use github actions for tests

Fix panic in store reader raw document iterator during segment merge

.github/workflows/test.yml

@@ -0,0 +1,24 @@
+name: Rust
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  build:
+
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Build
+      run: cargo build --verbose --workspace
+    - name: Run tests
+      run: cargo test --verbose --workspace
+    - name: Check Formatting
+      run: cargo fmt --all -- --check

CHANGELOG.md

@@ -1,3 +1,7 @@
+Tantivy 0.15.1
+=========================
+- Major bugfix. DocStore panics when first block is deleted. (@appaquet) #1077
+
 Tantivy 0.15.0
 =========================
 - API Changes. Using Range instead of (start, end) in the API and internals (`FileSlice`, `OwnedBytes`, `Snippets`, ...)
@@ -8,11 +12,19 @@ Tantivy 0.15.0
 - Bugfix consistent tie break handling in facet's topk (@hardikpnsp) #357
 - Date field support for range queries (@rihardsk) #516
 - Added lz4-flex as the default compression scheme in tantivy (@PSeitz) #1009
-- Renamed a lot of symbols to avoid all uppercasing on acronyms, as per new clippy recommendation. For instance, RAMDirectory -> RamDirectory. (@pmasurel)
+- Renamed a lot of symbols to avoid all uppercasing on acronyms, as per new clippy recommendation. For instance, RAMDirectory -> RamDirectory. (@fulmicoton)
 - Simplified positions index format (@fulmicoton) #1022
 - Moved bitpacking to bitpacker subcrate and add BlockedBitpacker, which bitpacks blocks of 128 elements (@PSeitz) #1030
 - Added support for more-like-this query in tantivy (@evanxg852000) #1011
-- Added support for sorting an index, e.g presorting documents in an index by a timestamp field. This can heavily improve performance for certain scenarios, by utilizing the sorted data (Top-n optimizations). #1026
+- Added support for sorting an index, e.g presorting documents in an index by a timestamp field. This can heavily improve performance for certain scenarios, by utilizing the sorted data (Top-n optimizations)(@PSeitz). #1026
+- Add iterator over documents in doc store (@PSeitz). #1044
+- Fix log merge policy (@PSeitz). #1043
+- Add detection to avoid small doc store blocks on merge (@PSeitz). #1054
+- Make doc store compression dynamic (@PSeitz). #1060
+- Switch to json for footer version handling (@PSeitz). #1060
+- Updated TermMerger implementation to rely on the union feature of the FST (@scampi) #469
+- Add boolean marking whether position is required in the query_terms API call (@fulmicoton). #1070
+
 
 Tantivy 0.14.0
 =========================

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.14.0"
+version = "0.15.1"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -20,8 +20,7 @@ once_cell = "1.7.2"
 regex ={ version = "1.5.4", default-features = false, features = ["std"] }
 tantivy-fst = "0.3"
 memmap = {version = "0.7", optional=true}
-lz4_flex = { version = "0.7.5", default-features = false, features = ["checked-decode"], optional = true }
-lz4 = { version = "1.23.2", optional = true }
+lz4_flex = { version = "0.8.0", default-features = false, features = ["checked-decode"], optional = true }
 brotli = { version = "3.3", optional = true }
 snap = { version = "1.0.5", optional = true }
 tempfile = { version = "3.2", optional = true }
@@ -34,7 +33,7 @@ levenshtein_automata = "0.2"
 uuid = { version = "0.8.2", features = ["v4", "serde"] }
 crossbeam = "0.8"
 futures = { version = "0.3.15", features = ["thread-pool"] }
-tantivy-query-grammar = { version="0.14.0", path="./query-grammar" }
+tantivy-query-grammar = { version="0.15.0", path="./query-grammar" }
 tantivy-bitpacker = { version="0.1", path="./bitpacker" }
 stable_deref_trait = "1.2"
 rust-stemmers = "1.2"
@@ -77,12 +76,13 @@ debug-assertions = true
 overflow-checks = true
 
 [features]
-default = ["mmap", "lz4-block-compression" ]
+default = ["mmap", "lz4-compression" ]
 mmap = ["fs2", "tempfile", "memmap"]
+
 brotli-compression = ["brotli"]
-lz4-compression = ["lz4"]
-lz4-block-compression = ["lz4_flex"]
+lz4-compression = ["lz4_flex"]
 snappy-compression = ["snap"]
+
 failpoints = ["fail/failpoints"]
 unstable = [] # useful for benches.
 wasm-bindgen = ["uuid/wasm-bindgen"]

appveyor.yml

@@ -18,5 +18,6 @@ install:
 build: false
 
 test_script:
-  - REM SET RUST_LOG=tantivy,test & cargo test --all --verbose --no-default-features --features lz4-block-compression --features mmap
+  - REM SET RUST_LOG=tantivy,test & cargo test --all --verbose --no-default-features --features lz4-compression --features mmap
+  - REM SET RUST_LOG=tantivy,test & cargo test test_store --verbose --no-default-features --features lz4-compression --features snappy-compression --features brotli-compression --features mmap
   - REM SET RUST_BACKTRACE=1 & cargo build --examples

bitpacker/Cargo.toml

@@ -2,6 +2,13 @@
 name = "tantivy-bitpacker"
 version = "0.1.0"
 edition = "2018"
+authors = ["Paul Masurel <paul.masurel@gmail.com>"]
+license = "MIT"
+categories = []
+description = """Tantivy-sub crate: bitpacking"""
+repository = "https://github.com/tantivy-search/tantivy"
+keywords = []
+
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 

bitpacker/src/bitpacker.rs

@@ -17,6 +17,7 @@ impl BitPacker {
         }
     }
 
+    #[inline]
     pub fn write<TWrite: io::Write>(
         &mut self,
         val: u64,
@@ -79,6 +80,7 @@ impl BitUnpacker {
         }
     }
 
+    #[inline]
     pub fn get(&self, idx: u64, data: &[u8]) -> u64 {
         if self.num_bits == 0 {
             return 0u64;

bitpacker/src/blocked_bitpacker.rs

@@ -80,6 +80,7 @@ impl BlockedBitpacker {
                 * std::mem::size_of_val(&self.buffer.get(0).cloned().unwrap_or_default())
     }
 
+    #[inline]
     pub fn add(&mut self, val: u64) {
         self.buffer.push(val);
         if self.buffer.len() == BLOCK_SIZE as usize {
@@ -122,6 +123,7 @@ impl BlockedBitpacker {
                 .resize(self.compressed_blocks.len() + 8, 0); // add padding for bitpacker
         }
     }
+    #[inline]
     pub fn get(&self, idx: usize) -> u64 {
         let metadata_pos = idx / BLOCK_SIZE as usize;
         let pos_in_block = idx % BLOCK_SIZE as usize;

examples/custom_collector.rs

@@ -10,7 +10,7 @@
 // ---
 // Importing tantivy...
 use tantivy::collector::{Collector, SegmentCollector};
-use tantivy::fastfield::FastFieldReader;
+use tantivy::fastfield::{DynamicFastFieldReader, FastFieldReader};
 use tantivy::query::QueryParser;
 use tantivy::schema::Field;
 use tantivy::schema::{Schema, FAST, INDEXED, TEXT};
@@ -98,7 +98,7 @@ impl Collector for StatsCollector {
 }
 
 struct StatsSegmentCollector {
-    fast_field_reader: FastFieldReader<u64>,
+    fast_field_reader: DynamicFastFieldReader<u64>,
     stats: Stats,
 }
 

examples/deleting_updating_documents.rs

@@ -90,7 +90,7 @@ fn main() -> tantivy::Result<()> {
 
     let frankenstein_isbn = Term::from_field_text(isbn, "978-9176370711");
 
-    // Oops our frankenstein doc seems mispelled
+    // Oops our frankenstein doc seems misspelled
     let frankenstein_doc_misspelled = extract_doc_given_isbn(&reader, &frankenstein_isbn)?.unwrap();
     assert_eq!(
         schema.to_json(&frankenstein_doc_misspelled),

examples/faceted_search.rs

@@ -92,7 +92,7 @@ fn main() -> tantivy::Result<()> {
 
     // Check the reference doc for different ways to create a `Facet` object.
     {
-        let facet = Facet::from_text("/Felidae/Pantherinae");
+        let facet = Facet::from("/Felidae/Pantherinae");
         let facet_term = Term::from_facet(classification, &facet);
         let facet_term_query = TermQuery::new(facet_term, IndexRecordOption::Basic);
         let mut facet_collector = FacetCollector::for_field(classification);

query-grammar/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy-query-grammar"
-version = "0.14.0"
+version = "0.15.0"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]

src/collector/facet_collector.rs

@@ -539,10 +539,10 @@ mod tests {
         let index = Index::create_in_ram(schema);
         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"),
-            facet_field => Facet::from_text(&"/subjects/A/b"),
-            facet_field => Facet::from_text(&"/subjects/B/b"),
+            facet_field => Facet::from_text(&"/subjects/A/a").unwrap(),
+            facet_field => Facet::from_text(&"/subjects/B/a").unwrap(),
+            facet_field => Facet::from_text(&"/subjects/A/b").unwrap(),
+            facet_field => Facet::from_text(&"/subjects/B/b").unwrap(),
         ));
         index_writer.commit().unwrap();
         let reader = index.reader().unwrap();
@@ -563,16 +563,16 @@ mod tests {
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
         index_writer.add_document(doc!(
-            facet_field => Facet::from_text(&"/A/A"),
+            facet_field => Facet::from_text(&"/A/A").unwrap(),
         ));
         index_writer.add_document(doc!(
-            facet_field => Facet::from_text(&"/A/B"),
+            facet_field => Facet::from_text(&"/A/B").unwrap(),
         ));
         index_writer.add_document(doc!(
-            facet_field => Facet::from_text(&"/A/C/A"),
+            facet_field => Facet::from_text(&"/A/C/A").unwrap(),
         ));
         index_writer.add_document(doc!(
-            facet_field => Facet::from_text(&"/D/C/A"),
+            facet_field => Facet::from_text(&"/D/C/A").unwrap(),
         ));
         index_writer.commit()?;
         let reader = index.reader()?;
@@ -580,7 +580,7 @@ mod tests {
         assert_eq!(searcher.num_docs(), 4);
 
         let count_facet = |facet_str: &str| {
-            let term = Term::from_facet(facet_field, &Facet::from_text(facet_str));
+            let term = Term::from_facet(facet_field, &Facet::from_text(facet_str).unwrap());
             searcher
                 .search(&TermQuery::new(term, IndexRecordOption::Basic), &Count)
                 .unwrap()

src/collector/filter_collector_wrapper.rs

@@ -12,7 +12,7 @@
 use std::marker::PhantomData;
 
 use crate::collector::{Collector, SegmentCollector};
-use crate::fastfield::{FastFieldReader, FastValue};
+use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
 use crate::schema::Field;
 use crate::{Score, SegmentReader, TantivyError};
 
@@ -155,7 +155,7 @@ where
     TPredicate: 'static,
     TPredicateValue: FastValue,
 {
-    fast_field_reader: FastFieldReader<TPredicateValue>,
+    fast_field_reader: DynamicFastFieldReader<TPredicateValue>,
     segment_collector: TSegmentCollector,
     predicate: TPredicate,
     t_predicate_value: PhantomData<TPredicateValue>,

src/collector/histogram_collector.rs

@@ -1,5 +1,5 @@
 use crate::collector::{Collector, SegmentCollector};
-use crate::fastfield::{FastFieldReader, FastValue};
+use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
 use crate::schema::{Field, Type};
 use crate::{DocId, Score};
 use fastdivide::DividerU64;
@@ -84,7 +84,7 @@ impl HistogramComputer {
 }
 pub struct SegmentHistogramCollector {
     histogram_computer: HistogramComputer,
-    ff_reader: FastFieldReader<u64>,
+    ff_reader: DynamicFastFieldReader<u64>,
 }
 
 impl SegmentCollector for SegmentHistogramCollector {

src/collector/tests.rs

@@ -1,6 +1,7 @@
 use super::*;
 use crate::core::SegmentReader;
 use crate::fastfield::BytesFastFieldReader;
+use crate::fastfield::DynamicFastFieldReader;
 use crate::fastfield::FastFieldReader;
 use crate::schema::Field;
 use crate::DocId;
@@ -162,7 +163,7 @@ pub struct FastFieldTestCollector {
 
 pub struct FastFieldSegmentCollector {
     vals: Vec<u64>,
-    reader: FastFieldReader<u64>,
+    reader: DynamicFastFieldReader<u64>,
 }
 
 impl FastFieldTestCollector {

src/collector/top_score_collector.rs

@@ -4,7 +4,7 @@ use crate::collector::tweak_score_top_collector::TweakedScoreTopCollector;
 use crate::collector::{
     CustomScorer, CustomSegmentScorer, ScoreSegmentTweaker, ScoreTweaker, SegmentCollector,
 };
-use crate::fastfield::FastFieldReader;
+use crate::fastfield::{DynamicFastFieldReader, FastFieldReader};
 use crate::query::Weight;
 use crate::schema::Field;
 use crate::DocAddress;
@@ -129,7 +129,7 @@ impl fmt::Debug for TopDocs {
 }
 
 struct ScorerByFastFieldReader {
-    ff_reader: FastFieldReader<u64>,
+    ff_reader: DynamicFastFieldReader<u64>,
 }
 
 impl CustomSegmentScorer<u64> for ScorerByFastFieldReader {
@@ -151,7 +151,7 @@ impl CustomScorer<u64> for ScorerByField {
         // mapping is monotonic, so it is sufficient to compute our top-K docs.
         //
         // The conversion will then happen only on the top-K docs.
-        let ff_reader: FastFieldReader<u64> = segment_reader
+        let ff_reader = segment_reader
             .fast_fields()
             .typed_fast_field_reader(self.field)?;
         Ok(ScorerByFastFieldReader { ff_reader })
@@ -401,6 +401,7 @@ impl TopDocs {
     /// # use tantivy::query::QueryParser;
     /// use tantivy::SegmentReader;
     /// use tantivy::collector::TopDocs;
+    /// use tantivy::fastfield::FastFieldReader;
     /// use tantivy::schema::Field;
     ///
     /// fn create_schema() -> Schema {
@@ -508,6 +509,7 @@ impl TopDocs {
     /// use tantivy::SegmentReader;
     /// use tantivy::collector::TopDocs;
     /// use tantivy::schema::Field;
+    /// use tantivy::fastfield::FastFieldReader;
     ///
     /// # fn create_schema() -> Schema {
     /// #    let mut schema_builder = Schema::builder();

src/common/mod.rs

@@ -8,7 +8,7 @@ pub use self::bitset::BitSet;
 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::serialize::{BinarySerializable, DeserializeFrom, FixedSize};
 pub use self::vint::{
     read_u32_vint, read_u32_vint_no_advance, serialize_vint_u32, write_u32_vint, VInt,
 };

src/common/serialize.rs

@@ -14,6 +14,20 @@ pub trait BinarySerializable: fmt::Debug + Sized {
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self>;
 }
 
+pub trait DeserializeFrom<T: BinarySerializable> {
+    fn deserialize(&mut self) -> io::Result<T>;
+}
+
+/// Implement deserialize from &[u8] for all types which implement BinarySerializable.
+///
+/// TryFrom would actually be preferrable, but not possible because of the orphan
+/// rules (not completely sure if this could be resolved)
+impl<T: BinarySerializable> DeserializeFrom<T> for &[u8] {
+    fn deserialize(&mut self) -> io::Result<T> {
+        T::deserialize(self)
+    }
+}
+
 /// `FixedSize` marks a `BinarySerializable` as
 /// always serializing to the same size.
 pub trait FixedSize: BinarySerializable {
@@ -61,6 +75,11 @@ impl<Left: BinarySerializable, Right: BinarySerializable> BinarySerializable for
         Ok((Left::deserialize(reader)?, Right::deserialize(reader)?))
     }
 }
+impl<Left: BinarySerializable + FixedSize, Right: BinarySerializable + FixedSize> FixedSize
+    for (Left, Right)
+{
+    const SIZE_IN_BYTES: usize = Left::SIZE_IN_BYTES + Right::SIZE_IN_BYTES;
+}
 
 impl BinarySerializable for u32 {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {

src/core/index.rs

@@ -76,7 +76,7 @@ fn load_metas(
 /// );
 ///
 /// let schema = schema_builder.build();
-/// let settings = IndexSettings{sort_by_field: Some(IndexSortByField{field:"number".to_string(), order:Order::Asc})};
+/// let settings = IndexSettings{sort_by_field: Some(IndexSortByField{field:"number".to_string(), order:Order::Asc}), ..Default::default()};
 /// let index = Index::builder().schema(schema).settings(settings).create_in_ram();
 ///
 /// ```
@@ -173,7 +173,7 @@ impl IndexBuilder {
             &directory,
         )?;
         let mut metas = IndexMeta::with_schema(self.get_expect_schema()?);
-        metas.index_settings = self.index_settings.clone();
+        metas.index_settings = self.index_settings;
         let index = Index::open_from_metas(directory, &metas, SegmentMetaInventory::default());
         Ok(index)
     }
@@ -460,6 +460,13 @@ impl Index {
     pub fn settings(&self) -> &IndexSettings {
         &self.settings
     }
+
+    /// Accessor to the index settings
+    ///
+    pub fn settings_mut(&mut self) -> &mut IndexSettings {
+        &mut self.settings
+    }
+
     /// Accessor to the index schema
     ///
     /// The schema is actually cloned.

src/core/index_meta.rs

@@ -1,7 +1,7 @@
 use super::SegmentComponent;
-use crate::core::SegmentId;
 use crate::schema::Schema;
 use crate::Opstamp;
+use crate::{core::SegmentId, store::Compressor};
 use census::{Inventory, TrackedObject};
 use serde::{Deserialize, Serialize};
 use std::path::PathBuf;
@@ -233,7 +233,11 @@ impl InnerSegmentMeta {
 pub struct IndexSettings {
     /// Sorts the documents by information
     /// provided in `IndexSortByField`
+    #[serde(skip_serializing_if = "Option::is_none")]
     pub sort_by_field: Option<IndexSortByField>,
+    /// The `Compressor` used to compress the doc store.
+    #[serde(default)]
+    pub docstore_compression: Compressor,
 }
 /// Settings to presort the documents in an index
 ///
@@ -255,6 +259,17 @@ pub enum Order {
     /// Descending Order
     Desc,
 }
+impl Order {
+    /// return if the Order is ascending
+    pub fn is_asc(&self) -> bool {
+        self == &Order::Asc
+    }
+    /// return if the Order is descending
+    pub fn is_desc(&self) -> bool {
+        self == &Order::Desc
+    }
+}
+
 /// Meta information about the `Index`.
 ///
 /// This object is serialized on disk in the `meta.json` file.
@@ -369,6 +384,7 @@ mod tests {
                     field: "text".to_string(),
                     order: Order::Asc,
                 }),
+                ..Default::default()
             },
             segments: Vec::new(),
             schema,
@@ -378,7 +394,7 @@ mod tests {
         let json = serde_json::ser::to_string(&index_metas).expect("serialization failed");
         assert_eq!(
             json,
-            r#"{"index_settings":{"sort_by_field":{"field":"text","order":"Asc"}},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","tokenizer":"default"},"stored":false}}],"opstamp":0}"#
+            r#"{"index_settings":{"sort_by_field":{"field":"text","order":"Asc"},"docstore_compression":"lz4"},"segments":[],"schema":[{"name":"text","type":"text","options":{"indexing":{"record":"position","tokenizer":"default"},"stored":false}}],"opstamp":0}"#
         );
     }
 }

src/directory/file_slice.rs

@@ -147,6 +147,13 @@ impl FileSlice {
         self.slice(from_offset..self.len())
     }
 
+    /// Returns a slice from the end.
+    ///
+    /// Equivalent to `.slice(self.len() - from_offset, self.len())`
+    pub fn slice_from_end(&self, from_offset: usize) -> FileSlice {
+        self.slice(self.len() - from_offset..self.len())
+    }
+
     /// Like `.slice(...)` but enforcing only the `to`
     /// boundary.
     ///

src/directory/footer.rs

@@ -1,69 +1,45 @@
-use crate::common::{BinarySerializable, CountingWriter, FixedSize, HasLen, VInt};
 use crate::directory::error::Incompatibility;
 use crate::directory::FileSlice;
-use crate::directory::{AntiCallToken, TerminatingWrite};
-use crate::Version;
+use crate::{
+    common::{BinarySerializable, CountingWriter, DeserializeFrom, FixedSize, HasLen},
+    directory::{AntiCallToken, TerminatingWrite},
+    Version, INDEX_FORMAT_VERSION,
+};
 use crc32fast::Hasher;
+use serde::{Deserialize, Serialize};
 use std::io;
 use std::io::Write;
 
-const FOOTER_MAX_LEN: usize = 10_000;
+const FOOTER_MAX_LEN: u32 = 50_000;
+
+/// The magic byte of the footer to identify corruption
+/// or an old version of the footer.
+const FOOTER_MAGIC_NUMBER: u32 = 1337;
 
 type CrcHashU32 = u32;
 
-#[derive(Debug, Clone, PartialEq)]
+/// A Footer is appended to every file
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
 pub struct Footer {
     pub version: Version,
-    pub meta: String,
-    pub versioned_footer: VersionedFooter,
-}
-
-/// Serialises the footer to a byte-array
-/// - versioned_footer_len : 4 bytes
-///-  versioned_footer: variable bytes
-/// - meta_len: 4 bytes
-/// - meta: variable bytes
-/// - version_len: 4 bytes
-/// - version json: variable bytes
-impl BinarySerializable for Footer {
-    fn serialize<W: io::Write>(&self, writer: &mut W) -> io::Result<()> {
-        BinarySerializable::serialize(&self.versioned_footer, writer)?;
-        BinarySerializable::serialize(&self.meta, writer)?;
-        let version_string =
-            serde_json::to_string(&self.version).map_err(|_err| io::ErrorKind::InvalidInput)?;
-        BinarySerializable::serialize(&version_string, writer)?;
-        Ok(())
-    }
-
-    fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
-        let versioned_footer = VersionedFooter::deserialize(reader)?;
-        let meta = String::deserialize(reader)?;
-        let version_json = String::deserialize(reader)?;
-        let version = serde_json::from_str(&version_json)?;
-        Ok(Footer {
-            version,
-            meta,
-            versioned_footer,
-        })
-    }
+    pub crc: CrcHashU32,
 }
 
 impl Footer {
-    pub fn new(versioned_footer: VersionedFooter) -> Self {
+    pub fn new(crc: CrcHashU32) -> Self {
         let version = crate::VERSION.clone();
-        let meta = version.to_string();
-        Footer {
-            version,
-            meta,
-            versioned_footer,
-        }
+        Footer { version, crc }
     }
 
+    pub fn crc(&self) -> CrcHashU32 {
+        self.crc
+    }
     pub fn append_footer<W: io::Write>(&self, mut write: &mut W) -> io::Result<()> {
         let mut counting_write = CountingWriter::wrap(&mut write);
-        self.serialize(&mut counting_write)?;
-        let written_len = counting_write.written_bytes();
-        (written_len as u32).serialize(write)?;
+        counting_write.write_all(serde_json::to_string(&self)?.as_ref())?;
+        let footer_payload_len = counting_write.written_bytes();
+        BinarySerializable::serialize(&(footer_payload_len as u32), write)?;
+        BinarySerializable::serialize(&(FOOTER_MAGIC_NUMBER as u32), write)?;
         Ok(())
     }
 
@@ -77,12 +53,47 @@ impl Footer {
                 ),
             ));
         }
-        let (body_footer, footer_len_file) = file.split_from_end(u32::SIZE_IN_BYTES);
-        let mut footer_len_bytes = footer_len_file.read_bytes()?;
-        let footer_len = u32::deserialize(&mut footer_len_bytes)? as usize;
-        let (body, footer) = body_footer.split_from_end(footer_len);
-        let mut footer_bytes = footer.read_bytes()?;
-        let footer = Footer::deserialize(&mut footer_bytes)?;
+
+        let footer_metadata_len = <(u32, u32)>::SIZE_IN_BYTES;
+        let (footer_len, footer_magic_byte): (u32, u32) = file
+            .slice_from_end(footer_metadata_len)
+            .read_bytes()?
+            .as_ref()
+            .deserialize()?;
+
+        if footer_magic_byte != FOOTER_MAGIC_NUMBER {
+            return Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                    "Footer magic byte mismatch. File corrupted or index was created using old an tantivy version which is not supported anymore. Please use tantivy 0.15 or above to recreate the index.",
+            ));
+        }
+
+        if footer_len > FOOTER_MAX_LEN {
+            return Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                format!(
+                    "Footer seems invalid as it suggests a footer len of {}. File is corrupted, \
+            or the index was created with a different & old version of tantivy.",
+                    footer_len
+                ),
+            ));
+        }
+        let total_footer_size = footer_len as usize + footer_metadata_len;
+        if file.len() < total_footer_size {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                format!(
+                    "File corrupted. The file is smaller than it's footer bytes (len={}).",
+                    total_footer_size
+                ),
+            ));
+        }
+
+        let footer: Footer = serde_json::from_slice(&file.read_bytes_slice(
+            file.len() - total_footer_size..file.len() - footer_metadata_len as usize,
+        )?)?;
+
+        let body = file.slice_to(file.len() - total_footer_size);
         Ok((footer, body))
     }
 
@@ -90,151 +101,16 @@ impl Footer {
     /// Has to be called after `extract_footer` to make sure it's not accessing uninitialised memory
     pub fn is_compatible(&self) -> Result<(), Incompatibility> {
         let library_version = crate::version();
-        match &self.versioned_footer {
-            VersionedFooter::V1 {
-                crc32: _crc,
-                store_compression,
-            } => {
-                if &library_version.store_compression != store_compression {
-                    return Err(Incompatibility::CompressionMismatch {
-                        library_compression_format: library_version.store_compression.to_string(),
-                        index_compression_format: store_compression.to_string(),
-                    });
-                }
-                Ok(())
-            }
-            VersionedFooter::V2 {
-                crc32: _crc,
-                store_compression,
-            } => {
-                if &library_version.store_compression != store_compression {
-                    return Err(Incompatibility::CompressionMismatch {
-                        library_compression_format: library_version.store_compression.to_string(),
-                        index_compression_format: store_compression.to_string(),
-                    });
-                }
-                Ok(())
-            }
-            VersionedFooter::V3 {
-                crc32: _crc,
-                store_compression,
-            } => {
-                if &library_version.store_compression != store_compression {
-                    return Err(Incompatibility::CompressionMismatch {
-                        library_compression_format: library_version.store_compression.to_string(),
-                        index_compression_format: store_compression.to_string(),
-                    });
-                }
-                Ok(())
-            }
-            VersionedFooter::UnknownVersion => Err(Incompatibility::IndexMismatch {
+        if self.version.index_format_version < 4
+            || self.version.index_format_version > INDEX_FORMAT_VERSION
+        {
+            return Err(Incompatibility::IndexMismatch {
                 library_version: library_version.clone(),
                 index_version: self.version.clone(),
-            }),
+            });
         }
-    }
-}
-
-/// Footer that includes a crc32 hash that enables us to checksum files in the index
-#[derive(Debug, Clone, PartialEq)]
-pub enum VersionedFooter {
-    UnknownVersion,
-    V1 {
-        crc32: CrcHashU32,
-        store_compression: String,
-    },
-    // Introduction of the Block WAND information.
-    V2 {
-        crc32: CrcHashU32,
-        store_compression: String,
-    },
-    // Block wand max termfred on 1 byte
-    V3 {
-        crc32: CrcHashU32,
-        store_compression: String,
-    },
-}
-
-impl BinarySerializable for VersionedFooter {
-    fn serialize<W: io::Write>(&self, writer: &mut W) -> io::Result<()> {
-        let mut buf = Vec::new();
-        match self {
-            VersionedFooter::V3 {
-                crc32,
-                store_compression: compression,
-            } => {
-                // Serializes a valid `VersionedFooter` or panics if the version is unknown
-                // [   version    |   crc_hash  | compression_mode ]
-                // [    0..4      |     4..8    |     variable     ]
-                BinarySerializable::serialize(&3u32, &mut buf)?;
-                BinarySerializable::serialize(crc32, &mut buf)?;
-                BinarySerializable::serialize(compression, &mut buf)?;
-            }
-            VersionedFooter::V2 { .. }
-            | VersionedFooter::V1 { .. }
-            | VersionedFooter::UnknownVersion => {
-                return Err(io::Error::new(
-                    io::ErrorKind::InvalidInput,
-                    "Cannot serialize an unknown versioned footer ",
-                ));
-            }
-        }
-        BinarySerializable::serialize(&VInt(buf.len() as u64), writer)?;
-        assert!(buf.len() <= FOOTER_MAX_LEN);
-        writer.write_all(&buf[..])?;
         Ok(())
     }
-
-    fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
-        let len = VInt::deserialize(reader)?.0 as usize;
-        if len > FOOTER_MAX_LEN {
-            return Err(io::Error::new(
-                io::ErrorKind::InvalidData,
-                format!(
-                    "Footer seems invalid as it suggests a footer len of {}. File is corrupted, \
-            or the index was created with a different & old version of tantivy.",
-                    len
-                ),
-            ));
-        }
-        let mut buf = vec![0u8; len];
-        reader.read_exact(&mut buf[..])?;
-        let mut cursor = &buf[..];
-        let version = u32::deserialize(&mut cursor)?;
-        if version > 3 {
-            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 if version == 2 {
-            VersionedFooter::V2 {
-                crc32,
-                store_compression,
-            }
-        } else {
-            assert_eq!(version, 3);
-            VersionedFooter::V3 {
-                crc32,
-                store_compression,
-            }
-        })
-    }
-}
-
-impl VersionedFooter {
-    pub fn crc(&self) -> Option<CrcHashU32> {
-        match self {
-            VersionedFooter::V3 { crc32, .. } => Some(*crc32),
-            VersionedFooter::V2 { crc32, .. } => Some(*crc32),
-            VersionedFooter::V1 { crc32, .. } => Some(*crc32),
-            VersionedFooter::UnknownVersion { .. } => None,
-        }
-    }
 }
 
 pub(crate) struct FooterProxy<W: TerminatingWrite> {
@@ -268,10 +144,7 @@ impl<W: TerminatingWrite> Write for FooterProxy<W> {
 impl<W: TerminatingWrite> TerminatingWrite for FooterProxy<W> {
     fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()> {
         let crc32 = self.hasher.take().unwrap().finalize();
-        let footer = Footer::new(VersionedFooter::V3 {
-            crc32,
-            store_compression: crate::store::COMPRESSION.to_string(),
-        });
+        let footer = Footer::new(crc32);
         let mut writer = self.writer.take().unwrap();
         footer.append_footer(&mut writer)?;
         writer.terminate()
@@ -281,140 +154,75 @@ impl<W: TerminatingWrite> TerminatingWrite for FooterProxy<W> {
 #[cfg(test)]
 mod tests {
 
-    use super::CrcHashU32;
-    use super::FooterProxy;
-    use crate::common::{BinarySerializable, VInt};
-    use crate::directory::footer::{Footer, VersionedFooter};
-    use crate::directory::TerminatingWrite;
-    use byteorder::{ByteOrder, LittleEndian};
-    use regex::Regex;
+    use crate::directory::footer::Footer;
+    use crate::directory::OwnedBytes;
+    use crate::{
+        common::BinarySerializable,
+        directory::{footer::FOOTER_MAGIC_NUMBER, FileSlice},
+    };
     use std::io;
 
     #[test]
-    fn test_versioned_footer() {
-        let mut vec = Vec::new();
-        let footer_proxy = FooterProxy::new(&mut vec);
-        assert!(footer_proxy.terminate().is_ok());
-        if crate::store::COMPRESSION == "lz4" {
-            assert_eq!(vec.len(), 158);
-        } else if crate::store::COMPRESSION == "snappy" {
-            assert_eq!(vec.len(), 167);
-        } else if crate::store::COMPRESSION == "lz4_block" {
-            assert_eq!(vec.len(), 176);
-        }
-        let footer = Footer::deserialize(&mut &vec[..]).unwrap();
-        assert!(matches!(
-           footer.versioned_footer,
-           VersionedFooter::V3 { store_compression, .. }
-           if store_compression == crate::store::COMPRESSION
-        ));
-        assert_eq!(&footer.version, crate::version());
+    fn test_deserialize_footer() {
+        let mut buf: Vec<u8> = vec![];
+        let footer = Footer::new(123);
+        footer.append_footer(&mut buf).unwrap();
+        let owned_bytes = OwnedBytes::new(buf);
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let (footer_deser, _body) = Footer::extract_footer(fileslice).unwrap();
+        assert_eq!(footer_deser.crc(), footer.crc());
     }
-
     #[test]
-    fn test_serialize_deserialize_footer() {
-        let mut buffer = Vec::new();
-        let crc32 = 123456u32;
-        let footer: Footer = Footer::new(VersionedFooter::V3 {
-            crc32,
-            store_compression: "lz4".to_string(),
-        });
-        footer.serialize(&mut buffer).unwrap();
-        let footer_deser = Footer::deserialize(&mut &buffer[..]).unwrap();
-        assert_eq!(footer_deser, footer);
+    fn test_deserialize_footer_missing_magic_byte() {
+        let mut buf: Vec<u8> = vec![];
+        BinarySerializable::serialize(&0_u32, &mut buf).unwrap();
+        let wrong_magic_byte: u32 = 5555;
+        BinarySerializable::serialize(&wrong_magic_byte, &mut buf).unwrap();
+
+        let owned_bytes = OwnedBytes::new(buf);
+
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let err = Footer::extract_footer(fileslice).unwrap_err();
+        assert_eq!(
+            err.to_string(),
+            "Footer magic byte mismatch. File corrupted or index was created using old an tantivy version which \
+            is not supported anymore. Please use tantivy 0.15 or above to recreate the index."
+        );
     }
-
     #[test]
-    fn footer_length() {
-        let crc32 = 1111111u32;
-        let versioned_footer = VersionedFooter::V3 {
-            crc32,
-            store_compression: "lz4".to_string(),
-        };
-        let mut buf = Vec::new();
-        versioned_footer.serialize(&mut buf).unwrap();
-        assert_eq!(buf.len(), 13);
-        let footer = Footer::new(versioned_footer);
-        let regex_ptn = Regex::new(
-            "tantivy v[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.{0,10}, index_format v[0-9]{1,5}",
-        )
-        .unwrap();
-        assert!(regex_ptn.is_match(&footer.meta));
-    }
+    fn test_deserialize_footer_wrong_filesize() {
+        let mut buf: Vec<u8> = vec![];
+        BinarySerializable::serialize(&100_u32, &mut buf).unwrap();
+        BinarySerializable::serialize(&FOOTER_MAGIC_NUMBER, &mut buf).unwrap();
 
-    #[test]
-    fn versioned_footer_from_bytes() {
-        let v_footer_bytes = vec![
-            // versionned footer length
-            12 | 128,
-            // index format version
-            3,
-            0,
-            0,
-            0,
-            // crc 32
-            12,
-            35,
-            89,
-            18,
-            // compression format
-            3 | 128,
-            b'l',
-            b'z',
-            b'4',
-        ];
-        let mut cursor = &v_footer_bytes[..];
-        let versioned_footer = VersionedFooter::deserialize(&mut cursor).unwrap();
-        assert!(cursor.is_empty());
-        let expected_crc: u32 = LittleEndian::read_u32(&v_footer_bytes[5..9]) as CrcHashU32;
-        let expected_versioned_footer: VersionedFooter = VersionedFooter::V3 {
-            crc32: expected_crc,
-            store_compression: "lz4".to_string(),
-        };
-        assert_eq!(versioned_footer, expected_versioned_footer);
-        let mut buffer = Vec::new();
-        assert!(versioned_footer.serialize(&mut buffer).is_ok());
-        assert_eq!(&v_footer_bytes[..], &buffer[..]);
-    }
+        let owned_bytes = OwnedBytes::new(buf);
 
-    #[test]
-    fn versioned_footer_panic() {
-        let v_footer_bytes = vec![6u8 | 128u8, 3u8, 0u8, 0u8, 1u8, 0u8, 0u8];
-        let mut b = &v_footer_bytes[..];
-        let versioned_footer = VersionedFooter::deserialize(&mut b).unwrap();
-        assert!(b.is_empty());
-        let expected_versioned_footer = VersionedFooter::UnknownVersion;
-        assert_eq!(versioned_footer, expected_versioned_footer);
-        let mut buf = Vec::new();
-        assert!(versioned_footer.serialize(&mut buf).is_err());
-    }
-
-    #[test]
-    #[cfg(not(feature = "lz4"))]
-    fn compression_mismatch() {
-        let crc32 = 1111111u32;
-        let versioned_footer = VersionedFooter::V1 {
-            crc32,
-            store_compression: "lz4".to_string(),
-        };
-        let footer = Footer::new(versioned_footer);
-        let res = footer.is_compatible();
-        assert!(res.is_err());
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let err = Footer::extract_footer(fileslice).unwrap_err();
+        assert_eq!(err.kind(), io::ErrorKind::UnexpectedEof);
+        assert_eq!(
+            err.to_string(),
+            "File corrupted. The file is smaller than it\'s footer bytes (len=108)."
+        );
     }
 
     #[test]
     fn test_deserialize_too_large_footer() {
-        let mut buf = vec![];
-        assert!(FooterProxy::new(&mut buf).terminate().is_ok());
-        let mut long_len_buf = [0u8; 10];
-        let num_bytes = VInt(super::FOOTER_MAX_LEN as u64 + 1u64).serialize_into(&mut long_len_buf);
-        buf[0..num_bytes].copy_from_slice(&long_len_buf[..num_bytes]);
-        let err = Footer::deserialize(&mut &buf[..]).unwrap_err();
+        let mut buf: Vec<u8> = vec![];
+
+        let footer_length = super::FOOTER_MAX_LEN + 1;
+        BinarySerializable::serialize(&footer_length, &mut buf).unwrap();
+        BinarySerializable::serialize(&FOOTER_MAGIC_NUMBER, &mut buf).unwrap();
+
+        let owned_bytes = OwnedBytes::new(buf);
+
+        let fileslice = FileSlice::new(Box::new(owned_bytes));
+        let err = Footer::extract_footer(fileslice).unwrap_err();
         assert_eq!(err.kind(), io::ErrorKind::InvalidData);
         assert_eq!(
             err.to_string(),
-            "Footer seems invalid as it suggests a footer len of 10001. File is corrupted, \
-            or the index was created with a different & old version of tantivy."
+            "Footer seems invalid as it suggests a footer len of 50001. File is corrupted, \
+    or the index was created with a different & old version of tantivy."
         );
     }
 }

src/directory/managed_directory.rs

@@ -245,11 +245,7 @@ impl ManagedDirectory {
         let mut hasher = Hasher::new();
         hasher.update(bytes.as_slice());
         let crc = hasher.finalize();
-        Ok(footer
-            .versioned_footer
-            .crc()
-            .map(|v| v == crc)
-            .unwrap_or(false))
+        Ok(footer.crc() == crc)
     }
 
     /// List files for which checksum does not match content

src/directory/mmap_directory.rs

@@ -593,7 +593,7 @@ mod tests {
 
             let mut index_writer = index.writer_for_tests().unwrap();
             let mut log_merge_policy = LogMergePolicy::default();
-            log_merge_policy.set_min_merge_size(3);
+            log_merge_policy.set_min_num_segments(3);
             index_writer.set_merge_policy(Box::new(log_merge_policy));
             for _num_commits in 0..10 {
                 for _ in 0..10 {

src/fastfield/bytes/reader.rs

@@ -1,7 +1,7 @@
+use crate::directory::FileSlice;
 use crate::directory::OwnedBytes;
-use crate::fastfield::FastFieldReader;
+use crate::fastfield::{BitpackedFastFieldReader, FastFieldReader, MultiValueLength};
 use crate::DocId;
-use crate::{directory::FileSlice, fastfield::MultiValueLength};
 
 /// Reader for byte array fast fields
 ///
@@ -15,13 +15,13 @@ use crate::{directory::FileSlice, fastfield::MultiValueLength};
 /// and the start index for the next document, and keeping the bytes in between.
 #[derive(Clone)]
 pub struct BytesFastFieldReader {
-    idx_reader: FastFieldReader<u64>,
+    idx_reader: BitpackedFastFieldReader<u64>,
     values: OwnedBytes,
 }
 
 impl BytesFastFieldReader {
     pub(crate) fn open(
-        idx_reader: FastFieldReader<u64>,
+        idx_reader: BitpackedFastFieldReader<u64>,
         values_file: FileSlice,
     ) -> crate::Result<BytesFastFieldReader> {
         let values = values_file.read_bytes()?;

src/fastfield/bytes/writer.rs

@@ -1,8 +1,11 @@
 use std::io;
 
+use crate::fastfield::serializer::FastFieldSerializer;
 use crate::schema::{Document, Field, Value};
 use crate::DocId;
-use crate::{fastfield::serializer::FastFieldSerializer, indexer::doc_id_mapping::DocIdMapping};
+use crate::{
+    fastfield::serializer::CompositeFastFieldSerializer, indexer::doc_id_mapping::DocIdMapping,
+};
 
 /// Writer for byte array (as in, any number of bytes per document) fast fields
 ///
@@ -104,7 +107,7 @@ impl BytesFastFieldWriter {
     /// Serializes the fast field values by pushing them to the `FastFieldSerializer`.
     pub fn serialize(
         &self,
-        serializer: &mut FastFieldSerializer,
+        serializer: &mut CompositeFastFieldSerializer,
         doc_id_map: Option<&DocIdMapping>,
     ) -> io::Result<()> {
         // writing the offset index

src/fastfield/facet_reader.rs

@@ -95,7 +95,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
-        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b")));
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
         index_writer.commit()?;
         let searcher = index.reader()?.searcher();
         let facet_reader = searcher
@@ -118,7 +118,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
-        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b")));
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
         index_writer.commit()?;
         let searcher = index.reader()?.searcher();
         let facet_reader = searcher
@@ -141,7 +141,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
-        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b")));
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
         index_writer.commit()?;
         let searcher = index.reader()?.searcher();
         let facet_reader = searcher
@@ -164,7 +164,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
-        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b")));
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
         index_writer.commit()?;
         let searcher = index.reader()?.searcher();
         let facet_reader = searcher
@@ -187,7 +187,7 @@ mod tests {
         let schema = schema_builder.build();
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
-        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b")));
+        index_writer.add_document(doc!(facet_field=>Facet::from_text("/a/b").unwrap()));
         index_writer.add_document(Document::default());
         index_writer.commit()?;
         let searcher = index.reader()?.searcher();

src/fastfield/mod.rs

@@ -29,8 +29,11 @@ pub use self::delete::DeleteBitSet;
 pub use self::error::{FastFieldNotAvailableError, Result};
 pub use self::facet_reader::FacetReader;
 pub use self::multivalued::{MultiValuedFastFieldReader, MultiValuedFastFieldWriter};
+pub use self::reader::BitpackedFastFieldReader;
+pub use self::reader::DynamicFastFieldReader;
 pub use self::reader::FastFieldReader;
 pub use self::readers::FastFieldReaders;
+pub use self::serializer::CompositeFastFieldSerializer;
 pub use self::serializer::FastFieldSerializer;
 pub use self::writer::{FastFieldsWriter, IntFastFieldWriter};
 use crate::schema::Cardinality;
@@ -57,7 +60,7 @@ mod writer;
 pub trait MultiValueLength {
     /// returns the num of values associated to a doc_id
     fn get_len(&self, doc_id: DocId) -> u64;
-    /// returns the sum of num of all values for all doc_ids
+    /// returns the sum of num values for all doc_ids
     fn get_total_len(&self) -> u64;
 }
 
@@ -211,7 +214,7 @@ mod tests {
     use super::*;
     use crate::common::CompositeFile;
     use crate::directory::{Directory, RamDirectory, WritePtr};
-    use crate::fastfield::FastFieldReader;
+    use crate::fastfield::BitpackedFastFieldReader;
     use crate::merge_policy::NoMergePolicy;
     use crate::schema::Field;
     use crate::schema::Schema;
@@ -236,7 +239,7 @@ mod tests {
 
     #[test]
     pub fn test_fastfield() {
-        let test_fastfield = FastFieldReader::<u64>::from(vec![100, 200, 300]);
+        let test_fastfield = BitpackedFastFieldReader::<u64>::from(vec![100, 200, 300]);
         assert_eq!(test_fastfield.get(0), 100);
         assert_eq!(test_fastfield.get(1), 200);
         assert_eq!(test_fastfield.get(2), 300);
@@ -254,7 +257,7 @@ mod tests {
         let directory: RamDirectory = RamDirectory::create();
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             fast_field_writers.add_document(&doc!(*FIELD=>13u64));
             fast_field_writers.add_document(&doc!(*FIELD=>14u64));
@@ -268,7 +271,7 @@ mod tests {
         assert_eq!(file.len(), 36 as usize);
         let composite_file = CompositeFile::open(&file)?;
         let file = composite_file.open_read(*FIELD).unwrap();
-        let fast_field_reader = FastFieldReader::<u64>::open(file)?;
+        let fast_field_reader = BitpackedFastFieldReader::<u64>::open(file)?;
         assert_eq!(fast_field_reader.get(0), 13u64);
         assert_eq!(fast_field_reader.get(1), 14u64);
         assert_eq!(fast_field_reader.get(2), 2u64);
@@ -281,7 +284,7 @@ mod tests {
         let directory: RamDirectory = RamDirectory::create();
         {
             let write: WritePtr = directory.open_write(Path::new("test"))?;
-            let mut serializer = FastFieldSerializer::from_write(write)?;
+            let mut serializer = CompositeFastFieldSerializer::from_write(write)?;
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             fast_field_writers.add_document(&doc!(*FIELD=>4u64));
             fast_field_writers.add_document(&doc!(*FIELD=>14_082_001u64));
@@ -300,7 +303,7 @@ mod tests {
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
             assert_eq!(fast_field_reader.get(0), 4u64);
             assert_eq!(fast_field_reader.get(1), 14_082_001u64);
             assert_eq!(fast_field_reader.get(2), 3_052u64);
@@ -321,7 +324,7 @@ mod tests {
 
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             for _ in 0..10_000 {
                 fast_field_writers.add_document(&doc!(*FIELD=>100_000u64));
@@ -336,7 +339,7 @@ mod tests {
         {
             let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
             for doc in 0..10_000 {
                 assert_eq!(fast_field_reader.get(doc), 100_000u64);
             }
@@ -351,7 +354,7 @@ mod tests {
 
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             // forcing the amplitude to be high
             fast_field_writers.add_document(&doc!(*FIELD=>0u64));
@@ -368,7 +371,7 @@ mod tests {
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
             assert_eq!(fast_field_reader.get(0), 0u64);
             for doc in 1..10_001 {
                 assert_eq!(
@@ -390,7 +393,7 @@ mod tests {
         let schema = schema_builder.build();
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
             for i in -100i64..10_000i64 {
                 let mut doc = Document::default();
@@ -407,7 +410,7 @@ mod tests {
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(i64_field).unwrap();
-            let fast_field_reader = FastFieldReader::<i64>::open(data)?;
+            let fast_field_reader = BitpackedFastFieldReader::<i64>::open(data)?;
 
             assert_eq!(fast_field_reader.min_value(), -100i64);
             assert_eq!(fast_field_reader.max_value(), 9_999i64);
@@ -433,7 +436,7 @@ mod tests {
 
         {
             let write: WritePtr = directory.open_write(Path::new("test")).unwrap();
-            let mut serializer = FastFieldSerializer::from_write(write).unwrap();
+            let mut serializer = CompositeFastFieldSerializer::from_write(write).unwrap();
             let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
             let doc = Document::default();
             fast_field_writers.add_document(&doc);
@@ -447,7 +450,7 @@ mod tests {
         {
             let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(i64_field).unwrap();
-            let fast_field_reader = FastFieldReader::<i64>::open(data)?;
+            let fast_field_reader = BitpackedFastFieldReader::<i64>::open(data)?;
             assert_eq!(fast_field_reader.get(0u32), 0i64);
         }
         Ok(())
@@ -468,7 +471,7 @@ mod tests {
         let directory = RamDirectory::create();
         {
             let write: WritePtr = directory.open_write(Path::new("test"))?;
-            let mut serializer = FastFieldSerializer::from_write(write)?;
+            let mut serializer = CompositeFastFieldSerializer::from_write(write)?;
             let mut fast_field_writers = FastFieldsWriter::from_schema(&SCHEMA);
             for &x in &permutation {
                 fast_field_writers.add_document(&doc!(*FIELD=>x));
@@ -480,7 +483,7 @@ mod tests {
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = FastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
 
             let mut a = 0u64;
             for _ in 0..n {

src/fastfield/multivalued/reader.rs

@@ -1,6 +1,6 @@
 use std::ops::Range;
 
-use crate::fastfield::{FastFieldReader, FastValue, MultiValueLength};
+use crate::fastfield::{BitpackedFastFieldReader, FastFieldReader, FastValue, MultiValueLength};
 use crate::DocId;
 
 /// Reader for a multivalued `u64` fast field.
@@ -13,14 +13,14 @@ use crate::DocId;
 ///
 #[derive(Clone)]
 pub struct MultiValuedFastFieldReader<Item: FastValue> {
-    idx_reader: FastFieldReader<u64>,
-    vals_reader: FastFieldReader<Item>,
+    idx_reader: BitpackedFastFieldReader<u64>,
+    vals_reader: BitpackedFastFieldReader<Item>,
 }
 
 impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
     pub(crate) fn open(
-        idx_reader: FastFieldReader<u64>,
-        vals_reader: FastFieldReader<Item>,
+        idx_reader: BitpackedFastFieldReader<u64>,
+        vals_reader: BitpackedFastFieldReader<Item>,
     ) -> MultiValuedFastFieldReader<Item> {
         MultiValuedFastFieldReader {
             idx_reader,

src/fastfield/multivalued/writer.rs

@@ -1,5 +1,6 @@
-use crate::fastfield::serializer::FastSingleFieldSerializer;
-use crate::fastfield::FastFieldSerializer;
+use crate::fastfield::serializer::DynamicFastFieldSerializer;
+use crate::fastfield::serializer::FastFieldSerializer;
+use crate::fastfield::CompositeFastFieldSerializer;
 use crate::postings::UnorderedTermId;
 use crate::schema::{Document, Field};
 use crate::termdict::TermOrdinal;
@@ -134,7 +135,7 @@ impl MultiValuedFastFieldWriter {
     ///
     pub fn serialize(
         &self,
-        serializer: &mut FastFieldSerializer,
+        serializer: &mut CompositeFastFieldSerializer,
         mapping_opt: Option<&FnvHashMap<UnorderedTermId, TermOrdinal>>,
         doc_id_map: Option<&DocIdMapping>,
     ) -> io::Result<()> {
@@ -154,7 +155,7 @@ impl MultiValuedFastFieldWriter {
         }
         {
             // writing the values themselves.
-            let mut value_serializer: FastSingleFieldSerializer<'_, _>;
+            let mut value_serializer: DynamicFastFieldSerializer<'_, _>;
             match mapping_opt {
                 Some(mapping) => {
                     value_serializer = serializer.new_u64_fast_field_with_idx(

src/fastfield/reader.rs

@@ -4,7 +4,7 @@ use crate::common::CompositeFile;
 use crate::directory::FileSlice;
 use crate::directory::OwnedBytes;
 use crate::directory::{Directory, RamDirectory, WritePtr};
-use crate::fastfield::{FastFieldSerializer, FastFieldsWriter};
+use crate::fastfield::{CompositeFastFieldSerializer, FastFieldsWriter};
 use crate::schema::Schema;
 use crate::schema::FAST;
 use crate::DocId;
@@ -14,12 +14,94 @@ use std::path::Path;
 use tantivy_bitpacker::compute_num_bits;
 use tantivy_bitpacker::BitUnpacker;
 
+/// FastFieldReader is the trait to access fast field data.
+pub trait FastFieldReader<Item: FastValue>: Clone {
+    /// Return the value associated to the given document.
+    ///
+    /// This accessor should return as fast as possible.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `doc` is greater than the segment
+    fn get(&self, doc: DocId) -> Item;
+
+    /// Fills an output buffer with the fast field values
+    /// associated with the `DocId` going from
+    /// `start` to `start + output.len()`.
+    ///
+    /// Regardless of the type of `Item`, this method works
+    /// - transmuting the output array
+    /// - extracting the `Item`s as if they were `u64`
+    /// - possibly converting the `u64` value to the right type.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `start + output.len()` is greater than
+    /// the segment's `maxdoc`.
+    fn get_range(&self, start: DocId, output: &mut [Item]);
+
+    /// Returns the minimum value for this fast field.
+    ///
+    /// The max value does not take in account of possible
+    /// deleted document, and should be considered as an upper bound
+    /// of the actual maximum value.
+    fn min_value(&self) -> Item;
+
+    /// Returns the maximum value for this fast field.
+    ///
+    /// The max value does not take in account of possible
+    /// deleted document, and should be considered as an upper bound
+    /// of the actual maximum value.
+    fn max_value(&self) -> Item;
+}
+
+#[derive(Clone)]
+/// DynamicFastFieldReader wraps different readers to access
+/// the various encoded fastfield data
+///
+pub enum DynamicFastFieldReader<Item: FastValue> {
+    /// Bitpacked compressed fastfield data.
+    Bitpacked(BitpackedFastFieldReader<Item>),
+}
+
+impl<Item: FastValue> DynamicFastFieldReader<Item> {
+    /// Returns correct the reader wrapped in the `DynamicFastFieldReader` enum for the data.
+    pub fn open(file: FileSlice) -> crate::Result<DynamicFastFieldReader<Item>> {
+        Ok(DynamicFastFieldReader::Bitpacked(
+            BitpackedFastFieldReader::open(file)?,
+        ))
+    }
+}
+
+impl<Item: FastValue> FastFieldReader<Item> for DynamicFastFieldReader<Item> {
+    fn get(&self, doc: DocId) -> Item {
+        match self {
+            Self::Bitpacked(reader) => reader.get(doc),
+        }
+    }
+    fn get_range(&self, start: DocId, output: &mut [Item]) {
+        match self {
+            Self::Bitpacked(reader) => reader.get_range(start, output),
+        }
+    }
+    fn min_value(&self) -> Item {
+        match self {
+            Self::Bitpacked(reader) => reader.min_value(),
+        }
+    }
+    fn max_value(&self) -> Item {
+        match self {
+            Self::Bitpacked(reader) => reader.max_value(),
+        }
+    }
+}
+
 /// Trait for accessing a fastfield.
 ///
 /// Depending on the field type, a different
 /// fast field is required.
 #[derive(Clone)]
-pub struct FastFieldReader<Item: FastValue> {
+pub struct BitpackedFastFieldReader<Item: FastValue> {
     bytes: OwnedBytes,
     bit_unpacker: BitUnpacker,
     min_value_u64: u64,
@@ -27,7 +109,7 @@ pub struct FastFieldReader<Item: FastValue> {
     _phantom: PhantomData<Item>,
 }
 
-impl<Item: FastValue> FastFieldReader<Item> {
+impl<Item: FastValue> BitpackedFastFieldReader<Item> {
     /// Opens a fast field given a file.
     pub fn open(file: FileSlice) -> crate::Result<Self> {
         let mut bytes = file.read_bytes()?;
@@ -36,7 +118,7 @@ impl<Item: FastValue> FastFieldReader<Item> {
         let max_value = min_value + amplitude;
         let num_bits = compute_num_bits(amplitude);
         let bit_unpacker = BitUnpacker::new(num_bits);
-        Ok(FastFieldReader {
+        Ok(BitpackedFastFieldReader {
             bytes,
             min_value_u64: min_value,
             max_value_u64: max_value,
@@ -44,19 +126,6 @@ impl<Item: FastValue> FastFieldReader<Item> {
             _phantom: PhantomData,
         })
     }
-
-    /// Return the value associated to the given document.
-    ///
-    /// This accessor should return as fast as possible.
-    ///
-    /// # Panics
-    ///
-    /// May panic if `doc` is greater than the segment
-    // `maxdoc`.
-    pub fn get(&self, doc: DocId) -> Item {
-        self.get_u64(u64::from(doc))
-    }
-
     pub(crate) fn get_u64(&self, doc: u64) -> Item {
         Item::from_u64(self.min_value_u64 + self.bit_unpacker.get(doc, &self.bytes))
     }
@@ -78,6 +147,20 @@ impl<Item: FastValue> FastFieldReader<Item> {
             *out = self.get_u64(start + (i as u64));
         }
     }
+}
+
+impl<Item: FastValue> FastFieldReader<Item> for BitpackedFastFieldReader<Item> {
+    /// Return the value associated to the given document.
+    ///
+    /// This accessor should return as fast as possible.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `doc` is greater than the segment
+    // `maxdoc`.
+    fn get(&self, doc: DocId) -> Item {
+        self.get_u64(u64::from(doc))
+    }
 
     /// Fills an output buffer with the fast field values
     /// associated with the `DocId` going from
@@ -92,7 +175,7 @@ impl<Item: FastValue> FastFieldReader<Item> {
     ///
     /// May panic if `start + output.len()` is greater than
     /// the segment's `maxdoc`.
-    pub fn get_range(&self, start: DocId, output: &mut [Item]) {
+    fn get_range(&self, start: DocId, output: &mut [Item]) {
         self.get_range_u64(u64::from(start), output);
     }
 
@@ -101,7 +184,7 @@ impl<Item: FastValue> FastFieldReader<Item> {
     /// The max value does not take in account of possible
     /// deleted document, and should be considered as an upper bound
     /// of the actual maximum value.
-    pub fn min_value(&self) -> Item {
+    fn min_value(&self) -> Item {
         Item::from_u64(self.min_value_u64)
     }
 
@@ -110,13 +193,13 @@ impl<Item: FastValue> FastFieldReader<Item> {
     /// The max value does not take in account of possible
     /// deleted document, and should be considered as an upper bound
     /// of the actual maximum value.
-    pub fn max_value(&self) -> Item {
+    fn max_value(&self) -> Item {
         Item::from_u64(self.max_value_u64)
     }
 }
 
-impl<Item: FastValue> From<Vec<Item>> for FastFieldReader<Item> {
-    fn from(vals: Vec<Item>) -> FastFieldReader<Item> {
+impl<Item: FastValue> From<Vec<Item>> for BitpackedFastFieldReader<Item> {
+    fn from(vals: Vec<Item>) -> BitpackedFastFieldReader<Item> {
         let mut schema_builder = Schema::builder();
         let field = schema_builder.add_u64_field("field", FAST);
         let schema = schema_builder.build();
@@ -126,7 +209,7 @@ impl<Item: FastValue> From<Vec<Item>> for FastFieldReader<Item> {
             let write: WritePtr = directory
                 .open_write(path)
                 .expect("With a RamDirectory, this should never fail.");
-            let mut serializer = FastFieldSerializer::from_write(write)
+            let mut serializer = CompositeFastFieldSerializer::from_write(write)
                 .expect("With a RamDirectory, this should never fail.");
             let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
             {
@@ -148,6 +231,6 @@ impl<Item: FastValue> From<Vec<Item>> for FastFieldReader<Item> {
         let field_file = composite_file
             .open_read(field)
             .expect("File component not found");
-        FastFieldReader::open(field_file).unwrap()
+        BitpackedFastFieldReader::open(field_file).unwrap()
     }
 }

src/fastfield/readers.rs

@@ -1,13 +1,16 @@
 use crate::common::CompositeFile;
 use crate::directory::FileSlice;
 use crate::fastfield::MultiValuedFastFieldReader;
+use crate::fastfield::{BitpackedFastFieldReader, FastFieldNotAvailableError};
 use crate::fastfield::{BytesFastFieldReader, FastValue};
-use crate::fastfield::{FastFieldNotAvailableError, FastFieldReader};
 use crate::schema::{Cardinality, Field, FieldType, Schema};
 use crate::space_usage::PerFieldSpaceUsage;
 use crate::TantivyError;
 
-/// Provides access to all of the FastFieldReader.
+use super::reader::DynamicFastFieldReader;
+use super::FastFieldReader;
+
+/// Provides access to all of the BitpackedFastFieldReader.
 ///
 /// Internally, `FastFieldReaders` have preloaded fast field readers,
 /// and just wraps several `HashMap`.
@@ -100,9 +103,9 @@ impl FastFieldReaders {
     pub(crate) fn typed_fast_field_reader<TFastValue: FastValue>(
         &self,
         field: Field,
-    ) -> crate::Result<FastFieldReader<TFastValue>> {
+    ) -> crate::Result<DynamicFastFieldReader<TFastValue>> {
         let fast_field_slice = self.fast_field_data(field, 0)?;
-        FastFieldReader::open(fast_field_slice)
+        DynamicFastFieldReader::open(fast_field_slice)
     }
 
     pub(crate) fn typed_fast_field_multi_reader<TFastValue: FastValue>(
@@ -111,16 +114,16 @@ impl FastFieldReaders {
     ) -> crate::Result<MultiValuedFastFieldReader<TFastValue>> {
         let fast_field_slice_idx = self.fast_field_data(field, 0)?;
         let fast_field_slice_vals = self.fast_field_data(field, 1)?;
-        let idx_reader = FastFieldReader::open(fast_field_slice_idx)?;
-        let vals_reader: FastFieldReader<TFastValue> =
-            FastFieldReader::open(fast_field_slice_vals)?;
+        let idx_reader = BitpackedFastFieldReader::open(fast_field_slice_idx)?;
+        let vals_reader: BitpackedFastFieldReader<TFastValue> =
+            BitpackedFastFieldReader::open(fast_field_slice_vals)?;
         Ok(MultiValuedFastFieldReader::open(idx_reader, vals_reader))
     }
 
     /// Returns the `u64` fast field reader reader associated to `field`.
     ///
     /// If `field` is not a u64 fast field, this method returns an Error.
-    pub fn u64(&self, field: Field) -> crate::Result<FastFieldReader<u64>> {
+    pub fn u64(&self, field: Field) -> crate::Result<DynamicFastFieldReader<u64>> {
         self.check_type(field, FastType::U64, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }
@@ -129,14 +132,14 @@ impl FastFieldReaders {
     /// field is effectively of type `u64` or not.
     ///
     /// If not, the fastfield reader will returns the u64-value associated to the original FastValue.
-    pub fn u64_lenient(&self, field: Field) -> crate::Result<FastFieldReader<u64>> {
+    pub fn u64_lenient(&self, field: Field) -> crate::Result<DynamicFastFieldReader<u64>> {
         self.typed_fast_field_reader(field)
     }
 
     /// Returns the `i64` fast field reader reader associated to `field`.
     ///
     /// If `field` is not a i64 fast field, this method returns an Error.
-    pub fn i64(&self, field: Field) -> crate::Result<FastFieldReader<i64>> {
+    pub fn i64(&self, field: Field) -> crate::Result<impl FastFieldReader<i64>> {
         self.check_type(field, FastType::I64, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }
@@ -144,7 +147,7 @@ impl FastFieldReaders {
     /// Returns the `i64` fast field reader reader associated to `field`.
     ///
     /// If `field` is not a i64 fast field, this method returns an Error.
-    pub fn date(&self, field: Field) -> crate::Result<FastFieldReader<crate::DateTime>> {
+    pub fn date(&self, field: Field) -> crate::Result<impl FastFieldReader<crate::DateTime>> {
         self.check_type(field, FastType::Date, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }
@@ -152,7 +155,7 @@ impl FastFieldReaders {
     /// Returns the `f64` fast field reader reader associated to `field`.
     ///
     /// If `field` is not a f64 fast field, this method returns an Error.
-    pub fn f64(&self, field: Field) -> crate::Result<FastFieldReader<f64>> {
+    pub fn f64(&self, field: Field) -> crate::Result<impl FastFieldReader<f64>> {
         self.check_type(field, FastType::F64, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }
@@ -213,7 +216,7 @@ impl FastFieldReaders {
                 )));
             }
             let fast_field_idx_file = self.fast_field_data(field, 0)?;
-            let idx_reader = FastFieldReader::open(fast_field_idx_file)?;
+            let idx_reader = BitpackedFastFieldReader::open(fast_field_idx_file)?;
             let data = self.fast_field_data(field, 1)?;
             BytesFastFieldReader::open(idx_reader, data)
         } else {

src/fastfield/serializer.rs

@@ -7,10 +7,10 @@ use std::io::{self, Write};
 use tantivy_bitpacker::compute_num_bits;
 use tantivy_bitpacker::BitPacker;
 
-/// `FastFieldSerializer` is in charge of serializing
+/// `CompositeFastFieldSerializer` is in charge of serializing
 /// fastfields on disk.
 ///
-/// Fast fields are encoded using bit-packing.
+/// Fast fields have differnt encodings like bit-packing.
 ///
 /// `FastFieldWriter`s are in charge of pushing the data to
 /// the serializer.
@@ -27,16 +27,16 @@ use tantivy_bitpacker::BitPacker;
 /// * ...
 /// * `close_field()`
 /// * `close()`
-pub struct FastFieldSerializer {
+pub struct CompositeFastFieldSerializer {
     composite_write: CompositeWrite<WritePtr>,
 }
 
-impl FastFieldSerializer {
+impl CompositeFastFieldSerializer {
     /// Constructor
-    pub fn from_write(write: WritePtr) -> io::Result<FastFieldSerializer> {
+    pub fn from_write(write: WritePtr) -> io::Result<CompositeFastFieldSerializer> {
         // just making room for the pointer to header.
         let composite_write = CompositeWrite::wrap(write);
-        Ok(FastFieldSerializer { composite_write })
+        Ok(CompositeFastFieldSerializer { composite_write })
     }
 
     /// Start serializing a new u64 fast field
@@ -45,7 +45,7 @@ impl FastFieldSerializer {
         field: Field,
         min_value: u64,
         max_value: u64,
-    ) -> io::Result<FastSingleFieldSerializer<'_, CountingWriter<WritePtr>>> {
+    ) -> io::Result<DynamicFastFieldSerializer<'_, CountingWriter<WritePtr>>> {
         self.new_u64_fast_field_with_idx(field, min_value, max_value, 0)
     }
 
@@ -56,9 +56,9 @@ impl FastFieldSerializer {
         min_value: u64,
         max_value: u64,
         idx: usize,
-    ) -> io::Result<FastSingleFieldSerializer<'_, CountingWriter<WritePtr>>> {
+    ) -> io::Result<DynamicFastFieldSerializer<'_, CountingWriter<WritePtr>>> {
         let field_write = self.composite_write.for_field_with_idx(field, idx);
-        FastSingleFieldSerializer::open(field_write, min_value, max_value)
+        DynamicFastFieldSerializer::open(field_write, min_value, max_value)
     }
 
     /// Start serializing a new [u8] fast field
@@ -79,14 +79,111 @@ impl FastFieldSerializer {
     }
 }
 
-pub struct FastSingleFieldSerializer<'a, W: Write> {
+#[derive(Debug, Clone)]
+pub struct EstimationStats {
+    min_value: u64,
+    max_value: u64,
+}
+/// The FastFieldSerializer trait is the common interface
+/// implemented by every fastfield serializer variant.
+///
+/// `DynamicFastFieldSerializer` is the enum wrapping all variants.
+/// It is used to create an serializer instance.
+pub trait FastFieldSerializer {
+    /// add value to serializer
+    fn add_val(&mut self, val: u64) -> io::Result<()>;
+    /// finish serializing a field.
+    fn close_field(self) -> io::Result<()>;
+}
+
+/// The FastFieldSerializerEstimate trait is required on all variants
+/// of fast field compressions, to decide which one to choose.
+pub trait FastFieldSerializerEstimate {
+    /// returns an estimate of the compression ratio.
+    fn estimate(
+        /*fastfield_accessor: impl FastFieldReader<u64>,*/ stats: EstimationStats,
+    ) -> (f32, &'static str);
+    /// the unique name of the compressor
+    fn name() -> &'static str;
+}
+
+pub enum DynamicFastFieldSerializer<'a, W: Write> {
+    Bitpacked(BitpackedFastFieldSerializer<'a, W>),
+}
+
+impl<'a, W: Write> DynamicFastFieldSerializer<'a, W> {
+    /// Creates a new fast field serializer.
+    ///
+    /// The serializer in fact encode the values by bitpacking
+    /// `(val - min_value)`.
+    ///
+    /// It requires a `min_value` and a `max_value` to compute
+    /// compute the minimum number of bits required to encode
+    /// values.
+    pub fn open(
+        write: &'a mut W,
+        min_value: u64,
+        max_value: u64,
+    ) -> io::Result<DynamicFastFieldSerializer<'a, W>> {
+        let stats = EstimationStats {
+            min_value,
+            max_value,
+        };
+        let (_ratio, name) = (
+            BitpackedFastFieldSerializer::<Vec<u8>>::estimate(stats),
+            BitpackedFastFieldSerializer::<Vec<u8>>::name(),
+        );
+        Self::open_from_name(write, min_value, max_value, name)
+    }
+
+    /// Creates a new fast field serializer.
+    ///
+    /// The serializer in fact encode the values by bitpacking
+    /// `(val - min_value)`.
+    ///
+    /// It requires a `min_value` and a `max_value` to compute
+    /// compute the minimum number of bits required to encode
+    /// values.
+    pub fn open_from_name(
+        write: &'a mut W,
+        min_value: u64,
+        max_value: u64,
+        name: &str,
+    ) -> io::Result<DynamicFastFieldSerializer<'a, W>> {
+        // Weirdly the W generic on BitpackedFastFieldSerializer needs to be set,
+        // although name() doesn't use it
+        let variant = if name == BitpackedFastFieldSerializer::<Vec<u8>>::name() {
+            DynamicFastFieldSerializer::Bitpacked(BitpackedFastFieldSerializer::open(
+                write, min_value, max_value,
+            )?)
+        } else {
+            panic!("unknown fastfield serializer {}", name);
+        };
+
+        Ok(variant)
+    }
+}
+impl<'a, W: Write> FastFieldSerializer for DynamicFastFieldSerializer<'a, W> {
+    fn add_val(&mut self, val: u64) -> io::Result<()> {
+        match self {
+            Self::Bitpacked(serializer) => serializer.add_val(val),
+        }
+    }
+    fn close_field(self) -> io::Result<()> {
+        match self {
+            Self::Bitpacked(serializer) => serializer.close_field(),
+        }
+    }
+}
+
+pub struct BitpackedFastFieldSerializer<'a, W: Write> {
     bit_packer: BitPacker,
     write: &'a mut W,
     min_value: u64,
     num_bits: u8,
 }
 
-impl<'a, W: Write> FastSingleFieldSerializer<'a, W> {
+impl<'a, W: Write> BitpackedFastFieldSerializer<'a, W> {
     /// Creates a new fast field serializer.
     ///
     /// The serializer in fact encode the values by bitpacking
@@ -99,34 +196,51 @@ impl<'a, W: Write> FastSingleFieldSerializer<'a, W> {
         write: &'a mut W,
         min_value: u64,
         max_value: u64,
-    ) -> io::Result<FastSingleFieldSerializer<'a, W>> {
+    ) -> io::Result<BitpackedFastFieldSerializer<'a, W>> {
         assert!(min_value <= max_value);
         min_value.serialize(write)?;
         let amplitude = max_value - min_value;
         amplitude.serialize(write)?;
         let num_bits = compute_num_bits(amplitude);
         let bit_packer = BitPacker::new();
-        Ok(FastSingleFieldSerializer {
+        Ok(BitpackedFastFieldSerializer {
             bit_packer,
             write,
             min_value,
             num_bits,
         })
     }
+}
 
+impl<'a, W: 'a + Write> FastFieldSerializer for BitpackedFastFieldSerializer<'a, W> {
     /// Pushes a new value to the currently open u64 fast field.
-    pub fn add_val(&mut self, val: u64) -> io::Result<()> {
+    fn add_val(&mut self, val: u64) -> io::Result<()> {
         let val_to_write: u64 = val - self.min_value;
         self.bit_packer
             .write(val_to_write, self.num_bits, &mut self.write)?;
         Ok(())
     }
-
-    pub fn close_field(mut self) -> io::Result<()> {
+    fn close_field(mut self) -> io::Result<()> {
         self.bit_packer.close(&mut self.write)
     }
 }
 
+impl<'a, W: 'a + Write> FastFieldSerializerEstimate for BitpackedFastFieldSerializer<'a, W> {
+    fn estimate(
+        /*_fastfield_accessor: impl FastFieldReader<u64>, */ stats: EstimationStats,
+    ) -> (f32, &'static str) {
+        let amplitude = stats.max_value - stats.min_value;
+        let num_bits = compute_num_bits(amplitude);
+        let num_bits_uncompressed = 64;
+        let ratio = num_bits as f32 / num_bits_uncompressed as f32;
+        let name = Self::name();
+        (ratio, name)
+    }
+    fn name() -> &'static str {
+        "Bitpacked"
+    }
+}
+
 pub struct FastBytesFieldSerializer<'a, W: Write> {
     write: &'a mut W,
 }

src/fastfield/writer.rs

@@ -1,6 +1,7 @@
 use super::multivalued::MultiValuedFastFieldWriter;
 use crate::common;
-use crate::fastfield::{BytesFastFieldWriter, FastFieldSerializer};
+use crate::fastfield::serializer::FastFieldSerializer;
+use crate::fastfield::{BytesFastFieldWriter, CompositeFastFieldSerializer};
 use crate::indexer::doc_id_mapping::DocIdMapping;
 use crate::postings::UnorderedTermId;
 use crate::schema::{Cardinality, Document, Field, FieldEntry, FieldType, Schema};
@@ -148,7 +149,7 @@ impl FastFieldsWriter {
     /// order to the fast field serializer.
     pub fn serialize(
         &self,
-        serializer: &mut FastFieldSerializer,
+        serializer: &mut CompositeFastFieldSerializer,
         mapping: &HashMap<Field, FnvHashMap<UnorderedTermId, TermOrdinal>>,
         doc_id_map: Option<&DocIdMapping>,
     ) -> io::Result<()> {
@@ -272,7 +273,7 @@ impl IntFastFieldWriter {
     /// Push the fast fields value to the `FastFieldWriter`.
     pub fn serialize(
         &self,
-        serializer: &mut FastFieldSerializer,
+        serializer: &mut CompositeFastFieldSerializer,
         doc_id_map: Option<&DocIdMapping>,
     ) -> io::Result<()> {
         let (min, max) = if self.val_min > self.val_max {

src/indexer/doc_id_mapping.rs

@@ -8,7 +8,6 @@ use crate::{
     DocId, IndexSortByField, Order, TantivyError,
 };
 use std::cmp::Reverse;
-
 /// Struct to provide mapping from old doc_id to new doc_id and vice versa
 pub struct DocIdMapping {
     new_doc_id_to_old: Vec<DocId>,
@@ -92,6 +91,7 @@ pub(crate) fn get_doc_id_mapping_from_field(
 
 #[cfg(test)]
 mod tests_indexsorting {
+    use crate::fastfield::FastFieldReader;
     use crate::{collector::TopDocs, query::QueryParser, schema::*};
     use crate::{schema::Schema, DocAddress};
     use crate::{Index, IndexSettings, IndexSortByField, Order};
@@ -175,6 +175,7 @@ mod tests_indexsorting {
                         field: "my_number".to_string(),
                         order: Order::Asc,
                     }),
+                    ..Default::default()
                 }),
                 option.clone(),
             );
@@ -206,6 +207,7 @@ mod tests_indexsorting {
                         field: "my_number".to_string(),
                         order: Order::Desc,
                     }),
+                    ..Default::default()
                 }),
                 option.clone(),
             );
@@ -264,6 +266,7 @@ mod tests_indexsorting {
                     field: "my_number".to_string(),
                     order: Order::Asc,
                 }),
+                ..Default::default()
             }),
             get_text_options(),
         );
@@ -288,6 +291,7 @@ mod tests_indexsorting {
                     field: "my_number".to_string(),
                     order: Order::Desc,
                 }),
+                ..Default::default()
             }),
             get_text_options(),
         );
@@ -322,6 +326,7 @@ mod tests_indexsorting {
                     field: "my_number".to_string(),
                     order: Order::Asc,
                 }),
+                ..Default::default()
             }),
             get_text_options(),
         );
@@ -352,6 +357,7 @@ mod tests_indexsorting {
                     field: "my_number".to_string(),
                     order: Order::Desc,
                 }),
+                ..Default::default()
             }),
             get_text_options(),
         );
@@ -387,6 +393,7 @@ mod tests_indexsorting {
                     field: "my_number".to_string(),
                     order: Order::Asc,
                 }),
+                ..Default::default()
             }),
             get_text_options(),
         );

src/indexer/index_writer.rs

@@ -945,7 +945,7 @@ mod tests {
         let index_writer = index.writer(3_000_000).unwrap();
         assert_eq!(
             format!("{:?}", index_writer.get_merge_policy()),
-            "LogMergePolicy { min_merge_size: 8, max_merge_size: 10000000, min_layer_size: 10000, \
+            "LogMergePolicy { min_num_segments: 8, max_docs_before_merge: 10000000, min_layer_size: 10000, \
              level_log_size: 0.75 }"
         );
         let merge_policy = Box::new(NoMergePolicy::default());

src/indexer/log_merge_policy.rs

@@ -1,19 +1,20 @@
 use super::merge_policy::{MergeCandidate, MergePolicy};
 use crate::core::SegmentMeta;
+use itertools::Itertools;
 use std::cmp;
 use std::f64;
 
 const DEFAULT_LEVEL_LOG_SIZE: f64 = 0.75;
 const DEFAULT_MIN_LAYER_SIZE: u32 = 10_000;
-const DEFAULT_MIN_MERGE_SIZE: usize = 8;
-const DEFAULT_MAX_MERGE_SIZE: usize = 10_000_000;
+const DEFAULT_MIN_NUM_SEGMENTS_IN_MERGE: usize = 8;
+const DEFAULT_MAX_DOCS_BEFORE_MERGE: usize = 10_000_000;
 
 /// `LogMergePolicy` tries to merge segments that have a similar number of
 /// documents.
 #[derive(Debug, Clone)]
 pub struct LogMergePolicy {
-    min_merge_size: usize,
-    max_merge_size: usize,
+    min_num_segments: usize,
+    max_docs_before_merge: usize,
     min_layer_size: u32,
     level_log_size: f64,
 }
@@ -23,15 +24,16 @@ impl LogMergePolicy {
         cmp::max(self.min_layer_size, size)
     }
 
-    /// Set the minimum number of segment that may be merge together.
-    pub fn set_min_merge_size(&mut self, min_merge_size: usize) {
-        self.min_merge_size = min_merge_size;
+    /// Set the minimum number of segments that may be merged together.
+    pub fn set_min_num_segments(&mut self, min_num_segments: usize) {
+        self.min_num_segments = min_num_segments;
     }
 
     /// Set the maximum number docs in a segment for it to be considered for
-    /// merging.
-    pub fn set_max_merge_size(&mut self, max_merge_size: usize) {
-        self.max_merge_size = max_merge_size;
+    /// merging. A segment can still reach more than max_docs, by merging many
+    /// smaller ones.
+    pub fn set_max_docs_before_merge(&mut self, max_docs_merge_size: usize) {
+        self.max_docs_before_merge = max_docs_merge_size;
     }
 
     /// Set the minimum segment size under which all segment belong
@@ -42,7 +44,7 @@ impl LogMergePolicy {
 
     /// Set the ratio between two consecutive levels.
     ///
-    /// Segment are group in levels according to their sizes.
+    /// Segments are grouped in levels according to their sizes.
     /// These levels are defined as intervals of exponentially growing sizes.
     /// level_log_size define the factor by which one should multiply the limit
     /// to reach a level, in order to get the limit to reach the following
@@ -54,52 +56,43 @@ impl LogMergePolicy {
 
 impl MergePolicy for LogMergePolicy {
     fn compute_merge_candidates(&self, segments: &[SegmentMeta]) -> Vec<MergeCandidate> {
-        let mut size_sorted_tuples = segments
+        let mut size_sorted_segments = segments
             .iter()
-            .map(SegmentMeta::num_docs)
-            .enumerate()
-            .filter(|(_, s)| s <= &(self.max_merge_size as u32))
-            .collect::<Vec<(usize, u32)>>();
+            .filter(|segment_meta| segment_meta.num_docs() <= (self.max_docs_before_merge as u32))
+            .collect::<Vec<&SegmentMeta>>();
 
-        size_sorted_tuples.sort_by(|x, y| y.1.cmp(&(x.1)));
-
-        if size_sorted_tuples.len() <= 1 {
-            return Vec::new();
-        }
-
-        let size_sorted_log_tuples: Vec<_> = size_sorted_tuples
-            .into_iter()
-            .map(|(ind, num_docs)| (ind, f64::from(self.clip_min_size(num_docs)).log2()))
-            .collect();
-
-        if let Some(&(first_ind, first_score)) = size_sorted_log_tuples.first() {
-            let mut current_max_log_size = first_score;
-            let mut levels = vec![vec![first_ind]];
-            for &(ind, score) in (&size_sorted_log_tuples).iter().skip(1) {
-                if score < (current_max_log_size - self.level_log_size) {
-                    current_max_log_size = score;
-                    levels.push(Vec::new());
-                }
-                levels.last_mut().unwrap().push(ind);
-            }
-            levels
-                .iter()
-                .filter(|level| level.len() >= self.min_merge_size)
-                .map(|ind_vec| {
-                    MergeCandidate(ind_vec.iter().map(|&ind| segments[ind].id()).collect())
-                })
-                .collect()
-        } else {
+        if size_sorted_segments.len() <= 1 {
             return vec![];
         }
+        size_sorted_segments.sort_by_key(|seg| std::cmp::Reverse(seg.num_docs()));
+
+        let mut current_max_log_size = f64::MAX;
+        let mut levels = vec![];
+        for (_, merge_group) in &size_sorted_segments.into_iter().group_by(|segment| {
+            let segment_log_size = f64::from(self.clip_min_size(segment.num_docs())).log2();
+            if segment_log_size < (current_max_log_size - self.level_log_size) {
+                // update current_max_log_size to create a new group
+                current_max_log_size = segment_log_size;
+            }
+            // return current_max_log_size to be grouped to the current group
+            current_max_log_size
+        }) {
+            levels.push(merge_group.collect::<Vec<&SegmentMeta>>());
+        }
+
+        levels
+            .iter()
+            .filter(|level| level.len() >= self.min_num_segments)
+            .map(|segments| MergeCandidate(segments.iter().map(|&seg| seg.id()).collect()))
+            .collect()
     }
 }
 
 impl Default for LogMergePolicy {
     fn default() -> LogMergePolicy {
         LogMergePolicy {
-            min_merge_size: DEFAULT_MIN_MERGE_SIZE,
-            max_merge_size: DEFAULT_MAX_MERGE_SIZE,
+            min_num_segments: DEFAULT_MIN_NUM_SEGMENTS_IN_MERGE,
+            max_docs_before_merge: DEFAULT_MAX_DOCS_BEFORE_MERGE,
             min_layer_size: DEFAULT_MIN_LAYER_SIZE,
             level_log_size: DEFAULT_LEVEL_LOG_SIZE,
         }
@@ -109,16 +102,79 @@ impl Default for LogMergePolicy {
 #[cfg(test)]
 mod tests {
     use super::*;
-    use crate::core::{SegmentId, SegmentMeta, SegmentMetaInventory};
-    use crate::indexer::merge_policy::MergePolicy;
+    use crate::{
+        core::{SegmentId, SegmentMeta, SegmentMetaInventory},
+        schema,
+    };
+    use crate::{indexer::merge_policy::MergePolicy, schema::INDEXED};
     use once_cell::sync::Lazy;
 
     static INVENTORY: Lazy<SegmentMetaInventory> = Lazy::new(SegmentMetaInventory::default);
 
+    use crate::Index;
+
+    #[test]
+    fn create_index_test_max_merge_issue_1035() {
+        let mut schema_builder = schema::Schema::builder();
+        let int_field = schema_builder.add_u64_field("intval", INDEXED);
+        let schema = schema_builder.build();
+
+        let index = Index::create_in_ram(schema);
+
+        {
+            let mut log_merge_policy = LogMergePolicy::default();
+            log_merge_policy.set_min_num_segments(1);
+            log_merge_policy.set_max_docs_before_merge(1);
+            log_merge_policy.set_min_layer_size(0);
+
+            let mut index_writer = index.writer_for_tests().unwrap();
+            index_writer.set_merge_policy(Box::new(log_merge_policy));
+
+            // after every commit the merge checker is started, it will merge only segments with 1
+            // element in it because of the max_merge_size.
+            index_writer.add_document(doc!(int_field=>1_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>2_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>3_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>4_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>5_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>6_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>7_u64));
+            assert!(index_writer.commit().is_ok());
+
+            index_writer.add_document(doc!(int_field=>8_u64));
+            assert!(index_writer.commit().is_ok());
+        }
+
+        let _segment_ids = index
+            .searchable_segment_ids()
+            .expect("Searchable segments failed.");
+
+        let reader = index.reader().unwrap();
+        let searcher = reader.searcher();
+        let segment_readers = searcher.segment_readers();
+        for segment in segment_readers {
+            if segment.num_docs() > 2 {
+                panic!("segment can't have more than two segments");
+            } // don't know how to wait for the merge, then it could be a simple eq
+        }
+    }
+
     fn test_merge_policy() -> LogMergePolicy {
         let mut log_merge_policy = LogMergePolicy::default();
-        log_merge_policy.set_min_merge_size(3);
-        log_merge_policy.set_max_merge_size(100_000);
+        log_merge_policy.set_min_num_segments(3);
+        log_merge_policy.set_max_docs_before_merge(100_000);
         log_merge_policy.set_min_layer_size(2);
         log_merge_policy
     }

src/indexer/merger.rs

@@ -1,5 +1,8 @@
 use super::doc_id_mapping::DocIdMapping;
+use crate::error::DataCorruption;
+use crate::fastfield::CompositeFastFieldSerializer;
 use crate::fastfield::DeleteBitSet;
+use crate::fastfield::DynamicFastFieldReader;
 use crate::fastfield::FastFieldReader;
 use crate::fastfield::FastFieldSerializer;
 use crate::fastfield::MultiValuedFastFieldReader;
@@ -86,7 +89,7 @@ pub struct IndexMerger {
 }
 
 fn compute_min_max_val(
-    u64_reader: &FastFieldReader<u64>,
+    u64_reader: &impl FastFieldReader<u64>,
     max_doc: DocId,
     delete_bitset_opt: Option<&DeleteBitSet>,
 ) -> Option<(u64, u64)> {
@@ -182,6 +185,10 @@ impl IndexMerger {
                 readers.push(reader);
             }
         }
+        if let Some(sort_by_field) = index_settings.sort_by_field.as_ref() {
+            readers = Self::sort_readers_by_min_sort_field(readers, sort_by_field)?;
+        }
+        // sort segments by their natural sort setting
         if max_doc >= MAX_DOC_LIMIT {
             let err_msg = format!(
                 "The segment resulting from this merge would have {} docs,\
@@ -191,13 +198,37 @@ impl IndexMerger {
             return Err(crate::TantivyError::InvalidArgument(err_msg));
         }
         Ok(IndexMerger {
-            schema,
             index_settings,
+            schema,
             readers,
             max_doc,
         })
     }
 
+    fn sort_readers_by_min_sort_field(
+        readers: Vec<SegmentReader>,
+        sort_by_field: &IndexSortByField,
+    ) -> crate::Result<Vec<SegmentReader>> {
+        // presort the readers by their min_values, so that when they are disjunct, we can use
+        // the regular merge logic (implicitly sorted)
+        let mut readers_with_min_sort_values = readers
+            .into_iter()
+            .map(|reader| {
+                let accessor = Self::get_sort_field_accessor(&reader, &sort_by_field)?;
+                Ok((reader, accessor.min_value()))
+            })
+            .collect::<crate::Result<Vec<_>>>()?;
+        if sort_by_field.order.is_asc() {
+            readers_with_min_sort_values.sort_by_key(|(_, min_val)| *min_val);
+        } else {
+            readers_with_min_sort_values.sort_by_key(|(_, min_val)| std::cmp::Reverse(*min_val));
+        }
+        Ok(readers_with_min_sort_values
+            .into_iter()
+            .map(|(reader, _)| reader)
+            .collect())
+    }
+
     fn write_fieldnorms(
         &self,
         mut fieldnorms_serializer: FieldNormsSerializer,
@@ -208,9 +239,14 @@ impl IndexMerger {
         for field in fields {
             fieldnorms_data.clear();
             if let Some(doc_id_mapping) = doc_id_mapping {
+                let fieldnorms_readers: Vec<FieldNormReader> = self
+                    .readers
+                    .iter()
+                    .map(|reader| reader.get_fieldnorms_reader(field))
+                    .collect::<Result<_, _>>()?;
                 for (doc_id, reader_with_ordinal) in doc_id_mapping {
                     let fieldnorms_reader =
-                        reader_with_ordinal.reader.get_fieldnorms_reader(field)?;
+                        &fieldnorms_readers[reader_with_ordinal.ordinal as usize];
                     let fieldnorm_id = fieldnorms_reader.fieldnorm_id(*doc_id);
                     fieldnorms_data.push(fieldnorm_id);
                 }
@@ -231,7 +267,7 @@ impl IndexMerger {
 
     fn write_fast_fields(
         &self,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         mut term_ord_mappings: HashMap<Field, TermOrdinalMapping>,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
     ) -> crate::Result<()> {
@@ -281,11 +317,11 @@ impl IndexMerger {
     fn write_single_fast_field(
         &self,
         field: Field,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
     ) -> crate::Result<()> {
         let (min_value, max_value) = self.readers.iter().map(|reader|{
-                let u64_reader: FastFieldReader<u64> = reader
+                let u64_reader: DynamicFastFieldReader<u64> = reader
                 .fast_fields()
                 .typed_fast_field_reader(field)
                 .expect("Failed to find a reader for single fast field. This is a tantivy bug and it should never happen.");
@@ -300,7 +336,7 @@ impl IndexMerger {
             .readers
             .iter()
             .map(|reader| {
-               let u64_reader: FastFieldReader<u64> = reader
+               let u64_reader: DynamicFastFieldReader<u64> = reader
                     .fast_fields()
                     .typed_fast_field_reader(field)
                     .expect("Failed to find a reader for single fast field. This is a tantivy bug and it should never happen.");
@@ -328,7 +364,7 @@ impl IndexMerger {
             let u64_readers = self.readers.iter()
                 .filter(|reader|reader.max_doc() != reader.delete_bitset().map(|bit_set|bit_set.len() as u32).unwrap_or(0))
                 .map(|reader|{
-                let u64_reader: FastFieldReader<u64> = reader
+                let u64_reader: DynamicFastFieldReader<u64> = reader
                 .fast_fields()
                 .typed_fast_field_reader(field)
                 .expect("Failed to find a reader for single fast field. This is a tantivy bug and it should never happen.");
@@ -354,6 +390,60 @@ impl IndexMerger {
         }
     }
 
+    /// Checks if the readers are disjunct for their sort property and in the correct order to be
+    /// able to just stack them.
+    pub(crate) fn is_disjunct_and_sorted_on_sort_property(
+        &self,
+        sort_by_field: &IndexSortByField,
+    ) -> crate::Result<bool> {
+        let reader_and_field_accessors = self.get_reader_with_sort_field_accessor(sort_by_field)?;
+
+        let everything_is_in_order = reader_and_field_accessors
+            .into_iter()
+            .map(|reader| reader.1)
+            .tuple_windows()
+            .all(|(field_accessor1, field_accessor2)| {
+                if sort_by_field.order.is_asc() {
+                    field_accessor1.max_value() <= field_accessor2.min_value()
+                } else {
+                    field_accessor1.min_value() >= field_accessor2.max_value()
+                }
+            });
+        Ok(everything_is_in_order)
+    }
+
+    pub(crate) fn get_sort_field_accessor(
+        reader: &SegmentReader,
+        sort_by_field: &IndexSortByField,
+    ) -> crate::Result<impl FastFieldReader<u64>> {
+        let field_id = expect_field_id_for_sort_field(&reader.schema(), &sort_by_field)?; // for now expect fastfield, but not strictly required
+        let value_accessor = reader.fast_fields().u64_lenient(field_id)?;
+        Ok(value_accessor)
+    }
+    /// Collecting value_accessors into a vec to bind the lifetime.
+    pub(crate) fn get_reader_with_sort_field_accessor<'a, 'b>(
+        &'a self,
+        sort_by_field: &'b IndexSortByField,
+    ) -> crate::Result<
+        Vec<(
+            SegmentReaderWithOrdinal<'a>,
+            impl FastFieldReader<u64> + Clone,
+        )>,
+    > {
+        let reader_and_field_accessors = self
+            .readers
+            .iter()
+            .enumerate()
+            .map(Into::into)
+            .map(|reader_with_ordinal: SegmentReaderWithOrdinal| {
+                let value_accessor =
+                    Self::get_sort_field_accessor(reader_with_ordinal.reader, sort_by_field)?;
+                Ok((reader_with_ordinal, value_accessor))
+            })
+            .collect::<crate::Result<Vec<_>>>()?;
+        Ok(reader_and_field_accessors)
+    }
+
     /// Generates the doc_id mapping where position in the vec=new
     /// doc_id.
     /// ReaderWithOrdinal will include the ordinal position of the
@@ -362,42 +452,26 @@ impl IndexMerger {
         &self,
         sort_by_field: &IndexSortByField,
     ) -> crate::Result<Vec<(DocId, SegmentReaderWithOrdinal)>> {
-        let reader_and_field_accessors = self
-            .readers
-            .iter()
-            .enumerate()
-            .map(|reader| {
-                let reader_with_ordinal: SegmentReaderWithOrdinal = reader.into();
-                let field_id = expect_field_id_for_sort_field(
-                    &reader_with_ordinal.reader.schema(),
-                    &sort_by_field,
-                )?; // for now expect fastfield, but not strictly required
-                let value_accessor = reader_with_ordinal
-                    .reader
-                    .fast_fields()
-                    .u64_lenient(field_id)?;
-                Ok((reader_with_ordinal, value_accessor))
-            })
-            .collect::<crate::Result<Vec<_>>>()?; // Collecting to bind the lifetime of value_accessor into the vec, or can't be used as a reference.
-                                                  // Loading the field accessor on demand causes a 15x regression
+        let reader_and_field_accessors = self.get_reader_with_sort_field_accessor(sort_by_field)?;
+        // Loading the field accessor on demand causes a 15x regression
 
         // create iterators over segment/sort_accessor/doc_id  tuple
-        let doc_id_reader_pair = reader_and_field_accessors
-            .iter()
-            .map(|reader_and_field_accessor| {
-                reader_and_field_accessor
-                    .0
-                    .reader
-                    .doc_ids_alive()
-                    .map(move |doc_id| {
-                        (
-                            doc_id,
-                            reader_and_field_accessor.0,
-                            &reader_and_field_accessor.1,
-                        )
-                    })
-            })
-            .collect::<Vec<_>>();
+        let doc_id_reader_pair =
+            reader_and_field_accessors
+                .iter()
+                .map(|reader_and_field_accessor| {
+                    reader_and_field_accessor
+                        .0
+                        .reader
+                        .doc_ids_alive()
+                        .map(move |doc_id| {
+                            (
+                                doc_id,
+                                reader_and_field_accessor.0,
+                                &reader_and_field_accessor.1,
+                            )
+                        })
+                });
 
         // create iterator tuple of (old doc_id, reader) in order of the new doc_ids
         let sorted_doc_ids: Vec<(DocId, SegmentReaderWithOrdinal)> = doc_id_reader_pair
@@ -424,7 +498,7 @@ impl IndexMerger {
     // is used to index the reader_and_field_accessors vec.
     fn write_1_n_fast_field_idx_generic(
         field: Field,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
         reader_and_field_accessors: &[(&SegmentReader, impl MultiValueLength)],
     ) -> crate::Result<()> {
@@ -479,7 +553,7 @@ impl IndexMerger {
     fn write_multi_value_fast_field_idx(
         &self,
         field: Field,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
     ) -> crate::Result<()> {
         let reader_and_field_accessors = self.readers.iter().map(|reader|{
@@ -501,7 +575,7 @@ impl IndexMerger {
         &self,
         field: Field,
         term_ordinal_mappings: &TermOrdinalMapping,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
     ) -> crate::Result<()> {
         // Multifastfield consists in 2 fastfields.
@@ -564,7 +638,7 @@ impl IndexMerger {
     fn write_multi_fast_field(
         &self,
         field: Field,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
     ) -> crate::Result<()> {
         // Multifastfield consists in 2 fastfields.
@@ -651,7 +725,7 @@ impl IndexMerger {
     fn write_bytes_fast_field(
         &self,
         field: Field,
-        fast_field_serializer: &mut FastFieldSerializer,
+        fast_field_serializer: &mut CompositeFastFieldSerializer,
         doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
     ) -> crate::Result<()> {
         let reader_and_field_accessors = self
@@ -797,13 +871,11 @@ impl IndexMerger {
             let mut total_doc_freq = 0;
 
             // Let's compute the list of non-empty posting lists
-            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];
+            for (segment_ord, term_info) in merged_terms.current_segment_ordinals_and_term_infos() {
+                let segment_reader = &self.readers[segment_ord];
                 let inverted_index: &InvertedIndexReader = &*field_readers[segment_ord];
                 let segment_postings = inverted_index
-                    .read_postings_from_terminfo(term_info, segment_postings_option)?;
+                    .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)
@@ -927,19 +999,41 @@ impl IndexMerger {
             .collect();
         if let Some(doc_id_mapping) = doc_id_mapping {
             for (old_doc_id, reader_with_ordinal) in doc_id_mapping {
-                let store_reader = &mut document_iterators[reader_with_ordinal.ordinal as usize];
-                let raw_doc = store_reader.next().expect(&format!(
-                    "unexpected missing document in docstore on merge, doc id {:?}",
-                    old_doc_id
-                ))?;
-                store_writer.store_bytes(raw_doc.get_bytes())?;
+                let doc_bytes_it = &mut document_iterators[reader_with_ordinal.ordinal as usize];
+                if let Some(doc_bytes_res) = doc_bytes_it.next() {
+                    let doc_bytes = doc_bytes_res?;
+                    store_writer.store_bytes(&doc_bytes)?;
+                } else {
+                    return Err(DataCorruption::comment_only(&format!(
+                        "unexpected missing document in docstore on merge, doc id {:?}",
+                        old_doc_id
+                    ))
+                    .into());
+                }
             }
         } else {
             for reader in &self.readers {
                 let store_reader = reader.get_store_reader()?;
-                if reader.num_deleted_docs() > 0 {
-                    for raw_doc in store_reader.iter_raw(reader.delete_bitset()) {
-                        store_writer.store_bytes(raw_doc?.get_bytes())?;
+                if reader.num_deleted_docs() > 0
+                    // If there is not enough data in the store, we avoid stacking in order to
+                    // avoid creating many small blocks in the doc store. Once we have 5 full blocks,
+                    // we start stacking. In the worst case 2/7 of the blocks would be very small.
+                    // [segment 1 - {1 doc}][segment 2 - {fullblock * 5}{1doc}]
+                    // => 5 * full blocks, 2 * 1 document blocks
+                    //
+                    // In a more realistic scenario the segments are of the same size, so 1/6 of
+                    // the doc stores would be on average half full, given total randomness (which
+                    // is not the case here, but not sure how it behaves exactly).
+                    //
+                    // https://github.com/tantivy-search/tantivy/issues/1053
+                    //
+                    // take 7 in order to not walk over all checkpoints.
+                    || store_reader.block_checkpoints().take(7).count() < 6
+                    || store_reader.compressor() != store_writer.compressor()
+                {
+                    for doc_bytes_res in store_reader.iter_raw(reader.delete_bitset()) {
+                        let doc_bytes = doc_bytes_res?;
+                        store_writer.store_bytes(&doc_bytes)?;
                     }
                 } else {
                     store_writer.stack(&store_reader)?;
@@ -958,7 +1052,13 @@ impl SerializableSegment for IndexMerger {
     ) -> crate::Result<u32> {
         let doc_id_mapping = if let Some(sort_by_field) = self.index_settings.sort_by_field.as_ref()
         {
-            Some(self.generate_doc_id_mapping(sort_by_field)?)
+            // If the documents are already sorted and stackable, we ignore the mapping and execute
+            // it as if there was no sorting
+            if self.is_disjunct_and_sorted_on_sort_property(sort_by_field)? {
+                None
+            } else {
+                Some(self.generate_doc_id_mapping(sort_by_field)?)
+            }
         } else {
             None
         };
@@ -993,6 +1093,7 @@ mod tests {
     use crate::collector::tests::{BytesFastFieldTestCollector, FastFieldTestCollector};
     use crate::collector::{Count, FacetCollector};
     use crate::core::Index;
+    use crate::fastfield::FastFieldReader;
     use crate::query::AllQuery;
     use crate::query::BooleanQuery;
     use crate::query::Scorer;
@@ -1470,31 +1571,65 @@ mod tests {
     }
     #[test]
     fn test_merge_facets_sort_none() {
-        test_merge_facets(None)
+        test_merge_facets(None, true)
     }
 
     #[test]
     fn test_merge_facets_sort_asc() {
-        // the data is already sorted asc, so this should have no effect, but go through the docid
-        // mapping code
-        test_merge_facets(Some(IndexSettings {
-            sort_by_field: Some(IndexSortByField {
-                field: "intval".to_string(),
-                order: Order::Asc,
+        // In the merge case this will go through the docid mapping code
+        test_merge_facets(
+            Some(IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "intval".to_string(),
+                    order: Order::Desc,
+                }),
+                ..Default::default()
             }),
-        }));
+            true,
+        );
+        // In the merge case this will not go through the docid mapping code, because the data is
+        // sorted and disjunct
+        test_merge_facets(
+            Some(IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "intval".to_string(),
+                    order: Order::Desc,
+                }),
+                ..Default::default()
+            }),
+            false,
+        );
     }
 
     #[test]
     fn test_merge_facets_sort_desc() {
-        test_merge_facets(Some(IndexSettings {
-            sort_by_field: Some(IndexSortByField {
-                field: "intval".to_string(),
-                order: Order::Desc,
+        // In the merge case this will go through the docid mapping code
+        test_merge_facets(
+            Some(IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "intval".to_string(),
+                    order: Order::Desc,
+                }),
+                ..Default::default()
             }),
-        }));
+            true,
+        );
+        // In the merge case this will not go through the docid mapping code, because the data is
+        // sorted and disjunct
+        test_merge_facets(
+            Some(IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "intval".to_string(),
+                    order: Order::Desc,
+                }),
+                ..Default::default()
+            }),
+            false,
+        );
     }
-    fn test_merge_facets(index_settings: Option<IndexSettings>) {
+    // force_segment_value_overlap forces the int value for sorting to have overlapping min and max
+    // ranges between segments so that merge algorithm can't apply certain optimizations
+    fn test_merge_facets(index_settings: Option<IndexSettings>, force_segment_value_overlap: bool) {
         let mut schema_builder = schema::Schema::builder();
         let facet_field = schema_builder.add_facet_field("facet", INDEXED);
         let int_options = IntOptions::default()
@@ -1511,32 +1646,47 @@ mod tests {
         let mut int_val = 0;
         {
             let mut index_writer = index.writer_for_tests().unwrap();
-            let mut index_doc = |index_writer: &mut IndexWriter, doc_facets: &[&str]| {
-                let mut doc = Document::default();
-                for facet in doc_facets {
-                    doc.add_facet(facet_field, Facet::from(facet));
-                }
-                doc.add_u64(int_field, int_val);
-                int_val += 1;
-                index_writer.add_document(doc);
-            };
+            let index_doc =
+                |index_writer: &mut IndexWriter, doc_facets: &[&str], int_val: &mut u64| {
+                    let mut doc = Document::default();
+                    for facet in doc_facets {
+                        doc.add_facet(facet_field, Facet::from(facet));
+                    }
+                    doc.add_u64(int_field, *int_val);
+                    *int_val += 1;
+                    index_writer.add_document(doc);
+                };
 
-            index_doc(&mut index_writer, &["/top/a/firstdoc", "/top/b"]);
-            index_doc(&mut index_writer, &["/top/a/firstdoc", "/top/b", "/top/c"]);
-            index_doc(&mut index_writer, &["/top/a", "/top/b"]);
-            index_doc(&mut index_writer, &["/top/a"]);
+            index_doc(
+                &mut index_writer,
+                &["/top/a/firstdoc", "/top/b"],
+                &mut int_val,
+            );
+            index_doc(
+                &mut index_writer,
+                &["/top/a/firstdoc", "/top/b", "/top/c"],
+                &mut int_val,
+            );
+            index_doc(&mut index_writer, &["/top/a", "/top/b"], &mut int_val);
+            index_doc(&mut index_writer, &["/top/a"], &mut int_val);
 
-            index_doc(&mut index_writer, &["/top/b", "/top/d"]);
-            index_doc(&mut index_writer, &["/top/d"]);
-            index_doc(&mut index_writer, &["/top/e"]);
+            index_doc(&mut index_writer, &["/top/b", "/top/d"], &mut int_val);
+            if force_segment_value_overlap {
+                index_doc(&mut index_writer, &["/top/d"], &mut 0);
+                index_doc(&mut index_writer, &["/top/e"], &mut 10);
+                index_writer.commit().expect("committed");
+                index_doc(&mut index_writer, &["/top/a"], &mut 5); // 5 is between 0 - 10 so the segments don' have disjunct ranges
+            } else {
+                index_doc(&mut index_writer, &["/top/d"], &mut int_val);
+                index_doc(&mut index_writer, &["/top/e"], &mut int_val);
+                index_writer.commit().expect("committed");
+                index_doc(&mut index_writer, &["/top/a"], &mut int_val);
+            }
+            index_doc(&mut index_writer, &["/top/b"], &mut int_val);
+            index_doc(&mut index_writer, &["/top/c"], &mut int_val);
             index_writer.commit().expect("committed");
 
-            index_doc(&mut index_writer, &["/top/a"]);
-            index_doc(&mut index_writer, &["/top/b"]);
-            index_doc(&mut index_writer, &["/top/c"]);
-            index_writer.commit().expect("committed");
-
-            index_doc(&mut index_writer, &["/top/e", "/top/f"]);
+            index_doc(&mut index_writer, &["/top/e", "/top/f"], &mut int_val);
             index_writer.commit().expect("committed");
         }
 
@@ -1821,7 +1971,7 @@ mod tests {
 
         // Make sure we'll attempt to merge every created segment
         let mut policy = crate::indexer::LogMergePolicy::default();
-        policy.set_min_merge_size(2);
+        policy.set_min_num_segments(2);
         writer.set_merge_policy(Box::new(policy));
 
         for i in 0..100 {

src/indexer/merger_sorted_index_test.rs

@@ -1,5 +1,6 @@
 #[cfg(test)]
 mod tests {
+    use crate::fastfield::FastFieldReader;
     use crate::{
         collector::TopDocs,
         schema::{Cardinality, TextFieldIndexing},
@@ -39,6 +40,7 @@ mod tests {
             let mut index_writer = index.writer_for_tests().unwrap();
 
             index_writer.add_document(doc!(int_field=>3_u64, facet_field=> Facet::from("/crime")));
+            index_writer.add_document(doc!(int_field=>6_u64, facet_field=> Facet::from("/crime")));
 
             assert!(index_writer.commit().is_ok());
             index_writer.add_document(doc!(int_field=>5_u64, facet_field=> Facet::from("/fanta")));
@@ -58,7 +60,12 @@ mod tests {
         index
     }
 
-    fn create_test_index(index_settings: Option<IndexSettings>) -> Index {
+    // force_disjunct_segment_sort_values forces the field, by which the index is sorted have disjunct
+    // ranges between segments, e.g. values in segment [1-3] [10 - 20] [50 - 500]
+    fn create_test_index(
+        index_settings: Option<IndexSettings>,
+        force_disjunct_segment_sort_values: bool,
+    ) -> Index {
         let mut schema_builder = schema::Schema::builder();
         let int_options = IntOptions::default()
             .set_fast(Cardinality::SingleValue)
@@ -92,6 +99,7 @@ mod tests {
         {
             let mut index_writer = index.writer_for_tests().unwrap();
 
+            // segment 1 - range 1-3
             index_writer.add_document(doc!(int_field=>1_u64));
             index_writer.add_document(
                 doc!(int_field=>3_u64, multi_numbers => 3_u64, multi_numbers => 4_u64, bytes_field => vec![1, 2, 3], text_field => "some text", facet_field=> Facet::from("/book/crime")),
@@ -102,13 +110,26 @@ mod tests {
             );
 
             assert!(index_writer.commit().is_ok());
+            // segment 2 - range 1-20 , with force_disjunct_segment_sort_values 10-20
             index_writer.add_document(doc!(int_field=>20_u64, multi_numbers => 20_u64));
-            index_writer.add_document(doc!(int_field=>1_u64, text_field=> "deleteme", facet_field=> Facet::from("/book/crime")));
+
+            let in_val = if force_disjunct_segment_sort_values {
+                10_u64
+            } else {
+                1
+            };
+            index_writer.add_document(doc!(int_field=>in_val, text_field=> "deleteme", facet_field=> Facet::from("/book/crime")));
             assert!(index_writer.commit().is_ok());
-            index_writer.add_document(
-                doc!(int_field=>10_u64, multi_numbers => 10_u64, multi_numbers => 11_u64, text_field=> "blubber", facet_field=> Facet::from("/book/fantasy")),
+            // segment 3 - range 5-1000, with force_disjunct_segment_sort_values 50-1000
+            let int_vals = if force_disjunct_segment_sort_values {
+                [100_u64, 50]
+            } else {
+                [10, 5]
+            };
+            index_writer.add_document( // position of this doc after delete in desc sorting = [2], in disjunct case [1]
+                doc!(int_field=>int_vals[0], multi_numbers => 10_u64, multi_numbers => 11_u64, text_field=> "blubber", facet_field=> Facet::from("/book/fantasy")),
             );
-            index_writer.add_document(doc!(int_field=>5_u64, text_field=> "deleteme"));
+            index_writer.add_document(doc!(int_field=>int_vals[1], text_field=> "deleteme"));
             index_writer.add_document(
                 doc!(int_field=>1_000u64, multi_numbers => 1001_u64, multi_numbers => 1002_u64, bytes_field => vec![5, 5],text_field => "the biggest num")
             );
@@ -136,17 +157,30 @@ mod tests {
                 field: "intval".to_string(),
                 order: Order::Desc,
             }),
+            ..Default::default()
         }));
     }
 
     #[test]
-    fn test_merge_sorted_index_desc() {
-        let index = create_test_index(Some(IndexSettings {
-            sort_by_field: Some(IndexSortByField {
-                field: "intval".to_string(),
-                order: Order::Desc,
+    fn test_merge_sorted_index_desc_not_disjunct() {
+        test_merge_sorted_index_desc_(false);
+    }
+    #[test]
+    fn test_merge_sorted_index_desc_disjunct() {
+        test_merge_sorted_index_desc_(true);
+    }
+
+    fn test_merge_sorted_index_desc_(force_disjunct_segment_sort_values: bool) {
+        let index = create_test_index(
+            Some(IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "intval".to_string(),
+                    order: Order::Desc,
+                }),
+                ..Default::default()
             }),
-        }));
+            force_disjunct_segment_sort_values,
+        );
 
         let int_field = index.schema().get_field("intval").unwrap();
         let reader = index.reader().unwrap();
@@ -160,8 +194,13 @@ mod tests {
         assert_eq!(fast_field.get(5u32), 1u64);
         assert_eq!(fast_field.get(4u32), 2u64);
         assert_eq!(fast_field.get(3u32), 3u64);
-        assert_eq!(fast_field.get(2u32), 10u64);
-        assert_eq!(fast_field.get(1u32), 20u64);
+        if force_disjunct_segment_sort_values {
+            assert_eq!(fast_field.get(2u32), 20u64);
+            assert_eq!(fast_field.get(1u32), 100u64);
+        } else {
+            assert_eq!(fast_field.get(2u32), 10u64);
+            assert_eq!(fast_field.get(1u32), 20u64);
+        }
         assert_eq!(fast_field.get(0u32), 1_000u64);
 
         // test new field norm mapping
@@ -169,8 +208,13 @@ mod tests {
             let my_text_field = index.schema().get_field("text_field").unwrap();
             let fieldnorm_reader = segment_reader.get_fieldnorms_reader(my_text_field).unwrap();
             assert_eq!(fieldnorm_reader.fieldnorm(0), 3); // the biggest num
-            assert_eq!(fieldnorm_reader.fieldnorm(1), 0);
-            assert_eq!(fieldnorm_reader.fieldnorm(2), 1); // blubber
+            if force_disjunct_segment_sort_values {
+                assert_eq!(fieldnorm_reader.fieldnorm(1), 1); // blubber
+                assert_eq!(fieldnorm_reader.fieldnorm(2), 0);
+            } else {
+                assert_eq!(fieldnorm_reader.fieldnorm(1), 0);
+                assert_eq!(fieldnorm_reader.fieldnorm(2), 1); // blubber
+            }
             assert_eq!(fieldnorm_reader.fieldnorm(3), 2); // some text
             assert_eq!(fieldnorm_reader.fieldnorm(5), 0);
         }
@@ -191,13 +235,22 @@ mod tests {
             };
 
             assert_eq!(do_search("some"), vec![3]);
-            assert_eq!(do_search("blubber"), vec![2]);
+            if force_disjunct_segment_sort_values {
+                assert_eq!(do_search("blubber"), vec![1]);
+            } else {
+                assert_eq!(do_search("blubber"), vec![2]);
+            }
             assert_eq!(do_search("biggest"), vec![0]);
         }
 
         // access doc store
         {
-            let doc = searcher.doc(DocAddress::new(0, 2)).unwrap();
+            let blubber_pos = if force_disjunct_segment_sort_values {
+                1
+            } else {
+                2
+            };
+            let doc = searcher.doc(DocAddress::new(0, blubber_pos)).unwrap();
             assert_eq!(
                 doc.get_first(my_text_field).unwrap().text(),
                 Some("blubber")
@@ -209,12 +262,16 @@ mod tests {
 
     #[test]
     fn test_merge_sorted_index_asc() {
-        let index = create_test_index(Some(IndexSettings {
-            sort_by_field: Some(IndexSortByField {
-                field: "intval".to_string(),
-                order: Order::Asc,
+        let index = create_test_index(
+            Some(IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "intval".to_string(),
+                    order: Order::Asc,
+                }),
+                ..Default::default()
             }),
-        }));
+            false,
+        );
 
         let int_field = index.schema().get_field("intval").unwrap();
         let multi_numbers = index.schema().get_field("multi_numbers").unwrap();
@@ -315,7 +372,6 @@ mod bench_sorted_index_merge {
     use crate::IndexSortByField;
     use crate::IndexWriter;
     use crate::Order;
-    use futures::executor::block_on;
     use test::{self, Bencher};
     fn create_index(sort_by_field: Option<IndexSortByField>) -> Index {
         let mut schema_builder = Schema::builder();
@@ -323,12 +379,12 @@ mod bench_sorted_index_merge {
             .set_fast(Cardinality::SingleValue)
             .set_indexed();
         let int_field = schema_builder.add_u64_field("intval", int_options);
-        let int_field = schema_builder.add_u64_field("intval", int_options);
         let schema = schema_builder.build();
 
-        let index_builder = Index::builder()
-            .schema(schema)
-            .settings(IndexSettings { sort_by_field });
+        let index_builder = Index::builder().schema(schema).settings(IndexSettings {
+            sort_by_field,
+            ..Default::default()
+        });
         let index = index_builder.create_in_ram().unwrap();
 
         {
@@ -366,7 +422,7 @@ mod bench_sorted_index_merge {
         b.iter(|| {
 
             let sorted_doc_ids = doc_id_mapping.iter().map(|(doc_id, reader)|{
-            let u64_reader: FastFieldReader<u64> = reader
+            let u64_reader: FastFieldReader<u64> = reader.reader
                 .fast_fields()
                 .typed_fast_field_reader(field)
                 .expect("Failed to find a reader for single fast field. This is a tantivy bug and it should never happen.");
@@ -391,7 +447,7 @@ mod bench_sorted_index_merge {
             order: Order::Desc,
         };
         let index = create_index(Some(sort_by_field.clone()));
-        let field = index.schema().get_field("intval").unwrap();
+        //let field = index.schema().get_field("intval").unwrap();
         let segments = index.searchable_segments().unwrap();
         let merger: IndexMerger =
             IndexMerger::open(index.schema(), index.settings().clone(), &segments[..])?;

src/indexer/segment_serializer.rs

@@ -1,6 +1,6 @@
 use crate::core::Segment;
 use crate::core::SegmentComponent;
-use crate::fastfield::FastFieldSerializer;
+use crate::fastfield::CompositeFastFieldSerializer;
 use crate::fieldnorm::FieldNormsSerializer;
 use crate::postings::InvertedIndexSerializer;
 use crate::store::StoreWriter;
@@ -10,7 +10,7 @@ use crate::store::StoreWriter;
 pub struct SegmentSerializer {
     segment: Segment,
     pub(crate) store_writer: StoreWriter,
-    fast_field_serializer: FastFieldSerializer,
+    fast_field_serializer: CompositeFastFieldSerializer,
     fieldnorms_serializer: Option<FieldNormsSerializer>,
     postings_serializer: InvertedIndexSerializer,
 }
@@ -33,15 +33,16 @@ impl SegmentSerializer {
         let store_write = segment.open_write(store_component)?;
 
         let fast_field_write = segment.open_write(SegmentComponent::FastFields)?;
-        let fast_field_serializer = FastFieldSerializer::from_write(fast_field_write)?;
+        let fast_field_serializer = CompositeFastFieldSerializer::from_write(fast_field_write)?;
 
         let fieldnorms_write = segment.open_write(SegmentComponent::FieldNorms)?;
         let fieldnorms_serializer = FieldNormsSerializer::from_write(fieldnorms_write)?;
 
         let postings_serializer = InvertedIndexSerializer::open(&mut segment)?;
+        let compressor = segment.index().settings().docstore_compression;
         Ok(SegmentSerializer {
             segment,
-            store_writer: StoreWriter::new(store_write),
+            store_writer: StoreWriter::new(store_write, compressor),
             fast_field_serializer,
             fieldnorms_serializer: Some(fieldnorms_serializer),
             postings_serializer,
@@ -67,7 +68,7 @@ impl SegmentSerializer {
     }
 
     /// Accessor to the `FastFieldSerializer`.
-    pub fn get_fast_field_serializer(&mut self) -> &mut FastFieldSerializer {
+    pub fn get_fast_field_serializer(&mut self) -> &mut CompositeFastFieldSerializer {
         &mut self.fast_field_serializer
     }
 

src/indexer/segment_writer.rs

@@ -345,8 +345,11 @@ fn write(
         let store_write = serializer
             .segment_mut()
             .open_write(SegmentComponent::Store)?;
-        let old_store_writer =
-            std::mem::replace(&mut serializer.store_writer, StoreWriter::new(store_write));
+        let compressor = serializer.segment().index().settings().docstore_compression;
+        let old_store_writer = std::mem::replace(
+            &mut serializer.store_writer,
+            StoreWriter::new(store_write, compressor),
+        );
         old_store_writer.close()?;
         let store_read = StoreReader::open(
             serializer
@@ -354,12 +357,9 @@ fn write(
                 .open_read(SegmentComponent::TempStore)?,
         )?;
         for old_doc_id in doc_id_map.iter_old_doc_ids() {
-            let raw_doc = store_read.get_raw(*old_doc_id)?;
-            serializer
-                .get_store_writer()
-                .store_bytes(raw_doc.get_bytes())?;
+            let doc_bytes = store_read.get_document_bytes(*old_doc_id)?;
+            serializer.get_store_writer().store_bytes(&doc_bytes)?;
         }
-        // TODO delete temp store
     }
     serializer.close()?;
     Ok(())

src/lib.rs

@@ -178,7 +178,7 @@ use once_cell::sync::Lazy;
 use serde::{Deserialize, Serialize};
 
 /// Index format version.
-const INDEX_FORMAT_VERSION: u32 = 3;
+const INDEX_FORMAT_VERSION: u32 = 4;
 
 /// Structure version for the index.
 #[derive(Clone, PartialEq, Eq, Serialize, Deserialize)]
@@ -187,7 +187,6 @@ pub struct Version {
     minor: u32,
     patch: u32,
     index_format_version: u32,
-    store_compression: String,
 }
 
 impl fmt::Debug for Version {
@@ -201,14 +200,13 @@ static VERSION: Lazy<Version> = Lazy::new(|| Version {
     minor: env!("CARGO_PKG_VERSION_MINOR").parse().unwrap(),
     patch: env!("CARGO_PKG_VERSION_PATCH").parse().unwrap(),
     index_format_version: INDEX_FORMAT_VERSION,
-    store_compression: crate::store::COMPRESSION.to_string(),
 });
 
 impl ToString for Version {
     fn to_string(&self) -> String {
         format!(
-            "tantivy v{}.{}.{}, index_format v{}, store_compression: {}",
-            self.major, self.minor, self.patch, self.index_format_version, self.store_compression
+            "tantivy v{}.{}.{}, index_format v{}",
+            self.major, self.minor, self.patch, self.index_format_version
         )
     }
 }
@@ -293,6 +291,7 @@ mod tests {
     use crate::collector::tests::TEST_COLLECTOR_WITH_SCORE;
     use crate::core::SegmentReader;
     use crate::docset::{DocSet, TERMINATED};
+    use crate::fastfield::FastFieldReader;
     use crate::query::BooleanQuery;
     use crate::schema::*;
     use crate::DocAddress;

src/query/boolean_query/boolean_query.rs

@@ -6,7 +6,7 @@ use crate::query::Weight;
 use crate::schema::IndexRecordOption;
 use crate::schema::Term;
 use crate::Searcher;
-use std::collections::BTreeSet;
+use std::collections::BTreeMap;
 
 /// The boolean query returns a set of documents
 /// that matches the Boolean combination of constituent subqueries.
@@ -159,9 +159,9 @@ impl Query for BooleanQuery {
         Ok(Box::new(BooleanWeight::new(sub_weights, scoring_enabled)))
     }
 
-    fn query_terms(&self, term_set: &mut BTreeSet<Term>) {
+    fn query_terms(&self, terms: &mut BTreeMap<Term, bool>) {
         for (_occur, subquery) in &self.subqueries {
-            subquery.query_terms(term_set);
+            subquery.query_terms(terms);
         }
     }
 }

src/query/boost_query.rs

@@ -2,7 +2,7 @@ use crate::fastfield::DeleteBitSet;
 use crate::query::explanation::does_not_match;
 use crate::query::{Explanation, Query, Scorer, Weight};
 use crate::{DocId, DocSet, Score, Searcher, SegmentReader, Term};
-use std::collections::BTreeSet;
+use std::collections::BTreeMap;
 use std::fmt;
 
 /// `BoostQuery` is a wrapper over a query used to boost its score.
@@ -48,8 +48,8 @@ impl Query for BoostQuery {
         Ok(boosted_weight)
     }
 
-    fn query_terms(&self, term_set: &mut BTreeSet<Term>) {
-        self.query.query_terms(term_set)
+    fn query_terms(&self, terms: &mut BTreeMap<Term, bool>) {
+        self.query.query_terms(terms)
     }
 }
 

src/query/mod.rs

@@ -11,7 +11,7 @@ mod exclude;
 mod explanation;
 mod fuzzy_query;
 mod intersection;
-mod mlt;
+mod more_like_this;
 mod phrase_query;
 mod query;
 mod query_parser;
@@ -46,7 +46,7 @@ pub use self::explanation::Explanation;
 pub(crate) use self::fuzzy_query::DfaWrapper;
 pub use self::fuzzy_query::FuzzyTermQuery;
 pub use self::intersection::intersect_scorers;
-pub use self::mlt::{MoreLikeThisQuery, MoreLikeThisQueryBuilder};
+pub use self::more_like_this::{MoreLikeThisQuery, MoreLikeThisQueryBuilder};
 pub use self::phrase_query::PhraseQuery;
 pub use self::query::{Query, QueryClone};
 pub use self::query_parser::QueryParser;
@@ -66,7 +66,7 @@ mod tests {
     use crate::schema::{Schema, TEXT};
     use crate::Index;
     use crate::Term;
-    use std::collections::BTreeSet;
+    use std::collections::BTreeMap;
 
     #[test]
     fn test_query_terms() {
@@ -78,49 +78,49 @@ mod tests {
         let term_a = Term::from_field_text(text_field, "a");
         let term_b = Term::from_field_text(text_field, "b");
         {
-            let mut terms_set: BTreeSet<Term> = BTreeSet::new();
+            let mut terms: BTreeMap<Term, bool> = Default::default();
             query_parser
                 .parse_query("a")
                 .unwrap()
-                .query_terms(&mut terms_set);
-            let terms: Vec<&Term> = terms_set.iter().collect();
-            assert_eq!(vec![&term_a], terms);
+                .query_terms(&mut terms);
+            let terms: Vec<(&Term, &bool)> = terms.iter().collect();
+            assert_eq!(vec![(&term_a, &false)], terms);
         }
         {
-            let mut terms_set: BTreeSet<Term> = BTreeSet::new();
+            let mut terms: BTreeMap<Term, bool> = Default::default();
             query_parser
                 .parse_query("a b")
                 .unwrap()
-                .query_terms(&mut terms_set);
-            let terms: Vec<&Term> = terms_set.iter().collect();
-            assert_eq!(vec![&term_a, &term_b], terms);
+                .query_terms(&mut terms);
+            let terms: Vec<(&Term, &bool)> = terms.iter().collect();
+            assert_eq!(vec![(&term_a, &false), (&term_b, &false)], terms);
         }
         {
-            let mut terms_set: BTreeSet<Term> = BTreeSet::new();
+            let mut terms: BTreeMap<Term, bool> = Default::default();
             query_parser
                 .parse_query("\"a b\"")
                 .unwrap()
-                .query_terms(&mut terms_set);
-            let terms: Vec<&Term> = terms_set.iter().collect();
-            assert_eq!(vec![&term_a, &term_b], terms);
+                .query_terms(&mut terms);
+            let terms: Vec<(&Term, &bool)> = terms.iter().collect();
+            assert_eq!(vec![(&term_a, &true), (&term_b, &true)], terms);
         }
         {
-            let mut terms_set: BTreeSet<Term> = BTreeSet::new();
+            let mut terms: BTreeMap<Term, bool> = Default::default();
             query_parser
                 .parse_query("a a a a a")
                 .unwrap()
-                .query_terms(&mut terms_set);
-            let terms: Vec<&Term> = terms_set.iter().collect();
-            assert_eq!(vec![&term_a], terms);
+                .query_terms(&mut terms);
+            let terms: Vec<(&Term, &bool)> = terms.iter().collect();
+            assert_eq!(vec![(&term_a, &false)], terms);
         }
         {
-            let mut terms_set: BTreeSet<Term> = BTreeSet::new();
+            let mut terms: BTreeMap<Term, bool> = Default::default();
             query_parser
                 .parse_query("a -b")
                 .unwrap()
-                .query_terms(&mut terms_set);
-            let terms: Vec<&Term> = terms_set.iter().collect();
-            assert_eq!(vec![&term_a, &term_b], terms);
+                .query_terms(&mut terms);
+            let terms: Vec<(&Term, &bool)> = terms.iter().collect();
+            assert_eq!(vec![(&term_a, &false), (&term_b, &false)], terms);
         }
     }
 }

src/query/more_like_this/mod.rs

@@ -1,5 +1,5 @@
-mod mlt;
+mod more_like_this;
 mod query;
 
-pub use self::mlt::MoreLikeThis;
+pub use self::more_like_this::MoreLikeThis;
 pub use self::query::{MoreLikeThisQuery, MoreLikeThisQueryBuilder};

src/query/more_like_this/more_like_this.rs

@@ -233,10 +233,9 @@ impl MoreLikeThis {
             }
             FieldType::U64(_) => {
                 for field_value in field_values {
-                    let val = field_value
-                        .value()
-                        .u64_value()
-                        .ok_or(TantivyError::InvalidArgument("invalid value".to_string()))?;
+                    let val = field_value.value().u64_value().ok_or_else(|| {
+                        TantivyError::InvalidArgument("invalid value".to_string())
+                    })?;
                     if !self.is_noise_word(val.to_string()) {
                         let term = Term::from_field_u64(field, val);
                         *term_frequencies.entry(term).or_insert(0) += 1;
@@ -249,7 +248,7 @@ impl MoreLikeThis {
                     let val = field_value
                         .value()
                         .date_value()
-                        .ok_or(TantivyError::InvalidArgument("invalid value".to_string()))?
+                        .ok_or_else(|| TantivyError::InvalidArgument("invalid value".to_string()))?
                         .timestamp();
                     if !self.is_noise_word(val.to_string()) {
                         let term = Term::from_field_i64(field, val);
@@ -259,10 +258,9 @@ impl MoreLikeThis {
             }
             FieldType::I64(_) => {
                 for field_value in field_values {
-                    let val = field_value
-                        .value()
-                        .i64_value()
-                        .ok_or(TantivyError::InvalidArgument("invalid value".to_string()))?;
+                    let val = field_value.value().i64_value().ok_or_else(|| {
+                        TantivyError::InvalidArgument("invalid value".to_string())
+                    })?;
                     if !self.is_noise_word(val.to_string()) {
                         let term = Term::from_field_i64(field, val);
                         *term_frequencies.entry(term).or_insert(0) += 1;
@@ -271,10 +269,9 @@ impl MoreLikeThis {
             }
             FieldType::F64(_) => {
                 for field_value in field_values {
-                    let val = field_value
-                        .value()
-                        .f64_value()
-                        .ok_or(TantivyError::InvalidArgument("invalid value".to_string()))?;
+                    let val = field_value.value().f64_value().ok_or_else(|| {
+                        TantivyError::InvalidArgument("invalid value".to_string())
+                    })?;
                     if !self.is_noise_word(val.to_string()) {
                         let term = Term::from_field_f64(field, val);
                         *term_frequencies.entry(term).or_insert(0) += 1;
@@ -306,7 +303,7 @@ impl MoreLikeThis {
         {
             return true;
         }
-        return self.stop_words.contains(&word);
+        self.stop_words.contains(&word)
     }
 
     /// Couputes the score for each term while ignoring not useful terms

src/query/phrase_query/phrase_query.rs

@@ -1,3 +1,5 @@
+use std::collections::BTreeMap;
+
 use super::PhraseWeight;
 use crate::core::searcher::Searcher;
 use crate::query::bm25::Bm25Weight;
@@ -5,7 +7,6 @@ use crate::query::Query;
 use crate::query::Weight;
 use crate::schema::IndexRecordOption;
 use crate::schema::{Field, Term};
-use std::collections::BTreeSet;
 
 /// `PhraseQuery` matches a specific sequence of words.
 ///
@@ -113,9 +114,9 @@ impl Query for PhraseQuery {
         Ok(Box::new(phrase_weight))
     }
 
-    fn query_terms(&self, term_set: &mut BTreeSet<Term>) {
-        for (_, query_term) in &self.phrase_terms {
-            term_set.insert(query_term.clone());
+    fn query_terms(&self, terms: &mut BTreeMap<Term, bool>) {
+        for (_, term) in &self.phrase_terms {
+            terms.insert(term.clone(), true);
         }
     }
 }

src/query/query.rs

@@ -4,7 +4,7 @@ use crate::query::Explanation;
 use crate::DocAddress;
 use crate::Term;
 use downcast_rs::impl_downcast;
-use std::collections::BTreeSet;
+use std::collections::BTreeMap;
 use std::fmt;
 
 /// The `Query` trait defines a set of documents and a scoring method
@@ -68,7 +68,10 @@ pub trait Query: QueryClone + Send + Sync + downcast_rs::Downcast + fmt::Debug {
 
     /// Extract all of the terms associated to the query and insert them in the
     /// term set given in arguments.
-    fn query_terms(&self, _term_set: &mut BTreeSet<Term>) {}
+    ///
+    /// Each term is associated with a boolean indicating whether
+    /// Positions are required or not.
+    fn query_terms(&self, _term_set: &mut BTreeMap<Term, bool>) {}
 }
 
 /// Implements `box_clone`.
@@ -95,8 +98,8 @@ impl Query for Box<dyn Query> {
         self.as_ref().count(searcher)
     }
 
-    fn query_terms(&self, term_set: &mut BTreeSet<Term<Vec<u8>>>) {
-        self.as_ref().query_terms(term_set);
+    fn query_terms(&self, terms: &mut BTreeMap<Term, bool>) {
+        self.as_ref().query_terms(terms);
     }
 }
 

src/query/query_parser/query_parser.rs

@@ -8,7 +8,7 @@ use crate::query::Query;
 use crate::query::RangeQuery;
 use crate::query::TermQuery;
 use crate::query::{AllQuery, BoostQuery};
-use crate::schema::{Facet, IndexRecordOption};
+use crate::schema::{Facet, FacetParseError, IndexRecordOption};
 use crate::schema::{Field, Schema};
 use crate::schema::{FieldType, Term};
 use crate::tokenizer::TokenizerManager;
@@ -68,6 +68,9 @@ pub enum QueryParserError {
     /// The format for the date field is not RFC 3339 compliant.
     #[error("The date field has an invalid format")]
     DateFormatError(chrono::ParseError),
+    /// The format for the facet field is invalid.
+    #[error("The facet field is malformed: {0}")]
+    FacetFormatError(FacetParseError),
 }
 
 impl From<ParseIntError> for QueryParserError {
@@ -88,6 +91,12 @@ impl From<chrono::ParseError> for QueryParserError {
     }
 }
 
+impl From<FacetParseError> for QueryParserError {
+    fn from(err: FacetParseError) -> QueryParserError {
+        QueryParserError::FacetFormatError(err)
+    }
+}
+
 /// Recursively remove empty clause from the AST
 ///
 /// Returns `None` iff the `logical_ast` ended up being empty.
@@ -358,10 +367,10 @@ impl QueryParser {
                     ))
                 }
             }
-            FieldType::HierarchicalFacet(_) => {
-                let facet = Facet::from_text(phrase);
-                Ok(vec![(0, Term::from_field_text(field, facet.encoded_str()))])
-            }
+            FieldType::HierarchicalFacet(_) => match Facet::from_text(phrase) {
+                Ok(facet) => Ok(vec![(0, Term::from_field_text(field, facet.encoded_str()))]),
+                Err(e) => Err(QueryParserError::from(e)),
+            },
             FieldType::Bytes(_) => {
                 let bytes = base64::decode(phrase).map_err(QueryParserError::ExpectedBase64)?;
                 let term = Term::from_field_bytes(field, &bytes);
@@ -1027,6 +1036,19 @@ mod test {
             .is_ok());
     }
 
+    #[test]
+    pub fn test_query_parser_expected_facet() {
+        let query_parser = make_query_parser();
+        match query_parser.parse_query("facet:INVALID") {
+            Ok(_) => panic!("should never succeed"),
+            Err(e) => assert_eq!(
+                "The facet field is malformed: Failed to parse the facet string: 'INVALID'",
+                format!("{}", e)
+            ),
+        }
+        assert!(query_parser.parse_query("facet:\"/foo/bar\"").is_ok());
+    }
+
     #[test]
     pub fn test_query_parser_not_empty_but_no_tokens() {
         let query_parser = make_query_parser();

src/query/term_query/term_query.rs

@@ -5,7 +5,7 @@ use crate::query::{Explanation, Query};
 use crate::schema::IndexRecordOption;
 use crate::Searcher;
 use crate::Term;
-use std::collections::BTreeSet;
+use std::collections::BTreeMap;
 use std::fmt;
 
 /// A Term query matches all of the documents
@@ -127,7 +127,7 @@ impl Query for TermQuery {
             self.specialized_weight(searcher, scoring_enabled)?,
         ))
     }
-    fn query_terms(&self, term_set: &mut BTreeSet<Term>) {
-        term_set.insert(self.term.clone());
+    fn query_terms(&self, terms: &mut BTreeMap<Term, bool>) {
+        terms.insert(self.term.clone(), false);
     }
 }

src/schema/facet.rs

@@ -20,6 +20,14 @@ pub const FACET_SEP_BYTE: u8 = 0u8;
 /// representation of facets. (It is the null codepoint.)
 pub const FACET_SEP_CHAR: char = '\u{0}';
 
+/// An error enum for facet parser.
+#[derive(Debug, PartialEq, Eq, Error)]
+pub enum FacetParseError {
+    /// The facet text representation is unparsable.
+    #[error("Failed to parse the facet string: '{0}'")]
+    FacetParseError(String),
+}
+
 /// A Facet represent a point in a given hierarchy.
 ///
 /// They are typically represented similarly to a filepath.
@@ -75,11 +83,47 @@ impl Facet {
     /// It is conceptually, if one of the steps of this path
     /// contains a `/` or a `\`, it should be escaped
     /// using an anti-slash `/`.
-    pub fn from_text<T>(path: &T) -> Facet
+    pub fn from_text<T>(path: &T) -> Result<Facet, FacetParseError>
     where
         T: ?Sized + AsRef<str>,
     {
-        From::from(path)
+        #[derive(Copy, Clone)]
+        enum State {
+            Escaped,
+            Idle,
+        }
+        let path_ref = path.as_ref();
+        if path_ref.is_empty() {
+            return Err(FacetParseError::FacetParseError(path_ref.to_string()));
+        }
+        if !path_ref.starts_with('/') {
+            return Err(FacetParseError::FacetParseError(path_ref.to_string()));
+        }
+        let mut facet_encoded = String::new();
+        let mut state = State::Idle;
+        let path_bytes = path_ref.as_bytes();
+        let mut last_offset = 1;
+        for i in 1..path_bytes.len() {
+            let c = path_bytes[i];
+            match (state, c) {
+                (State::Idle, ESCAPE_BYTE) => {
+                    facet_encoded.push_str(&path_ref[last_offset..i]);
+                    last_offset = i + 1;
+                    state = State::Escaped
+                }
+                (State::Idle, SLASH_BYTE) => {
+                    facet_encoded.push_str(&path_ref[last_offset..i]);
+                    facet_encoded.push(FACET_SEP_CHAR);
+                    last_offset = i + 1;
+                }
+                (State::Escaped, _escaped_char) => {
+                    state = State::Idle;
+                }
+                (State::Idle, _any_char) => {}
+            }
+        }
+        facet_encoded.push_str(&path_ref[last_offset..]);
+        Ok(Facet(facet_encoded))
     }
 
     /// Returns a `Facet` from an iterator over the different
@@ -137,39 +181,7 @@ impl Borrow<str> for Facet {
 
 impl<'a, T: ?Sized + AsRef<str>> From<&'a T> for Facet {
     fn from(path_asref: &'a T) -> Facet {
-        #[derive(Copy, Clone)]
-        enum State {
-            Escaped,
-            Idle,
-        }
-        let path: &str = path_asref.as_ref();
-        assert!(!path.is_empty());
-        assert!(path.starts_with('/'));
-        let mut facet_encoded = String::new();
-        let mut state = State::Idle;
-        let path_bytes = path.as_bytes();
-        let mut last_offset = 1;
-        for i in 1..path_bytes.len() {
-            let c = path_bytes[i];
-            match (state, c) {
-                (State::Idle, ESCAPE_BYTE) => {
-                    facet_encoded.push_str(&path[last_offset..i]);
-                    last_offset = i + 1;
-                    state = State::Escaped
-                }
-                (State::Idle, SLASH_BYTE) => {
-                    facet_encoded.push_str(&path[last_offset..i]);
-                    facet_encoded.push(FACET_SEP_CHAR);
-                    last_offset = i + 1;
-                }
-                (State::Escaped, _escaped_char) => {
-                    state = State::Idle;
-                }
-                (State::Idle, _any_char) => {}
-            }
-        }
-        facet_encoded.push_str(&path[last_offset..]);
-        Facet(facet_encoded)
+        Facet::from_text(path_asref).unwrap()
     }
 }
 
@@ -226,7 +238,7 @@ impl Debug for Facet {
 #[cfg(test)]
 mod tests {
 
-    use super::Facet;
+    use super::{Facet, FacetParseError};
 
     #[test]
     fn test_root() {
@@ -288,4 +300,12 @@ mod tests {
         let facet = Facet::from_path(v.iter());
         assert_eq!(facet.to_path_string(), "/");
     }
+
+    #[test]
+    fn test_from_text() {
+        assert_eq!(
+            Err(FacetParseError::FacetParseError("INVALID".to_string())),
+            Facet::from_text("INVALID")
+        );
+    }
 }

src/schema/mod.rs

@@ -128,6 +128,7 @@ pub use self::schema::{Schema, SchemaBuilder};
 pub use self::value::Value;
 
 pub use self::facet::Facet;
+pub use self::facet::FacetParseError;
 pub(crate) use self::facet::FACET_SEP_BYTE;
 pub use self::facet_options::FacetOptions;
 

src/snippet/mod.rs

@@ -7,7 +7,6 @@ use crate::{Document, Score};
 use htmlescape::encode_minimal;
 use std::cmp::Ordering;
 use std::collections::BTreeMap;
-use std::collections::BTreeSet;
 use std::ops::Range;
 
 const DEFAULT_MAX_NUM_CHARS: usize = 150;
@@ -239,10 +238,10 @@ impl SnippetGenerator {
         query: &dyn Query,
         field: Field,
     ) -> crate::Result<SnippetGenerator> {
-        let mut terms = BTreeSet::new();
+        let mut terms = BTreeMap::new();
         query.query_terms(&mut terms);
         let mut terms_text: BTreeMap<String, Score> = Default::default();
-        for term in terms {
+        for (term, _) in terms {
             if term.field() != field {
                 continue;
             }

src/store/compression_brotli.rs

@@ -1,10 +1,6 @@
 use std::io;
 
-/// Name of the compression scheme used in the doc store.
-///
-/// This name is appended to the version string of tantivy.
-pub const COMPRESSION: &'static str = "brotli";
-
+#[inline]
 pub fn compress(mut uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()> {
     let mut params = brotli::enc::BrotliEncoderParams::default();
     params.quality = 5;
@@ -13,6 +9,7 @@ pub fn compress(mut uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result
     Ok(())
 }
 
+#[inline]
 pub fn decompress(mut compressed: &[u8], decompressed: &mut Vec<u8>) -> io::Result<()> {
     decompressed.clear();
     brotli::BrotliDecompress(&mut compressed, decompressed)?;

src/store/compression_lz4.rs

@@ -1,22 +0,0 @@
-use std::io::{self, Read, Write};
-
-/// Name of the compression scheme used in the doc store.
-///
-/// This name is appended to the version string of tantivy.
-pub const COMPRESSION: &str = "lz4";
-
-pub fn compress(uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()> {
-    compressed.clear();
-    let mut encoder = lz4::EncoderBuilder::new().build(compressed)?;
-    encoder.write_all(&uncompressed)?;
-    let (_, encoder_result) = encoder.finish();
-    encoder_result?;
-    Ok(())
-}
-
-pub fn decompress(compressed: &[u8], decompressed: &mut Vec<u8>) -> io::Result<()> {
-    decompressed.clear();
-    let mut decoder = lz4::Decoder::new(compressed)?;
-    decoder.read_to_end(decompressed)?;
-    Ok(())
-}

src/store/compression_lz4_block.rs

@@ -2,38 +2,46 @@ use std::io::{self};
 
 use core::convert::TryInto;
 use lz4_flex::{compress_into, decompress_into};
-/// Name of the compression scheme used in the doc store.
-///
-/// This name is appended to the version string of tantivy.
-pub const COMPRESSION: &str = "lz4_block";
 
+#[inline]
 pub fn compress(uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()> {
     compressed.clear();
+    let maximum_ouput_size = lz4_flex::block::get_maximum_output_size(uncompressed.len());
+    compressed.reserve(maximum_ouput_size);
 
-    compressed.extend_from_slice(&[0, 0, 0, 0]);
-    compress_into(uncompressed, compressed);
+    unsafe {
+        compressed.set_len(maximum_ouput_size + 4);
+    }
+    let bytes_written = compress_into(uncompressed, compressed, 4)
+        .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err.to_string()))?;
     let num_bytes = uncompressed.len() as u32;
     compressed[0..4].copy_from_slice(&num_bytes.to_le_bytes());
+    unsafe {
+        compressed.set_len(bytes_written + 4);
+    }
     Ok(())
 }
 
+#[inline]
 pub fn decompress(compressed: &[u8], decompressed: &mut Vec<u8>) -> io::Result<()> {
     decompressed.clear();
-    //next lz4_flex version will support slice as input parameter.
-    //this will make the usage much less ugly
     let uncompressed_size_bytes: &[u8; 4] = compressed
         .get(..4)
         .ok_or(io::ErrorKind::InvalidData)?
         .try_into()
         .unwrap();
     let uncompressed_size = u32::from_le_bytes(*uncompressed_size_bytes) as usize;
-    // reserve more than required, because blocked writes may write out of bounds, will be improved
-    // with lz4_flex 1.0
-    decompressed.reserve(uncompressed_size + 4 + 24);
+    decompressed.reserve(uncompressed_size);
     unsafe {
         decompressed.set_len(uncompressed_size);
     }
-    decompress_into(&compressed[4..], decompressed)
+    let bytes_written = decompress_into(&compressed[4..], decompressed, 0)
         .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err.to_string()))?;
+    if bytes_written != uncompressed_size {
+        return Err(io::Error::new(
+            io::ErrorKind::InvalidData,
+            "doc store block not completely decompressed, data corruption".to_string(),
+        ));
+    }
     Ok(())
 }

src/store/compression_snap.rs

@@ -1,10 +1,6 @@
 use std::io::{self, Read, Write};
 
-/// Name of the compression scheme used in the doc store.
-///
-/// This name is appended to the version string of tantivy.
-pub const COMPRESSION: &str = "snappy";
-
+#[inline]
 pub fn compress(uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()> {
     compressed.clear();
     let mut encoder = snap::write::FrameEncoder::new(compressed);
@@ -13,6 +9,7 @@ pub fn compress(uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()>
     Ok(())
 }
 
+#[inline]
 pub fn decompress(compressed: &[u8], decompressed: &mut Vec<u8>) -> io::Result<()> {
     decompressed.clear();
     snap::read::FrameDecoder::new(compressed).read_to_end(decompressed)?;

src/store/compressors.rs

@@ -0,0 +1,134 @@
+use serde::{Deserialize, Serialize};
+use std::io;
+
+pub trait StoreCompressor {
+    fn compress(&self, uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()>;
+    fn decompress(&self, compressed: &[u8], decompressed: &mut Vec<u8>) -> io::Result<()>;
+    fn get_compressor_id() -> u8;
+}
+
+/// Compressor can be used on `IndexSettings` to choose
+/// the compressor used to compress the doc store.
+///
+/// The default is Lz4Block, but also depends on the enabled feature flags.
+#[derive(Clone, Debug, Copy, PartialEq, Eq, Serialize, Deserialize)]
+pub enum Compressor {
+    #[serde(rename = "lz4")]
+    /// Use the lz4 compressor (block format)
+    Lz4,
+    #[serde(rename = "brotli")]
+    /// Use the brotli compressor
+    Brotli,
+    #[serde(rename = "snappy")]
+    /// Use the snap compressor
+    Snappy,
+}
+
+impl Default for Compressor {
+    fn default() -> Self {
+        if cfg!(feature = "lz4-compression") {
+            Compressor::Lz4
+        } else if cfg!(feature = "brotli-compression") {
+            Compressor::Brotli
+        } else if cfg!(feature = "snappy-compression") {
+            Compressor::Snappy
+        } else {
+            panic!(
+                "all compressor feature flags like are disabled (e.g. lz4-compression), can't choose default compressor"
+            );
+        }
+    }
+}
+
+impl Compressor {
+    pub(crate) fn from_id(id: u8) -> Compressor {
+        match id {
+            1 => Compressor::Lz4,
+            2 => Compressor::Brotli,
+            3 => Compressor::Snappy,
+            _ => panic!("unknown compressor id {:?}", id),
+        }
+    }
+    pub(crate) fn get_id(&self) -> u8 {
+        match self {
+            Self::Lz4 => 1,
+            Self::Brotli => 2,
+            Self::Snappy => 3,
+        }
+    }
+    #[inline]
+    pub(crate) fn compress(&self, uncompressed: &[u8], compressed: &mut Vec<u8>) -> io::Result<()> {
+        match self {
+            Self::Lz4 => {
+                #[cfg(feature = "lz4-compression")]
+                {
+                    super::compression_lz4_block::compress(uncompressed, compressed)
+                }
+                #[cfg(not(feature = "lz4-compression"))]
+                {
+                    panic!("lz4-compression feature flag not activated");
+                }
+            }
+            Self::Brotli => {
+                #[cfg(feature = "brotli-compression")]
+                {
+                    super::compression_brotli::compress(uncompressed, compressed)
+                }
+                #[cfg(not(feature = "brotli-compression"))]
+                {
+                    panic!("brotli-compression-compression feature flag not activated");
+                }
+            }
+            Self::Snappy => {
+                #[cfg(feature = "snappy-compression")]
+                {
+                    super::compression_snap::compress(uncompressed, compressed)
+                }
+                #[cfg(not(feature = "snappy-compression"))]
+                {
+                    panic!("snappy-compression feature flag not activated");
+                }
+            }
+        }
+    }
+
+    #[inline]
+    pub(crate) fn decompress(
+        &self,
+        compressed: &[u8],
+        decompressed: &mut Vec<u8>,
+    ) -> io::Result<()> {
+        match self {
+            Self::Lz4 => {
+                #[cfg(feature = "lz4-compression")]
+                {
+                    super::compression_lz4_block::decompress(compressed, decompressed)
+                }
+                #[cfg(not(feature = "lz4-compression"))]
+                {
+                    panic!("lz4-compression feature flag not activated");
+                }
+            }
+            Self::Brotli => {
+                #[cfg(feature = "brotli-compression")]
+                {
+                    super::compression_brotli::decompress(compressed, decompressed)
+                }
+                #[cfg(not(feature = "brotli-compression"))]
+                {
+                    panic!("brotli-compression feature flag not activated");
+                }
+            }
+            Self::Snappy => {
+                #[cfg(feature = "snappy-compression")]
+                {
+                    super::compression_snap::decompress(compressed, decompressed)
+                }
+                #[cfg(not(feature = "snappy-compression"))]
+                {
+                    panic!("snappy-compression feature flag not activated");
+                }
+            }
+        }
+    }
+}

src/store/footer.rs

@@ -0,0 +1,69 @@
+use crate::{
+    common::{BinarySerializable, FixedSize, HasLen},
+    directory::FileSlice,
+    store::Compressor,
+};
+use std::io;
+
+#[derive(Debug, Clone, PartialEq)]
+pub struct DocStoreFooter {
+    pub offset: u64,
+    pub compressor: Compressor,
+}
+
+/// Serialises the footer to a byte-array
+/// - offset : 8 bytes
+///-  compressor id: 1 byte
+/// - reserved for future use: 15 bytes
+impl BinarySerializable for DocStoreFooter {
+    fn serialize<W: io::Write>(&self, writer: &mut W) -> io::Result<()> {
+        BinarySerializable::serialize(&self.offset, writer)?;
+        BinarySerializable::serialize(&self.compressor.get_id(), writer)?;
+        writer.write_all(&[0; 15])?;
+        Ok(())
+    }
+
+    fn deserialize<R: io::Read>(reader: &mut R) -> io::Result<Self> {
+        let offset = u64::deserialize(reader)?;
+        let compressor_id = u8::deserialize(reader)?;
+        let mut skip_buf = [0; 15];
+        reader.read_exact(&mut skip_buf)?;
+        Ok(DocStoreFooter {
+            offset,
+            compressor: Compressor::from_id(compressor_id),
+        })
+    }
+}
+
+impl FixedSize for DocStoreFooter {
+    const SIZE_IN_BYTES: usize = 24;
+}
+
+impl DocStoreFooter {
+    pub fn new(offset: u64, compressor: Compressor) -> Self {
+        DocStoreFooter { offset, compressor }
+    }
+
+    pub fn extract_footer(file: FileSlice) -> io::Result<(DocStoreFooter, FileSlice)> {
+        if file.len() < DocStoreFooter::SIZE_IN_BYTES {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                format!(
+                    "File corrupted. The file is smaller than Footer::SIZE_IN_BYTES (len={}).",
+                    file.len()
+                ),
+            ));
+        }
+        let (body, footer_slice) = file.split_from_end(DocStoreFooter::SIZE_IN_BYTES);
+        let mut footer_bytes = footer_slice.read_bytes()?;
+        let footer = DocStoreFooter::deserialize(&mut footer_bytes)?;
+        Ok((footer, body))
+    }
+}
+
+#[test]
+fn doc_store_footer_test() {
+    // This test is just to safe guard changes on the footer.
+    // When the doc store footer is updated, make sure to update also the serialize/deserialize methods
+    assert_eq!(core::mem::size_of::<DocStoreFooter>(), 16);
+}

src/store/mod.rs

@@ -33,73 +33,32 @@ and should rely on either
 
 !*/
 
+mod compressors;
+mod footer;
 mod index;
 mod reader;
 mod writer;
-pub use self::reader::RawDocument;
+pub use self::compressors::Compressor;
 pub use self::reader::StoreReader;
 pub use self::writer::StoreWriter;
 
-// compile_error doesn't scale very well, enum like feature flags would be great to have in Rust
-#[cfg(all(feature = "lz4", feature = "brotli"))]
-compile_error!("feature `lz4` or `brotli` must not be enabled together.");
-
-#[cfg(all(feature = "lz4_block", feature = "brotli"))]
-compile_error!("feature `lz4_block` or `brotli` must not be enabled together.");
-
-#[cfg(all(feature = "lz4_block", feature = "lz4"))]
-compile_error!("feature `lz4_block` or `lz4` must not be enabled together.");
-
-#[cfg(all(feature = "lz4_block", feature = "snap"))]
-compile_error!("feature `lz4_block` or `snap` must not be enabled together.");
-
-#[cfg(all(feature = "lz4", feature = "snap"))]
-compile_error!("feature `lz4` or `snap` must not be enabled together.");
-
-#[cfg(all(feature = "brotli", feature = "snap"))]
-compile_error!("feature `brotli` or `snap` must not be enabled together.");
-
-#[cfg(not(any(
-    feature = "lz4",
-    feature = "brotli",
-    feature = "lz4_flex",
-    feature = "snap"
-)))]
-compile_error!("all compressors are deactivated via feature-flags, check Cargo.toml for available decompressors.");
-
-#[cfg(feature = "lz4_flex")]
+#[cfg(feature = "lz4-compression")]
 mod compression_lz4_block;
-#[cfg(feature = "lz4_flex")]
-pub use self::compression_lz4_block::COMPRESSION;
-#[cfg(feature = "lz4_flex")]
-use self::compression_lz4_block::{compress, decompress};
 
-#[cfg(feature = "lz4")]
-mod compression_lz4;
-#[cfg(feature = "lz4")]
-pub use self::compression_lz4::COMPRESSION;
-#[cfg(feature = "lz4")]
-use self::compression_lz4::{compress, decompress};
-
-#[cfg(feature = "brotli")]
+#[cfg(feature = "brotli-compression")]
 mod compression_brotli;
-#[cfg(feature = "brotli")]
-pub use self::compression_brotli::COMPRESSION;
-#[cfg(feature = "brotli")]
-use self::compression_brotli::{compress, decompress};
 
-#[cfg(feature = "snap")]
+#[cfg(feature = "snappy-compression")]
 mod compression_snap;
-#[cfg(feature = "snap")]
-pub use self::compression_snap::COMPRESSION;
-#[cfg(feature = "snap")]
-use self::compression_snap::{compress, decompress};
 
 #[cfg(test)]
 pub mod tests {
 
+    use futures::executor::block_on;
+
     use super::*;
-    use crate::schema::{self, FieldValue, TextFieldIndexing};
+    use crate::fastfield::DeleteBitSet;
+    use crate::schema::{self, FieldValue, TextFieldIndexing, STORED, TEXT};
     use crate::schema::{Document, TextOptions};
     use crate::{
         directory::{Directory, RamDirectory, WritePtr},
@@ -108,28 +67,31 @@ pub mod tests {
     use crate::{schema::Schema, Index};
     use std::path::Path;
 
-    pub fn write_lorem_ipsum_store(writer: WritePtr, num_docs: usize) -> Schema {
-        let mut schema_builder = Schema::builder();
-        let field_body = schema_builder.add_text_field("body", TextOptions::default().set_stored());
-        let field_title =
-            schema_builder.add_text_field("title", TextOptions::default().set_stored());
-        let schema = schema_builder.build();
-        let lorem = String::from(
-            "Doc Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed \
+    const LOREM: &str = "Doc Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed \
              do eiusmod tempor incididunt ut labore et dolore magna aliqua. \
              Ut enim ad minim veniam, quis nostrud exercitation ullamco \
              laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure \
              dolor in reprehenderit in voluptate velit esse cillum dolore eu \
              fugiat nulla pariatur. Excepteur sint occaecat cupidatat non \
              proident, sunt in culpa qui officia deserunt mollit anim id est \
-             laborum.",
-        );
+             laborum.";
+
+    pub fn write_lorem_ipsum_store(
+        writer: WritePtr,
+        num_docs: usize,
+        compressor: Compressor,
+    ) -> Schema {
+        let mut schema_builder = Schema::builder();
+        let field_body = schema_builder.add_text_field("body", TextOptions::default().set_stored());
+        let field_title =
+            schema_builder.add_text_field("title", TextOptions::default().set_stored());
+        let schema = schema_builder.build();
         {
-            let mut store_writer = StoreWriter::new(writer);
+            let mut store_writer = StoreWriter::new(writer, compressor);
             for i in 0..num_docs {
                 let mut fields: Vec<FieldValue> = Vec::new();
                 {
-                    let field_value = FieldValue::new(field_body, From::from(lorem.clone()));
+                    let field_value = FieldValue::new(field_body, From::from(LOREM.to_string()));
                     fields.push(field_value);
                 }
                 {
@@ -146,16 +108,51 @@ pub mod tests {
         schema
     }
 
+    const NUM_DOCS: usize = 1_000;
     #[test]
-    fn test_store() -> crate::Result<()> {
+    fn test_doc_store_iter_with_delete_bug_1077() -> crate::Result<()> {
+        // this will cover deletion of the first element in a checkpoint
+        let deleted_docids = (200..300).collect::<Vec<_>>();
+        let delete_bitset = DeleteBitSet::for_test(&deleted_docids, NUM_DOCS as u32);
+
         let path = Path::new("store");
         let directory = RamDirectory::create();
         let store_wrt = directory.open_write(path)?;
-        let schema = write_lorem_ipsum_store(store_wrt, 1_000);
+        let schema = write_lorem_ipsum_store(store_wrt, NUM_DOCS, Compressor::Lz4);
         let field_title = schema.get_field("title").unwrap();
         let store_file = directory.open_read(path)?;
         let store = StoreReader::open(store_file)?;
-        for i in 0..1_000 {
+        for i in 0..NUM_DOCS as u32 {
+            assert_eq!(
+                *store
+                    .get(i)?
+                    .get_first(field_title)
+                    .unwrap()
+                    .text()
+                    .unwrap(),
+                format!("Doc {}", i)
+            );
+        }
+        for (_, doc) in store.iter(Some(&delete_bitset)).enumerate() {
+            let doc = doc?;
+            let title_content = doc.get_first(field_title).unwrap().text().unwrap();
+            if !title_content.starts_with("Doc ") {
+                panic!("unexpected title_content {}", title_content);
+            }
+        }
+
+        Ok(())
+    }
+
+    fn test_store(compressor: Compressor) -> crate::Result<()> {
+        let path = Path::new("store");
+        let directory = RamDirectory::create();
+        let store_wrt = directory.open_write(path)?;
+        let schema = write_lorem_ipsum_store(store_wrt, NUM_DOCS, compressor);
+        let field_title = schema.get_field("title").unwrap();
+        let store_file = directory.open_read(path)?;
+        let store = StoreReader::open(store_file)?;
+        for i in 0..NUM_DOCS as u32 {
             assert_eq!(
                 *store
                     .get(i)?
@@ -175,6 +172,22 @@ pub mod tests {
         Ok(())
     }
 
+    #[cfg(feature = "lz4-compression")]
+    #[test]
+    fn test_store_lz4_block() -> crate::Result<()> {
+        test_store(Compressor::Lz4)
+    }
+    #[cfg(feature = "snappy-compression")]
+    #[test]
+    fn test_store_snap() -> crate::Result<()> {
+        test_store(Compressor::Snappy)
+    }
+    #[cfg(feature = "brotli-compression")]
+    #[test]
+    fn test_store_brotli() -> crate::Result<()> {
+        test_store(Compressor::Brotli)
+    }
+
     #[test]
     fn test_store_with_delete() -> crate::Result<()> {
         let mut schema_builder = schema::Schema::builder();
@@ -215,6 +228,108 @@ pub mod tests {
         }
         Ok(())
     }
+
+    #[cfg(feature = "snappy-compression")]
+    #[cfg(feature = "lz4-compression")]
+    #[test]
+    fn test_merge_with_changed_compressor() -> crate::Result<()> {
+        let mut schema_builder = schema::Schema::builder();
+
+        let text_field = schema_builder.add_text_field("text_field", TEXT | STORED);
+        let schema = schema_builder.build();
+        let index_builder = Index::builder().schema(schema);
+
+        let mut index = index_builder.create_in_ram().unwrap();
+        index.settings_mut().docstore_compression = Compressor::Lz4;
+        {
+            let mut index_writer = index.writer_for_tests().unwrap();
+            // put enough data create enough blocks in the doc store to be considered for stacking
+            for _ in 0..200 {
+                index_writer.add_document(doc!(text_field=> LOREM));
+            }
+            assert!(index_writer.commit().is_ok());
+            for _ in 0..200 {
+                index_writer.add_document(doc!(text_field=> LOREM));
+            }
+            assert!(index_writer.commit().is_ok());
+        }
+        assert_eq!(
+            index.reader().unwrap().searcher().segment_readers()[0]
+                .get_store_reader()
+                .unwrap()
+                .compressor(),
+            Compressor::Lz4
+        );
+        // Change compressor, this disables stacking on merging
+        let index_settings = index.settings_mut();
+        index_settings.docstore_compression = Compressor::Snappy;
+        // Merging the segments
+        {
+            let segment_ids = index
+                .searchable_segment_ids()
+                .expect("Searchable segments failed.");
+            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());
+        }
+
+        let searcher = index.reader().unwrap().searcher();
+        assert_eq!(searcher.segment_readers().len(), 1);
+        let reader = searcher.segment_readers().iter().last().unwrap();
+        let store = reader.get_store_reader().unwrap();
+
+        for doc in store.iter(reader.delete_bitset()).take(50) {
+            assert_eq!(
+                *doc?.get_first(text_field).unwrap().text().unwrap(),
+                LOREM.to_string()
+            );
+        }
+        assert_eq!(store.compressor(), Compressor::Snappy);
+
+        Ok(())
+    }
+
+    #[test]
+    fn test_merge_of_small_segments() -> crate::Result<()> {
+        let mut schema_builder = schema::Schema::builder();
+
+        let text_field = schema_builder.add_text_field("text_field", TEXT | STORED);
+        let schema = schema_builder.build();
+        let index_builder = Index::builder().schema(schema);
+
+        let index = index_builder.create_in_ram().unwrap();
+
+        {
+            let mut index_writer = index.writer_for_tests().unwrap();
+
+            index_writer.add_document(doc!(text_field=> "1"));
+            assert!(index_writer.commit().is_ok());
+            index_writer.add_document(doc!(text_field=> "2"));
+            assert!(index_writer.commit().is_ok());
+            index_writer.add_document(doc!(text_field=> "3"));
+            assert!(index_writer.commit().is_ok());
+            index_writer.add_document(doc!(text_field=> "4"));
+            assert!(index_writer.commit().is_ok());
+            index_writer.add_document(doc!(text_field=> "5"));
+            assert!(index_writer.commit().is_ok());
+        }
+        // Merging the segments
+        {
+            let segment_ids = index
+                .searchable_segment_ids()
+                .expect("Searchable segments failed.");
+            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());
+        }
+
+        let searcher = index.reader().unwrap().searcher();
+        assert_eq!(searcher.segment_readers().len(), 1);
+        let reader = searcher.segment_readers().iter().last().unwrap();
+        let store = reader.get_store_reader().unwrap();
+        assert_eq!(store.block_checkpoints().count(), 1);
+        Ok(())
+    }
 }
 
 #[cfg(all(test, feature = "unstable"))]
@@ -223,6 +338,7 @@ mod bench {
     use super::tests::write_lorem_ipsum_store;
     use crate::directory::Directory;
     use crate::directory::RamDirectory;
+    use crate::store::Compressor;
     use crate::store::StoreReader;
     use std::path::Path;
     use test::Bencher;
@@ -233,7 +349,11 @@ mod bench {
         let directory = RamDirectory::create();
         let path = Path::new("store");
         b.iter(|| {
-            write_lorem_ipsum_store(directory.open_write(path).unwrap(), 1_000);
+            write_lorem_ipsum_store(
+                directory.open_write(path).unwrap(),
+                1_000,
+                Compressor::default(),
+            );
             directory.delete(path).unwrap();
         });
     }
@@ -242,11 +362,13 @@ mod bench {
     fn bench_store_decode(b: &mut Bencher) {
         let directory = RamDirectory::create();
         let path = Path::new("store");
-        write_lorem_ipsum_store(directory.open_write(path).unwrap(), 1_000);
+        write_lorem_ipsum_store(
+            directory.open_write(path).unwrap(),
+            1_000,
+            Compressor::default(),
+        );
         let store_file = directory.open_read(path).unwrap();
         let store = StoreReader::open(store_file).unwrap();
-        b.iter(|| {
-            store.get(12).unwrap();
-        });
+        b.iter(|| store.iter(None).collect::<Vec<_>>());
     }
 }

src/store/reader.rs

@@ -1,26 +1,29 @@
-use super::decompress;
-use super::index::SkipIndex;
-use crate::common::{BinarySerializable, HasLen};
+use super::Compressor;
+use super::{footer::DocStoreFooter, index::SkipIndex};
 use crate::directory::{FileSlice, OwnedBytes};
 use crate::schema::Document;
 use crate::space_usage::StoreSpaceUsage;
 use crate::store::index::Checkpoint;
 use crate::DocId;
-use crate::{common::VInt, fastfield::DeleteBitSet};
+use crate::{
+    common::{BinarySerializable, HasLen, VInt},
+    error::DataCorruption,
+    fastfield::DeleteBitSet,
+};
 use lru::LruCache;
 use std::io;
-use std::mem::size_of;
 use std::sync::atomic::{AtomicUsize, Ordering};
 use std::sync::{Arc, Mutex};
 
 const LRU_CACHE_CAPACITY: usize = 100;
 
-type Block = Arc<Vec<u8>>;
+type Block = OwnedBytes;
 
 type BlockCache = Arc<Mutex<LruCache<usize, Block>>>;
 
 /// Reads document off tantivy's [`Store`](./index.html)
 pub struct StoreReader {
+    compressor: Compressor,
     data: FileSlice,
     cache: BlockCache,
     cache_hits: Arc<AtomicUsize>,
@@ -32,11 +35,14 @@ pub struct StoreReader {
 impl StoreReader {
     /// Opens a store reader
     pub fn open(store_file: FileSlice) -> io::Result<StoreReader> {
-        let (data_file, offset_index_file) = split_file(store_file)?;
+        let (footer, data_and_offset) = DocStoreFooter::extract_footer(store_file)?;
+
+        let (data_file, offset_index_file) = data_and_offset.split(footer.offset as usize);
         let index_data = offset_index_file.read_bytes()?;
         let space_usage = StoreSpaceUsage::new(data_file.len(), offset_index_file.len());
         let skip_index = SkipIndex::open(index_data);
         Ok(StoreReader {
+            compressor: footer.compressor,
             data: data_file,
             cache: Arc::new(Mutex::new(LruCache::new(LRU_CACHE_CAPACITY))),
             cache_hits: Default::default(),
@@ -50,6 +56,10 @@ impl StoreReader {
         self.skip_index.checkpoints()
     }
 
+    pub(crate) fn compressor(&self) -> Compressor {
+        self.compressor
+    }
+
     fn block_checkpoint(&self, doc_id: DocId) -> Option<Checkpoint> {
         self.skip_index.seek(doc_id)
     }
@@ -72,9 +82,10 @@ impl StoreReader {
 
         let compressed_block = self.compressed_block(checkpoint)?;
         let mut decompressed_block = vec![];
-        decompress(compressed_block.as_slice(), &mut decompressed_block)?;
+        self.compressor
+            .decompress(compressed_block.as_slice(), &mut decompressed_block)?;
 
-        let block = Arc::new(decompressed_block);
+        let block = OwnedBytes::new(decompressed_block);
         self.cache
             .lock()
             .unwrap()
@@ -93,9 +104,8 @@ impl StoreReader {
     /// It should not be called to score documents
     /// for instance.
     pub fn get(&self, doc_id: DocId) -> crate::Result<Document> {
-        let raw_doc = self.get_raw(doc_id)?;
-        let mut cursor = raw_doc.get_bytes();
-        Ok(Document::deserialize(&mut cursor)?)
+        let mut doc_bytes = self.get_document_bytes(doc_id)?;
+        Ok(Document::deserialize(&mut doc_bytes)?)
     }
 
     /// Reads raw bytes of a given document. Returns `RawDocument`, which contains the block of a document and its start and end
@@ -106,7 +116,7 @@ impl StoreReader {
     /// so accessing docs from the same compressed block should be faster.
     /// For that reason a store reader should be kept and reused.
     ///
-    pub fn get_raw(&self, doc_id: DocId) -> crate::Result<RawDocument> {
+    pub fn get_document_bytes(&self, doc_id: DocId) -> crate::Result<OwnedBytes> {
         let checkpoint = self.block_checkpoint(doc_id).ok_or_else(|| {
             crate::TantivyError::InvalidArgument(format!("Failed to lookup Doc #{}.", doc_id))
         })?;
@@ -121,11 +131,7 @@ impl StoreReader {
         let doc_length = VInt::deserialize(&mut cursor)?.val() as usize;
         let start_pos = cursor_len_before - cursor.len();
         let end_pos = cursor_len_before - cursor.len() + doc_length;
-        Ok(RawDocument {
-            block,
-            start_pos,
-            end_pos,
-        })
+        Ok(block.slice(start_pos..end_pos))
     }
 
     /// Iterator over all Documents in their order as they are stored in the doc store.
@@ -135,10 +141,9 @@ impl StoreReader {
         &'b self,
         delete_bitset: Option<&'a DeleteBitSet>,
     ) -> impl Iterator<Item = crate::Result<Document>> + 'b {
-        self.iter_raw(delete_bitset).map(|raw_doc| {
-            let raw_doc = raw_doc?;
-            let mut cursor = raw_doc.get_bytes();
-            Ok(Document::deserialize(&mut cursor)?)
+        self.iter_raw(delete_bitset).map(|doc_bytes_res| {
+            let mut doc_bytes = doc_bytes_res?;
+            Ok(Document::deserialize(&mut doc_bytes)?)
         })
     }
 
@@ -148,7 +153,7 @@ impl StoreReader {
     pub(crate) fn iter_raw<'a: 'b, 'b>(
         &'b self,
         delete_bitset: Option<&'a DeleteBitSet>,
-    ) -> impl Iterator<Item = crate::Result<RawDocument>> + 'b {
+    ) -> impl Iterator<Item = crate::Result<OwnedBytes>> + 'b {
         let last_docid = self
             .block_checkpoints()
             .last()
@@ -158,9 +163,10 @@ impl StoreReader {
         let mut curr_checkpoint = checkpoint_block_iter.next();
         let mut curr_block = curr_checkpoint
             .as_ref()
-            .map(|checkpoint| self.read_block(&checkpoint));
+            .map(|checkpoint| self.read_block(&checkpoint).map_err(|e| e.kind())); // map error in order to enable cloning
         let mut block_start_pos = 0;
         let mut num_skipped = 0;
+        let mut reset_block_pos = false;
         (0..last_docid)
             .filter_map(move |doc_id| {
                 // filter_map is only used to resolve lifetime issues between the two closures on
@@ -170,36 +176,44 @@ impl StoreReader {
                     // we keep the number of skipped documents to move forward in the map block
                     num_skipped += 1;
                 }
+
                 // check move to next checkpoint
-                let mut reset_block_pos = false;
                 if doc_id >= curr_checkpoint.as_ref().unwrap().doc_range.end {
                     curr_checkpoint = checkpoint_block_iter.next();
                     curr_block = curr_checkpoint
                         .as_ref()
-                        .map(|checkpoint| self.read_block(&checkpoint));
+                        .map(|checkpoint| self.read_block(&checkpoint).map_err(|e| e.kind()));
                     reset_block_pos = true;
                     num_skipped = 0;
                 }
 
                 if alive {
-                    let ret = Some((
-                        curr_block.as_ref().unwrap().as_ref().unwrap().clone(), // todo forward errors
-                        num_skipped,
-                        reset_block_pos,
-                    ));
+                    let ret = Some((curr_block.clone(), num_skipped, reset_block_pos));
                     // the map block will move over the num_skipped, so we reset to 0
                     num_skipped = 0;
+                    reset_block_pos = false;
                     ret
                 } else {
                     None
                 }
             })
             .map(move |(block, num_skipped, reset_block_pos)| {
+                let block = block
+                    .ok_or_else(|| {
+                        DataCorruption::comment_only(
+                            "the current checkpoint in the doc store iterator is none, this should never happen",
+                        )
+                    })?
+                    .map_err(|error_kind| {
+                        std::io::Error::new(error_kind, "error when reading block in doc store")
+                    })?;
+                // this flag is set, when filter_map moved to the next block
                 if reset_block_pos {
                     block_start_pos = 0;
                 }
                 let mut cursor = &block[block_start_pos..];
                 let mut pos = 0;
+                // move forward 1 doc + num_skipped in block and return length of current doc
                 let doc_length = loop {
                     let doc_length = VInt::deserialize(&mut cursor)?.val() as usize;
                     let num_bytes_read = block[block_start_pos..].len() - cursor.len();
@@ -214,13 +228,9 @@ impl StoreReader {
                     }
                 };
                 let end_pos = block_start_pos + doc_length;
-                let raw_doc = RawDocument {
-                    block,
-                    start_pos: block_start_pos,
-                    end_pos,
-                };
+                let doc_bytes = block.slice(block_start_pos..end_pos);
                 block_start_pos = end_pos;
-                Ok(raw_doc)
+                Ok(doc_bytes)
             })
     }
 
@@ -230,31 +240,6 @@ impl StoreReader {
     }
 }
 
-/// Get the bytes of a serialized `Document` in a decompressed block.
-pub struct RawDocument {
-    /// the block of data containing multiple documents
-    block: Arc<Vec<u8>>,
-    /// start position of the document in the block
-    start_pos: usize,
-    /// end position of the document in the block
-    end_pos: usize,
-}
-
-impl RawDocument {
-    /// Get the bytes of a serialized `Document` in a decompressed block.
-    pub fn get_bytes(&self) -> &[u8] {
-        &self.block[self.start_pos..self.end_pos]
-    }
-}
-
-fn split_file(data: FileSlice) -> io::Result<(FileSlice, FileSlice)> {
-    let (data, footer_len_bytes) = data.split_from_end(size_of::<u64>());
-    let serialized_offset: OwnedBytes = footer_len_bytes.read_bytes()?;
-    let mut serialized_offset_buf = serialized_offset.as_slice();
-    let offset = u64::deserialize(&mut serialized_offset_buf)? as usize;
-    Ok(data.split(offset))
-}
-
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -272,7 +257,7 @@ mod tests {
         let directory = RamDirectory::create();
         let path = Path::new("store");
         let writer = directory.open_write(path)?;
-        let schema = write_lorem_ipsum_store(writer, 500);
+        let schema = write_lorem_ipsum_store(writer, 500, Compressor::default());
         let title = schema.get_field("title").unwrap();
         let store_file = directory.open_read(path)?;
         let store = StoreReader::open(store_file)?;
@@ -327,7 +312,7 @@ mod tests {
                 .unwrap()
                 .peek_lru()
                 .map(|(&k, _)| k as usize),
-            Some(9249)
+            Some(9210)
         );
 
         Ok(())

src/store/writer.rs

@@ -1,6 +1,6 @@
-use super::compress;
 use super::index::SkipIndexBuilder;
 use super::StoreReader;
+use super::{compressors::Compressor, footer::DocStoreFooter};
 use crate::common::CountingWriter;
 use crate::common::{BinarySerializable, VInt};
 use crate::directory::TerminatingWrite;
@@ -21,6 +21,7 @@ const BLOCK_SIZE: usize = 16_384;
 /// The skip list index on the other hand, is built in memory.
 ///
 pub struct StoreWriter {
+    compressor: Compressor,
     doc: DocId,
     first_doc_in_block: DocId,
     offset_index_writer: SkipIndexBuilder,
@@ -34,8 +35,9 @@ impl StoreWriter {
     ///
     /// The store writer will writes blocks on disc as
     /// document are added.
-    pub fn new(writer: WritePtr) -> StoreWriter {
+    pub fn new(writer: WritePtr, compressor: Compressor) -> StoreWriter {
         StoreWriter {
+            compressor,
             doc: 0,
             first_doc_in_block: 0,
             offset_index_writer: SkipIndexBuilder::new(),
@@ -45,6 +47,10 @@ impl StoreWriter {
         }
     }
 
+    pub(crate) fn compressor(&self) -> Compressor {
+        self.compressor
+    }
+
     /// The memory used (inclusive childs)
     pub fn mem_usage(&self) -> usize {
         self.intermediary_buffer.capacity() + self.current_block.capacity()
@@ -125,7 +131,8 @@ impl StoreWriter {
     fn write_and_compress_block(&mut self) -> io::Result<()> {
         assert!(self.doc > 0);
         self.intermediary_buffer.clear();
-        compress(&self.current_block[..], &mut self.intermediary_buffer)?;
+        self.compressor
+            .compress(&self.current_block[..], &mut self.intermediary_buffer)?;
         let start_offset = self.writer.written_bytes() as usize;
         self.writer.write_all(&self.intermediary_buffer)?;
         let end_offset = self.writer.written_bytes() as usize;
@@ -147,8 +154,9 @@ impl StoreWriter {
             self.write_and_compress_block()?;
         }
         let header_offset: u64 = self.writer.written_bytes() as u64;
+        let footer = DocStoreFooter::new(header_offset, self.compressor);
         self.offset_index_writer.write(&mut self.writer)?;
-        header_offset.serialize(&mut self.writer)?;
+        footer.serialize(&mut self.writer)?;
         self.writer.terminate()
     }
 }

src/termdict/fst_termdict/streamer.rs

@@ -78,8 +78,8 @@ pub struct TermStreamer<'a, A = AlwaysMatch>
 where
     A: Automaton,
 {
-    fst_map: &'a TermDictionary,
-    stream: Stream<'a, A>,
+    pub(crate) fst_map: &'a TermDictionary,
+    pub(crate) stream: Stream<'a, A>,
     term_ord: TermOrdinal,
     current_key: Vec<u8>,
     current_value: TermInfo,

src/termdict/merger.rs

@@ -1,32 +1,11 @@
+use crate::postings::TermInfo;
+use crate::termdict::TermDictionary;
 use crate::termdict::TermOrdinal;
 use crate::termdict::TermStreamer;
-use std::cmp::Ordering;
-use std::collections::BinaryHeap;
-
-pub struct HeapItem<'a> {
-    pub streamer: TermStreamer<'a>,
-    pub segment_ord: usize,
-}
-
-impl<'a> PartialEq for HeapItem<'a> {
-    fn eq(&self, other: &Self) -> bool {
-        self.segment_ord == other.segment_ord
-    }
-}
-
-impl<'a> Eq for HeapItem<'a> {}
-
-impl<'a> PartialOrd for HeapItem<'a> {
-    fn partial_cmp(&self, other: &HeapItem<'a>) -> Option<Ordering> {
-        Some(self.cmp(other))
-    }
-}
-
-impl<'a> Ord for HeapItem<'a> {
-    fn cmp(&self, other: &HeapItem<'a>) -> Ordering {
-        (&other.streamer.key(), &other.segment_ord).cmp(&(&self.streamer.key(), &self.segment_ord))
-    }
-}
+use tantivy_fst::map::OpBuilder;
+use tantivy_fst::map::Union;
+use tantivy_fst::raw::IndexedValue;
+use tantivy_fst::Streamer;
 
 /// Given a list of sorted term streams,
 /// returns an iterator over sorted unique terms.
@@ -34,61 +13,50 @@ impl<'a> Ord for HeapItem<'a> {
 /// The item yield is actually a pair with
 /// - the term
 /// - a slice with the ordinal of the segments containing
-/// the terms.
+/// the term.
 pub struct TermMerger<'a> {
-    heap: BinaryHeap<HeapItem<'a>>,
-    current_streamers: Vec<HeapItem<'a>>,
+    dictionaries: Vec<&'a TermDictionary>,
+    union: Union<'a>,
+    current_key: Vec<u8>,
+    current_segment_and_term_ordinals: Vec<IndexedValue>,
 }
 
 impl<'a> TermMerger<'a> {
     /// Stream of merged term dictionary
     ///
     pub fn new(streams: Vec<TermStreamer<'a>>) -> TermMerger<'a> {
+        let mut op_builder = OpBuilder::new();
+        let mut dictionaries = vec![];
+        for streamer in streams {
+            op_builder.push(streamer.stream);
+            dictionaries.push(streamer.fst_map);
+        }
         TermMerger {
-            heap: BinaryHeap::new(),
-            current_streamers: streams
-                .into_iter()
-                .enumerate()
-                .map(|(ord, streamer)| HeapItem {
-                    streamer,
-                    segment_ord: ord,
-                })
-                .collect(),
+            dictionaries,
+            union: op_builder.union(),
+            current_key: vec![],
+            current_segment_and_term_ordinals: vec![],
         }
     }
 
-    pub(crate) fn matching_segments<'b: 'a>(
-        &'b self,
-    ) -> impl 'b + Iterator<Item = (usize, TermOrdinal)> {
-        self.current_streamers
+    pub fn matching_segments<'b: 'a>(&'b self) -> impl 'b + Iterator<Item = (usize, TermOrdinal)> {
+        self.current_segment_and_term_ordinals
             .iter()
-            .map(|heap_item| (heap_item.segment_ord, heap_item.streamer.term_ord()))
-    }
-
-    fn advance_segments(&mut self) {
-        let streamers = &mut self.current_streamers;
-        let heap = &mut self.heap;
-        for mut heap_item in streamers.drain(..) {
-            if heap_item.streamer.advance() {
-                heap.push(heap_item);
-            }
-        }
+            .map(|iv| (iv.index, iv.value))
     }
 
     /// Advance the term iterator to the next term.
     /// Returns true if there is indeed another term
     /// False if there is none.
     pub fn advance(&mut self) -> bool {
-        self.advance_segments();
-        if let Some(head) = self.heap.pop() {
-            self.current_streamers.push(head);
-            while let Some(next_streamer) = self.heap.peek() {
-                if self.current_streamers[0].streamer.key() != next_streamer.streamer.key() {
-                    break;
-                }
-                let next_heap_it = self.heap.pop().unwrap(); // safe : we peeked beforehand
-                self.current_streamers.push(next_heap_it);
-            }
+        if let Some((k, values)) = self.union.next() {
+            self.current_key.clear();
+            self.current_key.extend_from_slice(k);
+            self.current_segment_and_term_ordinals.clear();
+            self.current_segment_and_term_ordinals
+                .extend_from_slice(values);
+            self.current_segment_and_term_ordinals
+                .sort_by_key(|iv| iv.index);
             true
         } else {
             false
@@ -101,16 +69,85 @@ impl<'a> TermMerger<'a> {
     /// iff advance() has been called before
     /// and "true" was returned.
     pub fn key(&self) -> &[u8] {
-        self.current_streamers[0].streamer.key()
+        &self.current_key
     }
 
-    /// Returns the sorted list of segment ordinals
-    /// that include the current term.
+    /// Iterator over (segment ordinal, TermInfo) pairs iterator sorted by the ordinal.
     ///
     /// This method may be called
     /// iff advance() has been called before
     /// and "true" was returned.
-    pub fn current_kvs(&self) -> &[HeapItem<'a>] {
-        &self.current_streamers[..]
+    pub fn current_segment_ordinals_and_term_infos<'b: 'a>(
+        &'b self,
+    ) -> impl 'b + Iterator<Item = (usize, TermInfo)> {
+        self.current_segment_and_term_ordinals
+            .iter()
+            .map(move |iv| {
+                (
+                    iv.index,
+                    self.dictionaries[iv.index].term_info_from_ord(iv.value),
+                )
+            })
+    }
+}
+
+#[cfg(all(test, feature = "unstable"))]
+mod bench {
+    use super::TermMerger;
+    use crate::directory::FileSlice;
+    use crate::postings::TermInfo;
+    use crate::termdict::{TermDictionary, TermDictionaryBuilder};
+    use rand::distributions::Alphanumeric;
+    use rand::{thread_rng, Rng};
+    use test::{self, Bencher};
+
+    fn make_term_info(term_ord: u64) -> TermInfo {
+        let offset = |term_ord: u64| (term_ord * 100 + term_ord * term_ord) as usize;
+        TermInfo {
+            doc_freq: term_ord as u32,
+            postings_range: offset(term_ord)..offset(term_ord + 1),
+            positions_range: offset(term_ord)..offset(term_ord + 1),
+        }
+    }
+
+    /// Create a dictionary of random strings.
+    fn rand_dict(num_terms: usize) -> crate::Result<TermDictionary> {
+        let buffer: Vec<u8> = {
+            let mut terms = vec![];
+            for _i in 0..num_terms {
+                let rand_string: String = thread_rng()
+                    .sample_iter(&Alphanumeric)
+                    .take(thread_rng().gen_range(30..42))
+                    .map(char::from)
+                    .collect();
+                terms.push(rand_string);
+            }
+            terms.sort();
+
+            let mut term_dictionary_builder = TermDictionaryBuilder::create(Vec::new())?;
+            for i in 0..num_terms {
+                term_dictionary_builder.insert(terms[i].as_bytes(), &make_term_info(i as u64))?;
+            }
+            term_dictionary_builder.finish()?
+        };
+        let file = FileSlice::from(buffer);
+        TermDictionary::open(file)
+    }
+
+    #[bench]
+    fn bench_termmerger(b: &mut Bencher) -> crate::Result<()> {
+        let dict1 = rand_dict(100_000)?;
+        let dict2 = rand_dict(100_000)?;
+        b.iter(|| -> crate::Result<u32> {
+            let stream1 = dict1.stream()?;
+            let stream2 = dict2.stream()?;
+            let mut merger = TermMerger::new(vec![stream1, stream2]);
+            let mut count = 0;
+            while merger.advance() {
+                count += 1;
+            }
+            Ok(count)
+        });
+        Ok(())
     }
 }