, BitSetDocSet, EmptyScorer, Explanation, Scorer, Weight,
-};
+use crate::query::{AutomatonWeight, BitSetDocSet, EmptyScorer, Explanation, Scorer, Weight};
 use crate::schema::{Field, IndexRecordOption};
 use crate::{DocId, DocSet, InvertedIndexReader, Score};
 
@@ -105,9 +103,18 @@ impl RegexPhraseWeight {
         term_info: &TermInfo,
         doc_bitset: &mut BitSet,
     ) -> crate::Result<()> {
-        let mut segment_postings =
-            inverted_index.read_postings_from_terminfo(term_info, IndexRecordOption::Basic)?;
-        segment_postings.fill_bitset(doc_bitset);
+        let mut block_segment_postings = inverted_index
+            .read_block_postings_from_terminfo(term_info, IndexRecordOption::Basic)?;
+        loop {
+            let docs = block_segment_postings.docs();
+            if docs.is_empty() {
+                break;
+            }
+            for &doc in docs {
+                doc_bitset.insert(doc);
+            }
+            block_segment_postings.advance();
+        }
         Ok(())
     }
 
@@ -181,7 +188,7 @@ impl RegexPhraseWeight {
         // - Bucket 1: Terms appearing in 0.1% to 1% of documents
         // - Bucket 2: Terms appearing in 1% to 10% of documents
         // - Bucket 3: Terms appearing in more than 10% of documents
-        let mut buckets: Vec<(BitSet, Vec<Box<dyn Postings>>)> = (0..4)
+        let mut buckets: Vec<(BitSet, Vec<SegmentPostings>)> = (0..4)
             .map(|_| (BitSet::with_max_value(max_doc), Vec::new()))
             .collect();
 
@@ -190,7 +197,7 @@ impl RegexPhraseWeight {
         for term_info in term_infos {
             let mut term_posting = inverted_index
                 .read_postings_from_terminfo(term_info, IndexRecordOption::WithFreqsAndPositions)?;
-            let num_docs = u32::from(term_posting.doc_freq());
+            let num_docs = term_posting.doc_freq();
 
             if num_docs < SPARSE_TERM_DOC_THRESHOLD {
                 let current_bucket = &mut sparse_buckets[0];
@@ -264,9 +271,9 @@ impl RegexPhraseWeight {
 impl Weight for RegexPhraseWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         if let Some(scorer) = self.phrase_scorer(reader, boost)? {
-            Ok(box_scorer(scorer))
+            Ok(Box::new(scorer))
         } else {
-            Ok(box_scorer(EmptyScorer))
+            Ok(Box::new(EmptyScorer))
         }
     }
 

src/query/range_query/fast_field_range_doc_set.rs

@@ -3,6 +3,7 @@ use std::ops::RangeInclusive;
 
 use columnar::Column;
 
+use crate::docset::SeekDangerResult;
 use crate::{DocId, DocSet, TERMINATED};
 
 /// Helper to have a cursor over a vec of docids
@@ -184,6 +185,37 @@ impl<T: Send + Sync + PartialOrd + Copy + Debug + 'static> DocSet for RangeDocSe
         doc
     }
 
+    /// `seek_danger` only needs to answer whether `target` itself matches, so it does a cheap
+    /// point lookup on the column instead of scanning forward to materialize the next match (the
+    /// expensive part of a regular `seek`).
+    fn seek_danger(&mut self, target: DocId) -> SeekDangerResult {
+        // Covers `target == TERMINATED` and any target past the last doc: no match is possible.
+        if target >= self.column.num_docs() {
+            return SeekDangerResult::SeekLowerBound(TERMINATED);
+        }
+
+        if self.is_last_seek_distance_large(target) {
+            self.reset_fetch_range();
+        }
+        self.last_seek_pos_opt = Some(target);
+
+        let is_match = self
+            .column
+            .values_for_doc(target)
+            .any(|value| self.value_range.contains(&value));
+        if is_match {
+            // Leave the docset in a valid state positioned on `target`, so `doc()` returns it and a
+            // following `advance()` resumes the scan right after it.
+            self.loaded_docs.get_cleared_data().push(target);
+            self.next_fetch_start = target + 1;
+            SeekDangerResult::Found
+        } else {
+            // `target` is not in the docset. The next match is strictly greater than `target`, so
+            // `target + 1` is a valid lower bound. We may leave the docset in an invalid state.
+            SeekDangerResult::SeekLowerBound(target + 1)
+        }
+    }
+
     fn size_hint(&self) -> u32 {
         // TODO: Implement a better size hint
         self.column.num_docs() / 10
@@ -209,12 +241,148 @@ impl<T: Send + Sync + PartialOrd + Copy + Debug + 'static> DocSet for RangeDocSe
 
 #[cfg(test)]
 mod tests {
-    use std::ops::Bound;
+    use std::ops::{Bound, RangeInclusive};
 
+    use columnar::Column;
+
+    use super::RangeDocSet;
     use crate::collector::Count;
     use crate::directory::RamDirectory;
+    use crate::docset::{SeekDangerResult, TERMINATED};
     use crate::query::RangeQuery;
-    use crate::{schema, IndexBuilder, TantivyDocument, Term};
+    use crate::{schema, DocSet, Index, IndexBuilder, TantivyDocument, Term};
+
+    /// Builds a single-segment index where doc `i` carries `values_for_doc(i)` in a u64 fast
+    /// field, then returns its column so we can drive a `RangeDocSet` directly.
+    fn build_u64_column(
+        num_docs: usize,
+        values_for_doc: impl Fn(usize) -> Vec<u64>,
+    ) -> Column<u64> {
+        let mut schema_builder = schema::SchemaBuilder::new();
+        let value_field = schema_builder.add_u64_field("value", schema::FAST);
+        let index = Index::create_in_ram(schema_builder.build());
+        {
+            let mut writer = index.writer_for_tests().unwrap();
+            for i in 0..num_docs {
+                let mut doc = TantivyDocument::new();
+                for v in values_for_doc(i) {
+                    doc.add_u64(value_field, v);
+                }
+                writer.add_document(doc).unwrap();
+            }
+            writer.commit().unwrap();
+        }
+        let searcher = index.reader().unwrap().searcher();
+        assert_eq!(searcher.segment_readers().len(), 1);
+        searcher
+            .segment_reader(0)
+            .fast_fields()
+            .u64("value")
+            .unwrap()
+    }
+
+    fn range_docset(
+        value_range: RangeInclusive<u64>,
+        num_docs: usize,
+        values_for_doc: impl Fn(usize) -> Vec<u64>,
+    ) -> RangeDocSet<u64> {
+        RangeDocSet::new(value_range, build_u64_column(num_docs, values_for_doc))
+    }
+
+    #[test]
+    fn seek_danger_found_leaves_valid_state() {
+        // Even docs match the range, odd docs do not.
+        let mut docset = range_docset(0..=0, 100, |i| vec![(i % 2) as u64]);
+
+        // Matching target: `Found`, and the docset is positioned exactly on it.
+        assert_eq!(docset.seek_danger(10), SeekDangerResult::Found);
+        assert_eq!(docset.doc(), 10);
+        // A following advance resumes the scan right after the found doc.
+        assert_eq!(docset.advance(), 12);
+        assert_eq!(docset.doc(), 12);
+    }
+
+    #[test]
+    fn seek_danger_miss_returns_lower_bound() {
+        let mut docset = range_docset(0..=0, 100, |i| vec![(i % 2) as u64]);
+
+        // Odd target does not match: lower bound is strictly greater than the target and never
+        // skips past the next real match (here doc 12, the first even doc after 11).
+        match docset.seek_danger(11) {
+            SeekDangerResult::SeekLowerBound(lower_bound) => {
+                assert!(lower_bound > 11);
+                assert!(lower_bound <= 12);
+            }
+            SeekDangerResult::Found => panic!("11 should not match"),
+        }
+        // After a miss we may be in an invalid state; another seek_danger recovers it.
+        assert_eq!(docset.seek_danger(12), SeekDangerResult::Found);
+        assert_eq!(docset.doc(), 12);
+    }
+
+    #[test]
+    fn seek_danger_terminated_and_out_of_bounds() {
+        let mut docset = range_docset(0..=0, 10, |i| vec![(i % 2) as u64]);
+        assert_eq!(
+            docset.seek_danger(TERMINATED),
+            SeekDangerResult::SeekLowerBound(TERMINATED)
+        );
+        // A target past the last doc has no possible match either.
+        assert_eq!(
+            docset.seek_danger(10),
+            SeekDangerResult::SeekLowerBound(TERMINATED)
+        );
+    }
+
+    #[test]
+    fn seek_danger_multivalued() {
+        // Doc `i` holds values [i, i+1]; the range {5} matches docs 4 and 5.
+        let mut docset = range_docset(5..=5, 20, |i| vec![i as u64, i as u64 + 1]);
+
+        assert_eq!(docset.seek_danger(4), SeekDangerResult::Found);
+        assert_eq!(docset.doc(), 4);
+        assert_eq!(docset.advance(), 5);
+        // No further match after doc 5.
+        assert_eq!(docset.advance(), TERMINATED);
+    }
+
+    #[test]
+    fn seek_danger_matches_seek() {
+        // Cross-check seek_danger against the true next match for every target, on a column with a
+        // few sparse matches.
+        let matches = [3u32, 7, 50, 51, 99];
+        let num_docs = 100;
+        let values_for_doc = |i: usize| {
+            vec![if matches.contains(&(i as u32)) {
+                1u64
+            } else {
+                0u64
+            }]
+        };
+
+        for target in 0..num_docs as u32 {
+            // The first matching doc greater than or equal to `target`, i.e. what `seek` returns.
+            let expected = matches
+                .iter()
+                .copied()
+                .find(|&m| m >= target)
+                .unwrap_or(TERMINATED);
+
+            let mut danger = range_docset(1..=1, num_docs, values_for_doc);
+            match danger.seek_danger(target) {
+                SeekDangerResult::Found => {
+                    assert_eq!(expected, target, "target {target} reported Found");
+                    assert_eq!(danger.doc(), target);
+                }
+                SeekDangerResult::SeekLowerBound(lower_bound) => {
+                    assert_ne!(expected, target, "target {target} should have been Found");
+                    assert!(lower_bound > target);
+                    // The lower bound must never skip past the true next match.
+                    assert!(lower_bound <= expected);
+                }
+            }
+        }
+    }
 
     #[test]
     fn range_query_fast_optional_field_minimum() {

src/query/range_query/range_query.rs

@@ -8,12 +8,10 @@ use super::range_query_fastfield::FastFieldRangeWeight;
 use crate::index::SegmentReader;
 use crate::query::explanation::does_not_match;
 use crate::query::range_query::is_type_valid_for_fastfield_range_query;
-use crate::query::{
-    box_scorer, BitSetDocSet, ConstScorer, EnableScoring, Explanation, Query, Scorer, Weight,
-};
+use crate::query::{BitSetDocSet, ConstScorer, EnableScoring, Explanation, Query, Scorer, Weight};
 use crate::schema::{Field, IndexRecordOption, Term, Type};
 use crate::termdict::{TermDictionary, TermStreamer};
-use crate::{DocId, DocSet, Score};
+use crate::{DocId, Score};
 
 /// `RangeQuery` matches all documents that have at least one term within a defined range.
 ///
@@ -230,12 +228,21 @@ impl Weight for InvertedIndexRangeWeight {
             }
             processed_count += 1;
             let term_info = term_range.value();
-            let mut postings =
-                inverted_index.read_postings_from_terminfo(term_info, IndexRecordOption::Basic)?;
-            postings.fill_bitset(&mut doc_bitset);
+            let mut block_segment_postings = inverted_index
+                .read_block_postings_from_terminfo(term_info, IndexRecordOption::Basic)?;
+            loop {
+                let docs = block_segment_postings.docs();
+                if docs.is_empty() {
+                    break;
+                }
+                for &doc in block_segment_postings.docs() {
+                    doc_bitset.insert(doc);
+                }
+                block_segment_postings.advance();
+            }
         }
         let doc_bitset = BitSetDocSet::from(doc_bitset);
-        Ok(box_scorer(ConstScorer::new(doc_bitset, boost)))
+        Ok(Box::new(ConstScorer::new(doc_bitset, boost)))
     }
 
     fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {

src/query/range_query/range_query_fastfield.rs

@@ -13,8 +13,7 @@ use common::bounds::{BoundsRange, TransformBound};
 
 use super::fast_field_range_doc_set::RangeDocSet;
 use crate::query::{
-    box_scorer, AllScorer, ConstScorer, EmptyScorer, EnableScoring, Explanation, Query, Scorer,
-    Weight,
+    AllScorer, ConstScorer, EmptyScorer, EnableScoring, Explanation, Query, Scorer, Weight,
 };
 use crate::schema::{Type, ValueBytes};
 use crate::{DocId, DocSet, Score, SegmentReader, TantivyError, Term};
@@ -56,7 +55,7 @@ impl Weight for FastFieldRangeWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         // Check if both bounds are Bound::Unbounded
         if self.bounds.is_unbounded() {
-            return Ok(box_scorer(AllScorer::new(reader.max_doc())));
+            return Ok(Box::new(AllScorer::new(reader.max_doc())));
         }
 
         let term = self
@@ -96,7 +95,7 @@ impl Weight for FastFieldRangeWeight {
                     let Some(str_dict_column): Option<StrColumn> =
                         reader.fast_fields().str(&field_name)?
                     else {
-                        return Ok(box_scorer(EmptyScorer));
+                        return Ok(Box::new(EmptyScorer));
                     };
                     let dict = str_dict_column.dictionary();
 
@@ -108,7 +107,7 @@ impl Weight for FastFieldRangeWeight {
                     let Some((column, _col_type)) = fast_field_reader
                         .u64_lenient_for_type(Some(&[ColumnType::Str]), &field_name)?
                     else {
-                        return Ok(box_scorer(EmptyScorer));
+                        return Ok(Box::new(EmptyScorer));
                     };
                     search_on_u64_ff(column, boost, BoundsRange::new(lower_bound, upper_bound))
                 }
@@ -120,7 +119,7 @@ impl Weight for FastFieldRangeWeight {
                     let Some((column, _col_type)) = fast_field_reader
                         .u64_lenient_for_type(Some(&[ColumnType::DateTime]), &field_name)?
                     else {
-                        return Ok(box_scorer(EmptyScorer));
+                        return Ok(Box::new(EmptyScorer));
                     };
                     let bounds = bounds.map_bound(|term| term.as_date().unwrap().to_u64());
                     search_on_u64_ff(
@@ -147,7 +146,7 @@ impl Weight for FastFieldRangeWeight {
             let Some(ip_addr_column): Option<Column<Ipv6Addr>> =
                 reader.fast_fields().column_opt(&field_name)?
             else {
-                return Ok(box_scorer(EmptyScorer));
+                return Ok(Box::new(EmptyScorer));
             };
             let value_range = bound_range_inclusive_ip(
                 &bounds.lower_bound,
@@ -156,11 +155,11 @@ impl Weight for FastFieldRangeWeight {
                 ip_addr_column.max_value(),
             );
             let docset = RangeDocSet::new(value_range, ip_addr_column);
-            Ok(box_scorer(ConstScorer::new(docset, boost)))
+            Ok(Box::new(ConstScorer::new(docset, boost)))
         } else if field_type.is_str() {
             let Some(str_dict_column): Option<StrColumn> = reader.fast_fields().str(&field_name)?
             else {
-                return Ok(box_scorer(EmptyScorer));
+                return Ok(Box::new(EmptyScorer));
             };
             let dict = str_dict_column.dictionary();
 
@@ -172,7 +171,7 @@ impl Weight for FastFieldRangeWeight {
             let Some((column, _col_type)) =
                 fast_field_reader.u64_lenient_for_type(None, &field_name)?
             else {
-                return Ok(box_scorer(EmptyScorer));
+                return Ok(Box::new(EmptyScorer));
             };
             search_on_u64_ff(column, boost, BoundsRange::new(lower_bound, upper_bound))
         } else if field_type.is_bytes() {
@@ -229,7 +228,7 @@ impl Weight for FastFieldRangeWeight {
                 &field_name,
             )?
             else {
-                return Ok(box_scorer(EmptyScorer));
+                return Ok(Box::new(EmptyScorer));
             };
             search_on_u64_ff(
                 column,
@@ -270,7 +269,7 @@ fn search_on_json_numerical_field(
     let Some((column, col_type)) =
         fast_field_reader.u64_lenient_for_type(allowed_column_types, field_name)?
     else {
-        return Ok(box_scorer(EmptyScorer));
+        return Ok(Box::new(EmptyScorer));
     };
     let actual_column_type: NumericalType = col_type
         .numerical_type()
@@ -428,18 +427,18 @@ fn search_on_u64_ff(
     )
     .unwrap_or(1..=0); // empty range
     if value_range.is_empty() {
-        return Ok(box_scorer(EmptyScorer));
+        return Ok(Box::new(EmptyScorer));
     }
     if col_min_value >= *value_range.start() && col_max_value <= *value_range.end() {
         // all values in the column are within the range.
         if column.index.get_cardinality() == Cardinality::Full {
             if boost != 1.0f32 {
-                return Ok(box_scorer(ConstScorer::new(
+                return Ok(Box::new(ConstScorer::new(
                     AllScorer::new(column.num_docs()),
                     boost,
                 )));
             } else {
-                return Ok(box_scorer(AllScorer::new(column.num_docs())));
+                return Ok(Box::new(AllScorer::new(column.num_docs())));
             }
         } else {
             // TODO Make it a field presence request for that specific column
@@ -447,7 +446,7 @@ fn search_on_u64_ff(
     }
 
     let docset = RangeDocSet::new(value_range, column);
-    Ok(box_scorer(ConstScorer::new(docset, boost)))
+    Ok(Box::new(ConstScorer::new(docset, boost)))
 }
 
 /// Returns true if the type maps to a u64 fast field

src/query/scorer.rs

@@ -1,11 +1,9 @@
-use std::mem::{transmute_copy, ManuallyDrop};
 use std::ops::DerefMut;
 
 use downcast_rs::impl_downcast;
 
 use crate::docset::DocSet;
-use crate::query::Explanation;
-use crate::{DocId, Score, TERMINATED};
+use crate::Score;
 
 /// Scored set of documents matching a query within a specific segment.
 ///
@@ -15,53 +13,6 @@ pub trait Scorer: downcast_rs::Downcast + DocSet + 'static {
     ///
     /// This method will perform a bit of computation and is not cached.
     fn score(&mut self) -> Score;
-
-    /// Calls `callback` with all of the `(doc, score)` for which score
-    /// is exceeding a given threshold.
-    ///
-    /// This method is useful for the TopDocs collector.
-    /// For all docsets, the blanket implementation has the benefit
-    /// of prefiltering (doc, score) pairs, avoiding the
-    /// virtual dispatch cost.
-    ///
-    /// More importantly, it makes it possible for scorers to implement
-    /// important optimization (e.g. BlockWAND for union).
-    fn for_each_pruning(
-        &mut self,
-        threshold: Score,
-        callback: &mut dyn FnMut(DocId, Score) -> Score,
-    ) {
-        for_each_pruning_scorer_default_impl(self, threshold, callback);
-    }
-
-    /// Calls `callback` with all of the `(doc, score)` in the scorer.
-    fn for_each(&mut self, callback: &mut dyn FnMut(DocId, Score)) {
-        let mut doc = self.doc();
-        while doc != TERMINATED {
-            callback(doc, self.score());
-            doc = self.advance();
-        }
-    }
-
-    /// Returns an explanation for the score of the current document.
-    fn explain(&mut self) -> Explanation {
-        let score = self.score();
-        let name = std::any::type_name_of_val(self);
-        Explanation::new(name, score)
-    }
-}
-
-/// Boxes a scorer. Prefer this to Box::new as it avoids double boxing
-/// when TScorer is already a Box<dyn Scorer>.
-pub fn box_scorer<TScorer: Scorer>(scorer: TScorer) -> Box<dyn Scorer> {
-    if std::any::TypeId::of::<TScorer>() == std::any::TypeId::of::<Box<dyn Scorer>>() {
-        unsafe {
-            let forget_me = ManuallyDrop::new(scorer);
-            transmute_copy::<TScorer, Box<dyn Scorer>>(&forget_me)
-        }
-    } else {
-        Box::new(scorer)
-    }
 }
 
 impl_downcast!(Scorer);
@@ -71,41 +22,4 @@ impl Scorer for Box<dyn Scorer> {
     fn score(&mut self) -> Score {
         self.deref_mut().score()
     }
-
-    fn for_each_pruning(
-        &mut self,
-        threshold: Score,
-        callback: &mut dyn FnMut(DocId, Score) -> Score,
-    ) {
-        self.deref_mut().for_each_pruning(threshold, callback);
-    }
-
-    fn for_each(&mut self, callback: &mut dyn FnMut(DocId, Score)) {
-        self.deref_mut().for_each(callback);
-    }
-}
-
-/// Calls `callback` with all of the `(doc, score)` for which score
-/// is exceeding a given threshold.
-///
-/// This method is useful for the [`TopDocs`](crate::collector::TopDocs) collector.
-/// For all docsets, the blanket implementation has the benefit
-/// of prefiltering (doc, score) pairs, avoiding the
-/// virtual dispatch cost.
-///
-/// More importantly, it makes it possible for scorers to implement
-/// important optimization (e.g. BlockWAND for union).
-pub(crate) fn for_each_pruning_scorer_default_impl<TScorer: Scorer + ?Sized>(
-    scorer: &mut TScorer,
-    mut threshold: Score,
-    callback: &mut dyn FnMut(DocId, Score) -> Score,
-) {
-    let mut doc = scorer.doc();
-    while doc != TERMINATED {
-        let score = scorer.score();
-        if score > threshold {
-            threshold = callback(doc, score);
-        }
-        doc = scorer.advance();
-    }
 }

src/query/term_query/mod.rs

@@ -4,9 +4,9 @@ mod term_weight;
 
 pub use self::term_query::TermQuery;
 pub use self::term_scorer::TermScorer;
-
 #[cfg(test)]
 mod tests {
+
     use crate::collector::TopDocs;
     use crate::docset::DocSet;
     use crate::postings::compression::COMPRESSION_BLOCK_SIZE;

src/query/term_query/term_scorer.rs

@@ -1,27 +1,23 @@
-use crate::codec::postings::{PostingsCodec, PostingsWithBlockMax};
-use crate::codec::{Codec, StandardCodec};
 use crate::docset::DocSet;
 use crate::fieldnorm::FieldNormReader;
-use crate::postings::Postings;
+use crate::postings::{BlockSegmentPostings, FreqReadingOption, Postings, SegmentPostings};
 use crate::query::bm25::Bm25Weight;
 use crate::query::{Explanation, Scorer};
 use crate::{DocId, Score};
 
 #[derive(Clone)]
-pub struct TermScorer<
-    TPostings: Postings = <<StandardCodec as Codec>::PostingsCodec as PostingsCodec>::Postings,
-> {
-    postings: TPostings,
+pub struct TermScorer {
+    postings: SegmentPostings,
     fieldnorm_reader: FieldNormReader,
     similarity_weight: Bm25Weight,
 }
 
-impl<TPostings: Postings> TermScorer<TPostings> {
+impl TermScorer {
     pub fn new(
-        postings: TPostings,
+        postings: SegmentPostings,
         fieldnorm_reader: FieldNormReader,
         similarity_weight: Bm25Weight,
-    ) -> TermScorer<TPostings> {
+    ) -> TermScorer {
         TermScorer {
             postings,
             fieldnorm_reader,
@@ -29,35 +25,10 @@ impl<TPostings: Postings> TermScorer<TPostings> {
         }
     }
 
-    pub fn term_freq(&self) -> u32 {
-        self.postings.term_freq()
+    pub(crate) fn seek_block(&mut self, target_doc: DocId) {
+        self.postings.block_cursor.seek_block(target_doc);
     }
 
-    pub fn fieldnorm_id(&self) -> u8 {
-        self.fieldnorm_reader.fieldnorm_id(self.doc())
-    }
-
-    pub fn max_score(&self) -> Score {
-        self.similarity_weight.max_score()
-    }
-}
-
-impl<TPostingsWithBlockMax: PostingsWithBlockMax> TermScorer<TPostingsWithBlockMax> {
-    pub(crate) fn last_doc_in_block(&self) -> DocId {
-        self.postings.last_doc_in_block()
-    }
-
-    /// Advances the term scorer to the block containing target_doc and returns
-    /// an upperbound for the score all of the documents in the block.
-    /// (BlockMax). This score is not guaranteed to be the
-    /// effective maximum score of the block.
-    pub(crate) fn seek_block_max(&mut self, target_doc: DocId) -> Score {
-        self.postings
-            .seek_block_max(target_doc, &self.fieldnorm_reader, &self.similarity_weight)
-    }
-}
-
-impl TermScorer {
     #[cfg(test)]
     pub fn create_for_test(
         doc_and_tfs: &[(DocId, u32)],
@@ -73,15 +44,75 @@ impl TermScorer {
                 .unwrap_or(0u32)
                 < fieldnorms.len() as u32
         );
-        type SegmentPostings = <<StandardCodec as Codec>::PostingsCodec as PostingsCodec>::Postings;
-        let segment_postings: SegmentPostings =
+        let segment_postings =
             SegmentPostings::create_from_docs_and_tfs(doc_and_tfs, Some(fieldnorms));
         let fieldnorm_reader = FieldNormReader::for_test(fieldnorms);
         TermScorer::new(segment_postings, fieldnorm_reader, similarity_weight)
     }
+
+    /// See `FreqReadingOption`.
+    pub(crate) fn freq_reading_option(&self) -> FreqReadingOption {
+        self.postings.block_cursor.freq_reading_option()
+    }
+
+    /// Returns the maximum score for the current block.
+    ///
+    /// In some rare case, the result may not be exact. In this case a lower value is returned,
+    /// (and may lead us to return a lesser document).
+    ///
+    /// At index time, we store the (fieldnorm_id, term frequency) pair that maximizes the
+    /// score assuming the average fieldnorm computed on this segment.
+    ///
+    /// Though extremely rare, it is theoretically possible that the actual average fieldnorm
+    /// is different enough from the current segment average fieldnorm that the maximum over a
+    /// specific is achieved on a different document.
+    ///
+    /// (The result is on the other hand guaranteed to be correct if there is only one segment).
+    pub fn block_max_score(&mut self) -> Score {
+        self.postings
+            .block_cursor
+            .block_max_score(&self.fieldnorm_reader, &self.similarity_weight)
+    }
+
+    pub fn term_freq(&self) -> u32 {
+        self.postings.term_freq()
+    }
+
+    pub fn fieldnorm_id(&self) -> u8 {
+        self.fieldnorm_reader.fieldnorm_id(self.doc())
+    }
+
+    pub fn explain(&self) -> Explanation {
+        let fieldnorm_id = self.fieldnorm_id();
+        let term_freq = self.term_freq();
+        self.similarity_weight.explain(fieldnorm_id, term_freq)
+    }
+
+    pub fn max_score(&self) -> Score {
+        self.similarity_weight.max_score()
+    }
+
+    pub fn last_doc_in_block(&self) -> DocId {
+        self.postings.block_cursor.skip_reader().last_doc_in_block()
+    }
+
+    /// Returns a mutable reference to the underlying block cursor.
+    pub(crate) fn block_cursor(&mut self) -> &mut BlockSegmentPostings {
+        &mut self.postings.block_cursor
+    }
+
+    /// Returns a reference to the fieldnorm reader for batch lookups.
+    pub(crate) fn fieldnorm_reader(&self) -> &FieldNormReader {
+        &self.fieldnorm_reader
+    }
+
+    /// Returns a reference to the BM25 weight for batch score computation.
+    pub(crate) fn bm25_weight(&self) -> &Bm25Weight {
+        &self.similarity_weight
+    }
 }
 
-impl<TPostings: Postings> DocSet for TermScorer<TPostings> {
+impl DocSet for TermScorer {
     #[inline]
     fn advance(&mut self) -> DocId {
         self.postings.advance()
@@ -109,19 +140,13 @@ impl<TPostings: Postings> DocSet for TermScorer<TPostings> {
     // and do not have access to x86 to investigate.
 }
 
-impl<TPostings: Postings> Scorer for TermScorer<TPostings> {
+impl Scorer for TermScorer {
     #[inline]
     fn score(&mut self) -> Score {
         let fieldnorm_id = self.fieldnorm_id();
         let term_freq = self.term_freq();
         self.similarity_weight.score(fieldnorm_id, term_freq)
     }
-
-    fn explain(&mut self) -> Explanation {
-        let fieldnorm_id = self.fieldnorm_id();
-        let term_freq = self.term_freq();
-        self.similarity_weight.explain(fieldnorm_id, term_freq)
-    }
 }
 
 #[cfg(test)]
@@ -130,7 +155,7 @@ mod tests {
 
     use crate::index::SegmentId;
     use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
-    use crate::indexer::NoMergePolicy;
+    use crate::merge_policy::NoMergePolicy;
     use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
     use crate::query::term_query::TermScorer;
     use crate::query::{Bm25Weight, EnableScoring, Scorer, TermQuery};
@@ -151,7 +176,7 @@ mod tests {
         crate::assert_nearly_equals!(max_scorer, 1.3990127);
         assert_eq!(term_scorer.doc(), 2);
         assert_eq!(term_scorer.term_freq(), 3);
-        assert_nearly_equals!(term_scorer.seek_block_max(2), 1.3676447);
+        assert_nearly_equals!(term_scorer.block_max_score(), 1.3676447);
         assert_nearly_equals!(term_scorer.score(), 1.0892314);
         assert_eq!(term_scorer.advance(), 3);
         assert_eq!(term_scorer.doc(), 3);
@@ -166,9 +191,9 @@ mod tests {
     }
 
     #[test]
-    fn test_term_scorer_shallow_advance() {
+    fn test_term_scorer_shallow_advance() -> crate::Result<()> {
         let bm25_weight = Bm25Weight::for_one_term(300, 1024, 10.0);
-        let mut doc_and_tfs = Vec::new();
+        let mut doc_and_tfs = vec![];
         for i in 0u32..300u32 {
             let doc = i * 10;
             doc_and_tfs.push((doc, 1u32 + doc % 3u32));
@@ -176,10 +201,11 @@ mod tests {
         let fieldnorms: Vec<u32> = std::iter::repeat_n(10u32, 3_000).collect();
         let mut term_scorer = TermScorer::create_for_test(&doc_and_tfs, &fieldnorms, bm25_weight);
         assert_eq!(term_scorer.doc(), 0u32);
-        term_scorer.seek_block_max(1289);
+        term_scorer.seek_block(1289);
         assert_eq!(term_scorer.doc(), 0u32);
         term_scorer.seek(1289);
         assert_eq!(term_scorer.doc(), 1290);
+        Ok(())
     }
 
     proptest! {
@@ -213,7 +239,7 @@ mod tests {
 
          let docs: Vec<DocId> = (0..term_doc_freq).map(|doc| doc as DocId).collect();
          for block in docs.chunks(COMPRESSION_BLOCK_SIZE) {
-             let block_max_score: Score = term_scorer.seek_block_max(0);
+             let block_max_score: Score = term_scorer.block_max_score();
              let mut block_max_score_computed: Score = 0.0;
              for &doc in block {
                 assert_eq!(term_scorer.doc(), doc);
@@ -241,24 +267,25 @@ mod tests {
         let fieldnorms: Vec<u32> = std::iter::repeat_n(20u32, 300).collect();
         let bm25_weight = Bm25Weight::for_one_term(10, 129, 20.0);
         let mut docs = TermScorer::create_for_test(&doc_tfs[..], &fieldnorms[..], bm25_weight);
-        assert_nearly_equals!(docs.seek_block_max(0), 2.5161593);
-        assert_nearly_equals!(docs.seek_block_max(135), 3.4597192);
+        assert_nearly_equals!(docs.block_max_score(), 2.5161593);
+        docs.seek_block(135);
+        assert_nearly_equals!(docs.block_max_score(), 3.4597192);
+        docs.seek_block(256);
         // the block is not loaded yet.
-        assert_nearly_equals!(docs.seek_block_max(256), 5.2971773);
+        assert_nearly_equals!(docs.block_max_score(), 5.2971773);
         assert_eq!(256, docs.seek(256));
-        assert_nearly_equals!(docs.seek_block_max(256), 3.9539647);
+        assert_nearly_equals!(docs.block_max_score(), 3.9539647);
     }
 
-    fn test_block_wand_aux(term_query: &TermQuery, searcher: &Searcher) {
-        let term_weight = term_query
-            .specialized_weight(EnableScoring::enabled_from_searcher(searcher))
-            .unwrap();
+    fn test_block_wand_aux(term_query: &TermQuery, searcher: &Searcher) -> crate::Result<()> {
+        let term_weight =
+            term_query.specialized_weight(EnableScoring::enabled_from_searcher(searcher))?;
         for reader in searcher.segment_readers() {
             let mut block_max_scores = vec![];
             let mut block_max_scores_b = vec![];
             let mut docs = vec![];
             {
-                let mut term_scorer = term_weight.term_scorer_for_test(reader, 1.0).unwrap();
+                let mut term_scorer = term_weight.term_scorer_for_test(reader, 1.0)?.unwrap();
                 while term_scorer.doc() != TERMINATED {
                     let mut score = term_scorer.score();
                     docs.push(term_scorer.doc());
@@ -272,10 +299,10 @@ mod tests {
                 }
             }
             {
-                let mut term_scorer = term_weight.term_scorer_for_test(reader, 1.0).unwrap();
+                let mut term_scorer = term_weight.term_scorer_for_test(reader, 1.0)?.unwrap();
                 for d in docs {
-                    let block_max_score = term_scorer.seek_block_max(d);
-                    block_max_scores_b.push(block_max_score);
+                    term_scorer.seek_block(d);
+                    block_max_scores_b.push(term_scorer.block_max_score());
                 }
             }
             for (l, r) in block_max_scores
@@ -286,18 +313,18 @@ mod tests {
                 assert_nearly_equals!(l, r);
             }
         }
+        Ok(())
     }
 
     #[ignore]
     #[test]
-    fn test_block_wand_long_test() {
+    fn test_block_wand_long_test() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut writer: IndexWriter = index
-            .writer_with_num_threads(3, 3 * MEMORY_BUDGET_NUM_BYTES_MIN)
-            .unwrap();
+        let mut writer: IndexWriter =
+            index.writer_with_num_threads(3, 3 * MEMORY_BUDGET_NUM_BYTES_MIN)?;
         use rand::Rng;
         let mut rng = rand::rng();
         writer.set_merge_policy(Box::new(NoMergePolicy));
@@ -305,15 +332,15 @@ mod tests {
             let term_freq = rng.random_range(1..10000);
             let words: Vec<&str> = std::iter::repeat_n("bbbb", term_freq).collect();
             let text = words.join(" ");
-            writer.add_document(doc!(text_field=>text)).unwrap();
+            writer.add_document(doc!(text_field=>text))?;
         }
-        writer.commit().unwrap();
+        writer.commit()?;
         let term_query = TermQuery::new(
             Term::from_field_text(text_field, "bbbb"),
             IndexRecordOption::WithFreqs,
         );
         let segment_ids: Vec<SegmentId>;
-        let reader = index.reader().unwrap();
+        let reader = index.reader()?;
         {
             let searcher = reader.searcher();
             segment_ids = searcher
@@ -321,14 +348,15 @@ mod tests {
                 .iter()
                 .map(|segment| segment.segment_id())
                 .collect();
-            test_block_wand_aux(&term_query, &searcher);
+            test_block_wand_aux(&term_query, &searcher)?;
         }
         writer.merge(&segment_ids[..]).wait().unwrap();
         {
-            reader.reload().unwrap();
+            reader.reload()?;
             let searcher = reader.searcher();
             assert_eq!(searcher.segment_readers().len(), 1);
-            test_block_wand_aux(&term_query, &searcher);
+            test_block_wand_aux(&term_query, &searcher)?;
         }
+        Ok(())
     }
 }

src/query/term_query/term_weight.rs

@@ -1,11 +1,12 @@
+use super::term_scorer::TermScorer;
 use crate::docset::{DocSet, COLLECT_BLOCK_BUFFER_LEN};
 use crate::fieldnorm::FieldNormReader;
 use crate::index::SegmentReader;
+use crate::postings::SegmentPostings;
 use crate::query::bm25::Bm25Weight;
 use crate::query::explanation::does_not_match;
-use crate::query::term_query::TermScorer;
-use crate::query::weight::for_each_docset_buffered;
-use crate::query::{box_scorer, AllScorer, AllWeight, EmptyScorer, Explanation, Scorer, Weight};
+use crate::query::weight::{for_each_docset_buffered, for_each_scorer};
+use crate::query::{AllScorer, AllWeight, EmptyScorer, Explanation, Scorer, Weight};
 use crate::schema::IndexRecordOption;
 use crate::{DocId, Score, TantivyError, Term};
 
@@ -17,7 +18,7 @@ pub struct TermWeight {
 }
 
 enum TermOrEmptyOrAllScorer {
-    TermScorer(Box<dyn Scorer>),
+    TermScorer(Box<TermScorer>),
     Empty,
     AllMatch(AllScorer),
 }
@@ -26,8 +27,8 @@ impl TermOrEmptyOrAllScorer {
     pub fn into_boxed_scorer(self) -> Box<dyn Scorer> {
         match self {
             TermOrEmptyOrAllScorer::TermScorer(scorer) => scorer,
-            TermOrEmptyOrAllScorer::Empty => box_scorer(EmptyScorer),
-            TermOrEmptyOrAllScorer::AllMatch(scorer) => box_scorer(scorer),
+            TermOrEmptyOrAllScorer::Empty => Box::new(EmptyScorer),
+            TermOrEmptyOrAllScorer::AllMatch(scorer) => Box::new(scorer),
         }
     }
 }
@@ -43,7 +44,6 @@ impl Weight for TermWeight {
                 if term_scorer.doc() > doc || term_scorer.seek(doc) != doc {
                     return Err(does_not_match(doc));
                 }
-                let mut term_scorer = term_scorer.downcast::<TermScorer>().ok().unwrap();
                 let mut explanation = term_scorer.explain();
                 explanation.add_context(format!("Term={:?}", self.term,));
                 Ok(explanation)
@@ -73,11 +73,11 @@ impl Weight for TermWeight {
     ) -> crate::Result<()> {
         match self.specialized_scorer(reader, 1.0)? {
             TermOrEmptyOrAllScorer::TermScorer(mut term_scorer) => {
-                term_scorer.for_each(callback);
+                for_each_scorer(&mut *term_scorer, callback);
             }
             TermOrEmptyOrAllScorer::Empty => {}
             TermOrEmptyOrAllScorer::AllMatch(mut all_scorer) => {
-                all_scorer.for_each(callback);
+                for_each_scorer(&mut all_scorer, callback);
             }
         }
         Ok(())
@@ -124,9 +124,11 @@ impl Weight for TermWeight {
         let specialized_scorer = self.specialized_scorer(reader, 1.0)?;
         match specialized_scorer {
             TermOrEmptyOrAllScorer::TermScorer(term_scorer) => {
-                reader
-                    .codec()
-                    .for_each_pruning(threshold, term_scorer, callback);
+                crate::query::boolean_query::block_wand_single_scorer(
+                    *term_scorer,
+                    threshold,
+                    callback,
+                );
             }
             TermOrEmptyOrAllScorer::Empty => {}
             TermOrEmptyOrAllScorer::AllMatch(_) => {
@@ -166,15 +168,12 @@ impl TermWeight {
         &self,
         reader: &SegmentReader,
         boost: Score,
-    ) -> Option<super::TermScorer> {
-        let scorer = self.specialized_scorer(reader, boost).unwrap();
-        match scorer {
-            TermOrEmptyOrAllScorer::TermScorer(scorer) => {
-                let term_scorer = scorer.downcast::<super::TermScorer>().ok()?;
-                Some(*term_scorer)
-            }
+    ) -> crate::Result<Option<TermScorer>> {
+        let scorer = self.specialized_scorer(reader, boost)?;
+        Ok(match scorer {
+            TermOrEmptyOrAllScorer::TermScorer(scorer) => Some(*scorer),
             _ => None,
-        }
+        })
     }
 
     fn specialized_scorer(
@@ -197,16 +196,14 @@ impl TermWeight {
             )));
         }
 
+        let segment_postings: SegmentPostings =
+            inverted_index.read_postings_from_terminfo(&term_info, self.index_record_option)?;
+
         let fieldnorm_reader = self.fieldnorm_reader(reader)?;
         let similarity_weight = self.similarity_weight.boost_by(boost);
-        let term_scorer = inverted_index.new_term_scorer(
-            &term_info,
-            self.index_record_option,
-            fieldnorm_reader,
-            similarity_weight,
-        )?;
-
-        Ok(TermOrEmptyOrAllScorer::TermScorer(term_scorer))
+        Ok(TermOrEmptyOrAllScorer::TermScorer(Box::new(
+            TermScorer::new(segment_postings, fieldnorm_reader, similarity_weight),
+        )))
     }
 
     fn fieldnorm_reader(&self, segment_reader: &SegmentReader) -> crate::Result<FieldNormReader> {

src/query/union/bitset_union.rs

@@ -1,7 +1,7 @@
 use std::cell::RefCell;
 
 use crate::docset::DocSet;
-use crate::postings::{DocFreq, Postings};
+use crate::postings::Postings;
 use crate::query::BitSetDocSet;
 use crate::DocId;
 
@@ -16,9 +16,6 @@ pub struct BitSetPostingUnion<TDocSet> {
     docsets: RefCell<Vec<TDocSet>>,
     /// The already unionized BitSet of the docsets
     bitset: BitSetDocSet,
-    /// The total number of documents in the union (regardless of the position we are in the
-    /// bitset).
-    doc_freq: u32,
 }
 
 impl<TDocSet: DocSet> BitSetPostingUnion<TDocSet> {
@@ -26,11 +23,9 @@ impl<TDocSet: DocSet> BitSetPostingUnion<TDocSet> {
         docsets: Vec<TDocSet>,
         bitset: BitSetDocSet,
     ) -> BitSetPostingUnion<TDocSet> {
-        let doc_freq = bitset.doc_freq();
         BitSetPostingUnion {
             docsets: RefCell::new(docsets),
             bitset,
-            doc_freq,
         }
     }
 }
@@ -51,10 +46,6 @@ impl<TDocSet: Postings> Postings for BitSetPostingUnion<TDocSet> {
         term_freq
     }
 
-    fn has_freq(&self) -> bool {
-        true
-    }
-
     fn append_positions_with_offset(&mut self, offset: u32, output: &mut Vec<u32>) {
         let curr_doc = self.bitset.doc();
         let mut docsets = self.docsets.borrow_mut();
@@ -73,10 +64,6 @@ impl<TDocSet: Postings> Postings for BitSetPostingUnion<TDocSet> {
         output.sort_unstable();
         output.dedup();
     }
-
-    fn doc_freq(&self) -> DocFreq {
-        DocFreq::Exact(self.doc_freq)
-    }
 }
 
 impl<TDocSet: DocSet> DocSet for BitSetPostingUnion<TDocSet> {

src/query/union/buffered_union.rs

@@ -31,7 +31,7 @@ where P: FnMut(&mut T) -> bool {
 /// Creates a `DocSet` that iterate through the union of two or more `DocSet`s.
 pub struct BufferedUnionScorer<TScorer, TScoreCombiner = DoNothingCombiner> {
     /// Active scorers (already filtered of `TERMINATED`).
-    scorers: Vec<TScorer>,
+    docsets: Vec<TScorer>,
     /// Sliding window presence map for upcoming docs.
     ///
     /// There are `HORIZON_NUM_TINYBITSETS` buckets, each covering
@@ -46,8 +46,6 @@ pub struct BufferedUnionScorer<TScorer, TScoreCombiner = DoNothingCombiner> {
     /// hit the same doc within the buffered window.
     scores: Box<[TScoreCombiner; HORIZON as usize]>,
     /// Start doc ID (inclusive) of the current sliding window.
-    /// None if the window is not loaded yet. This is true for a freshly created
-    /// BufferedUnionScorer.
     window_start_doc: DocId,
     /// Current doc ID of the union.
     doc: DocId,
@@ -83,81 +81,51 @@ fn refill<TScorer: Scorer, TScoreCombiner: ScoreCombiner>(
 }
 
 impl<TScorer: Scorer, TScoreCombiner: ScoreCombiner> BufferedUnionScorer<TScorer, TScoreCombiner> {
-    /// Returns the underlying scorers in the union.
-    pub fn into_scorers(self) -> Vec<TScorer> {
-        self.scorers
-    }
-
-    /// Accessor for the underlying scorers in the union.
-    pub fn scorers(&self) -> &[TScorer] {
-        &self.scorers[..]
-    }
-
     /// num_docs is the number of documents in the segment.
     pub(crate) fn build(
         docsets: Vec<TScorer>,
         score_combiner_fn: impl FnOnce() -> TScoreCombiner,
         num_docs: u32,
     ) -> BufferedUnionScorer<TScorer, TScoreCombiner> {
-        let score_combiner = score_combiner_fn();
-        let mut non_empty_docsets: Vec<TScorer> = docsets
+        let non_empty_docsets: Vec<TScorer> = docsets
             .into_iter()
             .filter(|docset| docset.doc() != TERMINATED)
             .collect();
-
-        let first_doc: DocId = non_empty_docsets
-            .iter()
-            .map(|docset| docset.doc())
-            .min()
-            .unwrap_or(TERMINATED);
-        let mut score_combiner_cloned = score_combiner;
-        let mut i = 0;
-        while i < non_empty_docsets.len() {
-            let should_remove_docset: bool = {
-                let non_empty_docset = &mut non_empty_docsets[i];
-                if non_empty_docset.doc() != first_doc {
-                    false
-                } else {
-                    score_combiner_cloned.update(non_empty_docset);
-                    non_empty_docsets[i].advance() == TERMINATED
-                }
-            };
-            if should_remove_docset {
-                non_empty_docsets.swap_remove(i);
-            } else {
-                i += 1;
-            }
-        }
-        let first_score: Score = score_combiner_cloned.score();
-        BufferedUnionScorer {
-            scorers: non_empty_docsets,
+        let mut union = BufferedUnionScorer {
+            docsets: non_empty_docsets,
             bitsets: Box::new([TinySet::empty(); HORIZON_NUM_TINYBITSETS]),
-            scores: Box::new([score_combiner; HORIZON as usize]),
+            scores: Box::new([score_combiner_fn(); HORIZON as usize]),
             bucket_idx: HORIZON_NUM_TINYBITSETS,
-            // That way we will be detected as outside the window,
-            window_start_doc: u32::MAX - HORIZON,
-            doc: first_doc,
-            score: first_score,
+            window_start_doc: 0,
+            doc: 0,
+            score: 0.0,
             num_docs,
+        };
+        if union.refill() {
+            union.advance();
+        } else {
+            union.doc = TERMINATED;
         }
+        union
     }
 
     fn refill(&mut self) -> bool {
-        let Some(min_doc) = self.scorers.iter().map(DocSet::doc).min() else {
-            return false;
-        };
-        // Reset the sliding window to start at the smallest doc
-        // across all scorers and prebuffer within the horizon.
-        self.window_start_doc = min_doc;
-        self.bucket_idx = 0;
-        self.doc = min_doc;
-        refill(
-            &mut self.scorers,
-            &mut self.bitsets,
-            &mut self.scores,
-            min_doc,
-        );
-        true
+        if let Some(min_doc) = self.docsets.iter().map(DocSet::doc).min() {
+            // Reset the sliding window to start at the smallest doc
+            // across all scorers and prebuffer within the horizon.
+            self.window_start_doc = min_doc;
+            self.bucket_idx = 0;
+            self.doc = min_doc;
+            refill(
+                &mut self.docsets,
+                &mut self.bitsets,
+                &mut self.scores,
+                min_doc,
+            );
+            true
+        } else {
+            false
+        }
     }
 
     #[inline]
@@ -179,7 +147,6 @@ impl<TScorer: Scorer, TScoreCombiner: ScoreCombiner> BufferedUnionScorer<TScorer
 
     fn is_in_horizon(&self, target: DocId) -> bool {
         // wrapping_sub, because target may be < window_start_doc
-        // in particular during initialization.
         let gap = target.wrapping_sub(self.window_start_doc);
         gap < HORIZON
     }
@@ -249,10 +216,11 @@ where
         if self.doc >= target {
             return self.doc;
         }
-        if self.is_in_horizon(target) {
+        let gap = target - self.window_start_doc;
+        if gap < HORIZON {
             // Our value is within the buffered horizon.
+
             // Skipping to corresponding bucket.
-            let gap = target.wrapping_sub(self.window_start_doc);
             let new_bucket_idx = gap as usize / 64;
             for obsolete_tinyset in &mut self.bitsets[self.bucket_idx..new_bucket_idx] {
                 obsolete_tinyset.clear();
@@ -271,14 +239,16 @@ where
             doc
         } else {
             // clear the buffered info.
-            self.bitsets.fill(TinySet::empty());
+            for obsolete_tinyset in self.bitsets.iter_mut() {
+                *obsolete_tinyset = TinySet::empty();
+            }
             for score_combiner in self.scores.iter_mut() {
                 score_combiner.clear();
             }
 
             // The target is outside of the buffered horizon.
             // advance all docsets to a doc >= to the target.
-            unordered_drain_filter(&mut self.scorers, |docset| {
+            unordered_drain_filter(&mut self.docsets, |docset| {
                 if docset.doc() < target {
                     docset.seek(target);
                 }
@@ -315,8 +285,8 @@ where
         let mut is_hit = false;
         let mut min_new_target = TERMINATED;
 
-        for scorer in self.scorers.iter_mut() {
-            match scorer.seek_danger(target) {
+        for docset in self.docsets.iter_mut() {
+            match docset.seek_danger(target) {
                 SeekDangerResult::Found => {
                     is_hit = true;
                     break;
@@ -345,11 +315,11 @@ where
     }
 
     fn size_hint(&self) -> u32 {
-        estimate_union(self.scorers.iter().map(DocSet::size_hint), self.num_docs)
+        estimate_union(self.docsets.iter().map(DocSet::size_hint), self.num_docs)
     }
 
     fn cost(&self) -> u64 {
-        self.scorers.iter().map(|docset| docset.cost()).sum()
+        self.docsets.iter().map(|docset| docset.cost()).sum()
     }
 
     // TODO Also implement `count` with deletes efficiently.
@@ -357,17 +327,21 @@ where
         if self.doc == TERMINATED {
             return 0;
         }
-        let mut count = 1 + self.bitsets[self.bucket_idx..HORIZON_NUM_TINYBITSETS]
+        let mut count = self.bitsets[self.bucket_idx..HORIZON_NUM_TINYBITSETS]
             .iter()
-            .copied()
-            .map(TinySet::len)
-            .sum::<u32>();
+            .map(|bitset| bitset.len())
+            .sum::<u32>()
+            + 1;
+        for bitset in self.bitsets.iter_mut() {
+            bitset.clear();
+        }
         while self.refill() {
-            count += self.bitsets.iter().copied().map(TinySet::len).sum::<u32>();
-            self.bitsets.fill(TinySet::empty());
+            count += self.bitsets.iter().map(|bitset| bitset.len()).sum::<u32>();
+            for bitset in self.bitsets.iter_mut() {
+                bitset.clear();
+            }
         }
         self.bucket_idx = HORIZON_NUM_TINYBITSETS;
-        self.doc = TERMINATED;
         count
     }
 }

src/query/union/simple_union.rs

@@ -1,5 +1,5 @@
 use crate::docset::{DocSet, TERMINATED};
-use crate::postings::{DocFreq, Postings};
+use crate::postings::Postings;
 use crate::DocId;
 
 /// A `SimpleUnion` is a `DocSet` that is the union of multiple `DocSet`.
@@ -56,22 +56,6 @@ impl<TDocSet: Postings> Postings for SimpleUnion<TDocSet> {
         term_freq
     }
 
-    fn has_freq(&self) -> bool {
-        true
-    }
-
-    /// We do not know the actual document frequency, so we return
-    /// the maximum document frequency of the docsets.
-    fn doc_freq(&self) -> DocFreq {
-        let approximate_doc_freq = self
-            .docsets
-            .iter()
-            .map(|docset| u32::from(docset.doc_freq()))
-            .max()
-            .unwrap_or(0u32);
-        DocFreq::Approximate(approximate_doc_freq)
-    }
-
     fn append_positions_with_offset(&mut self, offset: u32, output: &mut Vec<u32>) {
         for docset in &mut self.docsets {
             let doc = docset.doc();

src/query/weight.rs

@@ -1,9 +1,21 @@
 use super::Scorer;
 use crate::docset::COLLECT_BLOCK_BUFFER_LEN;
 use crate::index::SegmentReader;
-use crate::query::explanation::does_not_match;
 use crate::query::Explanation;
-use crate::{DocId, DocSet, Score};
+use crate::{DocId, DocSet, Score, TERMINATED};
+
+/// Iterates through all of the documents and scores matched by the DocSet
+/// `DocSet`.
+pub(crate) fn for_each_scorer<TScorer: Scorer + ?Sized>(
+    scorer: &mut TScorer,
+    callback: &mut dyn FnMut(DocId, Score),
+) {
+    let mut doc = scorer.doc();
+    while doc != TERMINATED {
+        callback(doc, scorer.score());
+        doc = scorer.advance();
+    }
+}
 
 /// Iterates through all of the documents matched by the DocSet
 /// `DocSet`.
@@ -22,6 +34,31 @@ pub(crate) fn for_each_docset_buffered<T: DocSet + ?Sized>(
     }
 }
 
+/// Calls `callback` with all of the `(doc, score)` for which score
+/// is exceeding a given threshold.
+///
+/// This method is useful for the [`TopDocs`](crate::collector::TopDocs) collector.
+/// For all docsets, the blanket implementation has the benefit
+/// of prefiltering (doc, score) pairs, avoiding the
+/// virtual dispatch cost.
+///
+/// More importantly, it makes it possible for scorers to implement
+/// important optimization (e.g. BlockWAND for union).
+pub(crate) fn for_each_pruning_scorer<TScorer: Scorer + ?Sized>(
+    scorer: &mut TScorer,
+    mut threshold: Score,
+    callback: &mut dyn FnMut(DocId, Score) -> Score,
+) {
+    let mut doc = scorer.doc();
+    while doc != TERMINATED {
+        let score = scorer.score();
+        if score > threshold {
+            threshold = callback(doc, score);
+        }
+        doc = scorer.advance();
+    }
+}
+
 /// A Weight is the specialization of a `Query`
 /// for a given set of segments.
 ///
@@ -35,13 +72,7 @@ pub trait Weight: Send + Sync + 'static {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>>;
 
     /// Returns an [`Explanation`] for the given document.
-    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
-        let mut scorer = self.scorer(reader, 1.0)?;
-        if scorer.doc() > doc || scorer.seek(doc) != doc {
-            return Err(does_not_match(doc));
-        }
-        Ok(scorer.explain())
-    }
+    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation>;
 
     /// Returns the number documents within the given [`SegmentReader`].
     fn count(&self, reader: &SegmentReader) -> crate::Result<u32> {
@@ -61,7 +92,7 @@ pub trait Weight: Send + Sync + 'static {
         callback: &mut dyn FnMut(DocId, Score),
     ) -> crate::Result<()> {
         let mut scorer = self.scorer(reader, 1.0)?;
-        scorer.for_each(callback);
+        for_each_scorer(scorer.as_mut(), callback);
         Ok(())
     }
 
@@ -96,7 +127,7 @@ pub trait Weight: Send + Sync + 'static {
         callback: &mut dyn FnMut(DocId, Score) -> Score,
     ) -> crate::Result<()> {
         let mut scorer = self.scorer(reader, 1.0)?;
-        scorer.for_each_pruning(threshold, callback);
+        for_each_pruning_scorer(scorer.as_mut(), threshold, callback);
         Ok(())
     }
 }

src/reader/mod.rs

@@ -7,7 +7,6 @@ use arc_swap::ArcSwap;
 pub use warming::Warmer;
 
 use self::warming::WarmingState;
-use crate::codec::Codec;
 use crate::core::searcher::{SearcherGeneration, SearcherInner};
 use crate::directory::{Directory, WatchCallback, WatchHandle, META_LOCK};
 use crate::store::DOCSTORE_CACHE_CAPACITY;
@@ -39,17 +38,17 @@ pub enum ReloadPolicy {
 /// - number of warming threads, for parallelizing warming work
 /// - The cache size of the underlying doc store readers.
 #[derive(Clone)]
-pub struct IndexReaderBuilder<C: Codec = crate::codec::StandardCodec> {
+pub struct IndexReaderBuilder {
     reload_policy: ReloadPolicy,
-    index: Index<C>,
+    index: Index,
     warmers: Vec<Weak<dyn Warmer>>,
     num_warming_threads: usize,
     doc_store_cache_num_blocks: usize,
 }
 
-impl<C: Codec> IndexReaderBuilder<C> {
+impl IndexReaderBuilder {
     #[must_use]
-    pub(crate) fn new(index: Index<C>) -> IndexReaderBuilder<C> {
+    pub(crate) fn new(index: Index) -> IndexReaderBuilder {
         IndexReaderBuilder {
             reload_policy: ReloadPolicy::OnCommitWithDelay,
             index,
@@ -64,7 +63,7 @@ impl<C: Codec> IndexReaderBuilder<C> {
     /// Building the reader is a non-trivial operation that requires
     /// to open different segment readers. It may take hundreds of milliseconds
     /// of time and it may return an error.
-    pub fn try_into(self) -> crate::Result<IndexReader<C>> {
+    pub fn try_into(self) -> crate::Result<IndexReader> {
         let searcher_generation_inventory = Inventory::default();
         let warming_state = WarmingState::new(
             self.num_warming_threads,
@@ -107,7 +106,7 @@ impl<C: Codec> IndexReaderBuilder<C> {
     ///
     /// See [`ReloadPolicy`] for more details.
     #[must_use]
-    pub fn reload_policy(mut self, reload_policy: ReloadPolicy) -> IndexReaderBuilder<C> {
+    pub fn reload_policy(mut self, reload_policy: ReloadPolicy) -> IndexReaderBuilder {
         self.reload_policy = reload_policy;
         self
     }
@@ -119,14 +118,14 @@ impl<C: Codec> IndexReaderBuilder<C> {
     pub fn doc_store_cache_num_blocks(
         mut self,
         doc_store_cache_num_blocks: usize,
-    ) -> IndexReaderBuilder<C> {
+    ) -> IndexReaderBuilder {
         self.doc_store_cache_num_blocks = doc_store_cache_num_blocks;
         self
     }
 
     /// Set the [`Warmer`]s that are invoked when reloading searchable segments.
     #[must_use]
-    pub fn warmers(mut self, warmers: Vec<Weak<dyn Warmer>>) -> IndexReaderBuilder<C> {
+    pub fn warmers(mut self, warmers: Vec<Weak<dyn Warmer>>) -> IndexReaderBuilder {
         self.warmers = warmers;
         self
     }
@@ -136,33 +135,33 @@ impl<C: Codec> IndexReaderBuilder<C> {
     /// This allows parallelizing warming work when there are multiple [`Warmer`] registered with
     /// the [`IndexReader`].
     #[must_use]
-    pub fn num_warming_threads(mut self, num_warming_threads: usize) -> IndexReaderBuilder<C> {
+    pub fn num_warming_threads(mut self, num_warming_threads: usize) -> IndexReaderBuilder {
         self.num_warming_threads = num_warming_threads;
         self
     }
 }
 
-impl<C: Codec> TryInto<IndexReader<C>> for IndexReaderBuilder<C> {
+impl TryInto<IndexReader> for IndexReaderBuilder {
     type Error = crate::TantivyError;
 
-    fn try_into(self) -> crate::Result<IndexReader<C>> {
+    fn try_into(self) -> crate::Result<IndexReader> {
         IndexReaderBuilder::try_into(self)
     }
 }
 
-struct InnerIndexReader<C: Codec> {
+struct InnerIndexReader {
     doc_store_cache_num_blocks: usize,
-    index: Index<C>,
+    index: Index,
     warming_state: WarmingState,
     searcher: arc_swap::ArcSwap<SearcherInner>,
     searcher_generation_counter: Arc<AtomicU64>,
     searcher_generation_inventory: Inventory<SearcherGeneration>,
 }
 
-impl<C: Codec> InnerIndexReader<C> {
+impl InnerIndexReader {
     fn new(
         doc_store_cache_num_blocks: usize,
-        index: Index<C>,
+        index: Index,
         warming_state: WarmingState,
         // The searcher_generation_inventory is not used as source, but as target to track the
         // loaded segments.
@@ -190,7 +189,7 @@ impl<C: Codec> InnerIndexReader<C> {
     ///
     /// This function acquires a lock to prevent GC from removing files
     /// as we are opening our index.
-    fn open_segment_readers(index: &Index<C>) -> crate::Result<Vec<SegmentReader>> {
+    fn open_segment_readers(index: &Index) -> crate::Result<Vec<SegmentReader>> {
         // Prevents segment files from getting deleted while we are in the process of opening them
         let _meta_lock = index.directory().acquire_lock(&META_LOCK)?;
         let searchable_segments = index.searchable_segments()?;
@@ -213,7 +212,7 @@ impl<C: Codec> InnerIndexReader<C> {
     }
 
     fn create_searcher(
-        index: &Index<C>,
+        index: &Index,
         doc_store_cache_num_blocks: usize,
         warming_state: &WarmingState,
         searcher_generation_counter: &Arc<AtomicU64>,
@@ -227,10 +226,9 @@ impl<C: Codec> InnerIndexReader<C> {
         );
 
         let schema = index.schema();
-        // SearcherInner uses Index<StandardCodec> since the codec doesn't affect reading
         let searcher = Arc::new(SearcherInner::new(
             schema,
-            index.with_standard_codec(),
+            index.clone(),
             segment_readers,
             searcher_generation,
             doc_store_cache_num_blocks,
@@ -266,14 +264,14 @@ impl<C: Codec> InnerIndexReader<C> {
 ///
 /// `IndexReader` just wraps an `Arc`.
 #[derive(Clone)]
-pub struct IndexReader<C: Codec = crate::codec::StandardCodec> {
-    inner: Arc<InnerIndexReader<C>>,
+pub struct IndexReader {
+    inner: Arc<InnerIndexReader>,
     _watch_handle_opt: Option<WatchHandle>,
 }
 
-impl<C: Codec> IndexReader<C> {
+impl IndexReader {
     #[cfg(test)]
-    pub(crate) fn index(&self) -> Index<C> {
+    pub(crate) fn index(&self) -> Index {
         self.inner.index.clone()
     }