TermInfo contain the end_offset of the postings.

We slice the ReadOnlySource tightly.

.gitignore

@@ -1,4 +1,5 @@
 tantivy.iml
+proptest-regressions
 *.swp
 target
 target/debug
@@ -11,3 +12,4 @@ cpp/simdcomp/bitpackingbenchmark
 *.bk
 .idea
 trace.dat
+cargo-timing*

CHANGELOG.md

@@ -1,5 +1,21 @@
+Tantivy 0.14.0
+=========================
+- Remove dependency to atomicwrites #833 .Implemented by @pmasurel upon suggestion and research from @asafigan). 
+
+Tantivy 0.13.1
+===================
+Made `Query` and `Collector` `Send + Sync`.
+Updated misc dependency versions.
+
 Tantivy 0.13.0
 ======================
+Tantivy 0.13 introduce a change in the index format that will require
+you to reindex your index (BlockWAND information are added in the skiplist). 
+The index size increase is minor as this information is only added for
+full blocks.
+If you have a massive index for which reindexing is not an option, please contact me
+so that we can discuss possible solutions.
+
 - Bugfix in `FuzzyTermQuery` not matching terms by prefix when it should (@Peachball)
 - Relaxed constraints on the custom/tweak score functions. At the segment level, they can be mut, and they are not required to be Sync + Send.
 - `MMapDirectory::open` does not return a `Result` anymore.
@@ -17,6 +33,8 @@ while doc != TERMINATED {
 The change made it possible to greatly simplify a lot of the docset's code.
 - Misc internal optimization and introduction of the `Scorer::for_each_pruning` function. (@fulmicoton)
 - Added an offset option to the Top(.*)Collectors. (@robyoung)
+- Added Block WAND. Performance on TOP-K on term-unions should be greatly increased. (@fulmicoton, and special thanks 
+to the PISA team for answering all my questions!)
 
 Tantivy 0.12.0
 ======================

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.12.0"
+version = "0.14.0-dev"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -13,21 +13,20 @@ keywords = ["search", "information", "retrieval"]
 edition = "2018"
 
 [dependencies]
-base64 = "0.12.0"
-byteorder = "1.0"
-crc32fast = "1.2.0"
-once_cell = "1.0"
-regex ={version = "1.3.0", default-features = false, features = ["std"]}
+base64 = "0.12"
+byteorder = "1"
+crc32fast = "1"
+once_cell = "1"
+regex ={version = "1", default-features = false, features = ["std"]}
 tantivy-fst = "0.3"
 memmap = {version = "0.7", optional=true}
-lz4 = {version="1.20", optional=true}
+lz4 = {version="1", optional=true}
 snap = "1"
-atomicwrites = {version="0.2.2", optional=true}
-tempfile = "3.0"
+tempfile = {version="3", optional=true}
 log = "0.4"
-serde = {version="1.0", features=["derive"]}
-serde_json = "1.0"
-num_cpus = "1.2"
+serde = {version="1", features=["derive"]}
+serde_json = "1"
+num_cpus = "1"
 fs2={version="0.4", optional=true}
 levenshtein_automata = "0.2"
 notify = {version="4", optional=true}
@@ -35,20 +34,20 @@ uuid = { version = "0.8", features = ["v4", "serde"] }
 crossbeam = "0.7"
 futures = {version = "0.3",  features=["thread-pool"] }
 owning_ref = "0.4"
-stable_deref_trait = "1.0.0"
-rust-stemmers = "1.2"
-downcast-rs = { version="1.0" }
-tantivy-query-grammar = { version="0.13", path="./query-grammar" }
+tantivy-query-grammar = { version="0.14.0-dev", path="./query-grammar" }
+stable_deref_trait = "1"
+rust-stemmers = "1"
+downcast-rs = "1"
 bitpacking = {version="0.8", default-features = false, features=["bitpacker4x"]}
 census = "0.4"
-fnv = "1.0.6"
+fnv = "1"
 owned-read = "0.4"
 failure = "0.1"
-htmlescape = "0.3.1"
+htmlescape = "0.3"
 fail = "0.4"
 murmurhash32 = "0.2"
 chrono = "0.4"
-smallvec = "1.0"
+smallvec = "1"
 rayon = "1"
 
 [target.'cfg(windows)'.dependencies]
@@ -75,7 +74,7 @@ overflow-checks = true
 
 [features]
 default = ["mmap"]
-mmap = ["atomicwrites", "fs2", "memmap", "notify"]
+mmap = ["fs2", "tempfile", "memmap", "notify"]
 lz4-compression = ["lz4"]
 failpoints = ["fail/failpoints"]
 unstable = [] # useful for benches.

README.md

@@ -34,11 +34,6 @@ Tantivy is, in fact, strongly inspired by Lucene's design.
 The following [benchmark](https://tantivy-search.github.io/bench/) break downs 
 performance for different type of queries / collection.
 
-
-In general, Tantivy tends to be 
-- slower than Lucene on union with a Top-K due to Block-WAND optimization.
-- faster than Lucene on intersection and phrase queries. 
-
 Your mileage WILL vary depending on the nature of queries and their load.
 
 # Features

doc/src/index-format.md

@@ -0,0 +1,50 @@
+
+# Managed files 
++----------+-----------+-------------------+
+| content  |  footer   |  footer_len: u32  |
++----------+-----------+-------------------+
+
+# Term Dictionary (Composite File)
+
++---------+---------------------------+------------------------+
+|   fst   | term_info_store           |   footer_len:  u64     |
++---------+---------------------------+------------------------+
+
+During a merge the term info store need to fit in memory.
+It has a cost of n bytes per term.
+
+# term_info_store
++-------------------+---------------------------+------------------------+
+|  len_block_meta   |  block_meta               |  term_infos            |
++-------------------+---------------------------+------------------------+
+
+# inverted_index
++------------------------+---------------------------+------------------------+
+|  total_num_tokens: u64 |  posting_lists.. |  term_infos            |
++------------------------+---------------------------+------------------------+
+
+# postings lists
++------------------------+---------------------------+------------------------+
+|    
++
+
+# composite file
++----------------+-----+----------------+----------------------+----------------+
+| field file 1   | ... | field field n  |composite file footer | footer len: u32| 
++----------------+-----+----------------+----------------------+----------------+
+
+# composite file footer
+
++-----------------+---------------------------------------+
+|num fields: vint | (file_addr, offset_delta: vint) []... |
++-----------------+---------------------------------------+
+
+# FileAddr 
++--------------+--------------+
+|  field: u32  |  idx: VInt   |
++--------------+--------------+
+
+# Posting lists 
++-----------------------------------------+
+| skip_reader 
++-----------------------------------------+

examples/basic_search.rs

@@ -112,18 +112,6 @@ fn main() -> tantivy::Result<()> {
             limbs and branches that arch over the pool"
     ));
 
-    index_writer.add_document(doc!(
-    title => "Of Mice and Men",
-    body => "A few miles south of Soledad, the Salinas River drops in close to the hillside \
-            bank and runs deep and green. The water is warm too, for it has slipped twinkling \
-            over the yellow sands in the sunlight before reaching the narrow pool. On one \
-            side of the river the golden foothill slopes curve up to the strong and rocky \
-            Gabilan Mountains, but on the valley side the water is lined with trees—willows \
-            fresh and green with every spring, carrying in their lower leaf junctures the \
-            debris of the winter’s flooding; and sycamores with mottled, white, recumbent \
-            limbs and branches that arch over the pool"
-    ));
-
     // Multivalued field just need to be repeated.
     index_writer.add_document(doc!(
     title => "Frankenstein",

examples/custom_collector.rs

@@ -14,7 +14,7 @@ use tantivy::fastfield::FastFieldReader;
 use tantivy::query::QueryParser;
 use tantivy::schema::Field;
 use tantivy::schema::{Schema, FAST, INDEXED, TEXT};
-use tantivy::{doc, Index, SegmentReader, TantivyError};
+use tantivy::{doc, Index, Score, SegmentReader, TantivyError};
 
 #[derive(Default)]
 struct Stats {
@@ -114,7 +114,7 @@ struct StatsSegmentCollector {
 impl SegmentCollector for StatsSegmentCollector {
     type Fruit = Option<Stats>;
 
-    fn collect(&mut self, doc: u32, _score: f32) {
+    fn collect(&mut self, doc: u32, _score: Score) {
         let value = self.fast_field_reader.get(doc) as f64;
         self.stats.count += 1;
         self.stats.sum += value;

query-grammar/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy-query-grammar"
-version = "0.13.0"
+version = "0.14.0-dev"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]

query-grammar/src/occur.rs

@@ -31,22 +31,12 @@ impl Occur {
 
     /// Compose two occur values.
     pub fn compose(left: Occur, right: Occur) -> Occur {
-        match left {
-            Occur::Should => right,
-            Occur::Must => {
-                if right == Occur::MustNot {
-                    Occur::MustNot
-                } else {
-                    Occur::Must
-                }
-            }
-            Occur::MustNot => {
-                if right == Occur::MustNot {
-                    Occur::Must
-                } else {
-                    Occur::MustNot
-                }
-            }
+        match (left, right) {
+            (Occur::Should, _) => right,
+            (Occur::Must, Occur::MustNot) => Occur::MustNot,
+            (Occur::Must, _) => Occur::Must,
+            (Occur::MustNot, Occur::MustNot) => Occur::Must,
+            (Occur::MustNot, _) => Occur::MustNot,
         }
     }
 }
@@ -56,3 +46,27 @@ impl fmt::Display for Occur {
         f.write_char(self.to_char())
     }
 }
+
+#[cfg(test)]
+mod test {
+    use crate::Occur;
+
+    #[test]
+    fn test_occur_compose() {
+        assert_eq!(Occur::compose(Occur::Should, Occur::Should), Occur::Should);
+        assert_eq!(Occur::compose(Occur::Should, Occur::Must), Occur::Must);
+        assert_eq!(
+            Occur::compose(Occur::Should, Occur::MustNot),
+            Occur::MustNot
+        );
+        assert_eq!(Occur::compose(Occur::Must, Occur::Should), Occur::Must);
+        assert_eq!(Occur::compose(Occur::Must, Occur::Must), Occur::Must);
+        assert_eq!(Occur::compose(Occur::Must, Occur::MustNot), Occur::MustNot);
+        assert_eq!(
+            Occur::compose(Occur::MustNot, Occur::Should),
+            Occur::MustNot
+        );
+        assert_eq!(Occur::compose(Occur::MustNot, Occur::Must), Occur::MustNot);
+        assert_eq!(Occur::compose(Occur::MustNot, Occur::MustNot), Occur::Must);
+    }
+}

query-grammar/src/query_grammar.rs

@@ -9,8 +9,10 @@ use combine::{
 
 fn field<'a>() -> impl Parser<&'a str, Output = String> {
     (
-        letter(),
-        many(satisfy(|c: char| c.is_alphanumeric() || c == '_')),
+        (letter().or(char('_'))),
+        many(satisfy(|c: char| {
+            c.is_alphanumeric() || c == '_' || c == '-'
+        })),
     )
         .skip(char(':'))
         .map(|(s1, s2): (char, String)| format!("{}{}", s1, s2))
@@ -180,7 +182,7 @@ fn occur_leaf<'a>() -> impl Parser<&'a str, Output = (Option<Occur>, UserInputAS
     (optional(occur_symbol()), boosted_leaf())
 }
 
-fn positive_float_number<'a>() -> impl Parser<&'a str, Output = f32> {
+fn positive_float_number<'a>() -> impl Parser<&'a str, Output = f64> {
     (many1(digit()), optional((char('.'), many1(digit())))).map(
         |(int_part, decimal_part_opt): (String, Option<(char, String)>)| {
             let mut float_str = int_part;
@@ -188,18 +190,18 @@ fn positive_float_number<'a>() -> impl Parser<&'a str, Output = f32> {
                 float_str.push(chr);
                 float_str.push_str(&decimal_str);
             }
-            float_str.parse::<f32>().unwrap()
+            float_str.parse::<f64>().unwrap()
         },
     )
 }
 
-fn boost<'a>() -> impl Parser<&'a str, Output = f32> {
+fn boost<'a>() -> impl Parser<&'a str, Output = f64> {
     (char('^'), positive_float_number()).map(|(_, boost)| boost)
 }
 
 fn boosted_leaf<'a>() -> impl Parser<&'a str, Output = UserInputAST> {
     (leaf(), optional(boost())).map(|(leaf, boost_opt)| match boost_opt {
-        Some(boost) if (boost - 1.0).abs() > std::f32::EPSILON => {
+        Some(boost) if (boost - 1.0).abs() > std::f64::EPSILON => {
             UserInputAST::Boost(Box::new(leaf), boost)
         }
         _ => leaf,
@@ -279,14 +281,16 @@ pub fn parse_to_ast<'a>() -> impl Parser<&'a str, Output = UserInputAST> {
 #[cfg(test)]
 mod test {
 
+    type TestParseResult = Result<(), StringStreamError>;
+
     use super::*;
     use combine::parser::Parser;
 
-    pub fn nearly_equals(a: f32, b: f32) -> bool {
+    pub fn nearly_equals(a: f64, b: f64) -> bool {
         (a - b).abs() < 0.0005 * (a + b).abs()
     }
 
-    fn assert_nearly_equals(expected: f32, val: f32) {
+    fn assert_nearly_equals(expected: f64, val: f64) {
         assert!(
             nearly_equals(val, expected),
             "Got {}, expected {}.",
@@ -296,14 +300,15 @@ mod test {
     }
 
     #[test]
-    fn test_occur_symbol() {
-        assert_eq!(super::occur_symbol().parse("-"), Ok((Occur::MustNot, "")));
-        assert_eq!(super::occur_symbol().parse("+"), Ok((Occur::Must, "")));
+    fn test_occur_symbol() -> TestParseResult {
+        assert_eq!(super::occur_symbol().parse("-")?, (Occur::MustNot, ""));
+        assert_eq!(super::occur_symbol().parse("+")?, (Occur::Must, ""));
+        Ok(())
     }
 
     #[test]
     fn test_positive_float_number() {
-        fn valid_parse(float_str: &str, expected_val: f32, expected_remaining: &str) {
+        fn valid_parse(float_str: &str, expected_val: f64, expected_remaining: &str) {
             let (val, remaining) = positive_float_number().parse(float_str).unwrap();
             assert_eq!(remaining, expected_remaining);
             assert_nearly_equals(val, expected_val);
@@ -311,9 +316,9 @@ mod test {
         fn error_parse(float_str: &str) {
             assert!(positive_float_number().parse(float_str).is_err());
         }
-        valid_parse("1.0", 1.0f32, "");
-        valid_parse("1", 1.0f32, "");
-        valid_parse("0.234234 aaa", 0.234234f32, " aaa");
+        valid_parse("1.0", 1.0, "");
+        valid_parse("1", 1.0, "");
+        valid_parse("0.234234 aaa", 0.234234f64, " aaa");
         error_parse(".3332");
         error_parse("1.");
         error_parse("-1.");
@@ -410,6 +415,25 @@ mod test {
         assert_eq!(format!("{:?}", ast), "\"abc\"");
     }
 
+    #[test]
+    fn test_field_name() -> TestParseResult {
+        assert_eq!(
+            super::field().parse("my-field-name:a")?,
+            ("my-field-name".to_string(), "a")
+        );
+        assert_eq!(
+            super::field().parse("my_field_name:a")?,
+            ("my_field_name".to_string(), "a")
+        );
+        assert!(super::field().parse(":a").is_err());
+        assert!(super::field().parse("-my_field:a").is_err());
+        assert_eq!(
+            super::field().parse("_my_field:a")?,
+            ("_my_field".to_string(), "a")
+        );
+        Ok(())
+    }
+
     #[test]
     fn test_range_parser() {
         // testing the range() parser separately

query-grammar/src/user_input_ast.rs

@@ -87,7 +87,7 @@ impl UserInputBound {
 pub enum UserInputAST {
     Clause(Vec<(Option<Occur>, UserInputAST)>),
     Leaf(Box<UserInputLeaf>),
-    Boost(Box<UserInputAST>, f32),
+    Boost(Box<UserInputAST>, f64),
 }
 
 impl UserInputAST {

src/collector/count_collector.rs

@@ -96,18 +96,18 @@ mod tests {
         }
         {
             let mut count_collector = SegmentCountCollector::default();
-            count_collector.collect(0u32, 1f32);
+            count_collector.collect(0u32, 1.0);
             assert_eq!(count_collector.harvest(), 1);
         }
         {
             let mut count_collector = SegmentCountCollector::default();
-            count_collector.collect(0u32, 1f32);
+            count_collector.collect(0u32, 1.0);
             assert_eq!(count_collector.harvest(), 1);
         }
         {
             let mut count_collector = SegmentCountCollector::default();
-            count_collector.collect(0u32, 1f32);
-            count_collector.collect(1u32, 1f32);
+            count_collector.collect(0u32, 1.0);
+            count_collector.collect(1u32, 1.0);
             assert_eq!(count_collector.harvest(), 2);
         }
     }

src/collector/custom_score_top_collector.rs

@@ -46,7 +46,7 @@ pub trait CustomScorer<TScore>: Sync {
 
 impl<TCustomScorer, TScore> Collector for CustomScoreTopCollector<TCustomScorer, TScore>
 where
-    TCustomScorer: CustomScorer<TScore>,
+    TCustomScorer: CustomScorer<TScore> + Send + Sync,
     TScore: 'static + PartialOrd + Clone + Send + Sync,
 {
     type Fruit = Vec<(TScore, DocAddress)>;

src/collector/facet_collector.rs

@@ -472,7 +472,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let num_facets: usize = 3 * 4 * 5;
         let facets: Vec<Facet> = (0..num_facets)
             .map(|mut n| {
@@ -531,7 +531,7 @@ mod tests {
         let facet_field = schema_builder.add_facet_field("facets");
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(
             facet_field => Facet::from_text(&"/subjects/A/a"),
             facet_field => Facet::from_text(&"/subjects/B/a"),
@@ -550,12 +550,12 @@ mod tests {
     }
 
     #[test]
-    fn test_doc_search_by_facet() {
+    fn test_doc_search_by_facet() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let facet_field = schema_builder.add_facet_field("facet");
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         index_writer.add_document(doc!(
             facet_field => Facet::from_text(&"/A/A"),
         ));
@@ -568,8 +568,8 @@ mod tests {
         index_writer.add_document(doc!(
             facet_field => Facet::from_text(&"/D/C/A"),
         ));
-        index_writer.commit().unwrap();
-        let reader = index.reader().unwrap();
+        index_writer.commit()?;
+        let reader = index.reader()?;
         let searcher = reader.searcher();
         assert_eq!(searcher.num_docs(), 4);
 
@@ -586,17 +586,17 @@ mod tests {
         assert_eq!(count_facet("/A/C"), 1);
         assert_eq!(count_facet("/A/C/A"), 1);
         assert_eq!(count_facet("/C/A"), 0);
+
+        let query_parser = QueryParser::for_index(&index, vec![]);
         {
-            let query_parser = QueryParser::for_index(&index, vec![]);
-            {
-                let query = query_parser.parse_query("facet:/A/B").unwrap();
-                assert_eq!(1, searcher.search(&query, &Count).unwrap());
-            }
-            {
-                let query = query_parser.parse_query("facet:/A").unwrap();
-                assert_eq!(3, searcher.search(&query, &Count).unwrap());
-            }
+            let query = query_parser.parse_query("facet:/A/B")?;
+            assert_eq!(1, searcher.search(&query, &Count).unwrap());
         }
+        {
+            let query = query_parser.parse_query("facet:/A")?;
+            assert_eq!(3, searcher.search(&query, &Count)?);
+        }
+        Ok(())
     }
 
     #[test]
@@ -631,7 +631,7 @@ mod tests {
             .collect();
         docs[..].shuffle(&mut thread_rng());
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         for doc in docs {
             index_writer.add_document(doc);
         }
@@ -684,7 +684,7 @@ mod bench {
         // 40425 docs
         docs[..].shuffle(&mut thread_rng());
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         for doc in docs {
             index_writer.add_document(doc);
         }

src/collector/int_facet_collector.rs

@@ -89,7 +89,7 @@ mod tests {
         let index = Index::create_in_ram(schema.clone());
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             {
                 for i in 0u64..10u64 {
                     index_writer.add_document(doc!(

src/collector/mod.rs

@@ -133,7 +133,7 @@ impl<T> Fruit for T where T: Send + downcast_rs::Downcast {}
 /// The collection logic itself is in the `SegmentCollector`.
 ///
 /// Segments are not guaranteed to be visited in any specific order.
-pub trait Collector: Sync {
+pub trait Collector: Sync + Send {
     /// `Fruit` is the type for the result of our collection.
     /// e.g. `usize` for the `Count` collector.
     type Fruit: Fruit;

src/collector/multi_collector.rs

@@ -55,7 +55,7 @@ impl<TCollector: Collector> Collector for CollectorWrapper<TCollector> {
 impl SegmentCollector for Box<dyn BoxableSegmentCollector> {
     type Fruit = Box<dyn Fruit>;
 
-    fn collect(&mut self, doc: u32, score: f32) {
+    fn collect(&mut self, doc: u32, score: Score) {
         self.as_mut().collect(doc, score);
     }
 
@@ -65,7 +65,7 @@ impl SegmentCollector for Box<dyn BoxableSegmentCollector> {
 }
 
 pub trait BoxableSegmentCollector {
-    fn collect(&mut self, doc: u32, score: f32);
+    fn collect(&mut self, doc: u32, score: Score);
     fn harvest_from_box(self: Box<Self>) -> Box<dyn Fruit>;
 }
 
@@ -74,7 +74,7 @@ pub struct SegmentCollectorWrapper<TSegmentCollector: SegmentCollector>(TSegment
 impl<TSegmentCollector: SegmentCollector> BoxableSegmentCollector
     for SegmentCollectorWrapper<TSegmentCollector>
 {
-    fn collect(&mut self, doc: u32, score: f32) {
+    fn collect(&mut self, doc: u32, score: Score) {
         self.0.collect(doc, score);
     }
 
@@ -259,7 +259,7 @@ mod tests {
 
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text=>"abc"));
             index_writer.add_document(doc!(text=>"abc abc abc"));
             index_writer.add_document(doc!(text=>"abc abc"));

src/collector/tests.rs

@@ -206,7 +206,7 @@ impl Collector for BytesFastFieldTestCollector {
 impl SegmentCollector for BytesFastFieldSegmentCollector {
     type Fruit = Vec<u8>;
 
-    fn collect(&mut self, doc: u32, _score: f32) {
+    fn collect(&mut self, doc: u32, _score: Score) {
         let data = self.reader.get_bytes(doc);
         self.vals.extend(data);
     }

src/collector/top_score_collector.rs

@@ -38,7 +38,7 @@ use std::fmt;
 /// let schema = schema_builder.build();
 /// let index = Index::create_in_ram(schema);
 ///
-/// let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
 /// index_writer.add_document(doc!(title => "The Name of the Wind"));
 /// index_writer.add_document(doc!(title => "The Diary of Muadib"));
 /// index_writer.add_document(doc!(title => "A Dairy Cow"));
@@ -52,8 +52,8 @@ use std::fmt;
 /// let query = query_parser.parse_query("diary").unwrap();
 /// let top_docs = searcher.search(&query, &TopDocs::with_limit(2)).unwrap();
 ///
-/// assert_eq!(&top_docs[0], &(0.7261542, DocAddress(0, 1)));
-/// assert_eq!(&top_docs[1], &(0.6099695, DocAddress(0, 3)));
+/// assert_eq!(top_docs[0].1, DocAddress(0, 1));
+/// assert_eq!(top_docs[1].1, DocAddress(0, 3));
 /// ```
 pub struct TopDocs(TopCollector<Score>);
 
@@ -123,7 +123,7 @@ impl TopDocs {
     /// let schema = schema_builder.build();
     /// let index = Index::create_in_ram(schema);
     ///
-    /// let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+    /// let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
     /// index_writer.add_document(doc!(title => "The Name of the Wind"));
     /// index_writer.add_document(doc!(title => "The Diary of Muadib"));
     /// index_writer.add_document(doc!(title => "A Dairy Cow"));
@@ -139,8 +139,8 @@ impl TopDocs {
     /// let top_docs = searcher.search(&query, &TopDocs::with_limit(2).and_offset(1)).unwrap();
     ///
     /// assert_eq!(top_docs.len(), 2);
-    /// assert_eq!(&top_docs[0], &(0.5204813, DocAddress(0, 4)));
-    /// assert_eq!(&top_docs[1], &(0.4793185, DocAddress(0, 3)));
+    /// assert_eq!(top_docs[0].1, DocAddress(0, 4));
+    /// assert_eq!(top_docs[1].1, DocAddress(0, 3));
     /// ```
     pub fn and_offset(self, offset: usize) -> TopDocs {
         TopDocs(self.0.and_offset(offset))
@@ -163,7 +163,7 @@ impl TopDocs {
     /// #   let schema = schema_builder.build();
     /// #  
     /// #   let index = Index::create_in_ram(schema);
-    /// #   let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
+    /// #   let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
     /// #   index_writer.add_document(doc!(title => "The Name of the Wind", rating => 92u64));
     /// #   index_writer.add_document(doc!(title => "The Diary of Muadib", rating => 97u64));
     /// #   index_writer.add_document(doc!(title => "A Dairy Cow", rating => 63u64));
@@ -264,7 +264,7 @@ impl TopDocs {
     /// fn create_index() -> tantivy::Result<Index> {
     ///   let schema = create_schema();
     ///   let index = Index::create_in_ram(schema);
-    ///   let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
+    ///   let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
     ///   let product_name = index.schema().get_field("product_name").unwrap();
     ///   let popularity: Field = index.schema().get_field("popularity").unwrap();
     ///   index_writer.add_document(doc!(product_name => "The Diary of Muadib", popularity => 1u64));
@@ -303,7 +303,7 @@ impl TopDocs {
     ///                 let popularity: u64 = popularity_reader.get(doc);
     ///                 // Well.. For the sake of the example we use a simple logarithm
     ///                 // function.
-    ///                 let popularity_boost_score = ((2u64 + popularity) as f32).log2();
+    ///                 let popularity_boost_score = ((2u64 + popularity) as Score).log2();
     ///                 popularity_boost_score * original_score
     ///             }
     ///           });
@@ -324,7 +324,7 @@ impl TopDocs {
     where
         TScore: 'static + Send + Sync + Clone + PartialOrd,
         TScoreSegmentTweaker: ScoreSegmentTweaker<TScore> + 'static,
-        TScoreTweaker: ScoreTweaker<TScore, Child = TScoreSegmentTweaker>,
+        TScoreTweaker: ScoreTweaker<TScore, Child = TScoreSegmentTweaker> + Send + Sync,
     {
         TweakedScoreTopCollector::new(score_tweaker, self.0.into_tscore())
     }
@@ -371,7 +371,7 @@ impl TopDocs {
     /// # fn main() -> tantivy::Result<()> {
     /// #   let schema = create_schema();
     /// #   let index = Index::create_in_ram(schema);
-    /// #   let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
+    /// #   let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
     /// #   let product_name = index.schema().get_field("product_name").unwrap();
     /// #   
     /// let popularity: Field = index.schema().get_field("popularity").unwrap();
@@ -438,7 +438,7 @@ impl TopDocs {
     where
         TScore: 'static + Send + Sync + Clone + PartialOrd,
         TCustomSegmentScorer: CustomSegmentScorer<TScore> + 'static,
-        TCustomScorer: CustomScorer<TScore, Child = TCustomSegmentScorer>,
+        TCustomScorer: CustomScorer<TScore, Child = TCustomSegmentScorer> + Send + Sync,
     {
         CustomScoreTopCollector::new(custom_score, self.0.into_tscore())
     }
@@ -479,7 +479,7 @@ impl Collector for TopDocs {
         let mut heap: BinaryHeap<ComparableDoc<Score, DocId>> = BinaryHeap::with_capacity(heap_len);
 
         if let Some(delete_bitset) = reader.delete_bitset() {
-            let mut threshold = f32::MIN;
+            let mut threshold = Score::MIN;
             weight.for_each_pruning(threshold, reader, &mut |doc, score| {
                 if delete_bitset.is_deleted(doc) {
                     return threshold;
@@ -491,16 +491,16 @@ impl Collector for TopDocs {
                 if heap.len() < heap_len {
                     heap.push(heap_item);
                     if heap.len() == heap_len {
-                        threshold = heap.peek().map(|el| el.feature).unwrap_or(f32::MIN);
+                        threshold = heap.peek().map(|el| el.feature).unwrap_or(Score::MIN);
                     }
                     return threshold;
                 }
                 *heap.peek_mut().unwrap() = heap_item;
-                threshold = heap.peek().map(|el| el.feature).unwrap_or(std::f32::MIN);
+                threshold = heap.peek().map(|el| el.feature).unwrap_or(Score::MIN);
                 threshold
             })?;
         } else {
-            weight.for_each_pruning(f32::MIN, reader, &mut |doc, score| {
+            weight.for_each_pruning(Score::MIN, reader, &mut |doc, score| {
                 let heap_item = ComparableDoc {
                     feature: score,
                     doc,
@@ -509,17 +509,17 @@ impl Collector for TopDocs {
                     heap.push(heap_item);
                     // TODO the threshold is suboptimal for heap.len == heap_len
                     if heap.len() == heap_len {
-                        return heap.peek().map(|el| el.feature).unwrap_or(f32::MIN);
+                        return heap.peek().map(|el| el.feature).unwrap_or(Score::MIN);
                     } else {
-                        return f32::MIN;
+                        return Score::MIN;
                     }
                 }
                 *heap.peek_mut().unwrap() = heap_item;
-                heap.peek().map(|el| el.feature).unwrap_or(std::f32::MIN)
+                heap.peek().map(|el| el.feature).unwrap_or(Score::MIN)
             })?;
         }
 
-        let fruit: Vec<(Score, DocAddress)> = heap
+        let fruit = heap
             .into_sorted_vec()
             .into_iter()
             .map(|cid| (cid.feature, DocAddress(segment_ord, cid.doc)))
@@ -561,7 +561,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
             index_writer.add_document(doc!(text_field=>"Hello happy tax payer."));
             index_writer.add_document(doc!(text_field=>"Droopy says hello happy tax payer"));
             index_writer.add_document(doc!(text_field=>"I like Droopy"));
@@ -570,6 +570,13 @@ mod tests {
         index
     }
 
+    fn assert_results_equals(results: &[(Score, DocAddress)], expected: &[(Score, DocAddress)]) {
+        for (result, expected) in results.iter().zip(expected.iter()) {
+            assert_eq!(result.1, expected.1);
+            crate::assert_nearly_equals!(result.0, expected.0);
+        }
+    }
+
     #[test]
     fn test_top_collector_not_at_capacity() {
         let index = make_index();
@@ -582,13 +589,13 @@ mod tests {
             .searcher()
             .search(&text_query, &TopDocs::with_limit(4))
             .unwrap();
-        assert_eq!(
-            score_docs,
-            vec![
+        assert_results_equals(
+            &score_docs,
+            &[
                 (0.81221175, DocAddress(0u32, 1)),
                 (0.5376842, DocAddress(0u32, 2)),
-                (0.48527452, DocAddress(0, 0))
-            ]
+                (0.48527452, DocAddress(0, 0)),
+            ],
         );
     }
 
@@ -604,7 +611,7 @@ mod tests {
             .searcher()
             .search(&text_query, &TopDocs::with_limit(4).and_offset(2))
             .unwrap();
-        assert_eq!(score_docs, vec![(0.48527452, DocAddress(0, 0))]);
+        assert_results_equals(&score_docs[..], &[(0.48527452, DocAddress(0, 0))]);
     }
 
     #[test]
@@ -619,12 +626,12 @@ mod tests {
             .searcher()
             .search(&text_query, &TopDocs::with_limit(2))
             .unwrap();
-        assert_eq!(
-            score_docs,
-            vec![
+        assert_results_equals(
+            &score_docs,
+            &[
                 (0.81221175, DocAddress(0u32, 1)),
                 (0.5376842, DocAddress(0u32, 2)),
-            ]
+            ],
         );
     }
 
@@ -640,12 +647,12 @@ mod tests {
             .searcher()
             .search(&text_query, &TopDocs::with_limit(2).and_offset(1))
             .unwrap();
-        assert_eq!(
-            score_docs,
-            vec![
+        assert_results_equals(
+            &score_docs[..],
+            &[
                 (0.5376842, DocAddress(0u32, 2)),
-                (0.48527452, DocAddress(0, 0))
-            ]
+                (0.48527452, DocAddress(0, 0)),
+            ],
         );
     }
 
@@ -706,8 +713,8 @@ mod tests {
         let top_collector = TopDocs::with_limit(4).order_by_u64_field(size);
         let top_docs: Vec<(u64, DocAddress)> = searcher.search(&query, &top_collector).unwrap();
         assert_eq!(
-            top_docs,
-            vec![
+            &top_docs[..],
+            &[
                 (64, DocAddress(0, 1)),
                 (16, DocAddress(0, 2)),
                 (12, DocAddress(0, 0))
@@ -814,7 +821,7 @@ mod tests {
     ) -> (Index, Box<dyn Query>) {
         let index = Index::create_in_ram(schema);
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
         doc_adder(&mut index_writer);
         index_writer.commit().unwrap();
         let query_parser = QueryParser::for_index(&index, vec![query_field]);

src/collector/tweak_score_top_collector.rs

@@ -49,7 +49,7 @@ pub trait ScoreTweaker<TScore>: Sync {
 
 impl<TScoreTweaker, TScore> Collector for TweakedScoreTopCollector<TScoreTweaker, TScore>
 where
-    TScoreTweaker: ScoreTweaker<TScore>,
+    TScoreTweaker: ScoreTweaker<TScore> + Send + Sync,
     TScore: 'static + PartialOrd + Clone + Send + Sync,
 {
     type Fruit = Vec<(TScore, DocAddress)>;

src/common/mod.rs

@@ -10,7 +10,9 @@ pub(crate) use self::bitset::TinySet;
 pub(crate) use self::composite_file::{CompositeFile, CompositeWrite};
 pub use self::counting_writer::CountingWriter;
 pub use self::serialize::{BinarySerializable, FixedSize};
-pub use self::vint::{read_u32_vint, serialize_vint_u32, write_u32_vint, VInt};
+pub use self::vint::{
+    read_u32_vint, read_u32_vint_no_advance, serialize_vint_u32, write_u32_vint, VInt,
+};
 pub use byteorder::LittleEndian as Endianness;
 
 /// Segment's max doc must be `< MAX_DOC_LIMIT`.

src/common/vint.rs

@@ -10,7 +10,7 @@ pub struct VInt(pub u64);
 
 const STOP_BIT: u8 = 128;
 
-pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
+pub fn serialize_vint_u32(val: u32, buf: &mut [u8; 8]) -> &[u8] {
     const START_2: u64 = 1 << 7;
     const START_3: u64 = 1 << 14;
     const START_4: u64 = 1 << 21;
@@ -29,7 +29,7 @@ pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
 
     let val = u64::from(val);
     const STOP_BIT: u64 = 128u64;
-    match val {
+    let (res, num_bytes) = match val {
         0..=STOP_1 => (val | STOP_BIT, 1),
         START_2..=STOP_2 => (
             (val & MASK_1) | ((val & MASK_2) << 1) | (STOP_BIT << (8)),
@@ -56,7 +56,9 @@ pub fn serialize_vint_u32(val: u32) -> (u64, usize) {
                 | (STOP_BIT << (8 * 4)),
             5,
         ),
-    }
+    };
+    LittleEndian::write_u64(&mut buf[..], res);
+    &buf[0..num_bytes]
 }
 
 /// Returns the number of bytes covered by a
@@ -85,23 +87,26 @@ fn vint_len(data: &[u8]) -> usize {
 /// If the buffer does not start by a valid
 /// vint payload
 pub fn read_u32_vint(data: &mut &[u8]) -> u32 {
-    let vlen = vint_len(*data);
+    let (result, vlen) = read_u32_vint_no_advance(*data);
+    *data = &data[vlen..];
+    result
+}
+
+pub fn read_u32_vint_no_advance(data: &[u8]) -> (u32, usize) {
+    let vlen = vint_len(data);
     let mut result = 0u32;
     let mut shift = 0u64;
     for &b in &data[..vlen] {
         result |= u32::from(b & 127u8) << shift;
         shift += 7;
     }
-    *data = &data[vlen..];
-    result
+    (result, vlen)
 }
-
 /// Write a `u32` as a vint payload.
 pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
-    let (val, num_bytes) = serialize_vint_u32(val);
-    let mut buffer = [0u8; 8];
-    LittleEndian::write_u64(&mut buffer, val);
-    writer.write_all(&buffer[..num_bytes])
+    let mut buf = [0u8; 8];
+    let data = serialize_vint_u32(val, &mut buf);
+    writer.write_all(&data)
 }
 
 impl VInt {
@@ -172,7 +177,6 @@ mod tests {
     use super::serialize_vint_u32;
     use super::VInt;
     use crate::common::BinarySerializable;
-    use byteorder::{ByteOrder, LittleEndian};
 
     fn aux_test_vint(val: u64) {
         let mut v = [14u8; 10];
@@ -208,12 +212,10 @@ mod tests {
 
     fn aux_test_serialize_vint_u32(val: u32) {
         let mut buffer = [0u8; 10];
-        let mut buffer2 = [0u8; 10];
+        let mut buffer2 = [0u8; 8];
         let len_vint = VInt(val as u64).serialize_into(&mut buffer);
-        let (vint, len) = serialize_vint_u32(val);
-        assert_eq!(len, len_vint, "len wrong for val {}", val);
-        LittleEndian::write_u64(&mut buffer2, vint);
-        assert_eq!(&buffer[..len], &buffer2[..len], "array wrong for {}", val);
+        let res2 = serialize_vint_u32(val, &mut buffer2);
+        assert_eq!(&buffer[..len_vint], res2, "array wrong for {}", val);
     }
 
     #[test]

src/core/index.rs

@@ -300,6 +300,15 @@ impl Index {
         )
     }
 
+    /// Helper to create an index writer for tests.
+    ///
+    /// That index writer only simply has a single thread and a heap of 5 MB.
+    /// Using a single thread gives us a deterministic allocation of DocId.
+    #[cfg(test)]
+    pub fn writer_for_tests(&self) -> crate::Result<IndexWriter> {
+        self.writer_with_num_threads(1, 10_000_000)
+    }
+
     /// Creates a multithreaded writer
     ///
     /// Tantivy will automatically define the number of threads to use.
@@ -502,7 +511,7 @@ mod tests {
             let schema = throw_away_schema();
             let field = schema.get_field("num_likes").unwrap();
             let mut index = Index::create_from_tempdir(schema).unwrap();
-            let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut writer = index.writer_for_tests().unwrap();
             writer.commit().unwrap();
             let reader = index
                 .reader_builder()
@@ -539,23 +548,33 @@ mod tests {
             test_index_on_commit_reload_policy_aux(field, &write_index, &reader);
         }
     }
-
     fn test_index_on_commit_reload_policy_aux(field: Field, index: &Index, reader: &IndexReader) {
         let mut reader_index = reader.index();
         let (sender, receiver) = crossbeam::channel::unbounded();
         let _watch_handle = reader_index.directory_mut().watch(Box::new(move || {
             let _ = sender.send(());
         }));
-        let mut writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut writer = index.writer_for_tests().unwrap();
         assert_eq!(reader.searcher().num_docs(), 0);
         writer.add_document(doc!(field=>1u64));
         writer.commit().unwrap();
-        assert!(receiver.recv().is_ok());
-        assert_eq!(reader.searcher().num_docs(), 1);
+        // We need a loop here because it is possible for notify to send more than
+        // one modify event. It was observed on CI on MacOS.
+        loop {
+            assert!(receiver.recv().is_ok());
+            if reader.searcher().num_docs() == 1 {
+                break;
+            }
+        }
         writer.add_document(doc!(field=>2u64));
         writer.commit().unwrap();
-        assert!(receiver.recv().is_ok());
-        assert_eq!(reader.searcher().num_docs(), 2);
+        // ... Same as above
+        loop {
+            assert!(receiver.recv().is_ok());
+            if reader.searcher().num_docs() == 2 {
+                break;
+            }
+        }
     }
 
     // This test will not pass on windows, because windows

src/core/index_meta.rs

@@ -116,6 +116,7 @@ impl SegmentMeta {
             SegmentComponent::FASTFIELDS => ".fast".to_string(),
             SegmentComponent::FIELDNORMS => ".fieldnorm".to_string(),
             SegmentComponent::DELETE => format!(".{}.del", self.delete_opstamp().unwrap_or(0)),
+            SegmentComponent::FIELDSTATS => ".fieldstats".to_string(),
         });
         PathBuf::from(path)
     }

src/core/inverted_index_reader.rs

@@ -1,9 +1,7 @@
-use crate::common::BinarySerializable;
 use crate::directory::ReadOnlySource;
 use crate::positions::PositionReader;
 use crate::postings::TermInfo;
 use crate::postings::{BlockSegmentPostings, SegmentPostings};
-use crate::schema::FieldType;
 use crate::schema::IndexRecordOption;
 use crate::schema::Term;
 use crate::termdict::TermDictionary;
@@ -37,14 +35,12 @@ impl InvertedIndexReader {
         postings_source: ReadOnlySource,
         positions_source: ReadOnlySource,
         positions_idx_source: ReadOnlySource,
+        total_num_tokens: u64,
         record_option: IndexRecordOption,
     ) -> InvertedIndexReader {
-        let total_num_tokens_data = postings_source.slice(0, 8);
-        let mut total_num_tokens_cursor = total_num_tokens_data.as_slice();
-        let total_num_tokens = u64::deserialize(&mut total_num_tokens_cursor).unwrap_or(0u64);
         InvertedIndexReader {
             termdict,
-            postings_source: postings_source.slice_from(8),
+            postings_source,
             positions_source,
             positions_idx_source,
             record_option,
@@ -54,10 +50,7 @@ impl InvertedIndexReader {
 
     /// Creates an empty `InvertedIndexReader` object, which
     /// contains no terms at all.
-    pub fn empty(field_type: &FieldType) -> InvertedIndexReader {
-        let record_option = field_type
-            .get_index_record_option()
-            .unwrap_or(IndexRecordOption::Basic);
+    pub fn empty(record_option: IndexRecordOption) -> InvertedIndexReader {
         InvertedIndexReader {
             termdict: TermDictionary::empty(),
             postings_source: ReadOnlySource::empty(),
@@ -93,7 +86,7 @@ impl InvertedIndexReader {
         term_info: &TermInfo,
         block_postings: &mut BlockSegmentPostings,
     ) {
-        let offset = term_info.postings_offset as usize;
+        let offset = term_info.postings_start_offset as usize;
         let end_source = self.postings_source.len();
         let postings_slice = self.postings_source.slice(offset, end_source);
         block_postings.reset(term_info.doc_freq, postings_slice);
@@ -121,8 +114,10 @@ impl InvertedIndexReader {
         term_info: &TermInfo,
         requested_option: IndexRecordOption,
     ) -> BlockSegmentPostings {
-        let offset = term_info.postings_offset as usize;
-        let postings_data = self.postings_source.slice_from(offset);
+        let postings_data = self.postings_source.slice(
+            term_info.postings_start_offset as usize,
+            term_info.postings_end_offset as usize,
+        );
         BlockSegmentPostings::from_data(
             term_info.doc_freq,
             postings_data,

src/core/segment_component.rs

@@ -24,14 +24,17 @@ pub enum SegmentComponent {
     /// Accessing a document from the store is relatively slow, as it
     /// requires to decompress the entire block it belongs to.
     STORE,
+
     /// Bitset describing which document of the segment is deleted.
     DELETE,
+
+    FIELDSTATS,
 }
 
 impl SegmentComponent {
     /// Iterates through the components.
     pub fn iterator() -> slice::Iter<'static, SegmentComponent> {
-        static SEGMENT_COMPONENTS: [SegmentComponent; 8] = [
+        static SEGMENT_COMPONENTS: [SegmentComponent; 9] = [
             SegmentComponent::POSTINGS,
             SegmentComponent::POSITIONS,
             SegmentComponent::POSITIONSSKIP,
@@ -40,6 +43,7 @@ impl SegmentComponent {
             SegmentComponent::TERMS,
             SegmentComponent::STORE,
             SegmentComponent::DELETE,
+            SegmentComponent::FIELDSTATS,
         ];
         SEGMENT_COMPONENTS.iter()
     }

src/core/segment_reader.rs

@@ -1,4 +1,3 @@
-use crate::common::CompositeFile;
 use crate::common::HasLen;
 use crate::core::InvertedIndexReader;
 use crate::core::Segment;
@@ -9,13 +8,14 @@ use crate::fastfield::DeleteBitSet;
 use crate::fastfield::FacetReader;
 use crate::fastfield::FastFieldReaders;
 use crate::fieldnorm::{FieldNormReader, FieldNormReaders};
-use crate::schema::Field;
 use crate::schema::FieldType;
 use crate::schema::Schema;
+use crate::schema::{Field, IndexRecordOption};
 use crate::space_usage::SegmentSpaceUsage;
 use crate::store::StoreReader;
 use crate::termdict::TermDictionary;
 use crate::DocId;
+use crate::{common::CompositeFile, postings::FieldStats};
 use fail::fail_point;
 use std::collections::HashMap;
 use std::fmt;
@@ -49,6 +49,7 @@ pub struct SegmentReader {
     positions_idx_composite: CompositeFile,
     fast_fields_readers: Arc<FastFieldReaders>,
     fieldnorm_readers: FieldNormReaders,
+    field_stats: FieldStats,
 
     store_source: ReadOnlySource,
     delete_bitset_opt: Option<DeleteBitSet>,
@@ -125,17 +126,15 @@ impl SegmentReader {
     ///
     /// They are simply stored as a fast field, serialized in
     /// the `.fieldnorm` file of the segment.
-    pub fn get_fieldnorms_reader(&self, field: Field) -> FieldNormReader {
-        if let Some(fieldnorm_source) = self.fieldnorm_readers.get_field(field) {
-            fieldnorm_source
-        } else {
+    pub fn get_fieldnorms_reader(&self, field: Field) -> crate::Result<FieldNormReader> {
+        self.fieldnorm_readers.get_field(field).ok_or_else(|| {
             let field_name = self.schema.get_field_name(field);
             let err_msg = format!(
                 "Field norm not found for field {:?}. Was it market as indexed during indexing.",
                 field_name
             );
-            panic!(err_msg);
-        }
+            crate::TantivyError::SchemaError(err_msg)
+        })
     }
 
     /// Accessor to the segment's `StoreReader`.
@@ -179,7 +178,10 @@ impl SegmentReader {
             Arc::new(FastFieldReaders::load_all(&schema, &fast_fields_composite)?);
 
         let fieldnorm_data = segment.open_read(SegmentComponent::FIELDNORMS)?;
-        let fieldnorm_readers = FieldNormReaders::new(fieldnorm_data)?;
+        let fieldnorm_readers = FieldNormReaders::open(fieldnorm_data)?;
+
+        let field_stats_data = segment.open_read(SegmentComponent::FIELDSTATS)?;
+        let field_stats = FieldStats::from_source(field_stats_data.as_slice())?;
 
         let delete_bitset_opt = if segment.meta().has_deletes() {
             let delete_data = segment.open_read(SegmentComponent::DELETE)?;
@@ -196,6 +198,7 @@ impl SegmentReader {
             postings_composite,
             fast_fields_readers: fast_field_readers,
             fieldnorm_readers,
+            field_stats,
             segment_id: segment.id(),
             store_source,
             delete_bitset_opt,
@@ -212,6 +215,11 @@ impl SegmentReader {
     /// The field reader is in charge of iterating through the
     /// term dictionary associated to a specific field,
     /// and opening the posting list associated to any term.
+    ///
+    /// If the field is marked as index, a warn is logged and an empty `InvertedIndexReader`
+    /// is returned.
+    /// Similarly if the field is marked as indexed but no term has been indexed for the given
+    /// index. an empty `InvertedIndexReader` is returned (but no warning is logged).
     pub fn inverted_index(&self, field: Field) -> Arc<InvertedIndexReader> {
         if let Some(inv_idx_reader) = self
             .inv_idx_reader_cache
@@ -226,21 +234,21 @@ impl SegmentReader {
         let record_option_opt = field_type.get_index_record_option();
 
         if record_option_opt.is_none() {
-            panic!("Field {:?} does not seem indexed.", field_entry.name());
+            warn!("Field {:?} does not seem indexed.", field_entry.name());
         }
 
-        let record_option = record_option_opt.unwrap();
-
         let postings_source_opt = self.postings_composite.open_read(field);
 
-        if postings_source_opt.is_none() {
+        if postings_source_opt.is_none() || record_option_opt.is_none() {
             // no documents in the segment contained this field.
             // As a result, no data is associated to the inverted index.
             //
             // Returns an empty inverted index.
-            return Arc::new(InvertedIndexReader::empty(field_type));
+            let record_option = record_option_opt.unwrap_or(IndexRecordOption::Basic);
+            return Arc::new(InvertedIndexReader::empty(record_option));
         }
 
+        let record_option = record_option_opt.unwrap();
         let postings_source = postings_source_opt.unwrap();
 
         let termdict_source = self.termdict_composite.open_read(field).expect(
@@ -257,11 +265,17 @@ impl SegmentReader {
             .open_read(field)
             .expect("Index corrupted. Failed to open field positions in composite file.");
 
+        let total_num_tokens = self
+            .field_stats
+            .get(field)
+            .map(|field_stat| field_stat.num_tokens())
+            .unwrap_or(0u64);
         let inv_idx_reader = Arc::new(InvertedIndexReader::new(
             TermDictionary::from_source(&termdict_source),
             postings_source,
             positions_source,
             positions_idx_source,
+            total_num_tokens,
             record_option,
         ));
 
@@ -339,7 +353,7 @@ mod test {
         let name = schema.get_field("name").unwrap();
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(name => "tantivy"));
             index_writer.add_document(doc!(name => "horse"));
             index_writer.add_document(doc!(name => "jockey"));

src/directory/footer.rs

@@ -183,23 +183,23 @@ impl BinarySerializable for VersionedFooter {
         reader.read_exact(&mut buf[..])?;
         let mut cursor = &buf[..];
         let version = u32::deserialize(&mut cursor)?;
-        if version == 1 {
-            let crc32 = u32::deserialize(&mut cursor)?;
-            let store_compression = String::deserialize(&mut cursor)?;
-            Ok(VersionedFooter::V1 {
-                crc32,
-                store_compression,
-            })
-        } else if version == 2 {
-            let crc32 = u32::deserialize(&mut cursor)?;
-            let store_compression = String::deserialize(&mut cursor)?;
-            Ok(VersionedFooter::V2 {
-                crc32,
-                store_compression,
-            })
-        } else {
-            Ok(VersionedFooter::UnknownVersion)
+        if version != 1 && version != 2 {
+            return Ok(VersionedFooter::UnknownVersion);
         }
+        let crc32 = u32::deserialize(&mut cursor)?;
+        let store_compression = String::deserialize(&mut cursor)?;
+        Ok(if version == 1 {
+            VersionedFooter::V1 {
+                crc32,
+                store_compression,
+            }
+        } else {
+            assert_eq!(version, 2);
+            VersionedFooter::V2 {
+                crc32,
+                store_compression,
+            }
+        })
     }
 }
 
@@ -271,7 +271,11 @@ mod tests {
         let mut vec = Vec::new();
         let footer_proxy = FooterProxy::new(&mut vec);
         assert!(footer_proxy.terminate().is_ok());
-        assert_eq!(vec.len(), 167);
+        if crate::store::COMPRESSION == "lz4" {
+            assert_eq!(vec.len(), 158);
+        } else {
+            assert_eq!(vec.len(), 167);
+        }
         let footer = Footer::deserialize(&mut &vec[..]).unwrap();
         assert!(matches!(
            footer.versioned_footer,

src/directory/managed_directory.rs

@@ -1,4 +1,4 @@
-use crate::core::MANAGED_FILEPATH;
+use crate::core::{MANAGED_FILEPATH, META_FILEPATH};
 use crate::directory::error::{DeleteError, IOError, LockError, OpenReadError, OpenWriteError};
 use crate::directory::footer::{Footer, FooterProxy};
 use crate::directory::DirectoryLock;
@@ -246,13 +246,15 @@ impl ManagedDirectory {
     /// List files for which checksum does not match content
     pub fn list_damaged(&self) -> result::Result<HashSet<PathBuf>, OpenReadError> {
         let mut hashset = HashSet::new();
-        let managed_paths = self
+        let mut managed_paths = self
             .meta_informations
             .read()
             .expect("Managed directory rlock poisoned in list damaged.")
             .managed_paths
             .clone();
 
+        managed_paths.remove(*META_FILEPATH);
+
         for path in managed_paths.into_iter() {
             if !self.validate_checksum(&path)? {
                 hashset.insert(path);

src/directory/mmap_directory.rs

@@ -487,11 +487,13 @@ impl Directory for MmapDirectory {
         }
     }
 
-    fn atomic_write(&mut self, path: &Path, data: &[u8]) -> io::Result<()> {
+    fn atomic_write(&mut self, path: &Path, content: &[u8]) -> io::Result<()> {
         debug!("Atomic Write {:?}", path);
+        let mut tempfile = tempfile::Builder::new().tempfile_in(&self.inner.root_path)?;
+        tempfile.write_all(content)?;
+        tempfile.flush()?;
         let full_path = self.resolve_path(path);
-        let meta_file = atomicwrites::AtomicFile::new(full_path, atomicwrites::AllowOverwrite);
-        meta_file.write(|f| f.write_all(data))?;
+        tempfile.into_temp_path().persist(full_path)?;
         Ok(())
     }
 
@@ -652,7 +654,7 @@ mod tests {
         {
             let index = Index::create(mmap_directory.clone(), schema).unwrap();
 
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             let mut log_merge_policy = LogMergePolicy::default();
             log_merge_policy.set_min_merge_size(3);
             index_writer.set_merge_policy(Box::new(log_merge_policy));

src/directory/tests.rs

@@ -211,19 +211,19 @@ fn test_watch(directory: &mut dyn Directory) {
         .unwrap();
 
     for i in 0..10 {
-        assert_eq!(i, counter.load(SeqCst));
+        assert!(i <= counter.load(SeqCst));
         assert!(directory
             .atomic_write(Path::new("meta.json"), b"random_test_data_2")
             .is_ok());
         assert_eq!(receiver.recv_timeout(Duration::from_millis(500)), Ok(i));
-        assert_eq!(i + 1, counter.load(SeqCst));
+        assert!(i + 1 <= counter.load(SeqCst)); // notify can trigger more than once.
     }
     mem::drop(watch_handle);
     assert!(directory
         .atomic_write(Path::new("meta.json"), b"random_test_data")
         .is_ok());
     assert!(receiver.recv_timeout(Duration::from_millis(500)).is_ok());
-    assert_eq!(10, counter.load(SeqCst));
+    assert!(10 <= counter.load(SeqCst));
 }
 
 fn test_lock_non_blocking(directory: &mut dyn Directory) {

src/directory/watch_event_router.rs

@@ -5,7 +5,7 @@ use std::sync::RwLock;
 use std::sync::Weak;
 
 /// Type alias for callbacks registered when watching files of a `Directory`.
-pub type WatchCallback = Box<dyn Fn() -> () + Sync + Send>;
+pub type WatchCallback = Box<dyn Fn() + Sync + Send>;
 
 /// Helper struct to implement the watch method in `Directory` implementations.
 ///
@@ -32,7 +32,7 @@ impl WatchHandle {
 }
 
 impl WatchCallbackList {
-    /// Suscribes a new callback and returns a handle that controls the lifetime of the callback.
+    /// Subscribes a new callback and returns a handle that controls the lifetime of the callback.
     pub fn subscribe(&self, watch_callback: WatchCallback) -> WatchHandle {
         let watch_callback_arc = Arc::new(watch_callback);
         let watch_callback_weak = Arc::downgrade(&watch_callback_arc);

src/fastfield/bytes/mod.rs

@@ -15,7 +15,7 @@ mod tests {
         let field = schema_builder.add_bytes_field("bytesfield");
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(field=>vec![0u8, 1, 2, 3]));
         index_writer.add_document(doc!(field=>vec![]));
         index_writer.add_document(doc!(field=>vec![255u8]));

src/fastfield/delete.rs

@@ -9,6 +9,8 @@ use std::io::Write;
 /// Write a delete `BitSet`
 ///
 /// where `delete_bitset` is the set of deleted `DocId`.
+/// Warning: this function does not call terminate. The caller is in charge of
+/// closing the writer properly.
 pub fn write_delete_bitset(
     delete_bitset: &BitSet,
     max_doc: u32,
@@ -42,6 +44,24 @@ pub struct DeleteBitSet {
 }
 
 impl DeleteBitSet {
+    #[cfg(test)]
+    pub(crate) fn for_test(docs: &[DocId], max_doc: u32) -> DeleteBitSet {
+        use crate::directory::{Directory, RAMDirectory, TerminatingWrite};
+        use std::path::Path;
+        assert!(docs.iter().all(|&doc| doc < max_doc));
+        let mut bitset = BitSet::with_max_value(max_doc);
+        for &doc in docs {
+            bitset.insert(doc);
+        }
+        let mut directory = RAMDirectory::create();
+        let path = Path::new("dummydeletebitset");
+        let mut wrt = directory.open_write(path).unwrap();
+        write_delete_bitset(&bitset, max_doc, &mut wrt).unwrap();
+        wrt.terminate().unwrap();
+        let source = directory.open_read(path).unwrap();
+        Self::open(source)
+    }
+
     /// Opens a delete bitset given its data source.
     pub fn open(data: ReadOnlySource) -> DeleteBitSet {
         let num_deleted: usize = data
@@ -83,42 +103,35 @@ impl HasLen for DeleteBitSet {
 
 #[cfg(test)]
 mod tests {
-    use super::*;
-    use crate::directory::*;
-    use std::path::PathBuf;
+    use super::DeleteBitSet;
+    use crate::common::HasLen;
 
-    fn test_delete_bitset_helper(bitset: &BitSet, max_doc: u32) {
-        let test_path = PathBuf::from("test");
-        let mut directory = RAMDirectory::create();
-        {
-            let mut writer = directory.open_write(&*test_path).unwrap();
-            write_delete_bitset(bitset, max_doc, &mut writer).unwrap();
-            writer.terminate().unwrap();
+    #[test]
+    fn test_delete_bitset_empty() {
+        let delete_bitset = DeleteBitSet::for_test(&[], 10);
+        for doc in 0..10 {
+            assert_eq!(delete_bitset.is_deleted(doc), !delete_bitset.is_alive(doc));
         }
-        let source = directory.open_read(&test_path).unwrap();
-        let delete_bitset = DeleteBitSet::open(source);
-        for doc in 0..max_doc {
-            assert_eq!(bitset.contains(doc), delete_bitset.is_deleted(doc as DocId));
-        }
-        assert_eq!(delete_bitset.len(), bitset.len());
+        assert_eq!(delete_bitset.len(), 0);
     }
 
     #[test]
     fn test_delete_bitset() {
-        {
-            let mut bitset = BitSet::with_max_value(10);
-            bitset.insert(1);
-            bitset.insert(9);
-            test_delete_bitset_helper(&bitset, 10);
-        }
-        {
-            let mut bitset = BitSet::with_max_value(8);
-            bitset.insert(1);
-            bitset.insert(2);
-            bitset.insert(3);
-            bitset.insert(5);
-            bitset.insert(7);
-            test_delete_bitset_helper(&bitset, 8);
+        let delete_bitset = DeleteBitSet::for_test(&[1, 9], 10);
+        assert!(delete_bitset.is_alive(0));
+        assert!(delete_bitset.is_deleted(1));
+        assert!(delete_bitset.is_alive(2));
+        assert!(delete_bitset.is_alive(3));
+        assert!(delete_bitset.is_alive(4));
+        assert!(delete_bitset.is_alive(5));
+        assert!(delete_bitset.is_alive(6));
+        assert!(delete_bitset.is_alive(6));
+        assert!(delete_bitset.is_alive(7));
+        assert!(delete_bitset.is_alive(8));
+        assert!(delete_bitset.is_deleted(9));
+        for doc in 0..10 {
+            assert_eq!(delete_bitset.is_deleted(doc), !delete_bitset.is_alive(doc));
         }
+        assert_eq!(delete_bitset.len(), 2);
     }
 }

src/fastfield/mod.rs

@@ -474,7 +474,7 @@ mod tests {
         let date_field = schema_builder.add_date_field("date", FAST);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.set_merge_policy(Box::new(NoMergePolicy));
         index_writer.add_document(doc!(date_field =>crate::chrono::prelude::Utc::now()));
         index_writer.commit().unwrap();
@@ -511,7 +511,7 @@ mod tests {
         );
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.set_merge_policy(Box::new(NoMergePolicy));
         index_writer.add_document(doc!(
             date_field => crate::DateTime::from_u64(1i64.to_u64()),

src/fastfield/multivalued/mod.rs

@@ -25,7 +25,7 @@ mod tests {
         );
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(field=>1u64, field=>3u64));
         index_writer.add_document(doc!());
         index_writer.add_document(doc!(field=>4u64));
@@ -64,7 +64,7 @@ mod tests {
             schema_builder.add_i64_field("time_stamp_i", IntOptions::default().set_stored());
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let first_time_stamp = chrono::Utc::now();
         index_writer.add_document(
             doc!(date_field=>first_time_stamp, date_field=>first_time_stamp, time_i=>1i64),
@@ -186,7 +186,7 @@ mod tests {
         );
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(field=> 1i64, field => 3i64));
         index_writer.add_document(doc!());
         index_writer.add_document(doc!(field=> -4i64));
@@ -221,7 +221,7 @@ mod tests {
         let field = schema_builder.add_facet_field("facetfield");
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         for i in 0..100_000 {
             index_writer.add_document(doc!(field=> Facet::from(format!("/lang/{}", i).as_str())));
         }

src/fastfield/multivalued/reader.rs

@@ -74,7 +74,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index
-            .writer_with_num_threads(1, 30_000_000)
+            .writer_for_tests()
             .expect("Failed to create index writer.");
         index_writer.add_document(doc!(
             facet_field => Facet::from("/category/cat2"),

src/fieldnorm/reader.rs

@@ -19,7 +19,7 @@ pub struct FieldNormReaders {
 
 impl FieldNormReaders {
     /// Creates a field norm reader.
-    pub fn new(source: ReadOnlySource) -> crate::Result<FieldNormReaders> {
+    pub fn open(source: ReadOnlySource) -> crate::Result<FieldNormReaders> {
         let data = CompositeFile::open(&source)?;
         Ok(FieldNormReaders {
             data: Arc::new(data),
@@ -103,11 +103,9 @@ impl FieldNormReader {
     pub fn fieldnorm_to_id(fieldnorm: u32) -> u8 {
         fieldnorm_to_id(fieldnorm)
     }
-}
 
-#[cfg(test)]
-impl From<&[u32]> for FieldNormReader {
-    fn from(field_norms: &[u32]) -> FieldNormReader {
+    #[cfg(test)]
+    pub fn for_test(field_norms: &[u32]) -> FieldNormReader {
         let field_norms_id = field_norms
             .iter()
             .cloned()
@@ -119,3 +117,20 @@ impl From<&[u32]> for FieldNormReader {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use crate::fieldnorm::FieldNormReader;
+
+    #[test]
+    fn test_from_fieldnorms_array() {
+        let fieldnorms = &[1, 2, 3, 4, 1_000_000];
+        let fieldnorm_reader = FieldNormReader::for_test(fieldnorms);
+        assert_eq!(fieldnorm_reader.num_docs(), 5);
+        assert_eq!(fieldnorm_reader.fieldnorm(0), 1);
+        assert_eq!(fieldnorm_reader.fieldnorm(1), 2);
+        assert_eq!(fieldnorm_reader.fieldnorm(2), 3);
+        assert_eq!(fieldnorm_reader.fieldnorm(3), 4);
+        assert_eq!(fieldnorm_reader.fieldnorm(4), 983_064);
+    }
+}

src/indexer/index_writer.rs

@@ -536,6 +536,7 @@ impl IndexWriter {
     /// when no documents are remaining.
     ///
     /// Returns the former segment_ready channel.
+    #[allow(unused_must_use)]
     fn recreate_document_channel(&mut self) -> OperationReceiver {
         let (document_sender, document_receiver): (OperationSender, OperationReceiver) =
             channel::bounded(PIPELINE_MAX_SIZE_IN_DOCS);
@@ -575,7 +576,7 @@ impl IndexWriter {
         //
         // This will drop the document queue, and the thread
         // should terminate.
-        mem::replace(self, new_index_writer);
+        *self = new_index_writer;
 
         // Drains the document receiver pipeline :
         // Workers don't need to index the pending documents.
@@ -799,7 +800,7 @@ mod tests {
         let mut schema_builder = schema::Schema::builder();
         let text_field = schema_builder.add_text_field("text", schema::TEXT);
         let index = Index::create_in_ram(schema_builder.build());
-        let index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let index_writer = index.writer_for_tests().unwrap();
         let operations = vec![
             UserOperation::Add(doc!(text_field=>"a")),
             UserOperation::Add(doc!(text_field=>"b")),
@@ -814,7 +815,7 @@ mod tests {
         let text_field = schema_builder.add_text_field("text", schema::TEXT);
         let index = Index::create_in_ram(schema_builder.build());
 
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(text_field => "hello1"));
         index_writer.add_document(doc!(text_field => "hello2"));
         assert!(index_writer.commit().is_ok());
@@ -863,7 +864,7 @@ mod tests {
             .reload_policy(ReloadPolicy::Manual)
             .try_into()
             .unwrap();
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let a_term = Term::from_field_text(text_field, "a");
         let b_term = Term::from_field_text(text_field, "b");
         let operations = vec![
@@ -925,8 +926,8 @@ mod tests {
     fn test_lockfile_already_exists_error_msg() {
         let schema_builder = schema::Schema::builder();
         let index = Index::create_in_ram(schema_builder.build());
-        let _index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-        match index.writer_with_num_threads(1, 3_000_000) {
+        let _index_writer = index.writer_for_tests().unwrap();
+        match index.writer_for_tests() {
             Err(err) => {
                 let err_msg = err.to_string();
                 assert!(err_msg.contains("already an `IndexWriter`"));
@@ -1260,7 +1261,7 @@ mod tests {
         let idfield = schema_builder.add_text_field("id", STRING);
         schema_builder.add_text_field("optfield", STRING);
         let index = Index::create_in_ram(schema_builder.build());
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(idfield=>"myid"));
         let commit = index_writer.commit();
         assert!(commit.is_ok());

src/indexer/merger.rs

@@ -12,26 +12,27 @@ use crate::fieldnorm::FieldNormsSerializer;
 use crate::fieldnorm::FieldNormsWriter;
 use crate::fieldnorm::{FieldNormReader, FieldNormReaders};
 use crate::indexer::SegmentSerializer;
-use crate::postings::InvertedIndexSerializer;
 use crate::postings::Postings;
+use crate::postings::{InvertedIndexSerializer, SegmentPostings};
 use crate::schema::Cardinality;
 use crate::schema::FieldType;
 use crate::schema::{Field, Schema};
 use crate::store::StoreWriter;
 use crate::termdict::TermMerger;
 use crate::termdict::TermOrdinal;
-use crate::{DocId, SegmentComponent};
+use crate::{DocId, InvertedIndexReader, SegmentComponent};
 use std::cmp;
 use std::collections::HashMap;
+use std::sync::Arc;
 
-fn compute_total_num_tokens(readers: &[SegmentReader], field: Field) -> u64 {
+fn compute_total_num_tokens(readers: &[SegmentReader], field: Field) -> crate::Result<u64> {
     let mut total_tokens = 0u64;
     let mut count: [usize; 256] = [0; 256];
     for reader in readers {
         if reader.has_deletes() {
             // if there are deletes, then we use an approximation
             // using the fieldnorm
-            let fieldnorms_reader = reader.get_fieldnorms_reader(field);
+            let fieldnorms_reader = reader.get_fieldnorms_reader(field)?;
             for doc in reader.doc_ids_alive() {
                 let fieldnorm_id = fieldnorms_reader.fieldnorm_id(doc);
                 count[fieldnorm_id as usize] += 1;
@@ -40,7 +41,7 @@ fn compute_total_num_tokens(readers: &[SegmentReader], field: Field) -> u64 {
             total_tokens += reader.inverted_index(field).total_num_tokens();
         }
     }
-    total_tokens
+    Ok(total_tokens
         + count
             .iter()
             .cloned()
@@ -48,7 +49,7 @@ fn compute_total_num_tokens(readers: &[SegmentReader], field: Field) -> u64 {
             .map(|(fieldnorm_ord, count)| {
                 count as u64 * u64::from(FieldNormReader::id_to_fieldnorm(fieldnorm_ord as u8))
             })
-            .sum::<u64>()
+            .sum::<u64>())
 }
 
 pub struct IndexMerger {
@@ -174,7 +175,7 @@ impl IndexMerger {
         for field in fields {
             fieldnorms_data.clear();
             for reader in &self.readers {
-                let fieldnorms_reader = reader.get_fieldnorms_reader(field);
+                let fieldnorms_reader = reader.get_fieldnorms_reader(field)?;
                 for doc_id in reader.doc_ids_alive() {
                     let fieldnorm_id = fieldnorms_reader.fieldnorm_id(doc_id);
                     fieldnorms_data.push(fieldnorm_id);
@@ -497,7 +498,7 @@ impl IndexMerger {
     ) -> crate::Result<Option<TermOrdinalMapping>> {
         let mut positions_buffer: Vec<u32> = Vec::with_capacity(1_000);
         let mut delta_computer = DeltaComputer::new();
-        let field_readers = self
+        let field_readers: Vec<Arc<InvertedIndexReader>> = self
             .readers
             .iter()
             .map(|reader| reader.inverted_index(indexed_field))
@@ -540,7 +541,7 @@ impl IndexMerger {
         // The total number of tokens will only be exact when there has been no deletes.
         //
         // Otherwise, we approximate by removing deleted documents proportionally.
-        let total_num_tokens: u64 = compute_total_num_tokens(&self.readers, indexed_field);
+        let total_num_tokens: u64 = compute_total_num_tokens(&self.readers, indexed_field)?;
 
         // Create the total list of doc ids
         // by stacking the doc ids from the different segment.
@@ -563,47 +564,45 @@ impl IndexMerger {
                          indexed. Have you modified the schema?",
         );
 
+        let mut segment_postings_containing_the_term: Vec<(usize, SegmentPostings)> = vec![];
+
         while merged_terms.advance() {
+            segment_postings_containing_the_term.clear();
             let term_bytes: &[u8] = merged_terms.key();
 
+            let mut total_doc_freq = 0;
+
             // Let's compute the list of non-empty posting lists
-            let segment_postings: Vec<_> = merged_terms
-                .current_kvs()
-                .iter()
-                .flat_map(|heap_item| {
-                    let segment_ord = heap_item.segment_ord;
-                    let term_info = heap_item.streamer.value();
-                    let segment_reader = &self.readers[heap_item.segment_ord];
-                    let inverted_index = segment_reader.inverted_index(indexed_field);
-                    let mut segment_postings = inverted_index
-                        .read_postings_from_terminfo(term_info, segment_postings_option);
-                    let mut doc = segment_postings.doc();
-                    while doc != TERMINATED {
-                        if !segment_reader.is_deleted(doc) {
-                            return Some((segment_ord, segment_postings));
-                        }
-                        doc = segment_postings.advance();
-                    }
-                    None
-                })
-                .collect();
+            for heap_item in merged_terms.current_kvs() {
+                let segment_ord = heap_item.segment_ord;
+                let term_info = heap_item.streamer.value();
+                let segment_reader = &self.readers[heap_item.segment_ord];
+                let inverted_index: &InvertedIndexReader = &*field_readers[segment_ord];
+                let segment_postings =
+                    inverted_index.read_postings_from_terminfo(term_info, segment_postings_option);
+                let delete_bitset_opt = segment_reader.delete_bitset();
+                let doc_freq = if let Some(delete_bitset) = delete_bitset_opt {
+                    segment_postings.doc_freq_given_deletes(delete_bitset)
+                } else {
+                    segment_postings.doc_freq()
+                };
+                if doc_freq > 0u32 {
+                    total_doc_freq += doc_freq;
+                    segment_postings_containing_the_term.push((segment_ord, segment_postings));
+                }
+            }
 
             // At this point, `segment_postings` contains the posting list
-            // of all of the segments containing the given term.
+            // of all of the segments containing the given term (and that are non-empty)
             //
             // These segments are non-empty and advance has already been called.
-            if segment_postings.is_empty() {
+            if total_doc_freq == 0u32 {
+                // All docs that used to contain the term have been deleted. The `term` will be
+                // entirely removed.
                 continue;
             }
-            // If not, the `term` will be entirely removed.
 
-            // We know that there is at least one document containing
-            // the term, so we add it.
-            let term_doc_freq = segment_postings
-                .iter()
-                .map(|(_, segment_posting)| segment_posting.doc_freq())
-                .sum();
-            let to_term_ord = field_serializer.new_term(term_bytes, term_doc_freq)?;
+            let to_term_ord = field_serializer.new_term(term_bytes, total_doc_freq)?;
 
             if let Some(ref mut term_ord_mapping) = term_ord_mapping_opt {
                 for (segment_ord, from_term_ord) in merged_terms.matching_segments() {
@@ -613,7 +612,9 @@ impl IndexMerger {
 
             // We can now serialize this postings, by pushing each document to the
             // postings serializer.
-            for (segment_ord, mut segment_postings) in segment_postings {
+            for (segment_ord, mut segment_postings) in
+                segment_postings_containing_the_term.drain(..)
+            {
                 let old_to_new_doc_id = &merged_doc_id_map[segment_ord];
 
                 let mut doc = segment_postings.doc();
@@ -686,7 +687,7 @@ impl SerializableSegment for IndexMerger {
         let fieldnorm_data = serializer
             .segment()
             .open_read(SegmentComponent::FIELDNORMS)?;
-        let fieldnorm_readers = FieldNormReaders::new(fieldnorm_data)?;
+        let fieldnorm_readers = FieldNormReaders::open(fieldnorm_data)?;
         let term_ord_mappings =
             self.write_postings(serializer.get_postings_serializer(), fieldnorm_readers)?;
         self.write_fast_fields(serializer.get_fast_field_serializer(), term_ord_mappings)?;
@@ -750,7 +751,7 @@ mod tests {
         };
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             {
                 // writing the segment
                 {
@@ -802,7 +803,7 @@ mod tests {
             let segment_ids = index
                 .searchable_segment_ids()
                 .expect("Searchable segments failed.");
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             block_on(index_writer.merge(&segment_ids)).expect("Merging failed");
             index_writer.wait_merging_threads().unwrap();
         }
@@ -903,7 +904,7 @@ mod tests {
         let score_field = schema_builder.add_u64_field("score", score_fieldtype);
         let bytes_score_field = schema_builder.add_bytes_field("score_bytes");
         let index = Index::create_in_ram(schema_builder.build());
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let reader = index.reader().unwrap();
         let search_term = |searcher: &Searcher, term: Term| {
             let collector = FastFieldTestCollector::for_field(score_field);
@@ -1210,7 +1211,7 @@ mod tests {
         let index = Index::create_in_ram(schema_builder.build());
         let reader = index.reader().unwrap();
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             let index_doc = |index_writer: &mut IndexWriter, doc_facets: &[&str]| {
                 let mut doc = Document::default();
                 for facet in doc_facets {
@@ -1275,7 +1276,7 @@ mod tests {
             let segment_ids = index
                 .searchable_segment_ids()
                 .expect("Searchable segments failed.");
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             block_on(index_writer.merge(&segment_ids)).expect("Merging failed");
             index_writer.wait_merging_threads().unwrap();
             reader.reload().unwrap();
@@ -1294,7 +1295,7 @@ mod tests {
 
         // Deleting one term
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             let facet = Facet::from_path(vec!["top", "a", "firstdoc"]);
             let facet_term = Term::from_facet(facet_field, &facet);
             index_writer.delete_term(facet_term);
@@ -1319,7 +1320,7 @@ mod tests {
         let mut schema_builder = schema::Schema::builder();
         let int_field = schema_builder.add_u64_field("intvals", INDEXED);
         let index = Index::create_in_ram(schema_builder.build());
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(int_field => 1u64));
         index_writer.commit().expect("commit failed");
         index_writer.add_document(doc!(int_field => 1u64));
@@ -1348,7 +1349,7 @@ mod tests {
         let index = Index::create_in_ram(schema_builder.build());
         let reader = index.reader().unwrap();
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             let mut doc = Document::default();
             doc.add_u64(int_field, 1);
             index_writer.add_document(doc.clone());
@@ -1387,7 +1388,7 @@ mod tests {
         let index = Index::create_in_ram(schema_builder.build());
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             let index_doc = |index_writer: &mut IndexWriter, int_vals: &[u64]| {
                 let mut doc = Document::default();
                 for &val in int_vals {
@@ -1461,7 +1462,7 @@ mod tests {
             let segment_ids = index
                 .searchable_segment_ids()
                 .expect("Searchable segments failed.");
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             assert!(block_on(index_writer.merge(&segment_ids)).is_ok());
             assert!(index_writer.wait_merging_threads().is_ok());
         }
@@ -1515,7 +1516,7 @@ mod tests {
 
         let index = Index::create_in_ram(builder.build());
 
-        let mut writer = index.writer_with_num_threads(1, 3_000_000)?;
+        let mut writer = index.writer_for_tests()?;
 
         // Make sure we'll attempt to merge every created segment
         let mut policy = crate::indexer::LogMergePolicy::default();
@@ -1547,7 +1548,7 @@ mod tests {
         let mut builder = schema::SchemaBuilder::new();
         let text = builder.add_text_field("text", TEXT);
         let index = Index::create_in_ram(builder.build());
-        let mut writer = index.writer_with_num_threads(1, 3_000_000)?;
+        let mut writer = index.writer_for_tests()?;
         let happy_term = Term::from_field_text(text, "happy");
         let term_query = TermQuery::new(happy_term, IndexRecordOption::WithFreqs);
         for _ in 0..62 {
@@ -1558,7 +1559,7 @@ mod tests {
         let searcher = reader.searcher();
         let mut term_scorer = term_query
             .specialized_weight(&searcher, true)
-            .specialized_scorer(searcher.segment_reader(0u32), 1.0f32)?;
+            .specialized_scorer(searcher.segment_reader(0u32), 1.0)?;
         assert_eq!(term_scorer.doc(), 0);
         assert_nearly_equals!(term_scorer.block_max_score(), 0.0079681855);
         assert_nearly_equals!(term_scorer.score(), 0.0079681855);
@@ -1573,7 +1574,7 @@ mod tests {
         for segment_reader in searcher.segment_readers() {
             let mut term_scorer = term_query
                 .specialized_weight(&searcher, true)
-                .specialized_scorer(segment_reader, 1.0f32)?;
+                .specialized_scorer(segment_reader, 1.0)?;
             // the difference compared to before is instrinsic to the bm25 formula. no worries there.
             for doc in segment_reader.doc_ids_alive() {
                 assert_eq!(term_scorer.doc(), doc);
@@ -1597,7 +1598,7 @@ mod tests {
         let segment_reader = searcher.segment_reader(0u32);
         let mut term_scorer = term_query
             .specialized_weight(&searcher, true)
-            .specialized_scorer(segment_reader, 1.0f32)?;
+            .specialized_scorer(segment_reader, 1.0)?;
         // the difference compared to before is instrinsic to the bm25 formula. no worries there.
         for doc in segment_reader.doc_ids_alive() {
             assert_eq!(term_scorer.doc(), doc);

src/indexer/mod.rs

@@ -29,8 +29,9 @@ pub use self::segment_writer::SegmentWriter;
 /// Alias for the default merge policy, which is the `LogMergePolicy`.
 pub type DefaultMergePolicy = LogMergePolicy;
 
+#[cfg(feature = "mmap")]
 #[cfg(test)]
-mod tests {
+mod tests_mmap {
     use crate::schema::{self, Schema};
     use crate::{Index, Term};
 
@@ -39,7 +40,7 @@ mod tests {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", schema::TEXT);
         let index = Index::create_from_tempdir(schema_builder.build()).unwrap();
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         // there must be one deleted document in the segment
         index_writer.add_document(doc!(text_field=>"b"));
         index_writer.delete_term(Term::from_field_text(text_field, "b"));

src/indexer/segment_updater.rs

@@ -130,15 +130,14 @@ fn merge(
     // An IndexMerger is like a "view" of our merged segments.
     let merger: IndexMerger = IndexMerger::open(index.schema(), &segments[..])?;
 
-    let merged_segment_id = merged_segment.id();
-
     // ... we just serialize this index merger in our new segment to merge the two segments.
-    let segment_serializer = SegmentSerializer::for_segment(merged_segment)?;
+    let segment_serializer = SegmentSerializer::for_segment(merged_segment.clone())?;
 
     let num_docs = merger.write(segment_serializer)?;
 
-    let segment_meta = index.new_segment_meta(merged_segment_id, num_docs);
+    let merged_segment_id = merged_segment.id();
 
+    let segment_meta = index.new_segment_meta(merged_segment_id, num_docs);
     Ok(SegmentEntry::new(segment_meta, delete_cursor, None))
 }
 
@@ -556,7 +555,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
 
         // writing the segment
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.set_merge_policy(Box::new(MergeWheneverPossible));
 
         {
@@ -609,7 +608,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
 
         // writing the segment
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
 
         {
             for _ in 0..100 {
@@ -680,7 +679,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
 
         // writing the segment
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
 
         {
             for _ in 0..100 {

src/indexer/segment_writer.rs

@@ -17,7 +17,6 @@ use crate::tokenizer::{TokenStreamChain, Tokenizer};
 use crate::Opstamp;
 use crate::{DocId, SegmentComponent};
 use std::io;
-use std::str;
 
 /// Computes the initial size of the hash table.
 ///
@@ -48,6 +47,7 @@ pub struct SegmentWriter {
     fieldnorms_writer: FieldNormsWriter,
     doc_opstamps: Vec<Opstamp>,
     tokenizers: Vec<Option<TextAnalyzer>>,
+    term_buffer: Term,
 }
 
 impl SegmentWriter {
@@ -91,6 +91,7 @@ impl SegmentWriter {
             fast_field_writers: FastFieldsWriter::from_schema(schema),
             doc_opstamps: Vec::with_capacity(1_000),
             tokenizers,
+            term_buffer: Term::new(),
         })
     }
 
@@ -128,24 +129,26 @@ impl SegmentWriter {
             if !field_options.is_indexed() {
                 continue;
             }
+            let (term_buffer, multifield_postings) =
+                (&mut self.term_buffer, &mut self.multifield_postings);
             match *field_options.field_type() {
                 FieldType::HierarchicalFacet => {
-                    let facets: Vec<&str> = field_values
-                        .iter()
-                        .flat_map(|field_value| match *field_value.value() {
-                            Value::Facet(ref facet) => Some(facet.encoded_str()),
-                            _ => {
-                                panic!("Expected hierarchical facet");
-                            }
-                        })
-                        .collect();
-                    let mut term = Term::for_field(field); // we set the Term
+                    term_buffer.set_field(field);
+                    let facets =
+                        field_values
+                            .iter()
+                            .flat_map(|field_value| match *field_value.value() {
+                                Value::Facet(ref facet) => Some(facet.encoded_str()),
+                                _ => {
+                                    panic!("Expected hierarchical facet");
+                                }
+                            });
                     for fake_str in facets {
                         let mut unordered_term_id_opt = None;
                         FacetTokenizer.token_stream(fake_str).process(&mut |token| {
-                            term.set_text(&token.text);
+                            term_buffer.set_text(&token.text);
                             let unordered_term_id =
-                                self.multifield_postings.subscribe(doc_id, &term);
+                                multifield_postings.subscribe(doc_id, &term_buffer);
                             unordered_term_id_opt = Some(unordered_term_id);
                         });
                         if let Some(unordered_term_id) = unordered_term_id_opt {
@@ -168,7 +171,6 @@ impl SegmentWriter {
                                 if let Some(last_token) = tok_str.tokens.last() {
                                     total_offset += last_token.offset_to;
                                 }
-
                                 token_streams
                                     .push(PreTokenizedStream::from(tok_str.clone()).into());
                             }
@@ -178,7 +180,6 @@ impl SegmentWriter {
                                 {
                                     offsets.push(total_offset);
                                     total_offset += text.len();
-
                                     token_streams.push(tokenizer.token_stream(text));
                                 }
                             }
@@ -190,8 +191,12 @@ impl SegmentWriter {
                         0
                     } else {
                         let mut token_stream = TokenStreamChain::new(offsets, token_streams);
-                        self.multifield_postings
-                            .index_text(doc_id, field, &mut token_stream)
+                        multifield_postings.index_text(
+                            doc_id,
+                            field,
+                            &mut token_stream,
+                            term_buffer,
+                        )
                     };
 
                     self.fieldnorms_writer.record(doc_id, field, num_tokens);
@@ -199,44 +204,36 @@ impl SegmentWriter {
                 FieldType::U64(ref int_option) => {
                     if int_option.is_indexed() {
                         for field_value in field_values {
-                            let term = Term::from_field_u64(
-                                field_value.field(),
-                                field_value.value().u64_value(),
-                            );
-                            self.multifield_postings.subscribe(doc_id, &term);
+                            term_buffer.set_field(field_value.field());
+                            term_buffer.set_u64(field_value.value().u64_value());
+                            multifield_postings.subscribe(doc_id, &term_buffer);
                         }
                     }
                 }
                 FieldType::Date(ref int_option) => {
                     if int_option.is_indexed() {
                         for field_value in field_values {
-                            let term = Term::from_field_i64(
-                                field_value.field(),
-                                field_value.value().date_value().timestamp(),
-                            );
-                            self.multifield_postings.subscribe(doc_id, &term);
+                            term_buffer.set_field(field_value.field());
+                            term_buffer.set_i64(field_value.value().date_value().timestamp());
+                            multifield_postings.subscribe(doc_id, &term_buffer);
                         }
                     }
                 }
                 FieldType::I64(ref int_option) => {
                     if int_option.is_indexed() {
                         for field_value in field_values {
-                            let term = Term::from_field_i64(
-                                field_value.field(),
-                                field_value.value().i64_value(),
-                            );
-                            self.multifield_postings.subscribe(doc_id, &term);
+                            term_buffer.set_field(field_value.field());
+                            term_buffer.set_i64(field_value.value().i64_value());
+                            multifield_postings.subscribe(doc_id, &term_buffer);
                         }
                     }
                 }
                 FieldType::F64(ref int_option) => {
                     if int_option.is_indexed() {
                         for field_value in field_values {
-                            let term = Term::from_field_f64(
-                                field_value.field(),
-                                field_value.value().f64_value(),
-                            );
-                            self.multifield_postings.subscribe(doc_id, &term);
+                            term_buffer.set_field(field_value.field());
+                            term_buffer.set_f64(field_value.value().f64_value());
+                            multifield_postings.subscribe(doc_id, &term_buffer);
                         }
                     }
                 }
@@ -287,7 +284,7 @@ fn write(
     let fieldnorm_data = serializer
         .segment()
         .open_read(SegmentComponent::FIELDNORMS)?;
-    let fieldnorm_readers = FieldNormReaders::new(fieldnorm_data)?;
+    let fieldnorm_readers = FieldNormReaders::open(fieldnorm_data)?;
     let term_ord_map =
         multifield_postings.serialize(serializer.get_postings_serializer(), fieldnorm_readers)?;
     fast_field_writers.serialize(serializer.get_fast_field_serializer(), &term_ord_map)?;

src/lib.rs

@@ -245,11 +245,10 @@ pub type DocId = u32;
 /// with opstamp `n+1`.
 pub type Opstamp = u64;
 
-/// A f32 that represents the relevance of the document to the query
+/// A Score that represents the relevance of the document to the query
 ///
-/// This is modelled internally as a `f32`. The
-/// larger the number, the more relevant the document
-/// to the search
+/// This is modelled internally as a `f32`. The larger the number, the more relevant
+/// the document to the search query.
 pub type Score = f32;
 
 /// A `SegmentLocalId` identifies a segment.
@@ -282,7 +281,6 @@ pub struct DocAddress(pub SegmentLocalId, pub DocId);
 
 #[cfg(test)]
 mod tests {
-
     use crate::collector::tests::TEST_COLLECTOR_WITH_SCORE;
     use crate::core::SegmentReader;
     use crate::docset::{DocSet, TERMINATED};
@@ -290,7 +288,6 @@ mod tests {
     use crate::schema::*;
     use crate::DocAddress;
     use crate::Index;
-    use crate::IndexWriter;
     use crate::Postings;
     use crate::ReloadPolicy;
     use rand::distributions::Bernoulli;
@@ -355,14 +352,14 @@ mod tests {
 
     #[test]
     #[cfg(feature = "mmap")]
-    fn test_indexing() {
+    fn test_indexing() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_from_tempdir(schema).unwrap();
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests()?;
             {
                 let doc = doc!(text_field=>"af b");
                 index_writer.add_document(doc);
@@ -377,29 +374,30 @@ mod tests {
             }
             assert!(index_writer.commit().is_ok());
         }
+        Ok(())
     }
 
     #[test]
-    fn test_docfreq1() {
+    fn test_docfreq1() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let index = Index::create_in_ram(schema_builder.build());
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         {
             index_writer.add_document(doc!(text_field=>"a b c"));
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
         {
             index_writer.add_document(doc!(text_field=>"a"));
             index_writer.add_document(doc!(text_field=>"a a"));
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
         {
             index_writer.add_document(doc!(text_field=>"c"));
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
         {
-            let reader = index.reader().unwrap();
+            let reader = index.reader()?;
             let searcher = reader.searcher();
             let term_a = Term::from_field_text(text_field, "a");
             assert_eq!(searcher.doc_freq(&term_a), 3);
@@ -410,67 +408,50 @@ mod tests {
             let term_d = Term::from_field_text(text_field, "d");
             assert_eq!(searcher.doc_freq(&term_d), 0);
         }
+        Ok(())
     }
 
     #[test]
-    fn test_fieldnorm_no_docs_with_field() {
+    fn test_fieldnorm_no_docs_with_field() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let title_field = schema_builder.add_text_field("title", TEXT);
         let text_field = schema_builder.add_text_field("text", TEXT);
         let index = Index::create_in_ram(schema_builder.build());
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(text_field=>"a b c"));
+        index_writer.commit()?;
+        let index_reader = index.reader()?;
+        let searcher = index_reader.searcher();
+        let reader = searcher.segment_reader(0);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            {
-                let doc = doc!(text_field=>"a b c");
-                index_writer.add_document(doc);
-            }
-            index_writer.commit().unwrap();
+            let fieldnorm_reader = reader.get_fieldnorms_reader(text_field)?;
+            assert_eq!(fieldnorm_reader.fieldnorm(0), 3);
         }
         {
-            let index_reader = index.reader().unwrap();
-            let searcher = index_reader.searcher();
-            let reader = searcher.segment_reader(0);
-            {
-                let fieldnorm_reader = reader.get_fieldnorms_reader(text_field);
-                assert_eq!(fieldnorm_reader.fieldnorm(0), 3);
-            }
-            {
-                let fieldnorm_reader = reader.get_fieldnorms_reader(title_field);
-                assert_eq!(fieldnorm_reader.fieldnorm_id(0), 0);
-            }
+            let fieldnorm_reader = reader.get_fieldnorms_reader(title_field)?;
+            assert_eq!(fieldnorm_reader.fieldnorm_id(0), 0);
         }
+        Ok(())
     }
 
     #[test]
-    fn test_fieldnorm() {
+    fn test_fieldnorm() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let index = Index::create_in_ram(schema_builder.build());
-        {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            {
-                let doc = doc!(text_field=>"a b c");
-                index_writer.add_document(doc);
-            }
-            {
-                let doc = doc!();
-                index_writer.add_document(doc);
-            }
-            {
-                let doc = doc!(text_field=>"a b");
-                index_writer.add_document(doc);
-            }
-            index_writer.commit().unwrap();
-        }
-        {
-            let reader = index.reader().unwrap();
-            let searcher = reader.searcher();
-            let segment_reader: &SegmentReader = searcher.segment_reader(0);
-            let fieldnorms_reader = segment_reader.get_fieldnorms_reader(text_field);
-            assert_eq!(fieldnorms_reader.fieldnorm(0), 3);
-            assert_eq!(fieldnorms_reader.fieldnorm(1), 0);
-            assert_eq!(fieldnorms_reader.fieldnorm(2), 2);
-        }
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(text_field=>"a b c"));
+        index_writer.add_document(doc!());
+        index_writer.add_document(doc!(text_field=>"a b"));
+        index_writer.commit()?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        let segment_reader: &SegmentReader = searcher.segment_reader(0);
+        let fieldnorms_reader = segment_reader.get_fieldnorms_reader(text_field)?;
+        assert_eq!(fieldnorms_reader.fieldnorm(0), 3);
+        assert_eq!(fieldnorms_reader.fieldnorm(1), 0);
+        assert_eq!(fieldnorms_reader.fieldnorm(2), 2);
+        Ok(())
     }
 
     fn advance_undeleted(docset: &mut dyn DocSet, reader: &SegmentReader) -> bool {
@@ -485,7 +466,7 @@ mod tests {
     }
 
     #[test]
-    fn test_delete_postings1() {
+    fn test_delete_postings1() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let term_abcd = Term::from_field_text(text_field, "abcd");
@@ -501,7 +482,7 @@ mod tests {
             .unwrap();
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests()?;
             // 0
             index_writer.add_document(doc!(text_field=>"a b"));
             // 1
@@ -517,10 +498,10 @@ mod tests {
             index_writer.add_document(doc!(text_field=>" b c"));
             // 5
             index_writer.add_document(doc!(text_field=>" a"));
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
         {
-            reader.reload().unwrap();
+            reader.reload()?;
             let searcher = reader.searcher();
             let segment_reader = searcher.segment_reader(0);
             let inverted_index = segment_reader.inverted_index(text_field);
@@ -548,15 +529,15 @@ mod tests {
         }
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests()?;
             // 0
             index_writer.add_document(doc!(text_field=>"a b"));
             // 1
             index_writer.delete_term(Term::from_field_text(text_field, "c"));
-            index_writer.rollback().unwrap();
+            index_writer.rollback()?;
         }
         {
-            reader.reload().unwrap();
+            reader.reload()?;
             let searcher = reader.searcher();
             let seg_reader = searcher.segment_reader(0);
             let inverted_index = seg_reader.inverted_index(term_abcd.field());
@@ -585,15 +566,15 @@ mod tests {
         }
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests()?;
             index_writer.add_document(doc!(text_field=>"a b"));
             index_writer.delete_term(Term::from_field_text(text_field, "c"));
-            index_writer.rollback().unwrap();
+            index_writer.rollback()?;
             index_writer.delete_term(Term::from_field_text(text_field, "a"));
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
         {
-            reader.reload().unwrap();
+            reader.reload()?;
             let searcher = reader.searcher();
             let segment_reader = searcher.segment_reader(0);
             let inverted_index = segment_reader.inverted_index(term_abcd.field());
@@ -625,19 +606,20 @@ mod tests {
                 assert!(!advance_undeleted(&mut postings, segment_reader));
             }
         }
+        Ok(())
     }
 
     #[test]
-    fn test_indexed_u64() {
+    fn test_indexed_u64() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let field = schema_builder.add_u64_field("value", INDEXED);
         let schema = schema_builder.build();
 
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         index_writer.add_document(doc!(field=>1u64));
-        index_writer.commit().unwrap();
-        let reader = index.reader().unwrap();
+        index_writer.commit()?;
+        let reader = index.reader()?;
         let searcher = reader.searcher();
         let term = Term::from_field_u64(field, 1u64);
         let mut postings = searcher
@@ -647,20 +629,21 @@ mod tests {
             .unwrap();
         assert_eq!(postings.doc(), 0);
         assert_eq!(postings.advance(), TERMINATED);
+        Ok(())
     }
 
     #[test]
-    fn test_indexed_i64() {
+    fn test_indexed_i64() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let value_field = schema_builder.add_i64_field("value", INDEXED);
         let schema = schema_builder.build();
 
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         let negative_val = -1i64;
         index_writer.add_document(doc!(value_field => negative_val));
-        index_writer.commit().unwrap();
-        let reader = index.reader().unwrap();
+        index_writer.commit()?;
+        let reader = index.reader()?;
         let searcher = reader.searcher();
         let term = Term::from_field_i64(value_field, negative_val);
         let mut postings = searcher
@@ -670,20 +653,21 @@ mod tests {
             .unwrap();
         assert_eq!(postings.doc(), 0);
         assert_eq!(postings.advance(), TERMINATED);
+        Ok(())
     }
 
     #[test]
-    fn test_indexed_f64() {
+    fn test_indexed_f64() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let value_field = schema_builder.add_f64_field("value", INDEXED);
         let schema = schema_builder.build();
 
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         let val = std::f64::consts::PI;
         index_writer.add_document(doc!(value_field => val));
-        index_writer.commit().unwrap();
-        let reader = index.reader().unwrap();
+        index_writer.commit()?;
+        let reader = index.reader()?;
         let searcher = reader.searcher();
         let term = Term::from_field_f64(value_field, val);
         let mut postings = searcher
@@ -693,26 +677,29 @@ mod tests {
             .unwrap();
         assert_eq!(postings.doc(), 0);
         assert_eq!(postings.advance(), TERMINATED);
+        Ok(())
     }
 
     #[test]
-    fn test_indexedfield_not_in_documents() {
+    fn test_indexedfield_not_in_documents() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let absent_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(2, 6_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         index_writer.add_document(doc!(text_field=>"a"));
         assert!(index_writer.commit().is_ok());
-        let reader = index.reader().unwrap();
+        let reader = index.reader()?;
         let searcher = reader.searcher();
         let segment_reader = searcher.segment_reader(0);
-        segment_reader.inverted_index(absent_field); //< should not panic
+        let inverted_index = segment_reader.inverted_index(absent_field); //< should not panic
+        assert_eq!(inverted_index.terms().num_terms(), 0);
+        Ok(())
     }
 
     #[test]
-    fn test_delete_postings2() {
+    fn test_delete_postings2() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
@@ -720,53 +707,40 @@ mod tests {
         let reader = index
             .reader_builder()
             .reload_policy(ReloadPolicy::Manual)
-            .try_into()
-            .unwrap();
+            .try_into()?;
 
         // writing the segment
-        let mut index_writer = index.writer_with_num_threads(2, 6_000_000).unwrap();
-
-        let add_document = |index_writer: &mut IndexWriter, val: &'static str| {
-            let doc = doc!(text_field=>val);
-            index_writer.add_document(doc);
-        };
-
-        let remove_document = |index_writer: &mut IndexWriter, val: &'static str| {
-            let delterm = Term::from_field_text(text_field, val);
-            index_writer.delete_term(delterm);
-        };
-
-        add_document(&mut index_writer, "63");
-        add_document(&mut index_writer, "70");
-        add_document(&mut index_writer, "34");
-        add_document(&mut index_writer, "1");
-        add_document(&mut index_writer, "38");
-        add_document(&mut index_writer, "33");
-        add_document(&mut index_writer, "40");
-        add_document(&mut index_writer, "17");
-        remove_document(&mut index_writer, "38");
-        remove_document(&mut index_writer, "34");
-        index_writer.commit().unwrap();
-        reader.reload().unwrap();
-        let searcher = reader.searcher();
-        assert_eq!(searcher.num_docs(), 6);
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(text_field=>"63"));
+        index_writer.add_document(doc!(text_field=>"70"));
+        index_writer.add_document(doc!(text_field=>"34"));
+        index_writer.add_document(doc!(text_field=>"1"));
+        index_writer.add_document(doc!(text_field=>"38"));
+        index_writer.add_document(doc!(text_field=>"33"));
+        index_writer.add_document(doc!(text_field=>"40"));
+        index_writer.add_document(doc!(text_field=>"17"));
+        index_writer.delete_term(Term::from_field_text(text_field, "38"));
+        index_writer.delete_term(Term::from_field_text(text_field, "34"));
+        index_writer.commit()?;
+        reader.reload()?;
+        assert_eq!(reader.searcher().num_docs(), 6);
+        Ok(())
     }
 
     #[test]
-    fn test_termfreq() {
+    fn test_termfreq() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            let doc = doc!(text_field=>"af af af bc bc");
-            index_writer.add_document(doc);
-            index_writer.commit().unwrap();
+            let mut index_writer = index.writer_for_tests()?;
+            index_writer.add_document(doc!(text_field=>"af af af bc bc"));
+            index_writer.commit()?;
         }
         {
-            let index_reader = index.reader().unwrap();
+            let index_reader = index.reader()?;
             let searcher = index_reader.searcher();
             let reader = searcher.segment_reader(0);
             let inverted_index = reader.inverted_index(text_field);
@@ -782,63 +756,63 @@ mod tests {
             assert_eq!(postings.term_freq(), 3);
             assert_eq!(postings.advance(), TERMINATED);
         }
+        Ok(())
     }
 
     #[test]
-    fn test_searcher_1() {
+    fn test_searcher_1() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let reader = index.reader().unwrap();
-        {
-            // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            index_writer.add_document(doc!(text_field=>"af af af b"));
-            index_writer.add_document(doc!(text_field=>"a b c"));
-            index_writer.add_document(doc!(text_field=>"a b c d"));
-            index_writer.commit().unwrap();
-        }
-        {
-            reader.reload().unwrap();
-            let searcher = reader.searcher();
-            let get_doc_ids = |terms: Vec<Term>| {
-                let query = BooleanQuery::new_multiterms_query(terms);
-                let topdocs = searcher.search(&query, &TEST_COLLECTOR_WITH_SCORE).unwrap();
-                topdocs.docs().to_vec()
-            };
-            assert_eq!(
-                get_doc_ids(vec![Term::from_field_text(text_field, "a")]),
-                vec![DocAddress(0, 1), DocAddress(0, 2)]
-            );
-            assert_eq!(
-                get_doc_ids(vec![Term::from_field_text(text_field, "af")]),
-                vec![DocAddress(0, 0)]
-            );
-            assert_eq!(
-                get_doc_ids(vec![Term::from_field_text(text_field, "b")]),
-                vec![DocAddress(0, 0), DocAddress(0, 1), DocAddress(0, 2)]
-            );
-            assert_eq!(
-                get_doc_ids(vec![Term::from_field_text(text_field, "c")]),
-                vec![DocAddress(0, 1), DocAddress(0, 2)]
-            );
-            assert_eq!(
-                get_doc_ids(vec![Term::from_field_text(text_field, "d")]),
-                vec![DocAddress(0, 2)]
-            );
-            assert_eq!(
-                get_doc_ids(vec![
-                    Term::from_field_text(text_field, "b"),
-                    Term::from_field_text(text_field, "a"),
-                ]),
-                vec![DocAddress(0, 0), DocAddress(0, 1), DocAddress(0, 2)]
-            );
-        }
+        let reader = index.reader()?;
+        // writing the segment
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(text_field=>"af af af b"));
+        index_writer.add_document(doc!(text_field=>"a b c"));
+        index_writer.add_document(doc!(text_field=>"a b c d"));
+        index_writer.commit()?;
+
+        reader.reload()?;
+        let searcher = reader.searcher();
+        let get_doc_ids = |terms: Vec<Term>| {
+            let query = BooleanQuery::new_multiterms_query(terms);
+            searcher
+                .search(&query, &TEST_COLLECTOR_WITH_SCORE)
+                .map(|topdocs| topdocs.docs().to_vec())
+        };
+        assert_eq!(
+            get_doc_ids(vec![Term::from_field_text(text_field, "a")])?,
+            vec![DocAddress(0, 1), DocAddress(0, 2)]
+        );
+        assert_eq!(
+            get_doc_ids(vec![Term::from_field_text(text_field, "af")])?,
+            vec![DocAddress(0, 0)]
+        );
+        assert_eq!(
+            get_doc_ids(vec![Term::from_field_text(text_field, "b")])?,
+            vec![DocAddress(0, 0), DocAddress(0, 1), DocAddress(0, 2)]
+        );
+        assert_eq!(
+            get_doc_ids(vec![Term::from_field_text(text_field, "c")])?,
+            vec![DocAddress(0, 1), DocAddress(0, 2)]
+        );
+        assert_eq!(
+            get_doc_ids(vec![Term::from_field_text(text_field, "d")])?,
+            vec![DocAddress(0, 2)]
+        );
+        assert_eq!(
+            get_doc_ids(vec![
+                Term::from_field_text(text_field, "b"),
+                Term::from_field_text(text_field, "a"),
+            ])?,
+            vec![DocAddress(0, 0), DocAddress(0, 1), DocAddress(0, 2)]
+        );
+        Ok(())
     }
 
     #[test]
-    fn test_searcher_2() {
+    fn test_searcher_2() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
@@ -846,19 +820,17 @@ mod tests {
         let reader = index
             .reader_builder()
             .reload_policy(ReloadPolicy::Manual)
-            .try_into()
-            .unwrap();
+            .try_into()?;
         assert_eq!(reader.searcher().num_docs(), 0u64);
-        {
-            // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-            index_writer.add_document(doc!(text_field=>"af b"));
-            index_writer.add_document(doc!(text_field=>"a b c"));
-            index_writer.add_document(doc!(text_field=>"a b c d"));
-            index_writer.commit().unwrap();
-        }
-        reader.reload().unwrap();
+        // writing the segment
+        let mut index_writer = index.writer_for_tests()?;
+        index_writer.add_document(doc!(text_field=>"af b"));
+        index_writer.add_document(doc!(text_field=>"a b c"));
+        index_writer.add_document(doc!(text_field=>"a b c d"));
+        index_writer.commit()?;
+        reader.reload()?;
         assert_eq!(reader.searcher().num_docs(), 3u64);
+        Ok(())
     }
 
     #[test]
@@ -880,7 +852,7 @@ mod tests {
     }
 
     #[test]
-    fn test_wrong_fast_field_type() {
+    fn test_wrong_fast_field_type() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let fast_field_unsigned = schema_builder.add_u64_field("unsigned", FAST);
         let fast_field_signed = schema_builder.add_i64_field("signed", FAST);
@@ -890,14 +862,14 @@ mod tests {
         let schema = schema_builder.build();
 
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 50_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         {
             let document =
                 doc!(fast_field_unsigned => 4u64, fast_field_signed=>4i64, fast_field_float=>4f64);
             index_writer.add_document(document);
-            index_writer.commit().unwrap();
+            index_writer.commit()?;
         }
-        let reader = index.reader().unwrap();
+        let reader = index.reader()?;
         let searcher = reader.searcher();
         let segment_reader: &SegmentReader = searcher.segment_reader(0);
         {
@@ -936,11 +908,12 @@ mod tests {
             let fast_field_reader = fast_field_reader_opt.unwrap();
             assert_eq!(fast_field_reader.get(0), 4f64)
         }
+        Ok(())
     }
 
     // motivated by #729
     #[test]
-    fn test_update_via_delete_insert() {
+    fn test_update_via_delete_insert() -> crate::Result<()> {
         use crate::collector::Count;
         use crate::indexer::NoMergePolicy;
         use crate::query::AllQuery;
@@ -954,17 +927,17 @@ mod tests {
         let schema = schema_builder.build();
 
         let index = Index::create_in_ram(schema.clone());
-        let index_reader = index.reader().unwrap();
+        let index_reader = index.reader()?;
 
-        let mut index_writer = index.writer(3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         index_writer.set_merge_policy(Box::new(NoMergePolicy));
 
         for doc_id in 0u64..DOC_COUNT {
             index_writer.add_document(doc!(id => doc_id));
         }
-        index_writer.commit().unwrap();
+        index_writer.commit()?;
 
-        index_reader.reload().unwrap();
+        index_reader.reload()?;
         let searcher = index_reader.searcher();
 
         assert_eq!(
@@ -975,12 +948,11 @@ mod tests {
         // update the 10 elements by deleting and re-adding
         for doc_id in 0u64..DOC_COUNT {
             index_writer.delete_term(Term::from_field_u64(id, doc_id));
-            index_writer.commit().unwrap();
-            index_reader.reload().unwrap();
-            let doc = doc!(id =>  doc_id);
-            index_writer.add_document(doc);
-            index_writer.commit().unwrap();
-            index_reader.reload().unwrap();
+            index_writer.commit()?;
+            index_reader.reload()?;
+            index_writer.add_document(doc!(id =>  doc_id));
+            index_writer.commit()?;
+            index_reader.reload()?;
             let searcher = index_reader.searcher();
             // The number of document should be stable.
             assert_eq!(
@@ -989,7 +961,7 @@ mod tests {
             );
         }
 
-        index_reader.reload().unwrap();
+        index_reader.reload()?;
         let searcher = index_reader.searcher();
         let segment_ids: Vec<SegmentId> = searcher
             .segment_readers()
@@ -998,12 +970,18 @@ mod tests {
             .collect();
         block_on(index_writer.merge(&segment_ids)).unwrap();
 
-        index_reader.reload().unwrap();
+        index_reader.reload()?;
         let searcher = index_reader.searcher();
+        assert_eq!(searcher.search(&AllQuery, &Count)?, DOC_COUNT as usize);
+        Ok(())
+    }
 
-        assert_eq!(
-            searcher.search(&AllQuery, &Count).unwrap(),
-            DOC_COUNT as usize
-        );
+    #[test]
+    fn test_validate_checksum() -> crate::Result<()> {
+        let index_path = tempfile::tempdir().expect("dir");
+        let schema = Schema::builder().build();
+        let index = Index::create_in_dir(&index_path, schema)?;
+        assert!(index.validate_checksum()?.is_empty());
+        Ok(())
     }
 }

src/positions/reader.rs

@@ -78,6 +78,7 @@ impl Positions {
     }
 }
 
+#[derive(Clone)]
 pub struct PositionReader {
     skip_read: OwnedRead,
     position_read: OwnedRead,

src/postings/block_segment_postings.rs

@@ -9,9 +9,9 @@ use crate::query::BM25Weight;
 use crate::schema::IndexRecordOption;
 use crate::{DocId, Score, TERMINATED};
 
-fn max_f32<I: Iterator<Item = f32>>(mut it: I) -> Option<f32> {
+fn max_score<I: Iterator<Item = Score>>(mut it: I) -> Option<Score> {
     if let Some(first) = it.next() {
-        Some(it.fold(first, f32::max))
+        Some(it.fold(first, Score::max))
     } else {
         None
     }
@@ -24,6 +24,7 @@ fn max_f32<I: Iterator<Item = f32>>(mut it: I) -> Option<f32> {
 ///
 /// While it is useful for some very specific high-performance
 /// use cases, you should prefer using `SegmentPostings` for most usage.
+#[derive(Clone)]
 pub struct BlockSegmentPostings {
     pub(crate) doc_decoder: BlockDecoder,
     loaded_offset: usize,
@@ -58,10 +59,14 @@ fn decode_vint_block(
     doc_offset: DocId,
     num_vint_docs: usize,
 ) {
-    doc_decoder.fill(TERMINATED);
-    let num_consumed_bytes = doc_decoder.uncompress_vint_sorted(data, doc_offset, num_vint_docs);
+    let num_consumed_bytes =
+        doc_decoder.uncompress_vint_sorted(data, doc_offset, num_vint_docs, TERMINATED);
     if let Some(freq_decoder) = freq_decoder_opt {
-        freq_decoder.uncompress_vint_unsorted(&data[num_consumed_bytes..], num_vint_docs);
+        freq_decoder.uncompress_vint_unsorted(
+            &data[num_consumed_bytes..],
+            num_vint_docs,
+            TERMINATED,
+        );
     }
 }
 
@@ -124,27 +129,33 @@ impl BlockSegmentPostings {
         fieldnorm_reader: &FieldNormReader,
         bm25_weight: &BM25Weight,
     ) -> Score {
-        let (block_max_score_cache, skip_reader, doc_decoder, freq_decoder) = (
-            &mut self.block_max_score_cache,
-            &self.skip_reader,
-            &self.doc_decoder,
-            &self.freq_decoder,
-        );
-        *block_max_score_cache.get_or_insert_with(|| {
-            skip_reader
-                .block_max_score(bm25_weight)
-                .or_else(|| {
-                    let docs = doc_decoder.output_array();
-                    let freqs = freq_decoder.output_array();
-                    max_f32(docs.iter().cloned().zip(freqs.iter().cloned()).map(
-                        |(doc, term_freq)| {
-                            let fieldnorm_id = fieldnorm_reader.fieldnorm_id(doc);
-                            bm25_weight.score(fieldnorm_id, term_freq)
-                        },
-                    ))
-                })
-                .unwrap_or(0f32)
-        })
+        if let Some(score) = self.block_max_score_cache {
+            return score;
+        }
+        if let Some(skip_reader_max_score) = self.skip_reader.block_max_score(bm25_weight) {
+            // if we are on a full block, the skip reader should have the block max information
+            // for us
+            self.block_max_score_cache = Some(skip_reader_max_score);
+            return skip_reader_max_score;
+        }
+        // this is the last block of the segment posting list.
+        // If it is actually loaded, we can compute block max manually.
+        if self.block_is_loaded() {
+            let docs = self.doc_decoder.output_array().iter().cloned();
+            let freqs = self.freq_decoder.output_array().iter().cloned();
+            let bm25_scores = docs.zip(freqs).map(|(doc, term_freq)| {
+                let fieldnorm_id = fieldnorm_reader.fieldnorm_id(doc);
+                bm25_weight.score(fieldnorm_id, term_freq)
+            });
+            let block_max_score = max_score(bm25_scores).unwrap_or(0.0);
+            self.block_max_score_cache = Some(block_max_score);
+            return block_max_score;
+        }
+        // We do not have access to any good block max value. We return bm25_weight.max_score()
+        // as it is a valid upperbound.
+        //
+        // We do not cache it however, so that it gets computed when once block is loaded.
+        bm25_weight.max_score()
     }
 
     pub(crate) fn freq_reading_option(&self) -> FreqReadingOption {
@@ -164,6 +175,7 @@ impl BlockSegmentPostings {
     pub(crate) fn reset(&mut self, doc_freq: u32, postings_data: ReadOnlySource) {
         let (skip_data_opt, postings_data) = split_into_skips_and_postings(doc_freq, postings_data);
         self.data = ReadOnlySource::new(postings_data);
+        self.block_max_score_cache = None;
         self.loaded_offset = std::usize::MAX;
         if let Some(skip_data) = skip_data_opt {
             self.skip_reader.reset(skip_data, doc_freq);
@@ -242,7 +254,7 @@ impl BlockSegmentPostings {
     /// If all docs are smaller than target, the block loaded may be empty,
     /// or be the last an incomplete VInt block.
     pub fn seek(&mut self, target_doc: DocId) {
-        self.skip_reader.seek(target_doc);
+        self.shallow_seek(target_doc);
         self.load_block();
     }
 
@@ -257,7 +269,9 @@ impl BlockSegmentPostings {
     /// If all docs are smaller than target, the block loaded may be empty,
     /// or be the last an incomplete VInt block.
     pub(crate) fn shallow_seek(&mut self, target_doc: DocId) {
-        self.skip_reader.seek(target_doc);
+        if self.skip_reader.seek(target_doc) {
+            self.block_max_score_cache = None;
+        }
     }
 
     pub(crate) fn block_is_loaded(&self) -> bool {
@@ -265,7 +279,6 @@ impl BlockSegmentPostings {
     }
 
     pub(crate) fn load_block(&mut self) {
-        self.block_max_score_cache = None;
         let offset = self.skip_reader.byte_offset();
         if self.loaded_offset == offset {
             return;
@@ -318,6 +331,7 @@ impl BlockSegmentPostings {
     /// Returns false iff there was no remaining blocks.
     pub fn advance(&mut self) {
         self.skip_reader.advance();
+        self.block_max_score_cache = None;
         self.load_block();
     }
 
@@ -441,7 +455,7 @@ mod tests {
         let int_field = schema_builder.add_u64_field("id", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         let mut last_doc = 0u32;
         for &doc in docs {
             for _ in last_doc..doc {
@@ -460,21 +474,19 @@ mod tests {
     }
 
     #[test]
-    fn test_block_segment_postings_seek2() {
+    fn test_block_segment_postings_seek() {
         let mut docs = vec![0];
         for i in 0..1300 {
             docs.push((i * i / 100) + i);
         }
         let mut block_postings = build_block_postings(&docs[..]);
         for i in vec![0, 424, 10000] {
-            block_postings.shallow_seek(i);
-            block_postings.load_block();
+            block_postings.seek(i);
             let docs = block_postings.docs();
             assert!(docs[0] <= i);
             assert!(docs.last().cloned().unwrap_or(0u32) >= i);
         }
-        block_postings.shallow_seek(100_000);
-        block_postings.load_block();
+        block_postings.seek(100_000);
         assert_eq!(block_postings.doc(COMPRESSION_BLOCK_SIZE - 1), TERMINATED);
     }
 
@@ -484,7 +496,7 @@ mod tests {
         let int_field = schema_builder.add_u64_field("id", INDEXED);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         // create two postings list, one containg even number,
         // the other containing odd numbers.
         for i in 0..6 {

src/postings/compression/mod.rs

@@ -1,5 +1,4 @@
 use crate::common::FixedSize;
-use crate::docset::TERMINATED;
 use bitpacking::{BitPacker, BitPacker4x};
 
 pub const COMPRESSION_BLOCK_SIZE: usize = BitPacker4x::BLOCK_LEN;
@@ -53,8 +52,10 @@ impl BlockEncoder {
 /// We ensure that the OutputBuffer is align on 128 bits
 /// in order to run SSE2 linear search on it.
 #[repr(align(128))]
+#[derive(Clone)]
 pub(crate) struct AlignedBuffer(pub [u32; COMPRESSION_BLOCK_SIZE]);
 
+#[derive(Clone)]
 pub struct BlockDecoder {
     bitpacker: BitPacker4x,
     output: AlignedBuffer,
@@ -107,10 +108,6 @@ impl BlockDecoder {
     pub fn output(&self, idx: usize) -> u32 {
         self.output.0[idx]
     }
-
-    pub fn fill(&mut self, val: u32) {
-        self.output.0.iter_mut().for_each(|el| *el = val);
-    }
 }
 
 pub trait VIntEncoder {
@@ -147,11 +144,14 @@ pub trait VIntDecoder {
     /// For instance, if delta encoded are `1, 3, 9`, and the
     /// `offset` is 5, then the output will be:
     /// `5 + 1 = 6, 6 + 3= 9, 9 + 9 = 18`
+    ///
+    /// The value given in `padding` will be used to fill the remaining `128 - num_els` values.
     fn uncompress_vint_sorted(
         &mut self,
         compressed_data: &[u8],
         offset: u32,
         num_els: usize,
+        padding: u32,
     ) -> usize;
 
     /// Uncompress an array of `u32s`, compressed using variable
@@ -159,7 +159,14 @@ pub trait VIntDecoder {
     ///
     /// The method takes a number of int to decompress, and returns
     /// the amount of bytes that were read to decompress them.
-    fn uncompress_vint_unsorted(&mut self, compressed_data: &[u8], num_els: usize) -> usize;
+    ///
+    /// The value given in `padding` will be used to fill the remaining `128 - num_els` values.
+    fn uncompress_vint_unsorted(
+        &mut self,
+        compressed_data: &[u8],
+        num_els: usize,
+        padding: u32,
+    ) -> usize;
 }
 
 impl VIntEncoder for BlockEncoder {
@@ -178,13 +185,21 @@ impl VIntDecoder for BlockDecoder {
         compressed_data: &[u8],
         offset: u32,
         num_els: usize,
+        padding: u32,
     ) -> usize {
         self.output_len = num_els;
+        self.output.0.iter_mut().for_each(|el| *el = padding);
         vint::uncompress_sorted(compressed_data, &mut self.output.0[..num_els], offset)
     }
 
-    fn uncompress_vint_unsorted(&mut self, compressed_data: &[u8], num_els: usize) -> usize {
+    fn uncompress_vint_unsorted(
+        &mut self,
+        compressed_data: &[u8],
+        num_els: usize,
+        padding: u32,
+    ) -> usize {
         self.output_len = num_els;
+        self.output.0.iter_mut().for_each(|el| *el = padding);
         vint::uncompress_unsorted(compressed_data, &mut self.output.0[..num_els])
     }
 }
@@ -193,6 +208,7 @@ impl VIntDecoder for BlockDecoder {
 pub mod tests {
 
     use super::*;
+    use crate::TERMINATED;
 
     #[test]
     fn test_encode_sorted_block() {
@@ -271,18 +287,20 @@ pub mod tests {
     }
     #[test]
     fn test_encode_vint() {
-        {
-            let expected_length = 154;
-            let mut encoder = BlockEncoder::new();
-            let input: Vec<u32> = (0u32..123u32).map(|i| 4 + i * 7 / 2).into_iter().collect();
-            for offset in &[0u32, 1u32, 2u32] {
-                let encoded_data = encoder.compress_vint_sorted(&input, *offset);
-                assert!(encoded_data.len() <= expected_length);
-                let mut decoder = BlockDecoder::default();
-                let consumed_num_bytes =
-                    decoder.uncompress_vint_sorted(&encoded_data, *offset, input.len());
-                assert_eq!(consumed_num_bytes, encoded_data.len());
-                assert_eq!(input, decoder.output_array());
+        const PADDING_VALUE: u32 = 234_234_345u32;
+        let expected_length = 154;
+        let mut encoder = BlockEncoder::new();
+        let input: Vec<u32> = (0u32..123u32).map(|i| 4 + i * 7 / 2).into_iter().collect();
+        for offset in &[0u32, 1u32, 2u32] {
+            let encoded_data = encoder.compress_vint_sorted(&input, *offset);
+            assert!(encoded_data.len() <= expected_length);
+            let mut decoder = BlockDecoder::default();
+            let consumed_num_bytes =
+                decoder.uncompress_vint_sorted(&encoded_data, *offset, input.len(), PADDING_VALUE);
+            assert_eq!(consumed_num_bytes, encoded_data.len());
+            assert_eq!(input, decoder.output_array());
+            for i in input.len()..COMPRESSION_BLOCK_SIZE {
+                assert_eq!(decoder.output(i), PADDING_VALUE);
             }
         }
     }
@@ -292,6 +310,7 @@ pub mod tests {
 mod bench {
 
     use super::*;
+    use crate::TERMINATED;
     use rand::rngs::StdRng;
     use rand::Rng;
     use rand::SeedableRng;
@@ -322,7 +341,7 @@ mod bench {
         let mut encoder = BlockEncoder::new();
         let data = generate_array(COMPRESSION_BLOCK_SIZE, 0.1);
         let (num_bits, compressed) = encoder.compress_block_sorted(&data, 0u32);
-        let mut decoder = BlockDecoder::new();
+        let mut decoder = BlockDecoder::default();
         b.iter(|| {
             decoder.uncompress_block_sorted(compressed, 0u32, num_bits);
         });
@@ -357,9 +376,9 @@ mod bench {
         let mut encoder = BlockEncoder::new();
         let data = generate_array(NUM_INTS_BENCH_VINT, 0.001);
         let compressed = encoder.compress_vint_sorted(&data, 0u32);
-        let mut decoder = BlockDecoder::new();
+        let mut decoder = BlockDecoder::default();
         b.iter(|| {
-            decoder.uncompress_vint_sorted(compressed, 0u32, NUM_INTS_BENCH_VINT);
+            decoder.uncompress_vint_sorted(compressed, 0u32, NUM_INTS_BENCH_VINT, TERMINATED);
         });
     }
 }

src/postings/mod.rs

@@ -5,6 +5,7 @@ Postings module (also called inverted index)
 mod block_search;
 mod block_segment_postings;
 pub(crate) mod compression;
+mod field_stats;
 mod postings;
 mod postings_writer;
 mod recorder;
@@ -15,6 +16,7 @@ mod stacker;
 mod term_info;
 
 pub(crate) use self::block_search::BlockSearcher;
+pub(crate) use self::field_stats::{FieldStat, FieldStats};
 
 pub(crate) use self::postings_writer::MultiFieldPostingsWriter;
 pub use self::serializer::{FieldSerializer, InvertedIndexSerializer};
@@ -65,47 +67,42 @@ pub mod tests {
     use std::iter;
 
     #[test]
-    pub fn test_position_write() {
+    pub fn test_position_write() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut segment = index.new_segment();
-        let mut posting_serializer = InvertedIndexSerializer::open(&mut segment).unwrap();
-        {
-            let mut field_serializer = posting_serializer
-                .new_field(text_field, 120 * 4, None)
-                .unwrap();
-            field_serializer.new_term("abc".as_bytes(), 12u32).unwrap();
-            for doc_id in 0u32..120u32 {
-                let delta_positions = vec![1, 2, 3, 2];
-                field_serializer
-                    .write_doc(doc_id, 4, &delta_positions)
-                    .unwrap();
-            }
-            field_serializer.close_term().unwrap();
+        let mut posting_serializer = InvertedIndexSerializer::open(&mut segment)?;
+        let mut field_serializer = posting_serializer.new_field(text_field, 120 * 4, None)?;
+        field_serializer.new_term("abc".as_bytes(), 12u32)?;
+        for doc_id in 0u32..120u32 {
+            let delta_positions = vec![1, 2, 3, 2];
+            field_serializer.write_doc(doc_id, 4, &delta_positions)?;
         }
-        posting_serializer.close().unwrap();
-        let read = segment.open_read(SegmentComponent::POSITIONS).unwrap();
+        field_serializer.close_term()?;
+        posting_serializer.close()?;
+        let read = segment.open_read(SegmentComponent::POSITIONS)?;
         assert!(read.len() <= 140);
+        Ok(())
     }
 
     #[test]
-    pub fn test_skip_positions() {
+    pub fn test_skip_positions() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();
         let title = schema_builder.add_text_field("title", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 30_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests()?;
         index_writer.add_document(doc!(title => r#"abc abc abc"#));
         index_writer.add_document(doc!(title => r#"abc be be be be abc"#));
         for _ in 0..1_000 {
             index_writer.add_document(doc!(title => r#"abc abc abc"#));
         }
         index_writer.add_document(doc!(title => r#"abc be be be be abc"#));
-        index_writer.commit().unwrap();
+        index_writer.commit()?;
 
-        let searcher = index.reader().unwrap().searcher();
+        let searcher = index.reader()?.searcher();
         let inverted_index = searcher.segment_reader(0u32).inverted_index(title);
         let term = Term::from_field_text(title, "abc");
         let mut positions = Vec::new();
@@ -160,6 +157,7 @@ pub mod tests {
             postings.positions(&mut positions);
             assert_eq!(&[0, 5], &positions[..]);
         }
+        Ok(())
     }
 
     #[test]
@@ -180,7 +178,7 @@ pub mod tests {
             .tokenizers()
             .register("simple_no_truncation", SimpleTokenizer);
         let reader = index.reader().unwrap();
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.set_merge_policy(Box::new(NoMergePolicy));
         {
             index_writer.add_document(doc!(text_field=>exceeding_token_text));
@@ -209,7 +207,7 @@ pub mod tests {
     }
 
     #[test]
-    pub fn test_position_and_fieldnorm1() {
+    pub fn test_position_and_fieldnorm1() -> crate::Result<()> {
         let mut positions = Vec::new();
         let mut schema_builder = Schema::builder();
         let text_field = schema_builder.add_text_field("text", TEXT);
@@ -221,42 +219,38 @@ pub mod tests {
             let mut segment_writer =
                 SegmentWriter::for_segment(3_000_000, segment.clone(), &schema).unwrap();
             {
-                let mut doc = Document::default();
                 // checking that position works if the field has two values
-                doc.add_text(text_field, "a b a c a d a a.");
-                doc.add_text(text_field, "d d d d a");
                 let op = AddOperation {
                     opstamp: 0u64,
-                    document: doc,
+                    document: doc!(
+                       text_field => "a b a c a d a a.",
+                       text_field => "d d d d a"
+                    ),
                 };
-                segment_writer.add_document(op, &schema).unwrap();
+                segment_writer.add_document(op, &schema)?;
             }
             {
-                let mut doc = Document::default();
-                doc.add_text(text_field, "b a");
                 let op = AddOperation {
                     opstamp: 1u64,
-                    document: doc,
+                    document: doc!(text_field => "b a"),
                 };
                 segment_writer.add_document(op, &schema).unwrap();
             }
             for i in 2..1000 {
-                let mut doc = Document::default();
-                let mut text = iter::repeat("e ").take(i).collect::<String>();
+                let mut text: String = iter::repeat("e ").take(i).collect();
                 text.push_str(" a");
-                doc.add_text(text_field, &text);
                 let op = AddOperation {
                     opstamp: 2u64,
-                    document: doc,
+                    document: doc!(text_field => text),
                 };
                 segment_writer.add_document(op, &schema).unwrap();
             }
-            segment_writer.finalize().unwrap();
+            segment_writer.finalize()?;
         }
         {
-            let segment_reader = SegmentReader::open(&segment).unwrap();
+            let segment_reader = SegmentReader::open(&segment)?;
             {
-                let fieldnorm_reader = segment_reader.get_fieldnorms_reader(text_field);
+                let fieldnorm_reader = segment_reader.get_fieldnorms_reader(text_field)?;
                 assert_eq!(fieldnorm_reader.fieldnorm(0), 8 + 5);
                 assert_eq!(fieldnorm_reader.fieldnorm(1), 2);
                 for i in 2..1000 {
@@ -316,6 +310,7 @@ pub mod tests {
                 assert_eq!(postings_e.doc(), TERMINATED);
             }
         }
+        Ok(())
     }
 
     #[test]
@@ -326,7 +321,7 @@ pub mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text_field => "g b b d c g c"));
             index_writer.add_document(doc!(text_field => "g a b b a d c g c"));
             assert!(index_writer.commit().is_ok());
@@ -358,7 +353,7 @@ pub mod tests {
 
             let index = Index::create_in_ram(schema);
             {
-                let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+                let mut index_writer = index.writer_for_tests().unwrap();
                 for i in 0u64..num_docs as u64 {
                     let doc = doc!(value_field => 2u64, value_field => i % 2u64);
                     index_writer.add_document(doc);
@@ -429,7 +424,7 @@ pub mod tests {
 
         // delete some of the documents
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.delete_term(term_0);
             assert!(index_writer.commit().is_ok());
         }
@@ -483,7 +478,7 @@ pub mod tests {
 
         // delete everything else
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.delete_term(term_1);
             assert!(index_writer.commit().is_ok());
         }
@@ -526,7 +521,7 @@ pub mod tests {
         let index = Index::create_in_ram(schema);
         let posting_list_size = 1_000_000;
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             for _ in 0..posting_list_size {
                 let mut doc = Document::default();
                 if rng.gen_bool(1f64 / 15f64) {
@@ -734,7 +729,7 @@ mod bench {
             let mut s = 0u32;
             while segment_postings.doc() != TERMINATED {
                 s += (segment_postings.doc() & n) % 1024;
-                segment_postings.advance()
+                segment_postings.advance();
             }
             s
         });

src/postings/postings_writer.rs

@@ -105,6 +105,7 @@ impl MultiFieldPostingsWriter {
         doc: DocId,
         field: Field,
         token_stream: &mut dyn TokenStream,
+        term_buffer: &mut Term,
     ) -> u32 {
         let postings_writer =
             self.per_field_postings_writers[field.field_id() as usize].deref_mut();
@@ -114,6 +115,7 @@ impl MultiFieldPostingsWriter {
             field,
             token_stream,
             &mut self.heap,
+            term_buffer,
         )
     }
 
@@ -220,13 +222,20 @@ pub trait PostingsWriter {
         field: Field,
         token_stream: &mut dyn TokenStream,
         heap: &mut MemoryArena,
+        term_buffer: &mut Term,
     ) -> u32 {
-        let mut term = Term::for_field(field);
+        term_buffer.set_field(field);
         let mut sink = |token: &Token| {
             // We skip all tokens with a len greater than u16.
             if token.text.len() <= MAX_TOKEN_LEN {
-                term.set_text(token.text.as_str());
-                self.subscribe(term_index, doc_id, token.position as u32, &term, heap);
+                term_buffer.set_text(token.text.as_str());
+                self.subscribe(
+                    term_index,
+                    doc_id,
+                    token.position as u32,
+                    &term_buffer,
+                    heap,
+                );
             } else {
                 info!(
                     "A token exceeding MAX_TOKEN_LEN ({}>{}) was dropped. Search for \

src/postings/recorder.rs

@@ -75,7 +75,9 @@ pub(crate) trait Recorder: Copy + 'static {
         serializer: &mut FieldSerializer<'_>,
         heap: &MemoryArena,
     ) -> io::Result<()>;
-    /// Returns the number of document containg this term.
+    /// Returns the number of document containing this term.
+    ///
+    /// Returns `None` if not available.
     fn term_doc_freq(&self) -> Option<u32>;
 }
 

src/postings/segment_postings.rs

@@ -10,10 +10,10 @@ use crate::postings::BlockSearcher;
 use crate::postings::Postings;
 
 use crate::schema::IndexRecordOption;
-use crate::DocId;
+use crate::{DocId, TERMINATED};
 
 use crate::directory::ReadOnlySource;
-use crate::fieldnorm::FieldNormReader;
+use crate::fastfield::DeleteBitSet;
 use crate::postings::BlockSegmentPostings;
 
 /// `SegmentPostings` represents the inverted list or postings associated to
@@ -21,6 +21,7 @@ use crate::postings::BlockSegmentPostings;
 ///
 /// As we iterate through the `SegmentPostings`, the frequencies are optionally decoded.
 /// Positions on the other hand, are optionally entirely decoded upfront.
+#[derive(Clone)]
 pub struct SegmentPostings {
     pub(crate) block_cursor: BlockSegmentPostings,
     cur: usize,
@@ -39,6 +40,25 @@ impl SegmentPostings {
         }
     }
 
+    /// Compute the number of non-deleted documents.
+    ///
+    /// This method will clone and scan through the posting lists.
+    /// (this is a rather expensive operation).
+    pub fn doc_freq_given_deletes(&self, delete_bitset: &DeleteBitSet) -> u32 {
+        let mut docset = self.clone();
+        let mut doc_freq = 0;
+        loop {
+            let doc = docset.doc();
+            if doc == TERMINATED {
+                return doc_freq;
+            }
+            if delete_bitset.is_alive(doc) {
+                doc_freq += 1u32;
+            }
+            docset.advance();
+        }
+    }
+
     /// Returns the overall number of documents in the block postings.
     /// It does not take in account whether documents are deleted or not.
     pub fn doc_freq(&self) -> u32 {
@@ -56,7 +76,8 @@ impl SegmentPostings {
     pub fn create_from_docs(docs: &[u32]) -> SegmentPostings {
         let mut buffer = Vec::new();
         {
-            let mut postings_serializer = PostingsSerializer::new(&mut buffer, false, false, None);
+            let mut postings_serializer =
+                PostingsSerializer::new(&mut buffer, 0.0, false, false, None);
             postings_serializer.new_term(docs.len() as u32);
             for &doc in docs {
                 postings_serializer.write_doc(doc, 1u32);
@@ -75,26 +96,48 @@ impl SegmentPostings {
     }
 
     /// Helper functions to create `SegmentPostings` for tests.
+    #[cfg(test)]
     pub fn create_from_docs_and_tfs(
         doc_and_tfs: &[(u32, u32)],
-        fieldnorm_reader: Option<FieldNormReader>,
-    ) -> crate::Result<SegmentPostings> {
-        let mut buffer = Vec::new();
-        let mut postings_serializer =
-            PostingsSerializer::new(&mut buffer, true, false, fieldnorm_reader);
+        fieldnorms: Option<&[u32]>,
+    ) -> SegmentPostings {
+        use crate::fieldnorm::FieldNormReader;
+        use crate::Score;
+        let mut buffer: Vec<u8> = Vec::new();
+        let fieldnorm_reader = fieldnorms.map(FieldNormReader::for_test);
+        let average_field_norm = fieldnorms
+            .map(|fieldnorms| {
+                if fieldnorms.len() == 0 {
+                    return 0.0;
+                }
+                let total_num_tokens: u64 = fieldnorms
+                    .iter()
+                    .map(|&fieldnorm| fieldnorm as u64)
+                    .sum::<u64>();
+                total_num_tokens as Score / fieldnorms.len() as Score
+            })
+            .unwrap_or(0.0);
+        let mut postings_serializer = PostingsSerializer::new(
+            &mut buffer,
+            average_field_norm,
+            true,
+            false,
+            fieldnorm_reader,
+        );
         postings_serializer.new_term(doc_and_tfs.len() as u32);
         for &(doc, tf) in doc_and_tfs {
             postings_serializer.write_doc(doc, tf);
         }
         postings_serializer
-            .close_term(doc_and_tfs.len() as u32)?;
+            .close_term(doc_and_tfs.len() as u32)
+            .unwrap();
         let block_segment_postings = BlockSegmentPostings::from_data(
             doc_and_tfs.len() as u32,
             ReadOnlySource::from(buffer),
             IndexRecordOption::WithFreqs,
             IndexRecordOption::WithFreqs,
         );
-        Ok(SegmentPostings::from_block_postings(block_segment_postings, None))
+        SegmentPostings::from_block_postings(block_segment_postings, None)
     }
 
     /// Reads a Segment postings from an &[u8]
@@ -121,7 +164,7 @@ impl DocSet for SegmentPostings {
     // next needs to be called a first time to point to the correct element.
     #[inline]
     fn advance(&mut self) -> DocId {
-        assert!(self.block_cursor.block_is_loaded());
+        debug_assert!(self.block_cursor.block_is_loaded());
         if self.cur == COMPRESSION_BLOCK_SIZE - 1 {
             self.cur = 0;
             self.block_cursor.advance();
@@ -228,6 +271,7 @@ mod tests {
     use crate::common::HasLen;
 
     use crate::docset::{DocSet, TERMINATED};
+    use crate::fastfield::DeleteBitSet;
     use crate::postings::postings::Postings;
 
     #[test]
@@ -250,4 +294,14 @@ mod tests {
         let postings = SegmentPostings::empty();
         assert_eq!(postings.term_freq(), 1);
     }
+
+    #[test]
+    fn test_doc_freq() {
+        let docs = SegmentPostings::create_from_docs(&[0, 2, 10]);
+        assert_eq!(docs.doc_freq(), 3);
+        let delete_bitset = DeleteBitSet::for_test(&[2], 12);
+        assert_eq!(docs.doc_freq_given_deletes(&delete_bitset), 2);
+        let all_deleted = DeleteBitSet::for_test(&[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11], 12);
+        assert_eq!(docs.doc_freq_given_deletes(&all_deleted), 0);
+    }
 }

src/postings/serializer.rs

@@ -1,5 +1,4 @@
-use super::TermInfo;
-use crate::common::{BinarySerializable, VInt};
+use super::{FieldStat, FieldStats, TermInfo};
 use crate::common::{CompositeWrite, CountingWriter};
 use crate::core::Segment;
 use crate::directory::WritePtr;
@@ -11,7 +10,11 @@ use crate::query::BM25Weight;
 use crate::schema::Schema;
 use crate::schema::{Field, FieldEntry, FieldType};
 use crate::termdict::{TermDictionaryBuilder, TermOrdinal};
-use crate::DocId;
+use crate::{
+    common::{BinarySerializable, VInt},
+    directory::TerminatingWrite,
+};
+use crate::{DocId, Score};
 use std::cmp::Ordering;
 use std::io::{self, Write};
 
@@ -51,6 +54,8 @@ pub struct InvertedIndexSerializer {
     postings_write: CompositeWrite<WritePtr>,
     positions_write: CompositeWrite<WritePtr>,
     positionsidx_write: CompositeWrite<WritePtr>,
+    field_stats: FieldStats,
+    field_stats_write: WritePtr,
     schema: Schema,
 }
 
@@ -61,6 +66,7 @@ impl InvertedIndexSerializer {
         postings_write: CompositeWrite<WritePtr>,
         positions_write: CompositeWrite<WritePtr>,
         positionsidx_write: CompositeWrite<WritePtr>,
+        field_stats_write: WritePtr,
         schema: Schema,
     ) -> crate::Result<InvertedIndexSerializer> {
         Ok(InvertedIndexSerializer {
@@ -68,18 +74,21 @@ impl InvertedIndexSerializer {
             postings_write,
             positions_write,
             positionsidx_write,
+            field_stats: FieldStats::default(),
+            field_stats_write,
             schema,
         })
     }
 
     /// Open a new `PostingsSerializer` for the given segment
     pub fn open(segment: &mut Segment) -> crate::Result<InvertedIndexSerializer> {
-        use crate::SegmentComponent::{POSITIONS, POSITIONSSKIP, POSTINGS, TERMS};
+        use crate::SegmentComponent::{FIELDSTATS, POSITIONS, POSITIONSSKIP, POSTINGS, TERMS};
         InvertedIndexSerializer::create(
             CompositeWrite::wrap(segment.open_write(TERMS)?),
             CompositeWrite::wrap(segment.open_write(POSTINGS)?),
             CompositeWrite::wrap(segment.open_write(POSITIONS)?),
             CompositeWrite::wrap(segment.open_write(POSITIONSSKIP)?),
+            segment.open_write(FIELDSTATS)?,
             segment.schema(),
         )
     }
@@ -94,15 +103,17 @@ impl InvertedIndexSerializer {
         total_num_tokens: u64,
         fieldnorm_reader: Option<FieldNormReader>,
     ) -> io::Result<FieldSerializer<'_>> {
+        self.field_stats
+            .insert(field, FieldStat::new(total_num_tokens));
         let field_entry: &FieldEntry = self.schema.get_field_entry(field);
         let term_dictionary_write = self.terms_write.for_field(field);
         let postings_write = self.postings_write.for_field(field);
-        total_num_tokens.serialize(postings_write)?;
         let positions_write = self.positions_write.for_field(field);
         let positionsidx_write = self.positionsidx_write.for_field(field);
         let field_type: FieldType = (*field_entry.field_type()).clone();
         FieldSerializer::create(
             &field_type,
+            total_num_tokens,
             term_dictionary_write,
             postings_write,
             positions_write,
@@ -112,7 +123,10 @@ impl InvertedIndexSerializer {
     }
 
     /// Closes the serializer.
-    pub fn close(self) -> io::Result<()> {
+    pub fn close(mut self) -> io::Result<()> {
+        self.field_stats
+            .serialize(self.field_stats_write.get_mut())?;
+        self.field_stats_write.terminate()?;
         self.terms_write.close()?;
         self.postings_write.close()?;
         self.positions_write.close()?;
@@ -135,6 +149,7 @@ pub struct FieldSerializer<'a> {
 impl<'a> FieldSerializer<'a> {
     fn create(
         field_type: &FieldType,
+        total_num_tokens: u64,
         term_dictionary_write: &'a mut CountingWriter<WritePtr>,
         postings_write: &'a mut CountingWriter<WritePtr>,
         positions_write: &'a mut CountingWriter<WritePtr>,
@@ -153,8 +168,13 @@ impl<'a> FieldSerializer<'a> {
             _ => (false, false),
         };
         let term_dictionary_builder = TermDictionaryBuilder::create(term_dictionary_write)?;
+        let average_fieldnorm = fieldnorm_reader
+            .as_ref()
+            .map(|ff_reader| (total_num_tokens as Score / ff_reader.num_docs() as Score))
+            .unwrap_or(0.0);
         let postings_serializer = PostingsSerializer::new(
             postings_write,
+            average_fieldnorm,
             term_freq_enabled,
             position_enabled,
             fieldnorm_reader,
@@ -183,7 +203,8 @@ impl<'a> FieldSerializer<'a> {
             .unwrap_or(0u64);
         TermInfo {
             doc_freq: 0,
-            postings_offset: self.postings_serializer.addr(),
+            postings_start_offset: self.postings_serializer.addr(),
+            postings_end_offset: 0u64,
             positions_idx,
         }
     }
@@ -197,6 +218,7 @@ impl<'a> FieldSerializer<'a> {
             !self.term_open,
             "Called new_term, while the previous term was not closed."
         );
+
         self.term_open = true;
         self.postings_serializer.clear();
         self.current_term_info = self.current_term_info();
@@ -236,10 +258,12 @@ impl<'a> FieldSerializer<'a> {
     /// using `VInt` encoding.
     pub fn close_term(&mut self) -> io::Result<()> {
         if self.term_open {
-            self.term_dictionary_builder
-                .insert_value(&self.current_term_info)?;
             self.postings_serializer
                 .close_term(self.current_term_info.doc_freq)?;
+            let end_offset = self.postings_serializer.addr();
+            self.current_term_info.postings_end_offset = end_offset;
+            self.term_dictionary_builder
+                .insert_value(&self.current_term_info)?;
             self.term_open = false;
         }
         Ok(())
@@ -322,29 +346,18 @@ pub struct PostingsSerializer<W: Write> {
     bm25_weight: Option<BM25Weight>,
 
     num_docs: u32, // Number of docs in the segment
-    avg_fieldnorm: f32, // Average number of term in the field for that segment.
+    avg_fieldnorm: Score, // Average number of term in the field for that segment.
                    // this value is used to compute the block wand information.
 }
 
-fn get_avg_fieldnorm(fieldnorm_reader: &FieldNormReader) -> f32 {
-    let num_docs = fieldnorm_reader.num_docs();
-    let sum_fieldnorm: f32 = (0u32..num_docs)
-        .map(|doc| fieldnorm_reader.fieldnorm(doc) as f32)
-        .sum();
-    sum_fieldnorm / (num_docs as f32)
-}
-
 impl<W: Write> PostingsSerializer<W> {
     pub fn new(
         write: W,
+        avg_fieldnorm: Score,
         termfreq_enabled: bool,
         termfreq_sum_enabled: bool,
         fieldnorm_reader: Option<FieldNormReader>,
     ) -> PostingsSerializer<W> {
-        let avg_fieldnorm: f32 = fieldnorm_reader
-            .as_ref()
-            .map(get_avg_fieldnorm)
-            .unwrap_or(0f32);
         let num_docs = fieldnorm_reader
             .as_ref()
             .map(|fieldnorm_reader| fieldnorm_reader.num_docs())
@@ -403,16 +416,14 @@ impl<W: Write> PostingsSerializer<W> {
                 let sum_freq = self.block.term_freqs().iter().cloned().sum();
                 self.skip_write.write_total_term_freq(sum_freq);
             }
-            let mut blockwand_params_opt = None;
+            let mut blockwand_params = (0u8, 0u32);
             if let Some(bm25_weight) = self.bm25_weight.as_ref() {
                 if let Some(fieldnorm_reader) = self.fieldnorm_reader.as_ref() {
-                    let docs = self.block.doc_ids();
-                    let term_freqs = self.block.term_freqs();
-                    blockwand_params_opt = docs
-                        .iter()
-                        .cloned()
-                        .map(|doc| fieldnorm_reader.fieldnorm_id(doc))
-                        .zip(term_freqs.iter().cloned())
+                    let docs = self.block.doc_ids().iter().cloned();
+                    let term_freqs = self.block.term_freqs().iter().cloned();
+                    let fieldnorms = docs.map(|doc| fieldnorm_reader.fieldnorm_id(doc));
+                    blockwand_params = fieldnorms
+                        .zip(term_freqs)
                         .max_by(
                             |(left_fieldnorm_id, left_term_freq),
                              (right_fieldnorm_id, right_term_freq)| {
@@ -424,10 +435,11 @@ impl<W: Write> PostingsSerializer<W> {
                                     .partial_cmp(&right_score)
                                     .unwrap_or(Ordering::Equal)
                             },
-                        );
+                        )
+                        .unwrap();
                 }
             }
-            let (fieldnorm_id, term_freq) = blockwand_params_opt.unwrap_or((0u8, 0u32));
+            let (fieldnorm_id, term_freq) = blockwand_params;
             self.skip_write.write_blockwand_max(fieldnorm_id, term_freq);
         }
         self.block.clear();

src/postings/skip.rs

@@ -1,4 +1,4 @@
-use crate::common::{BinarySerializable, VInt};
+use crate::common::{read_u32_vint_no_advance, serialize_vint_u32, BinarySerializable, VInt};
 use crate::directory::ReadOnlySource;
 use crate::postings::compression::{compressed_block_size, COMPRESSION_BLOCK_SIZE};
 use crate::query::BM25Weight;
@@ -43,7 +43,9 @@ impl SkipSerializer {
 
     pub fn write_blockwand_max(&mut self, fieldnorm_id: u8, term_freq: u32) {
         self.buffer.push(fieldnorm_id);
-        VInt(term_freq as u64).serialize_into_vec(&mut self.buffer);
+        let mut buf = [0u8; 8];
+        let bytes = serialize_vint_u32(term_freq, &mut buf);
+        self.buffer.extend_from_slice(bytes);
     }
 
     pub fn data(&self) -> &[u8] {
@@ -56,6 +58,7 @@ impl SkipSerializer {
     }
 }
 
+#[derive(Clone)]
 pub(crate) struct SkipReader {
     last_doc_in_block: DocId,
     pub(crate) last_doc_in_previous_block: DocId,
@@ -69,7 +72,7 @@ pub(crate) struct SkipReader {
     position_offset: u64,
 }
 
-#[derive(Clone, Copy, Debug)]
+#[derive(Clone, Eq, PartialEq, Copy, Debug)]
 pub(crate) enum BlockInfo {
     BitPacked {
         doc_num_bits: u8,
@@ -128,6 +131,10 @@ impl SkipReader {
         }
     }
 
+    // Returns the block max score for this block if available.
+    //
+    // The block max score is available for all full bitpacked block,
+    // but no available for the last VInt encoded incomplete block.
     pub fn block_max_score(&self, bm25_weight: &BM25Weight) -> Option<Score> {
         match self.block_info {
             BlockInfo::BitPacked {
@@ -169,9 +176,9 @@ impl SkipReader {
             IndexRecordOption::WithFreqs => {
                 let tf_num_bits = self.owned_read.get(1);
                 let block_wand_fieldnorm_id = self.owned_read.get(2);
-                self.owned_read.advance(3);
-                let block_wand_term_freq =
-                    VInt::deserialize_u64(&mut self.owned_read).unwrap() as u32;
+                let data = &self.owned_read.as_ref()[3..];
+                let (block_wand_term_freq, num_bytes) = read_u32_vint_no_advance(data);
+                self.owned_read.advance(3 + num_bytes);
                 self.block_info = BlockInfo::BitPacked {
                     doc_num_bits,
                     tf_num_bits,
@@ -207,9 +214,15 @@ impl SkipReader {
     ///
     /// If the target is larger than all documents, the skip_reader
     /// then advance to the last Variable In block.
-    pub fn seek(&mut self, target: DocId) {
-        while self.last_doc_in_block() < target {
+    pub fn seek(&mut self, target: DocId) -> bool {
+        if self.last_doc_in_block() >= target {
+            return false;
+        }
+        loop {
             self.advance();
+            if self.last_doc_in_block() >= target {
+                return true;
+            }
         }
     }
 
@@ -225,7 +238,7 @@ impl SkipReader {
                 self.byte_offset += compressed_block_size(doc_num_bits + tf_num_bits);
                 self.position_offset += tf_sum as u64;
             }
-            BlockInfo::VInt { num_docs} => {
+            BlockInfo::VInt { num_docs } => {
                 debug_assert_eq!(num_docs, self.remaining_docs);
                 self.remaining_docs = 0;
                 self.byte_offset = std::usize::MAX;
@@ -236,7 +249,9 @@ impl SkipReader {
             self.read_block_info();
         } else {
             self.last_doc_in_block = TERMINATED;
-            self.block_info = BlockInfo::VInt { num_docs: self.remaining_docs };
+            self.block_info = BlockInfo::VInt {
+                num_docs: self.remaining_docs,
+            };
         }
     }
 }
@@ -269,7 +284,7 @@ mod tests {
             IndexRecordOption::WithFreqs,
         );
         assert_eq!(skip_reader.last_doc_in_block(), 1u32);
-        assert!(matches!(
+        assert_eq!(
             skip_reader.block_info,
             BlockInfo::BitPacked {
                 doc_num_bits: 2u8,
@@ -278,10 +293,10 @@ mod tests {
                 block_wand_fieldnorm_id: 13,
                 block_wand_term_freq: 3
             }
-       ));
+        );
         skip_reader.advance();
         assert_eq!(skip_reader.last_doc_in_block(), 5u32);
-        assert!(matches!(
+        assert_eq!(
             skip_reader.block_info(),
             BlockInfo::BitPacked {
                 doc_num_bits: 5u8,
@@ -290,13 +305,13 @@ mod tests {
                 block_wand_fieldnorm_id: 8,
                 block_wand_term_freq: 2
             }
-       ));
+        );
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 3u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 3u32 });
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 });
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 });
     }
 
     #[test]
@@ -314,7 +329,7 @@ mod tests {
             IndexRecordOption::Basic,
         );
         assert_eq!(skip_reader.last_doc_in_block(), 1u32);
-        assert!(matches!(
+        assert_eq!(
             skip_reader.block_info(),
             BlockInfo::BitPacked {
                 doc_num_bits: 2u8,
@@ -323,10 +338,10 @@ mod tests {
                 block_wand_fieldnorm_id: 0,
                 block_wand_term_freq: 0
             }
-        ));
+        );
         skip_reader.advance();
         assert_eq!(skip_reader.last_doc_in_block(), 5u32);
-        assert!(matches!(
+        assert_eq!(
             skip_reader.block_info(),
             BlockInfo::BitPacked {
                 doc_num_bits: 5u8,
@@ -335,13 +350,13 @@ mod tests {
                 block_wand_fieldnorm_id: 0,
                 block_wand_term_freq: 0
             }
-        ));
+        );
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 3u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 3u32 });
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 });
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 });
     }
 
     #[test]
@@ -358,7 +373,7 @@ mod tests {
             IndexRecordOption::Basic,
         );
         assert_eq!(skip_reader.last_doc_in_block(), 1u32);
-        assert!(matches!(
+        assert_eq!(
             skip_reader.block_info(),
             BlockInfo::BitPacked {
                 doc_num_bits: 2u8,
@@ -367,8 +382,8 @@ mod tests {
                 block_wand_fieldnorm_id: 0,
                 block_wand_term_freq: 0
             }
-        ));
+        );
         skip_reader.advance();
-        assert!(matches!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 }));
+        assert_eq!(skip_reader.block_info(), BlockInfo::VInt { num_docs: 0u32 });
     }
 }

src/postings/term_info.rs

@@ -7,35 +7,49 @@ use std::io;
 pub struct TermInfo {
     /// Number of documents in the segment containing the term
     pub doc_freq: u32,
-    /// Start offset within the postings (`.idx`) file.
-    pub postings_offset: u64,
+    /// Start offset of the posting list within the postings (`.idx`) file.
+    pub postings_start_offset: u64,
+    /// End offset of the posting list within the postings (`.idx`) file.
+    pub postings_end_offset: u64,
     /// Start offset of the first block within the position (`.pos`) file.
     pub positions_idx: u64,
 }
 
+impl TermInfo {
+    pub(crate) fn posting_num_bytes(&self) -> u32 {
+        let num_bytes = self.postings_end_offset - self.postings_start_offset;
+        assert!(num_bytes <= std::u32::MAX as u64);
+        num_bytes as u32
+    }
+}
+
 impl FixedSize for TermInfo {
     /// Size required for the binary serialization of a `TermInfo` object.
     /// This is large, but in practise, `TermInfo` are encoded in blocks and
     /// only the first `TermInfo` of a block is serialized uncompressed.
     /// The subsequent `TermInfo` are delta encoded and bitpacked.
-    const SIZE_IN_BYTES: usize = u32::SIZE_IN_BYTES + 2 * u64::SIZE_IN_BYTES;
+    const SIZE_IN_BYTES: usize = 2 * u32::SIZE_IN_BYTES + 2 * u64::SIZE_IN_BYTES;
 }
 
 impl BinarySerializable for TermInfo {
     fn serialize<W: io::Write>(&self, writer: &mut W) -> io::Result<()> {
         self.doc_freq.serialize(writer)?;
-        self.postings_offset.serialize(writer)?;
+        self.postings_start_offset.serialize(writer)?;
+        self.posting_num_bytes().serialize(writer)?;
         self.positions_idx.serialize(writer)?;
         Ok(())
     }
 
     fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
         let doc_freq = u32::deserialize(reader)?;
-        let postings_offset = u64::deserialize(reader)?;
+        let postings_start_offset = u64::deserialize(reader)?;
+        let postings_num_bytes = u32::deserialize(reader)?;
+        let postings_end_offset = postings_start_offset + u64::from(postings_num_bytes);
         let positions_idx = u64::deserialize(reader)?;
         Ok(TermInfo {
             doc_freq,
-            postings_offset,
+            postings_start_offset,
+            postings_end_offset,
             positions_idx,
         })
     }

src/query/all_query.rs

@@ -9,7 +9,7 @@ use crate::Score;
 
 /// Query that matches all of the documents.
 ///
-/// All of the document get the score 1f32.
+/// All of the document get the score 1.0.
 #[derive(Clone, Debug)]
 pub struct AllQuery;
 
@@ -23,7 +23,7 @@ impl Query for AllQuery {
 pub struct AllWeight;
 
 impl Weight for AllWeight {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> crate::Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         let all_scorer = AllScorer {
             doc: 0u32,
             max_doc: reader.max_doc(),
@@ -35,7 +35,7 @@ impl Weight for AllWeight {
         if doc >= reader.max_doc() {
             return Err(does_not_match(doc));
         }
-        Ok(Explanation::new("AllQuery", 1f32))
+        Ok(Explanation::new("AllQuery", 1.0))
     }
 }
 
@@ -66,7 +66,7 @@ impl DocSet for AllScorer {
 
 impl Scorer for AllScorer {
     fn score(&mut self) -> Score {
-        1f32
+        1.0
     }
 }
 
@@ -83,7 +83,7 @@ mod tests {
         let field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(field=>"aaa"));
         index_writer.add_document(doc!(field=>"bbb"));
         index_writer.commit().unwrap();
@@ -100,7 +100,7 @@ mod tests {
         let weight = AllQuery.weight(&searcher, false).unwrap();
         {
             let reader = searcher.segment_reader(0);
-            let mut scorer = weight.scorer(reader, 1.0f32).unwrap();
+            let mut scorer = weight.scorer(reader, 1.0).unwrap();
             assert_eq!(scorer.doc(), 0u32);
             assert_eq!(scorer.advance(), 1u32);
             assert_eq!(scorer.doc(), 1u32);
@@ -108,7 +108,7 @@ mod tests {
         }
         {
             let reader = searcher.segment_reader(1);
-            let mut scorer = weight.scorer(reader, 1.0f32).unwrap();
+            let mut scorer = weight.scorer(reader, 1.0).unwrap();
             assert_eq!(scorer.doc(), 0u32);
             assert_eq!(scorer.advance(), TERMINATED);
         }
@@ -122,14 +122,14 @@ mod tests {
         let weight = AllQuery.weight(&searcher, false).unwrap();
         let reader = searcher.segment_reader(0);
         {
-            let mut scorer = weight.scorer(reader, 2.0f32).unwrap();
+            let mut scorer = weight.scorer(reader, 2.0).unwrap();
             assert_eq!(scorer.doc(), 0u32);
-            assert_eq!(scorer.score(), 2.0f32);
+            assert_eq!(scorer.score(), 2.0);
         }
         {
-            let mut scorer = weight.scorer(reader, 1.5f32).unwrap();
+            let mut scorer = weight.scorer(reader, 1.5).unwrap();
             assert_eq!(scorer.doc(), 0u32);
-            assert_eq!(scorer.score(), 1.5f32);
+            assert_eq!(scorer.score(), 1.5);
         }
     }
 }

src/query/automaton_weight.rs

@@ -5,9 +5,8 @@ use crate::query::{BitSetDocSet, Explanation};
 use crate::query::{Scorer, Weight};
 use crate::schema::{Field, IndexRecordOption};
 use crate::termdict::{TermDictionary, TermStreamer};
-use crate::DocId;
-use crate::Result;
 use crate::TantivyError;
+use crate::{DocId, Score};
 use std::sync::Arc;
 use tantivy_fst::Automaton;
 
@@ -40,7 +39,7 @@ impl<A> Weight for AutomatonWeight<A>
 where
     A: Automaton + Send + Sync + 'static,
 {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         let max_doc = reader.max_doc();
         let mut doc_bitset = BitSet::with_max_value(max_doc);
         let inverted_index = reader.inverted_index(self.field);
@@ -66,10 +65,10 @@ where
         Ok(Box::new(const_scorer))
     }
 
-    fn explain(&self, reader: &SegmentReader, doc: DocId) -> Result<Explanation> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
+        let mut scorer = self.scorer(reader, 1.0)?;
         if scorer.seek(doc) == doc {
-            Ok(Explanation::new("AutomatonScorer", 1.0f32))
+            Ok(Explanation::new("AutomatonScorer", 1.0))
         } else {
             Err(TantivyError::InvalidArgument(
                 "Document does not exist".to_string(),
@@ -91,7 +90,7 @@ mod tests {
         let mut schema = Schema::builder();
         let title = schema.add_text_field("title", STRING);
         let index = Index::create_in_ram(schema.build());
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(title=>"abc"));
         index_writer.add_document(doc!(title=>"bcd"));
         index_writer.add_document(doc!(title=>"abcd"));
@@ -144,13 +143,13 @@ mod tests {
         let reader = index.reader().unwrap();
         let searcher = reader.searcher();
         let mut scorer = automaton_weight
-            .scorer(searcher.segment_reader(0u32), 1.0f32)
+            .scorer(searcher.segment_reader(0u32), 1.0)
             .unwrap();
         assert_eq!(scorer.doc(), 0u32);
-        assert_eq!(scorer.score(), 1.0f32);
+        assert_eq!(scorer.score(), 1.0);
         assert_eq!(scorer.advance(), 2u32);
         assert_eq!(scorer.doc(), 2u32);
-        assert_eq!(scorer.score(), 1.0f32);
+        assert_eq!(scorer.score(), 1.0);
         assert_eq!(scorer.advance(), TERMINATED);
     }
 
@@ -162,9 +161,9 @@ mod tests {
         let reader = index.reader().unwrap();
         let searcher = reader.searcher();
         let mut scorer = automaton_weight
-            .scorer(searcher.segment_reader(0u32), 1.32f32)
+            .scorer(searcher.segment_reader(0u32), 1.32)
             .unwrap();
         assert_eq!(scorer.doc(), 0u32);
-        assert_eq!(scorer.score(), 1.32f32);
+        assert_eq!(scorer.score(), 1.32);
     }
 }

src/query/bitset/mod.rs

@@ -61,21 +61,23 @@ impl DocSet for BitSetDocSet {
     }
 
     fn seek(&mut self, target: DocId) -> DocId {
+        if target >= self.docs.max_value() {
+            self.doc = TERMINATED;
+            return TERMINATED;
+        }
         let target_bucket = target / 64u32;
-
-        // Mask for all of the bits greater or equal
-        // to our target document.
         if target_bucket > self.cursor_bucket {
             self.go_to_bucket(target_bucket);
             let greater_filter: TinySet = TinySet::range_greater_or_equal(target);
             self.cursor_tinybitset = self.cursor_tinybitset.intersect(greater_filter);
-            self.advance();
+            self.advance()
+        } else {
+            let mut doc = self.doc();
+            while doc < target {
+                doc = self.advance();
+            }
+            doc
         }
-        let mut doc = self.doc();
-        while doc < target {
-            doc = self.advance();
-        }
-        doc
     }
 
     /// Returns the current document
@@ -114,6 +116,13 @@ mod tests {
         assert_eq!(empty.advance(), TERMINATED)
     }
 
+    #[test]
+    fn test_seek_terminated() {
+        let bitset = BitSet::with_max_value(1000);
+        let mut empty = BitSetDocSet::from(bitset);
+        assert_eq!(empty.seek(TERMINATED), TERMINATED)
+    }
+
     fn test_go_through_sequential(docs: &[DocId]) {
         let mut docset = create_docbitset(docs, 1_000u32);
         for &doc in docs {

src/query/bm25.rs

@@ -6,21 +6,21 @@ use crate::Term;
 use serde::Deserialize;
 use serde::Serialize;
 
-const K1: f32 = 1.2;
-const B: f32 = 0.75;
+const K1: Score = 1.2;
+const B: Score = 0.75;
 
-fn idf(doc_freq: u64, doc_count: u64) -> f32 {
+fn idf(doc_freq: u64, doc_count: u64) -> Score {
     assert!(doc_count >= doc_freq, "{} >= {}", doc_count, doc_freq);
-    let x = ((doc_count - doc_freq) as f32 + 0.5) / (doc_freq as f32 + 0.5);
-    (1f32 + x).ln()
+    let x = ((doc_count - doc_freq) as Score + 0.5) / (doc_freq as Score + 0.5);
+    (1.0 + x).ln()
 }
 
-fn cached_tf_component(fieldnorm: u32, average_fieldnorm: f32) -> f32 {
-    K1 * (1f32 - B + B * fieldnorm as f32 / average_fieldnorm)
+fn cached_tf_component(fieldnorm: u32, average_fieldnorm: Score) -> Score {
+    K1 * (1.0 - B + B * fieldnorm as Score / average_fieldnorm)
 }
 
-fn compute_tf_cache(average_fieldnorm: f32) -> [f32; 256] {
-    let mut cache = [0f32; 256];
+fn compute_tf_cache(average_fieldnorm: Score) -> [Score; 256] {
+    let mut cache: [Score; 256] = [0.0; 256];
     for (fieldnorm_id, cache_mut) in cache.iter_mut().enumerate() {
         let fieldnorm = FieldNormReader::id_to_fieldnorm(fieldnorm_id as u8);
         *cache_mut = cached_tf_component(fieldnorm, average_fieldnorm);
@@ -30,19 +30,20 @@ fn compute_tf_cache(average_fieldnorm: f32) -> [f32; 256] {
 
 #[derive(Clone, PartialEq, Debug, Serialize, Deserialize)]
 pub struct BM25Params {
-    pub idf: f32,
-    pub avg_fieldnorm: f32,
+    pub idf: Score,
+    pub avg_fieldnorm: Score,
 }
 
+#[derive(Clone)]
 pub struct BM25Weight {
     idf_explain: Explanation,
-    weight: f32,
-    cache: [f32; 256],
-    average_fieldnorm: f32,
+    weight: Score,
+    cache: [Score; 256],
+    average_fieldnorm: Score,
 }
 
 impl BM25Weight {
-    pub fn boost_by(&self, boost: f32) -> BM25Weight {
+    pub fn boost_by(&self, boost: Score) -> BM25Weight {
         BM25Weight {
             idf_explain: self.idf_explain.clone(),
             weight: self.weight * boost,
@@ -69,7 +70,7 @@ impl BM25Weight {
             total_num_tokens += inverted_index.total_num_tokens();
             total_num_docs += u64::from(segment_reader.max_doc());
         }
-        let average_fieldnorm = total_num_tokens as f32 / total_num_docs as f32;
+        let average_fieldnorm = total_num_tokens as Score / total_num_docs as Score;
 
         if terms.len() == 1 {
             let term_doc_freq = searcher.doc_freq(&terms[0]);
@@ -81,26 +82,30 @@ impl BM25Weight {
                     let term_doc_freq = searcher.doc_freq(term);
                     idf(term_doc_freq, total_num_docs)
                 })
-                .sum::<f32>();
+                .sum::<Score>();
             let idf_explain = Explanation::new("idf", idf);
             BM25Weight::new(idf_explain, average_fieldnorm)
         }
     }
 
-    pub fn for_one_term(term_doc_freq: u64, total_num_docs: u64, avg_fieldnorm: f32) -> BM25Weight {
+    pub fn for_one_term(
+        term_doc_freq: u64,
+        total_num_docs: u64,
+        avg_fieldnorm: Score,
+    ) -> BM25Weight {
         let idf = idf(term_doc_freq, total_num_docs);
         let mut idf_explain =
             Explanation::new("idf, computed as log(1 + (N - n + 0.5) / (n + 0.5))", idf);
         idf_explain.add_const(
             "n, number of docs containing this term",
-            term_doc_freq as f32,
+            term_doc_freq as Score,
         );
-        idf_explain.add_const("N, total number of docs", total_num_docs as f32);
+        idf_explain.add_const("N, total number of docs", total_num_docs as Score);
         BM25Weight::new(idf_explain, avg_fieldnorm)
     }
 
-    fn new(idf_explain: Explanation, average_fieldnorm: f32) -> BM25Weight {
-        let weight = idf_explain.value() * (1f32 + K1);
+    fn new(idf_explain: Explanation, average_fieldnorm: Score) -> BM25Weight {
+        let weight = idf_explain.value() * (1.0 + K1);
         BM25Weight {
             idf_explain,
             weight,
@@ -119,8 +124,8 @@ impl BM25Weight {
     }
 
     #[inline(always)]
-    pub(crate) fn tf_factor(&self, fieldnorm_id: u8, term_freq: u32) -> f32 {
-        let term_freq = term_freq as f32;
+    pub(crate) fn tf_factor(&self, fieldnorm_id: u8, term_freq: u32) -> Score {
+        let term_freq = term_freq as Score;
         let norm = self.cache[fieldnorm_id as usize];
         term_freq / (term_freq + norm)
     }
@@ -131,7 +136,7 @@ impl BM25Weight {
         let score = self.score(fieldnorm_id, term_freq);
 
         let norm = self.cache[fieldnorm_id as usize];
-        let term_freq = term_freq as f32;
+        let term_freq = term_freq as Score;
         let right_factor = term_freq / (term_freq + norm);
 
         let mut tf_explanation = Explanation::new(
@@ -144,12 +149,12 @@ impl BM25Weight {
         tf_explanation.add_const("b, length normalization parameter", B);
         tf_explanation.add_const(
             "dl, length of field",
-            FieldNormReader::id_to_fieldnorm(fieldnorm_id) as f32,
+            FieldNormReader::id_to_fieldnorm(fieldnorm_id) as Score,
         );
         tf_explanation.add_const("avgdl, average length of field", self.average_fieldnorm);
 
         let mut explanation = Explanation::new("TermQuery, product of...", score);
-        explanation.add_detail(Explanation::new("(K1+1)", K1 + 1f32));
+        explanation.add_detail(Explanation::new("(K1+1)", K1 + 1.0));
         explanation.add_detail(self.idf_explain.clone());
         explanation.add_detail(tf_explanation);
         explanation
@@ -160,10 +165,11 @@ impl BM25Weight {
 mod tests {
 
     use super::idf;
-    use crate::assert_nearly_equals;
+    use crate::{assert_nearly_equals, Score};
 
     #[test]
     fn test_idf() {
-        assert_nearly_equals!(idf(1, 2), std::f32::consts::LN_2);
+        let score: Score = 2.0;
+        assert_nearly_equals!(idf(1, 2), score.ln());
     }
 }

src/query/boolean_query/block_wand.rs

@@ -1,8 +1,8 @@
 use crate::query::term_query::TermScorer;
 use crate::query::Scorer;
 use crate::{DocId, DocSet, Score, TERMINATED};
-use std::ops::DerefMut;
 use std::ops::Deref;
+use std::ops::DerefMut;
 
 /// Takes a term_scorers sorted by their current doc() and a threshold and returns
 /// Returns (pivot_len, pivot_ord) defined as follows:
@@ -13,8 +13,11 @@ use std::ops::Deref;
 /// We always have `before_pivot_len` < `pivot_len`.
 ///
 /// None is returned if we establish that no document can exceed the threshold.
-fn find_pivot_doc(term_scorers: &[TermScorerWithMaxScore], threshold: f32) -> Option<(usize, usize, DocId)> {
-    let mut max_score = 0.0f32;
+fn find_pivot_doc(
+    term_scorers: &[TermScorerWithMaxScore],
+    threshold: Score,
+) -> Option<(usize, usize, DocId)> {
+    let mut max_score = 0.0;
     let mut before_pivot_len = 0;
     let mut pivot_doc = TERMINATED;
     while before_pivot_len < term_scorers.len() {
@@ -32,44 +35,19 @@ fn find_pivot_doc(term_scorers: &[TermScorerWithMaxScore], threshold: f32) -> Op
     // Right now i is an ordinal, we want a len.
     let mut pivot_len = before_pivot_len + 1;
     // Some other term_scorer may be positioned on the same document.
-    pivot_len += term_scorers[pivot_len..].iter()
+    pivot_len += term_scorers[pivot_len..]
+        .iter()
         .take_while(|term_scorer| term_scorer.doc() == pivot_doc)
         .count();
     Some((before_pivot_len, pivot_len, pivot_doc))
 }
 
-struct TermScorerWithMaxScore<'a> {
-    scorer: &'a mut TermScorer,
-    max_score: f32,
-}
-
-impl<'a> From<&'a mut TermScorer> for TermScorerWithMaxScore<'a> {
-    fn from(scorer: &'a mut TermScorer) -> Self {
-        let max_score = scorer.max_score();
-        TermScorerWithMaxScore {
-            scorer,
-            max_score
-        }
-    }
-}
-
-impl<'a> Deref for TermScorerWithMaxScore<'a> {
-    type Target = TermScorer;
-
-    fn deref(&self) -> &Self::Target {
-        self.scorer
-    }
-}
-
-
-impl<'a> DerefMut for TermScorerWithMaxScore<'a> {
-    fn deref_mut(&mut self) -> &mut Self::Target {
-        self.scorer
-    }
-}
-
 // Before and after calling this method, scorers need to be sorted by their `.doc()`.
-fn block_max_was_too_low_advance_one_scorer(scorers: &mut Vec<TermScorerWithMaxScore>, pivot_len: usize) {
+fn block_max_was_too_low_advance_one_scorer(
+    scorers: &mut Vec<TermScorerWithMaxScore>,
+    pivot_len: usize,
+) {
+    debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
     let mut scorer_to_seek = pivot_len - 1;
     let mut doc_to_seek_after = scorers[scorer_to_seek].doc();
     for scorer_ord in (0..pivot_len - 1).rev() {
@@ -86,9 +64,9 @@ fn block_max_was_too_low_advance_one_scorer(scorers: &mut Vec<TermScorerWithMaxS
     }
     scorers[scorer_to_seek].seek(doc_to_seek_after + 1);
     restore_ordering(scorers, scorer_to_seek);
+    debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
 }
 
-
 // Given a list of term_scorers and a `ord` and assuming that `term_scorers[ord]` is sorted
 // except term_scorers[ord] that might be in advance compared to its ranks,
 // bubble up term_scorers[ord] in order to restore the ordering.
@@ -100,15 +78,19 @@ fn restore_ordering(term_scorers: &mut Vec<TermScorerWithMaxScore>, ord: usize)
         }
         term_scorers.swap(i, i - 1);
     }
+    debug_assert!(is_sorted(term_scorers.iter().map(|scorer| scorer.doc())));
 }
 
-
 // Attempts to advance all term_scorers between `&term_scorers[0..before_len]` to the pivot.
 // If this works, return true.
 // If this fails (ie: one of the term_scorer does not contain `pivot_doc` and seek goes past the
 // pivot), reorder the term_scorers to ensure the list is still sorted and returns `false`.
 // If a term_scorer reach TERMINATED in the process return false remove the term_scorer and return.
-fn align_scorers(term_scorers: &mut Vec<TermScorerWithMaxScore>, pivot_doc: DocId, before_pivot_len: usize) -> bool {
+fn align_scorers(
+    term_scorers: &mut Vec<TermScorerWithMaxScore>,
+    pivot_doc: DocId,
+    before_pivot_len: usize,
+) -> bool {
     debug_assert_ne!(pivot_doc, TERMINATED);
     for i in (0..before_pivot_len).rev() {
         let new_doc = term_scorers[i].seek(pivot_doc);
@@ -126,16 +108,20 @@ fn align_scorers(term_scorers: &mut Vec<TermScorerWithMaxScore>, pivot_doc: DocI
             return false;
         }
     }
-    return true;
+    true
 }
 
 // Assumes terms_scorers[..pivot_len] are positioned on the same doc (pivot_doc).
 // Advance term_scorers[..pivot_len] and out of these removes the terminated scores.
 // Restores the ordering of term_scorers.
 fn advance_all_scorers_on_pivot(term_scorers: &mut Vec<TermScorerWithMaxScore>, pivot_len: usize) {
+    for term_scorer in &mut term_scorers[..pivot_len] {
+        term_scorer.advance();
+    }
+    // TODO use drain_filter when available.
     let mut i = 0;
-    for _ in 0..pivot_len {
-        if term_scorers[i].advance() == TERMINATED {
+    while i != term_scorers.len() {
+        if term_scorers[i].doc() == TERMINATED {
             term_scorers.swap_remove(i);
         } else {
             i += 1;
@@ -146,61 +132,303 @@ fn advance_all_scorers_on_pivot(term_scorers: &mut Vec<TermScorerWithMaxScore>,
 
 pub fn block_wand(
     mut scorers: Vec<TermScorer>,
-    mut threshold: f32,
+    mut threshold: Score,
     callback: &mut dyn FnMut(u32, Score) -> Score,
 ) {
-    let mut scorers: Vec<TermScorerWithMaxScore> = scorers.iter_mut().map(TermScorerWithMaxScore::from).collect();
+    let mut scorers: Vec<TermScorerWithMaxScore> = scorers
+        .iter_mut()
+        .map(TermScorerWithMaxScore::from)
+        .collect();
     scorers.sort_by_key(|scorer| scorer.doc());
-    loop {
-        // At this point we need to ensure that the scorers are sorted!
-        if let Some((before_pivot_len, pivot_len, pivot_doc)) = find_pivot_doc(&scorers[..], threshold) {
-            debug_assert_ne!(pivot_doc, TERMINATED);
-            debug_assert!(before_pivot_len < pivot_len);
-            let block_max_score_upperbound: Score =  scorers[..pivot_len].iter_mut()
-                .map(|scorer| {
-                    scorer.shallow_seek(pivot_doc);
-                    scorer.block_max_score()
-                })
-                .sum();
+    // At this point we need to ensure that the scorers are sorted!
+    debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
+    while let Some((before_pivot_len, pivot_len, pivot_doc)) =
+        find_pivot_doc(&scorers[..], threshold)
+    {
+        debug_assert!(is_sorted(scorers.iter().map(|scorer| scorer.doc())));
+        debug_assert_ne!(pivot_doc, TERMINATED);
+        debug_assert!(before_pivot_len < pivot_len);
 
-            // Beware after shallow advance, skip readers can be in advance compared to
-            // the segment posting lists.
+        let block_max_score_upperbound: Score = scorers[..pivot_len]
+            .iter_mut()
+            .map(|scorer| {
+                scorer.shallow_seek(pivot_doc);
+                scorer.block_max_score()
+            })
+            .sum();
+
+        // Beware after shallow advance, skip readers can be in advance compared to
+        // the segment posting lists.
+        //
+        // `block_segment_postings.load_block()` need to be called separately.
+        if block_max_score_upperbound <= threshold {
+            // Block max condition was not reached
+            // We could get away by simply advancing the scorers to DocId + 1 but it would
+            // be inefficient. The optimization requires proper explanation and was
+            // isolated in a different function.
+            block_max_was_too_low_advance_one_scorer(&mut scorers, pivot_len);
+            continue;
+        }
+
+        // Block max condition is observed.
+        //
+        // Let's try and advance all scorers before the pivot to the pivot.
+        if !align_scorers(&mut scorers, pivot_doc, before_pivot_len) {
+            // At least of the scorer does not contain the pivot.
             //
-            // `block_segment_postings.load_block()` need to be called separately.
-            if block_max_score_upperbound <= threshold {
-                // Block max condition was not reached.
-                // We could get away by simply advancing the scorers to DocId + 1 but it would
-                // be inefficient. The optimization requires proper explanation and was
-                // isolated in a different function.
-                block_max_was_too_low_advance_one_scorer(&mut scorers, pivot_len);
-                continue;
-            }
+            // Let's stop scoring this pivot and go through the pivot selection again.
+            // Note that the current pivot is not necessarily a bad candidate and it
+            // may be picked again.
+            continue;
+        }
 
-            // Block max condition is observed.
-            //
-            // Let's try and advance all scorers before the pivot to the pivot.
-            if !align_scorers(&mut scorers, pivot_doc, before_pivot_len) {
-                // At least of the scorer does not contain the pivot.
-                //
-                // Let's stop scoring this pivot and go through the pivot selection again.
-                // Note that the current pivot is not necessarily a bad candidate and it
-                // may be picked again.
-                continue;
-            }
+        // At this point, all scorers are positioned on the doc.
+        let score = scorers[..pivot_len]
+            .iter_mut()
+            .map(|scorer| scorer.score())
+            .sum();
+        if score > threshold {
+            threshold = callback(pivot_doc, score);
+        }
+        // let's advance all of the scorers that are currently positioned on the pivot.
+        advance_all_scorers_on_pivot(&mut scorers, pivot_len);
+    }
+}
 
-            // At this point, all scorers are positioned on the doc.
-            let score = scorers[..pivot_len]
-                .iter_mut()
-                .map(|scorer| scorer.score())
-                .sum();
-            if score > threshold {
-                threshold = callback(pivot_doc, score);
-            }
-            // let's advance all of the scorers that are currently positioned on the pivot.
-            advance_all_scorers_on_pivot(&mut scorers, pivot_len);
+struct TermScorerWithMaxScore<'a> {
+    scorer: &'a mut TermScorer,
+    max_score: Score,
+}
 
-        } else {
-            return;
+impl<'a> From<&'a mut TermScorer> for TermScorerWithMaxScore<'a> {
+    fn from(scorer: &'a mut TermScorer) -> Self {
+        let max_score = scorer.max_score();
+        TermScorerWithMaxScore { scorer, max_score }
+    }
+}
+
+impl<'a> Deref for TermScorerWithMaxScore<'a> {
+    type Target = TermScorer;
+
+    fn deref(&self) -> &Self::Target {
+        self.scorer
+    }
+}
+
+impl<'a> DerefMut for TermScorerWithMaxScore<'a> {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        self.scorer
+    }
+}
+
+fn is_sorted<I: Iterator<Item = DocId>>(mut it: I) -> bool {
+    if let Some(first) = it.next() {
+        let mut prev = first;
+        for doc in it {
+            if doc < prev {
+                return false;
+            }
+            prev = doc;
+        }
+    }
+    true
+}
+#[cfg(test)]
+mod tests {
+    use crate::query::score_combiner::SumCombiner;
+    use crate::query::term_query::TermScorer;
+    use crate::query::Union;
+    use crate::query::{BM25Weight, Scorer};
+    use crate::{DocId, DocSet, Score, TERMINATED};
+    use proptest::prelude::*;
+    use std::cmp::Ordering;
+    use std::collections::BinaryHeap;
+    use std::iter;
+
+    struct Float(Score);
+
+    impl Eq for Float {}
+
+    impl PartialEq for Float {
+        fn eq(&self, other: &Self) -> bool {
+            self.cmp(&other) == Ordering::Equal
+        }
+    }
+
+    impl PartialOrd for Float {
+        fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+            Some(self.cmp(other))
+        }
+    }
+
+    impl Ord for Float {
+        fn cmp(&self, other: &Self) -> Ordering {
+            other.0.partial_cmp(&self.0).unwrap_or(Ordering::Equal)
+        }
+    }
+
+    fn nearly_equals(left: Score, right: Score) -> bool {
+        (left - right).abs() < 0.000001 * (left + right).abs()
+    }
+
+    fn compute_checkpoints_for_each_pruning(
+        term_scorers: Vec<TermScorer>,
+        n: usize,
+    ) -> Vec<(DocId, Score)> {
+        let mut heap: BinaryHeap<Float> = BinaryHeap::with_capacity(n);
+        let mut checkpoints: Vec<(DocId, Score)> = Vec::new();
+        let mut limit: Score = 0.0;
+        super::block_wand(term_scorers, Score::MIN, &mut |doc, score| {
+            heap.push(Float(score));
+            if heap.len() > n {
+                heap.pop().unwrap();
+            }
+            if heap.len() == n {
+                limit = heap.peek().unwrap().0;
+            }
+            if !nearly_equals(score, limit) {
+                checkpoints.push((doc, score));
+            }
+            return limit;
+        });
+        checkpoints
+    }
+
+    fn compute_checkpoints_manual(term_scorers: Vec<TermScorer>, n: usize) -> Vec<(DocId, Score)> {
+        let mut heap: BinaryHeap<Float> = BinaryHeap::with_capacity(n);
+        let mut checkpoints: Vec<(DocId, Score)> = Vec::new();
+        let mut scorer: Union<TermScorer, SumCombiner> = Union::from(term_scorers);
+
+        let mut limit = Score::MIN;
+        loop {
+            if scorer.doc() == TERMINATED {
+                break;
+            }
+            let doc = scorer.doc();
+            let score = scorer.score();
+            if score > limit {
+                heap.push(Float(score));
+                if heap.len() > n {
+                    heap.pop().unwrap();
+                }
+                if heap.len() == n {
+                    limit = heap.peek().unwrap().0;
+                }
+                if !nearly_equals(score, limit) {
+                    checkpoints.push((doc, score));
+                }
+            }
+            scorer.advance();
+        }
+        checkpoints
+    }
+
+    const MAX_TERM_FREQ: u32 = 100u32;
+
+    fn posting_list(max_doc: u32) -> BoxedStrategy<Vec<(DocId, u32)>> {
+        (1..max_doc + 1)
+            .prop_flat_map(move |doc_freq| {
+                (
+                    proptest::bits::bitset::sampled(doc_freq as usize, 0..max_doc as usize),
+                    proptest::collection::vec(1u32..MAX_TERM_FREQ, doc_freq as usize),
+                )
+            })
+            .prop_map(|(docset, term_freqs)| {
+                docset
+                    .iter()
+                    .map(|doc| doc as u32)
+                    .zip(term_freqs.iter().cloned())
+                    .collect::<Vec<_>>()
+            })
+            .boxed()
+    }
+
+    fn gen_term_scorers(num_scorers: usize) -> BoxedStrategy<(Vec<Vec<(DocId, u32)>>, Vec<u32>)> {
+        (1u32..100u32)
+            .prop_flat_map(move |max_doc: u32| {
+                (
+                    proptest::collection::vec(posting_list(max_doc), num_scorers),
+                    proptest::collection::vec(2u32..10u32 * MAX_TERM_FREQ, max_doc as usize),
+                )
+            })
+            .boxed()
+    }
+
+    fn test_block_wand_aux(posting_lists: &[Vec<(DocId, u32)>], fieldnorms: &[u32]) {
+        // We virtually repeat all docs 64 times in order to emulate blocks of 2 documents
+        // and surface blogs more easily.
+        const REPEAT: usize = 64;
+        let fieldnorms_expanded = fieldnorms
+            .iter()
+            .cloned()
+            .flat_map(|fieldnorm| iter::repeat(fieldnorm).take(REPEAT))
+            .collect::<Vec<u32>>();
+
+        let postings_lists_expanded: Vec<Vec<(DocId, u32)>> = posting_lists
+            .iter()
+            .map(|posting_list| {
+                posting_list
+                    .into_iter()
+                    .cloned()
+                    .flat_map(|(doc, term_freq)| {
+                        (0 as u32..REPEAT as u32).map(move |offset| {
+                            (
+                                doc * (REPEAT as u32) + offset,
+                                if offset == 0 { term_freq } else { 1 },
+                            )
+                        })
+                    })
+                    .collect::<Vec<(DocId, u32)>>()
+            })
+            .collect::<Vec<_>>();
+
+        let total_fieldnorms: u64 = fieldnorms_expanded
+            .iter()
+            .cloned()
+            .map(|fieldnorm| fieldnorm as u64)
+            .sum();
+        let average_fieldnorm = (total_fieldnorms as Score) / (fieldnorms_expanded.len() as Score);
+        let max_doc = fieldnorms_expanded.len();
+
+        let term_scorers: Vec<TermScorer> = postings_lists_expanded
+            .iter()
+            .map(|postings| {
+                let bm25_weight = BM25Weight::for_one_term(
+                    postings.len() as u64,
+                    max_doc as u64,
+                    average_fieldnorm,
+                );
+                TermScorer::create_for_test(postings, &fieldnorms_expanded[..], bm25_weight)
+            })
+            .collect();
+        for top_k in 1..4 {
+            let checkpoints_for_each_pruning =
+                compute_checkpoints_for_each_pruning(term_scorers.clone(), top_k);
+            let checkpoints_manual = compute_checkpoints_manual(term_scorers.clone(), top_k);
+            assert_eq!(checkpoints_for_each_pruning.len(), checkpoints_manual.len());
+            for (&(left_doc, left_score), &(right_doc, right_score)) in checkpoints_for_each_pruning
+                .iter()
+                .zip(checkpoints_manual.iter())
+            {
+                assert_eq!(left_doc, right_doc);
+                assert!(nearly_equals(left_score, right_score));
+            }
+        }
+    }
+
+    proptest! {
+        #![proptest_config(ProptestConfig::with_cases(500))]
+        #[test]
+        fn test_block_wand_two_term_scorers((posting_lists, fieldnorms) in gen_term_scorers(2)) {
+            test_block_wand_aux(&posting_lists[..], &fieldnorms[..]);
+        }
+    }
+
+    proptest! {
+        #![proptest_config(ProptestConfig::with_cases(500))]
+        #[test]
+        fn test_block_wand_three_term_scorers((posting_lists, fieldnorms) in gen_term_scorers(3)) {
+            test_block_wand_aux(&posting_lists[..], &fieldnorms[..]);
         }
     }
 }

src/query/boolean_query/boolean_weight.rs

@@ -78,7 +78,7 @@ impl BooleanWeight {
     fn per_occur_scorers(
         &self,
         reader: &SegmentReader,
-        boost: f32,
+        boost: Score,
     ) -> crate::Result<HashMap<Occur, Vec<Box<dyn Scorer>>>> {
         let mut per_occur_scorers: HashMap<Occur, Vec<Box<dyn Scorer>>> = HashMap::new();
         for &(ref occur, ref subweight) in &self.weights {
@@ -94,7 +94,7 @@ impl BooleanWeight {
     fn complex_scorer<TScoreCombiner: ScoreCombiner>(
         &self,
         reader: &SegmentReader,
-        boost: f32,
+        boost: Score,
     ) -> crate::Result<SpecializedScorer> {
         let mut per_occur_scorers = self.per_occur_scorers(reader, boost)?;
 
@@ -147,7 +147,7 @@ impl BooleanWeight {
 }
 
 impl Weight for BooleanWeight {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> crate::Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         if self.weights.is_empty() {
             Ok(Box::new(EmptyScorer))
         } else if self.weights.len() == 1 {
@@ -169,12 +169,12 @@ impl Weight for BooleanWeight {
     }
 
     fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+        let mut scorer = self.scorer(reader, 1.0)?;
         if scorer.seek(doc) != doc {
             return Err(does_not_match(doc));
         }
         if !self.scoring_enabled {
-            return Ok(Explanation::new("BooleanQuery with no scoring", 1f32));
+            return Ok(Explanation::new("BooleanQuery with no scoring", 1.0));
         }
 
         let mut explanation = Explanation::new("BooleanClause. Sum of ...", scorer.score());
@@ -193,7 +193,7 @@ impl Weight for BooleanWeight {
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score),
     ) -> crate::Result<()> {
-        let scorer = self.complex_scorer::<SumWithCoordsCombiner>(reader, 1.0f32)?;
+        let scorer = self.complex_scorer::<SumWithCoordsCombiner>(reader, 1.0)?;
         match scorer {
             SpecializedScorer::TermUnion(term_scorers) => {
                 let mut union_scorer =
@@ -219,11 +219,11 @@ impl Weight for BooleanWeight {
     /// important optimization (e.g. BlockWAND for union).
     fn for_each_pruning(
         &self,
-        threshold: f32,
+        threshold: Score,
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score) -> Score,
     ) -> crate::Result<()> {
-        let scorer = self.complex_scorer::<SumWithCoordsCombiner>(reader, 1.0f32)?;
+        let scorer = self.complex_scorer::<SumWithCoordsCombiner>(reader, 1.0)?;
         match scorer {
             SpecializedScorer::TermUnion(term_scorers) => {
                 super::block_wand(term_scorers, threshold, callback);

src/query/boolean_query/mod.rs

@@ -32,7 +32,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             {
                 index_writer.add_document(doc!(text_field => "a b c"));
                 index_writer.add_document(doc!(text_field => "a c"));
@@ -61,9 +61,7 @@ mod tests {
         let query = query_parser.parse_query("+a").unwrap();
         let searcher = index.reader().unwrap().searcher();
         let weight = query.weight(&searcher, true).unwrap();
-        let scorer = weight
-            .scorer(searcher.segment_reader(0u32), 1.0f32)
-            .unwrap();
+        let scorer = weight.scorer(searcher.segment_reader(0u32), 1.0).unwrap();
         assert!(scorer.is::<TermScorer>());
     }
 
@@ -75,17 +73,13 @@ mod tests {
         {
             let query = query_parser.parse_query("+a +b +c").unwrap();
             let weight = query.weight(&searcher, true).unwrap();
-            let scorer = weight
-                .scorer(searcher.segment_reader(0u32), 1.0f32)
-                .unwrap();
+            let scorer = weight.scorer(searcher.segment_reader(0u32), 1.0).unwrap();
             assert!(scorer.is::<Intersection<TermScorer>>());
         }
         {
             let query = query_parser.parse_query("+a +(b c)").unwrap();
             let weight = query.weight(&searcher, true).unwrap();
-            let scorer = weight
-                .scorer(searcher.segment_reader(0u32), 1.0f32)
-                .unwrap();
+            let scorer = weight.scorer(searcher.segment_reader(0u32), 1.0).unwrap();
             assert!(scorer.is::<Intersection<Box<dyn Scorer>>>());
         }
     }
@@ -98,9 +92,7 @@ mod tests {
         {
             let query = query_parser.parse_query("+a b").unwrap();
             let weight = query.weight(&searcher, true).unwrap();
-            let scorer = weight
-                .scorer(searcher.segment_reader(0u32), 1.0f32)
-                .unwrap();
+            let scorer = weight.scorer(searcher.segment_reader(0u32), 1.0).unwrap();
             assert!(scorer.is::<RequiredOptionalScorer<
                 Box<dyn Scorer>,
                 Box<dyn Scorer>,
@@ -110,9 +102,7 @@ mod tests {
         {
             let query = query_parser.parse_query("+a b").unwrap();
             let weight = query.weight(&searcher, false).unwrap();
-            let scorer = weight
-                .scorer(searcher.segment_reader(0u32), 1.0f32)
-                .unwrap();
+            let scorer = weight.scorer(searcher.segment_reader(0u32), 1.0).unwrap();
             assert!(scorer.is::<TermScorer>());
         }
     }
@@ -234,7 +224,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text_field => "a b c"));
             index_writer.add_document(doc!(text_field => "a c"));
             index_writer.add_document(doc!(text_field => "b c"));
@@ -255,17 +245,17 @@ mod tests {
         let boolean_weight = boolean_query.weight(&searcher, true).unwrap();
         {
             let mut boolean_scorer = boolean_weight
-                .scorer(searcher.segment_reader(0u32), 1.0f32)
+                .scorer(searcher.segment_reader(0u32), 1.0)
                 .unwrap();
             assert_eq!(boolean_scorer.doc(), 0u32);
-            assert_nearly_equals!(boolean_scorer.score(), 0.84163445f32);
+            assert_nearly_equals!(boolean_scorer.score(), 0.84163445);
         }
         {
             let mut boolean_scorer = boolean_weight
-                .scorer(searcher.segment_reader(0u32), 2.0f32)
+                .scorer(searcher.segment_reader(0u32), 2.0)
                 .unwrap();
             assert_eq!(boolean_scorer.doc(), 0u32);
-            assert_nearly_equals!(boolean_scorer.score(), 1.6832689f32);
+            assert_nearly_equals!(boolean_scorer.score(), 1.6832689);
         }
     }
 
@@ -295,170 +285,9 @@ mod tests {
                 (Occur::Must, make_term_query("a")),
                 (Occur::Must, make_term_query("b")),
             ]);
-            assert_eq!(score_docs(&boolean_query), vec![0.977973, 0.84699446]);
+            let scores = score_docs(&boolean_query);
+            assert_nearly_equals!(scores[0], 0.977973);
+            assert_nearly_equals!(scores[1], 0.84699446);
         }
     }
-
-    // motivated by #554
-    #[test]
-    fn test_bm25_several_fields() {
-        let mut schema_builder = Schema::builder();
-        let title = schema_builder.add_text_field("title", TEXT);
-        let text = schema_builder.add_text_field("text", TEXT);
-        let schema = schema_builder.build();
-        let index = Index::create_in_ram(schema);
-
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
-        index_writer.add_document(doc!(
-            // tf = 1 0
-            title =>  "Законы притяжения   Оксана Кулакова",
-            // tf = 1 0
-            text => "Законы притяжения Оксана Кулакова]  \n\nТема: Сексуальное искусство, Женственность\nТип товара: Запись вебинара (аудио)\nПродолжительность: 1,5 часа\n\nСсылка на вебинар:\n  ",
-        ));
-        index_writer.add_document(doc!(
-            // tf = 1 0
-            title => "Любимые русские пироги (Оксана Путан)",
-            // tf = 2 0
-            text => "http://i95.fastpic.ru/big/2017/0628/9a/615b9c8504d94a3893d7f496ac53539a.jpg \n\nОт издателя\nОксана Путан   профессиональный повар, автор кулинарных книг и известный кулинарный блогер. Ее рецепты отличаются практичностью, доступностью и пользуются огромной популярностью в русскоязычном интернете. Это третья книга автора о самом вкусном и ароматном   настоящих русских пирогах и выпечке!\nДаже новички на кухне легко готовят по ее рецептам. Оксана описывает процесс приготовления настолько подробно и понятно, что вам остается только наслаждаться готовкой и не тратить время на лишние усилия. Готовьте легко и просто!\n\nhttps://www.ozon.ru/context/detail/id/139872462/"
-        ));
-        index_writer.add_document(doc!(
-            // tf = 1 1
-            title =>  "PDF Мастер Класс \"Морячок\" (Оксана Лифенко)",
-            // tf = 0 0
-            text => "https://i.ibb.co/pzvHrDN/I3d U T6 Gg TM.jpg\nhttps://i.ibb.co/NFrb6v6/N0ls Z9nwjb U.jpg\nВ описание входит штаны, кофта, берет, матросский воротник. Описание продается в формате PDF, состоит из 12 страниц формата А4 и может быть напечатано на любом принтере.\nОписание предназначено для кукол BJD RealPuki от FairyLand, но может подойти и другим подобным куклам. Также вы можете вязать этот наряд из обычной пряжи, и он подойдет для куколок побольше.\nhttps://vk.com/market 95724412?w=product 95724412_2212"
-        ));
-        for _ in 0..1_000 {
-            index_writer.add_document(doc!(
-                title =>  "a b d e f g",
-                text => "maitre corbeau sur un arbre perche tenait dans son bec un fromage Maitre rnard par lodeur alleche lui tint a peu pres ce langage."
-            ));
-        }
-        index_writer.commit().unwrap();
-        let reader = index.reader().unwrap();
-        let searcher = reader.searcher();
-        let query_parser = QueryParser::for_index(&index, vec![title, text]);
-        let query = query_parser.parse_query("Оксана Лифенко").unwrap();
-        let weight = query.weight(&searcher, true).unwrap();
-        let mut scorer = weight
-            .scorer(searcher.segment_reader(0u32), 1.0f32)
-            .unwrap();
-        scorer.advance();
-
-        let explanation = query.explain(&searcher, DocAddress(0u32, 0u32)).unwrap();
-        assert_eq!(
-            explanation.to_pretty_json(),
-            r#"{
-  "value": 12.997711,
-  "description": "BooleanClause. Sum of ...",
-  "details": [
-    {
-      "value": 12.997711,
-      "description": "BooleanClause. Sum of ...",
-      "details": [
-        {
-          "value": 6.551476,
-          "description": "TermQuery, product of...",
-          "details": [
-            {
-              "value": 2.2,
-              "description": "(K1+1)"
-            },
-            {
-              "value": 5.658984,
-              "description": "idf, computed as log(1 + (N - n + 0.5) / (n + 0.5))",
-              "details": [
-                {
-                  "value": 3.0,
-                  "description": "n, number of docs containing this term"
-                },
-                {
-                  "value": 1003.0,
-                  "description": "N, total number of docs"
-                }
-              ]
-            },
-            {
-              "value": 0.5262329,
-              "description": "freq / (freq + k1 * (1 - b + b * dl / avgdl))",
-              "details": [
-                {
-                  "value": 1.0,
-                  "description": "freq, occurrences of term within document"
-                },
-                {
-                  "value": 1.2,
-                  "description": "k1, term saturation parameter"
-                },
-                {
-                  "value": 0.75,
-                  "description": "b, length normalization parameter"
-                },
-                {
-                  "value": 4.0,
-                  "description": "dl, length of field"
-                },
-                {
-                  "value": 5.997009,
-                  "description": "avgdl, average length of field"
-                }
-              ]
-            }
-          ]
-        },
-        {
-          "value": 6.446235,
-          "description": "TermQuery, product of...",
-          "details": [
-            {
-              "value": 2.2,
-              "description": "(K1+1)"
-            },
-            {
-              "value": 5.9954567,
-              "description": "idf, computed as log(1 + (N - n + 0.5) / (n + 0.5))",
-              "details": [
-                {
-                  "value": 2.0,
-                  "description": "n, number of docs containing this term"
-                },
-                {
-                  "value": 1003.0,
-                  "description": "N, total number of docs"
-                }
-              ]
-            },
-            {
-              "value": 0.4887212,
-              "description": "freq / (freq + k1 * (1 - b + b * dl / avgdl))",
-              "details": [
-                {
-                  "value": 1.0,
-                  "description": "freq, occurrences of term within document"
-                },
-                {
-                  "value": 1.2,
-                  "description": "k1, term saturation parameter"
-                },
-                {
-                  "value": 0.75,
-                  "description": "b, length normalization parameter"
-                },
-                {
-                  "value": 20.0,
-                  "description": "dl, length of field"
-                },
-                {
-                  "value": 24.123629,
-                  "description": "avgdl, average length of field"
-                }
-              ]
-            }
-          ]
-        }
-      ]
-    }
-  ]
-}"#
-        );
-    }
 }

src/query/boost_query.rs

@@ -1,7 +1,7 @@
 use crate::fastfield::DeleteBitSet;
 use crate::query::explanation::does_not_match;
 use crate::query::{Explanation, Query, Scorer, Weight};
-use crate::{DocId, DocSet, Searcher, SegmentReader, Term};
+use crate::{DocId, DocSet, Score, Searcher, SegmentReader, Term};
 use std::collections::BTreeSet;
 use std::fmt;
 
@@ -12,12 +12,12 @@ use std::fmt;
 /// factor.
 pub struct BoostQuery {
     query: Box<dyn Query>,
-    boost: f32,
+    boost: Score,
 }
 
 impl BoostQuery {
     /// Builds a boost query.
-    pub fn new(query: Box<dyn Query>, boost: f32) -> BoostQuery {
+    pub fn new(query: Box<dyn Query>, boost: Score) -> BoostQuery {
         BoostQuery { query, boost }
     }
 }
@@ -55,22 +55,22 @@ impl Query for BoostQuery {
 
 pub(crate) struct BoostWeight {
     weight: Box<dyn Weight>,
-    boost: f32,
+    boost: Score,
 }
 
 impl BoostWeight {
-    pub fn new(weight: Box<dyn Weight>, boost: f32) -> Self {
+    pub fn new(weight: Box<dyn Weight>, boost: Score) -> Self {
         BoostWeight { weight, boost }
     }
 }
 
 impl Weight for BoostWeight {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> crate::Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         self.weight.scorer(reader, boost * self.boost)
     }
 
     fn explain(&self, reader: &SegmentReader, doc: u32) -> crate::Result<Explanation> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+        let mut scorer = self.scorer(reader, 1.0)?;
         if scorer.seek(doc) != doc {
             return Err(does_not_match(doc));
         }
@@ -88,11 +88,11 @@ impl Weight for BoostWeight {
 
 pub(crate) struct BoostScorer<S: Scorer> {
     underlying: S,
-    boost: f32,
+    boost: Score,
 }
 
 impl<S: Scorer> BoostScorer<S> {
-    pub fn new(underlying: S, boost: f32) -> BoostScorer<S> {
+    pub fn new(underlying: S, boost: Score) -> BoostScorer<S> {
         BoostScorer { underlying, boost }
     }
 }
@@ -128,7 +128,7 @@ impl<S: Scorer> DocSet for BoostScorer<S> {
 }
 
 impl<S: Scorer> Scorer for BoostScorer<S> {
-    fn score(&mut self) -> f32 {
+    fn score(&mut self) -> Score {
         self.underlying.score() * self.boost
     }
 }
@@ -144,7 +144,7 @@ mod tests {
     fn test_boost_query_explain() {
         let schema = Schema::builder().build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(Document::new());
         assert!(index_writer.commit().is_ok());
         let reader = index.reader().unwrap();

src/query/empty_query.rs

@@ -34,7 +34,7 @@ impl Query for EmptyQuery {
 /// It is useful for tests and handling edge cases.
 pub struct EmptyWeight;
 impl Weight for EmptyWeight {
-    fn scorer(&self, _reader: &SegmentReader, _boost: f32) -> crate::Result<Box<dyn Scorer>> {
+    fn scorer(&self, _reader: &SegmentReader, _boost: Score) -> crate::Result<Box<dyn Scorer>> {
         Ok(Box::new(EmptyScorer))
     }
 
@@ -64,7 +64,7 @@ impl DocSet for EmptyScorer {
 
 impl Scorer for EmptyScorer {
     fn score(&mut self) -> Score {
-        0f32
+        0.0
     }
 }
 

src/query/explanation.rs

@@ -1,4 +1,4 @@
-use crate::{DocId, TantivyError};
+use crate::{DocId, Score, TantivyError};
 use serde::Serialize;
 use std::fmt;
 
@@ -13,7 +13,7 @@ pub(crate) fn does_not_match(doc: DocId) -> TantivyError {
 /// representation of this tree when debugging a given score.
 #[derive(Clone, Serialize)]
 pub struct Explanation {
-    value: f32,
+    value: Score,
     description: String,
     #[serde(skip_serializing_if = "Vec::is_empty")]
     details: Vec<Explanation>,
@@ -27,7 +27,7 @@ impl fmt::Debug for Explanation {
 
 impl Explanation {
     /// Creates a new explanation object.
-    pub fn new<T: ToString>(description: T, value: f32) -> Explanation {
+    pub fn new<T: ToString>(description: T, value: Score) -> Explanation {
         Explanation {
             value,
             description: description.to_string(),
@@ -36,7 +36,7 @@ impl Explanation {
     }
 
     /// Returns the value associated to the current node.
-    pub fn value(&self) -> f32 {
+    pub fn value(&self) -> Score {
         self.value
     }
 
@@ -48,7 +48,7 @@ impl Explanation {
     }
 
     /// Shortcut for `self.details.push(Explanation::new(name, value));`
-    pub fn add_const<T: ToString>(&mut self, name: T, value: f32) {
+    pub fn add_const<T: ToString>(&mut self, name: T, value: Score) {
         self.details.push(Explanation::new(name, value));
     }
 

src/query/fuzzy_query.rs

@@ -177,7 +177,7 @@ mod test {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(
                 country_field => "japan",
             ));
@@ -199,7 +199,7 @@ mod test {
                 .unwrap();
             assert_eq!(top_docs.len(), 1, "Expected only 1 document");
             let (score, _) = top_docs[0];
-            assert_nearly_equals!(1f32, score);
+            assert_nearly_equals!(1.0, score);
         }
 
         // fails because non-prefix Levenshtein distance is more than 1 (add 'a' and 'n')
@@ -223,7 +223,7 @@ mod test {
                 .unwrap();
             assert_eq!(top_docs.len(), 1, "Expected only 1 document");
             let (score, _) = top_docs[0];
-            assert_nearly_equals!(1f32, score);
+            assert_nearly_equals!(1.0, score);
         }
     }
 }

src/query/phrase_query/mod.rs

@@ -24,7 +24,7 @@ pub mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             for &text in texts {
                 let doc = doc!(text_field=>text);
                 index_writer.add_document(doc);
@@ -79,7 +79,7 @@ pub mod tests {
             .collect();
         let phrase_query = PhraseQuery::new(terms);
         let phrase_weight = phrase_query.phrase_weight(&searcher, false)?;
-        let mut phrase_scorer = phrase_weight.scorer(searcher.segment_reader(0), 1.0f32)?;
+        let mut phrase_scorer = phrase_weight.scorer(searcher.segment_reader(0), 1.0)?;
         assert_eq!(phrase_scorer.doc(), 1);
         assert_eq!(phrase_scorer.advance(), TERMINATED);
         Ok(())
@@ -135,7 +135,7 @@ pub mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text_field=>"a b c"));
             assert!(index_writer.commit().is_ok());
         }
@@ -186,7 +186,7 @@ pub mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text_field=>"b"));
             index_writer.add_document(doc!(text_field=>"a b"));
             index_writer.add_document(doc!(text_field=>"b a"));
@@ -217,7 +217,7 @@ pub mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text_field=>"a b c d e f g h"));
             assert!(index_writer.commit().is_ok());
         }

src/query/phrase_query/phrase_scorer.rs

@@ -3,7 +3,7 @@ use crate::fieldnorm::FieldNormReader;
 use crate::postings::Postings;
 use crate::query::bm25::BM25Weight;
 use crate::query::{Intersection, Scorer};
-use crate::DocId;
+use crate::{DocId, Score};
 use std::cmp::Ordering;
 
 struct PostingsWithOffset<TPostings> {
@@ -257,7 +257,7 @@ impl<TPostings: Postings> DocSet for PhraseScorer<TPostings> {
 }
 
 impl<TPostings: Postings> Scorer for PhraseScorer<TPostings> {
-    fn score(&mut self) -> f32 {
+    fn score(&mut self) -> Score {
         let doc = self.doc();
         let fieldnorm_id = self.fieldnorm_reader.fieldnorm_id(doc);
         self.similarity_weight

src/query/phrase_query/phrase_weight.rs

@@ -9,7 +9,7 @@ use crate::query::Weight;
 use crate::query::{EmptyScorer, Explanation};
 use crate::schema::IndexRecordOption;
 use crate::schema::Term;
-use crate::Result;
+use crate::Score;
 use crate::{DocId, DocSet};
 
 pub struct PhraseWeight {
@@ -32,7 +32,7 @@ impl PhraseWeight {
         }
     }
 
-    fn fieldnorm_reader(&self, reader: &SegmentReader) -> FieldNormReader {
+    fn fieldnorm_reader(&self, reader: &SegmentReader) -> crate::Result<FieldNormReader> {
         let field = self.phrase_terms[0].1.field();
         reader.get_fieldnorms_reader(field)
     }
@@ -40,10 +40,10 @@ impl PhraseWeight {
     fn phrase_scorer(
         &self,
         reader: &SegmentReader,
-        boost: f32,
-    ) -> Result<Option<PhraseScorer<SegmentPostings>>> {
+        boost: Score,
+    ) -> crate::Result<Option<PhraseScorer<SegmentPostings>>> {
         let similarity_weight = self.similarity_weight.boost_by(boost);
-        let fieldnorm_reader = self.fieldnorm_reader(reader);
+        let fieldnorm_reader = self.fieldnorm_reader(reader)?;
         if reader.has_deletes() {
             let mut term_postings_list = Vec::new();
             for &(offset, ref term) in &self.phrase_terms {
@@ -85,7 +85,7 @@ impl PhraseWeight {
 }
 
 impl Weight for PhraseWeight {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         if let Some(scorer) = self.phrase_scorer(reader, boost)? {
             Ok(Box::new(scorer))
         } else {
@@ -93,8 +93,8 @@ impl Weight for PhraseWeight {
         }
     }
 
-    fn explain(&self, reader: &SegmentReader, doc: DocId) -> Result<Explanation> {
-        let scorer_opt = self.phrase_scorer(reader, 1.0f32)?;
+    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
+        let scorer_opt = self.phrase_scorer(reader, 1.0)?;
         if scorer_opt.is_none() {
             return Err(does_not_match(doc));
         }
@@ -102,7 +102,7 @@ impl Weight for PhraseWeight {
         if scorer.seek(doc) != doc {
             return Err(does_not_match(doc));
         }
-        let fieldnorm_reader = self.fieldnorm_reader(reader);
+        let fieldnorm_reader = self.fieldnorm_reader(reader)?;
         let fieldnorm_id = fieldnorm_reader.fieldnorm_id(doc);
         let phrase_count = scorer.phrase_count();
         let mut explanation = Explanation::new("Phrase Scorer", scorer.score());
@@ -130,7 +130,7 @@ mod tests {
         ]);
         let phrase_weight = phrase_query.phrase_weight(&searcher, true).unwrap();
         let mut phrase_scorer = phrase_weight
-            .phrase_scorer(searcher.segment_reader(0u32), 1.0f32)
+            .phrase_scorer(searcher.segment_reader(0u32), 1.0)
             .unwrap()
             .unwrap();
         assert_eq!(phrase_scorer.doc(), 1);

src/query/query.rs

@@ -40,7 +40,7 @@ use std::fmt;
 ///
 /// When implementing a new type of `Query`, it is normal to implement a
 /// dedicated `Query`, `Weight` and `Scorer`.
-pub trait Query: QueryClone + downcast_rs::Downcast + fmt::Debug {
+pub trait Query: QueryClone + Send + Sync + downcast_rs::Downcast + fmt::Debug {
     /// Create the weight associated to a query.
     ///
     /// If scoring is not required, setting `scoring_enabled` to `false`

src/query/query_parser/logical_ast.rs

@@ -2,6 +2,7 @@ use crate::query::Occur;
 use crate::schema::Field;
 use crate::schema::Term;
 use crate::schema::Type;
+use crate::Score;
 use std::fmt;
 use std::ops::Bound;
 
@@ -21,12 +22,12 @@ pub enum LogicalLiteral {
 pub enum LogicalAST {
     Clause(Vec<(Occur, LogicalAST)>),
     Leaf(Box<LogicalLiteral>),
-    Boost(Box<LogicalAST>, f32),
+    Boost(Box<LogicalAST>, Score),
 }
 
 impl LogicalAST {
-    pub fn boost(self, boost: f32) -> LogicalAST {
-        if (boost - 1.0f32).abs() < std::f32::EPSILON {
+    pub fn boost(self, boost: Score) -> LogicalAST {
+        if (boost - 1.0).abs() < Score::EPSILON {
             self
         } else {
             LogicalAST::Boost(Box::new(self), boost)

src/query/query_parser/query_parser.rs

@@ -12,6 +12,7 @@ use crate::schema::{Facet, IndexRecordOption};
 use crate::schema::{Field, Schema};
 use crate::schema::{FieldType, Term};
 use crate::tokenizer::TokenizerManager;
+use crate::Score;
 use std::borrow::Cow;
 use std::collections::HashMap;
 use std::num::{ParseFloatError, ParseIntError};
@@ -172,7 +173,7 @@ pub struct QueryParser {
     default_fields: Vec<Field>,
     conjunction_by_default: bool,
     tokenizer_manager: TokenizerManager,
-    boost: HashMap<Field, f32>,
+    boost: HashMap<Field, Score>,
 }
 
 fn all_negative(ast: &LogicalAST) -> bool {
@@ -228,7 +229,7 @@ impl QueryParser {
     /// If the query defines a query boost through the query language (e.g: `country:France^3.0`),
     /// the two boosts (the one defined in the query, and the one defined in the `QueryParser`)
     /// are multiplied together.
-    pub fn set_field_boost(&mut self, field: Field, boost: f32) {
+    pub fn set_field_boost(&mut self, field: Field, boost: Score) {
         self.boost.insert(field, boost);
     }
 
@@ -440,14 +441,14 @@ impl QueryParser {
             }
             UserInputAST::Boost(ast, boost) => {
                 let ast = self.compute_logical_ast_with_occur(*ast)?;
-                Ok(ast.boost(boost))
+                Ok(ast.boost(boost as Score))
             }
             UserInputAST::Leaf(leaf) => self.compute_logical_ast_from_leaf(*leaf),
         }
     }
 
-    fn field_boost(&self, field: Field) -> f32 {
-        self.boost.get(&field).cloned().unwrap_or(1.0f32)
+    fn field_boost(&self, field: Field) -> Score {
+        self.boost.get(&field).cloned().unwrap_or(1.0)
     }
 
     fn compute_logical_ast_from_leaf(
@@ -658,7 +659,7 @@ mod test {
         let mut query_parser = make_query_parser();
         let schema = make_schema();
         let text_field = schema.get_field("text").unwrap();
-        query_parser.set_field_boost(text_field, 2.0f32);
+        query_parser.set_field_boost(text_field, 2.0);
         let query = query_parser.parse_query("text:hello").unwrap();
         assert_eq!(
             format!("{:?}", query),
@@ -671,7 +672,7 @@ mod test {
         let mut query_parser = make_query_parser();
         let schema = make_schema();
         let title_field = schema.get_field("title").unwrap();
-        query_parser.set_field_boost(title_field, 2.0f32);
+        query_parser.set_field_boost(title_field, 2.0);
         let query = query_parser.parse_query("title:[A TO B]").unwrap();
         assert_eq!(
             format!("{:?}", query),
@@ -684,7 +685,7 @@ mod test {
         let mut query_parser = make_query_parser();
         let schema = make_schema();
         let text_field = schema.get_field("text").unwrap();
-        query_parser.set_field_boost(text_field, 2.0f32);
+        query_parser.set_field_boost(text_field, 2.0);
         let query = query_parser.parse_query("text:hello^2").unwrap();
         assert_eq!(
             format!("{:?}", query),

src/query/range_query.rs

@@ -9,8 +9,7 @@ use crate::query::{Query, Scorer, Weight};
 use crate::schema::Type;
 use crate::schema::{Field, IndexRecordOption, Term};
 use crate::termdict::{TermDictionary, TermStreamer};
-use crate::DocId;
-use crate::Result;
+use crate::{DocId, Score};
 use std::collections::Bound;
 use std::ops::Range;
 
@@ -48,7 +47,7 @@ fn map_bound<TFrom, TTo, Transform: Fn(&TFrom) -> TTo>(
 /// let schema = schema_builder.build();
 ///
 /// let index = Index::create_in_ram(schema);
-/// let mut index_writer = index.writer_with_num_threads(1, 6_000_000)?;
+/// let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
 /// for year in 1950u64..2017u64 {
 ///     let num_docs_within_year = 10 + (year - 1950) * (year - 1950);
 ///     for _ in 0..num_docs_within_year {
@@ -246,7 +245,11 @@ impl RangeQuery {
 }
 
 impl Query for RangeQuery {
-    fn weight(&self, searcher: &Searcher, _scoring_enabled: bool) -> Result<Box<dyn Weight>> {
+    fn weight(
+        &self,
+        searcher: &Searcher,
+        _scoring_enabled: bool,
+    ) -> crate::Result<Box<dyn Weight>> {
         let schema = searcher.schema();
         let value_type = schema.get_field_entry(self.field).field_type().value_type();
         if value_type != self.value_type {
@@ -289,7 +292,7 @@ impl RangeWeight {
 }
 
 impl Weight for RangeWeight {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         let max_doc = reader.max_doc();
         let mut doc_bitset = BitSet::with_max_value(max_doc);
 
@@ -315,12 +318,12 @@ impl Weight for RangeWeight {
         Ok(Box::new(ConstScorer::new(doc_bitset, boost)))
     }
 
-    fn explain(&self, reader: &SegmentReader, doc: DocId) -> Result<Explanation> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
+        let mut scorer = self.scorer(reader, 1.0)?;
         if scorer.seek(doc) != doc {
             return Err(does_not_match(doc));
         }
-        Ok(Explanation::new("RangeQuery", 1.0f32))
+        Ok(Explanation::new("RangeQuery", 1.0))
     }
 }
 
@@ -328,8 +331,9 @@ impl Weight for RangeWeight {
 mod tests {
 
     use super::RangeQuery;
-    use crate::collector::Count;
-    use crate::schema::{Document, Field, Schema, INDEXED};
+    use crate::collector::{Count, TopDocs};
+    use crate::query::QueryParser;
+    use crate::schema::{Document, Field, Schema, INDEXED, TEXT};
     use crate::Index;
     use std::collections::Bound;
 
@@ -341,7 +345,7 @@ mod tests {
 
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 6_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             for year in 1950u64..2017u64 {
                 let num_docs_within_year = 10 + (year - 1950) * (year - 1950);
                 for _ in 0..num_docs_within_year {
@@ -476,4 +480,28 @@ mod tests {
             91
         );
     }
+
+    #[test]
+    fn test_bug_reproduce_range_query() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        schema_builder.add_text_field("title", TEXT);
+        schema_builder.add_i64_field("year", INDEXED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema.clone());
+        let mut index_writer = index.writer_for_tests()?;
+        let title = schema.get_field("title").unwrap();
+        let year = schema.get_field("year").unwrap();
+        index_writer.add_document(doc!(
+          title => "hemoglobin blood",
+          year => 1990 as i64
+        ));
+        index_writer.commit()?;
+        let reader = index.reader()?;
+        let searcher = reader.searcher();
+        let query_parser = QueryParser::for_index(&index, vec![title]);
+        let query = query_parser.parse_query("hemoglobin AND year:[1970 TO 1990]")?;
+        let top_docs = searcher.search(&query, &TopDocs::with_limit(10))?;
+        assert_eq!(top_docs.len(), 1);
+        Ok(())
+    }
 }

src/query/regex_query.rs

@@ -103,7 +103,7 @@ mod test {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(
                 country_field => "japan",
             ));
@@ -129,7 +129,7 @@ mod test {
                 .unwrap();
             assert_eq!(scored_docs.len(), 1, "Expected only 1 document");
             let (score, _) = scored_docs[0];
-            assert_nearly_equals!(1f32, score);
+            assert_nearly_equals!(1.0, score);
         }
         let top_docs = searcher
             .search(&query_matching_zero, &TopDocs::with_limit(2))

src/query/reqopt_scorer.rs

@@ -112,47 +112,47 @@ mod tests {
     fn test_reqopt_scorer() {
         let mut reqoptscorer: RequiredOptionalScorer<_, _, SumCombiner> =
             RequiredOptionalScorer::new(
-                ConstScorer::new(VecDocSet::from(vec![1, 3, 7, 8, 9, 10, 13, 15]), 1.0f32),
-                ConstScorer::new(VecDocSet::from(vec![1, 2, 7, 11, 12, 15]), 1.0f32),
+                ConstScorer::new(VecDocSet::from(vec![1, 3, 7, 8, 9, 10, 13, 15]), 1.0),
+                ConstScorer::new(VecDocSet::from(vec![1, 2, 7, 11, 12, 15]), 1.0),
             );
         {
             assert_eq!(reqoptscorer.doc(), 1);
-            assert_eq!(reqoptscorer.score(), 2f32);
+            assert_eq!(reqoptscorer.score(), 2.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 3);
             assert_eq!(reqoptscorer.doc(), 3);
-            assert_eq!(reqoptscorer.score(), 1f32);
+            assert_eq!(reqoptscorer.score(), 1.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 7);
             assert_eq!(reqoptscorer.doc(), 7);
-            assert_eq!(reqoptscorer.score(), 2f32);
+            assert_eq!(reqoptscorer.score(), 2.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 8);
             assert_eq!(reqoptscorer.doc(), 8);
-            assert_eq!(reqoptscorer.score(), 1f32);
+            assert_eq!(reqoptscorer.score(), 1.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 9);
             assert_eq!(reqoptscorer.doc(), 9);
-            assert_eq!(reqoptscorer.score(), 1f32);
+            assert_eq!(reqoptscorer.score(), 1.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 10);
             assert_eq!(reqoptscorer.doc(), 10);
-            assert_eq!(reqoptscorer.score(), 1f32);
+            assert_eq!(reqoptscorer.score(), 1.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 13);
             assert_eq!(reqoptscorer.doc(), 13);
-            assert_eq!(reqoptscorer.score(), 1f32);
+            assert_eq!(reqoptscorer.score(), 1.0);
         }
         {
             assert_eq!(reqoptscorer.advance(), 15);
             assert_eq!(reqoptscorer.doc(), 15);
-            assert_eq!(reqoptscorer.score(), 2f32);
+            assert_eq!(reqoptscorer.score(), 2.0);
         }
         assert_eq!(reqoptscorer.advance(), TERMINATED);
     }

src/query/score_combiner.rs

@@ -31,7 +31,7 @@ impl ScoreCombiner for DoNothingCombiner {
     fn clear(&mut self) {}
 
     fn score(&self) -> Score {
-        1f32
+        1.0
     }
 }
 
@@ -47,7 +47,7 @@ impl ScoreCombiner for SumCombiner {
     }
 
     fn clear(&mut self) {
-        self.score = 0f32;
+        self.score = 0.0;
     }
 
     fn score(&self) -> Score {
@@ -70,7 +70,7 @@ impl ScoreCombiner for SumWithCoordsCombiner {
     }
 
     fn clear(&mut self) {
-        self.score = 0f32;
+        self.score = 0.0;
         self.num_fields = 0;
     }
 

src/query/scorer.rs

@@ -35,14 +35,14 @@ pub struct ConstScorer<TDocSet: DocSet> {
 
 impl<TDocSet: DocSet> ConstScorer<TDocSet> {
     /// Creates a new `ConstScorer`.
-    pub fn new(docset: TDocSet, score: f32) -> ConstScorer<TDocSet> {
+    pub fn new(docset: TDocSet, score: Score) -> ConstScorer<TDocSet> {
         ConstScorer { docset, score }
     }
 }
 
 impl<TDocSet: DocSet> From<TDocSet> for ConstScorer<TDocSet> {
     fn from(docset: TDocSet) -> Self {
-        ConstScorer::new(docset, 1.0f32)
+        ConstScorer::new(docset, 1.0)
     }
 }
 

src/query/term_query/mod.rs

@@ -15,7 +15,7 @@ mod tests {
     use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
     use crate::query::{Query, QueryParser, Scorer, TermQuery};
     use crate::schema::{Field, IndexRecordOption, Schema, STRING, TEXT};
-    use crate::{Term, Index, TERMINATED};
+    use crate::{Index, Term, TERMINATED};
 
     #[test]
     pub fn test_term_query_no_freq() {
@@ -25,7 +25,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             let doc = doc!(text_field => "a");
             index_writer.add_document(doc);
             assert!(index_writer.commit().is_ok());
@@ -37,9 +37,9 @@ mod tests {
         );
         let term_weight = term_query.weight(&searcher, true).unwrap();
         let segment_reader = searcher.segment_reader(0);
-        let mut term_scorer = term_weight.scorer(segment_reader, 1.0f32).unwrap();
+        let mut term_scorer = term_weight.scorer(segment_reader, 1.0).unwrap();
         assert_eq!(term_scorer.doc(), 0);
-        assert_eq!(term_scorer.score(), 0.28768212);
+        assert_nearly_equals!(term_scorer.score(), 0.28768212);
     }
 
     #[test]
@@ -50,7 +50,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000)?;
+            let mut index_writer = index.writer_for_tests()?;
             for _ in 0..COMPRESSION_BLOCK_SIZE {
                 let doc = doc!(text_field => "a");
                 index_writer.add_document(doc);
@@ -64,7 +64,7 @@ mod tests {
         );
         let term_weight = term_query.weight(&searcher, true)?;
         let segment_reader = searcher.segment_reader(0);
-        let mut term_scorer = term_weight.scorer(segment_reader, 1.0f32)?;
+        let mut term_scorer = term_weight.scorer(segment_reader, 1.0)?;
         for i in 0u32..COMPRESSION_BLOCK_SIZE as u32 {
             assert_eq!(term_scorer.doc(), i);
             if i == COMPRESSION_BLOCK_SIZE as u32 - 1u32 {
@@ -86,7 +86,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         {
-            let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(
                 left_field => "left1 left2 left2 left2f2 left2f2 left3 abcde abcde abcde abcde abcde abcde abcde abcde abcde abcewde abcde abcde",
                 right_field => "right1 right2",
@@ -136,7 +136,7 @@ mod tests {
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 5_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(text_field=>"a b"));
         index_writer.add_document(doc!(text_field=>"a c"));
         index_writer.delete_term(Term::from_field_text(text_field, "b"));
@@ -153,7 +153,7 @@ mod tests {
         let text_field = schema_builder.add_text_field("text", TEXT);
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
-        let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+        let mut index_writer = index.writer_for_tests().unwrap();
         index_writer.add_document(doc!(text_field=>"a"));
         index_writer.add_document(doc!(text_field=>"a"));
         index_writer.commit()?;
@@ -161,7 +161,7 @@ mod tests {
         let term_query = TermQuery::new(term_a, IndexRecordOption::Basic);
         let searcher = index.reader()?.searcher();
         let term_weight = term_query.weight(&searcher, false)?;
-        let mut term_scorer = term_weight.scorer(searcher.segment_reader(0u32), 1.0f32)?;
+        let mut term_scorer = term_weight.scorer(searcher.segment_reader(0u32), 1.0)?;
         assert_eq!(term_scorer.doc(), 0u32);
         term_scorer.seek(1u32);
         assert_eq!(term_scorer.doc(), 1u32);

src/query/term_query/term_scorer.rs

@@ -8,6 +8,7 @@ use crate::postings::SegmentPostings;
 use crate::postings::{FreqReadingOption, Postings};
 use crate::query::bm25::BM25Weight;
 
+#[derive(Clone)]
 pub struct TermScorer {
     postings: SegmentPostings,
     fieldnorm_reader: FieldNormReader,
@@ -28,34 +29,28 @@ impl TermScorer {
     }
 
     pub(crate) fn shallow_seek(&mut self, target_doc: DocId) {
-        self.postings.block_cursor.shallow_seek(target_doc)
+        self.postings.block_cursor.shallow_seek(target_doc);
     }
 
-
     #[cfg(test)]
     pub fn create_for_test(
         doc_and_tfs: &[(DocId, u32)],
-        fieldnorm_vals: &[u32],
+        fieldnorms: &[u32],
         similarity_weight: BM25Weight,
-    ) -> crate::Result<TermScorer> {
+    ) -> TermScorer {
         assert!(!doc_and_tfs.is_empty());
-        assert!(doc_and_tfs.len() <= fieldnorm_vals.len());
-
-        let doc_freq = doc_and_tfs.len();
-        let max_doc = doc_and_tfs.last().unwrap().0 + 1;
-        let mut fieldnorms: Vec<u32> = std::iter::repeat(1).take(max_doc as usize).collect();
-
-        for i in 0..doc_freq {
-            let doc = doc_and_tfs[i].0;
-            let fieldnorm = fieldnorm_vals[i];
-            fieldnorms[doc as usize] = fieldnorm;
-        }
-        let fieldnorm_reader = FieldNormReader::from(&fieldnorms[..]);
-
+        assert!(
+            doc_and_tfs
+                .iter()
+                .map(|(doc, _tf)| *doc)
+                .max()
+                .unwrap_or(0u32)
+                < fieldnorms.len() as u32
+        );
         let segment_postings =
-            SegmentPostings::create_from_docs_and_tfs(doc_and_tfs, Some(fieldnorm_reader.clone()))?;
-
-        Ok(TermScorer::new(segment_postings, fieldnorm_reader, similarity_weight))
+            SegmentPostings::create_from_docs_and_tfs(doc_and_tfs, Some(fieldnorms));
+        let fieldnorm_reader = FieldNormReader::for_test(fieldnorms);
+        TermScorer::new(segment_postings, fieldnorm_reader, similarity_weight)
     }
 
     /// See `FreqReadingOption`.
@@ -86,10 +81,6 @@ impl TermScorer {
         self.postings.term_freq()
     }
 
-    pub fn doc_freq(&self) -> usize {
-        self.postings.doc_freq() as usize
-    }
-
     pub fn fieldnorm_id(&self) -> u8 {
         self.fieldnorm_reader.fieldnorm_id(self.doc())
     }
@@ -100,7 +91,7 @@ impl TermScorer {
         self.similarity_weight.explain(fieldnorm_id, term_freq)
     }
 
-    pub fn max_score(&self) -> f32 {
+    pub fn max_score(&self) -> Score {
         self.similarity_weight.max_score()
     }
 
@@ -137,48 +128,53 @@ impl Scorer for TermScorer {
 
 #[cfg(test)]
 mod tests {
-    use crate::assert_nearly_equals;
+    use crate::merge_policy::NoMergePolicy;
     use crate::postings::compression::COMPRESSION_BLOCK_SIZE;
     use crate::query::term_query::TermScorer;
-    use crate::query::{BM25Weight, Scorer};
+    use crate::query::{BM25Weight, Scorer, TermQuery};
+    use crate::schema::{IndexRecordOption, Schema, TEXT};
+    use crate::Score;
+    use crate::{assert_nearly_equals, Index, Searcher, SegmentId, Term};
     use crate::{DocId, DocSet, TERMINATED};
+    use futures::executor::block_on;
     use proptest::prelude::*;
 
     #[test]
     fn test_term_scorer_max_score() -> crate::Result<()> {
-        let bm25_weight = BM25Weight::for_one_term(3, 6, 10f32);
-        let mut term_scorer =
-            TermScorer::create_for_test(&[(2, 3), (3, 12), (7, 8)], &[10, 12, 100], bm25_weight)?;
+        let bm25_weight = BM25Weight::for_one_term(3, 6, 10.0);
+        let mut term_scorer = TermScorer::create_for_test(
+            &[(2, 3), (3, 12), (7, 8)],
+            &[0, 0, 10, 12, 0, 0, 0, 100],
+            bm25_weight,
+        );
         let max_scorer = term_scorer.max_score();
-        assert_eq!(max_scorer, 1.3990127f32);
+        crate::assert_nearly_equals!(max_scorer, 1.3990127);
         assert_eq!(term_scorer.doc(), 2);
         assert_eq!(term_scorer.term_freq(), 3);
-        assert_nearly_equals!(term_scorer.block_max_score(), 1.3676447f32);
-        assert_nearly_equals!(term_scorer.score(), 1.0892314f32);
+        assert_nearly_equals!(term_scorer.block_max_score(), 1.3676447);
+        assert_nearly_equals!(term_scorer.score(), 1.0892314);
         assert_eq!(term_scorer.advance(), 3);
         assert_eq!(term_scorer.doc(), 3);
         assert_eq!(term_scorer.term_freq(), 12);
-        assert_nearly_equals!(term_scorer.score(), 1.3676447f32);
+        assert_nearly_equals!(term_scorer.score(), 1.3676447);
         assert_eq!(term_scorer.advance(), 7);
         assert_eq!(term_scorer.doc(), 7);
         assert_eq!(term_scorer.term_freq(), 8);
-        assert_nearly_equals!(term_scorer.score(), 0.72015285f32);
+        assert_nearly_equals!(term_scorer.score(), 0.72015285);
         assert_eq!(term_scorer.advance(), TERMINATED);
         Ok(())
     }
 
-
     #[test]
     fn test_term_scorer_shallow_advance() -> crate::Result<()> {
-        let bm25_weight = BM25Weight::for_one_term(300, 1024, 10f32);
+        let bm25_weight = BM25Weight::for_one_term(300, 1024, 10.0);
         let mut doc_and_tfs = vec![];
         for i in 0u32..300u32 {
             let doc = i * 10;
             doc_and_tfs.push((doc, 1u32 + doc % 3u32));
         }
-        let fieldnorms: Vec<u32> = std::iter::repeat(10u32).take(1024).collect();
-        let mut term_scorer =
-            TermScorer::create_for_test(&doc_and_tfs, &fieldnorms, bm25_weight)?;
+        let fieldnorms: Vec<u32> = std::iter::repeat(10u32).take(3_000).collect();
+        let mut term_scorer = TermScorer::create_for_test(&doc_and_tfs, &fieldnorms, bm25_weight);
         assert_eq!(term_scorer.doc(), 0u32);
         term_scorer.shallow_seek(1289);
         assert_eq!(term_scorer.doc(), 0u32);
@@ -205,7 +201,7 @@ mod tests {
          let average_fieldnorm = fieldnorms
              .iter()
              .cloned()
-             .sum::<u32>() as f32 / term_doc_freq as f32;
+             .sum::<u32>() as Score / term_doc_freq as Score;
              // Average fieldnorm is over the entire index,
              // not necessarily the docs that are in the posting list.
              // For this reason we multiply by 1.1 to make a realistic value.
@@ -214,12 +210,12 @@ mod tests {
             average_fieldnorm);
 
          let mut term_scorer =
-              TermScorer::create_for_test(&doc_tfs[..], &fieldnorms[..], bm25_weight).unwrap();
+              TermScorer::create_for_test(&doc_tfs[..], &fieldnorms[..], bm25_weight);
 
          let docs: Vec<DocId> = (0..term_doc_freq).map(|doc| doc as DocId).collect();
          for block in docs.chunks(COMPRESSION_BLOCK_SIZE) {
-             let block_max_score = term_scorer.block_max_score();
-             let mut block_max_score_computed = 0.0f32;
+             let block_max_score: Score = term_scorer.block_max_score();
+             let mut block_max_score_computed: Score = 0.0;
              for &doc in block {
                 assert_eq!(term_scorer.doc(), doc);
                 block_max_score_computed = block_max_score_computed.max(term_scorer.score());
@@ -229,4 +225,113 @@ mod tests {
          }
         }
     }
+
+    #[test]
+    fn test_block_wand() {
+        let mut doc_tfs: Vec<(u32, u32)> = vec![];
+        for doc in 0u32..128u32 {
+            doc_tfs.push((doc, 1u32));
+        }
+        for doc in 128u32..256u32 {
+            doc_tfs.push((doc, if doc == 200 { 2u32 } else { 1u32 }));
+        }
+        doc_tfs.push((256, 1u32));
+        doc_tfs.push((257, 3u32));
+        doc_tfs.push((258, 1u32));
+
+        let fieldnorms: Vec<u32> = std::iter::repeat(20u32).take(300).collect();
+        let bm25_weight = BM25Weight::for_one_term(10, 129, 20.0);
+        let mut docs = TermScorer::create_for_test(&doc_tfs[..], &fieldnorms[..], bm25_weight);
+        assert_nearly_equals!(docs.block_max_score(), 2.5161593);
+        docs.shallow_seek(135);
+        assert_nearly_equals!(docs.block_max_score(), 3.4597192);
+        docs.shallow_seek(256);
+        // the block is not loaded yet.
+        assert_nearly_equals!(docs.block_max_score(), 5.2971773);
+        assert_eq!(256, docs.seek(256));
+        assert_nearly_equals!(docs.block_max_score(), 3.9539647);
+    }
+
+    fn test_block_wand_aux(term_query: &TermQuery, searcher: &Searcher) -> crate::Result<()> {
+        let term_weight = term_query.specialized_weight(&searcher, true);
+        for reader in searcher.segment_readers() {
+            let mut block_max_scores = vec![];
+            let mut block_max_scores_b = vec![];
+            let mut docs = vec![];
+            {
+                let mut term_scorer = term_weight.specialized_scorer(reader, 1.0)?;
+                while term_scorer.doc() != TERMINATED {
+                    let mut score = term_scorer.score();
+                    docs.push(term_scorer.doc());
+                    for _ in 0..128 {
+                        score = score.max(term_scorer.score());
+                        if term_scorer.advance() == TERMINATED {
+                            break;
+                        }
+                    }
+                    block_max_scores.push(score);
+                }
+            }
+            {
+                let mut term_scorer = term_weight.specialized_scorer(reader, 1.0)?;
+                for d in docs {
+                    term_scorer.shallow_seek(d);
+                    block_max_scores_b.push(term_scorer.block_max_score());
+                }
+            }
+            for (l, r) in block_max_scores
+                .iter()
+                .cloned()
+                .zip(block_max_scores_b.iter().cloned())
+            {
+                assert_nearly_equals!(l, r);
+            }
+        }
+        Ok(())
+    }
+
+    #[ignore]
+    #[test]
+    fn test_block_wand_long_test() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        let text_field = schema_builder.add_text_field("text", TEXT);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut writer = index.writer_with_num_threads(3, 30_000_000)?;
+        use rand::Rng;
+        let mut rng = rand::thread_rng();
+        writer.set_merge_policy(Box::new(NoMergePolicy));
+        for _ in 0..3_000 {
+            let term_freq = rng.gen_range(1, 10000);
+            let words: Vec<&str> = std::iter::repeat("bbbb").take(term_freq).collect();
+            let text = words.join(" ");
+            writer.add_document(doc!(text_field=>text));
+        }
+        writer.commit()?;
+        let term_query = TermQuery::new(
+            Term::from_field_text(text_field, &"bbbb"),
+            IndexRecordOption::WithFreqs,
+        );
+        let segment_ids: Vec<SegmentId>;
+        let reader = index.reader()?;
+        {
+            let searcher = reader.searcher();
+            segment_ids = searcher
+                .segment_readers()
+                .iter()
+                .map(|segment| segment.segment_id())
+                .collect();
+            test_block_wand_aux(&term_query, &searcher)?;
+        }
+        {
+            let _ = block_on(writer.merge(&segment_ids[..]));
+        }
+        {
+            reader.reload()?;
+            let searcher = reader.searcher();
+            assert_eq!(searcher.segment_readers().len(), 1);
+            test_block_wand_aux(&term_query, &searcher)?;
+        }
+        Ok(())
+    }
 }

src/query/term_query/term_weight.rs

@@ -4,11 +4,10 @@ use crate::docset::DocSet;
 use crate::postings::SegmentPostings;
 use crate::query::bm25::BM25Weight;
 use crate::query::explanation::does_not_match;
-use crate::query::weight::{for_each_pruning_scorer, for_each_scorer};
+use crate::query::weight::for_each_scorer;
 use crate::query::Weight;
 use crate::query::{Explanation, Scorer};
 use crate::schema::IndexRecordOption;
-use crate::Result;
 use crate::Term;
 use crate::{DocId, Score};
 
@@ -19,22 +18,22 @@ pub struct TermWeight {
 }
 
 impl Weight for TermWeight {
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> Result<Box<dyn Scorer>> {
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
         let term_scorer = self.specialized_scorer(reader, boost)?;
         Ok(Box::new(term_scorer))
     }
 
-    fn explain(&self, reader: &SegmentReader, doc: DocId) -> Result<Explanation> {
-        let mut scorer = self.specialized_scorer(reader, 1.0f32)?;
+    fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
+        let mut scorer = self.specialized_scorer(reader, 1.0)?;
         if scorer.seek(doc) != doc {
             return Err(does_not_match(doc));
         }
         Ok(scorer.explain())
     }
 
-    fn count(&self, reader: &SegmentReader) -> Result<u32> {
+    fn count(&self, reader: &SegmentReader) -> crate::Result<u32> {
         if let Some(delete_bitset) = reader.delete_bitset() {
-            Ok(self.scorer(reader, 1.0f32)?.count(delete_bitset))
+            Ok(self.scorer(reader, 1.0)?.count(delete_bitset))
         } else {
             let field = self.term.field();
             Ok(reader
@@ -52,7 +51,7 @@ impl Weight for TermWeight {
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score),
     ) -> crate::Result<()> {
-        let mut scorer = self.specialized_scorer(reader, 1.0f32)?;
+        let mut scorer = self.specialized_scorer(reader, 1.0)?;
         for_each_scorer(&mut scorer, callback);
         Ok(())
     }
@@ -69,12 +68,12 @@ impl Weight for TermWeight {
     /// important optimization (e.g. BlockWAND for union).
     fn for_each_pruning(
         &self,
-        threshold: f32,
+        threshold: Score,
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score) -> Score,
     ) -> crate::Result<()> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
-        for_each_pruning_scorer(&mut scorer, threshold, callback);
+        let scorer = self.specialized_scorer(reader, 1.0)?;
+        crate::query::boolean_query::block_wand(vec![scorer], threshold, callback);
         Ok(())
     }
 }
@@ -92,10 +91,14 @@ impl TermWeight {
         }
     }
 
-    pub fn specialized_scorer(&self, reader: &SegmentReader, boost: f32) -> Result<TermScorer> {
+    pub(crate) fn specialized_scorer(
+        &self,
+        reader: &SegmentReader,
+        boost: Score,
+    ) -> crate::Result<TermScorer> {
         let field = self.term.field();
         let inverted_index = reader.inverted_index(field);
-        let fieldnorm_reader = reader.get_fieldnorms_reader(field);
+        let fieldnorm_reader = reader.get_fieldnorms_reader(field)?;
         let similarity_weight = self.similarity_weight.boost_by(boost);
         let postings_opt: Option<SegmentPostings> =
             inverted_index.read_postings(&self.term, self.index_record_option);

src/query/union.rs

@@ -55,7 +55,7 @@ where
             cursor: HORIZON_NUM_TINYBITSETS,
             offset: 0,
             doc: 0,
-            score: 0f32,
+            score: 0.0,
         };
         if union.refill() {
             union.advance();
@@ -274,7 +274,7 @@ mod tests {
                 vals.iter()
                     .cloned()
                     .map(VecDocSet::from)
-                    .map(|docset| ConstScorer::new(docset, 1.0f32))
+                    .map(|docset| ConstScorer::new(docset, 1.0))
                     .collect::<Vec<ConstScorer<VecDocSet>>>(),
             )
         };
@@ -321,7 +321,7 @@ mod tests {
                     .iter()
                     .map(|docs| docs.clone())
                     .map(VecDocSet::from)
-                    .map(|docset| ConstScorer::new(docset, 1.0f32))
+                    .map(|docset| ConstScorer::new(docset, 1.0))
                     .collect::<Vec<_>>(),
             ));
             res
@@ -398,9 +398,9 @@ mod bench {
 
     use crate::query::score_combiner::DoNothingCombiner;
     use crate::query::{ConstScorer, Union, VecDocSet};
-    use crate::tests;
     use crate::DocId;
     use crate::DocSet;
+    use crate::{tests, TERMINATED};
     use test::Bencher;
 
     #[bench]
@@ -414,10 +414,12 @@ mod bench {
                 union_docset
                     .iter()
                     .map(|doc_ids| VecDocSet::from(doc_ids.clone()))
-                    .map(ConstScorer::new)
+                    .map(|docset| ConstScorer::new(docset, 1.0))
                     .collect::<Vec<_>>(),
             );
-            while v.advance() {}
+            while v.doc() != TERMINATED {
+                v.advance();
+            }
         });
     }
     #[bench]
@@ -432,10 +434,12 @@ mod bench {
                 union_docset
                     .iter()
                     .map(|doc_ids| VecDocSet::from(doc_ids.clone()))
-                    .map(ConstScorer::new)
+                    .map(|docset| ConstScorer::new(docset, 1.0))
                     .collect::<Vec<_>>(),
             );
-            while v.advance() {}
+            while v.doc() != TERMINATED {
+                v.advance();
+            }
         });
     }
 }

src/query/weight.rs

@@ -28,7 +28,7 @@ pub(crate) fn for_each_scorer<TScorer: Scorer + ?Sized>(
 /// important optimization (e.g. BlockWAND for union).
 pub(crate) fn for_each_pruning_scorer<TScorer: Scorer + ?Sized>(
     scorer: &mut TScorer,
-    mut threshold: f32,
+    mut threshold: Score,
     callback: &mut dyn FnMut(DocId, Score) -> Score,
 ) {
     let mut doc = scorer.doc();
@@ -51,14 +51,14 @@ pub trait Weight: Send + Sync + 'static {
     /// `boost` is a multiplier to apply to the score.
     ///
     /// See [`Query`](./trait.Query.html).
-    fn scorer(&self, reader: &SegmentReader, boost: f32) -> crate::Result<Box<dyn Scorer>>;
+    fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>>;
 
     /// Returns an `Explanation` for the given document.
     fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation>;
 
     /// Returns the number documents within the given `SegmentReader`.
     fn count(&self, reader: &SegmentReader) -> crate::Result<u32> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+        let mut scorer = self.scorer(reader, 1.0)?;
         if let Some(delete_bitset) = reader.delete_bitset() {
             Ok(scorer.count(delete_bitset))
         } else {
@@ -73,7 +73,7 @@ pub trait Weight: Send + Sync + 'static {
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score),
     ) -> crate::Result<()> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+        let mut scorer = self.scorer(reader, 1.0)?;
         for_each_scorer(scorer.as_mut(), callback);
         Ok(())
     }
@@ -90,11 +90,11 @@ pub trait Weight: Send + Sync + 'static {
     /// important optimization (e.g. BlockWAND for union).
     fn for_each_pruning(
         &self,
-        threshold: f32,
+        threshold: Score,
         reader: &SegmentReader,
         callback: &mut dyn FnMut(DocId, Score) -> Score,
     ) -> crate::Result<()> {
-        let mut scorer = self.scorer(reader, 1.0f32)?;
+        let mut scorer = self.scorer(reader, 1.0)?;
         for_each_pruning_scorer(scorer.as_mut(), threshold, callback);
         Ok(())
     }

src/reader/mod.rs

@@ -138,9 +138,11 @@ impl InnerIndexReader {
                 .collect::<crate::Result<_>>()?
         };
         let schema = self.index.schema();
-        let searchers = (0..self.num_searchers)
-            .map(|_| Searcher::new(schema.clone(), self.index.clone(), segment_readers.clone()))
-            .collect();
+        let searchers = std::iter::repeat_with(|| {
+            Searcher::new(schema.clone(), self.index.clone(), segment_readers.clone())
+        })
+        .take(self.num_searchers)
+        .collect();
         self.searcher_pool.publish_new_generation(searchers);
         Ok(())
     }

src/schema/document.rs

@@ -74,9 +74,8 @@ impl Document {
     }
 
     /// Add a text field.
-    pub fn add_text(&mut self, field: Field, text: &str) {
-        let value = Value::Str(String::from(text));
-        self.add(FieldValue::new(field, value));
+    pub fn add_text<S: ToString>(&mut self, field: Field, text: S) {
+        self.add(FieldValue::new(field, Value::Str(text.to_string())));
     }
 
     /// Add a pre-tokenized text field.
@@ -110,8 +109,8 @@ impl Document {
     }
 
     /// Add a bytes field
-    pub fn add_bytes(&mut self, field: Field, value: Vec<u8>) {
-        self.add(FieldValue::new(field, Value::Bytes(value)))
+    pub fn add_bytes<T: Into<Vec<u8>>>(&mut self, field: Field, value: T) {
+        self.add(FieldValue::new(field, Value::Bytes(value.into())))
     }
 
     /// Add a field value

src/schema/field_entry.rs

@@ -1,5 +1,5 @@
-use crate::schema::IntOptions;
 use crate::schema::TextOptions;
+use crate::schema::{is_valid_field_name, IntOptions};
 
 use crate::schema::FieldType;
 use serde::de::{self, MapAccess, Visitor};
@@ -24,6 +24,7 @@ impl FieldEntry {
     /// Creates a new u64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_text(field_name: String, text_options: TextOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::Str(text_options),
@@ -33,6 +34,7 @@ impl FieldEntry {
     /// Creates a new u64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_u64(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::U64(field_type),
@@ -42,6 +44,7 @@ impl FieldEntry {
     /// Creates a new i64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_i64(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::I64(field_type),
@@ -51,6 +54,7 @@ impl FieldEntry {
     /// Creates a new f64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_f64(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::F64(field_type),
@@ -60,6 +64,7 @@ impl FieldEntry {
     /// Creates a new date field entry in the schema, given
     /// a name, and some options.
     pub fn new_date(field_name: String, field_type: IntOptions) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::Date(field_type),
@@ -68,6 +73,7 @@ impl FieldEntry {
 
     /// Creates a field entry for a facet.
     pub fn new_facet(field_name: String) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::HierarchicalFacet,
@@ -76,6 +82,7 @@ impl FieldEntry {
 
     /// Creates a field entry for a bytes field
     pub fn new_bytes(field_name: String) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
         FieldEntry {
             name: field_name,
             field_type: FieldType::Bytes,
@@ -268,6 +275,12 @@ mod tests {
     use crate::schema::TEXT;
     use serde_json;
 
+    #[test]
+    #[should_panic]
+    fn test_invalid_field_name_should_panic() {
+        FieldEntry::new_text("-hello".to_string(), TEXT);
+    }
+
     #[test]
     fn test_json_serialization() {
         let field_value = FieldEntry::new_text(String::from("title"), TEXT);

src/schema/mod.rs

@@ -149,14 +149,16 @@ pub use self::int_options::IntOptions;
 use once_cell::sync::Lazy;
 use regex::Regex;
 
+/// Regular expression representing the restriction on a valid field names.
+pub const FIELD_NAME_PATTERN: &str = r#"^[_a-zA-Z][_\-a-zA-Z0-9]*$"#;
+
 /// Validator for a potential `field_name`.
 /// Returns true iff the name can be use for a field name.
 ///
 /// A field name must start by a letter `[a-zA-Z]`.
 /// The other characters can be any alphanumic character `[a-ZA-Z0-9]` or `_`.
 pub fn is_valid_field_name(field_name: &str) -> bool {
-    static FIELD_NAME_PTN: Lazy<Regex> =
-        Lazy::new(|| Regex::new("^[a-zA-Z][_a-zA-Z0-9]*$").unwrap());
+    static FIELD_NAME_PTN: Lazy<Regex> = Lazy::new(|| Regex::new(FIELD_NAME_PATTERN).unwrap());
     FIELD_NAME_PTN.is_match(field_name)
 }
 
@@ -170,6 +172,11 @@ mod tests {
         assert!(is_valid_field_name("text"));
         assert!(is_valid_field_name("text0"));
         assert!(!is_valid_field_name("0text"));
+        assert!(is_valid_field_name("field-name"));
+        assert!(is_valid_field_name("field_name"));
+        assert!(!is_valid_field_name("field!name"));
+        assert!(!is_valid_field_name("-fieldname"));
+        assert!(is_valid_field_name("_fieldname"));
         assert!(!is_valid_field_name(""));
         assert!(!is_valid_field_name("シャボン玉"));
         assert!(is_valid_field_name("my_text_field"));

src/schema/term.rs

@@ -4,7 +4,6 @@ use super::Field;
 use crate::common;
 use crate::schema::Facet;
 use crate::DateTime;
-use byteorder::{BigEndian, ByteOrder};
 use std::str;
 
 /// Size (in bytes) of the buffer of a int field.
@@ -19,6 +18,10 @@ where
     B: AsRef<[u8]>;
 
 impl Term {
+    pub(crate) fn new() -> Term {
+        Term(Vec::with_capacity(100))
+    }
+
     /// Builds a term given a field, and a i64-value
     ///
     /// Assuming the term has a field id of 1, and a i64 value of 3234,
@@ -93,6 +96,12 @@ impl Term {
         term
     }
 
+    pub(crate) fn from_field_bytes(field: Field, bytes: &[u8]) -> Term {
+        let mut term = Term::for_field(field);
+        term.set_bytes(bytes);
+        term
+    }
+
     /// Creates a new Term for a given field.
     pub(crate) fn for_field(field: Field) -> Term {
         let mut term = Term(Vec::with_capacity(100));
@@ -100,12 +109,10 @@ impl Term {
         term
     }
 
-    /// Returns the field.
-    pub fn set_field(&mut self, field: Field) {
-        if self.0.len() < 4 {
-            self.0.resize(4, 0u8);
-        }
-        BigEndian::write_u32(&mut self.0[0..4], field.field_id());
+    pub(crate) fn set_field(&mut self, field: Field) {
+        self.0.clear();
+        self.0
+            .extend_from_slice(&field.field_id().to_be_bytes()[..]);
     }
 
     /// Sets a u64 value in the term.
@@ -116,7 +123,7 @@ impl Term {
     /// the natural order of the values.
     pub fn set_u64(&mut self, val: u64) {
         self.0.resize(INT_TERM_LEN, 0u8);
-        BigEndian::write_u64(&mut self.0[4..], val);
+        self.0[4..12].copy_from_slice(val.to_be_bytes().as_ref());
     }
 
     /// Sets a `i64` value in the term.
@@ -134,12 +141,6 @@ impl Term {
         self.0.extend(bytes);
     }
 
-    pub(crate) fn from_field_bytes(field: Field, bytes: &[u8]) -> Term {
-        let mut term = Term::for_field(field);
-        term.set_bytes(bytes);
-        term
-    }
-
     /// Set the texts only, keeping the field untouched.
     pub fn set_text(&mut self, text: &str) {
         self.set_bytes(text.as_bytes());
@@ -157,7 +158,9 @@ where
 
     /// Returns the field.
     pub fn field(&self) -> Field {
-        Field::from_field_id(BigEndian::read_u32(&self.0.as_ref()[..4]))
+        let mut field_id_bytes = [0u8; 4];
+        field_id_bytes.copy_from_slice(&self.0.as_ref()[..4]);
+        Field::from_field_id(u32::from_be_bytes(field_id_bytes))
     }
 
     /// Returns the `u64` value stored in a term.
@@ -166,7 +169,9 @@ where
     /// ... or returns an invalid value
     /// if the term is not a `u64` field.
     pub fn get_u64(&self) -> u64 {
-        BigEndian::read_u64(&self.0.as_ref()[4..])
+        let mut field_id_bytes = [0u8; 8];
+        field_id_bytes.copy_from_slice(self.value_bytes());
+        u64::from_be_bytes(field_id_bytes)
     }
 
     /// Returns the `i64` value stored in a term.
@@ -175,7 +180,7 @@ where
     /// ... or returns an invalid value
     /// if the term is not a `i64` field.
     pub fn get_i64(&self) -> i64 {
-        common::u64_to_i64(BigEndian::read_u64(&self.0.as_ref()[4..]))
+        common::u64_to_i64(self.get_u64())
     }
 
     /// Returns the `f64` value stored in a term.
@@ -184,7 +189,7 @@ where
     /// ... or returns an invalid value
     /// if the term is not a `f64` field.
     pub fn get_f64(&self) -> f64 {
-        common::u64_to_f64(BigEndian::read_u64(&self.0.as_ref()[4..]))
+        common::u64_to_f64(self.get_u64())
     }
 
     /// Returns the text associated with the term.

src/schema/value.rs

@@ -221,6 +221,12 @@ impl<'a> From<&'a str> for Value {
     }
 }
 
+impl<'a> From<&'a [u8]> for Value {
+    fn from(bytes: &'a [u8]) -> Value {
+        Value::Bytes(bytes.to_vec())
+    }
+}
+
 impl<'a> From<Facet> for Value {
     fn from(facet: Facet) -> Value {
         Value::Facet(facet)

src/snippet/mod.rs

@@ -2,8 +2,8 @@ use crate::query::Query;
 use crate::schema::Field;
 use crate::schema::Value;
 use crate::tokenizer::{TextAnalyzer, Token};
-use crate::Document;
 use crate::Searcher;
+use crate::{Document, Score};
 use htmlescape::encode_minimal;
 use std::cmp::Ordering;
 use std::collections::BTreeMap;
@@ -30,7 +30,7 @@ impl HighlightSection {
 
 #[derive(Debug)]
 pub struct FragmentCandidate {
-    score: f32,
+    score: Score,
     start_offset: usize,
     stop_offset: usize,
     num_chars: usize,
@@ -58,7 +58,7 @@ impl FragmentCandidate {
     /// taking the token and terms, the token is added to the fragment.
     /// if the token is one of the terms, the score
     /// and highlighted fields are updated in the fragment.
-    fn try_add_token(&mut self, token: &Token, terms: &BTreeMap<String, f32>) {
+    fn try_add_token(&mut self, token: &Token, terms: &BTreeMap<String, Score>) {
         self.stop_offset = token.offset_to;
 
         if let Some(&score) = terms.get(&token.text.to_lowercase()) {
@@ -142,7 +142,7 @@ impl Snippet {
 fn search_fragments<'a>(
     tokenizer: &TextAnalyzer,
     text: &'a str,
-    terms: &BTreeMap<String, f32>,
+    terms: &BTreeMap<String, Score>,
     max_num_chars: usize,
 ) -> Vec<FragmentCandidate> {
     let mut token_stream = tokenizer.token_stream(text);
@@ -221,7 +221,7 @@ fn select_best_fragment_combination(fragments: &[FragmentCandidate], text: &str)
 /// #    let text_field = schema_builder.add_text_field("text", TEXT);
 /// #    let schema = schema_builder.build();
 /// #    let index = Index::create_in_ram(schema);
-/// #    let mut index_writer = index.writer_with_num_threads(1, 30_000_000)?;
+/// #    let mut index_writer = index.writer_with_num_threads(1, 10_000_000)?;
 /// #    let doc = doc!(text_field => r#"Comme je descendais des Fleuves impassibles,
 /// #   Je ne me sentis plus guidé par les haleurs :
 /// #  Des Peaux-Rouges criards les avaient pris pour cibles,
@@ -248,7 +248,7 @@ fn select_best_fragment_combination(fragments: &[FragmentCandidate], text: &str)
 /// # }
 /// ```
 pub struct SnippetGenerator {
-    terms_text: BTreeMap<String, f32>,
+    terms_text: BTreeMap<String, Score>,
     tokenizer: TextAnalyzer,
     field: Field,
     max_num_chars: usize,
@@ -263,12 +263,12 @@ impl SnippetGenerator {
     ) -> crate::Result<SnippetGenerator> {
         let mut terms = BTreeSet::new();
         query.query_terms(&mut terms);
-        let terms_text: BTreeMap<String, f32> = terms
+        let terms_text: BTreeMap<String, Score> = terms
             .into_iter()
             .filter(|term| term.field() == field)
             .flat_map(|term| {
                 let doc_freq = searcher.doc_freq(&term);
-                let score = 1f32 / (1f32 + doc_freq as f32);
+                let score = 1.0 / (1.0 + doc_freq as Score);
                 if doc_freq > 0 {
                     Some((term.text().to_string(), score))
                 } else {
@@ -291,7 +291,7 @@ impl SnippetGenerator {
     }
 
     #[cfg(test)]
-    pub fn terms_text(&self) -> &BTreeMap<String, f32> {
+    pub fn terms_text(&self) -> &BTreeMap<String, Score> {
         &self.terms_text
     }
 
@@ -373,8 +373,8 @@ Survey in 2016, 2017, and 2018."#;
     fn test_snippet_scored_fragment() {
         {
             let terms = btreemap! {
-                String::from("rust") =>1.0f32,
-                String::from("language") => 0.9f32
+                String::from("rust") =>1.0,
+                String::from("language") => 0.9
             };
             let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 20);
             {
@@ -387,8 +387,8 @@ Survey in 2016, 2017, and 2018."#;
         }
         {
             let terms = btreemap! {
-                String::from("rust") =>0.9f32,
-                String::from("language") => 1.0f32
+                String::from("rust") =>0.9,
+                String::from("language") => 1.0
             };
             let fragments = search_fragments(&From::from(SimpleTokenizer), TEST_TEXT, &terms, 20);
             //assert_eq!(fragments.len(), 7);
@@ -506,7 +506,7 @@ Survey in 2016, 2017, and 2018."#;
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(text_field => "a"));
             index_writer.add_document(doc!(text_field => "a"));
             index_writer.add_document(doc!(text_field => "a b"));
@@ -525,7 +525,7 @@ Survey in 2016, 2017, and 2018."#;
             let snippet_generator =
                 SnippetGenerator::create(&searcher, &*query, text_field).unwrap();
             assert_eq!(
-                &btreemap!("a".to_string() => 0.25f32),
+                &btreemap!("a".to_string() => 0.25),
                 snippet_generator.terms_text()
             );
         }
@@ -534,7 +534,7 @@ Survey in 2016, 2017, and 2018."#;
             let snippet_generator =
                 SnippetGenerator::create(&searcher, &*query, text_field).unwrap();
             assert_eq!(
-                &btreemap!("a".to_string() => 0.25f32, "b".to_string() => 0.5),
+                &btreemap!("a".to_string() => 0.25, "b".to_string() => 0.5),
                 snippet_generator.terms_text()
             );
         }
@@ -543,7 +543,7 @@ Survey in 2016, 2017, and 2018."#;
             let snippet_generator =
                 SnippetGenerator::create(&searcher, &*query, text_field).unwrap();
             assert_eq!(
-                &btreemap!("a".to_string() => 0.25f32, "b".to_string() => 0.5),
+                &btreemap!("a".to_string() => 0.25, "b".to_string() => 0.5),
                 snippet_generator.terms_text()
             );
         }
@@ -562,7 +562,7 @@ Survey in 2016, 2017, and 2018."#;
         let index = Index::create_in_ram(schema);
         {
             // writing the segment
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             {
                 let doc = doc ! (text_field => TEST_TEXT);
                 index_writer.add_document(doc);

src/space_usage/mod.rs

@@ -25,6 +25,8 @@ pub enum ComponentSpaceUsage {
     Store(StoreSpaceUsage),
     /// Some sort of raw byte count
     Basic(ByteCount),
+    ///
+    Unimplemented,
 }
 
 /// Represents combined space usage of an entire searcher and its component segments.
@@ -119,7 +121,7 @@ impl SegmentSpaceUsage {
     /// Clones the underlying data.
     /// Use the components directly if this is somehow in performance critical code.
     pub fn component(&self, component: SegmentComponent) -> ComponentSpaceUsage {
-        use self::ComponentSpaceUsage::*;
+        use self::ComponentSpaceUsage::{Basic, PerField, Store, Unimplemented};
         use crate::SegmentComponent::*;
         match component {
             POSTINGS => PerField(self.postings().clone()),
@@ -130,6 +132,7 @@ impl SegmentSpaceUsage {
             TERMS => PerField(self.termdict().clone()),
             STORE => Store(self.store().clone()),
             DELETE => Basic(self.deletes()),
+            FIELDSTATS => Unimplemented,
         }
     }
 
@@ -336,7 +339,7 @@ mod test {
         let index = Index::create_in_ram(schema.clone());
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(name => 1u64));
             index_writer.add_document(doc!(name => 2u64));
             index_writer.add_document(doc!(name => 10u64));
@@ -374,7 +377,7 @@ mod test {
         let index = Index::create_in_ram(schema.clone());
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(name => "hi"));
             index_writer.add_document(doc!(name => "this is a test"));
             index_writer.add_document(
@@ -414,7 +417,7 @@ mod test {
         let index = Index::create_in_ram(schema.clone());
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(name => "hi"));
             index_writer.add_document(doc!(name => "this is a test"));
             index_writer.add_document(
@@ -453,7 +456,7 @@ mod test {
         let index = Index::create_in_ram(schema.clone());
 
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             index_writer.add_document(doc!(name => 1u64));
             index_writer.add_document(doc!(name => 2u64));
             index_writer.add_document(doc!(name => 3u64));

src/termdict/mod.rs

@@ -44,11 +44,13 @@ mod tests {
 
     const BLOCK_SIZE: usize = 1_500;
 
-    fn make_term_info(val: u64) -> TermInfo {
+    fn make_term_info(term_ord: u64) -> TermInfo {
+        let offset = |term_ord: u64| term_ord * 100 + term_ord * term_ord;
         TermInfo {
-            doc_freq: val as u32,
-            positions_idx: val * 2u64,
-            postings_offset: val * 3u64,
+            doc_freq: term_ord as u32,
+            postings_start_offset: offset(term_ord),
+            postings_end_offset: offset(term_ord + 1),
+            positions_idx: offset(term_ord) * 2u64,
         }
     }
 
@@ -138,7 +140,7 @@ mod tests {
         let text_field = schema_builder.add_text_field("text", TEXT);
         let index = Index::create_in_ram(schema_builder.build());
         {
-            let mut index_writer = index.writer_with_num_threads(1, 3_000_000).unwrap();
+            let mut index_writer = index.writer_for_tests().unwrap();
             {
                 {
                     let mut doc = Document::default();
@@ -208,20 +210,14 @@ mod tests {
     }
 
     #[test]
-    fn test_stream_high_range_prefix_suffix() {
+    fn test_stream_high_range_prefix_suffix() -> std::io::Result<()> {
         let buffer: Vec<u8> = {
             let mut term_dictionary_builder = TermDictionaryBuilder::create(vec![]).unwrap();
             // term requires more than 16bits
-            term_dictionary_builder
-                .insert("abcdefghijklmnopqrstuvwxy", &make_term_info(1))
-                .unwrap();
-            term_dictionary_builder
-                .insert("abcdefghijklmnopqrstuvwxyz", &make_term_info(2))
-                .unwrap();
-            term_dictionary_builder
-                .insert("abr", &make_term_info(2))
-                .unwrap();
-            term_dictionary_builder.finish().unwrap()
+            term_dictionary_builder.insert("abcdefghijklmnopqrstuvwxy", &make_term_info(1))?;
+            term_dictionary_builder.insert("abcdefghijklmnopqrstuvwxyz", &make_term_info(2))?;
+            term_dictionary_builder.insert("abr", &make_term_info(3))?;
+            term_dictionary_builder.finish()?
         };
         let source = ReadOnlySource::from(buffer);
         let term_dictionary: TermDictionary = TermDictionary::from_source(&source);
@@ -229,12 +225,15 @@ mod tests {
         assert!(kv_stream.advance());
         assert_eq!(kv_stream.key(), "abcdefghijklmnopqrstuvwxy".as_bytes());
         assert_eq!(kv_stream.value(), &make_term_info(1));
+        dbg!(make_term_info(1));
         assert!(kv_stream.advance());
         assert_eq!(kv_stream.key(), "abcdefghijklmnopqrstuvwxyz".as_bytes());
         assert_eq!(kv_stream.value(), &make_term_info(2));
         assert!(kv_stream.advance());
         assert_eq!(kv_stream.key(), "abr".as_bytes());
+        assert_eq!(kv_stream.value(), &make_term_info(3));
         assert!(!kv_stream.advance());
+        Ok(())
     }
 
     #[test]

src/termdict/term_info_store.rs

@@ -57,21 +57,28 @@ impl TermInfoBlockMeta {
         self.doc_freq_nbits + self.postings_offset_nbits + self.positions_idx_nbits
     }
 
+    // Here inner_offset is the offset within the block, WITHOUT the first term_info.
+    // In other word, term_info #1,#2,#3 gets inner_offset 0,1,2... While term_info #0
+    // is encoded without bitpacking.
     fn deserialize_term_info(&self, data: &[u8], inner_offset: usize) -> TermInfo {
+        assert!(inner_offset < BLOCK_LEN - 1);
         let num_bits = self.num_bits() as usize;
         let mut cursor = num_bits * inner_offset;
 
+        let postings_start_offset = extract_bits(data, cursor, self.postings_offset_nbits);
+        let postings_end_offset = self.ref_term_info.postings_start_offset
+            + extract_bits(data, cursor + num_bits, self.postings_offset_nbits);
+        cursor += self.postings_offset_nbits as usize;
+
         let doc_freq = extract_bits(data, cursor, self.doc_freq_nbits) as u32;
         cursor += self.doc_freq_nbits as usize;
 
-        let postings_offset = extract_bits(data, cursor, self.postings_offset_nbits);
-        cursor += self.postings_offset_nbits as usize;
-
         let positions_idx = extract_bits(data, cursor, self.positions_idx_nbits);
 
         TermInfo {
             doc_freq,
-            postings_offset: postings_offset + self.ref_term_info.postings_offset,
+            postings_start_offset: postings_start_offset + self.ref_term_info.postings_start_offset,
+            postings_end_offset,
             positions_idx: positions_idx + self.ref_term_info.positions_idx,
         }
     }
@@ -126,14 +133,13 @@ impl TermInfoStore {
             .expect("Failed to deserialize terminfoblockmeta");
         let inner_offset = (term_ord as usize) % BLOCK_LEN;
         if inner_offset == 0 {
-            term_info_block_data.ref_term_info
-        } else {
-            let term_info_data = self.term_info_source.as_slice();
-            term_info_block_data.deserialize_term_info(
-                &term_info_data[term_info_block_data.offset as usize..],
-                inner_offset - 1,
-            )
+            return term_info_block_data.ref_term_info;
         }
+        let term_info_data = self.term_info_source.as_slice();
+        term_info_block_data.deserialize_term_info(
+            &term_info_data[term_info_block_data.offset as usize..],
+            inner_offset - 1,
+        )
     }
 
     pub fn num_terms(&self) -> usize {
@@ -154,16 +160,17 @@ fn bitpack_serialize<W: Write>(
     term_info_block_meta: &TermInfoBlockMeta,
     term_info: &TermInfo,
 ) -> io::Result<()> {
+    bit_packer.write(
+        term_info.postings_start_offset,
+        term_info_block_meta.postings_offset_nbits,
+        write,
+    )?;
     bit_packer.write(
         u64::from(term_info.doc_freq),
         term_info_block_meta.doc_freq_nbits,
         write,
     )?;
-    bit_packer.write(
-        term_info.postings_offset,
-        term_info_block_meta.postings_offset_nbits,
-        write,
-    )?;
+
     bit_packer.write(
         term_info.positions_idx,
         term_info_block_meta.positions_idx_nbits,
@@ -183,23 +190,27 @@ impl TermInfoStoreWriter {
     }
 
     fn flush_block(&mut self) -> io::Result<()> {
-        if self.term_infos.is_empty() {
-            return Ok(());
-        }
         let mut bit_packer = BitPacker::new();
         let ref_term_info = self.term_infos[0].clone();
+
+        let last_term_info = if let Some(last_term_info) = self.term_infos.last().cloned() {
+            last_term_info
+        } else {
+            return Ok(());
+        };
+        let postings_end_offset =
+            last_term_info.postings_end_offset - ref_term_info.postings_start_offset;
         for term_info in &mut self.term_infos[1..] {
-            term_info.postings_offset -= ref_term_info.postings_offset;
+            term_info.postings_start_offset -= ref_term_info.postings_start_offset;
             term_info.positions_idx -= ref_term_info.positions_idx;
         }
 
         let mut max_doc_freq: u32 = 0u32;
-        let mut max_postings_offset: u64 = 0u64;
-        let mut max_positions_idx: u64 = 0u64;
+        let max_postings_offset: u64 = postings_end_offset;
+        let max_positions_idx: u64 = last_term_info.positions_idx;
+
         for term_info in &self.term_infos[1..] {
             max_doc_freq = cmp::max(max_doc_freq, term_info.doc_freq);
-            max_postings_offset = cmp::max(max_postings_offset, term_info.postings_offset);
-            max_positions_idx = cmp::max(max_positions_idx, term_info.positions_idx);
         }
 
         let max_doc_freq_nbits: u8 = compute_num_bits(u64::from(max_doc_freq));
@@ -224,6 +235,12 @@ impl TermInfoStoreWriter {
             )?;
         }
 
+        bit_packer.write(
+            postings_end_offset,
+            term_info_block_meta.postings_offset_nbits,
+            &mut self.buffer_term_infos,
+        )?;
+
         // Block need end up at the end of a byte.
         bit_packer.flush(&mut self.buffer_term_infos)?;
         self.term_infos.clear();
@@ -232,6 +249,7 @@ impl TermInfoStoreWriter {
     }
 
     pub fn write_term_info(&mut self, term_info: &TermInfo) -> io::Result<()> {
+        assert!(term_info.postings_end_offset >= term_info.postings_start_offset);
         self.num_terms += 1u64;
         self.term_infos.push(term_info.clone());
         if self.term_infos.len() >= BLOCK_LEN {
@@ -291,10 +309,11 @@ mod tests {
     #[test]
     fn test_term_info_block_meta_serialization() {
         let term_info_block_meta = TermInfoBlockMeta {
-            offset: 2009,
+            offset: 2009u64,
             ref_term_info: TermInfo {
                 doc_freq: 512,
-                postings_offset: 51,
+                postings_start_offset: 51,
+                postings_end_offset: 57u64,
                 positions_idx: 3584,
             },
             doc_freq_nbits: 10,
@@ -312,10 +331,12 @@ mod tests {
     fn test_pack() {
         let mut store_writer = TermInfoStoreWriter::new();
         let mut term_infos = vec![];
+        let offset = |i| (i * 13 + i * i) as u64;
         for i in 0..1000 {
             let term_info = TermInfo {
                 doc_freq: i as u32,
-                postings_offset: (i / 10) as u64,
+                postings_start_offset: offset(i),
+                postings_end_offset: offset(i + 1),
                 positions_idx: (i * 7) as u64,
             };
             store_writer.write_term_info(&term_info).unwrap();
@@ -325,7 +346,12 @@ mod tests {
         store_writer.serialize(&mut buffer).unwrap();
         let term_info_store = TermInfoStore::open(&ReadOnlySource::from(buffer));
         for i in 0..1000 {
-            assert_eq!(term_info_store.get(i as u64), term_infos[i]);
+            assert_eq!(
+                term_info_store.get(i as u64),
+                term_infos[i],
+                "term info {}",
+                i
+            );
         }
     }
 }