tantivy document memory test

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.23.0"
+version = "0.22.0"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -15,15 +15,12 @@ rust-version = "1.63"
 exclude = ["benches/*.json", "benches/*.txt"]
 
 [dependencies]
-oneshot = "0.1.7"
+oneshot = "0.1.5"
 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 }
@@ -33,15 +30,14 @@ 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"
@@ -52,26 +48,26 @@ smallvec = "1.8.0"
 rayon = "1.5.2"
 lru = "0.12.0"
 fastdivide = "0.4.0"
-itertools = "0.13.0"
+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" }
-sketches-ddsketch = { version = "0.3.0", features = ["use_serde"] }
+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 = "0.8.0"
 rand = "0.8.5"
 maplit = "1.0.2"
 matches = "0.1.9"
@@ -86,6 +82,7 @@ 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 }
@@ -117,26 +114,17 @@ 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
@@ -157,7 +145,3 @@ harness = false
 [[bench]]
 name = "index-bench"
 harness = false
-
-[[bench]]
-name = "agg_bench"
-harness = false

benches/agg_bench.rs

@@ -1,419 +0,0 @@
-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, extendedstats_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 extendedstats_f64(index: &Index) {
-    let agg_req = json!({
-        "extendedstats_f64": { "extended_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)
-}

benches/index-bench.rs

@@ -18,7 +18,7 @@ fn benchmark(
         benchmark_dynamic_json(b, input, schema, commit, parse_json)
     } else {
         _benchmark(b, input, schema, commit, parse_json, |schema, doc_json| {
-            TantivyDocument::parse_json(schema, doc_json).unwrap()
+            TantivyDocument::parse_json(&schema, doc_json).unwrap()
         })
     }
 }
@@ -90,7 +90,8 @@ fn benchmark_dynamic_json(
 ) {
     let json_field = schema.get_field("json").unwrap();
     _benchmark(b, input, schema, commit, parse_json, |_schema, doc_json| {
-        let json_val: serde_json::Value = serde_json::from_str(doc_json).unwrap();
+        let json_val: serde_json::Map<String, serde_json::Value> =
+            serde_json::from_str(doc_json).unwrap();
         tantivy::doc!(json_field=>json_val)
     })
 }
@@ -137,16 +138,15 @@ pub fn hdfs_index_benchmark(c: &mut Criterion) {
     for (prefix, schema, is_dynamic) in benches {
         for commit in [false, true] {
             let suffix = if commit { "with-commit" } else { "no-commit" };
-            {
-                let parse_json = false;
+            for parse_json in [false] {
                 // for parse_json in [false, true] {
                 let suffix = if parse_json {
-                    format!("{suffix}-with-json-parsing")
+                    format!("{}-with-json-parsing", suffix)
                 } else {
-                    suffix.to_string()
+                    format!("{}", suffix)
                 };
 
-                let bench_name = format!("{prefix}{suffix}");
+                let bench_name = format!("{}{}", prefix, suffix);
                 group.bench_function(bench_name, |b| {
                     benchmark(b, HDFS_LOGS, schema.clone(), commit, parse_json, is_dynamic)
                 });

columnar/Cargo.toml

@@ -9,7 +9,7 @@ description = "column oriented storage for tantivy"
 categories = ["database-implementations", "data-structures", "compression"]
 
 [dependencies]
-itertools = "0.13.0"
+itertools = "0.12.0"
 fastdivide = "0.4.0"
 
 stacker = { version= "0.3", path = "../stacker", package="tantivy-stacker"}
@@ -23,12 +23,6 @@ downcast-rs = "1.2.0"
 proptest = "1"
 more-asserts = "0.3.1"
 rand = "0.8"
-binggan = "0.8.1"
-
-[[bench]]
-name = "bench_merge"
-harness = false
-
 
 [features]
 unstable = []

columnar/benches/bench_merge.rs

@@ -1,97 +0,0 @@
-#![feature(test)]
-extern crate test;
-
-use core::fmt;
-use std::fmt::{Display, Formatter};
-
-use binggan::{black_box, BenchRunner};
-use tantivy_columnar::*;
-
-enum Card {
-    Multi,
-    Sparse,
-    Dense,
-}
-impl Display for Card {
-    fn fmt(&self, f: &mut Formatter) -> fmt::Result {
-        match self {
-            Card::Multi => write!(f, "multi"),
-            Card::Sparse => write!(f, "sparse"),
-            Card::Dense => write!(f, "dense"),
-        }
-    }
-}
-
-const NUM_DOCS: u32 = 1_000_000;
-
-fn generate_columnar(card: Card, num_docs: u32) -> ColumnarReader {
-    use tantivy_columnar::ColumnarWriter;
-
-    let mut columnar_writer = ColumnarWriter::default();
-
-    match card {
-        Card::Multi => {
-            columnar_writer.record_numerical(0, "price", 10u64);
-            columnar_writer.record_numerical(0, "price", 10u64);
-        }
-        _ => {}
-    }
-
-    for i in 0..num_docs {
-        match card {
-            Card::Multi | Card::Sparse => {
-                if i % 8 == 0 {
-                    columnar_writer.record_numerical(i, "price", i as u64);
-                }
-            }
-            Card::Dense => {
-                if i % 6 == 0 {
-                    columnar_writer.record_numerical(i, "price", i as u64);
-                }
-            }
-        }
-    }
-
-    let mut wrt: Vec<u8> = Vec::new();
-    columnar_writer.serialize(num_docs, None, &mut wrt).unwrap();
-
-    ColumnarReader::open(wrt).unwrap()
-}
-fn main() {
-    let mut inputs = Vec::new();
-
-    let mut add_combo = |card1: Card, card2: Card| {
-        inputs.push((
-            format!("merge_{card1}_and_{card2}"),
-            vec![
-                generate_columnar(card1, NUM_DOCS),
-                generate_columnar(card2, NUM_DOCS),
-            ],
-        ));
-    };
-
-    add_combo(Card::Multi, Card::Multi);
-    add_combo(Card::Dense, Card::Dense);
-    add_combo(Card::Sparse, Card::Sparse);
-    add_combo(Card::Sparse, Card::Dense);
-    add_combo(Card::Multi, Card::Dense);
-    add_combo(Card::Multi, Card::Sparse);
-
-    let runner: BenchRunner = BenchRunner::new();
-    let mut group = runner.new_group();
-    for (input_name, columnar_readers) in inputs.iter() {
-        group.register_with_input(
-            input_name,
-            columnar_readers,
-            move |columnar_readers: &Vec<ColumnarReader>| {
-                let mut out = vec![];
-                let columnar_readers = columnar_readers.iter().collect::<Vec<_>>();
-                let merge_row_order = StackMergeOrder::stack(&columnar_readers[..]);
-
-                merge_columnar(&columnar_readers, &[], merge_row_order.into(), &mut out).unwrap();
-                black_box(out);
-            },
-        );
-    }
-    group.run();
-}

columnar/src/column_index/merge/mod.rs

@@ -73,18 +73,14 @@ fn detect_cardinality(
 pub fn merge_column_index<'a>(
     columns: &'a [ColumnIndex],
     merge_row_order: &'a MergeRowOrder,
-    num_values: u32,
 ) -> SerializableColumnIndex<'a> {
     // For simplification, we do not try to detect whether the cardinality could be
     // downgraded thanks to deletes.
     let cardinality_after_merge = detect_cardinality(columns, merge_row_order);
     match merge_row_order {
-        MergeRowOrder::Stack(stack_merge_order) => merge_column_index_stacked(
-            columns,
-            cardinality_after_merge,
-            stack_merge_order,
-            num_values,
-        ),
+        MergeRowOrder::Stack(stack_merge_order) => {
+            merge_column_index_stacked(columns, cardinality_after_merge, stack_merge_order)
+        }
         MergeRowOrder::Shuffled(complex_merge_order) => {
             merge_column_index_shuffled(columns, cardinality_after_merge, complex_merge_order)
         }
@@ -171,12 +167,8 @@ mod tests {
             ],
         )
         .into();
-        let merged_column_index = merge_column_index(&column_indexes[..], &merge_row_order, 3);
-        let SerializableColumnIndex::Multivalued {
-            indices: start_index_iterable,
-            ..
-        } = merged_column_index
-        else {
+        let merged_column_index = merge_column_index(&column_indexes[..], &merge_row_order);
+        let SerializableColumnIndex::Multivalued(start_index_iterable) = merged_column_index else {
             panic!("Excpected a multivalued index")
         };
         let start_indexes: Vec<RowId> = start_index_iterable.boxed_iter().collect();
@@ -208,12 +200,8 @@ mod tests {
             ],
         )
         .into();
-        let merged_column_index = merge_column_index(&column_indexes[..], &merge_row_order, 6);
-        let SerializableColumnIndex::Multivalued {
-            indices: start_index_iterable,
-            ..
-        } = merged_column_index
-        else {
+        let merged_column_index = merge_column_index(&column_indexes[..], &merge_row_order);
+        let SerializableColumnIndex::Multivalued(start_index_iterable) = merged_column_index else {
             panic!("Excpected a multivalued index")
         };
         let start_indexes: Vec<RowId> = start_index_iterable.boxed_iter().collect();

columnar/src/column_index/merge/shuffled.rs

@@ -22,10 +22,7 @@ pub fn merge_column_index_shuffled<'a>(
         Cardinality::Multivalued => {
             let multivalue_start_index =
                 merge_column_index_shuffled_multivalued(column_indexes, shuffle_merge_order);
-            SerializableColumnIndex::Multivalued {
-                indices: multivalue_start_index,
-                stats: None,
-            }
+            SerializableColumnIndex::Multivalued(multivalue_start_index)
         }
     }
 }

columnar/src/column_index/merge/stacked.rs

@@ -1,8 +1,6 @@
 use std::iter;
-use std::num::NonZeroU64;
 
 use crate::column_index::{SerializableColumnIndex, Set};
-use crate::column_values::ColumnStats;
 use crate::iterable::Iterable;
 use crate::{Cardinality, ColumnIndex, RowId, StackMergeOrder};
 
@@ -14,7 +12,6 @@ pub fn merge_column_index_stacked<'a>(
     columns: &'a [ColumnIndex],
     cardinality_after_merge: Cardinality,
     stack_merge_order: &'a StackMergeOrder,
-    num_values: u32,
 ) -> SerializableColumnIndex<'a> {
     match cardinality_after_merge {
         Cardinality::Full => SerializableColumnIndex::Full,
@@ -30,17 +27,7 @@ pub fn merge_column_index_stacked<'a>(
                 columns,
                 stack_merge_order,
             };
-            SerializableColumnIndex::Multivalued {
-                indices: Box::new(stacked_multivalued_index),
-                stats: Some(ColumnStats {
-                    gcd: NonZeroU64::new(1).unwrap(),
-                    // The values in the multivalue index are the positions of the values
-                    min_value: 0,
-                    max_value: num_values as u64,
-                    // This is num docs, but it starts at 0 so we need +1
-                    num_rows: stack_merge_order.num_rows() + 1,
-                }),
-            }
+            SerializableColumnIndex::Multivalued(Box::new(stacked_multivalued_index))
         }
     }
 }

columnar/src/column_index/multivalued_index.rs

@@ -6,29 +6,20 @@ use std::sync::Arc;
 use common::OwnedBytes;
 
 use crate::column_values::{
-    load_u64_based_column_values, serialize_u64_based_column_values,
-    serialize_u64_with_codec_and_stats, CodecType, ColumnStats, ColumnValues,
+    load_u64_based_column_values, serialize_u64_based_column_values, CodecType, ColumnValues,
 };
 use crate::iterable::Iterable;
 use crate::{DocId, RowId};
 
 pub fn serialize_multivalued_index(
     multivalued_index: &dyn Iterable<RowId>,
-    stats: Option<ColumnStats>,
     output: &mut impl Write,
 ) -> io::Result<()> {
-    if let Some(stats) = stats {
-        // TODO: Add something with higher compression that doesn't require a full scan upfront
-        let estimator = CodecType::Bitpacked.estimator();
-        assert!(!estimator.requires_full_scan());
-        serialize_u64_with_codec_and_stats(multivalued_index, estimator, stats, output)?;
-    } else {
-        serialize_u64_based_column_values(
-            multivalued_index,
-            &[CodecType::Bitpacked, CodecType::Linear],
-            output,
-        )?;
-    }
+    serialize_u64_based_column_values(
+        multivalued_index,
+        &[CodecType::Bitpacked, CodecType::Linear],
+        output,
+    )?;
     Ok(())
 }
 
@@ -61,7 +52,7 @@ impl From<Arc<dyn ColumnValues<RowId>>> for MultiValueIndex {
 impl MultiValueIndex {
     pub fn for_test(start_offsets: &[RowId]) -> MultiValueIndex {
         let mut buffer = Vec::new();
-        serialize_multivalued_index(&start_offsets, None, &mut buffer).unwrap();
+        serialize_multivalued_index(&start_offsets, &mut buffer).unwrap();
         let bytes = OwnedBytes::new(buffer);
         open_multivalued_index(bytes).unwrap()
     }

columnar/src/column_index/optional_index/mod.rs

@@ -196,7 +196,6 @@ impl Set<RowId> for OptionalIndex {
         } = row_addr_from_row_id(doc_id);
         let block_meta = self.block_metas[block_id as usize];
         let block = self.block(block_meta);
-
         let block_offset_row_id = match block {
             Block::Dense(dense_block) => dense_block.rank(in_block_row_id),
             Block::Sparse(sparse_block) => sparse_block.rank(in_block_row_id),

columnar/src/column_index/serialize.rs

@@ -6,7 +6,6 @@ use common::{CountingWriter, OwnedBytes};
 use crate::column_index::multivalued_index::serialize_multivalued_index;
 use crate::column_index::optional_index::serialize_optional_index;
 use crate::column_index::ColumnIndex;
-use crate::column_values::ColumnStats;
 use crate::iterable::Iterable;
 use crate::{Cardinality, RowId};
 
@@ -16,12 +15,9 @@ pub enum SerializableColumnIndex<'a> {
         non_null_row_ids: Box<dyn Iterable<RowId> + 'a>,
         num_rows: RowId,
     },
-    Multivalued {
-        /// Iterator emitting the indices for the index
-        indices: Box<dyn Iterable<RowId> + 'a>,
-        /// In the merge case we can precompute the column stats
-        stats: Option<ColumnStats>,
-    },
+    // TODO remove the Arc<dyn> apart from serialization this is not
+    // dynamic at all.
+    Multivalued(Box<dyn Iterable<RowId> + 'a>),
 }
 
 impl<'a> SerializableColumnIndex<'a> {
@@ -29,7 +25,7 @@ impl<'a> SerializableColumnIndex<'a> {
         match self {
             SerializableColumnIndex::Full => Cardinality::Full,
             SerializableColumnIndex::Optional { .. } => Cardinality::Optional,
-            SerializableColumnIndex::Multivalued { .. } => Cardinality::Multivalued,
+            SerializableColumnIndex::Multivalued(_) => Cardinality::Multivalued,
         }
     }
 }
@@ -48,10 +44,9 @@ pub fn serialize_column_index(
             non_null_row_ids,
             num_rows,
         } => serialize_optional_index(non_null_row_ids.as_ref(), num_rows, &mut output)?,
-        SerializableColumnIndex::Multivalued {
-            indices: multivalued_index,
-            stats,
-        } => serialize_multivalued_index(&*multivalued_index, stats, &mut output)?,
+        SerializableColumnIndex::Multivalued(multivalued_index) => {
+            serialize_multivalued_index(&*multivalued_index, &mut output)?
+        }
     }
     let column_index_num_bytes = output.written_bytes() as u32;
     Ok(column_index_num_bytes)

columnar/src/column_values/mod.rs

@@ -32,8 +32,7 @@ pub use u128_based::{
 };
 pub use u64_based::{
     load_u64_based_column_values, serialize_and_load_u64_based_column_values,
-    serialize_u64_based_column_values, serialize_u64_with_codec_and_stats, CodecType,
-    ALL_U64_CODEC_TYPES,
+    serialize_u64_based_column_values, CodecType, ALL_U64_CODEC_TYPES,
 };
 pub use vec_column::VecColumn;
 

columnar/src/column_values/u64_based/bitpacked.rs

@@ -128,9 +128,6 @@ impl ColumnCodecEstimator for BitpackedCodecEstimator {
         bit_packer.close(wrt)?;
         Ok(())
     }
-    fn codec_type(&self) -> super::CodecType {
-        super::CodecType::Bitpacked
-    }
 }
 
 pub struct BitpackedCodec;

columnar/src/column_values/u64_based/blockwise_linear.rs

@@ -163,10 +163,6 @@ impl ColumnCodecEstimator for BlockwiseLinearEstimator {
 
         Ok(())
     }
-
-    fn codec_type(&self) -> super::CodecType {
-        super::CodecType::BlockwiseLinear
-    }
 }
 
 pub struct BlockwiseLinearCodec;

columnar/src/column_values/u64_based/linear.rs

@@ -153,12 +153,6 @@ impl ColumnCodecEstimator for LinearCodecEstimator {
             self.collect_before_line_estimation(value);
         }
     }
-    fn requires_full_scan(&self) -> bool {
-        true
-    }
-    fn codec_type(&self) -> super::CodecType {
-        super::CodecType::Linear
-    }
 }
 
 impl LinearCodecEstimator {

columnar/src/column_values/u64_based/mod.rs

@@ -37,11 +37,7 @@ pub trait ColumnCodecEstimator<T = u64>: 'static {
     /// This method will be called for each element of the column during
     /// `estimation`.
     fn collect(&mut self, value: u64);
-    /// Returns true if the estimator needs a full pass over the column before serialization
-    fn requires_full_scan(&self) -> bool {
-        false
-    }
-    fn codec_type(&self) -> CodecType;
+    /// Finalizes the first pass phase.
     fn finalize(&mut self) {}
     /// Returns an accurate estimation of the number of bytes that will
     /// be used to represent this column.
@@ -154,45 +150,34 @@ pub fn serialize_u64_based_column_values<T: MonotonicallyMappableToU64>(
     wrt: &mut dyn Write,
 ) -> io::Result<()> {
     let mut stats_collector = StatsCollector::default();
-    let mut estimators: Vec<Box<dyn ColumnCodecEstimator>> = Vec::with_capacity(codec_types.len());
+    let mut estimators: Vec<(CodecType, Box<dyn ColumnCodecEstimator>)> =
+        Vec::with_capacity(codec_types.len());
     for &codec_type in codec_types {
-        estimators.push(codec_type.estimator());
+        estimators.push((codec_type, codec_type.estimator()));
     }
     for val in vals.boxed_iter() {
         let val_u64 = val.to_u64();
         stats_collector.collect(val_u64);
-        for estimator in &mut estimators {
+        for (_, estimator) in &mut estimators {
             estimator.collect(val_u64);
         }
     }
-    for estimator in &mut estimators {
+    for (_, estimator) in &mut estimators {
         estimator.finalize();
     }
     let stats = stats_collector.stats();
-    let (_, best_codec) = estimators
+    let (_, best_codec, best_codec_estimator) = estimators
         .into_iter()
-        .flat_map(|estimator| {
+        .flat_map(|(codec_type, estimator)| {
             let num_bytes = estimator.estimate(&stats)?;
-            Some((num_bytes, estimator))
+            Some((num_bytes, codec_type, estimator))
         })
-        .min_by_key(|(num_bytes, _)| *num_bytes)
+        .min_by_key(|(num_bytes, _, _)| *num_bytes)
         .ok_or_else(|| {
             io::Error::new(io::ErrorKind::InvalidData, "No available applicable codec.")
         })?;
-    serialize_u64_with_codec_and_stats(vals, best_codec, stats, wrt)?;
-    Ok(())
-}
-
-/// Serializes a given column of u64-mapped values.
-/// The codec estimator needs to be collected fully for the Line codec before calling this.
-pub fn serialize_u64_with_codec_and_stats<T: MonotonicallyMappableToU64>(
-    vals: &dyn Iterable<T>,
-    codec: Box<dyn ColumnCodecEstimator>,
-    stats: ColumnStats,
-    wrt: &mut dyn Write,
-) -> io::Result<()> {
-    codec.codec_type().to_code().serialize(wrt)?;
-    codec.serialize(
+    best_codec.to_code().serialize(wrt)?;
+    best_codec_estimator.serialize(
         &stats,
         &mut vals.boxed_iter().map(MonotonicallyMappableToU64::to_u64),
         wrt,

columnar/src/columnar/merge/mod.rs

@@ -3,7 +3,7 @@ mod merge_mapping;
 mod term_merger;
 
 use std::collections::{BTreeMap, HashSet};
-use std::io::{self};
+use std::io;
 use std::net::Ipv6Addr;
 use std::sync::Arc;
 
@@ -156,15 +156,8 @@ fn merge_column(
                     column_values.push(None);
                 }
             }
-            let num_values: u32 = column_values
-                .iter()
-                .map(|vals| vals.as_ref().map(|idx| idx.num_vals()).unwrap_or(0))
-                .sum();
-            let merged_column_index = crate::column_index::merge_column_index(
-                &column_indexes[..],
-                merge_row_order,
-                num_values,
-            );
+            let merged_column_index =
+                crate::column_index::merge_column_index(&column_indexes[..], merge_row_order);
             let merge_column_values = MergedColumnValues {
                 column_indexes: &column_indexes[..],
                 column_values: &column_values[..],
@@ -190,15 +183,8 @@ fn merge_column(
                 }
             }
 
-            let num_values: u32 = column_values
-                .iter()
-                .map(|vals| vals.as_ref().map(|idx| idx.num_vals()).unwrap_or(0))
-                .sum();
-            let merged_column_index = crate::column_index::merge_column_index(
-                &column_indexes[..],
-                merge_row_order,
-                num_values,
-            );
+            let merged_column_index =
+                crate::column_index::merge_column_index(&column_indexes[..], merge_row_order);
             let merge_column_values = MergedColumnValues {
                 column_indexes: &column_indexes[..],
                 column_values: &column_values,
@@ -228,19 +214,8 @@ fn merge_column(
                     }
                 }
             }
-            let num_values: u32 = bytes_columns
-                .iter()
-                .map(|vals| {
-                    vals.as_ref()
-                        .map(|idx| idx.term_ord_column.values.num_vals())
-                        .unwrap_or(0)
-                })
-                .sum();
-            let merged_column_index = crate::column_index::merge_column_index(
-                &column_indexes[..],
-                merge_row_order,
-                num_values,
-            );
+            let merged_column_index =
+                crate::column_index::merge_column_index(&column_indexes[..], merge_row_order);
             merge_bytes_or_str_column(merged_column_index, &bytes_columns, merge_row_order, wrt)?;
         }
     }

columnar/src/columnar/writer/mod.rs

@@ -59,6 +59,22 @@ 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()
@@ -159,8 +175,9 @@ impl ColumnarWriter {
                     },
                     &mut self.dictionaries,
                 );
-                hash_map.mutate_or_create(
-                    column_name.as_bytes(),
+                mutate_or_create_column(
+                    hash_map,
+                    column_name,
                     |column_opt: Option<StrOrBytesColumnWriter>| {
                         let mut column_writer = if let Some(column_writer) = column_opt {
                             column_writer
@@ -175,21 +192,24 @@ impl ColumnarWriter {
                 );
             }
             ColumnType::Bool => {
-                self.bool_field_hash_map.mutate_or_create(
-                    column_name.as_bytes(),
+                mutate_or_create_column(
+                    &mut self.bool_field_hash_map,
+                    column_name,
                     |column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
                 );
             }
             ColumnType::DateTime => {
-                self.datetime_field_hash_map.mutate_or_create(
-                    column_name.as_bytes(),
+                mutate_or_create_column(
+                    &mut self.datetime_field_hash_map,
+                    column_name,
                     |column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
                 );
             }
             ColumnType::I64 | ColumnType::F64 | ColumnType::U64 => {
                 let numerical_type = column_type.numerical_type().unwrap();
-                self.numerical_field_hash_map.mutate_or_create(
-                    column_name.as_bytes(),
+                mutate_or_create_column(
+                    &mut self.numerical_field_hash_map,
+                    column_name,
                     |column_opt: Option<NumericalColumnWriter>| {
                         let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
                         column.force_numerical_type(numerical_type);
@@ -197,8 +217,9 @@ impl ColumnarWriter {
                     },
                 );
             }
-            ColumnType::IpAddr => self.ip_addr_field_hash_map.mutate_or_create(
-                column_name.as_bytes(),
+            ColumnType::IpAddr => mutate_or_create_column(
+                &mut self.ip_addr_field_hash_map,
+                column_name,
                 |column_opt: Option<ColumnWriter>| column_opt.unwrap_or_default(),
             ),
         }
@@ -211,8 +232,9 @@ impl ColumnarWriter {
         numerical_value: T,
     ) {
         let (hash_map, arena) = (&mut self.numerical_field_hash_map, &mut self.arena);
-        hash_map.mutate_or_create(
-            column_name.as_bytes(),
+        mutate_or_create_column(
+            hash_map,
+            column_name,
             |column_opt: Option<NumericalColumnWriter>| {
                 let mut column: NumericalColumnWriter = column_opt.unwrap_or_default();
                 column.record_numerical_value(doc, numerical_value.into(), arena);
@@ -222,6 +244,10 @@ 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(),
@@ -235,30 +261,24 @@ 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);
-        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
-            },
-        );
+        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
+        });
     }
 
     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);
-        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
-            },
-        );
+        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
+        });
     }
 
     pub fn record_str(&mut self, doc: RowId, column_name: &str, value: &str) {
@@ -283,6 +303,10 @@ 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,
@@ -644,10 +668,7 @@ fn send_to_serialize_column_mappable_to_u128<
             let multivalued_index_builder = value_index_builders.borrow_multivalued_index_builder();
             consume_operation_iterator(op_iterator, multivalued_index_builder, values);
             let multivalued_index = multivalued_index_builder.finish(num_rows);
-            SerializableColumnIndex::Multivalued {
-                indices: Box::new(multivalued_index),
-                stats: Default::default(), // TODO: implement stats for u128
-            }
+            SerializableColumnIndex::Multivalued(Box::new(multivalued_index))
         }
     };
     crate::column::serialize_column_mappable_to_u128(
@@ -702,10 +723,7 @@ fn send_to_serialize_column_mappable_to_u64(
             if sort_values_within_row {
                 sort_values_within_row_in_place(multivalued_index, values);
             }
-            SerializableColumnIndex::Multivalued {
-                indices: Box::new(multivalued_index),
-                stats: None,
-            }
+            SerializableColumnIndex::Multivalued(Box::new(multivalued_index))
         }
     };
     crate::column::serialize_column_mappable_to_u64(

columnar/src/tests.rs

@@ -738,22 +738,35 @@ proptest! {
     #![proptest_config(ProptestConfig::with_cases(1000))]
     #[test]
     fn test_columnar_merge_proptest(columnar_docs in proptest::collection::vec(columnar_docs_strategy(), 2..=3)) {
-        test_columnar_docs(columnar_docs);
+        let columnar_readers: Vec<ColumnarReader> = columnar_docs.iter()
+            .map(|docs| build_columnar(&docs[..]))
+            .collect::<Vec<_>>();
+        let columnar_readers_arr: Vec<&ColumnarReader> = columnar_readers.iter().collect();
+        let mut output: Vec<u8> = Vec::new();
+        let stack_merge_order = StackMergeOrder::stack(&columnar_readers_arr[..]).into();
+        crate::merge_columnar(&columnar_readers_arr[..], &[], stack_merge_order, &mut output).unwrap();
+        let merged_columnar = ColumnarReader::open(output).unwrap();
+        let concat_rows: Vec<Vec<(&'static str, ColumnValue)>> = columnar_docs.iter().flatten().cloned().collect();
+        let expected_merged_columnar = build_columnar(&concat_rows[..]);
+        assert_columnar_eq_strict(&merged_columnar, &expected_merged_columnar);
     }
 }
 
-fn test_columnar_docs(columnar_docs: Vec<Vec<Vec<(&'static str, ColumnValue)>>>) {
+#[test]
+fn test_columnar_merging_empty_columnar() {
+    let columnar_docs: Vec<Vec<Vec<(&str, ColumnValue)>>> =
+        vec![vec![], vec![vec![("c1", ColumnValue::Str("a"))]]];
     let columnar_readers: Vec<ColumnarReader> = columnar_docs
         .iter()
         .map(|docs| build_columnar(&docs[..]))
         .collect::<Vec<_>>();
     let columnar_readers_arr: Vec<&ColumnarReader> = columnar_readers.iter().collect();
     let mut output: Vec<u8> = Vec::new();
-    let stack_merge_order = StackMergeOrder::stack(&columnar_readers_arr[..]).into();
+    let stack_merge_order = StackMergeOrder::stack(&columnar_readers_arr[..]);
     crate::merge_columnar(
         &columnar_readers_arr[..],
         &[],
-        stack_merge_order,
+        crate::MergeRowOrder::Stack(stack_merge_order),
         &mut output,
     )
     .unwrap();
@@ -764,24 +777,6 @@ fn test_columnar_docs(columnar_docs: Vec<Vec<Vec<(&'static str, ColumnValue)>>>)
     assert_columnar_eq_strict(&merged_columnar, &expected_merged_columnar);
 }
 
-#[test]
-fn test_columnar_merging_empty_columnar() {
-    let columnar_docs: Vec<Vec<Vec<(&str, ColumnValue)>>> =
-        vec![vec![], vec![vec![("c1", ColumnValue::Str("a"))]]];
-    test_columnar_docs(columnar_docs);
-}
-#[test]
-fn test_columnar_merging_simple() {
-    let columnar_docs: Vec<Vec<Vec<(&str, ColumnValue)>>> = vec![
-        vec![],
-        vec![vec![
-            ("c1", ColumnValue::Numerical(0u64.into())),
-            ("c1", ColumnValue::Numerical(0u64.into())),
-        ]],
-    ];
-    test_columnar_docs(columnar_docs);
-}
-
 #[test]
 fn test_columnar_merging_number_columns() {
     let columnar_docs: Vec<Vec<Vec<(&str, ColumnValue)>>> = vec![
@@ -798,7 +793,25 @@ fn test_columnar_merging_number_columns() {
             vec![("c2", ColumnValue::Numerical(u64::MAX.into()))],
         ],
     ];
-    test_columnar_docs(columnar_docs);
+    let columnar_readers: Vec<ColumnarReader> = columnar_docs
+        .iter()
+        .map(|docs| build_columnar(&docs[..]))
+        .collect::<Vec<_>>();
+    let columnar_readers_arr: Vec<&ColumnarReader> = columnar_readers.iter().collect();
+    let mut output: Vec<u8> = Vec::new();
+    let stack_merge_order = StackMergeOrder::stack(&columnar_readers_arr[..]);
+    crate::merge_columnar(
+        &columnar_readers_arr[..],
+        &[],
+        crate::MergeRowOrder::Stack(stack_merge_order),
+        &mut output,
+    )
+    .unwrap();
+    let merged_columnar = ColumnarReader::open(output).unwrap();
+    let concat_rows: Vec<Vec<(&'static str, ColumnValue)>> =
+        columnar_docs.iter().flatten().cloned().collect();
+    let expected_merged_columnar = build_columnar(&concat_rows[..]);
+    assert_columnar_eq_strict(&merged_columnar, &expected_merged_columnar);
 }
 
 // TODO add non trivial remap and merge

common/Cargo.toml

@@ -22,6 +22,3 @@ serde = { version = "1.0.136", features = ["derive"] }
 [dev-dependencies]
 proptest = "1.0.0"
 rand = "0.8.4"
-
-[features]
-unstable = [] # useful for benches.

common/src/vint.rs

@@ -151,7 +151,7 @@ pub fn read_u32_vint_no_advance(data: &[u8]) -> (u32, usize) {
     (result, vlen)
 }
 /// Write a `u32` as a vint payload.
-pub fn write_u32_vint<W: io::Write + ?Sized>(val: u32, writer: &mut W) -> io::Result<()> {
+pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
     let mut buf = [0u8; 8];
     let data = serialize_vint_u32(val, &mut buf);
     writer.write_all(data)

examples/custom_collector.rs

@@ -11,10 +11,9 @@ 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};
+use tantivy::{doc, Index, IndexWriter, Score, SegmentReader};
 
 #[derive(Default)]
 struct Stats {

examples/date_time_field.rs

@@ -4,7 +4,7 @@
 
 use tantivy::collector::TopDocs;
 use tantivy::query::QueryParser;
-use tantivy::schema::{DateOptions, Document, Schema, Value, INDEXED, STORED, STRING};
+use tantivy::schema::{DateOptions, Document, OwnedValue, Schema, INDEXED, STORED, STRING};
 use tantivy::{Index, IndexWriter, TantivyDocument};
 
 fn main() -> tantivy::Result<()> {
@@ -13,7 +13,7 @@ fn main() -> tantivy::Result<()> {
     let opts = DateOptions::from(INDEXED)
         .set_stored()
         .set_fast()
-        .set_precision(tantivy::schema::DateTimePrecision::Seconds);
+        .set_precision(tantivy::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);
@@ -61,12 +61,10 @@ fn main() -> tantivy::Result<()> {
         assert_eq!(count_docs.len(), 1);
         for (_score, doc_address) in count_docs {
             let retrieved_doc = searcher.doc::<TantivyDocument>(doc_address)?;
-            assert!(retrieved_doc
-                .get_first(occurred_at)
-                .unwrap()
-                .as_value()
-                .as_datetime()
-                .is_some(),);
+            assert!(matches!(
+                retrieved_doc.get_first(occurred_at),
+                Some(OwnedValue::Date(_))
+            ));
             assert_eq!(
                 retrieved_doc.to_json(&schema),
                 r#"{"event":["comment"],"occurred_at":["2022-06-22T13:00:00.22Z"]}"#

examples/doc_mem.rs

@@ -0,0 +1,335 @@
+#![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/faceted_search_with_tweaked_score.rs

@@ -51,7 +51,7 @@ fn main() -> tantivy::Result<()> {
     let reader = index.reader()?;
     let searcher = reader.searcher();
     {
-        let facets = [
+        let facets = vec![
             Facet::from("/ingredient/egg"),
             Facet::from("/ingredient/oil"),
             Facet::from("/ingredient/garlic"),
@@ -94,8 +94,9 @@ fn main() -> tantivy::Result<()> {
                     .doc::<TantivyDocument>(*doc_id)
                     .unwrap()
                     .get_first(title)
-                    .and_then(|v| v.as_str().map(|el| el.to_string()))
+                    .and_then(|v| v.as_str())
                     .unwrap()
+                    .to_owned()
             })
             .collect();
         assert_eq!(titles, vec!["Fried egg", "Egg rolls"]);

examples/index_from_multiple_threads.rs

@@ -61,7 +61,7 @@ fn main() -> tantivy::Result<()> {
                         debris of the winter’s flooding; and sycamores with mottled, white, recumbent \
                         limbs and branches that arch over the pool"
                     ))?;
-            println!("add doc {i} from thread 1 - opstamp {opstamp}");
+            println!("add doc {} from thread 1 - opstamp {}", i, opstamp);
             thread::sleep(Duration::from_millis(20));
         }
         Result::<(), TantivyError>::Ok(())
@@ -82,7 +82,7 @@ fn main() -> tantivy::Result<()> {
                     body => "Some great book description..."
                 ))?
             };
-            println!("add doc {i} from thread 2 - opstamp {opstamp}");
+            println!("add doc {} from thread 2 - opstamp {}", i, opstamp);
             thread::sleep(Duration::from_millis(10));
         }
         Result::<(), TantivyError>::Ok(())

examples/iterating_docs_and_positions.rs

@@ -7,11 +7,10 @@
 // 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, TERMINATED};
+use tantivy::{doc, DocSet, Index, IndexWriter, Postings, TERMINATED};
 
 fn main() -> tantivy::Result<()> {
     // We first create a schema for the sake of the

examples/warmer.rs

@@ -3,11 +3,10 @@ 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,
+    doc, DocAddress, DocId, Index, IndexWriter, Opstamp, Searcher, SearcherGeneration, SegmentId,
     SegmentReader, Warmer,
 };
 

query-grammar/src/query_grammar.rs

@@ -1,4 +1,3 @@
-use std::borrow::Cow;
 use std::iter::once;
 
 use nom::branch::alt;
@@ -20,7 +19,7 @@ use crate::Occur;
 // Note: '-' char is only forbidden at the beginning of a field name, would be clearer to add it to
 // special characters.
 const SPECIAL_CHARS: &[char] = &[
-    '+', '^', '`', ':', '{', '}', '"', '\'', '[', ']', '(', ')', '!', '\\', '*', ' ',
+    '+', '^', '`', ':', '{', '}', '"', '[', ']', '(', ')', '!', '\\', '*', ' ',
 ];
 
 /// consume a field name followed by colon. Return the field name with escape sequence
@@ -42,92 +41,36 @@ fn field_name(inp: &str) -> IResult<&str, String> {
     )(inp)
 }
 
-const ESCAPE_IN_WORD: &[char] = &['^', '`', ':', '{', '}', '"', '\'', '[', ']', '(', ')', '\\'];
-
-fn interpret_escape(source: &str) -> String {
-    let mut res = String::with_capacity(source.len());
-    let mut in_escape = false;
-    let require_escape = |c: char| c.is_whitespace() || ESCAPE_IN_WORD.contains(&c) || c == '-';
-
-    for c in source.chars() {
-        if in_escape {
-            if !require_escape(c) {
-                // we re-add the escape sequence
-                res.push('\\');
-            }
-            res.push(c);
-            in_escape = false;
-        } else if c == '\\' {
-            in_escape = true;
-        } else {
-            res.push(c);
-        }
-    }
-    res
-}
-
 /// Consume a word outside of any context.
 // TODO should support escape sequences
-fn word(inp: &str) -> IResult<&str, Cow<str>> {
+fn word(inp: &str) -> IResult<&str, &str> {
     map_res(
         recognize(tuple((
-            alt((
-                preceded(char('\\'), anychar),
-                satisfy(|c| !c.is_whitespace() && !ESCAPE_IN_WORD.contains(&c) && c != '-'),
-            )),
-            many0(alt((
-                preceded(char('\\'), anychar),
-                satisfy(|c: char| !c.is_whitespace() && !ESCAPE_IN_WORD.contains(&c)),
-            ))),
+            satisfy(|c| {
+                !c.is_whitespace()
+                    && !['-', '^', '`', ':', '{', '}', '"', '[', ']', '(', ')'].contains(&c)
+            }),
+            many0(satisfy(|c: char| {
+                !c.is_whitespace() && ![':', '^', '{', '}', '"', '[', ']', '(', ')'].contains(&c)
+            })),
         ))),
         |s| match s {
             "OR" | "AND" | "NOT" | "IN" => Err(Error::new(inp, ErrorKind::Tag)),
-            s if s.contains('\\') => Ok(Cow::Owned(interpret_escape(s))),
-            s => Ok(Cow::Borrowed(s)),
+            _ => Ok(s),
         },
     )(inp)
 }
 
-fn word_infallible(
-    delimiter: &str,
-    emit_error: bool,
-) -> impl Fn(&str) -> JResult<&str, Option<Cow<str>>> + '_ {
-    // emit error is set when receiving an unescaped `:` should emit an error
-
-    move |inp| {
-        map(
-            opt_i_err(
-                preceded(
-                    multispace0,
-                    recognize(many1(alt((
-                        preceded(char::<&str, _>('\\'), anychar),
-                        satisfy(|c| !c.is_whitespace() && !delimiter.contains(c)),
-                    )))),
-                ),
-                "expected word",
+fn word_infallible(delimiter: &str) -> impl Fn(&str) -> JResult<&str, Option<&str>> + '_ {
+    |inp| {
+        opt_i_err(
+            preceded(
+                multispace0,
+                recognize(many1(satisfy(|c| {
+                    !c.is_whitespace() && !delimiter.contains(c)
+                }))),
             ),
-            |(opt_s, mut errors)| match opt_s {
-                Some(s) => {
-                    if emit_error
-                        && (s
-                            .as_bytes()
-                            .windows(2)
-                            .any(|window| window[0] != b'\\' && window[1] == b':')
-                            || s.starts_with(':'))
-                    {
-                        errors.push(LenientErrorInternal {
-                            pos: inp.len(),
-                            message: "parsed possible invalid field as term".to_string(),
-                        });
-                    }
-                    if s.contains('\\') {
-                        (Some(Cow::Owned(interpret_escape(s))), errors)
-                    } else {
-                        (Some(Cow::Borrowed(s)), errors)
-                    }
-                }
-                None => (None, errors),
-            },
+            "expected word",
         )(inp)
     }
 }
@@ -216,7 +159,7 @@ fn simple_term_infallible(
                 (value((), char('\'')), simple_quotes),
             ),
             // numbers are parsed with words in this case, as we allow string starting with a -
-            map(word_infallible(delimiter, true), |(text, errors)| {
+            map(word_infallible(delimiter), |(text, errors)| {
                 (text.map(|text| (Delimiter::None, text.to_string())), errors)
             }),
         )(inp)
@@ -379,6 +322,15 @@ fn literal_no_group_infallible(inp: &str) -> JResult<&str, Option<UserInputAst>>
         |((field_name, _, leaf), mut errors)| {
             (
                 leaf.map(|leaf| {
+                    if matches!(&leaf, UserInputLeaf::Literal(literal)
+                            if literal.phrase.contains(':') && literal.delimiter == Delimiter::None)
+                        && field_name.is_none()
+                    {
+                        errors.push(LenientErrorInternal {
+                            pos: inp.len(),
+                            message: "parsed possible invalid field as term".to_string(),
+                        });
+                    }
                     if matches!(&leaf, UserInputLeaf::Literal(literal)
                             if literal.phrase == "NOT" && literal.delimiter == Delimiter::None)
                         && field_name.is_none()
@@ -497,20 +449,20 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
         tuple_infallible((
             opt_i(anychar),
             space0_infallible,
-            word_infallible("]}", false),
+            word_infallible("]}"),
             space1_infallible,
             opt_i_err(
                 terminated(tag("TO"), alt((value((), multispace1), value((), eof)))),
                 "missing keyword TO",
             ),
-            word_infallible("]}", false),
+            word_infallible("]}"),
             opt_i_err(one_of("]}"), "missing range delimiter"),
         )),
         |(
             (lower_bound_kind, _multispace0, lower, _multispace1, to, upper, upper_bound_kind),
             errs,
         )| {
-            let lower_bound = match (lower_bound_kind, lower.as_deref()) {
+            let lower_bound = match (lower_bound_kind, lower) {
                 (_, Some("*")) => UserInputBound::Unbounded,
                 (_, None) => UserInputBound::Unbounded,
                 // if it is some, TO was actually the bound (i.e. [TO TO something])
@@ -519,7 +471,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
                 (Some('{'), Some(bound)) => UserInputBound::Exclusive(bound.to_string()),
                 _ => unreachable!("precondition failed, range did not start with [ or {{"),
             };
-            let upper_bound = match (upper_bound_kind, upper.as_deref()) {
+            let upper_bound = match (upper_bound_kind, upper) {
                 (_, Some("*")) => UserInputBound::Unbounded,
                 (_, None) => UserInputBound::Unbounded,
                 (Some(']'), Some(bound)) => UserInputBound::Inclusive(bound.to_string()),
@@ -536,7 +488,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
             (
                 (
                     value((), tag(">=")),
-                    map(word_infallible("", false), |(bound, err)| {
+                    map(word_infallible(""), |(bound, err)| {
                         (
                             (
                                 bound
@@ -550,7 +502,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
                 ),
                 (
                     value((), tag("<=")),
-                    map(word_infallible("", false), |(bound, err)| {
+                    map(word_infallible(""), |(bound, err)| {
                         (
                             (
                                 UserInputBound::Unbounded,
@@ -564,7 +516,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
                 ),
                 (
                     value((), tag(">")),
-                    map(word_infallible("", false), |(bound, err)| {
+                    map(word_infallible(""), |(bound, err)| {
                         (
                             (
                                 bound
@@ -578,7 +530,7 @@ fn range_infallible(inp: &str) -> JResult<&str, UserInputLeaf> {
                 ),
                 (
                     value((), tag("<")),
-                    map(word_infallible("", false), |(bound, err)| {
+                    map(word_infallible(""), |(bound, err)| {
                         (
                             (
                                 UserInputBound::Unbounded,
@@ -1205,12 +1157,6 @@ mod test {
         test_parse_query_to_ast_helper("weight: <= 70", "\"weight\":{\"*\" TO \"70\"]");
 
         test_parse_query_to_ast_helper("weight: <= 70.5", "\"weight\":{\"*\" TO \"70.5\"]");
-
-        test_parse_query_to_ast_helper(">a", "{\"a\" TO \"*\"}");
-        test_parse_query_to_ast_helper(">=a", "[\"a\" TO \"*\"}");
-        test_parse_query_to_ast_helper("<a", "{\"*\" TO \"a\"}");
-        test_parse_query_to_ast_helper("<=a", "{\"*\" TO \"a\"]");
-        test_parse_query_to_ast_helper("<=bsd", "{\"*\" TO \"bsd\"]");
     }
 
     #[test]
@@ -1644,21 +1590,5 @@ mod test {
             r#"myfield:'hello\"happy\'tax'"#,
             r#""myfield":'hello"happy'tax'"#,
         );
-        // we don't process escape sequence for chars which don't require it
-        test_parse_query_to_ast_helper(r#"abc\*"#, r#"abc\*"#);
-    }
-
-    #[test]
-    fn test_queries_with_colons() {
-        test_parse_query_to_ast_helper(r#""abc:def""#, r#""abc:def""#);
-        test_parse_query_to_ast_helper(r#"'abc:def'"#, r#"'abc:def'"#);
-        test_parse_query_to_ast_helper(r#"abc\:def"#, r#"abc:def"#);
-        test_parse_query_to_ast_helper(r#""abc\:def""#, r#""abc:def""#);
-        test_parse_query_to_ast_helper(r#"'abc\:def'"#, r#"'abc:def'"#);
-    }
-
-    #[test]
-    fn test_invalid_field() {
-        test_is_parse_err(r#"!bc:def"#, "!bc:def");
     }
 }

src/aggregation/agg_bench.rs

@@ -0,0 +1,585 @@
+#[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

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

src/aggregation/agg_req.rs

@@ -34,7 +34,7 @@ use super::bucket::{
     DateHistogramAggregationReq, HistogramAggregation, RangeAggregation, TermsAggregation,
 };
 use super::metric::{
-    AverageAggregation, CountAggregation, ExtendedStatsAggregation, MaxAggregation, MinAggregation,
+    AverageAggregation, CountAggregation, MaxAggregation, MinAggregation,
     PercentilesAggregationReq, StatsAggregation, SumAggregation, TopHitsAggregation,
 };
 
@@ -146,11 +146,6 @@ pub enum AggregationVariants {
     /// extracted values.
     #[serde(rename = "stats")]
     Stats(StatsAggregation),
-    /// Computes a collection of estended statistics (`min`, `max`, `sum`, `count`, `avg`,
-    /// `sum_of_squares`, `variance`, `variance_sampling`, `std_deviation`,
-    /// `std_deviation_sampling`) over the  extracted values.
-    #[serde(rename = "extended_stats")]
-    ExtendedStats(ExtendedStatsAggregation),
     /// Computes the sum of the extracted values.
     #[serde(rename = "sum")]
     Sum(SumAggregation),
@@ -175,7 +170,6 @@ impl AggregationVariants {
             AggregationVariants::Max(max) => vec![max.field_name()],
             AggregationVariants::Min(min) => vec![min.field_name()],
             AggregationVariants::Stats(stats) => vec![stats.field_name()],
-            AggregationVariants::ExtendedStats(extended_stats) => vec![extended_stats.field_name()],
             AggregationVariants::Sum(sum) => vec![sum.field_name()],
             AggregationVariants::Percentiles(per) => vec![per.field_name()],
             AggregationVariants::TopHits(top_hits) => top_hits.field_names(),
@@ -203,12 +197,6 @@ impl AggregationVariants {
             _ => None,
         }
     }
-    pub(crate) fn as_top_hits(&self) -> Option<&TopHitsAggregation> {
-        match &self {
-            AggregationVariants::TopHits(top_hits) => Some(top_hits),
-            _ => None,
-        }
-    }
 
     pub(crate) fn as_percentile(&self) -> Option<&PercentilesAggregationReq> {
         match &self {

src/aggregation/agg_req_with_accessor.rs

@@ -11,14 +11,13 @@ use super::bucket::{
     DateHistogramAggregationReq, HistogramAggregation, RangeAggregation, TermsAggregation,
 };
 use super::metric::{
-    AverageAggregation, CountAggregation, ExtendedStatsAggregation, MaxAggregation, MinAggregation,
-    StatsAggregation, SumAggregation,
+    AverageAggregation, CountAggregation, MaxAggregation, MinAggregation, StatsAggregation,
+    SumAggregation,
 };
 use super::segment_agg_result::AggregationLimits;
 use super::VecWithNames;
 use crate::aggregation::{f64_to_fastfield_u64, Key};
-use crate::index::SegmentReader;
-use crate::SegmentOrdinal;
+use crate::{SegmentOrdinal, SegmentReader};
 
 #[derive(Default)]
 pub(crate) struct AggregationsWithAccessor {
@@ -276,10 +275,6 @@ impl AggregationWithAccessor {
                 field: ref field_name,
                 ..
             })
-            | ExtendedStats(ExtendedStatsAggregation {
-                field: ref field_name,
-                ..
-            })
             | Sum(SumAggregation {
                 field: ref field_name,
                 ..
@@ -339,8 +334,8 @@ fn get_missing_val(
         }
         _ => {
             return Err(crate::TantivyError::InvalidArgument(format!(
-                "Missing value {missing:?} for field {field_name} is not supported for column \
-                 type {column_type:?}"
+                "Missing value {:?} for field {} is not supported for column type {:?}",
+                missing, field_name, column_type
             )));
         }
     };
@@ -407,7 +402,7 @@ fn get_dynamic_columns(
         .iter()
         .map(|h| h.open())
         .collect::<io::Result<_>>()?;
-    assert!(!ff_fields.is_empty(), "field {field_name} not found");
+    assert!(!ff_fields.is_empty(), "field {} not found", field_name);
     Ok(cols)
 }
 

src/aggregation/agg_result.rs

@@ -8,9 +8,7 @@ use rustc_hash::FxHashMap;
 use serde::{Deserialize, Serialize};
 
 use super::bucket::GetDocCount;
-use super::metric::{
-    ExtendedStats, PercentilesMetricResult, SingleMetricResult, Stats, TopHitsMetricResult,
-};
+use super::metric::{PercentilesMetricResult, SingleMetricResult, Stats, TopHitsMetricResult};
 use super::{AggregationError, Key};
 use crate::TantivyError;
 
@@ -90,8 +88,6 @@ pub enum MetricResult {
     Min(SingleMetricResult),
     /// Stats metric result.
     Stats(Stats),
-    /// ExtendedStats metric result.
-    ExtendedStats(Box<ExtendedStats>),
     /// Sum metric result.
     Sum(SingleMetricResult),
     /// Percentiles metric result.
@@ -108,7 +104,6 @@ impl MetricResult {
             MetricResult::Max(max) => Ok(max.value),
             MetricResult::Min(min) => Ok(min.value),
             MetricResult::Stats(stats) => stats.get_value(agg_property),
-            MetricResult::ExtendedStats(extended_stats) => extended_stats.get_value(agg_property),
             MetricResult::Sum(sum) => Ok(sum.value),
             MetricResult::Percentiles(_) => Err(TantivyError::AggregationError(
                 AggregationError::InvalidRequest("percentiles can't be used to order".to_string()),

src/aggregation/bucket/histogram/histogram.rs

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

src/aggregation/bucket/term_agg.rs

@@ -324,11 +324,9 @@ impl SegmentAggregationCollector for SegmentTermCollector {
         }
 
         let mem_delta = self.get_memory_consumption() - mem_pre;
-        if mem_delta > 0 {
-            bucket_agg_accessor
-                .limits
-                .add_memory_consumed(mem_delta as u64)?;
-        }
+        bucket_agg_accessor
+            .limits
+            .add_memory_consumed(mem_delta as u64)?;
 
         Ok(())
     }
@@ -357,7 +355,8 @@ impl SegmentTermCollector {
     ) -> crate::Result<Self> {
         if field_type == ColumnType::Bytes {
             return Err(TantivyError::InvalidArgument(format!(
-                "terms aggregation is not supported for column type {field_type:?}"
+                "terms aggregation is not supported for column type {:?}",
+                field_type
             )));
         }
         let term_buckets = TermBuckets::default();

src/aggregation/collector.rs

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

src/aggregation/intermediate_agg_result.rs

@@ -19,8 +19,8 @@ use super::bucket::{
     GetDocCount, Order, OrderTarget, RangeAggregation, TermsAggregation,
 };
 use super::metric::{
-    IntermediateAverage, IntermediateCount, IntermediateExtendedStats, IntermediateMax,
-    IntermediateMin, IntermediateStats, IntermediateSum, PercentilesCollector, TopHitsTopNComputer,
+    IntermediateAverage, IntermediateCount, IntermediateMax, IntermediateMin, IntermediateStats,
+    IntermediateSum, PercentilesCollector, TopHitsTopNComputer,
 };
 use super::segment_agg_result::AggregationLimits;
 use super::{format_date, AggregationError, Key, SerializedKey};
@@ -215,9 +215,6 @@ pub(crate) fn empty_from_req(req: &Aggregation) -> IntermediateAggregationResult
         Stats(_) => IntermediateAggregationResult::Metric(IntermediateMetricResult::Stats(
             IntermediateStats::default(),
         )),
-        ExtendedStats(_) => IntermediateAggregationResult::Metric(
-            IntermediateMetricResult::ExtendedStats(IntermediateExtendedStats::default()),
-        ),
         Sum(_) => IntermediateAggregationResult::Metric(IntermediateMetricResult::Sum(
             IntermediateSum::default(),
         )),
@@ -225,7 +222,7 @@ pub(crate) fn empty_from_req(req: &Aggregation) -> IntermediateAggregationResult
             IntermediateMetricResult::Percentiles(PercentilesCollector::default()),
         ),
         TopHits(ref req) => IntermediateAggregationResult::Metric(
-            IntermediateMetricResult::TopHits(TopHitsTopNComputer::new(req)),
+            IntermediateMetricResult::TopHits(TopHitsTopNComputer::new(req.clone())),
         ),
     }
 }
@@ -285,8 +282,6 @@ pub enum IntermediateMetricResult {
     Min(IntermediateMin),
     /// Intermediate stats result.
     Stats(IntermediateStats),
-    /// Intermediate stats result.
-    ExtendedStats(IntermediateExtendedStats),
     /// Intermediate sum result.
     Sum(IntermediateSum),
     /// Intermediate top_hits result
@@ -311,9 +306,6 @@ impl IntermediateMetricResult {
             IntermediateMetricResult::Stats(intermediate_stats) => {
                 MetricResult::Stats(intermediate_stats.finalize())
             }
-            IntermediateMetricResult::ExtendedStats(intermediate_stats) => {
-                MetricResult::ExtendedStats(intermediate_stats.finalize())
-            }
             IntermediateMetricResult::Sum(intermediate_sum) => {
                 MetricResult::Sum(intermediate_sum.finalize().into())
             }
@@ -354,12 +346,6 @@ impl IntermediateMetricResult {
             ) => {
                 stats_left.merge_fruits(stats_right);
             }
-            (
-                IntermediateMetricResult::ExtendedStats(extended_stats_left),
-                IntermediateMetricResult::ExtendedStats(extended_stats_right),
-            ) => {
-                extended_stats_left.merge_fruits(extended_stats_right);
-            }
             (IntermediateMetricResult::Sum(sum_left), IntermediateMetricResult::Sum(sum_right)) => {
                 sum_left.merge_fruits(sum_right);
             }

src/aggregation/metric/mod.rs

@@ -18,7 +18,6 @@
 
 mod average;
 mod count;
-mod extended_stats;
 mod max;
 mod min;
 mod percentiles;
@@ -30,7 +29,6 @@ use std::collections::HashMap;
 
 pub use average::*;
 pub use count::*;
-pub use extended_stats::*;
 pub use max::*;
 pub use min::*;
 pub use percentiles::*;

src/aggregation/metric/stats.rs

@@ -1,5 +1,3 @@
-use std::fmt::Debug;
-
 use serde::{Deserialize, Serialize};
 
 use super::*;
@@ -87,15 +85,13 @@ impl Stats {
 #[derive(Clone, Debug, PartialEq, Serialize, Deserialize)]
 pub struct IntermediateStats {
     /// The number of extracted values.
-    pub(crate) count: u64,
+    count: u64,
     /// The sum of the extracted values.
-    pub(crate) sum: f64,
-    /// delta for sum needed for [Kahan algorithm for summation](https://en.wikipedia.org/wiki/Kahan_summation_algorithm)
-    pub(crate) delta: f64,
+    sum: f64,
     /// The min value.
-    pub(crate) min: f64,
+    min: f64,
     /// The max value.
-    pub(crate) max: f64,
+    max: f64,
 }
 
 impl Default for IntermediateStats {
@@ -103,7 +99,6 @@ impl Default for IntermediateStats {
         Self {
             count: 0,
             sum: 0.0,
-            delta: 0.0,
             min: f64::MAX,
             max: f64::MIN,
         }
@@ -114,13 +109,7 @@ impl IntermediateStats {
     /// Merges the other stats intermediate result into self.
     pub fn merge_fruits(&mut self, other: IntermediateStats) {
         self.count += other.count;
-
-        // kahan algorithm for sum
-        let y = other.sum - (self.delta + other.delta);
-        let t = self.sum + y;
-        self.delta = (t - self.sum) - y;
-        self.sum = t;
-
+        self.sum += other.sum;
         self.min = self.min.min(other.min);
         self.max = self.max.max(other.max);
     }
@@ -152,15 +141,9 @@ impl IntermediateStats {
     }
 
     #[inline]
-    pub(in crate::aggregation::metric) fn collect(&mut self, value: f64) {
+    fn collect(&mut self, value: f64) {
         self.count += 1;
-
-        // kahan algorithm for sum
-        let y = value - self.delta;
-        let t = self.sum + y;
-        self.delta = (t - self.sum) - y;
-        self.sum = t;
-
+        self.sum += value;
         self.min = self.min.min(value);
         self.max = self.max.max(value);
     }
@@ -305,6 +288,7 @@ impl SegmentAggregationCollector for SegmentStatsCollector {
 
 #[cfg(test)]
 mod tests {
+
     use serde_json::Value;
 
     use crate::aggregation::agg_req::{Aggregation, Aggregations};

src/aggregation/metric/top_hits.rs

@@ -1,7 +1,7 @@
 use std::collections::HashMap;
 use std::net::Ipv6Addr;
 
-use columnar::{Column, ColumnType, ColumnarReader, DynamicColumn};
+use columnar::{ColumnarReader, DynamicColumn};
 use common::json_path_writer::JSON_PATH_SEGMENT_SEP_STR;
 use common::DateTime;
 use regex::Regex;
@@ -131,8 +131,8 @@ impl<'de> Deserialize<'de> for KeyOrder {
         ))?;
         if key_order.next().is_some() {
             return Err(serde::de::Error::custom(format!(
-                "Expected exactly one key-value pair in sort parameter of top_hits, found \
-                 {key_order:?}"
+                "Expected exactly one key-value pair in sort parameter of top_hits, found {:?}",
+                key_order
             )));
         }
         Ok(Self { field, order })
@@ -144,22 +144,27 @@ fn globbed_string_to_regex(glob: &str) -> Result<Regex, crate::TantivyError> {
     // Replace `*` glob with `.*` regex
     let sanitized = format!("^{}$", regex::escape(glob).replace(r"\*", ".*"));
     Regex::new(&sanitized.replace('*', ".*")).map_err(|e| {
-        crate::TantivyError::SchemaError(format!("Invalid regex '{glob}' in docvalue_fields: {e}"))
+        crate::TantivyError::SchemaError(format!(
+            "Invalid regex '{}' in docvalue_fields: {}",
+            glob, e
+        ))
     })
 }
 
 fn use_doc_value_fields_err(parameter: &str) -> crate::Result<()> {
     Err(crate::TantivyError::AggregationError(
         AggregationError::InvalidRequest(format!(
-            "The `{parameter}` parameter is not supported, only `docvalue_fields` is supported in \
-             `top_hits` aggregation"
+            "The `{}` parameter is not supported, only `docvalue_fields` is supported in \
+             `top_hits` aggregation",
+            parameter
         )),
     ))
 }
 fn unsupported_err(parameter: &str) -> crate::Result<()> {
     Err(crate::TantivyError::AggregationError(
         AggregationError::InvalidRequest(format!(
-            "The `{parameter}` parameter is not supported in the `top_hits` aggregation"
+            "The `{}` parameter is not supported in the `top_hits` aggregation",
+            parameter
         )),
     ))
 }
@@ -212,7 +217,8 @@ impl TopHitsAggregation {
                     .collect::<Vec<_>>();
                 assert!(
                     !fields.is_empty(),
-                    "No fields matched the glob '{field}' in docvalue_fields"
+                    "No fields matched the glob '{}' in docvalue_fields",
+                    field
                 );
                 Ok(fields)
             })
@@ -248,7 +254,7 @@ impl TopHitsAggregation {
             .map(|field| {
                 let accessors = accessors
                     .get(field)
-                    .unwrap_or_else(|| panic!("field '{field}' not found in accessors"));
+                    .unwrap_or_else(|| panic!("field '{}' not found in accessors", field));
 
                 let values: Vec<FastFieldValue> = accessors
                     .iter()
@@ -443,10 +449,10 @@ impl std::cmp::PartialEq for TopHitsTopNComputer {
 
 impl TopHitsTopNComputer {
     /// Create a new TopHitsCollector
-    pub fn new(req: &TopHitsAggregation) -> Self {
+    pub fn new(req: TopHitsAggregation) -> Self {
         Self {
             top_n: TopNComputer::new(req.size + req.from.unwrap_or(0)),
-            req: req.clone(),
+            req,
         }
     }
 
@@ -491,6 +497,7 @@ impl TopHitsTopNComputer {
 pub(crate) struct TopHitsSegmentCollector {
     segment_ordinal: SegmentOrdinal,
     accessor_idx: usize,
+    req: TopHitsAggregation,
     top_n: TopNComputer<Vec<DocValueAndOrder>, DocAddress, false>,
 }
 
@@ -501,6 +508,7 @@ impl TopHitsSegmentCollector {
         segment_ordinal: SegmentOrdinal,
     ) -> Self {
         Self {
+            req: req.clone(),
             top_n: TopNComputer::new(req.size + req.from.unwrap_or(0)),
             segment_ordinal,
             accessor_idx,
@@ -509,13 +517,14 @@ impl TopHitsSegmentCollector {
     fn into_top_hits_collector(
         self,
         value_accessors: &HashMap<String, Vec<DynamicColumn>>,
-        req: &TopHitsAggregation,
     ) -> TopHitsTopNComputer {
-        let mut top_hits_computer = TopHitsTopNComputer::new(req);
+        let mut top_hits_computer = TopHitsTopNComputer::new(self.req.clone());
         let top_results = self.top_n.into_vec();
 
         for res in top_results {
-            let doc_value_fields = req.get_document_field_data(value_accessors, res.doc.doc_id);
+            let doc_value_fields = self
+                .req
+                .get_document_field_data(value_accessors, res.doc.doc_id);
             top_hits_computer.collect(
                 DocSortValuesAndFields {
                     sorts: res.feature,
@@ -527,15 +536,34 @@ impl TopHitsSegmentCollector {
 
         top_hits_computer
     }
+}
 
-    /// TODO add a specialized variant for a single sort field
-    fn collect_with(
+impl SegmentAggregationCollector for TopHitsSegmentCollector {
+    fn add_intermediate_aggregation_result(
+        self: Box<Self>,
+        agg_with_accessor: &crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
+        results: &mut crate::aggregation::intermediate_agg_result::IntermediateAggregationResults,
+    ) -> crate::Result<()> {
+        let name = agg_with_accessor.aggs.keys[self.accessor_idx].to_string();
+
+        let value_accessors = &agg_with_accessor.aggs.values[self.accessor_idx].value_accessors;
+
+        let intermediate_result =
+            IntermediateMetricResult::TopHits(self.into_top_hits_collector(value_accessors));
+        results.push(
+            name,
+            IntermediateAggregationResult::Metric(intermediate_result),
+        )
+    }
+
+    fn collect(
         &mut self,
         doc_id: crate::DocId,
-        req: &TopHitsAggregation,
-        accessors: &[(Column<u64>, ColumnType)],
+        agg_with_accessor: &mut crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let sorts: Vec<DocValueAndOrder> = req
+        let accessors = &agg_with_accessor.aggs.values[self.accessor_idx].accessors;
+        let sorts: Vec<DocValueAndOrder> = self
+            .req
             .sort
             .iter()
             .enumerate()
@@ -560,62 +588,15 @@ impl TopHitsSegmentCollector {
         );
         Ok(())
     }
-}
-
-impl SegmentAggregationCollector for TopHitsSegmentCollector {
-    fn add_intermediate_aggregation_result(
-        self: Box<Self>,
-        agg_with_accessor: &crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
-        results: &mut crate::aggregation::intermediate_agg_result::IntermediateAggregationResults,
-    ) -> crate::Result<()> {
-        let name = agg_with_accessor.aggs.keys[self.accessor_idx].to_string();
-
-        let value_accessors = &agg_with_accessor.aggs.values[self.accessor_idx].value_accessors;
-        let tophits_req = &agg_with_accessor.aggs.values[self.accessor_idx]
-            .agg
-            .agg
-            .as_top_hits()
-            .expect("aggregation request must be of type top hits");
-
-        let intermediate_result = IntermediateMetricResult::TopHits(
-            self.into_top_hits_collector(value_accessors, tophits_req),
-        );
-        results.push(
-            name,
-            IntermediateAggregationResult::Metric(intermediate_result),
-        )
-    }
-
-    /// TODO: Consider a caching layer to reduce the call overhead
-    fn collect(
-        &mut self,
-        doc_id: crate::DocId,
-        agg_with_accessor: &mut crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
-    ) -> crate::Result<()> {
-        let tophits_req = &agg_with_accessor.aggs.values[self.accessor_idx]
-            .agg
-            .agg
-            .as_top_hits()
-            .expect("aggregation request must be of type top hits");
-        let accessors = &agg_with_accessor.aggs.values[self.accessor_idx].accessors;
-        self.collect_with(doc_id, tophits_req, accessors)?;
-        Ok(())
-    }
 
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
         agg_with_accessor: &mut crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let tophits_req = &agg_with_accessor.aggs.values[self.accessor_idx]
-            .agg
-            .agg
-            .as_top_hits()
-            .expect("aggregation request must be of type top hits");
-        let accessors = &agg_with_accessor.aggs.values[self.accessor_idx].accessors;
         // TODO: Consider getting fields with the column block accessor.
         for doc in docs {
-            self.collect_with(*doc, tophits_req, accessors)?;
+            self.collect(*doc, agg_with_accessor)?;
         }
         Ok(())
     }

src/aggregation/mod.rs

@@ -143,6 +143,8 @@ use std::fmt::Display;
 #[cfg(test)]
 mod agg_tests;
 
+mod agg_bench;
+
 use core::fmt;
 
 pub use agg_limits::AggregationLimits;
@@ -158,14 +160,15 @@ use serde::de::{self, Visitor};
 use serde::{Deserialize, Deserializer, Serialize};
 
 fn parse_str_into_f64<E: de::Error>(value: &str) -> Result<f64, E> {
-    let parsed = value
-        .parse::<f64>()
-        .map_err(|_err| de::Error::custom(format!("Failed to parse f64 from string: {value:?}")))?;
+    let parsed = value.parse::<f64>().map_err(|_err| {
+        de::Error::custom(format!("Failed to parse f64 from string: {:?}", value))
+    })?;
 
     // Check if the parsed value is NaN or infinity
     if parsed.is_nan() || parsed.is_infinite() {
         Err(de::Error::custom(format!(
-            "Value is not a valid f64 (NaN or Infinity): {value:?}"
+            "Value is not a valid f64 (NaN or Infinity): {:?}",
+            value
         )))
     } else {
         Ok(parsed)

src/aggregation/segment_agg_result.rs

@@ -11,12 +11,12 @@ use super::agg_req_with_accessor::{AggregationWithAccessor, AggregationsWithAcce
 use super::bucket::{SegmentHistogramCollector, SegmentRangeCollector, SegmentTermCollector};
 use super::intermediate_agg_result::IntermediateAggregationResults;
 use super::metric::{
-    AverageAggregation, CountAggregation, ExtendedStatsAggregation, MaxAggregation, MinAggregation,
+    AverageAggregation, CountAggregation, MaxAggregation, MinAggregation,
     SegmentPercentilesCollector, SegmentStatsCollector, SegmentStatsType, StatsAggregation,
     SumAggregation,
 };
 use crate::aggregation::bucket::TermMissingAgg;
-use crate::aggregation::metric::{SegmentExtendedStatsCollector, TopHitsSegmentCollector};
+use crate::aggregation::metric::TopHitsSegmentCollector;
 
 pub(crate) trait SegmentAggregationCollector: CollectorClone + Debug {
     fn add_intermediate_aggregation_result(
@@ -148,9 +148,6 @@ pub(crate) fn build_single_agg_segment_collector(
             accessor_idx,
             *missing,
         ))),
-        ExtendedStats(ExtendedStatsAggregation { missing, sigma, .. }) => Ok(Box::new(
-            SegmentExtendedStatsCollector::from_req(req.field_type, *sigma, accessor_idx, *missing),
-        )),
         Sum(SumAggregation { missing, .. }) => Ok(Box::new(SegmentStatsCollector::from_req(
             req.field_type,
             SegmentStatsType::Sum,

src/collector/facet_collector.rs

@@ -598,7 +598,7 @@ mod tests {
                 let mid = n % 4;
                 n /= 4;
                 let leaf = n % 5;
-                Facet::from(&format!("/top{top}/mid{mid}/leaf{leaf}"))
+                Facet::from(&format!("/top{}/mid{}/leaf{}", top, mid, leaf))
             })
             .collect();
         for i in 0..num_facets * 10 {
@@ -737,7 +737,7 @@ mod tests {
             vec![("a", 10), ("b", 100), ("c", 7), ("d", 12), ("e", 21)]
                 .into_iter()
                 .flat_map(|(c, count)| {
-                    let facet = Facet::from(&format!("/facet/{c}"));
+                    let facet = Facet::from(&format!("/facet/{}", c));
                     let doc = doc!(facet_field => facet);
                     iter::repeat(doc).take(count)
                 })
@@ -785,7 +785,7 @@ mod tests {
         let docs: Vec<TantivyDocument> = vec![("b", 2), ("a", 2), ("c", 4)]
             .into_iter()
             .flat_map(|(c, count)| {
-                let facet = Facet::from(&format!("/facet/{c}"));
+                let facet = Facet::from(&format!("/facet/{}", c));
                 let doc = doc!(facet_field => facet);
                 iter::repeat(doc).take(count)
             })

src/collector/top_collector.rs

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

src/collector/top_score_collector.rs

@@ -871,10 +871,7 @@ mod tests {
     use crate::schema::{Field, Schema, FAST, STORED, TEXT};
     use crate::time::format_description::well_known::Rfc3339;
     use crate::time::OffsetDateTime;
-    use crate::{
-        assert_nearly_equals, DateTime, DocAddress, DocId, Index, IndexWriter, Order, Score,
-        SegmentReader,
-    };
+    use crate::{DateTime, DocAddress, DocId, Index, IndexWriter, Order, Score, SegmentReader};
 
     fn make_index() -> crate::Result<Index> {
         let mut schema_builder = Schema::builder();

src/core/executor.rs

@@ -1,25 +1,19 @@
-use std::sync::Arc;
-
-#[cfg(feature = "quickwit")]
-use futures_util::{future::Either, FutureExt};
+use rayon::{ThreadPool, ThreadPoolBuilder};
 
 use crate::TantivyError;
 
-/// Executor makes it possible to run tasks in single thread or
-/// in a thread pool.
-#[derive(Clone)]
+/// 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.
 pub enum Executor {
     /// Single thread variant of an Executor
     SingleThread,
     /// Thread pool variant of an Executor
-    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)
-    }
+    ThreadPool(ThreadPool),
 }
 
 impl Executor {
@@ -30,11 +24,11 @@ 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 = rayon::ThreadPoolBuilder::new()
+        let pool = ThreadPoolBuilder::new()
             .num_threads(num_threads)
             .thread_name(move |num| format!("{prefix}{num}"))
             .build()?;
-        Ok(Executor::ThreadPool(Arc::new(pool)))
+        Ok(Executor::ThreadPool(pool))
     }
 
     /// Perform a map in the thread pool.
@@ -97,36 +91,11 @@ 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]
@@ -178,62 +147,4 @@ 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;
-
-        let counter: Arc<AtomicU64> = Default::default();
-
-        let other_counter: Arc<AtomicU64> = Default::default();
-
-        let mut futures = Vec::new();
-        let mut other_futures = Vec::new();
-
-        let (tx, rx) = crossbeam_channel::bounded::<()>(0);
-        let rx = Arc::new(rx);
-        let executor = Executor::multi_thread(3, "search-test").unwrap();
-        for _ in 0..1000 {
-            let counter_clone: Arc<AtomicU64> = counter.clone();
-            let other_counter_clone: Arc<AtomicU64> = other_counter.clone();
-
-            let rx_clone = rx.clone();
-            let rx_clone2 = rx.clone();
-            let fut = executor.spawn_blocking(move || {
-                counter_clone.fetch_add(1, Ordering::SeqCst);
-                let _ = rx_clone.recv();
-            });
-            futures.push(fut);
-            let other_fut = executor.spawn_blocking(move || {
-                other_counter_clone.fetch_add(1, Ordering::SeqCst);
-                let _ = rx_clone2.recv();
-            });
-            other_futures.push(other_fut);
-        }
-
-        // We execute 100 futures.
-        for _ in 0..100 {
-            tx.send(()).unwrap();
-        }
-
-        let counter_val = counter.load(Ordering::SeqCst);
-        let other_counter_val = other_counter.load(Ordering::SeqCst);
-        assert!(counter_val >= 30);
-        assert!(other_counter_val >= 30);
-
-        drop(other_futures);
-
-        // We execute 100 futures.
-        for _ in 0..100 {
-            tx.send(()).unwrap();
-        }
-
-        let counter_val2 = counter.load(Ordering::SeqCst);
-        assert!(counter_val2 >= counter_val + 100 - 6);
-
-        let other_counter_val2 = other_counter.load(Ordering::SeqCst);
-        assert!(other_counter_val2 <= other_counter_val + 6);
-    }
 }

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 appears in the second object, a non increasing value would be pushed
+/// Worse, if a same term is 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)]
-pub(crate) struct IndexingPositionsPerPath {
+struct IndexingPositionsPerPath {
     positions_per_path: FxHashMap<u32, IndexingPosition>,
 }
 
@@ -58,9 +58,6 @@ 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.
@@ -71,6 +68,36 @@ 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,
@@ -99,7 +126,7 @@ fn index_json_object<'a, V: Value<'a>>(
 }
 
 #[allow(clippy::too_many_arguments)]
-pub(crate) fn index_json_value<'a, V: Value<'a>>(
+fn index_json_value<'a, V: Value<'a>>(
     doc: DocId,
     json_value: V,
     text_analyzer: &mut TextAnalyzer,
@@ -139,18 +166,12 @@ pub(crate) 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()),
                 );
-                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);
-                }
+                term_buffer.append_type_and_fast_value(val);
                 postings_writer.subscribe(doc, 0u32, term_buffer, ctx);
             }
             ReferenceValueLeaf::I64(val) => {
@@ -236,7 +257,10 @@ pub(crate) 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 fn convert_to_fast_value_and_append_to_json_term(mut term: Term, phrase: &str) -> Option<Term> {
+pub(crate) 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()
@@ -338,14 +362,14 @@ mod tests {
         let mut term = Term::from_field_json_path(field, "attributes.color", false);
         term.append_type_and_str("red");
         assert_eq!(
-            format!("{term:?}"),
+            format!("{:?}", term),
             "Term(field=1, type=Json, path=attributes.color, type=Str, \"red\")"
         );
 
         let mut term = Term::from_field_json_path(field, "attributes.dimensions.width", false);
         term.append_type_and_fast_value(400i64);
         assert_eq!(
-            format!("{term:?}"),
+            format!("{:?}", term),
             "Term(field=1, type=Json, path=attributes.dimensions.width, type=I64, 400)"
         );
     }

src/core/searcher.rs

@@ -4,13 +4,13 @@ use std::{fmt, io};
 
 use crate::collector::Collector;
 use crate::core::Executor;
-use crate::index::{SegmentId, SegmentReader};
+use crate::index::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, TrackedObject};
+use crate::{DocAddress, Index, Opstamp, SegmentId, TrackedObject};
 
 /// Identifies the searcher generation accessed by a [`Searcher`].
 ///
@@ -109,9 +109,8 @@ 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, executor).await
+        store_reader.get_async(doc_address.doc_id).await
     }
 
     /// Access the schema associated with the index of this searcher.

src/core/tests.rs

@@ -1,14 +1,12 @@
 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, ReloadPolicy,
-    TantivyDocument, Term,
+    Directory, DocSet, Index, IndexBuilder, IndexReader, IndexSettings, IndexWriter, Postings,
+    ReloadPolicy, SegmentId, TantivyDocument, Term,
 };
 
 #[test]
@@ -419,7 +417,7 @@ 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(75i64);
+    term.append_type_and_fast_value(75u64);
 
     let postings = inv_idx
         .read_postings(&term, IndexRecordOption::WithFreqsAndPositions)
@@ -453,7 +451,7 @@ 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(75i64);
+    term.append_type_and_fast_value(75u64);
 
     let mut postings = inv_idx
         .read_postings(&term, IndexRecordOption::WithFreqsAndPositions)

src/directory/mmap_directory.rs

@@ -566,7 +566,7 @@ mod tests {
         let mmap_directory = MmapDirectory::create_from_tempdir().unwrap();
         let num_paths = 10;
         let paths: Vec<PathBuf> = (0..num_paths)
-            .map(|i| PathBuf::from(&*format!("file_{i}")))
+            .map(|i| PathBuf::from(&*format!("file_{}", i)))
             .collect();
         {
             for path in &paths {

src/fastfield/facet_reader.rs

@@ -62,7 +62,8 @@ impl FacetReader {
 
 #[cfg(test)]
 mod tests {
-    use crate::schema::{Facet, FacetOptions, SchemaBuilder, Value, STORED};
+    use crate::schema::document::Value;
+    use crate::schema::{Facet, FacetOptions, SchemaBuilder, STORED};
     use crate::{DocAddress, Index, IndexWriter, TantivyDocument};
 
     #[test]
@@ -88,9 +89,7 @@ mod tests {
         let doc = searcher
             .doc::<TantivyDocument>(DocAddress::new(0u32, 0u32))
             .unwrap();
-        let value = doc
-            .get_first(facet_field)
-            .and_then(|v| v.as_value().as_facet());
+        let value = doc.get_first(facet_field).and_then(|v| v.as_facet());
         assert_eq!(value, None);
     }
 
@@ -147,11 +146,8 @@ 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())
-            .map(|facet| Facet::from_encoded_string(facet.to_string()));
-        assert_eq!(value, Facet::from_text("/a/b").ok());
+        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());
         Ok(())
     }
 

src/fastfield/mod.rs

@@ -80,7 +80,7 @@ mod tests {
     use std::path::Path;
 
     use columnar::StrColumn;
-    use common::{ByteCount, DateTimePrecision, HasLen, TerminatingWrite};
+    use common::{ByteCount, HasLen, TerminatingWrite};
     use once_cell::sync::Lazy;
     use rand::prelude::SliceRandom;
     use rand::rngs::StdRng;
@@ -88,15 +88,14 @@ mod tests {
 
     use super::*;
     use crate::directory::{Directory, RamDirectory, WritePtr};
-    use crate::index::SegmentId;
     use crate::merge_policy::NoMergePolicy;
     use crate::schema::{
-        DateOptions, Facet, FacetOptions, Field, JsonObjectOptions, Schema, SchemaBuilder,
-        TantivyDocument, TextOptions, FAST, INDEXED, STORED, STRING, TEXT,
+        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::{Index, IndexWriter, SegmentReader};
+    use crate::{DateOptions, DateTimePrecision, Index, IndexWriter, SegmentId, 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::{DateTimePrecision, JsonPathWriter};
+use common::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::{DocId, TantivyError};
+use crate::{DateTimePrecision, 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.
@@ -183,7 +183,8 @@ impl FastFieldsWriter {
                         .record_datetime(doc_id, field_name, truncated_datetime);
                 }
                 ReferenceValueLeaf::Facet(val) => {
-                    self.columnar_writer.record_str(doc_id, field_name, val);
+                    self.columnar_writer
+                        .record_str(doc_id, field_name, val.encoded_str());
                 }
                 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::thread::available_parallelism;
+use std::sync::Arc;
 
 use super::segment::Segment;
 use super::segment_reader::merge_field_meta_data;
@@ -252,8 +252,9 @@ impl IndexBuilder {
                 let field_type = entry.field_type().value_type();
                 if !supported_field_types.contains(&field_type) {
                     return Err(TantivyError::InvalidArgument(format!(
-                        "Unsupported field type in sort_by_field: {field_type:?}. Supported field \
-                         types: {supported_field_types:?} ",
+                        "Unsupported field type in sort_by_field: {:?}. Supported field types: \
+                         {:?} ",
+                        field_type, supported_field_types,
                     )));
                 }
             }
@@ -292,7 +293,7 @@ pub struct Index {
     directory: ManagedDirectory,
     schema: Schema,
     settings: IndexSettings,
-    executor: Executor,
+    executor: Arc<Executor>,
     tokenizers: TokenizerManager,
     fast_field_tokenizers: TokenizerManager,
     inventory: SegmentMetaInventory,
@@ -317,25 +318,29 @@ impl Index {
     ///
     /// By default the executor is single thread, and simply runs in the calling thread.
     pub fn search_executor(&self) -> &Executor {
-        &self.executor
+        self.executor.as_ref()
     }
 
     /// Replace the default single thread search executor pool
     /// by a thread pool with a given number of threads.
     pub fn set_multithread_executor(&mut self, num_threads: usize) -> crate::Result<()> {
-        self.executor = Executor::multi_thread(num_threads, "tantivy-search-")?;
+        self.executor = Arc::new(Executor::multi_thread(num_threads, "tantivy-search-")?);
         Ok(())
     }
 
     /// Custom thread pool by a outer thread pool.
-    pub fn set_executor(&mut self, executor: Executor) {
-        self.executor = executor;
+    pub fn set_shared_multithread_executor(
+        &mut self,
+        shared_thread_pool: Arc<Executor>,
+    ) -> crate::Result<()> {
+        self.executor = shared_thread_pool.clone();
+        Ok(())
     }
 
     /// Replace the default single thread search executor pool
     /// by a thread pool with as many threads as there are CPUs on the system.
     pub fn set_default_multithread_executor(&mut self) -> crate::Result<()> {
-        let default_num_threads = available_parallelism()?.get();
+        let default_num_threads = num_cpus::get();
         self.set_multithread_executor(default_num_threads)
     }
 
@@ -413,7 +418,7 @@ impl Index {
             schema,
             tokenizers: TokenizerManager::default(),
             fast_field_tokenizers: TokenizerManager::default(),
-            executor: Executor::single_thread(),
+            executor: Arc::new(Executor::single_thread()),
             inventory,
         }
     }
@@ -616,7 +621,7 @@ impl Index {
         &self,
         memory_budget_in_bytes: usize,
     ) -> crate::Result<IndexWriter<D>> {
-        let mut num_threads = std::cmp::min(available_parallelism()?.get(), MAX_NUM_THREAD);
+        let mut num_threads = std::cmp::min(num_cpus::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

@@ -1,3 +1,5 @@
+//! # 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/index/segment_reader.rs

@@ -318,14 +318,14 @@ impl SegmentReader {
                         if create_canonical {
                             // Without expand dots enabled dots need to be escaped.
                             let escaped_json_path = json_path.replace('.', "\\.");
-                            let full_path = format!("{field_name}.{escaped_json_path}");
+                            let full_path = format!("{}.{}", field_name, escaped_json_path);
                             let full_path_unescaped = format!("{}.{}", field_name, &json_path);
                             map_to_canonical.insert(full_path_unescaped, full_path.to_string());
                             full_path
                         } else {
                             // With expand dots enabled, we can use '.' instead of '\u{1}'.
                             json_path_sep_to_dot(&mut json_path);
-                            format!("{field_name}.{json_path}")
+                            format!("{}.{}", field_name, json_path)
                         }
                     };
                     indexed_fields.extend(

src/indexer/delete_queue.rs

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

src/indexer/doc_id_mapping.rs

@@ -306,10 +306,12 @@ mod tests_indexsorting {
         let my_string_field = index.schema().get_field("string_field").unwrap();
         let searcher = index.reader()?.searcher();
         {
-            assert!(searcher
-                .doc::<TantivyDocument>(DocAddress::new(0, 0))?
-                .get_first(my_string_field)
-                .is_none());
+            assert_eq!(
+                searcher
+                    .doc::<TantivyDocument>(DocAddress::new(0, 0))?
+                    .get_first(my_string_field),
+                None
+            );
             assert_eq!(
                 searcher
                     .doc::<TantivyDocument>(DocAddress::new(0, 3))?
@@ -342,7 +344,7 @@ mod tests_indexsorting {
                 Some("blublub")
             );
             let doc = searcher.doc::<TantivyDocument>(DocAddress::new(0, 4))?;
-            assert!(doc.get_first(my_string_field).is_none());
+            assert_eq!(doc.get_first(my_string_field), None);
         }
         // sort by field desc
         let index = create_test_index(

src/indexer/index_writer.rs

@@ -808,15 +808,16 @@ mod tests {
     use proptest::prop_oneof;
 
     use super::super::operation::UserOperation;
-    use crate::collector::{Count, TopDocs};
+    use crate::collector::TopDocs;
     use crate::directory::error::LockError;
     use crate::error::*;
     use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
     use crate::indexer::NoMergePolicy;
     use crate::query::{BooleanQuery, Occur, Query, QueryParser, TermQuery};
+    use crate::schema::document::Value;
     use crate::schema::{
         self, Facet, FacetOptions, IndexRecordOption, IpAddrOptions, NumericOptions, Schema,
-        TextFieldIndexing, TextOptions, Value, FAST, INDEXED, STORED, STRING, TEXT,
+        TextFieldIndexing, TextOptions, FAST, INDEXED, STORED, STRING, TEXT,
     };
     use crate::store::DOCSTORE_CACHE_CAPACITY;
     use crate::{
@@ -1572,74 +1573,20 @@ mod tests {
         Ok(())
     }
 
-    #[derive(Debug, Clone)]
+    #[derive(Debug, Clone, Copy)]
     enum IndexingOp {
-        AddMultipleDoc {
-            id: u64,
-            num_docs: u64,
-            value: IndexValue,
-        },
-        AddDoc {
-            id: u64,
-            value: IndexValue,
-        },
-        DeleteDoc {
-            id: u64,
-        },
-        DeleteDocQuery {
-            id: u64,
-        },
+        AddDoc { id: u64 },
+        DeleteDoc { id: u64 },
+        DeleteDocQuery { id: u64 },
         Commit,
         Merge,
     }
-    impl IndexingOp {
-        fn add(id: u64) -> Self {
-            IndexingOp::AddDoc {
-                id,
-                value: IndexValue::F64(id as f64),
-            }
-        }
-    }
-
-    use serde::Serialize;
-    #[derive(Debug, Clone, Serialize)]
-    #[serde(untagged)]
-    enum IndexValue {
-        Str(String),
-        F64(f64),
-        U64(u64),
-        I64(i64),
-    }
-    impl Default for IndexValue {
-        fn default() -> Self {
-            IndexValue::F64(0.0)
-        }
-    }
-
-    fn value_strategy() -> impl Strategy<Value = IndexValue> {
-        prop_oneof![
-            any::<f64>().prop_map(IndexValue::F64),
-            any::<u64>().prop_map(IndexValue::U64),
-            any::<i64>().prop_map(IndexValue::I64),
-            any::<String>().prop_map(IndexValue::Str),
-        ]
-    }
 
     fn balanced_operation_strategy() -> impl Strategy<Value = IndexingOp> {
         prop_oneof![
             (0u64..20u64).prop_map(|id| IndexingOp::DeleteDoc { id }),
             (0u64..20u64).prop_map(|id| IndexingOp::DeleteDocQuery { id }),
-            (0u64..20u64, value_strategy())
-                .prop_map(move |(id, value)| IndexingOp::AddDoc { id, value }),
-            ((0u64..20u64), (1u64..100), value_strategy()).prop_map(
-                move |(id, num_docs, value)| {
-                    IndexingOp::AddMultipleDoc {
-                        id,
-                        num_docs,
-                        value,
-                    }
-                }
-            ),
+            (0u64..20u64).prop_map(|id| IndexingOp::AddDoc { id }),
             (0u64..1u64).prop_map(|_| IndexingOp::Commit),
             (0u64..1u64).prop_map(|_| IndexingOp::Merge),
         ]
@@ -1649,17 +1596,7 @@ mod tests {
         prop_oneof![
             5 => (0u64..100u64).prop_map(|id| IndexingOp::DeleteDoc { id }),
             5 => (0u64..100u64).prop_map(|id| IndexingOp::DeleteDocQuery { id }),
-            50 => (0u64..100u64, value_strategy())
-                .prop_map(move |(id, value)| IndexingOp::AddDoc { id, value }),
-            50 => (0u64..100u64, (1u64..100), value_strategy()).prop_map(
-                move |(id, num_docs, value)| {
-                    IndexingOp::AddMultipleDoc {
-                        id,
-                        num_docs,
-                        value,
-                    }
-                }
-            ),
+            50 => (0u64..100u64).prop_map(|id| IndexingOp::AddDoc { id }),
             2 => (0u64..1u64).prop_map(|_| IndexingOp::Commit),
             1 => (0u64..1u64).prop_map(|_| IndexingOp::Merge),
         ]
@@ -1668,27 +1605,19 @@ mod tests {
     fn expected_ids(ops: &[IndexingOp]) -> (HashMap<u64, u64>, HashSet<u64>) {
         let mut existing_ids = HashMap::new();
         let mut deleted_ids = HashSet::new();
-        for op in ops {
+        for &op in ops {
             match op {
-                IndexingOp::AddDoc { id, value: _ } => {
-                    *existing_ids.entry(*id).or_insert(0) += 1;
-                    deleted_ids.remove(id);
-                }
-                IndexingOp::AddMultipleDoc {
-                    id,
-                    num_docs,
-                    value: _,
-                } => {
-                    *existing_ids.entry(*id).or_insert(0) += num_docs;
-                    deleted_ids.remove(id);
+                IndexingOp::AddDoc { id } => {
+                    *existing_ids.entry(id).or_insert(0) += 1;
+                    deleted_ids.remove(&id);
                 }
                 IndexingOp::DeleteDoc { id } => {
                     existing_ids.remove(&id);
-                    deleted_ids.insert(*id);
+                    deleted_ids.insert(id);
                 }
                 IndexingOp::DeleteDocQuery { id } => {
                     existing_ids.remove(&id);
-                    deleted_ids.insert(*id);
+                    deleted_ids.insert(id);
                 }
                 _ => {}
             }
@@ -1698,19 +1627,16 @@ mod tests {
 
     fn get_id_list(ops: &[IndexingOp]) -> Vec<u64> {
         let mut id_list = Vec::new();
-        for op in ops {
+        for &op in ops {
             match op {
-                IndexingOp::AddDoc { id, value: _ } => {
-                    id_list.push(*id);
-                }
-                IndexingOp::AddMultipleDoc { id, .. } => {
-                    id_list.push(*id);
+                IndexingOp::AddDoc { id } => {
+                    id_list.push(id);
                 }
                 IndexingOp::DeleteDoc { id } => {
-                    id_list.retain(|el| el != id);
+                    id_list.retain(|el| *el != id);
                 }
                 IndexingOp::DeleteDocQuery { id } => {
-                    id_list.retain(|el| el != id);
+                    id_list.retain(|el| *el != id);
                 }
                 _ => {}
             }
@@ -1791,59 +1717,42 @@ mod tests {
 
         let ip_from_id = |id| Ipv6Addr::from_u128(id as u128);
 
-        let add_docs = |index_writer: &mut IndexWriter,
-                        id: u64,
-                        value: IndexValue,
-                        num: u64|
-         -> crate::Result<()> {
-            let facet = Facet::from(&("/cola/".to_string() + &id.to_string()));
-            let ip = ip_from_id(id);
-            let doc = if !id_is_full_doc(id) {
-                // every 3rd doc has no ip field
-                doc!(
-                    id_field=>id,
-                )
-            } else {
-                let json = json!({"date1": format!("2022-{id}-01T00:00:01Z"), "date2": format!("{id}-05-01T00:00:01Z"), "id": id, "ip": ip.to_string(), "val": value});
-                doc!(id_field=>id,
-                        json_field=>json,
-                        bytes_field => id.to_le_bytes().as_slice(),
-                        id_opt_field => id,
-                        ip_field => ip,
-                        ips_field => ip,
-                        ips_field => ip,
-                        multi_numbers=> id,
-                        multi_numbers => id,
-                        bool_field => (id % 2u64) != 0,
-                        i64_field => id as i64,
-                        f64_field => id as f64,
-                        date_field => DateTime::from_timestamp_secs(id as i64),
-                        multi_bools => (id % 2u64) != 0,
-                        multi_bools => (id % 2u64) == 0,
-                        text_field => id.to_string(),
-                        facet_field => facet,
-                        large_text_field => LOREM,
-                        multi_text_fields => multi_text_field_text1,
-                        multi_text_fields => multi_text_field_text2,
-                        multi_text_fields => multi_text_field_text3,
-                )
-            };
-            for _ in 0..num {
-                index_writer.add_document(doc.clone())?;
-            }
-            Ok(())
-        };
-        for op in ops {
-            match op.clone() {
-                IndexingOp::AddMultipleDoc {
-                    id,
-                    num_docs,
-                    value,
-                } => {
-                    add_docs(&mut index_writer, id, value, num_docs)?;
-                }
-                IndexingOp::AddDoc { id, value } => {
-                    add_docs(&mut index_writer, id, value, 1)?;
+        for &op in ops {
+            match op {
+                IndexingOp::AddDoc { id } => {
+                    let facet = Facet::from(&("/cola/".to_string() + &id.to_string()));
+                    let ip = ip_from_id(id);
+
+                    if !id_is_full_doc(id) {
+                        // every 3rd doc has no ip field
+                        index_writer.add_document(doc!(
+                            id_field=>id,
+                        ))?;
+                    } else {
+                        let json = json!({"date1": format!("2022-{id}-01T00:00:01Z"), "date2": format!("{id}-05-01T00:00:01Z"), "id": id, "ip": ip.to_string()});
+                        index_writer.add_document(doc!(id_field=>id,
+                                json_field=>json,
+                                bytes_field => id.to_le_bytes().as_slice(),
+                                id_opt_field => id,
+                                ip_field => ip,
+                                ips_field => ip,
+                                ips_field => ip,
+                                multi_numbers=> id,
+                                multi_numbers => id,
+                                bool_field => (id % 2u64) != 0,
+                                i64_field => id as i64,
+                                f64_field => id as f64,
+                                date_field => DateTime::from_timestamp_secs(id as i64),
+                                multi_bools => (id % 2u64) != 0,
+                                multi_bools => (id % 2u64) == 0,
+                                text_field => id.to_string(),
+                                facet_field => facet,
+                                large_text_field => LOREM,
+                                multi_text_fields => multi_text_field_text1,
+                                multi_text_fields => multi_text_field_text2,
+                                multi_text_fields => multi_text_field_text3,
+                        ))?;
+                    }
                 }
                 IndexingOp::DeleteDoc { id } => {
                     index_writer.delete_term(Term::from_field_u64(id_field, id));
@@ -2071,13 +1980,7 @@ mod tests {
                 .unwrap();
             // test store iterator
             for doc in store_reader.iter::<TantivyDocument>(segment_reader.alive_bitset()) {
-                let id = doc
-                    .unwrap()
-                    .get_first(id_field)
-                    .unwrap()
-                    .as_value()
-                    .as_u64()
-                    .unwrap();
+                let id = doc.unwrap().get_first(id_field).unwrap().as_u64().unwrap();
                 assert!(expected_ids_and_num_occurrences.contains_key(&id));
             }
             // test store random access
@@ -2110,7 +2013,7 @@ mod tests {
                     let mut bool2 = doc.get_all(multi_bools);
                     assert_eq!(bool, bool2.next().unwrap().as_bool().unwrap());
                     assert_ne!(bool, bool2.next().unwrap().as_bool().unwrap());
-                    assert!(bool2.next().is_none())
+                    assert_eq!(None, bool2.next())
                 }
             }
         }
@@ -2124,22 +2027,18 @@ mod tests {
 
             top_docs.iter().map(|el| el.1).collect::<Vec<_>>()
         };
-        let count_search = |term: &str, field| {
-            let query = QueryParser::for_index(&index, vec![field])
-                .parse_query(term)
-                .unwrap();
-            searcher.search(&query, &Count).unwrap()
-        };
 
-        let count_search2 = |term: Term| {
+        let do_search2 = |term: Term| {
             let query = TermQuery::new(term, IndexRecordOption::Basic);
-            searcher.search(&query, &Count).unwrap()
+            let top_docs: Vec<(f32, DocAddress)> =
+                searcher.search(&query, &TopDocs::with_limit(1000)).unwrap();
+
+            top_docs.iter().map(|el| el.1).collect::<Vec<_>>()
         };
 
         for (id, count) in &expected_ids_and_num_occurrences {
-            // skip expensive queries
             let (existing_id, count) = (*id, *count);
-            let get_num_hits = |field| count_search(&existing_id.to_string(), field) as u64;
+            let get_num_hits = |field| do_search(&existing_id.to_string(), field).len() as u64;
             assert_eq!(get_num_hits(id_field), count);
             if !id_is_full_doc(existing_id) {
                 continue;
@@ -2149,31 +2048,29 @@ mod tests {
             assert_eq!(get_num_hits(f64_field), count);
 
             // Test multi text
-            if num_docs_with_values < 1000 {
-                assert_eq!(
-                    do_search("\"test1 test2\"", multi_text_fields).len(),
-                    num_docs_with_values
-                );
-                assert_eq!(
-                    do_search("\"test2 test3\"", multi_text_fields).len(),
-                    num_docs_with_values
-                );
-            }
+            assert_eq!(
+                do_search("\"test1 test2\"", multi_text_fields).len(),
+                num_docs_with_values
+            );
+            assert_eq!(
+                do_search("\"test2 test3\"", multi_text_fields).len(),
+                num_docs_with_values
+            );
 
             // Test bytes
             let term = Term::from_field_bytes(bytes_field, existing_id.to_le_bytes().as_slice());
-            assert_eq!(count_search2(term) as u64, count);
+            assert_eq!(do_search2(term).len() as u64, count);
 
             // Test date
             let term = Term::from_field_date(
                 date_field,
                 DateTime::from_timestamp_secs(existing_id as i64),
             );
-            assert_eq!(count_search2(term) as u64, count);
+            assert_eq!(do_search2(term).len() as u64, count);
         }
         for deleted_id in deleted_ids {
             let assert_field = |field| {
-                assert_eq!(count_search(&deleted_id.to_string(), field) as u64, 0);
+                assert_eq!(do_search(&deleted_id.to_string(), field).len() as u64, 0);
             };
             assert_field(text_field);
             assert_field(f64_field);
@@ -2182,12 +2079,12 @@ mod tests {
 
             // Test bytes
             let term = Term::from_field_bytes(bytes_field, deleted_id.to_le_bytes().as_slice());
-            assert_eq!(count_search2(term), 0);
+            assert_eq!(do_search2(term).len() as u64, 0);
 
             // Test date
             let term =
                 Term::from_field_date(date_field, DateTime::from_timestamp_secs(deleted_id as i64));
-            assert_eq!(count_search2(term), 0);
+            assert_eq!(do_search2(term).len() as u64, 0);
         }
         // search ip address
         //
@@ -2196,13 +2093,13 @@ mod tests {
             if !id_is_full_doc(existing_id) {
                 continue;
             }
-            let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+            let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
             let ip_addr = Ipv6Addr::from_u128(existing_id as u128);
             // Test incoming ip as ipv6
             assert_eq!(do_search_ip_field(&format!("\"{ip_addr}\"")), count);
 
             let term = Term::from_field_ip_addr(ip_field, ip_addr);
-            assert_eq!(count_search2(term) as u64, count);
+            assert_eq!(do_search2(term).len() as u64, count);
 
             // Test incoming ip as ipv4
             if let Some(ip_addr) = ip_addr.to_ipv4_mapped() {
@@ -2219,7 +2116,7 @@ mod tests {
         if !sample.is_empty() {
             let (left_sample, right_sample) = sample.split_at(sample.len() / 2);
 
-            let calc_expected_count = |sample: &[(&u64, &u64)]| {
+            let expected_count = |sample: &[(&u64, &u64)]| {
                 sample
                     .iter()
                     .filter(|(id, _)| id_is_full_doc(**id))
@@ -2235,17 +2132,18 @@ mod tests {
             }
 
             // Query first half
-            let expected_count = calc_expected_count(left_sample);
-            if !left_sample.is_empty() && expected_count < 1000 {
+            if !left_sample.is_empty() {
+                let expected_count = expected_count(left_sample);
+
                 let start_range = *left_sample[0].0;
                 let end_range = *left_sample.last().unwrap().0;
                 let query = gen_query_inclusive("id_opt", start_range, end_range);
-                assert_eq!(count_search(&query, id_opt_field) as u64, expected_count);
+                assert_eq!(do_search(&query, id_opt_field).len() as u64, expected_count);
 
                 // Range query on ip field
                 let ip1 = ip_from_id(start_range);
                 let ip2 = ip_from_id(end_range);
-                let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+                let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
                 let query = gen_query_inclusive("ip", ip1, ip2);
                 assert_eq!(do_search_ip_field(&query), expected_count);
                 let query = gen_query_inclusive("ip", "*", ip2);
@@ -2257,19 +2155,19 @@ mod tests {
                 assert_eq!(do_search_ip_field(&query), expected_count);
             }
             // Query second half
-            let expected_count = calc_expected_count(right_sample);
-            if !right_sample.is_empty() && expected_count < 1000 {
+            if !right_sample.is_empty() {
+                let expected_count = expected_count(right_sample);
                 let start_range = *right_sample[0].0;
                 let end_range = *right_sample.last().unwrap().0;
                 // Range query on id opt field
                 let query =
                     gen_query_inclusive("id_opt", start_range.to_string(), end_range.to_string());
-                assert_eq!(count_search(&query, id_opt_field) as u64, expected_count);
+                assert_eq!(do_search(&query, id_opt_field).len() as u64, expected_count);
 
                 // Range query on ip field
                 let ip1 = ip_from_id(start_range);
                 let ip2 = ip_from_id(end_range);
-                let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+                let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
                 let query = gen_query_inclusive("ip", ip1, ip2);
                 assert_eq!(do_search_ip_field(&query), expected_count);
                 let query = gen_query_inclusive("ip", ip1, "*");
@@ -2294,7 +2192,7 @@ mod tests {
             };
             let ip = ip_from_id(existing_id);
 
-            let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+            let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
             // Range query on single value field
             let query = gen_query_inclusive("ip", ip, ip);
             assert_eq!(do_search_ip_field(&query), count);
@@ -2354,7 +2252,7 @@ mod tests {
 
     #[test]
     fn test_fast_field_range() {
-        let ops: Vec<_> = (0..1000).map(|id| IndexingOp::add(id)).collect();
+        let ops: Vec<_> = (0..1000).map(|id| IndexingOp::AddDoc { id }).collect();
         assert!(test_operation_strategy(&ops, false, true).is_ok());
     }
 
@@ -2362,8 +2260,8 @@ mod tests {
     fn test_sort_index_on_opt_field_regression() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(81),
-                IndexingOp::add(70),
+                IndexingOp::AddDoc { id: 81 },
+                IndexingOp::AddDoc { id: 70 },
                 IndexingOp::DeleteDoc { id: 70 }
             ],
             true,
@@ -2372,45 +2270,14 @@ mod tests {
         .is_ok());
     }
 
-    #[test]
-    fn test_simple_multiple_doc() {
-        assert!(test_operation_strategy(
-            &[
-                IndexingOp::AddMultipleDoc {
-                    id: 7,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-                IndexingOp::AddMultipleDoc {
-                    id: 92,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-                IndexingOp::AddMultipleDoc {
-                    id: 30,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-                IndexingOp::AddMultipleDoc {
-                    id: 33,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-            ],
-            true,
-            false
-        )
-        .is_ok());
-    }
-
     #[test]
     fn test_ip_range_query_multivalue_bug() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(2),
+                IndexingOp::AddDoc { id: 2 },
                 IndexingOp::Commit,
-                IndexingOp::add(1),
-                IndexingOp::add(1),
+                IndexingOp::AddDoc { id: 1 },
+                IndexingOp::AddDoc { id: 1 },
                 IndexingOp::Commit,
                 IndexingOp::Merge
             ],
@@ -2424,11 +2291,11 @@ mod tests {
     fn test_ff_num_ips_regression() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(13),
-                IndexingOp::add(1),
+                IndexingOp::AddDoc { id: 13 },
+                IndexingOp::AddDoc { id: 1 },
                 IndexingOp::Commit,
                 IndexingOp::DeleteDocQuery { id: 13 },
-                IndexingOp::add(1),
+                IndexingOp::AddDoc { id: 1 },
                 IndexingOp::Commit,
             ],
             false,
@@ -2440,7 +2307,7 @@ mod tests {
     #[test]
     fn test_minimal_sort_force_end_merge() {
         assert!(test_operation_strategy(
-            &[IndexingOp::add(23), IndexingOp::add(13),],
+            &[IndexingOp::AddDoc { id: 23 }, IndexingOp::AddDoc { id: 13 },],
             false,
             false
         )
@@ -2501,8 +2368,8 @@ mod tests {
     fn test_minimal_sort_force_end_merge_with_delete() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(23),
-                IndexingOp::add(13),
+                IndexingOp::AddDoc { id: 23 },
+                IndexingOp::AddDoc { id: 13 },
                 IndexingOp::DeleteDoc { id: 13 }
             ],
             true,
@@ -2515,8 +2382,8 @@ mod tests {
     fn test_minimal_no_sort_no_force_end_merge() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(23),
-                IndexingOp::add(13),
+                IndexingOp::AddDoc { id: 23 },
+                IndexingOp::AddDoc { id: 13 },
                 IndexingOp::DeleteDoc { id: 13 }
             ],
             false,
@@ -2527,7 +2394,7 @@ mod tests {
 
     #[test]
     fn test_minimal_sort_merge() {
-        assert!(test_operation_strategy(&[IndexingOp::add(3),], true, true).is_ok());
+        assert!(test_operation_strategy(&[IndexingOp::AddDoc { id: 3 },], true, true).is_ok());
     }
 
     use proptest::prelude::*;
@@ -2623,14 +2490,14 @@ mod tests {
     fn test_delete_bug_reproduction_ip_addr() {
         use IndexingOp::*;
         let ops = &[
-            IndexingOp::add(1),
-            IndexingOp::add(2),
+            AddDoc { id: 1 },
+            AddDoc { id: 2 },
             Commit,
-            IndexingOp::add(3),
+            AddDoc { id: 3 },
             DeleteDoc { id: 1 },
             Commit,
             Merge,
-            IndexingOp::add(4),
+            AddDoc { id: 4 },
             Commit,
         ];
         test_operation_strategy(&ops[..], false, true).unwrap();
@@ -2639,13 +2506,7 @@ mod tests {
     #[test]
     fn test_merge_regression_1() {
         use IndexingOp::*;
-        let ops = &[
-            IndexingOp::add(15),
-            Commit,
-            IndexingOp::add(9),
-            Commit,
-            Merge,
-        ];
+        let ops = &[AddDoc { id: 15 }, Commit, AddDoc { id: 9 }, Commit, Merge];
         test_operation_strategy(&ops[..], false, true).unwrap();
     }
 
@@ -2653,9 +2514,9 @@ mod tests {
     fn test_range_query_bug_1() {
         use IndexingOp::*;
         let ops = &[
-            IndexingOp::add(9),
-            IndexingOp::add(0),
-            IndexingOp::add(13),
+            AddDoc { id: 9 },
+            AddDoc { id: 0 },
+            AddDoc { id: 13 },
             Commit,
         ];
         test_operation_strategy(&ops[..], false, true).unwrap();
@@ -2663,11 +2524,12 @@ mod tests {
 
     #[test]
     fn test_range_query_bug_2() {
+        use IndexingOp::*;
         let ops = &[
-            IndexingOp::add(3),
-            IndexingOp::add(6),
-            IndexingOp::add(9),
-            IndexingOp::add(10),
+            AddDoc { id: 3 },
+            AddDoc { id: 6 },
+            AddDoc { id: 9 },
+            AddDoc { id: 10 },
         ];
         test_operation_strategy(&ops[..], false, false).unwrap();
     }
@@ -2689,7 +2551,7 @@ mod tests {
         assert!(test_operation_strategy(
             &[
                 IndexingOp::DeleteDoc { id: 0 },
-                IndexingOp::add(6),
+                IndexingOp::AddDoc { id: 6 },
                 IndexingOp::DeleteDocQuery { id: 11 },
                 IndexingOp::Commit,
                 IndexingOp::Merge,
@@ -2706,13 +2568,10 @@ mod tests {
     fn test_bug_1617_2() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::AddDoc {
-                    id: 13,
-                    value: Default::default()
-                },
+                IndexingOp::AddDoc { id: 13 },
                 IndexingOp::DeleteDoc { id: 13 },
                 IndexingOp::Commit,
-                IndexingOp::add(30),
+                IndexingOp::AddDoc { id: 30 },
                 IndexingOp::Commit,
                 IndexingOp::Merge,
             ],

src/indexer/log_merge_policy.rs

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

src/indexer/merge_operation.rs

@@ -1,8 +1,7 @@
 use std::collections::HashSet;
 use std::ops::Deref;
 
-use crate::index::SegmentId;
-use crate::{Inventory, Opstamp, TrackedObject};
+use crate::{Inventory, Opstamp, SegmentId, 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, SegmentComponent, SegmentReader};
+use crate::index::{Segment, SegmentReader};
 use crate::indexer::doc_id_mapping::{MappingType, SegmentDocIdMapping};
 use crate::indexer::SegmentSerializer;
 use crate::postings::{InvertedIndexSerializer, Postings, SegmentPostings};
@@ -21,7 +21,8 @@ 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, SegmentOrdinal,
+    DocAddress, DocId, IndexSettings, IndexSortByField, InvertedIndexReader, Order,
+    SegmentComponent, SegmentOrdinal,
 };
 
 /// Segment's max doc must be `< MAX_DOC_LIMIT`.
@@ -787,25 +788,23 @@ impl IndexMerger {
 mod tests {
 
     use columnar::Column;
-    use proptest::prop_oneof;
-    use proptest::strategy::Strategy;
     use schema::FAST;
 
     use crate::collector::tests::{
         BytesFastFieldTestCollector, FastFieldTestCollector, TEST_COLLECTOR_WITH_SCORE,
     };
     use crate::collector::{Count, FacetCollector};
-    use crate::index::{Index, SegmentId};
-    use crate::indexer::NoMergePolicy;
+    use crate::index::Index;
     use crate::query::{AllQuery, BooleanQuery, EnableScoring, Scorer, TermQuery};
+    use crate::schema::document::Value;
     use crate::schema::{
         Facet, FacetOptions, IndexRecordOption, NumericOptions, TantivyDocument, Term,
-        TextFieldIndexing, Value, INDEXED, TEXT,
+        TextFieldIndexing, INDEXED, TEXT,
     };
     use crate::time::OffsetDateTime;
     use crate::{
         assert_nearly_equals, schema, DateTime, DocAddress, DocId, DocSet, IndexSettings,
-        IndexSortByField, IndexWriter, Order, Searcher,
+        IndexSortByField, IndexWriter, Order, Searcher, SegmentId,
     };
 
     #[test]
@@ -912,24 +911,15 @@ mod tests {
             }
             {
                 let doc = searcher.doc::<TantivyDocument>(DocAddress::new(0, 0))?;
-                assert_eq!(
-                    doc.get_first(text_field).unwrap().as_value().as_str(),
-                    Some("af b")
-                );
+                assert_eq!(doc.get_first(text_field).unwrap().as_str(), Some("af b"));
             }
             {
                 let doc = searcher.doc::<TantivyDocument>(DocAddress::new(0, 1))?;
-                assert_eq!(
-                    doc.get_first(text_field).unwrap().as_value().as_str(),
-                    Some("a b c")
-                );
+                assert_eq!(doc.get_first(text_field).unwrap().as_str(), Some("a b c"));
             }
             {
                 let doc = searcher.doc::<TantivyDocument>(DocAddress::new(0, 2))?;
-                assert_eq!(
-                    doc.get_first(text_field).unwrap().as_value().as_str(),
-                    Some("a b c d")
-                );
+                assert_eq!(doc.get_first(text_field).unwrap().as_str(), Some("a b c d"));
             }
             {
                 let doc = searcher.doc::<TantivyDocument>(DocAddress::new(0, 3))?;
@@ -1534,112 +1524,6 @@ mod tests {
         Ok(())
     }
 
-    #[derive(Debug, Clone, Copy, Eq, PartialEq)]
-    enum IndexingOp {
-        ZeroVal,
-        OneVal { val: u64 },
-        TwoVal { val: u64 },
-        Commit,
-    }
-
-    fn balanced_operation_strategy() -> impl Strategy<Value = IndexingOp> {
-        prop_oneof![
-            (0u64..1u64).prop_map(|_| IndexingOp::ZeroVal),
-            (0u64..1u64).prop_map(|val| IndexingOp::OneVal { val }),
-            (0u64..1u64).prop_map(|val| IndexingOp::TwoVal { val }),
-            (0u64..1u64).prop_map(|_| IndexingOp::Commit),
-        ]
-    }
-
-    use proptest::prelude::*;
-    proptest! {
-        #[test]
-        fn test_merge_columnar_int_proptest(ops in proptest::collection::vec(balanced_operation_strategy(), 1..20)) {
-            assert!(test_merge_int_fields(&ops[..]).is_ok());
-        }
-    }
-    fn test_merge_int_fields(ops: &[IndexingOp]) -> crate::Result<()> {
-        if ops.iter().all(|op| *op == IndexingOp::Commit) {
-            return Ok(());
-        }
-        let expected_doc_and_vals: Vec<(u32, Vec<u64>)> = ops
-            .iter()
-            .filter(|op| *op != &IndexingOp::Commit)
-            .map(|op| match op {
-                IndexingOp::ZeroVal => vec![],
-                IndexingOp::OneVal { val } => vec![*val],
-                IndexingOp::TwoVal { val } => vec![*val, *val],
-                IndexingOp::Commit => unreachable!(),
-            })
-            .enumerate()
-            .map(|(id, val)| (id as u32, val))
-            .collect();
-
-        let mut schema_builder = schema::Schema::builder();
-        let int_options = NumericOptions::default().set_fast().set_indexed();
-        let int_field = schema_builder.add_u64_field("intvals", int_options);
-        let index = Index::create_in_ram(schema_builder.build());
-        {
-            let mut index_writer = index.writer_for_tests()?;
-            index_writer.set_merge_policy(Box::new(NoMergePolicy));
-            let index_doc = |index_writer: &mut IndexWriter, int_vals: &[u64]| {
-                let mut doc = TantivyDocument::default();
-                for &val in int_vals {
-                    doc.add_u64(int_field, val);
-                }
-                index_writer.add_document(doc).unwrap();
-            };
-
-            for op in ops {
-                match op {
-                    IndexingOp::ZeroVal => index_doc(&mut index_writer, &[]),
-                    IndexingOp::OneVal { val } => index_doc(&mut index_writer, &[*val]),
-                    IndexingOp::TwoVal { val } => index_doc(&mut index_writer, &[*val, *val]),
-                    IndexingOp::Commit => {
-                        index_writer.commit().expect("commit failed");
-                    }
-                }
-            }
-            index_writer.commit().expect("commit failed");
-        }
-        {
-            let mut segment_ids = index.searchable_segment_ids()?;
-            segment_ids.sort();
-            let mut index_writer: IndexWriter = index.writer_for_tests()?;
-            index_writer.merge(&segment_ids).wait()?;
-            index_writer.wait_merging_threads()?;
-        }
-        let reader = index.reader()?;
-        reader.reload()?;
-
-        let mut vals: Vec<u64> = Vec::new();
-        let mut test_vals = move |col: &Column<u64>, doc: DocId, expected: &[u64]| {
-            vals.clear();
-            vals.extend(col.values_for_doc(doc));
-            assert_eq!(&vals[..], expected);
-        };
-
-        let mut test_col = move |col: &Column<u64>, column_expected: &[(u32, Vec<u64>)]| {
-            for (doc_id, vals) in column_expected.iter() {
-                test_vals(col, *doc_id, vals);
-            }
-        };
-
-        {
-            let searcher = reader.searcher();
-            let segment = searcher.segment_reader(0u32);
-            let col = segment
-                .fast_fields()
-                .column_opt::<u64>("intvals")
-                .unwrap()
-                .unwrap();
-
-            test_col(&col, &expected_doc_and_vals);
-        }
-
-        Ok(())
-    }
-
     #[test]
     fn test_merge_multivalued_int_fields_simple() -> crate::Result<()> {
         let mut schema_builder = schema::Schema::builder();

src/indexer/merger_sorted_index_test.rs

@@ -3,15 +3,15 @@ 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::{
         self, BytesOptions, Facet, FacetOptions, IndexRecordOption, NumericOptions,
-        TextFieldIndexing, TextOptions, Value,
+        TextFieldIndexing, TextOptions,
     };
     use crate::{
-        DocAddress, DocSet, IndexSettings, IndexSortByField, IndexWriter, Order, TantivyDocument,
-        Term,
+        DocAddress, DocSet, IndexSettings, IndexSortByField, IndexWriter, Order, Postings,
+        TantivyDocument, Term,
     };
 
     fn create_test_index_posting_list_issue(index_settings: Option<IndexSettings>) -> Index {
@@ -280,16 +280,13 @@ mod tests {
                 .doc::<TantivyDocument>(DocAddress::new(0, blubber_pos))
                 .unwrap();
             assert_eq!(
-                doc.get_first(my_text_field).unwrap().as_value().as_str(),
+                doc.get_first(my_text_field).unwrap().as_str(),
                 Some("blubber")
             );
             let doc = searcher
                 .doc::<TantivyDocument>(DocAddress::new(0, 0))
                 .unwrap();
-            assert_eq!(
-                doc.get_first(int_field).unwrap().as_value().as_u64(),
-                Some(1000)
-            );
+            assert_eq!(doc.get_first(int_field).unwrap().as_u64(), Some(1000));
         }
     }
 

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", format!("num{field_name_in}"): 10})))
+            .add_document(doc!(field=>json!({format!("{field_name_in}"): "test1"})))
             .unwrap();
         index_writer
             .add_document(doc!(field=>json!({format!("a{field_name_in}"): "test2"})))
@@ -216,7 +216,7 @@ mod tests_mmap {
         let test_query = |query_str: &str| {
             let query = parse_query.parse_query(query_str).unwrap();
             let num_docs = searcher.search(&query, &Count).unwrap();
-            assert_eq!(num_docs, 1, "{query_str}");
+            assert_eq!(num_docs, 1, "{}", query_str);
         };
         test_query(format!("json.{field_name_out}:test1").as_str());
         test_query(format!("json.a{field_name_out}:test2").as_str());
@@ -260,64 +260,6 @@ 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]
@@ -590,10 +532,10 @@ mod tests_mmap {
         let query_parser = QueryParser::for_index(&index, vec![]);
         // Test if field name can be queried
         for (indexed_field, val) in fields_and_vals.iter() {
-            let query_str = &format!("{indexed_field}:{val}");
+            let query_str = &format!("{}:{}", indexed_field, val);
             let query = query_parser.parse_query(query_str).unwrap();
             let count_docs = searcher.search(&*query, &TopDocs::with_limit(2)).unwrap();
-            assert!(!count_docs.is_empty(), "{indexed_field}:{val}");
+            assert!(!count_docs.is_empty(), "{}:{}", indexed_field, val);
         }
         // Test if field name can be used for aggregation
         for (field_name, val) in fields_and_vals.iter() {

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, SegmentComponent};
+use crate::index::Segment;
 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, Value};
+use crate::schema::document::{Document, ReferenceValue, 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, TantivyError};
+use crate::{DocId, Opstamp, SegmentComponent, TantivyError};
 
 /// Computes the initial size of the hash table.
 ///
@@ -68,7 +68,6 @@ 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,
@@ -120,7 +119,6 @@ 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,
@@ -202,10 +200,12 @@ impl SegmentWriter {
             match field_entry.field_type() {
                 FieldType::Facet(_) => {
                     let mut facet_tokenizer = FacetTokenizer::default(); // this can be global
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
-                        let facet_str = value.as_facet().ok_or_else(make_schema_error)?;
+                        let facet = value.as_facet().ok_or_else(make_schema_error)?;
+                        let facet_str = facet.encoded_str();
                         let mut facet_tokenizer = facet_tokenizer.token_stream(facet_str);
                         let mut indexing_position = IndexingPosition::default();
                         postings_writer.index_text(
@@ -219,15 +219,16 @@ impl SegmentWriter {
                 }
                 FieldType::Str(_) => {
                     let mut indexing_position = IndexingPosition::default();
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         let mut token_stream = if let Some(text) = value.as_str() {
                             let text_analyzer =
                                 &mut self.per_field_text_analyzers[field.field_id() as usize];
                             text_analyzer.token_stream(text)
-                        } else if let Some(tok_str) = value.into_pre_tokenized_text() {
-                            BoxTokenStream::new(PreTokenizedStream::from(*tok_str.clone()))
+                        } else if let Some(tok_str) = value.as_pre_tokenized_text() {
+                            BoxTokenStream::new(PreTokenizedStream::from(tok_str.clone()))
                         } else {
                             continue;
                         };
@@ -248,8 +249,9 @@ impl SegmentWriter {
                 }
                 FieldType::U64(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         num_vals += 1;
                         let u64_val = value.as_u64().ok_or_else(make_schema_error)?;
@@ -262,8 +264,10 @@ impl SegmentWriter {
                 }
                 FieldType::Date(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value_access = value_access as D::Value<'_>;
+                        let value = value_access.as_value();
 
                         num_vals += 1;
                         let date_val = value.as_datetime().ok_or_else(make_schema_error)?;
@@ -277,8 +281,9 @@ impl SegmentWriter {
                 }
                 FieldType::I64(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         num_vals += 1;
                         let i64_val = value.as_i64().ok_or_else(make_schema_error)?;
@@ -291,8 +296,10 @@ impl SegmentWriter {
                 }
                 FieldType::F64(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
+
                         num_vals += 1;
                         let f64_val = value.as_f64().ok_or_else(make_schema_error)?;
                         term_buffer.set_f64(f64_val);
@@ -304,8 +311,10 @@ impl SegmentWriter {
                 }
                 FieldType::Bool(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
+
                         num_vals += 1;
                         let bool_val = value.as_bool().ok_or_else(make_schema_error)?;
                         term_buffer.set_bool(bool_val);
@@ -317,8 +326,10 @@ impl SegmentWriter {
                 }
                 FieldType::Bytes(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
+
                         num_vals += 1;
                         let bytes = value.as_bytes().ok_or_else(make_schema_error)?;
                         term_buffer.set_bytes(bytes);
@@ -331,29 +342,32 @@ 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();
 
-                    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,
-                        );
-                    }
+                        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,
+                    )?;
                 }
                 FieldType::IpAddr(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         num_vals += 1;
                         let ip_addr = value.as_ip_addr().ok_or_else(make_schema_error)?;
@@ -484,19 +498,19 @@ mod tests {
     use crate::collector::{Count, TopDocs};
     use crate::directory::RamDirectory;
     use crate::fastfield::FastValue;
-    use crate::postings::{Postings, TermInfo};
+    use crate::postings::TermInfo;
     use crate::query::{PhraseQuery, QueryParser};
+    use crate::schema::document::Value;
     use crate::schema::{
-        Document, IndexRecordOption, OwnedValue, Schema, TextFieldIndexing, TextOptions, Value,
-        STORED, STRING, TEXT,
+        Document, IndexRecordOption, 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, TantivyDocument, Term,
-        TERMINATED,
+        DateTime, Directory, DocAddress, DocSet, Index, IndexWriter, Postings, TantivyDocument,
+        Term, TERMINATED,
     };
 
     #[test]
@@ -541,15 +555,9 @@ mod tests {
         let reader = StoreReader::open(directory.open_read(path).unwrap(), 0).unwrap();
         let doc = reader.get::<TantivyDocument>(0).unwrap();
 
-        assert_eq!(doc.field_values().count(), 2);
-        assert_eq!(
-            doc.get_all(text_field).next().unwrap().as_value().as_str(),
-            Some("A")
-        );
-        assert_eq!(
-            doc.get_all(text_field).nth(1).unwrap().as_value().as_str(),
-            Some("title")
-        );
+        assert_eq!(doc.field_values().len(), 2);
+        assert_eq!(doc.field_values()[0].value().as_str(), Some("A"));
+        assert_eq!(doc.field_values()[1].value().as_str(), Some("title"));
     }
     #[test]
     fn test_simple_json_indexing() {
@@ -589,51 +597,12 @@ 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();
         let json_field = schema_builder.add_json_field("json", STORED | TEXT);
         let schema = schema_builder.build();
-        let json_val: serde_json::Value = serde_json::from_str(
+        let json_val: serde_json::Map<String, serde_json::Value> = serde_json::from_str(
             r#"{
             "toto": "titi",
             "float": -0.2,
@@ -661,10 +630,14 @@ mod tests {
                 doc_id: 0u32,
             })
             .unwrap();
-        let serdeser_json_val = serde_json::from_str::<serde_json::Value>(&doc.to_json(&schema))
+        let serdeser_json_val = serde_json::from_str::<serde_json::Map<String, serde_json::Value>>(
+            &doc.to_json(&schema),
+        )
+        .unwrap()
+        .get("json")
+        .unwrap()[0]
+            .as_object()
             .unwrap()
-            .get("json")
-            .unwrap()[0]
             .clone();
         assert_eq!(json_val, serdeser_json_val);
         let segment_reader = searcher.segment_reader(0u32);
@@ -828,7 +801,7 @@ mod tests {
         let mut schema_builder = Schema::builder();
         let json_field = schema_builder.add_json_field("json", STRING);
         let schema = schema_builder.build();
-        let json_val: serde_json::Value =
+        let json_val: serde_json::Map<String, serde_json::Value> =
             serde_json::from_str(r#"{"mykey": "two tokens"}"#).unwrap();
         let doc = doc!(json_field=>json_val);
         let index = Index::create_in_ram(schema);
@@ -868,7 +841,7 @@ mod tests {
         let mut schema_builder = Schema::builder();
         let json_field = schema_builder.add_json_field("json", TEXT);
         let schema = schema_builder.build();
-        let json_val: serde_json::Value = serde_json::from_str(
+        let json_val: serde_json::Map<String, serde_json::Value> = serde_json::from_str(
             r#"{"mykey": [{"field": "hello happy tax payer"}, {"field": "nothello"}]}"#,
         )
         .unwrap();

src/lib.rs

@@ -216,6 +216,11 @@ 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};
@@ -223,10 +228,16 @@ pub use crate::directory::Directory;
 #[allow(deprecated)] // Remove with index sorting
 pub use crate::index::{
     Index, IndexBuilder, IndexMeta, IndexSettings, IndexSortByField, InvertedIndexReader, Order,
-    Segment, SegmentMeta, SegmentReader,
+    Segment, SegmentComponent, SegmentId, 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::schema::{Document, TantivyDocument, Term};
+pub use crate::postings::Postings;
+pub use crate::schema::{DateOptions, DateTimePrecision, Document, TantivyDocument, Term};
 
 /// Index format version.
 const INDEX_FORMAT_VERSION: u32 = 6;
@@ -381,10 +392,9 @@ 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, ReloadPolicy};
+    use crate::{DateTime, DocAddress, Index, IndexWriter, Postings, ReloadPolicy};
 
     pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
         let mut buffer = Vec::new();
@@ -397,20 +407,16 @@ pub mod tests {
     #[macro_export]
     macro_rules! assert_nearly_equals {
         ($left:expr, $right:expr) => {{
-            assert_nearly_equals!($left, $right, 0.0005);
-        }};
-        ($left:expr, $right:expr, $epsilon:expr) => {{
-            match (&$left, &$right, &$epsilon) {
-                (left_val, right_val, epsilon_val) => {
+            match (&$left, &$right) {
+                (left_val, right_val) => {
                     let diff = (left_val - right_val).abs();
-
-                    if diff > *epsilon_val {
+                    let add = left_val.abs() + right_val.abs();
+                    if diff > 0.0005 * add {
                         panic!(
-                            r#"assertion failed: `abs(left-right)>epsilon`
-    left: `{:?}`,
-    right: `{:?}`,
-    epsilon: `{:?}`"#,
-                            &*left_val, &*right_val, &*epsilon_val
+                            r#"assertion failed: `(left ~= right)`
+  left: `{:?}`,
+ right: `{:?}`"#,
+                            &*left_val, &*right_val
                         )
                     }
                 }
@@ -440,6 +446,7 @@ pub mod tests {
     }
 
     #[test]
+    #[cfg(not(feature = "lz4"))]
     fn test_version_string() {
         use regex::Regex;
         let regex_ptn = Regex::new(
@@ -939,7 +946,7 @@ pub mod tests {
         let mut schema_builder = Schema::builder();
         let json_field = schema_builder.add_json_field("json", STORED | TEXT);
         let schema = schema_builder.build();
-        let json_val: serde_json::Value = serde_json::from_str(
+        let json_val: serde_json::Map<String, serde_json::Value> = serde_json::from_str(
             r#"{
             "signed": 2,
             "float": 2.0,
@@ -1029,16 +1036,13 @@ pub mod tests {
                             text_field => "some other value",
                             other_text_field => "short");
         assert_eq!(document.len(), 3);
-        let values: Vec<OwnedValue> = document.get_all(text_field).map(OwnedValue::from).collect();
+        let values: Vec<&OwnedValue> = document.get_all(text_field).collect();
         assert_eq!(values.len(), 2);
-        assert_eq!(values[0].as_ref().as_str(), Some("tantivy"));
-        assert_eq!(values[1].as_ref().as_str(), Some("some other value"));
-        let values: Vec<OwnedValue> = document
-            .get_all(other_text_field)
-            .map(OwnedValue::from)
-            .collect();
+        assert_eq!(values[0].as_str(), Some("tantivy"));
+        assert_eq!(values[1].as_str(), Some("some other value"));
+        let values: Vec<&OwnedValue> = document.get_all(other_text_field).collect();
         assert_eq!(values.len(), 1);
-        assert_eq!(values[0].as_ref().as_str(), Some("short"));
+        assert_eq!(values[0].as_str(), Some("short"));
     }
 
     #[test]
@@ -1105,9 +1109,9 @@ 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/macros.rs

@@ -41,7 +41,6 @@
 /// );
 /// # }
 /// ```
-
 #[macro_export]
 macro_rules! doc(
     () => {
@@ -53,7 +52,7 @@ macro_rules! doc(
         {
             let mut document = $crate::TantivyDocument::default();
             $(
-                document.add_field_value($field, &$value);
+                document.add_field_value($field, $value);
             )*
             document
         }

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::index::SegmentComponent::{Positions, Postings, Terms};
+        use crate::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

@@ -1,9 +1,8 @@
 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};
+use crate::{DocId, DocSet, Score, Searcher, SegmentReader};
 
 /// `EmptyQuery` is a dummy `Query` in which no document matches.
 ///

src/query/fuzzy_query.rs

@@ -138,7 +138,8 @@ impl FuzzyTermQuery {
                 if json_path_type != Type::Str {
                     return Err(InvalidArgument(format!(
                         "The fuzzy term query requires a string path type for a json term. Found \
-                         {json_path_type:?}"
+                         {:?}",
+                        json_path_type
                     )));
                 }
             }

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().ok_or_else(|| {
+                        value.as_facet().map(|f| f.encoded_str()).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/query_parser/logical_ast.rs

@@ -2,7 +2,7 @@ use std::fmt;
 use std::ops::Bound;
 
 use crate::query::Occur;
-use crate::schema::{Term, Type};
+use crate::schema::{Field, Term, Type};
 use crate::Score;
 
 #[derive(Clone)]
@@ -20,6 +20,8 @@ pub enum LogicalLiteral {
         upper: Bound<Term>,
     },
     Set {
+        field: Field,
+        value_type: Type,
         elements: Vec<Term>,
     },
     All,

src/query/query_parser/query_parser.rs

@@ -832,11 +832,17 @@ impl QueryParser {
                 let (field, json_path) = try_tuple!(self
                     .split_full_path(&full_path)
                     .ok_or_else(|| QueryParserError::FieldDoesNotExist(full_path.clone())));
+                let field_entry = self.schema.get_field_entry(field);
+                let value_type = field_entry.field_type().value_type();
                 let (elements, errors) = elements
                     .into_iter()
                     .map(|element| self.compute_boundary_term(field, json_path, &element))
                     .partition_result();
-                let logical_ast = LogicalAst::Leaf(Box::new(LogicalLiteral::Set { elements }));
+                let logical_ast = LogicalAst::Leaf(Box::new(LogicalLiteral::Set {
+                    elements,
+                    field,
+                    value_type,
+                }));
                 (Some(logical_ast), errors)
             }
             UserInputLeaf::Exists { .. } => (

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 {
-        self.column.num_docs()
+        0 // heuristic possible by checking number of hits when fetching a block
     }
 }
 

src/query/regex_query.rs

@@ -185,7 +185,7 @@ mod test {
             Err(crate::TantivyError::InvalidArgument(msg)) => {
                 assert!(msg.contains("error: unclosed group"))
             }
-            res => panic!("unexpected result: {res:?}"),
+            res => panic!("unexpected result: {:?}", res),
         }
     }
 }

src/query/term_query/term_scorer.rs

@@ -127,7 +127,6 @@ 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;
@@ -135,7 +134,8 @@ 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, Term, TERMINATED,
+        assert_nearly_equals, DocId, DocSet, Index, IndexWriter, Score, Searcher, SegmentId, Term,
+        TERMINATED,
     };
 
     #[test]

src/reader/warming.rs

@@ -179,10 +179,9 @@ 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};
+    use crate::{Index, IndexSettings, ReloadPolicy, Searcher, SegmentId};
 
     #[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.encoded_str()).into());
+        let result = serialize_value(ReferenceValueLeaf::Facet(&facet).into());
         let value = deserialize_value(result);
         assert_eq!(value, crate::schema::OwnedValue::Facet(facet));
 
@@ -881,8 +881,7 @@ mod tests {
             text: "hello, world".to_string(),
             tokens: vec![Token::default(), Token::default()],
         };
-        let result =
-            serialize_value(ReferenceValueLeaf::PreTokStr(pre_tok_str.clone().into()).into());
+        let result = serialize_value(ReferenceValueLeaf::PreTokStr(&pre_tok_str).into());
         let value = deserialize_value(result);
         assert_eq!(value, crate::schema::OwnedValue::PreTokStr(pre_tok_str));
     }

src/schema/document/default_document.rs

@@ -1,64 +1,93 @@
 use std::collections::{BTreeMap, HashMap, HashSet};
-use std::io::{self, Read, Write};
 use std::net::Ipv6Addr;
 
-use columnar::MonotonicallyMappableToU128;
-use common::{read_u32_vint_no_advance, serialize_vint_u32, BinarySerializable, DateTime, VInt};
+use common::DateTime;
 use serde_json::Map;
-pub use CompactDoc as TantivyDocument;
 
-use super::{ReferenceValue, ReferenceValueLeaf, Value};
 use crate::schema::document::{
     DeserializeError, Document, DocumentDeserialize, DocumentDeserializer,
 };
 use crate::schema::field_type::ValueParsingError;
-use crate::schema::{Facet, Field, NamedFieldDocument, OwnedValue, Schema};
+use crate::schema::field_value::FieldValueIter;
+use crate::schema::{Facet, Field, FieldValue, NamedFieldDocument, OwnedValue, Schema};
 use crate::tokenizer::PreTokenizedString;
 
-#[repr(packed)]
-#[derive(Debug, Clone)]
-/// A field value pair in the compact tantivy document
-struct FieldValueAddr {
-    pub field: u16,
-    pub value_addr: ValueAddr,
+/// TantivyDocument provides a default implementation of the `Document` trait.
+/// It is the object that can be indexed and then searched for.
+///
+/// Documents are fundamentally a collection of unordered couples `(field, value)`.
+/// In this list, one field may appear more than once.
+#[derive(Clone, Debug, serde::Serialize, serde::Deserialize, Default)]
+pub struct TantivyDocument {
+    field_values: Vec<FieldValue>,
 }
 
-#[derive(Debug, Clone)]
-/// The default document in tantivy. It encodes data in a compact form.
-pub struct CompactDoc {
-    /// `node_data` is a vec of bytes, where each value is serialized into bytes and stored. It
-    /// includes all the data of the document and also metadata like where the nodes are located
-    /// in an object or array.
-    pub node_data: Vec<u8>,
-    /// The root (Field, Value) pairs
-    field_values: Vec<FieldValueAddr>,
-}
+impl Document for TantivyDocument {
+    type Value<'a> = &'a OwnedValue;
+    type FieldsValuesIter<'a> = FieldValueIter<'a>;
 
-impl Default for CompactDoc {
-    fn default() -> Self {
-        Self::new()
+    fn iter_fields_and_values(&self) -> Self::FieldsValuesIter<'_> {
+        FieldValueIter(self.field_values.iter())
     }
 }
 
-impl CompactDoc {
-    /// Creates a new, empty document object
-    /// The reserved capacity is for the total serialized data
-    pub fn with_capacity(bytes: usize) -> CompactDoc {
-        CompactDoc {
-            node_data: Vec::with_capacity(bytes),
-            field_values: Vec::with_capacity(4),
+impl DocumentDeserialize for TantivyDocument {
+    fn deserialize<'de, D>(mut deserializer: D) -> Result<Self, DeserializeError>
+    where D: DocumentDeserializer<'de> {
+        let mut field_values = Vec::with_capacity(deserializer.size_hint());
+
+        while let Some((field, value)) = deserializer.next_field()? {
+            field_values.push(FieldValue::new(field, value));
         }
-    }
 
+        Ok(Self { field_values })
+    }
+}
+
+impl From<Vec<FieldValue>> for TantivyDocument {
+    fn from(field_values: Vec<FieldValue>) -> Self {
+        Self { field_values }
+    }
+}
+
+impl PartialEq for TantivyDocument {
+    fn eq(&self, other: &Self) -> bool {
+        // super slow, but only here for tests
+        let convert_to_comparable_map = |field_values: &[FieldValue]| {
+            let mut field_value_set: HashMap<Field, HashSet<String>> = Default::default();
+            for field_value in field_values.iter() {
+                let value = serde_json::to_string(field_value.value()).unwrap();
+                field_value_set
+                    .entry(field_value.field())
+                    .or_default()
+                    .insert(value);
+            }
+            field_value_set
+        };
+        let self_field_values: HashMap<Field, HashSet<String>> =
+            convert_to_comparable_map(&self.field_values);
+        let other_field_values: HashMap<Field, HashSet<String>> =
+            convert_to_comparable_map(&other.field_values);
+        self_field_values.eq(&other_field_values)
+    }
+}
+
+impl Eq for TantivyDocument {}
+
+impl IntoIterator for TantivyDocument {
+    type Item = FieldValue;
+
+    type IntoIter = std::vec::IntoIter<FieldValue>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        self.field_values.into_iter()
+    }
+}
+
+impl TantivyDocument {
     /// Creates a new, empty document object
-    pub fn new() -> CompactDoc {
-        CompactDoc::with_capacity(1024)
-    }
-
-    /// Skrinks the capacity of the document to fit the data
-    pub fn shrink_to_fit(&mut self) {
-        self.node_data.shrink_to_fit();
-        self.field_values.shrink_to_fit();
+    pub fn new() -> TantivyDocument {
+        TantivyDocument::default()
     }
 
     /// Returns the length of the document.
@@ -70,111 +99,83 @@ impl CompactDoc {
     pub fn add_facet<F>(&mut self, field: Field, path: F)
     where Facet: From<F> {
         let facet = Facet::from(path);
-        self.add_leaf_field_value(field, ReferenceValueLeaf::Facet(facet.encoded_str()));
+        let value = OwnedValue::Facet(facet);
+        self.add_field_value(field, value);
     }
 
     /// Add a text field.
-    pub fn add_text<S: AsRef<str>>(&mut self, field: Field, text: S) {
-        self.add_leaf_field_value(field, ReferenceValueLeaf::Str(text.as_ref()));
+    pub fn add_text<S: ToString>(&mut self, field: Field, text: S) {
+        let value = OwnedValue::Str(text.to_string());
+        self.add_field_value(field, value);
     }
 
     /// Add a pre-tokenized text field.
     pub fn add_pre_tokenized_text(&mut self, field: Field, pre_tokenized_text: PreTokenizedString) {
-        self.add_leaf_field_value(field, pre_tokenized_text);
+        self.add_field_value(field, pre_tokenized_text);
     }
 
     /// Add a u64 field
     pub fn add_u64(&mut self, field: Field, value: u64) {
-        self.add_leaf_field_value(field, value);
+        self.add_field_value(field, value);
     }
 
     /// Add a IP address field. Internally only Ipv6Addr is used.
     pub fn add_ip_addr(&mut self, field: Field, value: Ipv6Addr) {
-        self.add_leaf_field_value(field, value);
+        self.add_field_value(field, value);
     }
 
     /// Add a i64 field
     pub fn add_i64(&mut self, field: Field, value: i64) {
-        self.add_leaf_field_value(field, value);
+        self.add_field_value(field, value);
     }
 
     /// Add a f64 field
     pub fn add_f64(&mut self, field: Field, value: f64) {
-        self.add_leaf_field_value(field, value);
+        self.add_field_value(field, value);
     }
 
     /// Add a bool field
     pub fn add_bool(&mut self, field: Field, value: bool) {
-        self.add_leaf_field_value(field, value);
+        self.add_field_value(field, value);
     }
 
     /// Add a date field with unspecified time zone offset
     pub fn add_date(&mut self, field: Field, value: DateTime) {
-        self.add_leaf_field_value(field, value);
+        self.add_field_value(field, value);
     }
 
     /// Add a bytes field
-    pub fn add_bytes(&mut self, field: Field, value: &[u8]) {
-        self.add_leaf_field_value(field, value);
+    pub fn add_bytes<T: Into<Vec<u8>>>(&mut self, field: Field, value: T) {
+        self.add_field_value(field, value.into());
     }
 
     /// Add a dynamic object field
     pub fn add_object(&mut self, field: Field, object: BTreeMap<String, OwnedValue>) {
-        self.add_field_value(field, &OwnedValue::from(object));
+        self.add_field_value(field, object);
     }
 
     /// Add a (field, value) to the document.
-    ///
-    /// `OwnedValue` implements Value, which should be easiest to use, but is not the most
-    /// performant.
-    pub fn add_field_value<'a, V: Value<'a>>(&mut self, field: Field, value: V) {
-        let field_value = FieldValueAddr {
-            field: field
-                .field_id()
-                .try_into()
-                .expect("support only up to u16::MAX field ids"),
-            value_addr: self.add_value(value),
-        };
-        self.field_values.push(field_value);
-    }
-
-    /// Add a (field, leaf value) to the document.
-    /// Leaf values don't have nested values.
-    pub fn add_leaf_field_value<'a, T: Into<ReferenceValueLeaf<'a>>>(
-        &mut self,
-        field: Field,
-        typed_val: T,
-    ) {
+    pub fn add_field_value<T: Into<OwnedValue>>(&mut self, field: Field, typed_val: T) {
         let value = typed_val.into();
-        let field_value = FieldValueAddr {
-            field: field
-                .field_id()
-                .try_into()
-                .expect("support only up to u16::MAX field ids"),
-            value_addr: self.add_value_leaf(value),
-        };
+        let field_value = FieldValue { field, value };
         self.field_values.push(field_value);
     }
 
     /// field_values accessor
-    pub fn field_values(&self) -> impl Iterator<Item = (Field, CompactDocValue<'_>)> {
-        self.field_values.iter().map(|field_val| {
-            let field = Field::from_field_id(field_val.field as u32);
-            let val = self.get_compact_doc_value(field_val.value_addr);
-            (field, val)
-        })
+    pub fn field_values(&self) -> &[FieldValue] {
+        &self.field_values
     }
 
-    /// Returns all of the `ReferenceValue`s associated the given field
-    pub fn get_all(&self, field: Field) -> impl Iterator<Item = CompactDocValue<'_>> + '_ {
+    /// Returns all of the `FieldValue`s associated the given field
+    pub fn get_all(&self, field: Field) -> impl Iterator<Item = &OwnedValue> {
         self.field_values
             .iter()
-            .filter(move |field_value| Field::from_field_id(field_value.field as u32) == field)
-            .map(|val| self.get_compact_doc_value(val.value_addr))
+            .filter(move |field_value| field_value.field() == field)
+            .map(FieldValue::value)
     }
 
-    /// Returns the first `ReferenceValue` associated the given field
-    pub fn get_first(&self, field: Field) -> Option<CompactDocValue<'_>> {
+    /// Returns the first `FieldValue` associated the given field
+    pub fn get_first(&self, field: Field) -> Option<&OwnedValue> {
         self.get_all(field).next()
     }
 
@@ -182,12 +183,12 @@ impl CompactDoc {
     pub fn convert_named_doc(
         schema: &Schema,
         named_doc: NamedFieldDocument,
-    ) -> Result<Self, DocParsingError> {
-        let mut document = Self::new();
+    ) -> Result<TantivyDocument, DocParsingError> {
+        let mut document = TantivyDocument::new();
         for (field_name, values) in named_doc.0 {
             if let Ok(field) = schema.get_field(&field_name) {
                 for value in values {
-                    document.add_field_value(field, &value);
+                    document.add_field_value(field, value);
                 }
             }
         }
@@ -195,7 +196,7 @@ impl CompactDoc {
     }
 
     /// Build a document object from a json-object.
-    pub fn parse_json(schema: &Schema, doc_json: &str) -> Result<Self, DocParsingError> {
+    pub fn parse_json(schema: &Schema, doc_json: &str) -> Result<TantivyDocument, DocParsingError> {
         let json_obj: Map<String, serde_json::Value> =
             serde_json::from_str(doc_json).map_err(|_| DocParsingError::invalid_json(doc_json))?;
         Self::from_json_object(schema, json_obj)
@@ -205,8 +206,8 @@ impl CompactDoc {
     pub fn from_json_object(
         schema: &Schema,
         json_obj: Map<String, serde_json::Value>,
-    ) -> Result<Self, DocParsingError> {
-        let mut doc = Self::default();
+    ) -> Result<TantivyDocument, DocParsingError> {
+        let mut doc = TantivyDocument::default();
         for (field_name, json_value) in json_obj {
             if let Ok(field) = schema.get_field(&field_name) {
                 let field_entry = schema.get_field_entry(field);
@@ -217,482 +218,20 @@ impl CompactDoc {
                             let value = field_type
                                 .value_from_json(json_item)
                                 .map_err(|e| DocParsingError::ValueError(field_name.clone(), e))?;
-                            doc.add_field_value(field, &value);
+                            doc.add_field_value(field, value);
                         }
                     }
                     _ => {
                         let value = field_type
                             .value_from_json(json_value)
                             .map_err(|e| DocParsingError::ValueError(field_name.clone(), e))?;
-                        doc.add_field_value(field, &value);
+                        doc.add_field_value(field, value);
                     }
                 }
             }
         }
         Ok(doc)
     }
-
-    fn add_value_leaf(&mut self, leaf: ReferenceValueLeaf) -> ValueAddr {
-        let type_id = ValueType::from(&leaf);
-        // Write into `node_data` and return u32 position as its address
-        // Null and bool are inlined into the address
-        let val_addr = match leaf {
-            ReferenceValueLeaf::Null => 0,
-            ReferenceValueLeaf::Str(bytes) => {
-                write_bytes_into(&mut self.node_data, bytes.as_bytes())
-            }
-            ReferenceValueLeaf::Facet(bytes) => {
-                write_bytes_into(&mut self.node_data, bytes.as_bytes())
-            }
-            ReferenceValueLeaf::Bytes(bytes) => write_bytes_into(&mut self.node_data, bytes),
-            ReferenceValueLeaf::U64(num) => write_into(&mut self.node_data, num),
-            ReferenceValueLeaf::I64(num) => write_into(&mut self.node_data, num),
-            ReferenceValueLeaf::F64(num) => write_into(&mut self.node_data, num),
-            ReferenceValueLeaf::Bool(b) => b as u32,
-            ReferenceValueLeaf::Date(date) => {
-                write_into(&mut self.node_data, date.into_timestamp_nanos())
-            }
-            ReferenceValueLeaf::IpAddr(num) => write_into(&mut self.node_data, num.to_u128()),
-            ReferenceValueLeaf::PreTokStr(pre_tok) => write_into(&mut self.node_data, *pre_tok),
-        };
-        ValueAddr { type_id, val_addr }
-    }
-    /// Adds a value and returns in address into the
-    fn add_value<'a, V: Value<'a>>(&mut self, value: V) -> ValueAddr {
-        let value = value.as_value();
-        let type_id = ValueType::from(&value);
-        match value {
-            ReferenceValue::Leaf(leaf) => self.add_value_leaf(leaf),
-            ReferenceValue::Array(elements) => {
-                // addresses of the elements in node_data
-                // Reusing a vec would be nicer, but it's not easy because of the recursion
-                // A global vec would work if every writer get it's discriminator
-                let mut addresses = Vec::new();
-                for elem in elements {
-                    let value_addr = self.add_value(elem);
-                    write_into(&mut addresses, value_addr);
-                }
-                ValueAddr {
-                    type_id,
-                    val_addr: write_bytes_into(&mut self.node_data, &addresses),
-                }
-            }
-            ReferenceValue::Object(entries) => {
-                // addresses of the elements in node_data
-                let mut addresses = Vec::new();
-                for (key, value) in entries {
-                    let key_addr = self.add_value_leaf(ReferenceValueLeaf::Str(key));
-                    let value_addr = self.add_value(value);
-                    write_into(&mut addresses, key_addr);
-                    write_into(&mut addresses, value_addr);
-                }
-                ValueAddr {
-                    type_id,
-                    val_addr: write_bytes_into(&mut self.node_data, &addresses),
-                }
-            }
-        }
-    }
-
-    /// Get CompactDocValue for address
-    fn get_compact_doc_value(&self, value_addr: ValueAddr) -> CompactDocValue<'_> {
-        CompactDocValue {
-            container: self,
-            value_addr,
-        }
-    }
-
-    /// get &[u8] reference from node_data
-    fn extract_bytes(&self, addr: Addr) -> &[u8] {
-        binary_deserialize_bytes(self.get_slice(addr))
-    }
-
-    /// get &str reference from node_data
-    fn extract_str(&self, addr: Addr) -> &str {
-        let data = self.extract_bytes(addr);
-        // Utf-8 checks would have a noticeable performance overhead here
-        unsafe { std::str::from_utf8_unchecked(data) }
-    }
-
-    /// deserialized owned value from node_data
-    fn read_from<T: BinarySerializable>(&self, addr: Addr) -> io::Result<T> {
-        let data_slice = &self.node_data[addr as usize..];
-        let mut cursor = std::io::Cursor::new(data_slice);
-        T::deserialize(&mut cursor)
-    }
-
-    /// get slice from address. The returned slice is open ended
-    fn get_slice(&self, addr: Addr) -> &[u8] {
-        &self.node_data[addr as usize..]
-    }
-}
-
-/// BinarySerializable alternative to read references
-fn binary_deserialize_bytes(data: &[u8]) -> &[u8] {
-    let (len, bytes_read) = read_u32_vint_no_advance(data);
-    &data[bytes_read..bytes_read + len as usize]
-}
-
-/// Write bytes and return the position of the written data.
-///
-/// BinarySerializable alternative to write references
-fn write_bytes_into(vec: &mut Vec<u8>, data: &[u8]) -> u32 {
-    let pos = vec.len() as u32;
-    let mut buf = [0u8; 8];
-    let len_vint_bytes = serialize_vint_u32(data.len() as u32, &mut buf);
-    vec.extend_from_slice(len_vint_bytes);
-    vec.extend_from_slice(data);
-    pos
-}
-
-/// Serialize and return the position
-fn write_into<T: BinarySerializable>(vec: &mut Vec<u8>, value: T) -> u32 {
-    let pos = vec.len() as u32;
-    value.serialize(vec).unwrap();
-    pos
-}
-
-impl PartialEq for CompactDoc {
-    fn eq(&self, other: &Self) -> bool {
-        // super slow, but only here for tests
-        let convert_to_comparable_map = |doc: &CompactDoc| {
-            let mut field_value_set: HashMap<Field, HashSet<String>> = Default::default();
-            for field_value in doc.field_values.iter() {
-                let value: OwnedValue = doc.get_compact_doc_value(field_value.value_addr).into();
-                let value = serde_json::to_string(&value).unwrap();
-                field_value_set
-                    .entry(Field::from_field_id(field_value.field as u32))
-                    .or_default()
-                    .insert(value);
-            }
-            field_value_set
-        };
-        let self_field_values: HashMap<Field, HashSet<String>> = convert_to_comparable_map(self);
-        let other_field_values: HashMap<Field, HashSet<String>> = convert_to_comparable_map(other);
-        self_field_values.eq(&other_field_values)
-    }
-}
-
-impl Eq for CompactDoc {}
-
-impl DocumentDeserialize for CompactDoc {
-    fn deserialize<'de, D>(mut deserializer: D) -> Result<Self, DeserializeError>
-    where D: DocumentDeserializer<'de> {
-        let mut doc = CompactDoc::default();
-        // TODO: Deserializing into OwnedValue is wasteful. The deserializer should be able to work
-        // on slices and referenced data.
-        while let Some((field, value)) = deserializer.next_field::<OwnedValue>()? {
-            doc.add_field_value(field, &value);
-        }
-        Ok(doc)
-    }
-}
-
-/// A value of Compact Doc needs a reference to the container to extract its payload
-#[derive(Debug, Clone, Copy)]
-pub struct CompactDocValue<'a> {
-    container: &'a CompactDoc,
-    value_addr: ValueAddr,
-}
-impl PartialEq for CompactDocValue<'_> {
-    fn eq(&self, other: &Self) -> bool {
-        let value1: OwnedValue = (*self).into();
-        let value2: OwnedValue = (*other).into();
-        value1 == value2
-    }
-}
-impl<'a> From<CompactDocValue<'a>> for OwnedValue {
-    fn from(value: CompactDocValue) -> Self {
-        value.as_value().into()
-    }
-}
-impl<'a> Value<'a> for CompactDocValue<'a> {
-    type ArrayIter = CompactDocArrayIter<'a>;
-
-    type ObjectIter = CompactDocObjectIter<'a>;
-
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        self.get_ref_value().unwrap()
-    }
-}
-impl<'a> CompactDocValue<'a> {
-    fn get_ref_value(&self) -> io::Result<ReferenceValue<'a, CompactDocValue<'a>>> {
-        let addr = self.value_addr.val_addr;
-        match self.value_addr.type_id {
-            ValueType::Null => Ok(ReferenceValueLeaf::Null.into()),
-            ValueType::Str => {
-                let str_ref = self.container.extract_str(addr);
-                Ok(ReferenceValueLeaf::Str(str_ref).into())
-            }
-            ValueType::Facet => {
-                let str_ref = self.container.extract_str(addr);
-                Ok(ReferenceValueLeaf::Facet(str_ref).into())
-            }
-            ValueType::Bytes => {
-                let data = self.container.extract_bytes(addr);
-                Ok(ReferenceValueLeaf::Bytes(data).into())
-            }
-            ValueType::U64 => self
-                .container
-                .read_from::<u64>(addr)
-                .map(ReferenceValueLeaf::U64)
-                .map(Into::into),
-            ValueType::I64 => self
-                .container
-                .read_from::<i64>(addr)
-                .map(ReferenceValueLeaf::I64)
-                .map(Into::into),
-            ValueType::F64 => self
-                .container
-                .read_from::<f64>(addr)
-                .map(ReferenceValueLeaf::F64)
-                .map(Into::into),
-            ValueType::Bool => Ok(ReferenceValueLeaf::Bool(addr != 0).into()),
-            ValueType::Date => self
-                .container
-                .read_from::<i64>(addr)
-                .map(|ts| ReferenceValueLeaf::Date(DateTime::from_timestamp_nanos(ts)))
-                .map(Into::into),
-            ValueType::IpAddr => self
-                .container
-                .read_from::<u128>(addr)
-                .map(|num| ReferenceValueLeaf::IpAddr(Ipv6Addr::from_u128(num)))
-                .map(Into::into),
-            ValueType::PreTokStr => self
-                .container
-                .read_from::<PreTokenizedString>(addr)
-                .map(Into::into)
-                .map(ReferenceValueLeaf::PreTokStr)
-                .map(Into::into),
-            ValueType::Object => Ok(ReferenceValue::Object(CompactDocObjectIter::new(
-                self.container,
-                addr,
-            )?)),
-            ValueType::Array => Ok(ReferenceValue::Array(CompactDocArrayIter::new(
-                self.container,
-                addr,
-            )?)),
-        }
-    }
-}
-
-/// The address in the vec
-type Addr = u32;
-
-#[derive(Clone, Copy, Default)]
-#[repr(packed)]
-/// The value type and the address to its payload in the container.
-struct ValueAddr {
-    type_id: ValueType,
-    /// This is the address to the value in the vec, except for bool and null, which are inlined
-    val_addr: Addr,
-}
-impl BinarySerializable for ValueAddr {
-    fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        self.type_id.serialize(writer)?;
-        VInt(self.val_addr as u64).serialize(writer)
-    }
-
-    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        let type_id = ValueType::deserialize(reader)?;
-        let val_addr = VInt::deserialize(reader)?.0 as u32;
-        Ok(ValueAddr { type_id, val_addr })
-    }
-}
-impl std::fmt::Debug for ValueAddr {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        let val_addr = self.val_addr;
-        f.write_fmt(format_args!("{:?} at {:?}", self.type_id, val_addr))
-    }
-}
-
-/// A enum representing a value for tantivy to index.
-///
-/// Any changes need to be reflected in `BinarySerializable` for `ValueType`
-///
-/// We can't use [schema::Type] or [columnar::ColumnType] here, because they are missing
-/// some items like Array and PreTokStr.
-#[derive(Default, Clone, Copy, Debug, PartialEq)]
-#[repr(u8)]
-pub enum ValueType {
-    /// A null value.
-    #[default]
-    Null = 0,
-    /// The str type is used for any text information.
-    Str = 1,
-    /// Unsigned 64-bits Integer `u64`
-    U64 = 2,
-    /// Signed 64-bits Integer `i64`
-    I64 = 3,
-    /// 64-bits Float `f64`
-    F64 = 4,
-    /// Date/time with nanoseconds precision
-    Date = 5,
-    /// Facet
-    Facet = 6,
-    /// Arbitrarily sized byte array
-    Bytes = 7,
-    /// IpV6 Address. Internally there is no IpV4, it needs to be converted to `Ipv6Addr`.
-    IpAddr = 8,
-    /// Bool value
-    Bool = 9,
-    /// Pre-tokenized str type,
-    PreTokStr = 10,
-    /// Object
-    Object = 11,
-    /// Pre-tokenized str type,
-    Array = 12,
-}
-
-impl BinarySerializable for ValueType {
-    fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        (*self as u8).serialize(writer)?;
-        Ok(())
-    }
-
-    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        let num = u8::deserialize(reader)?;
-        let type_id = if (0..=12).contains(&num) {
-            unsafe { std::mem::transmute(num) }
-        } else {
-            return Err(io::Error::new(
-                io::ErrorKind::InvalidData,
-                format!("Invalid value type id: {num}"),
-            ));
-        };
-        Ok(type_id)
-    }
-}
-
-impl<'a, V: Value<'a>> From<&ReferenceValue<'a, V>> for ValueType {
-    fn from(value: &ReferenceValue<'a, V>) -> Self {
-        match value {
-            ReferenceValue::Leaf(leaf) => leaf.into(),
-            ReferenceValue::Array(_) => ValueType::Array,
-            ReferenceValue::Object(_) => ValueType::Object,
-        }
-    }
-}
-impl<'a> From<&ReferenceValueLeaf<'a>> for ValueType {
-    fn from(value: &ReferenceValueLeaf<'a>) -> Self {
-        match value {
-            ReferenceValueLeaf::Null => ValueType::Null,
-            ReferenceValueLeaf::Str(_) => ValueType::Str,
-            ReferenceValueLeaf::U64(_) => ValueType::U64,
-            ReferenceValueLeaf::I64(_) => ValueType::I64,
-            ReferenceValueLeaf::F64(_) => ValueType::F64,
-            ReferenceValueLeaf::Bool(_) => ValueType::Bool,
-            ReferenceValueLeaf::Date(_) => ValueType::Date,
-            ReferenceValueLeaf::IpAddr(_) => ValueType::IpAddr,
-            ReferenceValueLeaf::PreTokStr(_) => ValueType::PreTokStr,
-            ReferenceValueLeaf::Facet(_) => ValueType::Facet,
-            ReferenceValueLeaf::Bytes(_) => ValueType::Bytes,
-        }
-    }
-}
-
-#[derive(Debug, Clone)]
-/// The Iterator for the object values in the compact document
-pub struct CompactDocObjectIter<'a> {
-    container: &'a CompactDoc,
-    node_addresses_slice: &'a [u8],
-}
-
-impl<'a> CompactDocObjectIter<'a> {
-    fn new(container: &'a CompactDoc, addr: Addr) -> io::Result<Self> {
-        // Objects are `&[ValueAddr]` serialized into bytes
-        let node_addresses_slice = container.extract_bytes(addr);
-        Ok(Self {
-            container,
-            node_addresses_slice,
-        })
-    }
-}
-
-impl<'a> Iterator for CompactDocObjectIter<'a> {
-    type Item = (&'a str, CompactDocValue<'a>);
-
-    fn next(&mut self) -> Option<Self::Item> {
-        if self.node_addresses_slice.is_empty() {
-            return None;
-        }
-        let key_addr = ValueAddr::deserialize(&mut self.node_addresses_slice).ok()?;
-        let key = self.container.extract_str(key_addr.val_addr);
-        let value = ValueAddr::deserialize(&mut self.node_addresses_slice).ok()?;
-        let value = CompactDocValue {
-            container: self.container,
-            value_addr: value,
-        };
-        Some((key, value))
-    }
-}
-
-#[derive(Debug, Clone)]
-/// The Iterator for the array values in the compact document
-pub struct CompactDocArrayIter<'a> {
-    container: &'a CompactDoc,
-    node_addresses_slice: &'a [u8],
-}
-
-impl<'a> CompactDocArrayIter<'a> {
-    fn new(container: &'a CompactDoc, addr: Addr) -> io::Result<Self> {
-        // Arrays are &[ValueAddr] serialized into bytes
-        let node_addresses_slice = container.extract_bytes(addr);
-        Ok(Self {
-            container,
-            node_addresses_slice,
-        })
-    }
-}
-
-impl<'a> Iterator for CompactDocArrayIter<'a> {
-    type Item = CompactDocValue<'a>;
-
-    fn next(&mut self) -> Option<Self::Item> {
-        if self.node_addresses_slice.is_empty() {
-            return None;
-        }
-        let value = ValueAddr::deserialize(&mut self.node_addresses_slice).ok()?;
-        let value = CompactDocValue {
-            container: self.container,
-            value_addr: value,
-        };
-        Some(value)
-    }
-}
-
-impl Document for CompactDoc {
-    type Value<'a> = CompactDocValue<'a>;
-    type FieldsValuesIter<'a> = FieldValueIterRef<'a>;
-
-    fn iter_fields_and_values(&self) -> Self::FieldsValuesIter<'_> {
-        FieldValueIterRef {
-            slice: self.field_values.iter(),
-            container: self,
-        }
-    }
-}
-
-/// A helper wrapper for creating an iterator over the field values
-pub struct FieldValueIterRef<'a> {
-    slice: std::slice::Iter<'a, FieldValueAddr>,
-    container: &'a CompactDoc,
-}
-
-impl<'a> Iterator for FieldValueIterRef<'a> {
-    type Item = (Field, CompactDocValue<'a>);
-
-    fn next(&mut self) -> Option<Self::Item> {
-        self.slice.next().map(|field_value| {
-            (
-                Field::from_field_id(field_value.field as u32),
-                CompactDocValue::<'a> {
-                    container: self.container,
-                    value_addr: field_value.value_addr,
-                },
-            )
-        })
-    }
 }
 
 /// Error that may happen when deserializing
@@ -725,40 +264,7 @@ mod tests {
         let text_field = schema_builder.add_text_field("title", TEXT);
         let mut doc = TantivyDocument::default();
         doc.add_text(text_field, "My title");
-        assert_eq!(doc.field_values().count(), 1);
-
-        let schema = schema_builder.build();
-        let _val = doc.get_first(text_field).unwrap();
-        let _json = doc.to_named_doc(&schema);
-    }
-
-    #[test]
-    fn test_json_value() {
-        let json_str = r#"{ 
-            "toto": "titi",
-            "float": -0.2,
-            "bool": true,
-            "unsigned": 1,
-            "signed": -2,
-            "complexobject": {
-                "field.with.dot": 1
-            },
-            "date": "1985-04-12T23:20:50.52Z",
-            "my_arr": [2, 3, {"my_key": "two tokens"}, 4, {"nested_array": [2, 5, 6, [7, 8, {"a": [{"d": {"e":[99]}}, 9000]}, 9, 10], [5, 5]]}]
-        }"#;
-        let json_val: std::collections::BTreeMap<String, OwnedValue> =
-            serde_json::from_str(json_str).unwrap();
-
-        let mut schema_builder = Schema::builder();
-        let json_field = schema_builder.add_json_field("json", TEXT);
-        let mut doc = TantivyDocument::default();
-        doc.add_object(json_field, json_val);
-
-        let schema = schema_builder.build();
-        let json = doc.to_json(&schema);
-        let actual_json: serde_json::Value = serde_json::from_str(&json).unwrap();
-        let expected_json: serde_json::Value = serde_json::from_str(json_str).unwrap();
-        assert_eq!(actual_json["json"][0], expected_json);
+        assert_eq!(doc.field_values().len(), 1);
     }
 
     // TODO: Should this be re-added with the serialize method

src/schema/document/existing_type_impls.rs

@@ -5,39 +5,21 @@
 //! and don't care about some of the more specialised types or only want to customise
 //! part of the document structure.
 use std::collections::{btree_map, hash_map, BTreeMap, HashMap};
-use std::iter::Empty;
-use std::net::Ipv6Addr;
 
-use common::DateTime;
 use serde_json::Number;
-use time::format_description::well_known::Rfc3339;
-use time::OffsetDateTime;
 
-use super::facet::Facet;
 use super::ReferenceValueLeaf;
 use crate::schema::document::{
     ArrayAccess, DeserializeError, Document, DocumentDeserialize, DocumentDeserializer,
     ObjectAccess, ReferenceValue, Value, ValueDeserialize, ValueDeserializer, ValueVisitor,
 };
 use crate::schema::Field;
-use crate::tokenizer::PreTokenizedString;
 
 // Serde compatibility support.
-pub fn can_be_rfc3339_date_time(text: &str) -> bool {
-    if let Some(&first_byte) = text.as_bytes().first() {
-        if first_byte.is_ascii_digit() {
-            return true;
-        }
-    }
-
-    false
-}
-
 impl<'a> Value<'a> for &'a serde_json::Value {
     type ArrayIter = std::slice::Iter<'a, serde_json::Value>;
     type ObjectIter = JsonObjectIter<'a>;
 
-    #[inline]
     fn as_value(&self) -> ReferenceValue<'a, Self> {
         match self {
             serde_json::Value::Null => ReferenceValueLeaf::Null.into(),
@@ -53,19 +35,7 @@ impl<'a> Value<'a> for &'a serde_json::Value {
                     panic!("Unsupported serde_json number {number}");
                 }
             }
-            serde_json::Value::String(text) => {
-                if can_be_rfc3339_date_time(text) {
-                    match OffsetDateTime::parse(text, &Rfc3339) {
-                        Ok(dt) => {
-                            let dt_utc = dt.to_offset(time::UtcOffset::UTC);
-                            ReferenceValueLeaf::Date(DateTime::from_utc(dt_utc)).into()
-                        }
-                        Err(_) => ReferenceValueLeaf::Str(text).into(),
-                    }
-                } else {
-                    ReferenceValueLeaf::Str(text).into()
-                }
-            }
+            serde_json::Value::String(val) => ReferenceValueLeaf::Str(val).into(),
             serde_json::Value::Array(elements) => ReferenceValue::Array(elements.iter()),
             serde_json::Value::Object(object) => {
                 ReferenceValue::Object(JsonObjectIter(object.iter()))
@@ -74,126 +44,6 @@ impl<'a> Value<'a> for &'a serde_json::Value {
     }
 }
 
-impl<'a> Value<'a> for &'a String {
-    type ArrayIter = Empty<&'a String>;
-    type ObjectIter = Empty<(&'a str, &'a String)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Str(self))
-    }
-}
-
-impl<'a> Value<'a> for &'a Facet {
-    type ArrayIter = Empty<&'a Facet>;
-    type ObjectIter = Empty<(&'a str, &'a Facet)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Facet(self.encoded_str()))
-    }
-}
-
-impl<'a> Value<'a> for &'a u64 {
-    type ArrayIter = Empty<&'a u64>;
-    type ObjectIter = Empty<(&'a str, &'a u64)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::U64(**self))
-    }
-}
-
-impl<'a> Value<'a> for &'a i64 {
-    type ArrayIter = Empty<&'a i64>;
-    type ObjectIter = Empty<(&'a str, &'a i64)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::I64(**self))
-    }
-}
-impl<'a> Value<'a> for &'a f64 {
-    type ArrayIter = Empty<&'a f64>;
-    type ObjectIter = Empty<(&'a str, &'a f64)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::F64(**self))
-    }
-}
-impl<'a> Value<'a> for &'a bool {
-    type ArrayIter = Empty<&'a bool>;
-    type ObjectIter = Empty<(&'a str, &'a bool)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Bool(**self))
-    }
-}
-impl<'a> Value<'a> for &'a str {
-    type ArrayIter = Empty<&'a str>;
-    type ObjectIter = Empty<(&'a str, &'a str)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Str(self))
-    }
-}
-impl<'a> Value<'a> for &'a &'a str {
-    type ArrayIter = Empty<&'a &'a str>;
-    type ObjectIter = Empty<(&'a str, &'a &'a str)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Str(self))
-    }
-}
-
-impl<'a> Value<'a> for &'a [u8] {
-    type ArrayIter = Empty<&'a [u8]>;
-    type ObjectIter = Empty<(&'a str, &'a [u8])>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Bytes(self))
-    }
-}
-
-impl<'a> Value<'a> for &'a &'a [u8] {
-    type ArrayIter = Empty<&'a &'a [u8]>;
-    type ObjectIter = Empty<(&'a str, &'a &'a [u8])>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Bytes(self))
-    }
-}
-
-impl<'a> Value<'a> for &'a Vec<u8> {
-    type ArrayIter = Empty<&'a Vec<u8>>;
-    type ObjectIter = Empty<(&'a str, &'a Vec<u8>)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Bytes(self))
-    }
-}
-
-impl<'a> Value<'a> for &'a DateTime {
-    type ArrayIter = Empty<&'a DateTime>;
-    type ObjectIter = Empty<(&'a str, &'a DateTime)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::Date(**self))
-    }
-}
-impl<'a> Value<'a> for &'a Ipv6Addr {
-    type ArrayIter = Empty<&'a Ipv6Addr>;
-    type ObjectIter = Empty<(&'a str, &'a Ipv6Addr)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::IpAddr(**self))
-    }
-}
-impl<'a> Value<'a> for &'a PreTokenizedString {
-    type ArrayIter = Empty<&'a PreTokenizedString>;
-    type ObjectIter = Empty<(&'a str, &'a PreTokenizedString)>;
-    #[inline]
-    fn as_value(&self) -> ReferenceValue<'a, Self> {
-        ReferenceValue::Leaf(ReferenceValueLeaf::PreTokStr(Box::new((*self).clone())))
-    }
-}
-
 impl ValueDeserialize for serde_json::Value {
     fn deserialize<'de, D>(deserializer: D) -> Result<Self, DeserializeError>
     where D: ValueDeserializer<'de> {

src/schema/document/mod.rs

@@ -172,9 +172,7 @@ pub use self::de::{
     ArrayAccess, DeserializeError, DocumentDeserialize, DocumentDeserializer, ObjectAccess,
     ValueDeserialize, ValueDeserializer, ValueType, ValueVisitor,
 };
-pub use self::default_document::{
-    CompactDocArrayIter, CompactDocObjectIter, CompactDocValue, DocParsingError, TantivyDocument,
-};
+pub use self::default_document::{DocParsingError, TantivyDocument};
 pub use self::owned_value::OwnedValue;
 pub(crate) use self::se::BinaryDocumentSerializer;
 pub use self::value::{ReferenceValue, ReferenceValueLeaf, Value};
@@ -235,7 +233,7 @@ pub trait Document: Send + Sync + 'static {
             let field_name = schema.get_field_name(field);
             let values: Vec<OwnedValue> = field_values
                 .into_iter()
-                .map(|val| OwnedValue::from(val.as_value()))
+                .map(|val| val.as_value().into())
                 .collect();
             field_map.insert(field_name.to_string(), values);
         }

src/schema/document/owned_value.rs

@@ -8,7 +8,6 @@ use serde::de::{MapAccess, SeqAccess};
 use time::format_description::well_known::Rfc3339;
 use time::OffsetDateTime;
 
-use super::existing_type_impls::can_be_rfc3339_date_time;
 use super::ReferenceValueLeaf;
 use crate::schema::document::{
     ArrayAccess, DeserializeError, ObjectAccess, ReferenceValue, Value, ValueDeserialize,
@@ -66,13 +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.clone().into()).into(),
+            OwnedValue::PreTokStr(val) => ReferenceValueLeaf::PreTokStr(val).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.encoded_str()).into(),
+            OwnedValue::Facet(val) => ReferenceValueLeaf::Facet(val).into(),
             OwnedValue::Bytes(val) => ReferenceValueLeaf::Bytes(val).into(),
             OwnedValue::IpAddr(val) => ReferenceValueLeaf::IpAddr(*val).into(),
             OwnedValue::Array(array) => ReferenceValue::Array(array.iter()),
@@ -184,7 +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 (k, v) in obj {
+                for &(ref k, ref v) in obj {
                     map.serialize_entry(k, v)?;
                 }
                 map.end()
@@ -278,13 +277,11 @@ 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(Facet::from_encoded_string(val.to_string()))
-                }
+                ReferenceValueLeaf::Facet(val) => OwnedValue::Facet(val.clone()),
                 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())
@@ -376,6 +373,16 @@ impl From<BTreeMap<String, OwnedValue>> for OwnedValue {
     }
 }
 
+fn can_be_rfc3339_date_time(text: &str) -> bool {
+    if let Some(&first_byte) = text.as_bytes().first() {
+        if first_byte.is_ascii_digit() {
+            return true;
+        }
+    }
+
+    false
+}
+
 impl From<serde_json::Value> for OwnedValue {
     fn from(value: serde_json::Value) -> Self {
         match value {
@@ -463,7 +470,6 @@ mod tests {
         let mut doc = TantivyDocument::default();
         doc.add_bytes(bytes_field, "".as_bytes());
         let json_string = doc.to_json(&schema);
-
         assert_eq!(json_string, r#"{"my_bytes":[""]}"#);
     }
 

src/schema/document/se.rs

@@ -25,7 +25,6 @@ where W: Write
 
     /// Attempts to serialize a given document and write the output
     /// to the writer.
-    #[inline]
     pub(crate) fn serialize_doc<D>(&mut self, doc: &D) -> io::Result<()>
     where D: Document {
         let stored_field_values = || {
@@ -58,8 +57,9 @@ where W: Write
             return Err(io::Error::new(
                 io::ErrorKind::Other,
                 format!(
-                    "Unexpected number of entries written to serializer, expected \
-                     {num_field_values} entries, got {actual_length} entries",
+                    "Unexpected number of entries written to serializer, expected {} entries, got \
+                     {} entries",
+                    num_field_values, actual_length,
                 ),
             ));
         }
@@ -121,7 +121,7 @@ where W: Write
                 ReferenceValueLeaf::Facet(val) => {
                     self.write_type_code(type_codes::HIERARCHICAL_FACET_CODE)?;
 
-                    Cow::Borrowed(val).serialize(self.writer)
+                    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.encoded_str()).into());
+        let result = serialize_value(ReferenceValueLeaf::Facet(&facet).into());
         let expected = binary_repr!(
             type_codes::HIERARCHICAL_FACET_CODE => Facet::from_text("/hello/world").unwrap(),
         );
@@ -441,8 +441,7 @@ mod tests {
             text: "hello, world".to_string(),
             tokens: vec![Token::default(), Token::default()],
         };
-        let result =
-            serialize_value(ReferenceValueLeaf::PreTokStr(pre_tok_str.clone().into()).into());
+        let result = serialize_value(ReferenceValueLeaf::PreTokStr(&pre_tok_str).into());
         let expected = binary_repr!(
             type_codes::EXT_CODE, type_codes::TOK_STR_EXT_CODE => pre_tok_str,
         );
@@ -679,7 +678,6 @@ mod tests {
         );
     }
 
-    #[inline]
     fn serialize_doc<D: Document>(doc: &D, schema: &Schema) -> Vec<u8> {
         let mut writer = Vec::new();
 

src/schema/document/value.rs

@@ -3,6 +3,7 @@ use std::net::Ipv6Addr;
 
 use common::DateTime;
 
+use crate::schema::Facet;
 use crate::tokenizer::PreTokenizedString;
 
 /// A single field value.
@@ -18,7 +19,16 @@ pub trait Value<'a>: Send + Sync + Debug {
     fn as_value(&self) -> ReferenceValue<'a, Self>;
 
     #[inline]
-    /// If the Value is a leaf, returns the associated leaf. Returns None otherwise.
+    /// Returns if the value is `null` or not.
+    fn is_null(&self) -> bool {
+        matches!(
+            self.as_value(),
+            ReferenceValue::Leaf(ReferenceValueLeaf::Null)
+        )
+    }
+
+    #[inline]
+    /// If the Value is a String, returns the associated str. Returns None otherwise.
     fn as_leaf(&self) -> Option<ReferenceValueLeaf<'a>> {
         if let ReferenceValue::Leaf(val) = self.as_value() {
             Some(val)
@@ -72,9 +82,8 @@ 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<Box<PreTokenizedString>> {
-        self.as_leaf()
-            .and_then(|leaf| leaf.into_pre_tokenized_text())
+    fn as_pre_tokenized_text(&self) -> Option<&'a PreTokenizedString> {
+        self.as_leaf().and_then(|leaf| leaf.as_pre_tokenized_text())
     }
 
     #[inline]
@@ -85,7 +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 str> {
+    fn as_facet(&self) -> Option<&'a Facet> {
         self.as_leaf().and_then(|leaf| leaf.as_facet())
     }
 
@@ -108,10 +117,22 @@ pub trait Value<'a>: Send + Sync + Debug {
             None
         }
     }
+
+    #[inline]
+    /// Returns true if the Value is an array.
+    fn is_array(&self) -> bool {
+        matches!(self.as_value(), ReferenceValue::Object(_))
+    }
+
+    #[inline]
+    /// Returns true if the Value is an object.
+    fn is_object(&self) -> bool {
+        matches!(self.as_value(), ReferenceValue::Object(_))
+    }
 }
 
 /// A enum representing a leaf value for tantivy to index.
-#[derive(Clone, Debug, PartialEq)]
+#[derive(Clone, Copy, Debug, PartialEq)]
 pub enum ReferenceValueLeaf<'a> {
     /// A null value.
     Null,
@@ -125,9 +146,8 @@ pub enum ReferenceValueLeaf<'a> {
     F64(f64),
     /// Date/time with nanoseconds precision
     Date(DateTime),
-    /// Facet string needs to match the format of
-    /// [Facet::encoded_str](crate::schema::Facet::encoded_str).
-    Facet(&'a str),
+    /// Facet
+    Facet(&'a Facet),
     /// Arbitrarily sized byte array
     Bytes(&'a [u8]),
     /// IpV6 Address. Internally there is no IpV4, it needs to be converted to `Ipv6Addr`.
@@ -135,70 +155,7 @@ pub enum ReferenceValueLeaf<'a> {
     /// Bool value
     Bool(bool),
     /// Pre-tokenized str type,
-    PreTokStr(Box<PreTokenizedString>),
-}
-
-impl From<u64> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(value: u64) -> Self {
-        ReferenceValueLeaf::U64(value)
-    }
-}
-
-impl From<i64> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(value: i64) -> Self {
-        ReferenceValueLeaf::I64(value)
-    }
-}
-
-impl From<f64> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(value: f64) -> Self {
-        ReferenceValueLeaf::F64(value)
-    }
-}
-
-impl From<bool> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(value: bool) -> Self {
-        ReferenceValueLeaf::Bool(value)
-    }
-}
-
-impl<'a> From<&'a str> for ReferenceValueLeaf<'a> {
-    #[inline]
-    fn from(value: &'a str) -> Self {
-        ReferenceValueLeaf::Str(value)
-    }
-}
-
-impl<'a> From<&'a [u8]> for ReferenceValueLeaf<'a> {
-    #[inline]
-    fn from(value: &'a [u8]) -> Self {
-        ReferenceValueLeaf::Bytes(value)
-    }
-}
-
-impl From<DateTime> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(value: DateTime) -> Self {
-        ReferenceValueLeaf::Date(value)
-    }
-}
-
-impl From<Ipv6Addr> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(value: Ipv6Addr) -> Self {
-        ReferenceValueLeaf::IpAddr(value)
-    }
-}
-
-impl From<PreTokenizedString> for ReferenceValueLeaf<'_> {
-    #[inline]
-    fn from(val: PreTokenizedString) -> Self {
-        ReferenceValueLeaf::PreTokStr(Box::new(val))
-    }
+    PreTokStr(&'a PreTokenizedString),
 }
 
 impl<'a, T: Value<'a> + ?Sized> From<ReferenceValueLeaf<'a>> for ReferenceValue<'a, T> {
@@ -302,9 +259,9 @@ impl<'a> ReferenceValueLeaf<'a> {
     }
 
     #[inline]
-    /// 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 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 let Self::PreTokStr(val) = self {
             Some(val)
         } else {
@@ -324,7 +281,7 @@ 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 str> {
+    pub fn as_facet(&self) -> Option<&'a Facet> {
         if let Self::Facet(val) = self {
             Some(val)
         } else {
@@ -365,16 +322,6 @@ 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> {
@@ -418,11 +365,10 @@ where V: Value<'a>
     }
 
     #[inline]
-    /// 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())
+    /// 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())
     }
 
     #[inline]
@@ -433,7 +379,7 @@ 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 str> {
+    pub fn as_facet(&self) -> Option<&'a Facet> {
         self.as_leaf().and_then(|leaf| leaf.as_facet())
     }
 

src/schema/field_type.rs

@@ -568,21 +568,21 @@ mod tests {
         let schema = schema_builder.build();
         let doc = TantivyDocument::parse_json(&schema, r#"{"id": 100}"#).unwrap();
         assert_eq!(
-            OwnedValue::Str("100".to_string()),
-            doc.get_first(text_field).unwrap().into()
+            &OwnedValue::Str("100".to_string()),
+            doc.get_first(text_field).unwrap()
         );
 
         let doc = TantivyDocument::parse_json(&schema, r#"{"id": true}"#).unwrap();
         assert_eq!(
-            OwnedValue::Str("true".to_string()),
-            doc.get_first(text_field).unwrap().into()
+            &OwnedValue::Str("true".to_string()),
+            doc.get_first(text_field).unwrap()
         );
 
         // Not sure if this null coercion is the best approach
         let doc = TantivyDocument::parse_json(&schema, r#"{"id": null}"#).unwrap();
         assert_eq!(
-            OwnedValue::Str("null".to_string()),
-            doc.get_first(text_field).unwrap().into()
+            &OwnedValue::Str("null".to_string()),
+            doc.get_first(text_field).unwrap()
         );
     }
 
@@ -595,18 +595,9 @@ mod tests {
         let schema = schema_builder.build();
         let doc_json = r#"{"i64": "100", "u64": "100", "f64": "100"}"#;
         let doc = TantivyDocument::parse_json(&schema, doc_json).unwrap();
-        assert_eq!(
-            OwnedValue::I64(100),
-            doc.get_first(i64_field).unwrap().into()
-        );
-        assert_eq!(
-            OwnedValue::U64(100),
-            doc.get_first(u64_field).unwrap().into()
-        );
-        assert_eq!(
-            OwnedValue::F64(100.0),
-            doc.get_first(f64_field).unwrap().into()
-        );
+        assert_eq!(&OwnedValue::I64(100), doc.get_first(i64_field).unwrap());
+        assert_eq!(&OwnedValue::U64(100), doc.get_first(u64_field).unwrap());
+        assert_eq!(&OwnedValue::F64(100.0), doc.get_first(f64_field).unwrap());
     }
 
     #[test]
@@ -616,17 +607,11 @@ mod tests {
         let schema = schema_builder.build();
         let doc_json = r#"{"bool": "true"}"#;
         let doc = TantivyDocument::parse_json(&schema, doc_json).unwrap();
-        assert_eq!(
-            OwnedValue::Bool(true),
-            doc.get_first(bool_field).unwrap().into()
-        );
+        assert_eq!(&OwnedValue::Bool(true), doc.get_first(bool_field).unwrap());
 
         let doc_json = r#"{"bool": "false"}"#;
         let doc = TantivyDocument::parse_json(&schema, doc_json).unwrap();
-        assert_eq!(
-            OwnedValue::Bool(false),
-            doc.get_first(bool_field).unwrap().into()
-        );
+        assert_eq!(&OwnedValue::Bool(false), doc.get_first(bool_field).unwrap());
     }
 
     #[test]
@@ -659,7 +644,7 @@ mod tests {
         let schema = schema_builder.build();
         let doc_json = r#"{"date": "2019-10-12T07:20:50.52+02:00"}"#;
         let doc = TantivyDocument::parse_json(&schema, doc_json).unwrap();
-        let date = OwnedValue::from(doc.get_first(date_field).unwrap());
+        let date = doc.get_first(date_field).unwrap();
         // Time zone is converted to UTC
         assert_eq!("Date(2019-10-12T05:20:50.52Z)", format!("{date:?}"));
     }

src/schema/field_value.rs

@@ -0,0 +1,46 @@
+use crate::schema::{Field, OwnedValue};
+
+/// `FieldValue` holds together a `Field` and its `Value`.
+#[allow(missing_docs)]
+#[derive(Debug, Clone, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
+pub struct FieldValue {
+    pub field: Field,
+    pub value: OwnedValue,
+}
+
+impl FieldValue {
+    /// Constructor
+    pub fn new(field: Field, value: OwnedValue) -> FieldValue {
+        FieldValue { field, value }
+    }
+
+    /// Field accessor
+    pub fn field(&self) -> Field {
+        self.field
+    }
+
+    /// Value accessor
+    pub fn value(&self) -> &OwnedValue {
+        &self.value
+    }
+}
+
+impl From<FieldValue> for OwnedValue {
+    fn from(field_value: FieldValue) -> Self {
+        field_value.value
+    }
+}
+
+/// A helper wrapper for creating standard iterators
+/// out of the fields iterator trait.
+pub struct FieldValueIter<'a>(pub(crate) std::slice::Iter<'a, FieldValue>);
+
+impl<'a> Iterator for FieldValueIter<'a> {
+    type Item = (Field, &'a OwnedValue);
+
+    fn next(&mut self) -> Option<Self::Item> {
+        self.0
+            .next()
+            .map(|field_value| (field_value.field, &field_value.value))
+    }
+}

src/schema/flags.rs

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

src/schema/mod.rs

@@ -114,6 +114,7 @@ pub(crate) mod term;
 
 mod field_entry;
 mod field_type;
+mod field_value;
 
 mod bytes_options;
 mod date_time_options;
@@ -137,6 +138,7 @@ pub use self::facet_options::FacetOptions;
 pub use self::field::Field;
 pub use self::field_entry::FieldEntry;
 pub use self::field_type::{FieldType, Type};
+pub use self::field_value::FieldValue;
 pub use self::flags::{COERCE, FAST, INDEXED, STORED};
 pub use self::index_record_option::IndexRecordOption;
 pub use self::ip_options::{IntoIpv6Addr, IpAddrOptions};

src/schema/schema.rs

@@ -645,15 +645,15 @@ mod tests {
         let doc =
             TantivyDocument::convert_named_doc(&schema, NamedFieldDocument(named_doc_map)).unwrap();
         assert_eq!(
-            doc.get_all(title).map(OwnedValue::from).collect::<Vec<_>>(),
+            doc.get_all(title).collect::<Vec<_>>(),
             vec![
-                OwnedValue::from("title1".to_string()),
-                OwnedValue::from("title2".to_string())
+                &OwnedValue::from("title1".to_string()),
+                &OwnedValue::from("title2".to_string())
             ]
         );
         assert_eq!(
-            doc.get_all(val).map(OwnedValue::from).collect::<Vec<_>>(),
-            vec![OwnedValue::from(14u64), OwnedValue::from(-1i64)]
+            doc.get_all(val).collect::<Vec<_>>(),
+            vec![&OwnedValue::from(14u64), &OwnedValue::from(-1i64)]
         );
     }
 
@@ -682,7 +682,7 @@ mod tests {
         let schema = schema_builder.build();
         {
             let doc = TantivyDocument::parse_json(&schema, "{}").unwrap();
-            assert!(doc.field_values().next().is_none());
+            assert!(doc.field_values().is_empty());
         }
         {
             let doc = TantivyDocument::parse_json(

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::index::SegmentComponent::*;
+        use crate::SegmentComponent::*;
         match component {
             Postings => PerField(self.postings().clone()),
             Positions => PerField(self.positions().clone()),

src/store/mod.rs

@@ -59,8 +59,9 @@ pub mod tests {
     use super::*;
     use crate::directory::{Directory, RamDirectory, WritePtr};
     use crate::fastfield::AliveBitSet;
+    use crate::schema::document::Value;
     use crate::schema::{
-        self, Schema, TantivyDocument, TextFieldIndexing, TextOptions, Value, STORED, TEXT,
+        self, Schema, TantivyDocument, TextFieldIndexing, TextOptions, STORED, TEXT,
     };
     use crate::{Index, IndexWriter, Term};
 
@@ -91,8 +92,8 @@ pub mod tests {
                 StoreWriter::new(writer, compressor, blocksize, separate_thread).unwrap();
             for i in 0..num_docs {
                 let mut doc = TantivyDocument::default();
-                doc.add_text(field_body, LOREM);
-                doc.add_text(field_title, format!("Doc {i}"));
+                doc.add_field_value(field_body, LOREM.to_string());
+                doc.add_field_value(field_title, format!("Doc {i}"));
                 store_writer.store(&doc, &schema).unwrap();
             }
             store_writer.close().unwrap();
@@ -118,11 +119,10 @@ pub mod tests {
         let store = StoreReader::open(store_file, 10)?;
         for i in 0..NUM_DOCS as u32 {
             assert_eq!(
-                store
+                *store
                     .get::<TantivyDocument>(i)?
                     .get_first(field_title)
                     .unwrap()
-                    .as_value()
                     .as_str()
                     .unwrap(),
                 format!("Doc {i}")
@@ -131,13 +131,7 @@ pub mod tests {
 
         for doc in store.iter::<TantivyDocument>(Some(&alive_bitset)) {
             let doc = doc?;
-            let title_content = doc
-                .get_first(field_title)
-                .unwrap()
-                .as_value()
-                .as_str()
-                .unwrap()
-                .to_string();
+            let title_content = doc.get_first(field_title).unwrap().as_str().unwrap();
             if !title_content.starts_with("Doc ") {
                 panic!("unexpected title_content {title_content}");
             }

src/store/reader.rs

@@ -18,8 +18,6 @@ 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;
 
@@ -343,11 +341,7 @@ 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,
-        executor: &Executor,
-    ) -> io::Result<Block> {
+    async fn read_block_async(&self, checkpoint: &Checkpoint) -> 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);
@@ -359,12 +353,8 @@ impl StoreReader {
             .read_bytes_async()
             .await?;
 
-        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?);
+        let decompressed_block =
+            OwnedBytes::new(self.decompressor.decompress(compressed_block.as_ref())?);
 
         self.cache
             .put_into_cache(cache_key, decompressed_block.clone());
@@ -373,23 +363,15 @@ impl StoreReader {
     }
 
     /// Reads raw bytes of a given document asynchronously.
-    pub async fn get_document_bytes_async(
-        &self,
-        doc_id: DocId,
-        executor: &Executor,
-    ) -> crate::Result<OwnedBytes> {
+    pub async fn get_document_bytes_async(&self, doc_id: DocId) -> crate::Result<OwnedBytes> {
         let checkpoint = self.block_checkpoint(doc_id)?;
-        let block = self.read_block_async(&checkpoint, executor).await?;
+        let block = self.read_block_async(&checkpoint).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,
-        executor: &Executor,
-    ) -> crate::Result<D> {
-        let mut doc_bytes = self.get_document_bytes_async(doc_id, executor).await?;
+    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?;
 
         let deserializer = BinaryDocumentDeserializer::from_reader(&mut doc_bytes)
             .map_err(crate::TantivyError::from)?;
@@ -403,7 +385,8 @@ mod tests {
 
     use super::*;
     use crate::directory::RamDirectory;
-    use crate::schema::{Field, TantivyDocument, Value};
+    use crate::schema::document::Value;
+    use crate::schema::{Field, TantivyDocument};
     use crate::store::tests::write_lorem_ipsum_store;
     use crate::store::Compressor;
     use crate::Directory;
@@ -411,7 +394,7 @@ mod tests {
     const BLOCK_SIZE: usize = 16_384;
 
     fn get_text_field<'a>(doc: &'a TantivyDocument, field: &'a Field) -> Option<&'a str> {
-        doc.get_first(*field).and_then(|f| f.as_value().as_str())
+        doc.get_first(*field).and_then(|f| f.as_str())
     }
 
     #[test]

src/termdict/fst_termdict/termdict.rs

@@ -93,7 +93,7 @@ fn open_fst_index(fst_file: FileSlice) -> io::Result<tantivy_fst::Map<OwnedBytes
     let fst = Fst::new(bytes).map_err(|err| {
         io::Error::new(
             io::ErrorKind::InvalidData,
-            format!("Fst data is corrupted: {err:?}"),
+            format!("Fst data is corrupted: {:?}", err),
         )
     })?;
     Ok(tantivy_fst::Map::from(fst))

src/termdict/tests.rs

@@ -95,7 +95,7 @@ fn test_term_dictionary_simple() -> crate::Result<()> {
 #[test]
 fn test_term_dictionary_stream() -> crate::Result<()> {
     let ids: Vec<_> = (0u32..10_000u32)
-        .map(|i| (format!("doc{i:0>6}"), i))
+        .map(|i| (format!("doc{:0>6}", i), i))
         .collect();
     let buffer: Vec<u8> = {
         let mut term_dictionary_builder = TermDictionaryBuilder::create(vec![]).unwrap();
@@ -156,7 +156,7 @@ fn test_stream_high_range_prefix_suffix() -> crate::Result<()> {
 #[test]
 fn test_stream_range() -> crate::Result<()> {
     let ids: Vec<_> = (0u32..10_000u32)
-        .map(|i| (format!("doc{i:0>6}"), i))
+        .map(|i| (format!("doc{:0>6}", i), i))
         .collect();
     let buffer: Vec<u8> = {
         let mut term_dictionary_builder = TermDictionaryBuilder::create(vec![]).unwrap();

src/tokenizer/facet_tokenizer.rs

@@ -96,7 +96,7 @@ mod tests {
         {
             let mut add_token = |token: &Token| {
                 let facet = Facet::from_encoded(token.text.as_bytes().to_owned()).unwrap();
-                tokens.push(format!("{facet}"));
+                tokens.push(format!("{}", facet));
             };
             FacetTokenizer::default()
                 .token_stream(facet.encoded_str())
@@ -116,7 +116,7 @@ mod tests {
         {
             let mut add_token = |token: &Token| {
                 let facet = Facet::from_encoded(token.text.as_bytes().to_owned()).unwrap(); // ok test
-                tokens.push(format!("{facet}"));
+                tokens.push(format!("{}", facet));
             };
             FacetTokenizer::default()
                 .token_stream(facet.encoded_str()) // ok test