oneshot 0.1.7

Now that the is_closed changed has been merge upstream, we can rely on
that.

This commit is a "hotfix" because we don't want to rely on
some of the commit in main just yet

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.22.0"
+version = "0.23.0"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -15,12 +15,16 @@ rust-version = "1.63"
 exclude = ["benches/*.json", "benches/*.txt"]
 
 [dependencies]
-oneshot = "0.1.5"
+# Switch back to the non-forked oneshot crate once https://github.com/faern/oneshot/pull/35 is merged
+oneshot = "0.1.7"
 base64 = "0.22.0"
 byteorder = "1.4.3"
 crc32fast = "1.3.2"
 once_cell = "1.10.0"
-regex = { version = "1.5.5", default-features = false, features = ["std", "unicode"] }
+regex = { version = "1.5.5", default-features = false, features = [
+    "std",
+    "unicode",
+] }
 aho-corasick = "1.0"
 tantivy-fst = "0.5"
 memmap2 = { version = "0.9.0", optional = true }
@@ -30,14 +34,15 @@ tempfile = { version = "3.3.0", optional = true }
 log = "0.4.16"
 serde = { version = "1.0.136", features = ["derive"] }
 serde_json = "1.0.79"
-num_cpus = "1.13.1"
 fs4 = { version = "0.8.0", optional = true }
 levenshtein_automata = "0.2.1"
 uuid = { version = "1.0.0", features = ["v4", "serde"] }
 crossbeam-channel = "0.5.4"
 rust-stemmers = "1.2.0"
 downcast-rs = "1.2.0"
-bitpacking = { version = "0.9.2", default-features = false, features = ["bitpacker4x"] }
+bitpacking = { version = "0.9.2", default-features = false, features = [
+    "bitpacker4x",
+] }
 census = "0.4.2"
 rustc-hash = "1.1.0"
 thiserror = "1.0.30"
@@ -52,22 +57,22 @@ itertools = "0.12.0"
 measure_time = "0.8.2"
 arc-swap = "1.5.0"
 
-columnar = { version= "0.3", path="./columnar", package ="tantivy-columnar" }
-sstable = { version= "0.3", path="./sstable", package ="tantivy-sstable", optional = true }
-stacker = { version= "0.3", path="./stacker", package ="tantivy-stacker" }
-query-grammar = { version= "0.22.0", path="./query-grammar", package = "tantivy-query-grammar" }
-tantivy-bitpacker = { version= "0.6", path="./bitpacker" }
-common = { version= "0.7", path = "./common/", package = "tantivy-common" }
-tokenizer-api = { version= "0.3", path="./tokenizer-api", package="tantivy-tokenizer-api" }
+columnar = { version = "0.3", path = "./columnar", package = "tantivy-columnar" }
+sstable = { version = "0.3", path = "./sstable", package = "tantivy-sstable", optional = true }
+stacker = { version = "0.3", path = "./stacker", package = "tantivy-stacker" }
+query-grammar = { version = "0.22.0", path = "./query-grammar", package = "tantivy-query-grammar" }
+tantivy-bitpacker = { version = "0.6", path = "./bitpacker" }
+common = { version = "0.7", path = "./common/", package = "tantivy-common" }
+tokenizer-api = { version = "0.3", path = "./tokenizer-api", package = "tantivy-tokenizer-api" }
 sketches-ddsketch = { version = "0.2.1", features = ["use_serde"] }
 futures-util = { version = "0.3.28", optional = true }
 fnv = "1.0.7"
-mediumvec = "1.3.0"
 
 [target.'cfg(windows)'.dependencies]
 winapi = "0.3.9"
 
 [dev-dependencies]
+binggan = "0.6.2"
 rand = "0.8.5"
 maplit = "1.0.2"
 matches = "0.1.9"
@@ -82,7 +87,6 @@ time = { version = "0.3.10", features = ["serde-well-known", "macros"] }
 postcard = { version = "1.0.4", features = [
   "use-std",
 ], default-features = false }
-peakmem-alloc = "0.3.0"
 
 [target.'cfg(not(windows))'.dev-dependencies]
 criterion = { version = "0.5", default-features = false }
@@ -114,17 +118,26 @@ lz4-compression = ["lz4_flex"]
 zstd-compression = ["zstd"]
 
 failpoints = ["fail", "fail/failpoints"]
-unstable = [] # useful for benches.
+unstable = []                            # useful for benches.
 
 quickwit = ["sstable", "futures-util"]
 
-# Compares only the hash of a string when indexing data. 
+# Compares only the hash of a string when indexing data.
 # Increases indexing speed, but may lead to extremely rare missing terms, when there's a hash collision.
 # Uses 64bit ahash.
 compare_hash_only = ["stacker/compare_hash_only"]
 
 [workspace]
-members = ["query-grammar", "bitpacker", "common", "ownedbytes", "stacker", "sstable", "tokenizer-api", "columnar"]
+members = [
+    "query-grammar",
+    "bitpacker",
+    "common",
+    "ownedbytes",
+    "stacker",
+    "sstable",
+    "tokenizer-api",
+    "columnar",
+]
 
 # Following the "fail" crate best practises, we isolate
 # tests that define specific behavior in fail check points
@@ -145,3 +158,7 @@ harness = false
 [[bench]]
 name = "index-bench"
 harness = false
+
+[[bench]]
+name = "agg_bench"
+harness = false

benches/agg_bench.rs

@@ -0,0 +1,413 @@
+use binggan::{black_box, InputGroup, PeakMemAlloc, INSTRUMENTED_SYSTEM};
+use rand::prelude::SliceRandom;
+use rand::rngs::StdRng;
+use rand::{Rng, SeedableRng};
+use rand_distr::Distribution;
+use serde_json::json;
+use tantivy::aggregation::agg_req::Aggregations;
+use tantivy::aggregation::AggregationCollector;
+use tantivy::query::{AllQuery, TermQuery};
+use tantivy::schema::{IndexRecordOption, Schema, TextFieldIndexing, FAST, STRING};
+use tantivy::{doc, Index, Term};
+
+#[global_allocator]
+pub static GLOBAL: &PeakMemAlloc<std::alloc::System> = &INSTRUMENTED_SYSTEM;
+
+/// Mini macro to register a function via its name
+/// runner.register("average_u64", move |index| average_u64(index));
+macro_rules! register {
+    ($runner:expr, $func:ident) => {
+        $runner.register(stringify!($func), move |index| $func(index))
+    };
+}
+
+fn main() {
+    let inputs = vec![
+        ("full", get_test_index_bench(Cardinality::Full).unwrap()),
+        (
+            "dense",
+            get_test_index_bench(Cardinality::OptionalDense).unwrap(),
+        ),
+        (
+            "sparse",
+            get_test_index_bench(Cardinality::OptionalSparse).unwrap(),
+        ),
+        (
+            "multivalue",
+            get_test_index_bench(Cardinality::Multivalued).unwrap(),
+        ),
+    ];
+
+    bench_agg(InputGroup::new_with_inputs(inputs));
+}
+
+fn bench_agg(mut group: InputGroup<Index>) {
+    group.set_alloc(GLOBAL); // Set the peak mem allocator. This will enable peak memory reporting.
+    register!(group, average_u64);
+    register!(group, average_f64);
+    register!(group, average_f64_u64);
+    register!(group, stats_f64);
+    register!(group, percentiles_f64);
+    register!(group, terms_few);
+    register!(group, terms_many);
+    register!(group, terms_many_order_by_term);
+    register!(group, terms_many_with_top_hits);
+    register!(group, terms_many_with_avg_sub_agg);
+    register!(group, terms_many_json_mixed_type_with_sub_agg_card);
+    register!(group, range_agg);
+    register!(group, range_agg_with_avg_sub_agg);
+    register!(group, range_agg_with_term_agg_few);
+    register!(group, range_agg_with_term_agg_many);
+    register!(group, histogram);
+    register!(group, histogram_hard_bounds);
+    register!(group, histogram_with_avg_sub_agg);
+    register!(group, avg_and_range_with_avg_sub_agg);
+
+    group.run();
+}
+
+fn exec_term_with_agg(index: &Index, agg_req: serde_json::Value) {
+    let agg_req: Aggregations = serde_json::from_value(agg_req).unwrap();
+
+    let reader = index.reader().unwrap();
+    let text_field = reader.searcher().schema().get_field("text").unwrap();
+    let term_query = TermQuery::new(
+        Term::from_field_text(text_field, "cool"),
+        IndexRecordOption::Basic,
+    );
+    let collector = get_collector(agg_req);
+    let searcher = reader.searcher();
+    black_box(searcher.search(&term_query, &collector).unwrap());
+}
+
+fn average_u64(index: &Index) {
+    let agg_req = json!({
+        "average": { "avg": { "field": "score", } }
+    });
+    exec_term_with_agg(index, agg_req)
+}
+fn average_f64(index: &Index) {
+    let agg_req = json!({
+        "average": { "avg": { "field": "score_f64", } }
+    });
+    exec_term_with_agg(index, agg_req)
+}
+fn average_f64_u64(index: &Index) {
+    let agg_req = json!({
+        "average_f64": { "avg": { "field": "score_f64" } },
+        "average": { "avg": { "field": "score" } },
+    });
+    exec_term_with_agg(index, agg_req)
+}
+fn stats_f64(index: &Index) {
+    let agg_req = json!({
+        "average_f64": { "stats": { "field": "score_f64", } }
+    });
+    exec_term_with_agg(index, agg_req)
+}
+
+fn percentiles_f64(index: &Index) {
+    let agg_req = json!({
+      "mypercentiles": {
+        "percentiles": {
+          "field": "score_f64",
+          "percents": [ 95, 99, 99.9 ]
+        }
+      }
+    });
+    execute_agg(index, agg_req);
+}
+fn terms_few(index: &Index) {
+    let agg_req = json!({
+        "my_texts": { "terms": { "field": "text_few_terms" } },
+    });
+    execute_agg(index, agg_req);
+}
+fn terms_many(index: &Index) {
+    let agg_req = json!({
+        "my_texts": { "terms": { "field": "text_many_terms" } },
+    });
+    execute_agg(index, agg_req);
+}
+fn terms_many_order_by_term(index: &Index) {
+    let agg_req = json!({
+        "my_texts": { "terms": { "field": "text_many_terms", "order": { "_key": "desc" } } },
+    });
+    execute_agg(index, agg_req);
+}
+fn terms_many_with_top_hits(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": { "field": "text_many_terms" },
+            "aggs": {
+                "top_hits": { "top_hits":
+                    {
+                        "sort": [
+                            { "score": "desc" }
+                        ],
+                        "size": 2,
+                        "doc_value_fields": ["score_f64"]
+                    }
+                }
+            }
+        },
+    });
+    execute_agg(index, agg_req);
+}
+fn terms_many_with_avg_sub_agg(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": { "field": "text_many_terms" },
+            "aggs": {
+                "average_f64": { "avg": { "field": "score_f64" } }
+            }
+        },
+    });
+    execute_agg(index, agg_req);
+}
+fn terms_many_json_mixed_type_with_sub_agg_card(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": { "field": "json.mixed_type" },
+            "aggs": {
+                "average_f64": { "avg": { "field": "score_f64" } }
+            }
+        },
+    });
+    execute_agg(index, agg_req);
+}
+
+fn execute_agg(index: &Index, agg_req: serde_json::Value) {
+    let agg_req: Aggregations = serde_json::from_value(agg_req).unwrap();
+    let collector = get_collector(agg_req);
+
+    let reader = index.reader().unwrap();
+    let searcher = reader.searcher();
+    black_box(searcher.search(&AllQuery, &collector).unwrap());
+}
+fn range_agg(index: &Index) {
+    let agg_req = json!({
+        "range_f64": { "range": { "field": "score_f64", "ranges": [
+            { "from": 3, "to": 7000 },
+            { "from": 7000, "to": 20000 },
+            { "from": 20000, "to": 30000 },
+            { "from": 30000, "to": 40000 },
+            { "from": 40000, "to": 50000 },
+            { "from": 50000, "to": 60000 }
+        ] } },
+    });
+    execute_agg(index, agg_req);
+}
+fn range_agg_with_avg_sub_agg(index: &Index) {
+    let agg_req = json!({
+        "rangef64": {
+            "range": {
+                "field": "score_f64",
+                "ranges": [
+                    { "from": 3, "to": 7000 },
+                    { "from": 7000, "to": 20000 },
+                    { "from": 20000, "to": 30000 },
+                    { "from": 30000, "to": 40000 },
+                    { "from": 40000, "to": 50000 },
+                    { "from": 50000, "to": 60000 }
+                ]
+            },
+            "aggs": {
+                "average_f64": { "avg": { "field": "score_f64" } }
+            }
+        },
+    });
+    execute_agg(index, agg_req);
+}
+
+fn range_agg_with_term_agg_few(index: &Index) {
+    let agg_req = json!({
+        "rangef64": {
+            "range": {
+                "field": "score_f64",
+                "ranges": [
+                    { "from": 3, "to": 7000 },
+                    { "from": 7000, "to": 20000 },
+                    { "from": 20000, "to": 30000 },
+                    { "from": 30000, "to": 40000 },
+                    { "from": 40000, "to": 50000 },
+                    { "from": 50000, "to": 60000 }
+                ]
+            },
+            "aggs": {
+                "my_texts": { "terms": { "field": "text_few_terms" } },
+            }
+        },
+    });
+    execute_agg(index, agg_req);
+}
+fn range_agg_with_term_agg_many(index: &Index) {
+    let agg_req = json!({
+        "rangef64": {
+            "range": {
+                "field": "score_f64",
+                "ranges": [
+                    { "from": 3, "to": 7000 },
+                    { "from": 7000, "to": 20000 },
+                    { "from": 20000, "to": 30000 },
+                    { "from": 30000, "to": 40000 },
+                    { "from": 40000, "to": 50000 },
+                    { "from": 50000, "to": 60000 }
+                ]
+            },
+            "aggs": {
+                "my_texts": { "terms": { "field": "text_many_terms" } },
+            }
+        },
+    });
+    execute_agg(index, agg_req);
+}
+fn histogram(index: &Index) {
+    let agg_req = json!({
+        "rangef64": {
+            "histogram": {
+                "field": "score_f64",
+                "interval": 100 // 1000 buckets
+            },
+        }
+    });
+    execute_agg(index, agg_req);
+}
+fn histogram_hard_bounds(index: &Index) {
+    let agg_req = json!({
+        "rangef64": { "histogram": { "field": "score_f64", "interval": 100, "hard_bounds": { "min": 1000, "max": 300000 } } },
+    });
+    execute_agg(index, agg_req);
+}
+fn histogram_with_avg_sub_agg(index: &Index) {
+    let agg_req = json!({
+        "rangef64": {
+            "histogram": { "field": "score_f64", "interval": 100 },
+            "aggs": {
+                "average_f64": { "avg": { "field": "score_f64" } }
+            }
+        }
+    });
+    execute_agg(index, agg_req);
+}
+fn avg_and_range_with_avg_sub_agg(index: &Index) {
+    let agg_req = json!({
+        "rangef64": {
+            "range": {
+                "field": "score_f64",
+                "ranges": [
+                    { "from": 3, "to": 7000 },
+                    { "from": 7000, "to": 20000 },
+                    { "from": 20000, "to": 60000 }
+                ]
+            },
+            "aggs": {
+                "average_in_range": { "avg": { "field": "score" } }
+            }
+        },
+        "average": { "avg": { "field": "score" } }
+    });
+    execute_agg(index, agg_req);
+}
+
+#[derive(Clone, Copy, Hash, Default, Debug, PartialEq, Eq, PartialOrd, Ord)]
+enum Cardinality {
+    /// All documents contain exactly one value.
+    /// `Full` is the default for auto-detecting the Cardinality, since it is the most strict.
+    #[default]
+    Full = 0,
+    /// All documents contain at most one value.
+    OptionalDense = 1,
+    /// All documents may contain any number of values.
+    Multivalued = 2,
+    /// 1 / 20 documents has a value
+    OptionalSparse = 3,
+}
+
+fn get_collector(agg_req: Aggregations) -> AggregationCollector {
+    AggregationCollector::from_aggs(agg_req, Default::default())
+}
+
+fn get_test_index_bench(cardinality: Cardinality) -> tantivy::Result<Index> {
+    let mut schema_builder = Schema::builder();
+    let text_fieldtype = tantivy::schema::TextOptions::default()
+        .set_indexing_options(
+            TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
+        )
+        .set_stored();
+    let text_field = schema_builder.add_text_field("text", text_fieldtype);
+    let json_field = schema_builder.add_json_field("json", FAST);
+    let text_field_many_terms = schema_builder.add_text_field("text_many_terms", STRING | FAST);
+    let text_field_few_terms = schema_builder.add_text_field("text_few_terms", STRING | FAST);
+    let score_fieldtype = tantivy::schema::NumericOptions::default().set_fast();
+    let score_field = schema_builder.add_u64_field("score", score_fieldtype.clone());
+    let score_field_f64 = schema_builder.add_f64_field("score_f64", score_fieldtype.clone());
+    let score_field_i64 = schema_builder.add_i64_field("score_i64", score_fieldtype);
+    let index = Index::create_from_tempdir(schema_builder.build())?;
+    let few_terms_data = ["INFO", "ERROR", "WARN", "DEBUG"];
+
+    let lg_norm = rand_distr::LogNormal::new(2.996f64, 0.979f64).unwrap();
+
+    let many_terms_data = (0..150_000)
+        .map(|num| format!("author{}", num))
+        .collect::<Vec<_>>();
+    {
+        let mut rng = StdRng::from_seed([1u8; 32]);
+        let mut index_writer = index.writer_with_num_threads(1, 200_000_000)?;
+        // To make the different test cases comparable we just change one doc to force the
+        // cardinality
+        if cardinality == Cardinality::OptionalDense {
+            index_writer.add_document(doc!())?;
+        }
+        if cardinality == Cardinality::Multivalued {
+            index_writer.add_document(doc!(
+                json_field => json!({"mixed_type": 10.0}),
+                json_field => json!({"mixed_type": 10.0}),
+                text_field => "cool",
+                text_field => "cool",
+                text_field_many_terms => "cool",
+                text_field_many_terms => "cool",
+                text_field_few_terms => "cool",
+                text_field_few_terms => "cool",
+                score_field => 1u64,
+                score_field => 1u64,
+                score_field_f64 => lg_norm.sample(&mut rng),
+                score_field_f64 => lg_norm.sample(&mut rng),
+                score_field_i64 => 1i64,
+                score_field_i64 => 1i64,
+            ))?;
+        }
+        let mut doc_with_value = 1_000_000;
+        if cardinality == Cardinality::OptionalSparse {
+            doc_with_value /= 20;
+        }
+        let _val_max = 1_000_000.0;
+        for _ in 0..doc_with_value {
+            let val: f64 = rng.gen_range(0.0..1_000_000.0);
+            let json = if rng.gen_bool(0.1) {
+                // 10% are numeric values
+                json!({ "mixed_type": val })
+            } else {
+                json!({"mixed_type": many_terms_data.choose(&mut rng).unwrap().to_string()})
+            };
+            index_writer.add_document(doc!(
+                text_field => "cool",
+                json_field => json,
+                text_field_many_terms => many_terms_data.choose(&mut rng).unwrap().to_string(),
+                text_field_few_terms => few_terms_data.choose(&mut rng).unwrap().to_string(),
+                score_field => val as u64,
+                score_field_f64 => lg_norm.sample(&mut rng),
+                score_field_i64 => val as i64,
+            ))?;
+            if cardinality == Cardinality::OptionalSparse {
+                for _ in 0..20 {
+                    index_writer.add_document(doc!(text_field => "cool"))?;
+                }
+            }
+        }
+        // writing the segment
+        index_writer.commit()?;
+    }
+
+    Ok(index)
+}

columnar/src/columnar/writer/mod.rs

@@ -59,22 +59,6 @@ pub struct ColumnarWriter {
     buffers: SpareBuffers,
 }
 
-#[inline]
-fn mutate_or_create_column<V, TMutator>(
-    arena_hash_map: &mut ArenaHashMap,
-    column_name: &str,
-    updater: TMutator,
-) where
-    V: Copy + 'static,
-    TMutator: FnMut(Option<V>) -> V,
-{
-    assert!(
-        !column_name.as_bytes().contains(&0u8),
-        "key may not contain the 0 byte"
-    );
-    arena_hash_map.mutate_or_create(column_name.as_bytes(), updater);
-}
-
 impl ColumnarWriter {
     pub fn mem_usage(&self) -> usize {
         self.arena.mem_usage()
@@ -175,9 +159,8 @@ impl ColumnarWriter {
                     },
                     &mut self.dictionaries,
                 );
-                mutate_or_create_column(
-                    hash_map,
-                    column_name,
+                hash_map.mutate_or_create(
+                    column_name.as_bytes(),
                     |column_opt: Option<StrOrBytesColumnWriter>| {
                         let mut column_writer = if let Some(column_writer) = column_opt {
                             column_writer
@@ -192,24 +175,21 @@ impl ColumnarWriter {
                 );
             }
             ColumnType::Bool => {
-                mutate_or_create_column(
-                    &mut self.bool_field_hash_map,
-                    column_name,
+                self.bool_field_hash_map.mutate_or_create(
+                    column_name.as_bytes(),
                     |column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
                 );
             }
             ColumnType::DateTime => {
-                mutate_or_create_column(
-                    &mut self.datetime_field_hash_map,
-                    column_name,
+                self.datetime_field_hash_map.mutate_or_create(
+                    column_name.as_bytes(),
                     |column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
                 );
             }
             ColumnType::I64 | ColumnType::F64 | ColumnType::U64 => {
                 let numerical_type = column_type.numerical_type().unwrap();
-                mutate_or_create_column(
-                    &mut self.numerical_field_hash_map,
-                    column_name,
+                self.numerical_field_hash_map.mutate_or_create(
+                    column_name.as_bytes(),
                     |column_opt: Option<NumericalColumnWriter>| {
                         let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
                         column.force_numerical_type(numerical_type);
@@ -217,9 +197,8 @@ impl ColumnarWriter {
                     },
                 );
             }
-            ColumnType::IpAddr => mutate_or_create_column(
-                &mut self.ip_addr_field_hash_map,
-                column_name,
+            ColumnType::IpAddr => self.ip_addr_field_hash_map.mutate_or_create(
+                column_name.as_bytes(),
                 |column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
             ),
         }
@@ -232,9 +211,8 @@ impl ColumnarWriter {
         numerical_value: T,
     ) {
         let (hash_map, arena) = (&mut self.numerical_field_hash_map, &mut self.arena);
-        mutate_or_create_column(
-            hash_map,
-            column_name,
+        hash_map.mutate_or_create(
+            column_name.as_bytes(),
             |column_opt: Option<NumericalColumnWriter>| {
                 let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
                 column.record_numerical_value(doc, numerical_value.into(), arena);
@@ -244,10 +222,6 @@ impl ColumnarWriter {
     }
 
     pub fn record_ip_addr(&mut self, doc: RowId, column_name: &str, ip_addr: Ipv6Addr) {
-        assert!(
-            !column_name.as_bytes().contains(&0u8),
-            "key may not contain the 0 byte"
-        );
         let (hash_map, arena) = (&mut self.ip_addr_field_hash_map, &mut self.arena);
         hash_map.mutate_or_create(
             column_name.as_bytes(),
@@ -261,24 +235,30 @@ impl ColumnarWriter {
 
     pub fn record_bool(&mut self, doc: RowId, column_name: &str, val: bool) {
         let (hash_map, arena) = (&mut self.bool_field_hash_map, &mut self.arena);
-        mutate_or_create_column(hash_map, column_name, |column_opt: Option<ColumnWriter>| {
-            let mut column: ColumnWriter = column_opt.unwrap_or_default();
-            column.record(doc, val, arena);
-            column
-        });
+        hash_map.mutate_or_create(
+            column_name.as_bytes(),
+            |column_opt: Option<ColumnWriter>| {
+                let mut column: ColumnWriter = column_opt.unwrap_or_default();
+                column.record(doc, val, arena);
+                column
+            },
+        );
     }
 
     pub fn record_datetime(&mut self, doc: RowId, column_name: &str, datetime: common::DateTime) {
         let (hash_map, arena) = (&mut self.datetime_field_hash_map, &mut self.arena);
-        mutate_or_create_column(hash_map, column_name, |column_opt: Option<ColumnWriter>| {
-            let mut column: ColumnWriter = column_opt.unwrap_or_default();
-            column.record(
-                doc,
-                NumericalValue::I64(datetime.into_timestamp_nanos()),
-                arena,
-            );
-            column
-        });
+        hash_map.mutate_or_create(
+            column_name.as_bytes(),
+            |column_opt: Option<ColumnWriter>| {
+                let mut column: ColumnWriter = column_opt.unwrap_or_default();
+                column.record(
+                    doc,
+                    NumericalValue::I64(datetime.into_timestamp_nanos()),
+                    arena,
+                );
+                column
+            },
+        );
     }
 
     pub fn record_str(&mut self, doc: RowId, column_name: &str, value: &str) {
@@ -303,10 +283,6 @@ impl ColumnarWriter {
     }
 
     pub fn record_bytes(&mut self, doc: RowId, column_name: &str, value: &[u8]) {
-        assert!(
-            !column_name.as_bytes().contains(&0u8),
-            "key may not contain the 0 byte"
-        );
         let (hash_map, arena, dictionaries) = (
             &mut self.bytes_field_hash_map,
             &mut self.arena,

examples/custom_collector.rs

@@ -11,9 +11,10 @@ use columnar::Column;
 // ---
 // Importing tantivy...
 use tantivy::collector::{Collector, SegmentCollector};
+use tantivy::index::SegmentReader;
 use tantivy::query::QueryParser;
 use tantivy::schema::{Schema, FAST, INDEXED, TEXT};
-use tantivy::{doc, Index, IndexWriter, Score, SegmentReader};
+use tantivy::{doc, Index, IndexWriter, Score};
 
 #[derive(Default)]
 struct Stats {

examples/date_time_field.rs

@@ -13,7 +13,7 @@ fn main() -> tantivy::Result<()> {
     let opts = DateOptions::from(INDEXED)
         .set_stored()
         .set_fast()
-        .set_precision(tantivy::DateTimePrecision::Seconds);
+        .set_precision(tantivy::schema::DateTimePrecision::Seconds);
     // Add `occurred_at` date field type
     let occurred_at = schema_builder.add_date_field("occurred_at", opts);
     let event_type = schema_builder.add_text_field("event", STRING | STORED);

examples/doc_mem.rs

@@ -1,335 +0,0 @@
-#![allow(unused_imports)]
-#![allow(dead_code)]
-use std::alloc::System;
-use std::env::args;
-use std::net::Ipv6Addr;
-
-use columnar::{MonotonicallyMappableToU128, MonotonicallyMappableToU64};
-use common::{BinarySerializable, CountingWriter, DateTime, FixedSize};
-use peakmem_alloc::*;
-use tantivy::schema::{Field, FieldValue, OwnedValue, FAST, INDEXED, STRING, TEXT};
-use tantivy::tokenizer::PreTokenizedString;
-use tantivy::{doc, TantivyDocument};
-
-const GH_LOGS: &str = include_str!("../benches/gh.json");
-const HDFS_LOGS: &str = include_str!("../benches/hdfs.json");
-
-#[global_allocator]
-static GLOBAL: &PeakMemAlloc<System> = &INSTRUMENTED_SYSTEM;
-
-fn main() {
-    dbg!(std::mem::size_of::<TantivyDocument>());
-    dbg!(std::mem::size_of::<DocContainerRef>());
-    dbg!(std::mem::size_of::<OwnedValue>());
-    dbg!(std::mem::size_of::<OwnedValueMedVec>());
-    dbg!(std::mem::size_of::<ValueContainerRef>());
-    dbg!(std::mem::size_of::<mediumvec::vec32::Vec32::<u8>>());
-
-    let filter = args().nth(1);
-    measure_fn(
-        test_hdfs::<TantivyDocument>,
-        "hdfs TantivyDocument",
-        &filter,
-    );
-    measure_fn(
-        test_hdfs::<TantivyDocumentMedVec>,
-        "hdfs TantivyDocumentMedVec",
-        &filter,
-    );
-    measure_fn(
-        test_hdfs::<DocContainerRef>,
-        "hdfs DocContainerRef",
-        &filter,
-    );
-    measure_fn(test_gh::<TantivyDocument>, "gh TantivyDocument", &filter);
-    measure_fn(
-        test_gh::<TantivyDocumentMedVec>,
-        "gh TantivyDocumentMedVec",
-        &filter,
-    );
-    measure_fn(test_gh::<DocContainerRef>, "gh DocContainerRef", &filter);
-}
-fn measure_fn<F: FnOnce()>(f: F, name: &str, filter: &Option<std::string::String>) {
-    if let Some(filter) = filter {
-        if !name.contains(filter) {
-            return;
-        }
-    }
-    GLOBAL.reset_peak_memory();
-    f();
-    println!("Peak Memory {} : {:#?}", GLOBAL.get_peak_memory(), name);
-}
-fn test_hdfs<T: From<TantivyDocument>>() {
-    let schema = {
-        let mut schema_builder = tantivy::schema::SchemaBuilder::new();
-        schema_builder.add_u64_field("timestamp", INDEXED);
-        schema_builder.add_text_field("body", TEXT);
-        schema_builder.add_text_field("severity", STRING);
-        schema_builder.build()
-    };
-    let mut docs: Vec<T> = Vec::with_capacity(HDFS_LOGS.lines().count());
-    for doc_json in HDFS_LOGS.lines() {
-        let doc = TantivyDocument::parse_json(&schema, doc_json)
-            .unwrap()
-            .into();
-        docs.push(doc);
-    }
-}
-
-fn test_gh<T: From<TantivyDocument>>() {
-    let schema = {
-        let mut schema_builder = tantivy::schema::SchemaBuilder::new();
-        schema_builder.add_json_field("json", FAST);
-        schema_builder.build()
-    };
-    let mut docs: Vec<T> = Vec::with_capacity(GH_LOGS.lines().count());
-    for doc_json in GH_LOGS.lines() {
-        let json_field = schema.get_field("json").unwrap();
-
-        let json_val: serde_json::Map<String, serde_json::Value> =
-            serde_json::from_str(doc_json).unwrap();
-        let doc = tantivy::doc!(json_field=>json_val).into();
-        docs.push(doc);
-    }
-}
-
-#[derive(Clone, Debug, Default)]
-#[allow(dead_code)]
-pub struct TantivyDocumentMedVec {
-    field_values: mediumvec::Vec32<FieldValueMedVec>,
-}
-
-#[derive(Debug, Clone, PartialEq)]
-pub struct FieldValueMedVec {
-    pub field: Field,
-    pub value: OwnedValueMedVec,
-}
-
-/// This is a owned variant of `Value`, that can be passed around without lifetimes.
-/// Represents the value of a any field.
-/// It is an enum over all over all of the possible field type.
-#[derive(Debug, Clone, PartialEq)]
-pub enum OwnedValueMedVec {
-    /// A null value.
-    Null,
-    /// The str type is used for any text information.
-    Str(mediumvec::vec32::Vec32<u8>),
-    /// Unsigned 64-bits Integer `u64`
-    U64(u64),
-    /// Signed 64-bits Integer `i64`
-    I64(i64),
-    /// 64-bits Float `f64`
-    F64(f64),
-    /// Bool value
-    Bool(bool),
-    /// Date/time with nanoseconds precision
-    Date(DateTime),
-    Array(mediumvec::vec32::Vec32<Self>),
-    /// Dynamic object value.
-    Object(mediumvec::vec32::Vec32<(String, Self)>),
-    /// IpV6 Address. Internally there is no IpV4, it needs to be converted to `Ipv6Addr`.
-    IpAddr(Ipv6Addr),
-    /// Pre-tokenized str type,
-    PreTokStr(Box<PreTokenizedString>),
-    /// Arbitrarily sized byte array
-    Bytes(mediumvec::vec32::Vec32<u8>),
-}
-
-impl From<TantivyDocument> for TantivyDocumentMedVec {
-    fn from(doc: TantivyDocument) -> Self {
-        let field_values = doc
-            .into_iter()
-            .map(|fv| FieldValueMedVec {
-                field: fv.field,
-                value: fv.value.into(),
-            })
-            .collect();
-        TantivyDocumentMedVec { field_values }
-    }
-}
-impl From<OwnedValue> for OwnedValueMedVec {
-    fn from(value: OwnedValue) -> Self {
-        match value {
-            OwnedValue::Null => OwnedValueMedVec::Null,
-            OwnedValue::Str(s) => {
-                let bytes = s.into_bytes();
-                let vec = mediumvec::vec32::Vec32::from_vec(bytes);
-                OwnedValueMedVec::Str(vec)
-            }
-            OwnedValue::U64(u) => OwnedValueMedVec::U64(u),
-            OwnedValue::I64(i) => OwnedValueMedVec::I64(i),
-            OwnedValue::F64(f) => OwnedValueMedVec::F64(f),
-            OwnedValue::Bool(b) => OwnedValueMedVec::Bool(b),
-            OwnedValue::Date(d) => OwnedValueMedVec::Date(d),
-            OwnedValue::Array(arr) => {
-                let arr = arr.into_iter().map(|v| v.into()).collect();
-                OwnedValueMedVec::Array(arr)
-            }
-            OwnedValue::Object(obj) => {
-                let obj = obj.into_iter().map(|(k, v)| (k, v.into())).collect();
-                OwnedValueMedVec::Object(obj)
-            }
-            OwnedValue::IpAddr(ip) => OwnedValueMedVec::IpAddr(ip),
-            _ => panic!("Unsupported value type {:?}", value),
-        }
-    }
-}
-
-#[repr(packed)]
-pub struct FieldValueContainerRef {
-    pub field: u16,
-    pub value: ValueContainerRef,
-}
-
-#[repr(packed)]
-struct DocContainerRef {
-    container: OwnedValueRefContainer,
-    field_values: mediumvec::Vec32<FieldValueContainerRef>,
-}
-
-#[derive(Default)]
-struct OwnedValueRefContainer {
-    nodes: mediumvec::Vec32<ValueContainerRef>,
-    node_data: mediumvec::Vec32<u8>,
-}
-impl OwnedValueRefContainer {
-    fn shrink_to_fit(&mut self) {
-        self.nodes.shrink_to_fit();
-        self.node_data.shrink_to_fit();
-    }
-}
-
-impl From<TantivyDocument> for DocContainerRef {
-    fn from(doc: TantivyDocument) -> Self {
-        let mut container = OwnedValueRefContainer::default();
-        let field_values = doc
-            .into_iter()
-            .map(|fv| FieldValueContainerRef {
-                field: fv.field.field_id().try_into().unwrap(),
-                value: container.add_value(fv.value),
-            })
-            .collect();
-        container.shrink_to_fit();
-        Self {
-            field_values,
-            container,
-        }
-    }
-}
-
-// References to positions in two array, one for the OwnedValueRef and the other for the encoded
-// bytes
-#[derive(Debug, Clone, PartialEq)]
-pub enum ValueContainerRef {
-    /// A null value.
-    Null,
-    /// The str type is used for any text information.
-    Str(u32),
-    /// Unsigned 64-bits Integer `u64`
-    U64(u32), // position of the serialized 8 bytes in the data array
-    /// Signed 64-bits Integer `i64`
-    I64(u32), // position of the serialized 8 bytes in the data array
-    /// 64-bits Float `f64`
-    F64(u32), // position of the serialized 8 bytes in the data array
-    /// Bool value
-    Bool(bool), // inlined bool
-    /// Date/time with nanoseconds precision
-    Date(u32), // position of the serialized 8 byte in the data array
-    Array(NodeAddress),
-    /// Dynamic object value.
-    Object(NodeAddress),
-    /// IpV6 Address. Internally there is no IpV4, it needs to be converted to `Ipv6Addr`.
-    IpAddr(u32), // position of the serialized 16 bytes in the data array
-    /// Arbitrarily sized byte array
-    Bytes(u32),
-}
-
-#[derive(Debug, Clone, PartialEq)]
-pub struct NodeAddress {
-    pos: u32,
-    num_nodes: u32,
-}
-
-impl OwnedValueRefContainer {
-    pub fn add_value(&mut self, value: OwnedValue) -> ValueContainerRef {
-        match value {
-            OwnedValue::Null => ValueContainerRef::Null,
-            OwnedValue::U64(num) => ValueContainerRef::U64(write_into(&mut self.node_data, num)),
-            OwnedValue::I64(num) => ValueContainerRef::I64(write_into(&mut self.node_data, num)),
-            OwnedValue::F64(num) => ValueContainerRef::F64(write_into(&mut self.node_data, num)),
-            OwnedValue::Bool(b) => ValueContainerRef::Bool(b),
-            OwnedValue::Date(date) => ValueContainerRef::Date(write_into(
-                &mut self.node_data,
-                date.into_timestamp_nanos(),
-            )),
-            OwnedValue::Str(bytes) => {
-                ValueContainerRef::Str(write_into(&mut self.node_data, bytes))
-            }
-            OwnedValue::Bytes(bytes) => {
-                ValueContainerRef::Bytes(write_into(&mut self.node_data, bytes))
-            }
-            OwnedValue::Array(elements) => {
-                let pos = self.nodes.len() as u32;
-                let len = elements.len() as u32;
-                for elem in elements {
-                    let ref_elem = self.add_value(elem);
-                    self.nodes.push(ref_elem);
-                }
-                ValueContainerRef::Array(NodeAddress {
-                    pos,
-                    num_nodes: len,
-                })
-            }
-            OwnedValue::Object(entries) => {
-                let pos = self.nodes.len() as u32;
-                let len = entries.len() as u32;
-                for (key, value) in entries {
-                    let ref_key = self.add_value(OwnedValue::Str(key));
-                    let ref_value = self.add_value(value);
-                    self.nodes.push(ref_key);
-                    self.nodes.push(ref_value);
-                }
-                ValueContainerRef::Object(NodeAddress {
-                    pos,
-                    num_nodes: len,
-                })
-            }
-            OwnedValue::IpAddr(num) => {
-                ValueContainerRef::IpAddr(write_into(&mut self.node_data, num.to_u128()))
-            }
-            OwnedValue::PreTokStr(_) => todo!(),
-            OwnedValue::Facet(_) => todo!(),
-        }
-    }
-}
-
-fn write_into<T: BinarySerializable>(data: &mut mediumvec::Vec32<u8>, value: T) -> u32 {
-    let pos = data.len() as u32;
-    data.as_vec(|vec| value.serialize(vec).unwrap());
-    pos
-}
-
-fn write_into_2<T: BinarySerializable>(data: &mut mediumvec::Vec32<u8>, value: T) -> NodeAddress {
-    let pos = data.len() as u32;
-    let mut len = 0;
-    data.as_vec(|vec| {
-        let mut wrt = CountingWriter::wrap(vec);
-        value.serialize(&mut wrt).unwrap();
-        len = wrt.written_bytes() as u32;
-    });
-    NodeAddress {
-        pos,
-        num_nodes: len,
-    }
-}
-
-// impl From<ContainerDocRef> for TantivyDocument {
-// fn from(doc: ContainerDocRef) -> Self {
-// let mut doc2 = TantivyDocument::new();
-// for fv in doc.field_values {
-// let field = Field::from_field_id(fv.field as u32);
-// let value = doc.container.get_value(fv.value);
-// doc2.add(FieldValue::new(field, value));
-//}
-// doc2
-//}

examples/iterating_docs_and_positions.rs

@@ -7,10 +7,11 @@
 // the list of documents containing a term, getting
 // its term frequency, and accessing its positions.
 
+use tantivy::postings::Postings;
 // ---
 // Importing tantivy...
 use tantivy::schema::*;
-use tantivy::{doc, DocSet, Index, IndexWriter, Postings, TERMINATED};
+use tantivy::{doc, DocSet, Index, IndexWriter, TERMINATED};
 
 fn main() -> tantivy::Result<()> {
     // We first create a schema for the sake of the

examples/warmer.rs

@@ -3,10 +3,11 @@ use std::collections::{HashMap, HashSet};
 use std::sync::{Arc, RwLock, Weak};
 
 use tantivy::collector::TopDocs;
+use tantivy::index::SegmentId;
 use tantivy::query::QueryParser;
 use tantivy::schema::{Schema, FAST, TEXT};
 use tantivy::{
-    doc, DocAddress, DocId, Index, IndexWriter, Opstamp, Searcher, SearcherGeneration, SegmentId,
+    doc, DocAddress, DocId, Index, IndexWriter, Opstamp, Searcher, SearcherGeneration,
     SegmentReader, Warmer,
 };
 

src/aggregation/agg_bench.rs

@@ -1,585 +0,0 @@
-#[cfg(all(test, feature = "unstable"))]
-mod bench {
-
-    use rand::prelude::SliceRandom;
-    use rand::rngs::StdRng;
-    use rand::{Rng, SeedableRng};
-    use rand_distr::Distribution;
-    use serde_json::json;
-    use test::{self, Bencher};
-
-    use crate::aggregation::agg_req::Aggregations;
-    use crate::aggregation::AggregationCollector;
-    use crate::query::{AllQuery, TermQuery};
-    use crate::schema::{IndexRecordOption, Schema, TextFieldIndexing, FAST, STRING};
-    use crate::{Index, Term};
-
-    #[derive(Clone, Copy, Hash, Default, Debug, PartialEq, Eq, PartialOrd, Ord)]
-    enum Cardinality {
-        /// All documents contain exactly one value.
-        /// `Full` is the default for auto-detecting the Cardinality, since it is the most strict.
-        #[default]
-        Full = 0,
-        /// All documents contain at most one value.
-        Optional = 1,
-        /// All documents may contain any number of values.
-        Multivalued = 2,
-        /// 1 / 20 documents has a value
-        Sparse = 3,
-    }
-
-    fn get_collector(agg_req: Aggregations) -> AggregationCollector {
-        AggregationCollector::from_aggs(agg_req, Default::default())
-    }
-
-    fn get_test_index_bench(cardinality: Cardinality) -> crate::Result<Index> {
-        let mut schema_builder = Schema::builder();
-        let text_fieldtype = crate::schema::TextOptions::default()
-            .set_indexing_options(
-                TextFieldIndexing::default().set_index_option(IndexRecordOption::WithFreqs),
-            )
-            .set_stored();
-        let text_field = schema_builder.add_text_field("text", text_fieldtype);
-        let json_field = schema_builder.add_json_field("json", FAST);
-        let text_field_many_terms = schema_builder.add_text_field("text_many_terms", STRING | FAST);
-        let text_field_few_terms = schema_builder.add_text_field("text_few_terms", STRING | FAST);
-        let score_fieldtype = crate::schema::NumericOptions::default().set_fast();
-        let score_field = schema_builder.add_u64_field("score", score_fieldtype.clone());
-        let score_field_f64 = schema_builder.add_f64_field("score_f64", score_fieldtype.clone());
-        let score_field_i64 = schema_builder.add_i64_field("score_i64", score_fieldtype);
-        let index = Index::create_from_tempdir(schema_builder.build())?;
-        let few_terms_data = ["INFO", "ERROR", "WARN", "DEBUG"];
-
-        let lg_norm = rand_distr::LogNormal::new(2.996f64, 0.979f64).unwrap();
-
-        let many_terms_data = (0..150_000)
-            .map(|num| format!("author{}", num))
-            .collect::<Vec<_>>();
-        {
-            let mut rng = StdRng::from_seed([1u8; 32]);
-            let mut index_writer = index.writer_with_num_threads(1, 200_000_000)?;
-            // To make the different test cases comparable we just change one doc to force the
-            // cardinality
-            if cardinality == Cardinality::Optional {
-                index_writer.add_document(doc!())?;
-            }
-            if cardinality == Cardinality::Multivalued {
-                index_writer.add_document(doc!(
-                    json_field => json!({"mixed_type": 10.0}),
-                    json_field => json!({"mixed_type": 10.0}),
-                    text_field => "cool",
-                    text_field => "cool",
-                    text_field_many_terms => "cool",
-                    text_field_many_terms => "cool",
-                    text_field_few_terms => "cool",
-                    text_field_few_terms => "cool",
-                    score_field => 1u64,
-                    score_field => 1u64,
-                    score_field_f64 => lg_norm.sample(&mut rng),
-                    score_field_f64 => lg_norm.sample(&mut rng),
-                    score_field_i64 => 1i64,
-                    score_field_i64 => 1i64,
-                ))?;
-            }
-            let mut doc_with_value = 1_000_000;
-            if cardinality == Cardinality::Sparse {
-                doc_with_value /= 20;
-            }
-            let _val_max = 1_000_000.0;
-            for _ in 0..doc_with_value {
-                let val: f64 = rng.gen_range(0.0..1_000_000.0);
-                let json = if rng.gen_bool(0.1) {
-                    // 10% are numeric values
-                    json!({ "mixed_type": val })
-                } else {
-                    json!({"mixed_type": many_terms_data.choose(&mut rng).unwrap().to_string()})
-                };
-                index_writer.add_document(doc!(
-                    text_field => "cool",
-                    json_field => json,
-                    text_field_many_terms => many_terms_data.choose(&mut rng).unwrap().to_string(),
-                    text_field_few_terms => few_terms_data.choose(&mut rng).unwrap().to_string(),
-                    score_field => val as u64,
-                    score_field_f64 => lg_norm.sample(&mut rng),
-                    score_field_i64 => val as i64,
-                ))?;
-                if cardinality == Cardinality::Sparse {
-                    for _ in 0..20 {
-                        index_writer.add_document(doc!(text_field => "cool"))?;
-                    }
-                }
-            }
-            // writing the segment
-            index_writer.commit()?;
-        }
-
-        Ok(index)
-    }
-
-    use paste::paste;
-    #[macro_export]
-    macro_rules! bench_all_cardinalities {
-        (  $x:ident ) => {
-            paste! {
-                #[bench]
-                fn $x(b: &mut Bencher) {
-                    [<$x _card>](b, Cardinality::Full)
-                }
-
-                #[bench]
-                fn [<$x _opt>](b: &mut Bencher) {
-                    [<$x _card>](b, Cardinality::Optional)
-                }
-
-                #[bench]
-                fn [<$x _multi>](b: &mut Bencher) {
-                    [<$x _card>](b, Cardinality::Multivalued)
-                }
-
-                #[bench]
-                fn [<$x _sparse>](b: &mut Bencher) {
-                    [<$x _card>](b, Cardinality::Sparse)
-                }
-
-            }
-        };
-    }
-
-    bench_all_cardinalities!(bench_aggregation_average_u64);
-
-    fn bench_aggregation_average_u64_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-        let text_field = reader.searcher().schema().get_field("text").unwrap();
-
-        b.iter(|| {
-            let term_query = TermQuery::new(
-                Term::from_field_text(text_field, "cool"),
-                IndexRecordOption::Basic,
-            );
-
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "average": { "avg": { "field": "score", } }
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&term_query, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_stats_f64);
-
-    fn bench_aggregation_stats_f64_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-        let text_field = reader.searcher().schema().get_field("text").unwrap();
-
-        b.iter(|| {
-            let term_query = TermQuery::new(
-                Term::from_field_text(text_field, "cool"),
-                IndexRecordOption::Basic,
-            );
-
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "average_f64": { "stats": { "field": "score_f64", } }
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&term_query, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_average_f64);
-
-    fn bench_aggregation_average_f64_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-        let text_field = reader.searcher().schema().get_field("text").unwrap();
-
-        b.iter(|| {
-            let term_query = TermQuery::new(
-                Term::from_field_text(text_field, "cool"),
-                IndexRecordOption::Basic,
-            );
-
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "average_f64": { "avg": { "field": "score_f64", } }
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&term_query, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_percentiles_f64);
-
-    fn bench_aggregation_percentiles_f64_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req_str = r#"
-            {
-              "mypercentiles": {
-                "percentiles": {
-                  "field": "score_f64",
-                  "percents": [ 95, 99, 99.9 ]
-                }
-              }
-            } "#;
-            let agg_req_1: Aggregations = serde_json::from_str(agg_req_str).unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_average_u64_and_f64);
-
-    fn bench_aggregation_average_u64_and_f64_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-        let text_field = reader.searcher().schema().get_field("text").unwrap();
-
-        b.iter(|| {
-            let term_query = TermQuery::new(
-                Term::from_field_text(text_field, "cool"),
-                IndexRecordOption::Basic,
-            );
-
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "average_f64": { "avg": { "field": "score_f64" } },
-                "average": { "avg": { "field": "score" } },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&term_query, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_terms_few);
-
-    fn bench_aggregation_terms_few_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req: Aggregations = serde_json::from_value(json!({
-                "my_texts": { "terms": { "field": "text_few_terms" } },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_terms_many_with_top_hits_agg);
-
-    fn bench_aggregation_terms_many_with_top_hits_agg_card(
-        b: &mut Bencher,
-        cardinality: Cardinality,
-    ) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req: Aggregations = serde_json::from_value(json!({
-                "my_texts": {
-                    "terms": { "field": "text_many_terms" },
-                    "aggs": {
-                        "top_hits": { "top_hits":
-                            {
-                                "sort": [
-                                    { "score": "desc" }
-                                ],
-                                "size": 2,
-                                "doc_value_fields": ["score_f64"]
-                            }
-                        }
-                    }
-                },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_terms_many_with_sub_agg);
-
-    fn bench_aggregation_terms_many_with_sub_agg_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req: Aggregations = serde_json::from_value(json!({
-                "my_texts": {
-                    "terms": { "field": "text_many_terms" },
-                    "aggs": {
-                        "average_f64": { "avg": { "field": "score_f64" } }
-                    }
-                },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_terms_many_json_mixed_type_with_sub_agg);
-
-    fn bench_aggregation_terms_many_json_mixed_type_with_sub_agg_card(
-        b: &mut Bencher,
-        cardinality: Cardinality,
-    ) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req: Aggregations = serde_json::from_value(json!({
-                "my_texts": {
-                    "terms": { "field": "json.mixed_type" },
-                    "aggs": {
-                        "average_f64": { "avg": { "field": "score_f64" } }
-                    }
-                },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_terms_many2);
-
-    fn bench_aggregation_terms_many2_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req: Aggregations = serde_json::from_value(json!({
-                "my_texts": { "terms": { "field": "text_many_terms" } },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_terms_many_order_by_term);
-
-    fn bench_aggregation_terms_many_order_by_term_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req: Aggregations = serde_json::from_value(json!({
-                "my_texts": { "terms": { "field": "text_many_terms", "order": { "_key": "desc" } } },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_range_only);
-
-    fn bench_aggregation_range_only_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "range_f64": { "range": { "field": "score_f64", "ranges": [
-                    { "from": 3, "to": 7000 },
-                    { "from": 7000, "to": 20000 },
-                    { "from": 20000, "to": 30000 },
-                    { "from": 30000, "to": 40000 },
-                    { "from": 40000, "to": 50000 },
-                    { "from": 50000, "to": 60000 }
-                ] } },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_range_with_avg);
-
-    fn bench_aggregation_range_with_avg_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "rangef64": {
-                    "range": {
-                        "field": "score_f64",
-                        "ranges": [
-                            { "from": 3, "to": 7000 },
-                            { "from": 7000, "to": 20000 },
-                            { "from": 20000, "to": 30000 },
-                            { "from": 30000, "to": 40000 },
-                            { "from": 40000, "to": 50000 },
-                            { "from": 50000, "to": 60000 }
-                        ]
-                    },
-                    "aggs": {
-                        "average_f64": { "avg": { "field": "score_f64" } }
-                    }
-                },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    // hard bounds has a different algorithm, because it actually limits collection range
-    //
-    bench_all_cardinalities!(bench_aggregation_histogram_only_hard_bounds);
-
-    fn bench_aggregation_histogram_only_hard_bounds_card(
-        b: &mut Bencher,
-        cardinality: Cardinality,
-    ) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "rangef64": { "histogram": { "field": "score_f64", "interval": 100, "hard_bounds": { "min": 1000, "max": 300000 } } },
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_histogram_with_avg);
-
-    fn bench_aggregation_histogram_with_avg_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "rangef64": {
-                    "histogram": { "field": "score_f64", "interval": 100 },
-                    "aggs": {
-                        "average_f64": { "avg": { "field": "score_f64" } }
-                    }
-                }
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_histogram_only);
-
-    fn bench_aggregation_histogram_only_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-
-        b.iter(|| {
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "rangef64": {
-                    "histogram": {
-                        "field": "score_f64",
-                        "interval": 100 // 1000 buckets
-                    },
-                }
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&AllQuery, &collector).unwrap()
-        });
-    }
-
-    bench_all_cardinalities!(bench_aggregation_avg_and_range_with_avg);
-
-    fn bench_aggregation_avg_and_range_with_avg_card(b: &mut Bencher, cardinality: Cardinality) {
-        let index = get_test_index_bench(cardinality).unwrap();
-        let reader = index.reader().unwrap();
-        let text_field = reader.searcher().schema().get_field("text").unwrap();
-
-        b.iter(|| {
-            let term_query = TermQuery::new(
-                Term::from_field_text(text_field, "cool"),
-                IndexRecordOption::Basic,
-            );
-
-            let agg_req_1: Aggregations = serde_json::from_value(json!({
-                "rangef64": {
-                    "range": {
-                        "field": "score_f64",
-                        "ranges": [
-                            { "from": 3, "to": 7000 },
-                            { "from": 7000, "to": 20000 },
-                            { "from": 20000, "to": 60000 }
-                        ]
-                    },
-                    "aggs": {
-                        "average_in_range": { "avg": { "field": "score" } }
-                    }
-                },
-                "average": { "avg": { "field": "score" } }
-            }))
-            .unwrap();
-
-            let collector = get_collector(agg_req_1);
-
-            let searcher = reader.searcher();
-            searcher.search(&term_query, &collector).unwrap()
-        });
-    }
-}

src/aggregation/agg_limits.rs

@@ -81,10 +81,11 @@ impl AggregationLimits {
         }
     }
 
-    pub(crate) fn add_memory_consumed(&self, num_bytes: u64) -> crate::Result<()> {
-        self.memory_consumption
-            .fetch_add(num_bytes, Ordering::Relaxed);
-        validate_memory_consumption(&self.memory_consumption, self.memory_limit)?;
+    pub(crate) fn add_memory_consumed(&self, add_num_bytes: u64) -> crate::Result<()> {
+        let prev_value = self
+            .memory_consumption
+            .fetch_add(add_num_bytes, Ordering::Relaxed);
+        validate_memory_consumption(prev_value + add_num_bytes, self.memory_limit)?;
         Ok(())
     }
 
@@ -94,11 +95,11 @@ impl AggregationLimits {
 }
 
 fn validate_memory_consumption(
-    memory_consumption: &AtomicU64,
+    memory_consumption: u64,
     memory_limit: ByteCount,
 ) -> Result<(), AggregationError> {
     // Load the estimated memory consumed by the aggregations
-    let memory_consumed: ByteCount = memory_consumption.load(Ordering::Relaxed).into();
+    let memory_consumed: ByteCount = memory_consumption.into();
     if memory_consumed > memory_limit {
         return Err(AggregationError::MemoryExceeded {
             limit: memory_limit,
@@ -118,10 +119,11 @@ pub struct ResourceLimitGuard {
 }
 
 impl ResourceLimitGuard {
-    pub(crate) fn add_memory_consumed(&self, num_bytes: u64) -> crate::Result<()> {
-        self.memory_consumption
-            .fetch_add(num_bytes, Ordering::Relaxed);
-        validate_memory_consumption(&self.memory_consumption, self.memory_limit)?;
+    pub(crate) fn add_memory_consumed(&self, add_num_bytes: u64) -> crate::Result<()> {
+        let prev_value = self
+            .memory_consumption
+            .fetch_add(add_num_bytes, Ordering::Relaxed);
+        validate_memory_consumption(prev_value + add_num_bytes, self.memory_limit)?;
         Ok(())
     }
 }

src/aggregation/agg_req_with_accessor.rs

@@ -17,7 +17,8 @@ use super::metric::{
 use super::segment_agg_result::AggregationLimits;
 use super::VecWithNames;
 use crate::aggregation::{f64_to_fastfield_u64, Key};
-use crate::{SegmentOrdinal, SegmentReader};
+use crate::index::SegmentReader;
+use crate::SegmentOrdinal;
 
 #[derive(Default)]
 pub(crate) struct AggregationsWithAccessor {

src/aggregation/bucket/histogram/histogram.rs

@@ -331,9 +331,11 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
         }
 
         let mem_delta = self.get_memory_consumption() - mem_pre;
-        bucket_agg_accessor
-            .limits
-            .add_memory_consumed(mem_delta as u64)?;
+        if mem_delta > 0 {
+            bucket_agg_accessor
+                .limits
+                .add_memory_consumed(mem_delta as u64)?;
+        }
 
         Ok(())
     }

src/aggregation/bucket/term_agg.rs

@@ -324,9 +324,11 @@ impl SegmentAggregationCollector for SegmentTermCollector {
         }
 
         let mem_delta = self.get_memory_consumption() - mem_pre;
-        bucket_agg_accessor
-            .limits
-            .add_memory_consumed(mem_delta as u64)?;
+        if mem_delta > 0 {
+            bucket_agg_accessor
+                .limits
+                .add_memory_consumed(mem_delta as u64)?;
+        }
 
         Ok(())
     }

src/aggregation/collector.rs

@@ -8,7 +8,8 @@ use super::segment_agg_result::{
 };
 use crate::aggregation::agg_req_with_accessor::get_aggs_with_segment_accessor_and_validate;
 use crate::collector::{Collector, SegmentCollector};
-use crate::{DocId, SegmentOrdinal, SegmentReader, TantivyError};
+use crate::index::SegmentReader;
+use crate::{DocId, SegmentOrdinal, TantivyError};
 
 /// The default max bucket count, before the aggregation fails.
 pub const DEFAULT_BUCKET_LIMIT: u32 = 65000;

src/aggregation/mod.rs

@@ -143,8 +143,6 @@ use std::fmt::Display;
 #[cfg(test)]
 mod agg_tests;
 
-mod agg_bench;
-
 use core::fmt;
 
 pub use agg_limits::AggregationLimits;

src/collector/top_collector.rs

@@ -4,7 +4,8 @@ use std::marker::PhantomData;
 use serde::{Deserialize, Serialize};
 
 use super::top_score_collector::TopNComputer;
-use crate::{DocAddress, DocId, SegmentOrdinal, SegmentReader};
+use crate::index::SegmentReader;
+use crate::{DocAddress, DocId, SegmentOrdinal};
 
 /// Contains a feature (field, score, etc.) of a document along with the document address.
 ///

src/core/executor.rs

@@ -1,19 +1,25 @@
-use rayon::{ThreadPool, ThreadPoolBuilder};
+use std::sync::Arc;
+
+#[cfg(feature = "quickwit")]
+use futures_util::{future::Either, FutureExt};
 
 use crate::TantivyError;
 
-/// Search executor whether search request are single thread or multithread.
-///
-/// We don't expose Rayon thread pool directly here for several reasons.
-///
-/// First dependency hell. It is not a good idea to expose the
-/// API of a dependency, knowing it might conflict with a different version
-/// used by the client. Second, we may stop using rayon in the future.
+/// Executor makes it possible to run tasks in single thread or
+/// in a thread pool.
+#[derive(Clone)]
 pub enum Executor {
     /// Single thread variant of an Executor
     SingleThread,
     /// Thread pool variant of an Executor
-    ThreadPool(ThreadPool),
+    ThreadPool(Arc<rayon::ThreadPool>),
+}
+
+#[cfg(feature = "quickwit")]
+impl From<Arc<rayon::ThreadPool>> for Executor {
+    fn from(thread_pool: Arc<rayon::ThreadPool>) -> Self {
+        Executor::ThreadPool(thread_pool)
+    }
 }
 
 impl Executor {
@@ -24,11 +30,11 @@ impl Executor {
 
     /// Creates an Executor that dispatches the tasks in a thread pool.
     pub fn multi_thread(num_threads: usize, prefix: &'static str) -> crate::Result<Executor> {
-        let pool = ThreadPoolBuilder::new()
+        let pool = rayon::ThreadPoolBuilder::new()
             .num_threads(num_threads)
             .thread_name(move |num| format!("{prefix}{num}"))
             .build()?;
-        Ok(Executor::ThreadPool(pool))
+        Ok(Executor::ThreadPool(Arc::new(pool)))
     }
 
     /// Perform a map in the thread pool.
@@ -91,11 +97,36 @@ impl Executor {
             }
         }
     }
+
+    /// Spawn a task on the pool, returning a future completing on task success.
+    ///
+    /// If the task panic, returns `Err(())`.
+    #[cfg(feature = "quickwit")]
+    pub fn spawn_blocking<T: Send + 'static>(
+        &self,
+        cpu_intensive_task: impl FnOnce() -> T + Send + 'static,
+    ) -> impl std::future::Future<Output = Result<T, ()>> {
+        match self {
+            Executor::SingleThread => Either::Left(std::future::ready(Ok(cpu_intensive_task()))),
+            Executor::ThreadPool(pool) => {
+                let (sender, receiver) = oneshot::channel();
+                pool.spawn(|| {
+                    if sender.is_closed() {
+                        return;
+                    }
+                    let task_result = cpu_intensive_task();
+                    let _ = sender.send(task_result);
+                });
+
+                let res = receiver.map(|res| res.map_err(|_| ()));
+                Either::Right(res)
+            }
+        }
+    }
 }
 
 #[cfg(test)]
 mod tests {
-
     use super::Executor;
 
     #[test]
@@ -147,4 +178,34 @@ mod tests {
             assert_eq!(result[i], i * 2);
         }
     }
+
+    #[cfg(feature = "quickwit")]
+    #[test]
+    fn test_cancel_cpu_intensive_tasks() {
+        use std::sync::atomic::{AtomicU64, Ordering};
+        use std::sync::Arc;
+        use std::time::Duration;
+
+        let counter: Arc<AtomicU64> = Default::default();
+        let mut futures = Vec::new();
+        let executor = Executor::multi_thread(3, "search-test").unwrap();
+        for _ in 0..1_000 {
+            let counter_clone = counter.clone();
+            let fut = executor.spawn_blocking(move || {
+                std::thread::sleep(Duration::from_millis(4));
+                counter_clone.fetch_add(1, Ordering::SeqCst)
+            });
+            futures.push(fut);
+        }
+        std::thread::sleep(Duration::from_millis(5));
+        // The first few num_cores tasks should run, but the other should get cancelled.
+        drop(futures);
+        while Arc::strong_count(&counter) > 1 {
+            std::thread::sleep(Duration::from_millis(10));
+        }
+        // with ideal timing, we expect the result to always be 6, but as long as we run some, and
+        // cancelled most, the test is a success
+        assert!(counter.load(Ordering::SeqCst) > 0);
+        assert!(counter.load(Ordering::SeqCst) < 50);
+    }
 }

src/core/json_utils.rs

@@ -31,7 +31,7 @@ use crate::{DateTime, DocId, Term};
 /// position 1.
 /// As a result, with lemmatization, "The Smiths" will match our object.
 ///
-/// Worse, if a same term is appears in the second object, a non increasing value would be pushed
+/// Worse, if a same term appears in the second object, a non increasing value would be pushed
 /// to the position recorder probably provoking a panic.
 ///
 /// This problem is solved for regular multivalued object by offsetting the position
@@ -50,7 +50,7 @@ use crate::{DateTime, DocId, Term};
 /// We can therefore afford working with a map that is not imperfect. It is fine if several
 /// path map to the same index position as long as the probability is relatively low.
 #[derive(Default)]
-struct IndexingPositionsPerPath {
+pub(crate) struct IndexingPositionsPerPath {
     positions_per_path: FxHashMap<u32, IndexingPosition>,
 }
 
@@ -58,6 +58,9 @@ impl IndexingPositionsPerPath {
     fn get_position_from_id(&mut self, id: u32) -> &mut IndexingPosition {
         self.positions_per_path.entry(id).or_default()
     }
+    pub fn clear(&mut self) {
+        self.positions_per_path.clear();
+    }
 }
 
 /// Convert JSON_PATH_SEGMENT_SEP to a dot.
@@ -68,36 +71,6 @@ pub fn json_path_sep_to_dot(path: &mut str) {
     }
 }
 
-#[allow(clippy::too_many_arguments)]
-pub(crate) fn index_json_values<'a, V: Value<'a>>(
-    doc: DocId,
-    json_visitors: impl Iterator<Item = crate::Result<V::ObjectIter>>,
-    text_analyzer: &mut TextAnalyzer,
-    expand_dots_enabled: bool,
-    term_buffer: &mut Term,
-    postings_writer: &mut dyn PostingsWriter,
-    json_path_writer: &mut JsonPathWriter,
-    ctx: &mut IndexingContext,
-) -> crate::Result<()> {
-    json_path_writer.clear();
-    json_path_writer.set_expand_dots(expand_dots_enabled);
-    let mut positions_per_path: IndexingPositionsPerPath = Default::default();
-    for json_visitor_res in json_visitors {
-        let json_visitor = json_visitor_res?;
-        index_json_object::<V>(
-            doc,
-            json_visitor,
-            text_analyzer,
-            term_buffer,
-            json_path_writer,
-            postings_writer,
-            ctx,
-            &mut positions_per_path,
-        );
-    }
-    Ok(())
-}
-
 #[allow(clippy::too_many_arguments)]
 fn index_json_object<'a, V: Value<'a>>(
     doc: DocId,
@@ -126,7 +99,7 @@ fn index_json_object<'a, V: Value<'a>>(
 }
 
 #[allow(clippy::too_many_arguments)]
-fn index_json_value<'a, V: Value<'a>>(
+pub(crate) fn index_json_value<'a, V: Value<'a>>(
     doc: DocId,
     json_value: V,
     text_analyzer: &mut TextAnalyzer,
@@ -166,12 +139,18 @@ fn index_json_value<'a, V: Value<'a>>(
                 );
             }
             ReferenceValueLeaf::U64(val) => {
+                // try to parse to i64, since when querying we will apply the same logic and prefer
+                // i64 values
                 set_path_id(
                     term_buffer,
                     ctx.path_to_unordered_id
                         .get_or_allocate_unordered_id(json_path_writer.as_str()),
                 );
-                term_buffer.append_type_and_fast_value(val);
+                if let Ok(i64_val) = val.try_into() {
+                    term_buffer.append_type_and_fast_value::<i64>(i64_val);
+                } else {
+                    term_buffer.append_type_and_fast_value(val);
+                }
                 postings_writer.subscribe(doc, 0u32, term_buffer, ctx);
             }
             ReferenceValueLeaf::I64(val) => {
@@ -257,10 +236,7 @@ fn index_json_value<'a, V: Value<'a>>(
 /// Tries to infer a JSON type from a string and append it to the term.
 ///
 /// The term must be json + JSON path.
-pub(crate) fn convert_to_fast_value_and_append_to_json_term(
-    mut term: Term,
-    phrase: &str,
-) -> Option<Term> {
+pub fn convert_to_fast_value_and_append_to_json_term(mut term: Term, phrase: &str) -> Option<Term> {
     assert_eq!(
         term.value()
             .as_json_value_bytes()

src/core/searcher.rs

@@ -4,13 +4,13 @@ use std::{fmt, io};
 
 use crate::collector::Collector;
 use crate::core::Executor;
-use crate::index::SegmentReader;
+use crate::index::{SegmentId, SegmentReader};
 use crate::query::{Bm25StatisticsProvider, EnableScoring, Query};
 use crate::schema::document::DocumentDeserialize;
 use crate::schema::{Schema, Term};
 use crate::space_usage::SearcherSpaceUsage;
 use crate::store::{CacheStats, StoreReader};
-use crate::{DocAddress, Index, Opstamp, SegmentId, TrackedObject};
+use crate::{DocAddress, Index, Opstamp, TrackedObject};
 
 /// Identifies the searcher generation accessed by a [`Searcher`].
 ///
@@ -109,8 +109,9 @@ impl Searcher {
         &self,
         doc_address: DocAddress,
     ) -> crate::Result<D> {
+        let executor = self.inner.index.search_executor();
         let store_reader = &self.inner.store_readers[doc_address.segment_ord as usize];
-        store_reader.get_async(doc_address.doc_id).await
+        store_reader.get_async(doc_address.doc_id, executor).await
     }
 
     /// Access the schema associated with the index of this searcher.

src/core/tests.rs

@@ -1,12 +1,14 @@
 use crate::collector::Count;
 use crate::directory::{RamDirectory, WatchCallback};
+use crate::index::SegmentId;
 use crate::indexer::{LogMergePolicy, NoMergePolicy};
+use crate::postings::Postings;
 use crate::query::TermQuery;
 use crate::schema::{Field, IndexRecordOption, Schema, INDEXED, STRING, TEXT};
 use crate::tokenizer::TokenizerManager;
 use crate::{
-    Directory, DocSet, Index, IndexBuilder, IndexReader, IndexSettings, IndexWriter, Postings,
-    ReloadPolicy, SegmentId, TantivyDocument, Term,
+    Directory, DocSet, Index, IndexBuilder, IndexReader, IndexSettings, IndexWriter, ReloadPolicy,
+    TantivyDocument, Term,
 };
 
 #[test]
@@ -417,7 +419,7 @@ fn test_non_text_json_term_freq() {
     let inv_idx = segment_reader.inverted_index(field).unwrap();
 
     let mut term = Term::from_field_json_path(field, "tenant_id", false);
-    term.append_type_and_fast_value(75u64);
+    term.append_type_and_fast_value(75i64);
 
     let postings = inv_idx
         .read_postings(&term, IndexRecordOption::WithFreqsAndPositions)
@@ -451,7 +453,7 @@ fn test_non_text_json_term_freq_bitpacked() {
     let inv_idx = segment_reader.inverted_index(field).unwrap();
 
     let mut term = Term::from_field_json_path(field, "tenant_id", false);
-    term.append_type_and_fast_value(75u64);
+    term.append_type_and_fast_value(75i64);
 
     let mut postings = inv_idx
         .read_postings(&term, IndexRecordOption::WithFreqsAndPositions)

src/fastfield/facet_reader.rs

@@ -146,8 +146,11 @@ mod tests {
         facet_ords.extend(facet_reader.facet_ords(0u32));
         assert_eq!(&facet_ords, &[0u64]);
         let doc = searcher.doc::<TantivyDocument>(DocAddress::new(0u32, 0u32))?;
-        let value: Option<&Facet> = doc.get_first(facet_field).and_then(|v| v.as_facet());
-        assert_eq!(value, Facet::from_text("/a/b").ok().as_ref());
+        let value: Option<Facet> = doc
+            .get_first(facet_field)
+            .and_then(|v| v.as_facet())
+            .map(|facet| Facet::from_encoded_string(facet.to_string()));
+        assert_eq!(value, Facet::from_text("/a/b").ok());
         Ok(())
     }
 

src/fastfield/mod.rs

@@ -80,7 +80,7 @@ mod tests {
     use std::path::Path;
 
     use columnar::StrColumn;
-    use common::{ByteCount, HasLen, TerminatingWrite};
+    use common::{ByteCount, DateTimePrecision, HasLen, TerminatingWrite};
     use once_cell::sync::Lazy;
     use rand::prelude::SliceRandom;
     use rand::rngs::StdRng;
@@ -88,14 +88,15 @@ mod tests {
 
     use super::*;
     use crate::directory::{Directory, RamDirectory, WritePtr};
+    use crate::index::SegmentId;
     use crate::merge_policy::NoMergePolicy;
     use crate::schema::{
-        Facet, FacetOptions, Field, JsonObjectOptions, Schema, SchemaBuilder, TantivyDocument,
-        TextOptions, FAST, INDEXED, STORED, STRING, TEXT,
+        DateOptions, Facet, FacetOptions, Field, JsonObjectOptions, Schema, SchemaBuilder,
+        TantivyDocument, TextOptions, FAST, INDEXED, STORED, STRING, TEXT,
     };
     use crate::time::OffsetDateTime;
     use crate::tokenizer::{LowerCaser, RawTokenizer, TextAnalyzer, TokenizerManager};
-    use crate::{DateOptions, DateTimePrecision, Index, IndexWriter, SegmentId, SegmentReader};
+    use crate::{Index, IndexWriter, SegmentReader};
 
     pub static SCHEMA: Lazy<Schema> = Lazy::new(|| {
         let mut schema_builder = Schema::builder();

src/fastfield/writer.rs

@@ -1,14 +1,14 @@
 use std::io;
 
 use columnar::{ColumnarWriter, NumericalValue};
-use common::JsonPathWriter;
+use common::{DateTimePrecision, JsonPathWriter};
 use tokenizer_api::Token;
 
 use crate::indexer::doc_id_mapping::DocIdMapping;
 use crate::schema::document::{Document, ReferenceValue, ReferenceValueLeaf, Value};
 use crate::schema::{value_type_to_column_type, Field, FieldType, Schema, Type};
 use crate::tokenizer::{TextAnalyzer, TokenizerManager};
-use crate::{DateTimePrecision, DocId, TantivyError};
+use crate::{DocId, TantivyError};
 
 /// Only index JSON down to a depth of 20.
 /// This is mostly to guard us from a stack overflow triggered by malicious input.
@@ -183,8 +183,7 @@ impl FastFieldsWriter {
                         .record_datetime(doc_id, field_name, truncated_datetime);
                 }
                 ReferenceValueLeaf::Facet(val) => {
-                    self.columnar_writer
-                        .record_str(doc_id, field_name, val.encoded_str());
+                    self.columnar_writer.record_str(doc_id, field_name, val);
                 }
                 ReferenceValueLeaf::Bytes(val) => {
                     self.columnar_writer.record_bytes(doc_id, field_name, val);

src/index/index.rs

@@ -3,7 +3,7 @@ use std::fmt;
 #[cfg(feature = "mmap")]
 use std::path::Path;
 use std::path::PathBuf;
-use std::sync::Arc;
+use std::thread::available_parallelism;
 
 use super::segment::Segment;
 use super::segment_reader::merge_field_meta_data;
@@ -293,7 +293,7 @@ pub struct Index {
     directory: ManagedDirectory,
     schema: Schema,
     settings: IndexSettings,
-    executor: Arc<Executor>,
+    executor: Executor,
     tokenizers: TokenizerManager,
     fast_field_tokenizers: TokenizerManager,
     inventory: SegmentMetaInventory,
@@ -318,29 +318,25 @@ impl Index {
     ///
     /// By default the executor is single thread, and simply runs in the calling thread.
     pub fn search_executor(&self) -> &Executor {
-        self.executor.as_ref()
+        &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 = Arc::new(Executor::multi_thread(num_threads, "tantivy-search-")?);
+        self.executor = Executor::multi_thread(num_threads, "tantivy-search-")?;
         Ok(())
     }
 
     /// Custom thread pool by a outer thread pool.
-    pub fn set_shared_multithread_executor(
-        &mut self,
-        shared_thread_pool: Arc<Executor>,
-    ) -> crate::Result<()> {
-        self.executor = shared_thread_pool.clone();
-        Ok(())
+    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 = num_cpus::get();
+        let default_num_threads = available_parallelism()?.get();
         self.set_multithread_executor(default_num_threads)
     }
 
@@ -418,7 +414,7 @@ impl Index {
             schema,
             tokenizers: TokenizerManager::default(),
             fast_field_tokenizers: TokenizerManager::default(),
-            executor: Arc::new(Executor::single_thread()),
+            executor: Executor::single_thread(),
             inventory,
         }
     }
@@ -621,7 +617,7 @@ impl Index {
         &self,
         memory_budget_in_bytes: usize,
     ) -> crate::Result<IndexWriter<D>> {
-        let mut num_threads = std::cmp::min(num_cpus::get(), MAX_NUM_THREAD);
+        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 {
             num_threads = (memory_budget_in_bytes / MEMORY_BUDGET_NUM_BYTES_MIN).max(1);

src/index/mod.rs

@@ -1,5 +1,3 @@
-//! # Index Module
-//!
 //! The `index` module in Tantivy contains core components to read and write indexes.
 //!
 //! It contains `Index` and `Segment`, where a `Index` consists of one or more `Segment`s.

src/indexer/delete_queue.rs

@@ -246,8 +246,9 @@ impl DeleteCursor {
 mod tests {
 
     use super::{DeleteOperation, DeleteQueue};
+    use crate::index::SegmentReader;
     use crate::query::{Explanation, Scorer, Weight};
-    use crate::{DocId, Score, SegmentReader};
+    use crate::{DocId, Score};
 
     struct DummyWeight;
     impl Weight for DummyWeight {

src/indexer/log_merge_policy.rs

@@ -144,9 +144,9 @@ mod tests {
     use once_cell::sync::Lazy;
 
     use super::*;
-    use crate::index::SegmentMetaInventory;
+    use crate::index::{SegmentId, SegmentMetaInventory};
+    use crate::schema;
     use crate::schema::INDEXED;
-    use crate::{schema, SegmentId};
 
     static INVENTORY: Lazy<SegmentMetaInventory> = Lazy::new(SegmentMetaInventory::default);
 

src/indexer/merge_operation.rs

@@ -1,7 +1,8 @@
 use std::collections::HashSet;
 use std::ops::Deref;
 
-use crate::{Inventory, Opstamp, SegmentId, TrackedObject};
+use crate::index::SegmentId;
+use crate::{Inventory, Opstamp, TrackedObject};
 
 #[derive(Default)]
 pub(crate) struct MergeOperationInventory(Inventory<InnerMergeOperation>);

src/indexer/merger.rs

@@ -13,7 +13,7 @@ use crate::docset::{DocSet, TERMINATED};
 use crate::error::DataCorruption;
 use crate::fastfield::{AliveBitSet, FastFieldNotAvailableError};
 use crate::fieldnorm::{FieldNormReader, FieldNormReaders, FieldNormsSerializer, FieldNormsWriter};
-use crate::index::{Segment, SegmentReader};
+use crate::index::{Segment, SegmentComponent, SegmentReader};
 use crate::indexer::doc_id_mapping::{MappingType, SegmentDocIdMapping};
 use crate::indexer::SegmentSerializer;
 use crate::postings::{InvertedIndexSerializer, Postings, SegmentPostings};
@@ -21,8 +21,7 @@ use crate::schema::{value_type_to_column_type, Field, FieldType, Schema};
 use crate::store::StoreWriter;
 use crate::termdict::{TermMerger, TermOrdinal};
 use crate::{
-    DocAddress, DocId, IndexSettings, IndexSortByField, InvertedIndexReader, Order,
-    SegmentComponent, SegmentOrdinal,
+    DocAddress, DocId, IndexSettings, IndexSortByField, InvertedIndexReader, Order, SegmentOrdinal,
 };
 
 /// Segment's max doc must be `< MAX_DOC_LIMIT`.
@@ -794,7 +793,7 @@ mod tests {
         BytesFastFieldTestCollector, FastFieldTestCollector, TEST_COLLECTOR_WITH_SCORE,
     };
     use crate::collector::{Count, FacetCollector};
-    use crate::index::Index;
+    use crate::index::{Index, SegmentId};
     use crate::query::{AllQuery, BooleanQuery, EnableScoring, Scorer, TermQuery};
     use crate::schema::document::Value;
     use crate::schema::{
@@ -804,7 +803,7 @@ mod tests {
     use crate::time::OffsetDateTime;
     use crate::{
         assert_nearly_equals, schema, DateTime, DocAddress, DocId, DocSet, IndexSettings,
-        IndexSortByField, IndexWriter, Order, Searcher, SegmentId,
+        IndexSortByField, IndexWriter, Order, Searcher,
     };
 
     #[test]

src/indexer/merger_sorted_index_test.rs

@@ -3,6 +3,7 @@ mod tests {
     use crate::collector::TopDocs;
     use crate::fastfield::AliveBitSet;
     use crate::index::Index;
+    use crate::postings::Postings;
     use crate::query::QueryParser;
     use crate::schema::document::Value;
     use crate::schema::{
@@ -10,8 +11,8 @@ mod tests {
         TextFieldIndexing, TextOptions,
     };
     use crate::{
-        DocAddress, DocSet, IndexSettings, IndexSortByField, IndexWriter, Order, Postings,
-        TantivyDocument, Term,
+        DocAddress, DocSet, IndexSettings, IndexSortByField, IndexWriter, Order, TantivyDocument,
+        Term,
     };
 
     fn create_test_index_posting_list_issue(index_settings: Option<IndexSettings>) -> Index {

src/indexer/mod.rs

@@ -182,7 +182,7 @@ mod tests_mmap {
         let index = Index::create_in_ram(schema_builder.build());
         let mut index_writer = index.writer_for_tests().unwrap();
         index_writer
-            .add_document(doc!(field=>json!({format!("{field_name_in}"): "test1"})))
+            .add_document(doc!(field=>json!({format!("{field_name_in}"): "test1", format!("num{field_name_in}"): 10})))
             .unwrap();
         index_writer
             .add_document(doc!(field=>json!({format!("a{field_name_in}"): "test2"})))
@@ -260,6 +260,64 @@ mod tests_mmap {
             "test6",
         );
         test_agg(format!("json.{field_name_out}a").as_str(), "test7");
+
+        // `.` is stored as `\u{0001}` internally in tantivy
+        let field_name_out_internal = if field_name_out == "." {
+            "\u{0001}"
+        } else {
+            field_name_out
+        };
+
+        let mut fields = reader.searcher().segment_readers()[0]
+            .inverted_index(field)
+            .unwrap()
+            .list_encoded_fields()
+            .unwrap();
+        assert_eq!(fields.len(), 8);
+        fields.sort();
+        let mut expected_fields = vec![
+            (format!("a{field_name_out_internal}"), Type::Str),
+            (format!("a{field_name_out_internal}a"), Type::Str),
+            (
+                format!("a{field_name_out_internal}a{field_name_out_internal}"),
+                Type::Str,
+            ),
+            (
+                format!("a{field_name_out_internal}\u{1}ab{field_name_out_internal}"),
+                Type::Str,
+            ),
+            (
+                format!("a{field_name_out_internal}\u{1}a{field_name_out_internal}"),
+                Type::Str,
+            ),
+            (format!("{field_name_out_internal}a"), Type::Str),
+            (format!("{field_name_out_internal}"), Type::Str),
+            (format!("num{field_name_out_internal}"), Type::I64),
+        ];
+        expected_fields.sort();
+        assert_eq!(fields, expected_fields);
+        // Check columnar reader
+        let mut columns = reader.searcher().segment_readers()[0]
+            .fast_fields()
+            .columnar()
+            .list_columns()
+            .unwrap()
+            .into_iter()
+            .map(|(name, _)| name)
+            .collect::<Vec<_>>();
+        let mut expected_columns = vec![
+            format!("json\u{1}{field_name_out_internal}"),
+            format!("json\u{1}{field_name_out_internal}a"),
+            format!("json\u{1}a{field_name_out_internal}"),
+            format!("json\u{1}a{field_name_out_internal}a"),
+            format!("json\u{1}a{field_name_out_internal}a{field_name_out_internal}"),
+            format!("json\u{1}a{field_name_out_internal}\u{1}ab{field_name_out_internal}"),
+            format!("json\u{1}a{field_name_out_internal}\u{1}a{field_name_out_internal}"),
+            format!("json\u{1}num{field_name_out_internal}"),
+        ];
+        columns.sort();
+        expected_columns.sort();
+        assert_eq!(columns, expected_columns);
     }
 
     #[test]

src/indexer/segment_writer.rs

@@ -5,20 +5,20 @@ use tokenizer_api::BoxTokenStream;
 
 use super::doc_id_mapping::{get_doc_id_mapping_from_field, DocIdMapping};
 use super::operation::AddOperation;
-use crate::core::json_utils::index_json_values;
 use crate::fastfield::FastFieldsWriter;
 use crate::fieldnorm::{FieldNormReaders, FieldNormsWriter};
-use crate::index::Segment;
+use crate::index::{Segment, SegmentComponent};
 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,
 };
-use crate::schema::document::{Document, ReferenceValue, Value};
+use crate::schema::document::{Document, Value};
 use crate::schema::{FieldEntry, FieldType, Schema, Term, DATE_TIME_PRECISION_INDEXED};
 use crate::store::{StoreReader, StoreWriter};
 use crate::tokenizer::{FacetTokenizer, PreTokenizedStream, TextAnalyzer, Tokenizer};
-use crate::{DocId, Opstamp, SegmentComponent, TantivyError};
+use crate::{DocId, Opstamp, TantivyError};
 
 /// Computes the initial size of the hash table.
 ///
@@ -68,6 +68,7 @@ pub struct SegmentWriter {
     pub(crate) fast_field_writers: FastFieldsWriter,
     pub(crate) fieldnorms_writer: FieldNormsWriter,
     pub(crate) json_path_writer: JsonPathWriter,
+    pub(crate) json_positions_per_path: IndexingPositionsPerPath,
     pub(crate) doc_opstamps: Vec<Opstamp>,
     per_field_text_analyzers: Vec<TextAnalyzer>,
     term_buffer: Term,
@@ -119,6 +120,7 @@ impl SegmentWriter {
             per_field_postings_writers,
             fieldnorms_writer: FieldNormsWriter::for_schema(&schema),
             json_path_writer: JsonPathWriter::default(),
+            json_positions_per_path: IndexingPositionsPerPath::default(),
             segment_serializer,
             fast_field_writers: FastFieldsWriter::from_schema_and_tokenizer_manager(
                 &schema,
@@ -204,8 +206,7 @@ impl SegmentWriter {
                         // Used to help with linting and type checking.
                         let value = value_access as D::Value<'_>;
 
-                        let facet = value.as_facet().ok_or_else(make_schema_error)?;
-                        let facet_str = facet.encoded_str();
+                        let facet_str = value.as_facet().ok_or_else(make_schema_error)?;
                         let mut facet_tokenizer = facet_tokenizer.token_stream(facet_str);
                         let mut indexing_position = IndexingPosition::default();
                         postings_writer.index_text(
@@ -228,7 +229,7 @@ impl SegmentWriter {
                                 &mut self.per_field_text_analyzers[field.field_id() as usize];
                             text_analyzer.token_stream(text)
                         } else if let Some(tok_str) = value.as_pre_tokenized_text() {
-                            BoxTokenStream::new(PreTokenizedStream::from(tok_str.clone()))
+                            BoxTokenStream::new(PreTokenizedStream::from(*tok_str.clone()))
                         } else {
                             continue;
                         };
@@ -342,26 +343,24 @@ impl SegmentWriter {
                 FieldType::JsonObject(json_options) => {
                     let text_analyzer =
                         &mut self.per_field_text_analyzers[field.field_id() as usize];
-                    let json_values_it = values.map(|value_access| {
-                        // Used to help with linting and type checking.
-                        let value_access = value_access as D::Value<'_>;
-                        let value = value_access.as_value();
 
-                        match value {
-                            ReferenceValue::Object(object_iter) => Ok(object_iter),
-                            _ => Err(make_schema_error()),
-                        }
-                    });
-                    index_json_values::<D::Value<'_>>(
-                        doc_id,
-                        json_values_it,
-                        text_analyzer,
-                        json_options.is_expand_dots_enabled(),
-                        term_buffer,
-                        postings_writer,
-                        &mut self.json_path_writer,
-                        ctx,
-                    )?;
+                    self.json_positions_per_path.clear();
+                    self.json_path_writer
+                        .set_expand_dots(json_options.is_expand_dots_enabled());
+                    for json_value in values {
+                        self.json_path_writer.clear();
+
+                        index_json_value(
+                            doc_id,
+                            json_value,
+                            text_analyzer,
+                            term_buffer,
+                            &mut self.json_path_writer,
+                            postings_writer,
+                            ctx,
+                            &mut self.json_positions_per_path,
+                        );
+                    }
                 }
                 FieldType::IpAddr(_) => {
                     let mut num_vals = 0;
@@ -498,19 +497,20 @@ mod tests {
     use crate::collector::{Count, TopDocs};
     use crate::directory::RamDirectory;
     use crate::fastfield::FastValue;
-    use crate::postings::TermInfo;
+    use crate::postings::{Postings, TermInfo};
     use crate::query::{PhraseQuery, QueryParser};
     use crate::schema::document::Value;
     use crate::schema::{
-        Document, IndexRecordOption, Schema, TextFieldIndexing, TextOptions, STORED, STRING, TEXT,
+        Document, IndexRecordOption, OwnedValue, Schema, TextFieldIndexing, TextOptions, STORED,
+        STRING, TEXT,
     };
     use crate::store::{Compressor, StoreReader, StoreWriter};
     use crate::time::format_description::well_known::Rfc3339;
     use crate::time::OffsetDateTime;
     use crate::tokenizer::{PreTokenizedString, Token};
     use crate::{
-        DateTime, Directory, DocAddress, DocSet, Index, IndexWriter, Postings, TantivyDocument,
-        Term, TERMINATED,
+        DateTime, Directory, DocAddress, DocSet, Index, IndexWriter, TantivyDocument, Term,
+        TERMINATED,
     };
 
     #[test]
@@ -597,6 +597,45 @@ mod tests {
         assert_eq!(score_docs.len(), 2);
     }
 
+    #[test]
+    fn test_flat_json_indexing() {
+        // A JSON Object that contains mixed values on the first level
+        let mut schema_builder = Schema::builder();
+        let json_field = schema_builder.add_json_field("json", STORED | STRING);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema.clone());
+        let mut writer = index.writer_for_tests().unwrap();
+        // Text, i64, u64
+        writer.add_document(doc!(json_field=>"b")).unwrap();
+        writer
+            .add_document(doc!(json_field=>OwnedValue::I64(10i64)))
+            .unwrap();
+        writer
+            .add_document(doc!(json_field=>OwnedValue::U64(55u64)))
+            .unwrap();
+        writer
+            .add_document(doc!(json_field=>json!({"my_field": "a"})))
+            .unwrap();
+        writer.commit().unwrap();
+
+        let search_and_expect = |query| {
+            let query_parser = QueryParser::for_index(&index, vec![json_field]);
+            let text_query = query_parser.parse_query(query).unwrap();
+            let score_docs: Vec<(_, DocAddress)> = index
+                .reader()
+                .unwrap()
+                .searcher()
+                .search(&text_query, &TopDocs::with_limit(4))
+                .unwrap();
+            assert_eq!(score_docs.len(), 1);
+        };
+
+        search_and_expect("my_field:a");
+        search_and_expect("b");
+        search_and_expect("10");
+        search_and_expect("55");
+    }
+
     #[test]
     fn test_json_indexing() {
         let mut schema_builder = Schema::builder();

src/lib.rs

@@ -216,11 +216,6 @@ use once_cell::sync::Lazy;
 use serde::{Deserialize, Serialize};
 
 pub use self::docset::{DocSet, COLLECT_BLOCK_BUFFER_LEN, TERMINATED};
-#[deprecated(
-    since = "0.22.0",
-    note = "Will be removed in tantivy 0.23. Use export from snippet module instead"
-)]
-pub use self::snippet::{Snippet, SnippetGenerator};
 #[doc(hidden)]
 pub use crate::core::json_utils;
 pub use crate::core::{Executor, Searcher, SearcherGeneration};
@@ -228,16 +223,10 @@ pub use crate::directory::Directory;
 #[allow(deprecated)] // Remove with index sorting
 pub use crate::index::{
     Index, IndexBuilder, IndexMeta, IndexSettings, IndexSortByField, InvertedIndexReader, Order,
-    Segment, SegmentComponent, SegmentId, SegmentMeta, SegmentReader,
+    Segment, SegmentMeta, SegmentReader,
 };
-#[deprecated(
-    since = "0.22.0",
-    note = "Will be removed in tantivy 0.23. Use export from indexer module instead"
-)]
-pub use crate::indexer::PreparedCommit;
 pub use crate::indexer::{IndexWriter, SingleSegmentIndexWriter};
-pub use crate::postings::Postings;
-pub use crate::schema::{DateOptions, DateTimePrecision, Document, TantivyDocument, Term};
+pub use crate::schema::{Document, TantivyDocument, Term};
 
 /// Index format version.
 const INDEX_FORMAT_VERSION: u32 = 6;
@@ -392,9 +381,10 @@ pub mod tests {
     use crate::docset::{DocSet, TERMINATED};
     use crate::index::SegmentReader;
     use crate::merge_policy::NoMergePolicy;
+    use crate::postings::Postings;
     use crate::query::BooleanQuery;
     use crate::schema::*;
-    use crate::{DateTime, DocAddress, Index, IndexWriter, Postings, ReloadPolicy};
+    use crate::{DateTime, DocAddress, Index, IndexWriter, ReloadPolicy};
 
     pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
         let mut buffer = Vec::new();
@@ -446,7 +436,6 @@ pub mod tests {
     }
 
     #[test]
-    #[cfg(not(feature = "lz4"))]
     fn test_version_string() {
         use regex::Regex;
         let regex_ptn = Regex::new(
@@ -1109,9 +1098,9 @@ pub mod tests {
     #[test]
     fn test_update_via_delete_insert() -> crate::Result<()> {
         use crate::collector::Count;
+        use crate::index::SegmentId;
         use crate::indexer::NoMergePolicy;
         use crate::query::AllQuery;
-        use crate::SegmentId;
 
         const DOC_COUNT: u64 = 2u64;
 

src/postings/serializer.rs

@@ -56,7 +56,7 @@ pub struct InvertedIndexSerializer {
 impl InvertedIndexSerializer {
     /// Open a new `InvertedIndexSerializer` for the given segment
     pub fn open(segment: &mut Segment) -> crate::Result<InvertedIndexSerializer> {
-        use crate::SegmentComponent::{Positions, Postings, Terms};
+        use crate::index::SegmentComponent::{Positions, Postings, Terms};
         let inv_index_serializer = InvertedIndexSerializer {
             terms_write: CompositeWrite::wrap(segment.open_write(Terms)?),
             postings_write: CompositeWrite::wrap(segment.open_write(Postings)?),

src/query/empty_query.rs

@@ -1,8 +1,9 @@
 use super::Scorer;
 use crate::docset::TERMINATED;
+use crate::index::SegmentReader;
 use crate::query::explanation::does_not_match;
 use crate::query::{EnableScoring, Explanation, Query, Weight};
-use crate::{DocId, DocSet, Score, Searcher, SegmentReader};
+use crate::{DocId, DocSet, Score, Searcher};
 
 /// `EmptyQuery` is a dummy `Query` in which no document matches.
 ///

src/query/more_like_this/more_like_this.rs

@@ -180,7 +180,7 @@ impl MoreLikeThis {
                 let facets: Vec<&str> = values
                     .iter()
                     .map(|value| {
-                        value.as_facet().map(|f| f.encoded_str()).ok_or_else(|| {
+                        value.as_facet().ok_or_else(|| {
                             TantivyError::InvalidArgument("invalid field value".to_string())
                         })
                     })
@@ -220,7 +220,7 @@ impl MoreLikeThis {
                         let mut token_stream = tokenizer.token_stream(text);
                         token_stream.process(sink);
                     } else if let Some(tok_str) = value.as_pre_tokenized_text() {
-                        let mut token_stream = PreTokenizedStream::from(tok_str.clone());
+                        let mut token_stream = PreTokenizedStream::from(*tok_str.clone());
                         token_stream.process(sink);
                     }
                 }

src/query/range_query/fast_field_range_query.rs

@@ -174,7 +174,7 @@ impl<T: Send + Sync + PartialOrd + Copy + Debug + 'static> DocSet for RangeDocSe
     }
 
     fn size_hint(&self) -> u32 {
-        0 // heuristic possible by checking number of hits when fetching a block
+        self.column.num_docs()
     }
 }
 

src/query/term_query/term_scorer.rs

@@ -127,6 +127,7 @@ impl Scorer for TermScorer {
 mod tests {
     use proptest::prelude::*;
 
+    use crate::index::SegmentId;
     use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
     use crate::merge_policy::NoMergePolicy;
     use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
@@ -134,8 +135,7 @@ mod tests {
     use crate::query::{Bm25Weight, EnableScoring, Scorer, TermQuery};
     use crate::schema::{IndexRecordOption, Schema, TEXT};
     use crate::{
-        assert_nearly_equals, DocId, DocSet, Index, IndexWriter, Score, Searcher, SegmentId, Term,
-        TERMINATED,
+        assert_nearly_equals, DocId, DocSet, Index, IndexWriter, Score, Searcher, Term, TERMINATED,
     };
 
     #[test]

src/reader/warming.rs

@@ -179,9 +179,10 @@ mod tests {
     use super::Warmer;
     use crate::core::searcher::SearcherGeneration;
     use crate::directory::RamDirectory;
+    use crate::index::SegmentId;
     use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
     use crate::schema::{Schema, INDEXED};
-    use crate::{Index, IndexSettings, ReloadPolicy, Searcher, SegmentId};
+    use crate::{Index, IndexSettings, ReloadPolicy, Searcher};
 
     #[derive(Default)]
     struct TestWarmer {

src/schema/document/de.rs

@@ -873,7 +873,7 @@ mod tests {
         );
 
         let facet = Facet::from_text("/hello/world").unwrap();
-        let result = serialize_value(ReferenceValueLeaf::Facet(&facet).into());
+        let result = serialize_value(ReferenceValueLeaf::Facet(facet.encoded_str()).into());
         let value = deserialize_value(result);
         assert_eq!(value, crate::schema::OwnedValue::Facet(facet));
 
@@ -881,7 +881,8 @@ mod tests {
             text: "hello, world".to_string(),
             tokens: vec![Token::default(), Token::default()],
         };
-        let result = serialize_value(ReferenceValueLeaf::PreTokStr(&pre_tok_str).into());
+        let result =
+            serialize_value(ReferenceValueLeaf::PreTokStr(pre_tok_str.clone().into()).into());
         let value = deserialize_value(result);
         assert_eq!(value, crate::schema::OwnedValue::PreTokStr(pre_tok_str));
     }

src/schema/document/owned_value.rs

@@ -65,13 +65,13 @@ impl<'a> Value<'a> for &'a OwnedValue {
         match self {
             OwnedValue::Null => ReferenceValueLeaf::Null.into(),
             OwnedValue::Str(val) => ReferenceValueLeaf::Str(val).into(),
-            OwnedValue::PreTokStr(val) => ReferenceValueLeaf::PreTokStr(val).into(),
+            OwnedValue::PreTokStr(val) => ReferenceValueLeaf::PreTokStr(val.clone().into()).into(),
             OwnedValue::U64(val) => ReferenceValueLeaf::U64(*val).into(),
             OwnedValue::I64(val) => ReferenceValueLeaf::I64(*val).into(),
             OwnedValue::F64(val) => ReferenceValueLeaf::F64(*val).into(),
             OwnedValue::Bool(val) => ReferenceValueLeaf::Bool(*val).into(),
             OwnedValue::Date(val) => ReferenceValueLeaf::Date(*val).into(),
-            OwnedValue::Facet(val) => ReferenceValueLeaf::Facet(val).into(),
+            OwnedValue::Facet(val) => ReferenceValueLeaf::Facet(val.encoded_str()).into(),
             OwnedValue::Bytes(val) => ReferenceValueLeaf::Bytes(val).into(),
             OwnedValue::IpAddr(val) => ReferenceValueLeaf::IpAddr(*val).into(),
             OwnedValue::Array(array) => ReferenceValue::Array(array.iter()),
@@ -183,7 +183,7 @@ impl serde::Serialize for OwnedValue {
             OwnedValue::Bytes(ref bytes) => serializer.serialize_str(&BASE64.encode(bytes)),
             OwnedValue::Object(ref obj) => {
                 let mut map = serializer.serialize_map(Some(obj.len()))?;
-                for &(ref k, ref v) in obj {
+                for (k, v) in obj {
                     map.serialize_entry(k, v)?;
                 }
                 map.end()
@@ -277,11 +277,13 @@ impl<'a, V: Value<'a>> From<ReferenceValue<'a, V>> for OwnedValue {
                 ReferenceValueLeaf::I64(val) => OwnedValue::I64(val),
                 ReferenceValueLeaf::F64(val) => OwnedValue::F64(val),
                 ReferenceValueLeaf::Date(val) => OwnedValue::Date(val),
-                ReferenceValueLeaf::Facet(val) => OwnedValue::Facet(val.clone()),
+                ReferenceValueLeaf::Facet(val) => {
+                    OwnedValue::Facet(Facet::from_encoded_string(val.to_string()))
+                }
                 ReferenceValueLeaf::Bytes(val) => OwnedValue::Bytes(val.to_vec()),
                 ReferenceValueLeaf::IpAddr(val) => OwnedValue::IpAddr(val),
                 ReferenceValueLeaf::Bool(val) => OwnedValue::Bool(val),
-                ReferenceValueLeaf::PreTokStr(val) => OwnedValue::PreTokStr(val.clone()),
+                ReferenceValueLeaf::PreTokStr(val) => OwnedValue::PreTokStr(*val.clone()),
             },
             ReferenceValue::Array(val) => {
                 OwnedValue::Array(val.map(|v| v.as_value().into()).collect())

src/schema/document/se.rs

@@ -121,7 +121,7 @@ where W: Write
                 ReferenceValueLeaf::Facet(val) => {
                     self.write_type_code(type_codes::HIERARCHICAL_FACET_CODE)?;
 
-                    val.serialize(self.writer)
+                    Cow::Borrowed(val).serialize(self.writer)
                 }
                 ReferenceValueLeaf::Bytes(val) => {
                     self.write_type_code(type_codes::BYTES_CODE)?;
@@ -428,7 +428,7 @@ mod tests {
         );
 
         let facet = Facet::from_text("/hello/world").unwrap();
-        let result = serialize_value(ReferenceValueLeaf::Facet(&facet).into());
+        let result = serialize_value(ReferenceValueLeaf::Facet(facet.encoded_str()).into());
         let expected = binary_repr!(
             type_codes::HIERARCHICAL_FACET_CODE => Facet::from_text("/hello/world").unwrap(),
         );
@@ -441,7 +441,8 @@ mod tests {
             text: "hello, world".to_string(),
             tokens: vec![Token::default(), Token::default()],
         };
-        let result = serialize_value(ReferenceValueLeaf::PreTokStr(&pre_tok_str).into());
+        let result =
+            serialize_value(ReferenceValueLeaf::PreTokStr(pre_tok_str.clone().into()).into());
         let expected = binary_repr!(
             type_codes::EXT_CODE, type_codes::TOK_STR_EXT_CODE => pre_tok_str,
         );

src/schema/document/value.rs

@@ -3,7 +3,6 @@ use std::net::Ipv6Addr;
 
 use common::DateTime;
 
-use crate::schema::Facet;
 use crate::tokenizer::PreTokenizedString;
 
 /// A single field value.
@@ -28,7 +27,7 @@ pub trait Value<'a>: Send + Sync + Debug {
     }
 
     #[inline]
-    /// If the Value is a String, returns the associated str. Returns None otherwise.
+    /// If the Value is a leaf, returns the associated leaf. Returns None otherwise.
     fn as_leaf(&self) -> Option<ReferenceValueLeaf<'a>> {
         if let ReferenceValue::Leaf(val) = self.as_value() {
             Some(val)
@@ -82,8 +81,9 @@ pub trait Value<'a>: Send + Sync + Debug {
     #[inline]
     /// If the Value is a pre-tokenized string, returns the associated string. Returns None
     /// otherwise.
-    fn as_pre_tokenized_text(&self) -> Option<&'a PreTokenizedString> {
-        self.as_leaf().and_then(|leaf| leaf.as_pre_tokenized_text())
+    fn as_pre_tokenized_text(&self) -> Option<Box<PreTokenizedString>> {
+        self.as_leaf()
+            .and_then(|leaf| leaf.into_pre_tokenized_text())
     }
 
     #[inline]
@@ -94,7 +94,7 @@ pub trait Value<'a>: Send + Sync + Debug {
 
     #[inline]
     /// If the Value is a facet, returns the associated facet. Returns None otherwise.
-    fn as_facet(&self) -> Option<&'a Facet> {
+    fn as_facet(&self) -> Option<&'a str> {
         self.as_leaf().and_then(|leaf| leaf.as_facet())
     }
 
@@ -132,7 +132,7 @@ pub trait Value<'a>: Send + Sync + Debug {
 }
 
 /// A enum representing a leaf value for tantivy to index.
-#[derive(Clone, Copy, Debug, PartialEq)]
+#[derive(Clone, Debug, PartialEq)]
 pub enum ReferenceValueLeaf<'a> {
     /// A null value.
     Null,
@@ -146,8 +146,9 @@ pub enum ReferenceValueLeaf<'a> {
     F64(f64),
     /// Date/time with nanoseconds precision
     Date(DateTime),
-    /// Facet
-    Facet(&'a Facet),
+    /// Facet string needs to match the format of
+    /// [Facet::encoded_str](crate::schema::Facet::encoded_str).
+    Facet(&'a str),
     /// Arbitrarily sized byte array
     Bytes(&'a [u8]),
     /// IpV6 Address. Internally there is no IpV4, it needs to be converted to `Ipv6Addr`.
@@ -155,7 +156,7 @@ pub enum ReferenceValueLeaf<'a> {
     /// Bool value
     Bool(bool),
     /// Pre-tokenized str type,
-    PreTokStr(&'a PreTokenizedString),
+    PreTokStr(Box<PreTokenizedString>),
 }
 
 impl<'a, T: Value<'a> + ?Sized> From<ReferenceValueLeaf<'a>> for ReferenceValue<'a, T> {
@@ -259,9 +260,9 @@ impl<'a> ReferenceValueLeaf<'a> {
     }
 
     #[inline]
-    /// If the Value is a pre-tokenized string, returns the associated string. Returns None
-    /// otherwise.
-    pub fn as_pre_tokenized_text(&self) -> Option<&'a PreTokenizedString> {
+    /// If the Value is a pre-tokenized string, consumes it and returns the string.
+    /// Returns None otherwise.
+    pub fn into_pre_tokenized_text(self) -> Option<Box<PreTokenizedString>> {
         if let Self::PreTokStr(val) = self {
             Some(val)
         } else {
@@ -281,7 +282,7 @@ impl<'a> ReferenceValueLeaf<'a> {
 
     #[inline]
     /// If the Value is a facet, returns the associated facet. Returns None otherwise.
-    pub fn as_facet(&self) -> Option<&'a Facet> {
+    pub fn as_facet(&self) -> Option<&'a str> {
         if let Self::Facet(val) = self {
             Some(val)
         } else {
@@ -322,6 +323,16 @@ where V: Value<'a>
         }
     }
 
+    #[inline]
+    /// If the Value is a leaf, consume it and return the leaf. Returns None otherwise.
+    pub fn into_leaf(self) -> Option<ReferenceValueLeaf<'a>> {
+        if let Self::Leaf(val) = self {
+            Some(val)
+        } else {
+            None
+        }
+    }
+
     #[inline]
     /// If the Value is a String, returns the associated str. Returns None otherwise.
     pub fn as_str(&self) -> Option<&'a str> {
@@ -365,10 +376,11 @@ where V: Value<'a>
     }
 
     #[inline]
-    /// If the Value is a pre-tokenized string, returns the associated string. Returns None
-    /// otherwise.
-    pub fn as_pre_tokenized_text(&self) -> Option<&'a PreTokenizedString> {
-        self.as_leaf().and_then(|leaf| leaf.as_pre_tokenized_text())
+    /// If the Value is a pre-tokenized string, consumes it and returns the string.
+    /// Returns None otherwise.
+    pub fn into_pre_tokenized_text(self) -> Option<Box<PreTokenizedString>> {
+        self.into_leaf()
+            .and_then(|leaf| leaf.into_pre_tokenized_text())
     }
 
     #[inline]
@@ -379,7 +391,7 @@ where V: Value<'a>
 
     #[inline]
     /// If the Value is a facet, returns the associated facet. Returns None otherwise.
-    pub fn as_facet(&self) -> Option<&'a Facet> {
+    pub fn as_facet(&self) -> Option<&'a str> {
         self.as_leaf().and_then(|leaf| leaf.as_facet())
     }
 

src/schema/flags.rs

@@ -1,7 +1,6 @@
 use std::ops::BitOr;
 
-use crate::schema::{NumericOptions, TextOptions};
-use crate::DateOptions;
+use crate::schema::{DateOptions, NumericOptions, TextOptions};
 
 #[derive(Clone)]
 pub struct StoredFlag;

src/space_usage/mod.rs

@@ -12,8 +12,8 @@ use std::collections::HashMap;
 use common::ByteCount;
 use serde::{Deserialize, Serialize};
 
+use crate::index::SegmentComponent;
 use crate::schema::Field;
-use crate::SegmentComponent;
 
 /// Enum containing any of the possible space usage results for segment components.
 pub enum ComponentSpaceUsage {
@@ -115,7 +115,7 @@ impl SegmentSpaceUsage {
     /// Use the components directly if this is somehow in performance critical code.
     pub fn component(&self, component: SegmentComponent) -> ComponentSpaceUsage {
         use self::ComponentSpaceUsage::*;
-        use crate::SegmentComponent::*;
+        use crate::index::SegmentComponent::*;
         match component {
             Postings => PerField(self.postings().clone()),
             Positions => PerField(self.positions().clone()),

src/store/reader.rs

@@ -18,6 +18,8 @@ use crate::schema::document::{BinaryDocumentDeserializer, DocumentDeserialize};
 use crate::space_usage::StoreSpaceUsage;
 use crate::store::index::Checkpoint;
 use crate::DocId;
+#[cfg(feature = "quickwit")]
+use crate::Executor;
 
 pub(crate) const DOCSTORE_CACHE_CAPACITY: usize = 100;
 
@@ -341,7 +343,11 @@ impl StoreReader {
     /// In most cases use [`get_async`](Self::get_async)
     ///
     /// Loads and decompresses a block asynchronously.
-    async fn read_block_async(&self, checkpoint: &Checkpoint) -> io::Result<Block> {
+    async fn read_block_async(
+        &self,
+        checkpoint: &Checkpoint,
+        executor: &Executor,
+    ) -> io::Result<Block> {
         let cache_key = checkpoint.byte_range.start;
         if let Some(block) = self.cache.get_from_cache(checkpoint.byte_range.start) {
             return Ok(block);
@@ -353,8 +359,12 @@ impl StoreReader {
             .read_bytes_async()
             .await?;
 
-        let decompressed_block =
-            OwnedBytes::new(self.decompressor.decompress(compressed_block.as_ref())?);
+        let decompressor = self.decompressor;
+        let maybe_decompressed_block = executor
+            .spawn_blocking(move || decompressor.decompress(compressed_block.as_ref()))
+            .await
+            .expect("decompression panicked");
+        let decompressed_block = OwnedBytes::new(maybe_decompressed_block?);
 
         self.cache
             .put_into_cache(cache_key, decompressed_block.clone());
@@ -363,15 +373,23 @@ impl StoreReader {
     }
 
     /// Reads raw bytes of a given document asynchronously.
-    pub async fn get_document_bytes_async(&self, doc_id: DocId) -> crate::Result<OwnedBytes> {
+    pub async fn get_document_bytes_async(
+        &self,
+        doc_id: DocId,
+        executor: &Executor,
+    ) -> crate::Result<OwnedBytes> {
         let checkpoint = self.block_checkpoint(doc_id)?;
-        let block = self.read_block_async(&checkpoint).await?;
+        let block = self.read_block_async(&checkpoint, executor).await?;
         Self::get_document_bytes_from_block(block, doc_id, &checkpoint)
     }
 
     /// Fetches a document asynchronously. Async version of [`get`](Self::get).
-    pub async fn get_async<D: DocumentDeserialize>(&self, doc_id: DocId) -> crate::Result<D> {
-        let mut doc_bytes = self.get_document_bytes_async(doc_id).await?;
+    pub async fn get_async<D: DocumentDeserialize>(
+        &self,
+        doc_id: DocId,
+        executor: &Executor,
+    ) -> crate::Result<D> {
+        let mut doc_bytes = self.get_document_bytes_async(doc_id, executor).await?;
 
         let deserializer = BinaryDocumentDeserializer::from_reader(&mut doc_bytes)
             .map_err(crate::TantivyError::from)?;