move rand to optional dependencies

closes #1117

.github/workflows/coverage.yml

@@ -0,0 +1,27 @@
+name:                           coverage
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+jobs:
+  test:
+    name:                       coverage
+    runs-on:                    ubuntu-latest
+    container:
+      image:                    xd009642/tarpaulin:develop-nightly
+      options:                  --security-opt seccomp=unconfined
+    steps:
+      - name:                   Checkout repository
+        uses:                   actions/checkout@v2
+
+      - name:                   Generate code coverage
+        run: |
+          cargo +nightly tarpaulin --verbose --all-features --workspace --timeout 120 --out Xml
+
+      - name:                   Upload to codecov.io
+        uses:                   codecov/codecov-action@v1
+        with:
+          # token:                ${{secrets.CODECOV_TOKEN}} # not required for public repos
+          fail_ci_if_error:     true

.github/workflows/test.yml

@@ -18,7 +18,13 @@ jobs:
     - uses: actions/checkout@v2
     - name: Build
       run: cargo build --verbose --workspace
+    - name: Install latest nightly to test also against unstable feature flag
+      uses: actions-rs/toolchain@v1
+      with:
+            toolchain: nightly
+            override: true
+            components: rustfmt
     - name: Run tests
-      run: cargo test --verbose --workspace
+      run: cargo test --all-features --verbose --workspace
     - name: Check Formatting
       run: cargo fmt --all -- --check

CHANGELOG.md

@@ -1,3 +1,8 @@
+Tantivy 0.15.3
+=========================
+- Major bugfix. Deleting documents was broken when the index was sorted by a field. (@appaquet, @fulmicoton) #1101
+
+
 Tantivy 0.15.2
 ========================
 - Major bugfix. DocStore still panics when a deleted doc is at the beginning of a block. (@appaquet) #1088

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tantivy"
-version = "0.15.2"
+version = "0.16.0-dev"
 authors = ["Paul Masurel <paul.masurel@gmail.com>"]
 license = "MIT"
 categories = ["database-implementations", "data-structures"]
@@ -35,6 +35,9 @@ crossbeam = "0.8"
 futures = { version = "0.3.15", features = ["thread-pool"] }
 tantivy-query-grammar = { version="0.15.0", path="./query-grammar" }
 tantivy-bitpacker = { version="0.1", path="./bitpacker" }
+common = { version="0.1", path="./common" }
+fastfield_codecs = { version="0.1", path="./fastfield_codecs", default-features = false }
+ownedbytes = { version="0.1", path="./ownedbytes" }
 stable_deref_trait = "1.2"
 rust-stemmers = "1.2"
 downcast-rs = "1.2"
@@ -51,6 +54,7 @@ rayon = "1.5"
 lru = "0.6.5"
 fastdivide = "0.3"
 itertools = "0.10.0"
+measure_time = "0.7.0"
 
 [target.'cfg(windows)'.dependencies]
 winapi = "0.3.9"
@@ -88,7 +92,7 @@ unstable = [] # useful for benches.
 wasm-bindgen = ["uuid/wasm-bindgen"]
 
 [workspace]
-members = ["query-grammar", "bitpacker"]
+members = ["query-grammar", "bitpacker", "common", "fastfield_codecs", "ownedbytes"]
 
 [badges]
 travis-ci = { repository = "tantivy-search/tantivy" }

benches/analyzer.rs

@@ -1,7 +1,7 @@
 use criterion::{criterion_group, criterion_main, Criterion};
 use tantivy::tokenizer::TokenizerManager;
 
-const ALICE_TXT: &'static str = include_str!("alice.txt");
+const ALICE_TXT: &str = include_str!("alice.txt");
 
 pub fn criterion_benchmark(c: &mut Criterion) {
     let tokenizer_manager = TokenizerManager::default();

bitpacker/src/bitpacker.rs

@@ -49,6 +49,7 @@ impl BitPacker {
             let bytes = self.mini_buffer.to_le_bytes();
             output.write_all(&bytes[..num_bytes])?;
             self.mini_buffer_written = 0;
+            self.mini_buffer = 0;
         }
         Ok(())
     }
@@ -61,7 +62,7 @@ impl BitPacker {
     }
 }
 
-#[derive(Clone)]
+#[derive(Clone, Debug, Default)]
 pub struct BitUnpacker {
     num_bits: u64,
     mask: u64,

common/Cargo.toml

@@ -0,0 +1,12 @@
+[package]
+name = "common"
+version = "0.1.0"
+authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
+license = "MIT"
+edition = "2018"
+description = "common traits and utility functions used by multiple tantivy subcrates"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+byteorder = "1.4.3"

common/src/lib.rs

@@ -0,0 +1,9 @@
+pub use byteorder::LittleEndian as Endianness;
+
+mod serialize;
+mod vint;
+mod writer;
+
+pub use serialize::{BinarySerializable, DeserializeFrom, FixedSize};
+pub use vint::{read_u32_vint, read_u32_vint_no_advance, serialize_vint_u32, write_u32_vint, VInt};
+pub use writer::{AntiCallToken, CountingWriter, TerminatingWrite};

common/src/serialize.rs

@@ -1,5 +1,5 @@
-use crate::common::Endianness;
-use crate::common::VInt;
+use crate::Endianness;
+use crate::VInt;
 use byteorder::{ReadBytesExt, WriteBytesExt};
 use std::fmt;
 use std::io;
@@ -160,6 +160,28 @@ impl FixedSize for u8 {
     const SIZE_IN_BYTES: usize = 1;
 }
 
+impl BinarySerializable for bool {
+    fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
+        let val = if *self { 1 } else { 0 };
+        writer.write_u8(val)
+    }
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<bool> {
+        let val = reader.read_u8()?;
+        match val {
+            0 => Ok(false),
+            1 => Ok(true),
+            _ => Err(io::Error::new(
+                io::ErrorKind::InvalidData,
+                "invalid bool value on deserialization, data corrupted",
+            )),
+        }
+    }
+}
+
+impl FixedSize for bool {
+    const SIZE_IN_BYTES: usize = 1;
+}
+
 impl BinarySerializable for String {
     fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
         let data: &[u8] = self.as_bytes();
@@ -180,9 +202,9 @@ impl BinarySerializable for String {
 #[cfg(test)]
 pub mod test {
 
+    use super::VInt;
     use super::*;
-    use crate::common::VInt;
-
+    use crate::serialize::BinarySerializable;
     pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
         let mut buffer = Vec::new();
         O::default().serialize(&mut buffer).unwrap();

common/src/vint.rs

@@ -106,7 +106,7 @@ pub fn read_u32_vint_no_advance(data: &[u8]) -> (u32, usize) {
 pub fn write_u32_vint<W: io::Write>(val: u32, writer: &mut W) -> io::Result<()> {
     let mut buf = [0u8; 8];
     let data = serialize_vint_u32(val, &mut buf);
-    writer.write_all(&data)
+    writer.write_all(data)
 }
 
 impl VInt {
@@ -175,14 +175,14 @@ impl BinarySerializable for VInt {
 mod tests {
 
     use super::serialize_vint_u32;
+    use super::BinarySerializable;
     use super::VInt;
-    use crate::common::BinarySerializable;
 
     fn aux_test_vint(val: u64) {
         let mut v = [14u8; 10];
         let num_bytes = VInt(val).serialize_into(&mut v);
-        for i in num_bytes..10 {
-            assert_eq!(v[i], 14u8);
+        for el in &v[num_bytes..10] {
+            assert_eq!(el, &14u8);
         }
         assert!(num_bytes > 0);
         if num_bytes < 10 {

common/src/writer.rs

@@ -1,7 +1,4 @@
-use crate::directory::AntiCallToken;
-use crate::directory::TerminatingWrite;
-use std::io;
-use std::io::Write;
+use std::io::{self, BufWriter, Write};
 
 pub struct CountingWriter<W> {
     underlying: W,
@@ -16,41 +13,87 @@ impl<W: Write> CountingWriter<W> {
         }
     }
 
+    #[inline]
     pub fn written_bytes(&self) -> u64 {
         self.written_bytes
     }
 
     /// Returns the underlying write object.
     /// Note that this method does not trigger any flushing.
+    #[inline]
     pub fn finish(self) -> W {
         self.underlying
     }
 }
 
 impl<W: Write> Write for CountingWriter<W> {
+    #[inline]
     fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
         let written_size = self.underlying.write(buf)?;
         self.written_bytes += written_size as u64;
         Ok(written_size)
     }
 
+    #[inline]
     fn write_all(&mut self, buf: &[u8]) -> io::Result<()> {
         self.underlying.write_all(buf)?;
         self.written_bytes += buf.len() as u64;
         Ok(())
     }
 
+    #[inline]
     fn flush(&mut self) -> io::Result<()> {
         self.underlying.flush()
     }
 }
 
 impl<W: TerminatingWrite> TerminatingWrite for CountingWriter<W> {
+    #[inline]
     fn terminate_ref(&mut self, token: AntiCallToken) -> io::Result<()> {
         self.underlying.terminate_ref(token)
     }
 }
 
+/// Struct used to prevent from calling [`terminate_ref`](trait.TerminatingWrite#method.terminate_ref) directly
+///
+/// The point is that while the type is public, it cannot be built by anyone
+/// outside of this module.
+pub struct AntiCallToken(());
+
+/// Trait used to indicate when no more write need to be done on a writer
+pub trait TerminatingWrite: Write {
+    /// Indicate that the writer will no longer be used. Internally call terminate_ref.
+    fn terminate(mut self) -> io::Result<()>
+    where
+        Self: Sized,
+    {
+        self.terminate_ref(AntiCallToken(()))
+    }
+
+    /// You should implement this function to define custom behavior.
+    /// This function should flush any buffer it may hold.
+    fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()>;
+}
+
+impl<W: TerminatingWrite + ?Sized> TerminatingWrite for Box<W> {
+    fn terminate_ref(&mut self, token: AntiCallToken) -> io::Result<()> {
+        self.as_mut().terminate_ref(token)
+    }
+}
+
+impl<W: TerminatingWrite> TerminatingWrite for BufWriter<W> {
+    fn terminate_ref(&mut self, a: AntiCallToken) -> io::Result<()> {
+        self.flush()?;
+        self.get_mut().terminate_ref(a)
+    }
+}
+
+impl<'a> TerminatingWrite for &'a mut Vec<u8> {
+    fn terminate_ref(&mut self, _a: AntiCallToken) -> io::Result<()> {
+        self.flush()
+    }
+}
+
 #[cfg(test)]
 mod test {
 

examples/custom_collector.rs

@@ -139,7 +139,7 @@ fn main() -> tantivy::Result<()> {
     //
     // Lets index a bunch of fake documents for the sake of
     // this example.
-    let index = Index::create_in_ram(schema.clone());
+    let index = Index::create_in_ram(schema);
 
     let mut index_writer = index.writer(50_000_000)?;
     index_writer.add_document(doc!(

examples/faceted_search_with_tweaked_score.rs

@@ -12,7 +12,7 @@ fn main() -> tantivy::Result<()> {
     let ingredient = schema_builder.add_facet_field("ingredient", INDEXED);
 
     let schema = schema_builder.build();
-    let index = Index::create_in_ram(schema.clone());
+    let index = Index::create_in_ram(schema);
 
     let mut index_writer = index.writer(30_000_000)?;
 
@@ -51,7 +51,7 @@ fn main() -> tantivy::Result<()> {
         let query = BooleanQuery::new_multiterms_query(
             facets
                 .iter()
-                .map(|key| Term::from_facet(ingredient, &key))
+                .map(|key| Term::from_facet(ingredient, key))
                 .collect(),
         );
         let top_docs_by_custom_score =

examples/iterating_docs_and_positions.rs

@@ -22,7 +22,7 @@ fn main() -> tantivy::Result<()> {
     let title = schema_builder.add_text_field("title", TEXT | STORED);
     let schema = schema_builder.build();
 
-    let index = Index::create_in_ram(schema.clone());
+    let index = Index::create_in_ram(schema);
 
     let mut index_writer = index.writer_with_num_threads(1, 50_000_000)?;
     index_writer.add_document(doc!(title => "The Old Man and the Sea"));

examples/pre_tokenized_text.rs

@@ -82,7 +82,7 @@ fn main() -> tantivy::Result<()> {
         }]
     }"#;
 
-    let short_man_doc = schema.parse_document(&short_man_json)?;
+    let short_man_doc = schema.parse_document(short_man_json)?;
 
     index_writer.add_document(short_man_doc);
 

examples/snippet.rs

@@ -25,7 +25,7 @@ fn main() -> tantivy::Result<()> {
     let schema = schema_builder.build();
 
     // # Indexing documents
-    let index = Index::create_in_dir(&index_path, schema.clone())?;
+    let index = Index::create_in_dir(&index_path, schema)?;
 
     let mut index_writer = index.writer(50_000_000)?;
 

examples/working_with_json.rs

@@ -1,4 +1,3 @@
-use tantivy;
 use tantivy::schema::*;
 
 // # Document from json
@@ -22,7 +21,7 @@ fn main() -> tantivy::Result<()> {
     }"#;
 
     // We can parse our document
-    let _mice_and_men_doc = schema.parse_document(&mice_and_men_doc_json)?;
+    let _mice_and_men_doc = schema.parse_document(mice_and_men_doc_json)?;
 
     // Multi-valued field are allowed, they are
     // expressed in JSON by an array.
@@ -31,7 +30,7 @@ fn main() -> tantivy::Result<()> {
        "title": ["Frankenstein", "The Modern Prometheus"],
        "year": 1818
     }"#;
-    let _frankenstein_doc = schema.parse_document(&frankenstein_json)?;
+    let _frankenstein_doc = schema.parse_document(frankenstein_json)?;
 
     // Note that the schema is saved in your index directory.
     //

fastfield_codecs/Cargo.toml

@@ -0,0 +1,24 @@
+[package]
+name = "fastfield_codecs"
+version = "0.1.0"
+authors = ["Pascal Seitz <pascal@quickwit.io>"]
+license = "MIT"
+edition = "2018"
+description = "Fast field codecs used by tantivy"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+common = { path = "../common/" }
+tantivy-bitpacker = { path = "../bitpacker/" }
+prettytable-rs = {version="0.8.0", optional= true}
+rand = {version="0.8.3", optional= true}
+
+[dev-dependencies]
+more-asserts = "0.2.1"
+rand = "0.8.3"
+
+[features]
+bin = ["prettytable-rs", "rand"]
+default = ["bin"]
+

fastfield_codecs/README.md

@@ -0,0 +1,68 @@
+
+
+# Fast Field Codecs
+
+This crate contains various fast field codecs, used to compress/decompress fast field data in tantivy.
+
+## Contributing
+
+Contributing is pretty straightforward. Since the bitpacking is the simplest compressor, you can check it for reference.
+
+A codec needs to implement 2 traits:
+
+- A reader implementing `FastFieldCodecReader` to read the codec.
+- A serializer implementing `FastFieldCodecSerializer` for compression estimation and codec name + id.
+
+### Tests
+
+Once the traits are implemented test and benchmark integration is pretty easy (see `test_with_codec_data_sets` and `bench.rs`).
+
+Make sure to add the codec to the main.rs, which tests the compression ratio and estimation against different data sets. You can run it with:
+```
+cargo run --features bin
+```
+
+### TODO
+- Add real world data sets in comparison
+- Add codec to cover sparse data sets
+
+
+### Codec Comparison
+```
++----------------------------------+-------------------+------------------------+
+|                                  | Compression Ratio | Compression Estimation |
++----------------------------------+-------------------+------------------------+
+| Autoincrement                    |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.000039572664    | 0.000004396963         |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.1477348         | 0.17275847             |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.28126493        | 0.28125                |
++----------------------------------+-------------------+------------------------+
+| Monotonically increasing concave |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.25003937        | 0.26562938             |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.190665          | 0.1883836              |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.31251436        | 0.3125                 |
++----------------------------------+-------------------+------------------------+
+| Monotonically increasing convex  |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.25003937        | 0.28125438             |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.18676           | 0.2040086              |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.31251436        | 0.3125                 |
++----------------------------------+-------------------+------------------------+
+| Almost monotonically increasing  |                   |                        |
++----------------------------------+-------------------+------------------------+
+| LinearInterpol                   | 0.14066513        | 0.1562544              |
++----------------------------------+-------------------+------------------------+
+| MultiLinearInterpol              | 0.16335973        | 0.17275847             |
++----------------------------------+-------------------+------------------------+
+| Bitpacked                        | 0.28126493        | 0.28125                |
++----------------------------------+-------------------+------------------------+
+
+```

fastfield_codecs/benches/bench.rs

@@ -0,0 +1,108 @@
+#![feature(test)]
+
+extern crate test;
+
+#[cfg(test)]
+mod tests {
+    use fastfield_codecs::{
+        bitpacked::{BitpackedFastFieldReader, BitpackedFastFieldSerializer},
+        linearinterpol::{LinearInterpolFastFieldReader, LinearInterpolFastFieldSerializer},
+        multilinearinterpol::{
+            MultiLinearInterpolFastFieldReader, MultiLinearInterpolFastFieldSerializer,
+        },
+        *,
+    };
+
+    fn get_data() -> Vec<u64> {
+        let mut data: Vec<_> = (100..55000_u64)
+            .map(|num| num + rand::random::<u8>() as u64)
+            .collect();
+        data.push(99_000);
+        data.insert(1000, 2000);
+        data.insert(2000, 100);
+        data.insert(3000, 4100);
+        data.insert(4000, 100);
+        data.insert(5000, 800);
+        data
+    }
+
+    fn value_iter() -> impl Iterator<Item = u64> {
+        0..20_000
+    }
+    fn bench_get<S: FastFieldCodecSerializer, R: FastFieldCodecReader>(
+        b: &mut Bencher,
+        data: &[u64],
+    ) {
+        let mut bytes = vec![];
+        S::serialize(
+            &mut bytes,
+            &data,
+            stats_from_vec(data),
+            data.iter().cloned(),
+            data.iter().cloned(),
+        )
+        .unwrap();
+        let reader = R::open_from_bytes(&bytes).unwrap();
+        b.iter(|| {
+            for pos in value_iter() {
+                reader.get_u64(pos as u64, &bytes);
+            }
+        });
+    }
+    fn bench_create<S: FastFieldCodecSerializer>(b: &mut Bencher, data: &[u64]) {
+        let mut bytes = vec![];
+        b.iter(|| {
+            S::serialize(
+                &mut bytes,
+                &data,
+                stats_from_vec(data),
+                data.iter().cloned(),
+                data.iter().cloned(),
+            )
+            .unwrap();
+        });
+    }
+
+    use test::Bencher;
+    #[bench]
+    fn bench_fastfield_bitpack_create(b: &mut Bencher) {
+        let data: Vec<_> = get_data();
+        bench_create::<BitpackedFastFieldSerializer>(b, &data);
+    }
+    #[bench]
+    fn bench_fastfield_linearinterpol_create(b: &mut Bencher) {
+        let data: Vec<_> = get_data();
+        bench_create::<LinearInterpolFastFieldSerializer>(b, &data);
+    }
+    #[bench]
+    fn bench_fastfield_multilinearinterpol_create(b: &mut Bencher) {
+        let data: Vec<_> = get_data();
+        bench_create::<MultiLinearInterpolFastFieldSerializer>(b, &data);
+    }
+    #[bench]
+    fn bench_fastfield_bitpack_get(b: &mut Bencher) {
+        let data: Vec<_> = get_data();
+        bench_get::<BitpackedFastFieldSerializer, BitpackedFastFieldReader>(b, &data);
+    }
+    #[bench]
+    fn bench_fastfield_linearinterpol_get(b: &mut Bencher) {
+        let data: Vec<_> = get_data();
+        bench_get::<LinearInterpolFastFieldSerializer, LinearInterpolFastFieldReader>(b, &data);
+    }
+    #[bench]
+    fn bench_fastfield_multilinearinterpol_get(b: &mut Bencher) {
+        let data: Vec<_> = get_data();
+        bench_get::<MultiLinearInterpolFastFieldSerializer, MultiLinearInterpolFastFieldReader>(
+            b, &data,
+        );
+    }
+    pub fn stats_from_vec(data: &[u64]) -> FastFieldStats {
+        let min_value = data.iter().cloned().min().unwrap_or(0);
+        let max_value = data.iter().cloned().max().unwrap_or(0);
+        FastFieldStats {
+            min_value,
+            max_value,
+            num_vals: data.len() as u64,
+        }
+    }
+}

fastfield_codecs/src/bitpacked.rs

@@ -0,0 +1,176 @@
+use crate::FastFieldCodecReader;
+use crate::FastFieldCodecSerializer;
+use crate::FastFieldDataAccess;
+use crate::FastFieldStats;
+use common::BinarySerializable;
+use std::io::{self, Write};
+use tantivy_bitpacker::compute_num_bits;
+use tantivy_bitpacker::BitPacker;
+
+use tantivy_bitpacker::BitUnpacker;
+
+/// Depending on the field type, a different
+/// fast field is required.
+#[derive(Clone)]
+pub struct BitpackedFastFieldReader {
+    bit_unpacker: BitUnpacker,
+    pub min_value_u64: u64,
+    pub max_value_u64: u64,
+}
+
+impl<'data> FastFieldCodecReader for BitpackedFastFieldReader {
+    /// Opens a fast field given a file.
+    fn open_from_bytes(bytes: &[u8]) -> io::Result<Self> {
+        let (_data, mut footer) = bytes.split_at(bytes.len() - 16);
+        let min_value = u64::deserialize(&mut footer)?;
+        let amplitude = u64::deserialize(&mut footer)?;
+        let max_value = min_value + amplitude;
+        let num_bits = compute_num_bits(amplitude);
+        let bit_unpacker = BitUnpacker::new(num_bits);
+        Ok(BitpackedFastFieldReader {
+            min_value_u64: min_value,
+            max_value_u64: max_value,
+            bit_unpacker,
+        })
+    }
+    #[inline]
+    fn get_u64(&self, doc: u64, data: &[u8]) -> u64 {
+        self.min_value_u64 + self.bit_unpacker.get(doc, data)
+    }
+    #[inline]
+    fn min_value(&self) -> u64 {
+        self.min_value_u64
+    }
+    #[inline]
+    fn max_value(&self) -> u64 {
+        self.max_value_u64
+    }
+}
+pub struct BitpackedFastFieldSerializerLegacy<'a, W: 'a + Write> {
+    bit_packer: BitPacker,
+    write: &'a mut W,
+    min_value: u64,
+    amplitude: u64,
+    num_bits: u8,
+}
+
+impl<'a, W: Write> BitpackedFastFieldSerializerLegacy<'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<BitpackedFastFieldSerializerLegacy<'a, W>> {
+        assert!(min_value <= max_value);
+        let amplitude = max_value - min_value;
+        let num_bits = compute_num_bits(amplitude);
+        let bit_packer = BitPacker::new();
+        Ok(BitpackedFastFieldSerializerLegacy {
+            bit_packer,
+            write,
+            min_value,
+            amplitude,
+            num_bits,
+        })
+    }
+    /// Pushes a new value to the currently open u64 fast field.
+    #[inline]
+    pub 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<()> {
+        self.bit_packer.close(&mut self.write)?;
+        self.min_value.serialize(&mut self.write)?;
+        self.amplitude.serialize(&mut self.write)?;
+        Ok(())
+    }
+}
+
+pub struct BitpackedFastFieldSerializer {}
+
+impl FastFieldCodecSerializer for BitpackedFastFieldSerializer {
+    const NAME: &'static str = "Bitpacked";
+    const ID: u8 = 1;
+    /// Serializes data with the BitpackedFastFieldSerializer.
+    ///
+    /// 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.
+    fn serialize(
+        write: &mut impl Write,
+        _fastfield_accessor: &impl FastFieldDataAccess,
+        stats: FastFieldStats,
+        data_iter: impl Iterator<Item = u64>,
+        _data_iter1: impl Iterator<Item = u64>,
+    ) -> io::Result<()> {
+        let mut serializer =
+            BitpackedFastFieldSerializerLegacy::open(write, stats.min_value, stats.max_value)?;
+
+        for val in data_iter {
+            serializer.add_val(val)?;
+        }
+        serializer.close_field()?;
+
+        Ok(())
+    }
+    fn is_applicable(
+        _fastfield_accessor: &impl FastFieldDataAccess,
+        _stats: FastFieldStats,
+    ) -> bool {
+        true
+    }
+    fn estimate(_fastfield_accessor: &impl FastFieldDataAccess, stats: FastFieldStats) -> f32 {
+        let amplitude = stats.max_value - stats.min_value;
+        let num_bits = compute_num_bits(amplitude);
+        let num_bits_uncompressed = 64;
+        num_bits as f32 / num_bits_uncompressed as f32
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::tests::get_codec_test_data_sets;
+
+    fn create_and_validate(data: &[u64], name: &str) {
+        crate::tests::create_and_validate::<BitpackedFastFieldSerializer, BitpackedFastFieldReader>(
+            data, name,
+        );
+    }
+
+    #[test]
+    fn test_with_codec_data_sets() {
+        let data_sets = get_codec_test_data_sets();
+        for (mut data, name) in data_sets {
+            create_and_validate(&data, name);
+            data.reverse();
+            create_and_validate(&data, name);
+        }
+    }
+
+    #[test]
+    fn bitpacked_fast_field_rand() {
+        for _ in 0..500 {
+            let mut data = (0..1 + rand::random::<u8>() as usize)
+                .map(|_| rand::random::<i64>() as u64 / 2)
+                .collect::<Vec<_>>();
+            create_and_validate(&data, "rand");
+
+            data.reverse();
+            create_and_validate(&data, "rand");
+        }
+    }
+}

fastfield_codecs/src/lib.rs

@@ -0,0 +1,225 @@
+#[cfg(test)]
+#[macro_use]
+extern crate more_asserts;
+
+use std::io;
+use std::io::Write;
+
+pub mod bitpacked;
+pub mod linearinterpol;
+pub mod multilinearinterpol;
+
+pub trait FastFieldCodecReader: Sized {
+    /// reads the metadata and returns the CodecReader
+    fn open_from_bytes(bytes: &[u8]) -> std::io::Result<Self>;
+
+    fn get_u64(&self, doc: u64, data: &[u8]) -> u64;
+
+    fn min_value(&self) -> u64;
+    fn max_value(&self) -> u64;
+}
+
+/// The FastFieldSerializerEstimate trait is required on all variants
+/// of fast field compressions, to decide which one to choose.
+pub trait FastFieldCodecSerializer {
+    /// A codex needs to provide a unique name and id, which is
+    /// used for debugging and de/serialization.
+    const NAME: &'static str;
+    const ID: u8;
+
+    /// Check if the Codec is able to compress the data
+    fn is_applicable(fastfield_accessor: &impl FastFieldDataAccess, stats: FastFieldStats) -> bool;
+
+    /// Returns an estimate of the compression ratio.
+    /// The baseline is uncompressed 64bit data.
+    ///
+    /// It could make sense to also return a value representing
+    /// computational complexity.
+    fn estimate(fastfield_accessor: &impl FastFieldDataAccess, stats: FastFieldStats) -> f32;
+
+    /// Serializes the data using the serializer into write.
+    /// There are multiple iterators, in case the codec needs to read the data multiple times.
+    /// The iterators should be preferred over using fastfield_accessor for performance reasons.
+    fn serialize(
+        write: &mut impl Write,
+        fastfield_accessor: &impl FastFieldDataAccess,
+        stats: FastFieldStats,
+        data_iter: impl Iterator<Item = u64>,
+        data_iter1: impl Iterator<Item = u64>,
+    ) -> io::Result<()>;
+}
+
+/// FastFieldDataAccess is the trait to access fast field data during serialization and estimation.
+pub trait FastFieldDataAccess {
+    /// Return the value associated to the given position.
+    ///
+    /// Whenever possible use the Iterator passed to the fastfield creation instead, for performance reasons.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `position` is greater than the index.
+    fn get_val(&self, position: u64) -> u64;
+}
+
+#[derive(Debug, Clone)]
+pub struct FastFieldStats {
+    pub min_value: u64,
+    pub max_value: u64,
+    pub num_vals: u64,
+}
+
+impl<'a> FastFieldDataAccess for &'a [u64] {
+    fn get_val(&self, position: u64) -> u64 {
+        self[position as usize]
+    }
+}
+
+impl FastFieldDataAccess for Vec<u64> {
+    fn get_val(&self, position: u64) -> u64 {
+        self[position as usize]
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::{
+        bitpacked::{BitpackedFastFieldReader, BitpackedFastFieldSerializer},
+        linearinterpol::{LinearInterpolFastFieldReader, LinearInterpolFastFieldSerializer},
+        multilinearinterpol::{
+            MultiLinearInterpolFastFieldReader, MultiLinearInterpolFastFieldSerializer,
+        },
+    };
+
+    pub fn create_and_validate<S: FastFieldCodecSerializer, R: FastFieldCodecReader>(
+        data: &[u64],
+        name: &str,
+    ) -> (f32, f32) {
+        if !S::is_applicable(&data, crate::tests::stats_from_vec(data)) {
+            return (f32::MAX, 0.0);
+        }
+        let estimation = S::estimate(&data, crate::tests::stats_from_vec(data));
+        let mut out = vec![];
+        S::serialize(
+            &mut out,
+            &data,
+            crate::tests::stats_from_vec(data),
+            data.iter().cloned(),
+            data.iter().cloned(),
+        )
+        .unwrap();
+
+        let reader = R::open_from_bytes(&out).unwrap();
+        for (doc, orig_val) in data.iter().enumerate() {
+            let val = reader.get_u64(doc as u64, &out);
+            if val != *orig_val {
+                panic!(
+                    "val {:?} does not match orig_val {:?}, in data set {}, data {:?}",
+                    val, orig_val, name, data
+                );
+            }
+        }
+        let actual_compression = data.len() as f32 / out.len() as f32;
+        (estimation, actual_compression)
+    }
+    pub fn get_codec_test_data_sets() -> Vec<(Vec<u64>, &'static str)> {
+        let mut data_and_names = vec![];
+
+        let data = (10..=20_u64).collect::<Vec<_>>();
+        data_and_names.push((data, "simple monotonically increasing"));
+
+        data_and_names.push((
+            vec![5, 6, 7, 8, 9, 10, 99, 100],
+            "offset in linear interpol",
+        ));
+        data_and_names.push((vec![5, 50, 3, 13, 1, 1000, 35], "rand small"));
+        data_and_names.push((vec![10], "single value"));
+
+        data_and_names
+    }
+
+    fn test_codec<S: FastFieldCodecSerializer, R: FastFieldCodecReader>() {
+        let codec_name = S::NAME;
+        for (data, data_set_name) in get_codec_test_data_sets() {
+            let (estimate, actual) =
+                crate::tests::create_and_validate::<S, R>(&data, data_set_name);
+            let result = if estimate == f32::MAX {
+                "Disabled".to_string()
+            } else {
+                format!("Estimate {:?} Actual {:?} ", estimate, actual)
+            };
+            println!(
+                "Codec {}, DataSet {}, {}",
+                codec_name, data_set_name, result
+            );
+        }
+    }
+    #[test]
+    fn test_codec_bitpacking() {
+        test_codec::<BitpackedFastFieldSerializer, BitpackedFastFieldReader>();
+    }
+    #[test]
+    fn test_codec_interpolation() {
+        test_codec::<LinearInterpolFastFieldSerializer, LinearInterpolFastFieldReader>();
+    }
+    #[test]
+    fn test_codec_multi_interpolation() {
+        test_codec::<MultiLinearInterpolFastFieldSerializer, MultiLinearInterpolFastFieldReader>();
+    }
+
+    use super::*;
+    pub fn stats_from_vec(data: &[u64]) -> FastFieldStats {
+        let min_value = data.iter().cloned().min().unwrap_or(0);
+        let max_value = data.iter().cloned().max().unwrap_or(0);
+        FastFieldStats {
+            min_value,
+            max_value,
+            num_vals: data.len() as u64,
+        }
+    }
+
+    #[test]
+    fn estimation_good_interpolation_case() {
+        let data = (10..=20000_u64).collect::<Vec<_>>();
+
+        let linear_interpol_estimation =
+            LinearInterpolFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(linear_interpol_estimation, 0.01);
+
+        let multi_linear_interpol_estimation =
+            MultiLinearInterpolFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(multi_linear_interpol_estimation, 0.2);
+        assert_le!(linear_interpol_estimation, multi_linear_interpol_estimation);
+
+        let bitpacked_estimation =
+            BitpackedFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(linear_interpol_estimation, bitpacked_estimation);
+    }
+    #[test]
+    fn estimation_test_bad_interpolation_case() {
+        let data = vec![200, 10, 10, 10, 10, 1000, 20];
+
+        let linear_interpol_estimation =
+            LinearInterpolFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(linear_interpol_estimation, 0.32);
+
+        let bitpacked_estimation =
+            BitpackedFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(bitpacked_estimation, linear_interpol_estimation);
+    }
+    #[test]
+    fn estimation_test_bad_interpolation_case_monotonically_increasing() {
+        let mut data = (200..=20000_u64).collect::<Vec<_>>();
+        data.push(1_000_000);
+
+        // in this case the linear interpolation can't in fact not be worse than bitpacking,
+        // but the estimator adds some threshold, which leads to estimated worse behavior
+        let linear_interpol_estimation =
+            LinearInterpolFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(linear_interpol_estimation, 0.35);
+
+        let bitpacked_estimation =
+            BitpackedFastFieldSerializer::estimate(&data, stats_from_vec(&data));
+        assert_le!(bitpacked_estimation, 0.32);
+        assert_le!(bitpacked_estimation, linear_interpol_estimation);
+    }
+}

fastfield_codecs/src/linearinterpol.rs

@@ -0,0 +1,295 @@
+use crate::FastFieldCodecReader;
+use crate::FastFieldCodecSerializer;
+use crate::FastFieldDataAccess;
+use crate::FastFieldStats;
+use std::io::{self, Read, Write};
+use std::ops::Sub;
+use tantivy_bitpacker::compute_num_bits;
+use tantivy_bitpacker::BitPacker;
+
+use common::BinarySerializable;
+use common::FixedSize;
+use tantivy_bitpacker::BitUnpacker;
+
+/// Depending on the field type, a different
+/// fast field is required.
+#[derive(Clone)]
+pub struct LinearInterpolFastFieldReader {
+    bit_unpacker: BitUnpacker,
+    pub footer: LinearInterpolFooter,
+    pub slope: f32,
+}
+
+#[derive(Clone, Debug)]
+pub struct LinearInterpolFooter {
+    pub relative_max_value: u64,
+    pub offset: u64,
+    pub first_val: u64,
+    pub last_val: u64,
+    pub num_vals: u64,
+    pub min_value: u64,
+    pub max_value: u64,
+}
+
+impl BinarySerializable for LinearInterpolFooter {
+    fn serialize<W: Write>(&self, write: &mut W) -> io::Result<()> {
+        self.relative_max_value.serialize(write)?;
+        self.offset.serialize(write)?;
+        self.first_val.serialize(write)?;
+        self.last_val.serialize(write)?;
+        self.num_vals.serialize(write)?;
+        self.min_value.serialize(write)?;
+        self.max_value.serialize(write)?;
+        Ok(())
+    }
+
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<LinearInterpolFooter> {
+        Ok(LinearInterpolFooter {
+            relative_max_value: u64::deserialize(reader)?,
+            offset: u64::deserialize(reader)?,
+            first_val: u64::deserialize(reader)?,
+            last_val: u64::deserialize(reader)?,
+            num_vals: u64::deserialize(reader)?,
+            min_value: u64::deserialize(reader)?,
+            max_value: u64::deserialize(reader)?,
+        })
+    }
+}
+
+impl FixedSize for LinearInterpolFooter {
+    const SIZE_IN_BYTES: usize = 56;
+}
+
+impl FastFieldCodecReader for LinearInterpolFastFieldReader {
+    /// Opens a fast field given a file.
+    fn open_from_bytes(bytes: &[u8]) -> io::Result<Self> {
+        let (_data, mut footer) = bytes.split_at(bytes.len() - LinearInterpolFooter::SIZE_IN_BYTES);
+        let footer = LinearInterpolFooter::deserialize(&mut footer)?;
+        let slope = get_slope(footer.first_val, footer.last_val, footer.num_vals);
+
+        let num_bits = compute_num_bits(footer.relative_max_value);
+        let bit_unpacker = BitUnpacker::new(num_bits);
+        Ok(LinearInterpolFastFieldReader {
+            bit_unpacker,
+            footer,
+            slope,
+        })
+    }
+    #[inline]
+    fn get_u64(&self, doc: u64, data: &[u8]) -> u64 {
+        let calculated_value = get_calculated_value(self.footer.first_val, doc, self.slope);
+        (calculated_value + self.bit_unpacker.get(doc, data)) - self.footer.offset
+    }
+
+    #[inline]
+    fn min_value(&self) -> u64 {
+        self.footer.min_value
+    }
+    #[inline]
+    fn max_value(&self) -> u64 {
+        self.footer.max_value
+    }
+}
+
+/// Fastfield serializer, which tries to guess values by linear interpolation
+/// and stores the difference bitpacked.
+pub struct LinearInterpolFastFieldSerializer {}
+
+#[inline]
+fn get_slope(first_val: u64, last_val: u64, num_vals: u64) -> f32 {
+    if num_vals <= 1 {
+        return 0.0;
+    }
+    //  We calculate the slope with f64 high precision and use the result in lower precision f32
+    //  This is done in order to handle estimations for very large values like i64::MAX
+    ((last_val as f64 - first_val as f64) / (num_vals as u64 - 1) as f64) as f32
+}
+
+#[inline]
+fn get_calculated_value(first_val: u64, pos: u64, slope: f32) -> u64 {
+    first_val + (pos as f32 * slope) as u64
+}
+
+impl FastFieldCodecSerializer for LinearInterpolFastFieldSerializer {
+    const NAME: &'static str = "LinearInterpol";
+    const ID: u8 = 2;
+    /// Creates a new fast field serializer.
+    fn serialize(
+        write: &mut impl Write,
+        fastfield_accessor: &impl FastFieldDataAccess,
+        stats: FastFieldStats,
+        data_iter: impl Iterator<Item = u64>,
+        data_iter1: impl Iterator<Item = u64>,
+    ) -> io::Result<()> {
+        assert!(stats.min_value <= stats.max_value);
+
+        let first_val = fastfield_accessor.get_val(0);
+        let last_val = fastfield_accessor.get_val(stats.num_vals as u64 - 1);
+        let slope = get_slope(first_val, last_val, stats.num_vals);
+        // calculate offset to ensure all values are positive
+        let mut offset = 0;
+        let mut rel_positive_max = 0;
+        for (pos, actual_value) in data_iter1.enumerate() {
+            let calculated_value = get_calculated_value(first_val, pos as u64, slope);
+            if calculated_value > actual_value {
+                // negative value we need to apply an offset
+                // we ignore negative values in the max value calculation, because negative values
+                // will be offset to 0
+                offset = offset.max(calculated_value - actual_value);
+            } else {
+                //positive value no offset reuqired
+                rel_positive_max = rel_positive_max.max(actual_value - calculated_value);
+            }
+        }
+
+        // rel_positive_max will be adjusted by offset
+        let relative_max_value = rel_positive_max + offset;
+
+        let num_bits = compute_num_bits(relative_max_value);
+        let mut bit_packer = BitPacker::new();
+        for (pos, val) in data_iter.enumerate() {
+            let calculated_value = get_calculated_value(first_val, pos as u64, slope);
+            let diff = (val + offset) - calculated_value;
+            bit_packer.write(diff, num_bits, write)?;
+        }
+        bit_packer.close(write)?;
+
+        let footer = LinearInterpolFooter {
+            relative_max_value,
+            offset,
+            first_val,
+            last_val,
+            num_vals: stats.num_vals,
+            min_value: stats.min_value,
+            max_value: stats.max_value,
+        };
+        footer.serialize(write)?;
+        Ok(())
+    }
+    fn is_applicable(
+        _fastfield_accessor: &impl FastFieldDataAccess,
+        stats: FastFieldStats,
+    ) -> bool {
+        if stats.num_vals < 3 {
+            return false; //disable compressor for this case
+        }
+        // On serialisation the offset is added to the actual value.
+        // We need to make sure this won't run into overflow calculation issues.
+        // For this we take the maximum theroretical offset and add this to the max value.
+        // If this doesn't overflow the algortihm should be fine
+        let theorethical_maximum_offset = stats.max_value - stats.min_value;
+        if stats
+            .max_value
+            .checked_add(theorethical_maximum_offset)
+            .is_none()
+        {
+            return false;
+        }
+        true
+    }
+    /// estimation for linear interpolation is hard because, you don't know
+    /// where the local maxima for the deviation of the calculated value are and
+    /// the offset to shift all values to >=0 is also unknown.
+    fn estimate(fastfield_accessor: &impl FastFieldDataAccess, stats: FastFieldStats) -> f32 {
+        let first_val = fastfield_accessor.get_val(0);
+        let last_val = fastfield_accessor.get_val(stats.num_vals as u64 - 1);
+        let slope = get_slope(first_val, last_val, stats.num_vals);
+
+        // let's sample at 0%, 5%, 10% .. 95%, 100%
+        let num_vals = stats.num_vals as f32 / 100.0;
+        let sample_positions = (0..20)
+            .map(|pos| (num_vals * pos as f32 * 5.0) as usize)
+            .collect::<Vec<_>>();
+
+        let max_distance = sample_positions
+            .iter()
+            .map(|pos| {
+                let calculated_value = get_calculated_value(first_val, *pos as u64, slope);
+                let actual_value = fastfield_accessor.get_val(*pos as u64);
+                distance(calculated_value, actual_value)
+            })
+            .max()
+            .unwrap_or(0);
+
+        // the theory would be that we don't have the actual max_distance, but we are close within 50%
+        // threshold.
+        // It is multiplied by 2 because in a log case scenario the line would be as much above as
+        // below. So the offset would = max_distance
+        //
+        let relative_max_value = (max_distance as f32 * 1.5) * 2.0;
+
+        let num_bits = compute_num_bits(relative_max_value as u64) as u64 * stats.num_vals as u64
+            + LinearInterpolFooter::SIZE_IN_BYTES as u64;
+        let num_bits_uncompressed = 64 * stats.num_vals;
+        num_bits as f32 / num_bits_uncompressed as f32
+    }
+}
+
+#[inline]
+fn distance<T: Sub<Output = T> + Ord>(x: T, y: T) -> T {
+    if x < y {
+        y - x
+    } else {
+        x - y
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::tests::get_codec_test_data_sets;
+
+    fn create_and_validate(data: &[u64], name: &str) {
+        crate::tests::create_and_validate::<
+            LinearInterpolFastFieldSerializer,
+            LinearInterpolFastFieldReader,
+        >(data, name);
+    }
+
+    #[test]
+    fn test_with_codec_data_sets() {
+        let data_sets = get_codec_test_data_sets();
+        for (mut data, name) in data_sets {
+            create_and_validate(&data, name);
+            data.reverse();
+            create_and_validate(&data, name);
+        }
+    }
+    #[test]
+    fn linear_interpol_fast_field_test_large_amplitude() {
+        let data = vec![
+            i64::MAX as u64 / 2,
+            i64::MAX as u64 / 3,
+            i64::MAX as u64 / 2,
+        ];
+
+        create_and_validate(&data, "large amplitude");
+    }
+    #[test]
+    fn linear_interpol_fast_concave_data() {
+        let data = vec![0, 1, 2, 5, 8, 10, 20, 50];
+        create_and_validate(&data, "concave data");
+    }
+    #[test]
+    fn linear_interpol_fast_convex_data() {
+        let data = vec![0, 40, 60, 70, 75, 77];
+        create_and_validate(&data, "convex data");
+    }
+    #[test]
+    fn linear_interpol_fast_field_test_simple() {
+        let data = (10..=20_u64).collect::<Vec<_>>();
+
+        create_and_validate(&data, "simple monotonically");
+    }
+
+    #[test]
+    fn linear_interpol_fast_field_rand() {
+        for _ in 0..5000 {
+            let mut data = (0..50).map(|_| rand::random::<u64>()).collect::<Vec<_>>();
+            create_and_validate(&data, "random");
+
+            data.reverse();
+            create_and_validate(&data, "random");
+        }
+    }
+}

fastfield_codecs/src/main.rs

@@ -0,0 +1,127 @@
+#[macro_use]
+extern crate prettytable;
+use fastfield_codecs::{
+    linearinterpol::LinearInterpolFastFieldSerializer,
+    multilinearinterpol::MultiLinearInterpolFastFieldSerializer, FastFieldCodecSerializer,
+    FastFieldStats,
+};
+use prettytable::{Cell, Row, Table};
+
+fn main() {
+    let mut table = Table::new();
+
+    // Add a row per time
+    table.add_row(row!["", "Compression Ratio", "Compression Estimation"]);
+
+    for (data, data_set_name) in get_codec_test_data_sets() {
+        let mut results = vec![];
+        let res = serialize_with_codec::<LinearInterpolFastFieldSerializer>(&data);
+        results.push(res);
+        let res = serialize_with_codec::<MultiLinearInterpolFastFieldSerializer>(&data);
+        results.push(res);
+        let res = serialize_with_codec::<fastfield_codecs::bitpacked::BitpackedFastFieldSerializer>(
+            &data,
+        );
+        results.push(res);
+
+        //let best_estimation_codec = results
+        //.iter()
+        //.min_by(|res1, res2| res1.partial_cmp(&res2).unwrap())
+        //.unwrap();
+        let best_compression_ratio_codec = results
+            .iter()
+            .min_by(|res1, res2| res1.partial_cmp(res2).unwrap())
+            .cloned()
+            .unwrap();
+
+        table.add_row(Row::new(vec![Cell::new(data_set_name).style_spec("Bbb")]));
+        for (is_applicable, est, comp, name) in results {
+            let (est_cell, ratio_cell) = if !is_applicable {
+                ("Codec Disabled".to_string(), "".to_string())
+            } else {
+                (est.to_string(), comp.to_string())
+            };
+            #[allow(clippy::all)]
+            let style = if comp == best_compression_ratio_codec.1 {
+                "Fb"
+            } else {
+                ""
+            };
+
+            table.add_row(Row::new(vec![
+                Cell::new(&name.to_string()).style_spec("bFg"),
+                Cell::new(&ratio_cell).style_spec(style),
+                Cell::new(&est_cell).style_spec(""),
+            ]));
+        }
+    }
+
+    table.printstd();
+}
+
+pub fn get_codec_test_data_sets() -> Vec<(Vec<u64>, &'static str)> {
+    let mut data_and_names = vec![];
+
+    let data = (1000..=200_000_u64).collect::<Vec<_>>();
+    data_and_names.push((data, "Autoincrement"));
+
+    let mut current_cumulative = 0;
+    let data = (1..=200_000_u64)
+        .map(|num| {
+            let num = (num as f32 + num as f32).log10() as u64;
+            current_cumulative += num;
+            current_cumulative
+        })
+        .collect::<Vec<_>>();
+    //let data = (1..=200000_u64).map(|num| num + num).collect::<Vec<_>>();
+    data_and_names.push((data, "Monotonically increasing concave"));
+
+    let mut current_cumulative = 0;
+    let data = (1..=200_000_u64)
+        .map(|num| {
+            let num = (200_000.0 - num as f32).log10() as u64;
+            current_cumulative += num;
+            current_cumulative
+        })
+        .collect::<Vec<_>>();
+    data_and_names.push((data, "Monotonically increasing convex"));
+
+    let data = (1000..=200_000_u64)
+        .map(|num| num + rand::random::<u8>() as u64)
+        .collect::<Vec<_>>();
+    data_and_names.push((data, "Almost monotonically increasing"));
+
+    data_and_names
+}
+
+pub fn serialize_with_codec<S: FastFieldCodecSerializer>(
+    data: &[u64],
+) -> (bool, f32, f32, &'static str) {
+    let is_applicable = S::is_applicable(&data, stats_from_vec(data));
+    if !is_applicable {
+        return (false, 0.0, 0.0, S::NAME);
+    }
+    let estimation = S::estimate(&data, stats_from_vec(data));
+    let mut out = vec![];
+    S::serialize(
+        &mut out,
+        &data,
+        stats_from_vec(data),
+        data.iter().cloned(),
+        data.iter().cloned(),
+    )
+    .unwrap();
+
+    let actual_compression = out.len() as f32 / (data.len() * 8) as f32;
+    (true, estimation, actual_compression, S::NAME)
+}
+
+pub fn stats_from_vec(data: &[u64]) -> FastFieldStats {
+    let min_value = data.iter().cloned().min().unwrap_or(0);
+    let max_value = data.iter().cloned().max().unwrap_or(0);
+    FastFieldStats {
+        min_value,
+        max_value,
+        num_vals: data.len() as u64,
+    }
+}

fastfield_codecs/src/multilinearinterpol.rs

@@ -0,0 +1,411 @@
+use crate::FastFieldCodecReader;
+use crate::FastFieldCodecSerializer;
+use crate::FastFieldDataAccess;
+use crate::FastFieldStats;
+use common::CountingWriter;
+use std::io::{self, Read, Write};
+use std::ops::Sub;
+use tantivy_bitpacker::compute_num_bits;
+use tantivy_bitpacker::BitPacker;
+
+use common::BinarySerializable;
+use common::DeserializeFrom;
+use tantivy_bitpacker::BitUnpacker;
+
+const CHUNK_SIZE: u64 = 512;
+
+/// Depending on the field type, a different
+/// fast field is required.
+#[derive(Clone)]
+pub struct MultiLinearInterpolFastFieldReader {
+    pub footer: MultiLinearInterpolFooter,
+}
+
+#[derive(Clone, Debug, Default)]
+struct Function {
+    // The offset in the data is required, because we have diffrent bit_widths per block
+    data_start_offset: u64,
+    // start_pos in the block will be CHUNK_SIZE * BLOCK_NUM
+    start_pos: u64,
+    // only used during serialization, 0 after deserialization
+    end_pos: u64,
+    // only used during serialization, 0 after deserialization
+    value_start_pos: u64,
+    // only used during serialization, 0 after deserialization
+    value_end_pos: u64,
+    slope: f32,
+    // The offset so that all values are positive when writing them
+    positive_val_offset: u64,
+    num_bits: u8,
+    bit_unpacker: BitUnpacker,
+}
+
+impl Function {
+    fn calc_slope(&mut self) {
+        let num_vals = self.end_pos - self.start_pos;
+        get_slope(self.value_start_pos, self.value_end_pos, num_vals);
+    }
+    // split the interpolation into two function, change self and return the second split
+    fn split(&mut self, split_pos: u64, split_pos_value: u64) -> Function {
+        let mut new_function = Function {
+            start_pos: split_pos,
+            end_pos: self.end_pos,
+            value_start_pos: split_pos_value,
+            value_end_pos: self.value_end_pos,
+            ..Default::default()
+        };
+        new_function.calc_slope();
+        self.end_pos = split_pos;
+        self.value_end_pos = split_pos_value;
+        self.calc_slope();
+        new_function
+    }
+}
+
+impl BinarySerializable for Function {
+    fn serialize<W: Write>(&self, write: &mut W) -> io::Result<()> {
+        self.data_start_offset.serialize(write)?;
+        self.value_start_pos.serialize(write)?;
+        self.positive_val_offset.serialize(write)?;
+        self.slope.serialize(write)?;
+        self.num_bits.serialize(write)?;
+        Ok(())
+    }
+
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Function> {
+        let data_start_offset = u64::deserialize(reader)?;
+        let value_start_pos = u64::deserialize(reader)?;
+        let offset = u64::deserialize(reader)?;
+        let slope = f32::deserialize(reader)?;
+        let num_bits = u8::deserialize(reader)?;
+        let interpolation = Function {
+            data_start_offset,
+            value_start_pos,
+            positive_val_offset: offset,
+            num_bits,
+            bit_unpacker: BitUnpacker::new(num_bits),
+            slope,
+            ..Default::default()
+        };
+
+        Ok(interpolation)
+    }
+}
+
+#[derive(Clone, Debug)]
+pub struct MultiLinearInterpolFooter {
+    pub num_vals: u64,
+    pub min_value: u64,
+    pub max_value: u64,
+    interpolations: Vec<Function>,
+}
+
+impl BinarySerializable for MultiLinearInterpolFooter {
+    fn serialize<W: Write>(&self, write: &mut W) -> io::Result<()> {
+        let mut out = vec![];
+        self.num_vals.serialize(&mut out)?;
+        self.min_value.serialize(&mut out)?;
+        self.max_value.serialize(&mut out)?;
+        self.interpolations.serialize(&mut out)?;
+        write.write_all(&out)?;
+        (out.len() as u32).serialize(write)?;
+        Ok(())
+    }
+
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<MultiLinearInterpolFooter> {
+        let mut footer = MultiLinearInterpolFooter {
+            num_vals: u64::deserialize(reader)?,
+            min_value: u64::deserialize(reader)?,
+            max_value: u64::deserialize(reader)?,
+            interpolations: Vec::<Function>::deserialize(reader)?,
+        };
+        for (num, interpol) in footer.interpolations.iter_mut().enumerate() {
+            interpol.start_pos = CHUNK_SIZE * num as u64;
+        }
+        Ok(footer)
+    }
+}
+
+#[inline]
+fn get_interpolation_position(doc: u64) -> usize {
+    let index = doc / CHUNK_SIZE;
+    index as usize
+}
+
+#[inline]
+fn get_interpolation_function(doc: u64, interpolations: &[Function]) -> &Function {
+    &interpolations[get_interpolation_position(doc)]
+}
+
+impl FastFieldCodecReader for MultiLinearInterpolFastFieldReader {
+    /// Opens a fast field given a file.
+    fn open_from_bytes(bytes: &[u8]) -> io::Result<Self> {
+        let footer_len: u32 = (&bytes[bytes.len() - 4..]).deserialize()?;
+
+        let (_data, mut footer) = bytes.split_at(bytes.len() - (4 + footer_len) as usize);
+        let footer = MultiLinearInterpolFooter::deserialize(&mut footer)?;
+
+        Ok(MultiLinearInterpolFastFieldReader { footer })
+    }
+
+    #[inline]
+    fn get_u64(&self, doc: u64, data: &[u8]) -> u64 {
+        let interpolation = get_interpolation_function(doc, &self.footer.interpolations);
+        let doc = doc - interpolation.start_pos;
+        let calculated_value =
+            get_calculated_value(interpolation.value_start_pos, doc, interpolation.slope);
+        let diff = interpolation
+            .bit_unpacker
+            .get(doc, &data[interpolation.data_start_offset as usize..]);
+        (calculated_value + diff) - interpolation.positive_val_offset
+    }
+
+    #[inline]
+    fn min_value(&self) -> u64 {
+        self.footer.min_value
+    }
+    #[inline]
+    fn max_value(&self) -> u64 {
+        self.footer.max_value
+    }
+}
+
+#[inline]
+fn get_slope(first_val: u64, last_val: u64, num_vals: u64) -> f32 {
+    ((last_val as f64 - first_val as f64) / (num_vals as u64 - 1) as f64) as f32
+}
+
+#[inline]
+fn get_calculated_value(first_val: u64, pos: u64, slope: f32) -> u64 {
+    (first_val as i64 + (pos as f32 * slope) as i64) as u64
+}
+
+/// Same as LinearInterpolFastFieldSerializer, but working on chunks of CHUNK_SIZE elements.
+pub struct MultiLinearInterpolFastFieldSerializer {}
+
+impl FastFieldCodecSerializer for MultiLinearInterpolFastFieldSerializer {
+    const NAME: &'static str = "MultiLinearInterpol";
+    const ID: u8 = 3;
+    /// Creates a new fast field serializer.
+    fn serialize(
+        write: &mut impl Write,
+        fastfield_accessor: &impl FastFieldDataAccess,
+        stats: FastFieldStats,
+        data_iter: impl Iterator<Item = u64>,
+        _data_iter1: impl Iterator<Item = u64>,
+    ) -> io::Result<()> {
+        assert!(stats.min_value <= stats.max_value);
+
+        let first_val = fastfield_accessor.get_val(0);
+        let last_val = fastfield_accessor.get_val(stats.num_vals as u64 - 1);
+
+        let mut first_function = Function {
+            end_pos: stats.num_vals,
+            value_start_pos: first_val,
+            value_end_pos: last_val,
+            ..Default::default()
+        };
+        first_function.calc_slope();
+        let mut interpolations = vec![first_function];
+
+        // Since we potentially apply multiple passes over the data, the data is cached.
+        // Multiple iteration can be expensive (merge with index sorting can add lot of overhead per
+        // iteration)
+        let data = data_iter.collect::<Vec<_>>();
+
+        //// let's split this into chunks of CHUNK_SIZE
+        for data_pos in (0..data.len() as u64).step_by(CHUNK_SIZE as usize).skip(1) {
+            let new_fun = {
+                let current_interpolation = interpolations.last_mut().unwrap();
+                current_interpolation.split(data_pos, data[data_pos as usize])
+            };
+            interpolations.push(new_fun);
+        }
+        // calculate offset and max (-> numbits) for each function
+        for interpolation in &mut interpolations {
+            let mut offset = 0;
+            let mut rel_positive_max = 0;
+            for (pos, actual_value) in data
+                [interpolation.start_pos as usize..interpolation.end_pos as usize]
+                .iter()
+                .cloned()
+                .enumerate()
+            {
+                let calculated_value = get_calculated_value(
+                    interpolation.value_start_pos,
+                    pos as u64,
+                    interpolation.slope,
+                );
+                if calculated_value > actual_value {
+                    // negative value we need to apply an offset
+                    // we ignore negative values in the max value calculation, because negative values
+                    // will be offset to 0
+                    offset = offset.max(calculated_value - actual_value);
+                } else {
+                    //positive value no offset reuqired
+                    rel_positive_max = rel_positive_max.max(actual_value - calculated_value);
+                }
+            }
+
+            interpolation.positive_val_offset = offset;
+            interpolation.num_bits = compute_num_bits(rel_positive_max + offset);
+        }
+        let mut bit_packer = BitPacker::new();
+
+        let write = &mut CountingWriter::wrap(write);
+        for interpolation in &mut interpolations {
+            interpolation.data_start_offset = write.written_bytes();
+            let num_bits = interpolation.num_bits;
+            for (pos, actual_value) in data
+                [interpolation.start_pos as usize..interpolation.end_pos as usize]
+                .iter()
+                .cloned()
+                .enumerate()
+            {
+                let calculated_value = get_calculated_value(
+                    interpolation.value_start_pos,
+                    pos as u64,
+                    interpolation.slope,
+                );
+                let diff = (actual_value + interpolation.positive_val_offset) - calculated_value;
+                bit_packer.write(diff, num_bits, write)?;
+            }
+            bit_packer.flush(write)?;
+        }
+        bit_packer.close(write)?;
+
+        let footer = MultiLinearInterpolFooter {
+            num_vals: stats.num_vals,
+            min_value: stats.min_value,
+            max_value: stats.max_value,
+            interpolations,
+        };
+        footer.serialize(write)?;
+        Ok(())
+    }
+
+    fn is_applicable(
+        _fastfield_accessor: &impl FastFieldDataAccess,
+        stats: FastFieldStats,
+    ) -> bool {
+        if stats.num_vals < 5_000 {
+            return false;
+        }
+        // On serialization the offset is added to the actual value.
+        // We need to make sure this won't run into overflow calculation issues.
+        // For this we take the maximum theroretical offset and add this to the max value.
+        // If this doesn't overflow the algortihm should be fine
+        let theorethical_maximum_offset = stats.max_value - stats.min_value;
+        if stats
+            .max_value
+            .checked_add(theorethical_maximum_offset)
+            .is_none()
+        {
+            return false;
+        }
+        true
+    }
+    /// estimation for linear interpolation is hard because, you don't know
+    /// where the local maxima are for the deviation of the calculated value and
+    /// the offset is also unknown.
+    fn estimate(fastfield_accessor: &impl FastFieldDataAccess, stats: FastFieldStats) -> f32 {
+        let first_val_in_first_block = fastfield_accessor.get_val(0);
+        let last_elem_in_first_chunk = CHUNK_SIZE.min(stats.num_vals);
+        let last_val_in_first_block =
+            fastfield_accessor.get_val(last_elem_in_first_chunk as u64 - 1);
+        let slope = get_slope(
+            first_val_in_first_block,
+            last_val_in_first_block,
+            stats.num_vals,
+        );
+
+        // let's sample at 0%, 5%, 10% .. 95%, 100%, but for the first block only
+        let sample_positions = (0..20)
+            .map(|pos| (last_elem_in_first_chunk as f32 / 100.0 * pos as f32 * 5.0) as usize)
+            .collect::<Vec<_>>();
+
+        let max_distance = sample_positions
+            .iter()
+            .map(|pos| {
+                let calculated_value =
+                    get_calculated_value(first_val_in_first_block, *pos as u64, slope);
+                let actual_value = fastfield_accessor.get_val(*pos as u64);
+                distance(calculated_value, actual_value)
+            })
+            .max()
+            .unwrap();
+
+        // Estimate one block and extrapolate the cost to all blocks.
+        // the theory would be that we don't have the actual max_distance, but we are close within 50%
+        // threshold.
+        // It is multiplied by 2 because in a log case scenario the line would be as much above as
+        // below. So the offset would = max_distance
+        //
+        let relative_max_value = (max_distance as f32 * 1.5) * 2.0;
+
+        let num_bits = compute_num_bits(relative_max_value as u64) as u64 * stats.num_vals as u64
+            // function metadata per block
+            + 29 * (stats.num_vals / CHUNK_SIZE);
+        let num_bits_uncompressed = 64 * stats.num_vals;
+        num_bits as f32 / num_bits_uncompressed as f32
+    }
+}
+
+fn distance<T: Sub<Output = T> + Ord>(x: T, y: T) -> T {
+    if x < y {
+        y - x
+    } else {
+        x - y
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::tests::get_codec_test_data_sets;
+
+    fn create_and_validate(data: &[u64], name: &str) {
+        crate::tests::create_and_validate::<
+            MultiLinearInterpolFastFieldSerializer,
+            MultiLinearInterpolFastFieldReader,
+        >(data, name);
+    }
+
+    #[test]
+    fn test_with_codec_data_sets() {
+        let data_sets = get_codec_test_data_sets();
+        for (mut data, name) in data_sets {
+            create_and_validate(&data, name);
+            data.reverse();
+            create_and_validate(&data, name);
+        }
+    }
+    #[test]
+    fn test_simple() {
+        let data = (10..=20_u64).collect::<Vec<_>>();
+        create_and_validate(&data, "simple monotonically");
+    }
+
+    #[test]
+    fn border_cases_1() {
+        let data = (0..1024).collect::<Vec<_>>();
+        create_and_validate(&data, "border case");
+    }
+    #[test]
+    fn border_case_2() {
+        let data = (0..1025).collect::<Vec<_>>();
+        create_and_validate(&data, "border case");
+    }
+    #[test]
+    fn rand() {
+        for _ in 0..10 {
+            let mut data = (5_000..20_000)
+                .map(|_| rand::random::<u64>() as u64)
+                .collect::<Vec<_>>();
+            create_and_validate(&data, "random");
+
+            data.reverse();
+            create_and_validate(&data, "random");
+        }
+    }
+}

ownedbytes/Cargo.toml

@@ -0,0 +1,10 @@
+[package]
+authors = ["Paul Masurel <paul@quickwit.io>", "Pascal Seitz <pascal@quickwit.io>"]
+name = "ownedbytes"
+version = "0.1.0"
+edition = "2018"
+description = "Expose data as static slice"
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+stable_deref_trait = "1.2.0"

ownedbytes/src/lib.rs

@@ -0,0 +1,290 @@
+use stable_deref_trait::StableDeref;
+use std::convert::TryInto;
+use std::mem;
+use std::ops::{Deref, Range};
+use std::sync::Arc;
+use std::{fmt, io};
+
+/// An OwnedBytes simply wraps an object that owns a slice of data and exposes
+/// this data as a static slice.
+///
+/// The backing object is required to be `StableDeref`.
+#[derive(Clone)]
+pub struct OwnedBytes {
+    data: &'static [u8],
+    box_stable_deref: Arc<dyn Deref<Target = [u8]> + Sync + Send>,
+}
+
+impl OwnedBytes {
+    /// Creates an empty `OwnedBytes`.
+    pub fn empty() -> OwnedBytes {
+        OwnedBytes::new(&[][..])
+    }
+
+    /// Creates an `OwnedBytes` intance given a `StableDeref` object.
+    pub fn new<T: StableDeref + Deref<Target = [u8]> + 'static + Send + Sync>(
+        data_holder: T,
+    ) -> OwnedBytes {
+        let box_stable_deref = Arc::new(data_holder);
+        let bytes: &[u8] = box_stable_deref.as_ref();
+        let data = unsafe { mem::transmute::<_, &'static [u8]>(bytes.deref()) };
+        OwnedBytes {
+            data,
+            box_stable_deref,
+        }
+    }
+
+    /// creates a fileslice that is just a view over a slice of the data.
+    pub fn slice(&self, range: Range<usize>) -> Self {
+        OwnedBytes {
+            data: &self.data[range],
+            box_stable_deref: self.box_stable_deref.clone(),
+        }
+    }
+
+    /// Returns the underlying slice of data.
+    /// `Deref` and `AsRef` are also available.
+    #[inline]
+    pub fn as_slice(&self) -> &[u8] {
+        self.data
+    }
+
+    /// Returns the len of the slice.
+    #[inline]
+    pub fn len(&self) -> usize {
+        self.data.len()
+    }
+
+    /// Splits the OwnedBytes into two OwnedBytes `(left, right)`.
+    ///
+    /// Left will hold `split_len` bytes.
+    ///
+    /// This operation is cheap and does not require to copy any memory.
+    /// On the other hand, both `left` and `right` retain a handle over
+    /// the entire slice of memory. In other words, the memory will only
+    /// be released when both left and right are dropped.
+    pub fn split(self, split_len: usize) -> (OwnedBytes, OwnedBytes) {
+        let right_box_stable_deref = self.box_stable_deref.clone();
+        let left = OwnedBytes {
+            data: &self.data[..split_len],
+            box_stable_deref: self.box_stable_deref,
+        };
+        let right = OwnedBytes {
+            data: &self.data[split_len..],
+            box_stable_deref: right_box_stable_deref,
+        };
+        (left, right)
+    }
+
+    /// Returns true iff this `OwnedBytes` is empty.
+    #[inline]
+    pub fn is_empty(&self) -> bool {
+        self.as_slice().is_empty()
+    }
+
+    /// Drops the left most `advance_len` bytes.
+    ///
+    /// See also [.clip(clip_len: usize))](#method.clip).
+    #[inline]
+    pub fn advance(&mut self, advance_len: usize) {
+        self.data = &self.data[advance_len..]
+    }
+
+    /// Reads an `u8` from the `OwnedBytes` and advance by one byte.
+    #[inline]
+    pub fn read_u8(&mut self) -> u8 {
+        assert!(!self.is_empty());
+
+        let byte = self.as_slice()[0];
+        self.advance(1);
+        byte
+    }
+
+    /// Reads an `u64` encoded as little-endian from the `OwnedBytes` and advance by 8 bytes.
+    #[inline]
+    pub fn read_u64(&mut self) -> u64 {
+        assert!(self.len() > 7);
+
+        let octlet: [u8; 8] = self.as_slice()[..8].try_into().unwrap();
+        self.advance(8);
+        u64::from_le_bytes(octlet)
+    }
+}
+
+impl fmt::Debug for OwnedBytes {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        // We truncate the bytes in order to make sure the debug string
+        // is not too long.
+        let bytes_truncated: &[u8] = if self.len() > 8 {
+            &self.as_slice()[..10]
+        } else {
+            self.as_slice()
+        };
+        write!(f, "OwnedBytes({:?}, len={})", bytes_truncated, self.len())
+    }
+}
+
+impl Deref for OwnedBytes {
+    type Target = [u8];
+
+    #[inline]
+    fn deref(&self) -> &Self::Target {
+        self.as_slice()
+    }
+}
+
+impl io::Read for OwnedBytes {
+    #[inline]
+    fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
+        let read_len = {
+            let data = self.as_slice();
+            if data.len() >= buf.len() {
+                let buf_len = buf.len();
+                buf.copy_from_slice(&data[..buf_len]);
+                buf.len()
+            } else {
+                let data_len = data.len();
+                buf[..data_len].copy_from_slice(data);
+                data_len
+            }
+        };
+        self.advance(read_len);
+        Ok(read_len)
+    }
+    #[inline]
+    fn read_to_end(&mut self, buf: &mut Vec<u8>) -> io::Result<usize> {
+        let read_len = {
+            let data = self.as_slice();
+            buf.extend(data);
+            data.len()
+        };
+        self.advance(read_len);
+        Ok(read_len)
+    }
+    #[inline]
+    fn read_exact(&mut self, buf: &mut [u8]) -> io::Result<()> {
+        let read_len = self.read(buf)?;
+        if read_len != buf.len() {
+            return Err(io::Error::new(
+                io::ErrorKind::UnexpectedEof,
+                "failed to fill whole buffer",
+            ));
+        }
+        Ok(())
+    }
+}
+
+impl AsRef<[u8]> for OwnedBytes {
+    #[inline]
+    fn as_ref(&self) -> &[u8] {
+        self.as_slice()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use std::io::{self, Read};
+
+    use super::OwnedBytes;
+
+    #[test]
+    fn test_owned_bytes_debug() {
+        let short_bytes = OwnedBytes::new(b"abcd".as_ref());
+        assert_eq!(
+            format!("{:?}", short_bytes),
+            "OwnedBytes([97, 98, 99, 100], len=4)"
+        );
+        let long_bytes = OwnedBytes::new(b"abcdefghijklmnopq".as_ref());
+        assert_eq!(
+            format!("{:?}", long_bytes),
+            "OwnedBytes([97, 98, 99, 100, 101, 102, 103, 104, 105, 106], len=17)"
+        );
+    }
+
+    #[test]
+    fn test_owned_bytes_read() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcdefghiklmnopqrstuvwxyz".as_ref());
+        {
+            let mut buf = [0u8; 5];
+            bytes.read_exact(&mut buf[..]).unwrap();
+            assert_eq!(&buf, b"abcde");
+            assert_eq!(bytes.as_slice(), b"fghiklmnopqrstuvwxyz")
+        }
+        {
+            let mut buf = [0u8; 2];
+            bytes.read_exact(&mut buf[..]).unwrap();
+            assert_eq!(&buf, b"fg");
+            assert_eq!(bytes.as_slice(), b"hiklmnopqrstuvwxyz")
+        }
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_right_at_the_end() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
+        let mut buf = [0u8; 5];
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 5);
+        assert_eq!(&buf, b"abcde");
+        assert_eq!(bytes.as_slice(), b"");
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 0);
+        assert_eq!(&buf, b"abcde");
+        Ok(())
+    }
+    #[test]
+    fn test_owned_bytes_read_incomplete() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
+        let mut buf = [0u8; 7];
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 5);
+        assert_eq!(&buf[..5], b"abcde");
+        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_to_end() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
+        let mut buf = Vec::new();
+        bytes.read_to_end(&mut buf)?;
+        assert_eq!(buf.as_slice(), b"abcde".as_ref());
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_u8() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"\xFF".as_ref());
+        assert_eq!(bytes.read_u8(), 255);
+        assert_eq!(bytes.len(), 0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_read_u64() -> io::Result<()> {
+        let mut bytes = OwnedBytes::new(b"\0\xFF\xFF\xFF\xFF\xFF\xFF\xFF".as_ref());
+        assert_eq!(bytes.read_u64(), u64::MAX - 255);
+        assert_eq!(bytes.len(), 0);
+        Ok(())
+    }
+
+    #[test]
+    fn test_owned_bytes_split() {
+        let bytes = OwnedBytes::new(b"abcdefghi".as_ref());
+        let (left, right) = bytes.split(3);
+        assert_eq!(left.as_slice(), b"abc");
+        assert_eq!(right.as_slice(), b"defghi");
+    }
+
+    #[test]
+    fn test_owned_bytes_split_boundary() {
+        let bytes = OwnedBytes::new(b"abcdefghi".as_ref());
+        {
+            let (left, right) = bytes.clone().split(0);
+            assert_eq!(left.as_slice(), b"");
+            assert_eq!(right.as_slice(), b"abcdefghi");
+        }
+        {
+            let (left, right) = bytes.split(9);
+            assert_eq!(left.as_slice(), b"abcdefghi");
+            assert_eq!(right.as_slice(), b"");
+        }
+    }
+}

query-grammar/Cargo.toml

@@ -14,3 +14,5 @@ edition = "2018"
 
 [dependencies]
 combine = {version="4", default-features=false, features=[] }
+once_cell = "1.7.2"
+regex ={ version = "1.5.4", default-features = false, features = ["std"] }

query-grammar/src/query_grammar.rs

@@ -1,21 +1,44 @@
 use super::user_input_ast::{UserInputAst, UserInputBound, UserInputLeaf, UserInputLiteral};
 use crate::Occur;
-use combine::parser::char::{char, digit, letter, space, spaces, string};
+use combine::parser::char::{char, digit, space, spaces, string};
+use combine::parser::range::{take_while, take_while1};
+use combine::parser::repeat::escaped;
 use combine::parser::Parser;
 use combine::{
     attempt, choice, eof, many, many1, one_of, optional, parser, satisfy, skip_many1, value,
 };
 use combine::{error::StringStreamError, parser::combinator::recognize};
+use once_cell::sync::Lazy;
+use regex::Regex;
 
-fn field<'a>() -> impl Parser<&'a str, Output = String> {
-    (
-        (letter().or(char('_'))),
-        many(satisfy(|c: char| {
-            c.is_alphanumeric() || c == '_' || c == '-'
-        })),
-    )
-        .skip(char(':'))
-        .map(|(s1, s2): (char, String)| format!("{}{}", s1, s2))
+// Note: '-' char is only forbidden at the beginning of a field name, would be clearer to add it to special characters.
+const SPECIAL_CHARS: &[char] = &[
+    '+', '^', '`', ':', '{', '}', '"', '[', ']', '(', ')', '~', '!', '\\', '*', ' ',
+];
+const ESCAPED_SPECIAL_CHARS_PATTERN: &str = r#"\\(\+|\^|`|:|\{|\}|"|\[|\]|\(|\)|\~|!|\\|\*| )"#;
+
+/// Parses a field_name
+/// A field name must have at least one character and be followed by a colon.
+/// All characters are allowed including special characters `SPECIAL_CHARS`, but these
+/// need to be escaped with a backslack character '\'.
+fn field_name<'a>() -> impl Parser<&'a str, Output = String> {
+    static ESCAPED_SPECIAL_CHARS_RE: Lazy<Regex> =
+        Lazy::new(|| Regex::new(ESCAPED_SPECIAL_CHARS_PATTERN).unwrap());
+
+    recognize::<String, _, _>(escaped(
+        (
+            take_while1(|c| !SPECIAL_CHARS.contains(&c) && c != '-'),
+            take_while(|c| !SPECIAL_CHARS.contains(&c)),
+        ),
+        '\\',
+        satisfy(|c| SPECIAL_CHARS.contains(&c)),
+    ))
+    .skip(char(':'))
+    .map(|s| ESCAPED_SPECIAL_CHARS_RE.replace_all(&s, "$1").to_string())
+    .and_then(|s: String| match s.is_empty() {
+        true => Err(StringStreamError::UnexpectedParse),
+        _ => Ok(s),
+    })
 }
 
 fn word<'a>() -> impl Parser<&'a str, Output = String> {
@@ -98,7 +121,7 @@ fn term_val<'a>() -> impl Parser<&'a str, Output = String> {
 
 fn term_query<'a>() -> impl Parser<&'a str, Output = UserInputLiteral> {
     let term_val_with_field = negative_number().or(term_val());
-    (field(), term_val_with_field).map(|(field_name, phrase)| UserInputLiteral {
+    (field_name(), term_val_with_field).map(|(field_name, phrase)| UserInputLiteral {
         field_name: Some(field_name),
         phrase,
     })
@@ -195,7 +218,7 @@ fn range<'a>() -> impl Parser<&'a str, Output = UserInputLeaf> {
     );
 
     (
-        optional(field()).skip(spaces()),
+        optional(field_name()).skip(spaces()),
         // try elastic first, if it matches, the range is unbounded
         attempt(elastic_unbounded_range).or(lower_to_upper),
     )
@@ -464,21 +487,21 @@ mod test {
 
     #[test]
     fn test_parse_elastic_query_ranges() {
-        test_parse_query_to_ast_helper("title: >a", "title:{\"a\" TO \"*\"}");
-        test_parse_query_to_ast_helper("title:>=a", "title:[\"a\" TO \"*\"}");
-        test_parse_query_to_ast_helper("title: <a", "title:{\"*\" TO \"a\"}");
-        test_parse_query_to_ast_helper("title:<=a", "title:{\"*\" TO \"a\"]");
-        test_parse_query_to_ast_helper("title:<=bsd", "title:{\"*\" TO \"bsd\"]");
+        test_parse_query_to_ast_helper("title: >a", "\"title\":{\"a\" TO \"*\"}");
+        test_parse_query_to_ast_helper("title:>=a", "\"title\":[\"a\" TO \"*\"}");
+        test_parse_query_to_ast_helper("title: <a", "\"title\":{\"*\" TO \"a\"}");
+        test_parse_query_to_ast_helper("title:<=a", "\"title\":{\"*\" TO \"a\"]");
+        test_parse_query_to_ast_helper("title:<=bsd", "\"title\":{\"*\" TO \"bsd\"]");
 
-        test_parse_query_to_ast_helper("weight: >70", "weight:{\"70\" TO \"*\"}");
-        test_parse_query_to_ast_helper("weight:>=70", "weight:[\"70\" TO \"*\"}");
-        test_parse_query_to_ast_helper("weight: <70", "weight:{\"*\" TO \"70\"}");
-        test_parse_query_to_ast_helper("weight:<=70", "weight:{\"*\" TO \"70\"]");
-        test_parse_query_to_ast_helper("weight: >60.7", "weight:{\"60.7\" TO \"*\"}");
+        test_parse_query_to_ast_helper("weight: >70", "\"weight\":{\"70\" TO \"*\"}");
+        test_parse_query_to_ast_helper("weight:>=70", "\"weight\":[\"70\" TO \"*\"}");
+        test_parse_query_to_ast_helper("weight: <70", "\"weight\":{\"*\" TO \"70\"}");
+        test_parse_query_to_ast_helper("weight:<=70", "\"weight\":{\"*\" TO \"70\"]");
+        test_parse_query_to_ast_helper("weight: >60.7", "\"weight\":{\"60.7\" TO \"*\"}");
 
-        test_parse_query_to_ast_helper("weight: <= 70", "weight:{\"*\" TO \"70\"]");
+        test_parse_query_to_ast_helper("weight: <= 70", "\"weight\":{\"*\" TO \"70\"]");
 
-        test_parse_query_to_ast_helper("weight: <= 70.5", "weight:{\"*\" TO \"70.5\"]");
+        test_parse_query_to_ast_helper("weight: <= 70.5", "\"weight\":{\"*\" TO \"70.5\"]");
     }
 
     #[test]
@@ -491,22 +514,43 @@ mod test {
     #[test]
     fn test_field_name() -> TestParseResult {
         assert_eq!(
-            super::field().parse("my-field-name:a")?,
-            ("my-field-name".to_string(), "a")
+            super::field_name().parse(".my.field.name:a"),
+            Ok((".my.field.name".to_string(), "a"))
         );
         assert_eq!(
-            super::field().parse("my_field_name:a")?,
-            ("my_field_name".to_string(), "a")
+            super::field_name().parse("my\\ field\\ name:a"),
+            Ok(("my field name".to_string(), "a"))
         );
-        assert!(super::field().parse(":a").is_err());
-        assert!(super::field().parse("-my_field:a").is_err());
+        assert!(super::field_name().parse("my field:a").is_err());
         assert_eq!(
-            super::field().parse("_my_field:a")?,
+            super::field_name().parse("\\(1\\+1\\):2"),
+            Ok(("(1+1)".to_string(), "2"))
+        );
+        assert_eq!(
+            super::field_name().parse("my_field_name:a"),
+            Ok(("my_field_name".to_string(), "a"))
+        );
+        assert!(super::field_name().parse("my_field_name").is_err());
+        assert!(super::field_name().parse(":a").is_err());
+        assert!(super::field_name().parse("-my_field:a").is_err());
+        assert_eq!(
+            super::field_name().parse("_my_field:a")?,
             ("_my_field".to_string(), "a")
         );
         Ok(())
     }
 
+    #[test]
+    fn test_field_name_re() {
+        let escaped_special_chars_re = Regex::new(ESCAPED_SPECIAL_CHARS_PATTERN).unwrap();
+        for special_char in SPECIAL_CHARS.iter() {
+            assert_eq!(
+                escaped_special_chars_re.replace_all(&format!("\\{}", special_char), "$1"),
+                special_char.to_string()
+            );
+        }
+    }
+
     #[test]
     fn test_range_parser() {
         // testing the range() parser separately
@@ -600,12 +644,14 @@ mod test {
 
     #[test]
     fn test_single_term_with_field() {
-        test_parse_query_to_ast_helper("abc:toto", "abc:\"toto\"");
+        test_parse_query_to_ast_helper("abc:toto", "\"abc\":\"toto\"");
     }
 
     #[test]
     fn test_single_term_with_float() {
-        test_parse_query_to_ast_helper("abc:1.1", "abc:\"1.1\"");
+        test_parse_query_to_ast_helper("abc:1.1", "\"abc\":\"1.1\"");
+        test_parse_query_to_ast_helper("a.b.c:1.1", "\"a.b.c\":\"1.1\"");
+        test_parse_query_to_ast_helper("a\\ b\\ c:1.1", "\"a b c\":\"1.1\"");
     }
 
     #[test]
@@ -621,22 +667,27 @@ mod test {
     #[test]
     fn test_parse_test_query_other() {
         test_parse_query_to_ast_helper("(+a +b) d", "(*(+\"a\" +\"b\") *\"d\")");
-        test_parse_query_to_ast_helper("+abc:toto", "abc:\"toto\"");
-        test_parse_query_to_ast_helper("(+abc:toto -titi)", "(+abc:\"toto\" -\"titi\")");
-        test_parse_query_to_ast_helper("-abc:toto", "(-abc:\"toto\")");
-        test_parse_query_to_ast_helper("abc:a b", "(*abc:\"a\" *\"b\")");
-        test_parse_query_to_ast_helper("abc:\"a b\"", "abc:\"a b\"");
-        test_parse_query_to_ast_helper("foo:[1 TO 5]", "foo:[\"1\" TO \"5\"]");
+        test_parse_query_to_ast_helper("+abc:toto", "\"abc\":\"toto\"");
+        test_parse_query_to_ast_helper("+a\\+b\\+c:toto", "\"a+b+c\":\"toto\"");
+        test_parse_query_to_ast_helper("(+abc:toto -titi)", "(+\"abc\":\"toto\" -\"titi\")");
+        test_parse_query_to_ast_helper("-abc:toto", "(-\"abc\":\"toto\")");
+        test_is_parse_err("--abc:toto");
+        test_parse_query_to_ast_helper("abc:a b", "(*\"abc\":\"a\" *\"b\")");
+        test_parse_query_to_ast_helper("abc:\"a b\"", "\"abc\":\"a b\"");
+        test_parse_query_to_ast_helper("foo:[1 TO 5]", "\"foo\":[\"1\" TO \"5\"]");
     }
 
     #[test]
     fn test_parse_query_with_range() {
         test_parse_query_to_ast_helper("[1 TO 5]", "[\"1\" TO \"5\"]");
-        test_parse_query_to_ast_helper("foo:{a TO z}", "foo:{\"a\" TO \"z\"}");
-        test_parse_query_to_ast_helper("foo:[1 TO toto}", "foo:[\"1\" TO \"toto\"}");
-        test_parse_query_to_ast_helper("foo:[* TO toto}", "foo:{\"*\" TO \"toto\"}");
-        test_parse_query_to_ast_helper("foo:[1 TO *}", "foo:[\"1\" TO \"*\"}");
-        test_parse_query_to_ast_helper("foo:[1.1 TO *}", "foo:[\"1.1\" TO \"*\"}");
+        test_parse_query_to_ast_helper("foo:{a TO z}", "\"foo\":{\"a\" TO \"z\"}");
+        test_parse_query_to_ast_helper("foo:[1 TO toto}", "\"foo\":[\"1\" TO \"toto\"}");
+        test_parse_query_to_ast_helper("foo:[* TO toto}", "\"foo\":{\"*\" TO \"toto\"}");
+        test_parse_query_to_ast_helper("foo:[1 TO *}", "\"foo\":[\"1\" TO \"*\"}");
+        test_parse_query_to_ast_helper(
+            "1.2.foo.bar:[1.1 TO *}",
+            "\"1.2.foo.bar\":[\"1.1\" TO \"*\"}",
+        );
         test_is_parse_err("abc +    ");
     }
 }

query-grammar/src/user_input_ast.rs

@@ -24,7 +24,7 @@ impl Debug for UserInputLeaf {
                 ref upper,
             } => {
                 if let Some(ref field) = field {
-                    write!(formatter, "{}:", field)?;
+                    write!(formatter, "\"{}\":", field)?;
                 }
                 lower.display_lower(formatter)?;
                 write!(formatter, " TO ")?;
@@ -45,7 +45,7 @@ pub struct UserInputLiteral {
 impl fmt::Debug for UserInputLiteral {
     fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> Result<(), fmt::Error> {
         match self.field_name {
-            Some(ref field_name) => write!(formatter, "{}:\"{}\"", field_name, self.phrase),
+            Some(ref field_name) => write!(formatter, "\"{}\":\"{}\"", field_name, self.phrase),
             None => write!(formatter, "\"{}\"", self.phrase),
         }
     }
@@ -79,7 +79,7 @@ impl UserInputBound {
         match *self {
             UserInputBound::Inclusive(ref contents) => contents,
             UserInputBound::Exclusive(ref contents) => contents,
-            UserInputBound::Unbounded => &"*",
+            UserInputBound::Unbounded => "*",
         }
     }
 }

src/collector/facet_collector.rs

@@ -297,10 +297,8 @@ impl Collector for FacetCollector {
                                 if depth == collapse_depth + 1 {
                                     collapsed_id = collapse_facet_ords.len();
                                     collapse_facet_ords.push(facet_streamer.term_ord());
-                                    collapse_mapping.push(collapsed_id);
-                                } else {
-                                    collapse_mapping.push(collapsed_id);
                                 }
+                                collapse_mapping.push(collapsed_id);
                             }
                             break;
                         }

src/collector/filter_collector_wrapper.rs

@@ -16,7 +16,7 @@ use crate::fastfield::{DynamicFastFieldReader, FastFieldReader, FastValue};
 use crate::schema::Field;
 use crate::{Score, SegmentReader, TantivyError};
 
-/// The `FilterCollector` collector filters docs using a u64 fast field value and a predicate.
+/// The `FilterCollector` collector filters docs using a fast field value and a predicate.
 /// Only the documents for which the predicate returned "true" will be passed on to the next collector.
 ///
 /// ```rust

src/collector/multi_collector.rs

@@ -276,7 +276,7 @@ mod tests {
         let mut collectors = MultiCollector::new();
         let topdocs_handler = collectors.add_collector(TopDocs::with_limit(2));
         let count_handler = collectors.add_collector(Count);
-        let mut multifruits = searcher.search(&query, &mut collectors).unwrap();
+        let mut multifruits = searcher.search(&query, &collectors).unwrap();
 
         assert_eq!(count_handler.extract(&mut multifruits), 5);
         assert_eq!(topdocs_handler.extract(&mut multifruits).len(), 2);

src/collector/top_score_collector.rs

@@ -1080,7 +1080,7 @@ mod tests {
         query: &str,
         query_field: Field,
         schema: Schema,
-        mut doc_adder: impl FnMut(&mut IndexWriter) -> (),
+        mut doc_adder: impl FnMut(&mut IndexWriter),
     ) -> (Index, Box<dyn Query>) {
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_with_num_threads(1, 10_000_000).unwrap();

src/common/mod.rs

@@ -1,18 +1,15 @@
 mod bitset;
 mod composite_file;
-mod counting_writer;
-mod serialize;
-mod vint;
 
 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, DeserializeFrom, FixedSize};
-pub use self::vint::{
+pub use byteorder::LittleEndian as Endianness;
+pub use common::CountingWriter;
+pub use common::{
     read_u32_vint, read_u32_vint_no_advance, serialize_vint_u32, write_u32_vint, VInt,
 };
-pub use byteorder::LittleEndian as Endianness;
+pub use common::{BinarySerializable, DeserializeFrom, FixedSize};
 
 /// Segment's max doc must be `< MAX_DOC_LIMIT`.
 ///
@@ -103,8 +100,8 @@ pub fn u64_to_f64(val: u64) -> f64 {
 #[cfg(test)]
 pub(crate) mod test {
 
-    pub use super::serialize::test::fixed_size_test;
     use super::{f64_to_u64, i64_to_u64, u64_to_f64, u64_to_i64};
+    use common::{BinarySerializable, FixedSize};
     use proptest::prelude::*;
     use std::f64;
     use tantivy_bitpacker::compute_num_bits;
@@ -118,6 +115,12 @@ pub(crate) mod test {
         assert_eq!(u64_to_f64(f64_to_u64(val)), val);
     }
 
+    pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
+        let mut buffer = Vec::new();
+        O::default().serialize(&mut buffer).unwrap();
+        assert_eq!(buffer.len(), O::SIZE_IN_BYTES);
+    }
+
     proptest! {
         #[test]
         fn test_f64_converter_monotonicity_proptest((left, right) in (proptest::num::f64::NORMAL, proptest::num::f64::NORMAL)) {

src/core/index.rs

@@ -42,7 +42,7 @@ fn load_metas(
             "Meta file does not contain valid utf8 file.".to_string(),
         )
     })?;
-    IndexMeta::deserialize(&meta_string, &inventory)
+    IndexMeta::deserialize(&meta_string, inventory)
         .map_err(|e| {
             DataCorruption::new(
                 META_FILEPATH.to_path_buf(),
@@ -604,7 +604,7 @@ mod tests {
         .is_ok());
         assert!(Index::exists(&directory).unwrap());
         assert!(Index::create(
-            directory.clone(),
+            directory,
             Schema::builder().build(),
             IndexSettings::default()
         )

src/core/index_meta.rs

@@ -369,7 +369,6 @@ mod tests {
         schema::{Schema, TEXT},
         IndexSettings, IndexSortByField, Order,
     };
-    use serde_json;
 
     #[test]
     fn test_serialize_metas() {

src/core/mod.rs

@@ -16,7 +16,6 @@ pub use self::index_meta::{
 pub use self::inverted_index_reader::InvertedIndexReader;
 pub use self::searcher::Searcher;
 pub use self::segment::Segment;
-pub use self::segment::SerializableSegment;
 pub use self::segment_component::SegmentComponent;
 pub use self::segment_id::SegmentId;
 pub use self::segment_reader::SegmentReader;

src/core/segment.rs

@@ -1,13 +1,12 @@
 use super::SegmentComponent;
+use crate::core::Index;
 use crate::core::SegmentId;
 use crate::core::SegmentMeta;
 use crate::directory::error::{OpenReadError, OpenWriteError};
 use crate::directory::Directory;
 use crate::directory::{FileSlice, WritePtr};
-use crate::indexer::segment_serializer::SegmentSerializer;
 use crate::schema::Schema;
 use crate::Opstamp;
-use crate::{core::Index, indexer::doc_id_mapping::DocIdMapping};
 use std::fmt;
 use std::path::PathBuf;
 
@@ -90,20 +89,3 @@ impl Segment {
         Ok(write)
     }
 }
-
-pub trait SerializableSegment {
-    /// Writes a view of a segment by pushing information
-    /// to the `SegmentSerializer`.
-    ///
-    /// # Returns
-    /// The number of documents in the segment.
-    ///
-    /// doc_id_map is used when index is created and sorted, to map to the new doc_id order.
-    /// It is not used by the `IndexMerger`, since the doc_id_mapping on cross-segments works
-    /// differently
-    fn write(
-        &self,
-        serializer: SegmentSerializer,
-        doc_id_map: Option<&DocIdMapping>,
-    ) -> crate::Result<u32>;
-}

src/core/segment_id.rs

@@ -22,7 +22,7 @@ use std::sync::atomic;
 pub struct SegmentId(Uuid);
 
 #[cfg(test)]
-static AUTO_INC_COUNTER: Lazy<atomic::AtomicUsize> = Lazy::new(|| atomic::AtomicUsize::default());
+static AUTO_INC_COUNTER: Lazy<atomic::AtomicUsize> = Lazy::new(atomic::AtomicUsize::default);
 
 #[cfg(test)]
 const ZERO_ARRAY: [u8; 8] = [0u8; 8];
@@ -108,6 +108,12 @@ impl fmt::Debug for SegmentId {
     }
 }
 
+impl fmt::Display for SegmentId {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "Seg({:?})", self.short_uuid_string())
+    }
+}
+
 impl PartialOrd for SegmentId {
     fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
         Some(self.cmp(other))

src/core/segment_reader.rs

@@ -1,4 +1,3 @@
-use crate::common::HasLen;
 use crate::core::InvertedIndexReader;
 use crate::core::Segment;
 use crate::core::SegmentComponent;
@@ -32,9 +31,6 @@ use std::{collections::HashMap, io};
 ///
 /// The segment reader has a very low memory footprint,
 /// as close to all of the memory data is mmapped.
-///
-///
-/// TODO fix not decoding docfreq
 #[derive(Clone)]
 pub struct SegmentReader {
     inv_idx_reader_cache: Arc<RwLock<HashMap<Field, Arc<InvertedIndexReader>>>>,
@@ -57,17 +53,12 @@ pub struct SegmentReader {
 impl SegmentReader {
     /// Returns the highest document id ever attributed in
     /// this segment + 1.
-    /// Today, `tantivy` does not handle deletes, so it happens
-    /// to also be the number of documents in the index.
     pub fn max_doc(&self) -> DocId {
         self.max_doc
     }
 
-    /// Returns the number of documents.
+    /// Returns the number of alive documents.
     /// Deleted documents are not counted.
-    ///
-    /// Today, `tantivy` does not handle deletes so max doc and
-    /// num_docs are the same.
     pub fn num_docs(&self) -> DocId {
         self.num_docs
     }
@@ -81,7 +72,7 @@ impl SegmentReader {
     /// deleted in the segment.
     pub fn num_deleted_docs(&self) -> DocId {
         self.delete_bitset()
-            .map(|delete_set| delete_set.len() as DocId)
+            .map(|delete_set| delete_set.num_deleted() as DocId)
             .unwrap_or(0u32)
     }
 
@@ -329,6 +320,32 @@ mod test {
     use crate::schema::{Schema, Term, STORED, TEXT};
     use crate::DocId;
 
+    #[test]
+    fn test_num_alive() -> crate::Result<()> {
+        let mut schema_builder = Schema::builder();
+        schema_builder.add_text_field("name", TEXT | STORED);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema.clone());
+        let name = schema.get_field("name").unwrap();
+
+        {
+            let mut index_writer = index.writer_for_tests()?;
+            index_writer.add_document(doc!(name => "tantivy"));
+            index_writer.add_document(doc!(name => "horse"));
+            index_writer.add_document(doc!(name => "jockey"));
+            index_writer.add_document(doc!(name => "cap"));
+            // we should now have one segment with two docs
+            index_writer.delete_term(Term::from_field_text(name, "horse"));
+            index_writer.delete_term(Term::from_field_text(name, "cap"));
+
+            // ok, now we should have a deleted doc
+            index_writer.commit()?;
+        }
+        let searcher = index.reader()?.searcher();
+        assert_eq!(2, searcher.segment_reader(0).num_docs());
+        assert_eq!(4, searcher.segment_reader(0).max_doc());
+        Ok(())
+    }
     #[test]
     fn test_alive_docs_iterator() -> crate::Result<()> {
         let mut schema_builder = Schema::builder();

src/directory/file_slice.rs

@@ -211,7 +211,7 @@ mod tests {
             assert_eq!(right.read_bytes()?.as_slice(), b"");
         }
         {
-            let (left, right) = file_slice.clone().split_from_end(2);
+            let (left, right) = file_slice.split_from_end(2);
             assert_eq!(left.read_bytes()?.as_slice(), b"abcd");
             assert_eq!(right.read_bytes()?.as_slice(), b"ef");
         }

src/directory/managed_directory.rs

@@ -402,10 +402,8 @@ mod tests_mmap_specific {
             // The file should still be in the list of managed file and
             // eventually be deleted once mmap is released.
             assert!(managed_directory.garbage_collect(|| living_files).is_ok());
-            assert!(!managed_directory.exists(test_path1).unwrap());
-        } else {
-            assert!(!managed_directory.exists(test_path1).unwrap());
         }
+        assert!(!managed_directory.exists(test_path1).unwrap());
     }
 
     #[test]
@@ -430,7 +428,7 @@ mod tests_mmap_specific {
         assert_eq!(read_file.as_slice(), &[3u8, 4u8, 5u8]);
         assert!(managed_directory.list_damaged().unwrap().is_empty());
 
-        let mut corrupted_path = tempdir_path.clone();
+        let mut corrupted_path = tempdir_path;
         corrupted_path.push(test_path2);
         let mut file = OpenOptions::new().write(true).open(&corrupted_path)?;
         file.write_all(&[255u8])?;

src/directory/mod.rs

@@ -28,7 +28,9 @@ pub use self::file_slice::{FileHandle, FileSlice};
 pub use self::owned_bytes::OwnedBytes;
 pub use self::ram_directory::RamDirectory;
 pub use self::watch_event_router::{WatchCallback, WatchCallbackList, WatchHandle};
-use std::io::{self, BufWriter, Write};
+pub use common::AntiCallToken;
+pub use common::TerminatingWrite;
+use std::io::BufWriter;
 use std::path::PathBuf;
 
 /// Outcome of the Garbage collection
@@ -50,47 +52,6 @@ pub use self::mmap_directory::MmapDirectory;
 
 pub use self::managed_directory::ManagedDirectory;
 
-/// Struct used to prevent from calling [`terminate_ref`](trait.TerminatingWrite#method.terminate_ref) directly
-///
-/// The point is that while the type is public, it cannot be built by anyone
-/// outside of this module.
-pub struct AntiCallToken(());
-
-/// Trait used to indicate when no more write need to be done on a writer
-pub trait TerminatingWrite: Write {
-    /// Indicate that the writer will no longer be used. Internally call terminate_ref.
-    fn terminate(mut self) -> io::Result<()>
-    where
-        Self: Sized,
-    {
-        self.terminate_ref(AntiCallToken(()))
-    }
-
-    /// You should implement this function to define custom behavior.
-    /// This function should flush any buffer it may hold.
-    fn terminate_ref(&mut self, _: AntiCallToken) -> io::Result<()>;
-}
-
-impl<W: TerminatingWrite + ?Sized> TerminatingWrite for Box<W> {
-    fn terminate_ref(&mut self, token: AntiCallToken) -> io::Result<()> {
-        self.as_mut().terminate_ref(token)
-    }
-}
-
-impl<W: TerminatingWrite> TerminatingWrite for BufWriter<W> {
-    fn terminate_ref(&mut self, a: AntiCallToken) -> io::Result<()> {
-        self.flush()?;
-        self.get_mut().terminate_ref(a)
-    }
-}
-
-#[cfg(test)]
-impl<'a> TerminatingWrite for &'a mut Vec<u8> {
-    fn terminate_ref(&mut self, _a: AntiCallToken) -> io::Result<()> {
-        self.flush()
-    }
-}
-
 /// Write object for Directory.
 ///
 /// `WritePtr` are required to implement both Write

src/directory/owned_bytes.rs

@@ -1,290 +1,11 @@
 use crate::directory::FileHandle;
-use stable_deref_trait::StableDeref;
-use std::convert::TryInto;
-use std::mem;
-use std::ops::{Deref, Range};
-use std::sync::Arc;
-use std::{fmt, io};
+use std::io;
+use std::ops::Range;
 
-/// An OwnedBytes simply wraps an object that owns a slice of data and exposes
-/// this data as a static slice.
-///
-/// The backing object is required to be `StableDeref`.
-#[derive(Clone)]
-pub struct OwnedBytes {
-    data: &'static [u8],
-    box_stable_deref: Arc<dyn Deref<Target = [u8]> + Sync + Send>,
-}
+pub use ownedbytes::OwnedBytes;
 
 impl FileHandle for OwnedBytes {
     fn read_bytes(&self, range: Range<usize>) -> io::Result<OwnedBytes> {
         Ok(self.slice(range))
     }
 }
-
-impl OwnedBytes {
-    /// Creates an empty `OwnedBytes`.
-    pub fn empty() -> OwnedBytes {
-        OwnedBytes::new(&[][..])
-    }
-
-    /// Creates an `OwnedBytes` intance given a `StableDeref` object.
-    pub fn new<T: StableDeref + Deref<Target = [u8]> + 'static + Send + Sync>(
-        data_holder: T,
-    ) -> OwnedBytes {
-        let box_stable_deref = Arc::new(data_holder);
-        let bytes: &[u8] = box_stable_deref.as_ref();
-        let data = unsafe { mem::transmute::<_, &'static [u8]>(bytes.deref()) };
-        OwnedBytes {
-            data,
-            box_stable_deref,
-        }
-    }
-
-    /// creates a fileslice that is just a view over a slice of the data.
-    pub fn slice(&self, range: Range<usize>) -> Self {
-        OwnedBytes {
-            data: &self.data[range],
-            box_stable_deref: self.box_stable_deref.clone(),
-        }
-    }
-
-    /// Returns the underlying slice of data.
-    /// `Deref` and `AsRef` are also available.
-    #[inline]
-    pub fn as_slice(&self) -> &[u8] {
-        self.data
-    }
-
-    /// Returns the len of the slice.
-    #[inline]
-    pub fn len(&self) -> usize {
-        self.data.len()
-    }
-
-    /// Splits the OwnedBytes into two OwnedBytes `(left, right)`.
-    ///
-    /// Left will hold `split_len` bytes.
-    ///
-    /// This operation is cheap and does not require to copy any memory.
-    /// On the other hand, both `left` and `right` retain a handle over
-    /// the entire slice of memory. In other words, the memory will only
-    /// be released when both left and right are dropped.
-    pub fn split(self, split_len: usize) -> (OwnedBytes, OwnedBytes) {
-        let right_box_stable_deref = self.box_stable_deref.clone();
-        let left = OwnedBytes {
-            data: &self.data[..split_len],
-            box_stable_deref: self.box_stable_deref,
-        };
-        let right = OwnedBytes {
-            data: &self.data[split_len..],
-            box_stable_deref: right_box_stable_deref,
-        };
-        (left, right)
-    }
-
-    /// Returns true iff this `OwnedBytes` is empty.
-    #[inline]
-    pub fn is_empty(&self) -> bool {
-        self.as_slice().is_empty()
-    }
-
-    /// Drops the left most `advance_len` bytes.
-    ///
-    /// See also [.clip(clip_len: usize))](#method.clip).
-    #[inline]
-    pub fn advance(&mut self, advance_len: usize) {
-        self.data = &self.data[advance_len..]
-    }
-
-    /// Reads an `u8` from the `OwnedBytes` and advance by one byte.
-    pub fn read_u8(&mut self) -> u8 {
-        assert!(!self.is_empty());
-
-        let byte = self.as_slice()[0];
-        self.advance(1);
-        byte
-    }
-
-    /// Reads an `u64` encoded as little-endian from the `OwnedBytes` and advance by 8 bytes.
-    pub fn read_u64(&mut self) -> u64 {
-        assert!(self.len() > 7);
-
-        let octlet: [u8; 8] = self.as_slice()[..8].try_into().unwrap();
-        self.advance(8);
-        u64::from_le_bytes(octlet)
-    }
-}
-
-impl fmt::Debug for OwnedBytes {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        // We truncate the bytes in order to make sure the debug string
-        // is not too long.
-        let bytes_truncated: &[u8] = if self.len() > 8 {
-            &self.as_slice()[..10]
-        } else {
-            self.as_slice()
-        };
-        write!(f, "OwnedBytes({:?}, len={})", bytes_truncated, self.len())
-    }
-}
-
-impl Deref for OwnedBytes {
-    type Target = [u8];
-
-    fn deref(&self) -> &Self::Target {
-        self.as_slice()
-    }
-}
-
-impl io::Read for OwnedBytes {
-    fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
-        let read_len = {
-            let data = self.as_slice();
-            if data.len() >= buf.len() {
-                let buf_len = buf.len();
-                buf.copy_from_slice(&data[..buf_len]);
-                buf.len()
-            } else {
-                let data_len = data.len();
-                buf[..data_len].copy_from_slice(data);
-                data_len
-            }
-        };
-        self.advance(read_len);
-        Ok(read_len)
-    }
-    fn read_to_end(&mut self, buf: &mut Vec<u8>) -> io::Result<usize> {
-        let read_len = {
-            let data = self.as_slice();
-            buf.extend(data);
-            data.len()
-        };
-        self.advance(read_len);
-        Ok(read_len)
-    }
-    fn read_exact(&mut self, buf: &mut [u8]) -> io::Result<()> {
-        let read_len = self.read(buf)?;
-        if read_len != buf.len() {
-            return Err(io::Error::new(
-                io::ErrorKind::UnexpectedEof,
-                "failed to fill whole buffer",
-            ));
-        }
-        Ok(())
-    }
-}
-
-impl AsRef<[u8]> for OwnedBytes {
-    fn as_ref(&self) -> &[u8] {
-        self.as_slice()
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use std::io::{self, Read};
-
-    use super::OwnedBytes;
-
-    #[test]
-    fn test_owned_bytes_debug() {
-        let short_bytes = OwnedBytes::new(b"abcd".as_ref());
-        assert_eq!(
-            format!("{:?}", short_bytes),
-            "OwnedBytes([97, 98, 99, 100], len=4)"
-        );
-        let long_bytes = OwnedBytes::new(b"abcdefghijklmnopq".as_ref());
-        assert_eq!(
-            format!("{:?}", long_bytes),
-            "OwnedBytes([97, 98, 99, 100, 101, 102, 103, 104, 105, 106], len=17)"
-        );
-    }
-
-    #[test]
-    fn test_owned_bytes_read() -> io::Result<()> {
-        let mut bytes = OwnedBytes::new(b"abcdefghiklmnopqrstuvwxyz".as_ref());
-        {
-            let mut buf = [0u8; 5];
-            bytes.read_exact(&mut buf[..]).unwrap();
-            assert_eq!(&buf, b"abcde");
-            assert_eq!(bytes.as_slice(), b"fghiklmnopqrstuvwxyz")
-        }
-        {
-            let mut buf = [0u8; 2];
-            bytes.read_exact(&mut buf[..]).unwrap();
-            assert_eq!(&buf, b"fg");
-            assert_eq!(bytes.as_slice(), b"hiklmnopqrstuvwxyz")
-        }
-        Ok(())
-    }
-
-    #[test]
-    fn test_owned_bytes_read_right_at_the_end() -> io::Result<()> {
-        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
-        let mut buf = [0u8; 5];
-        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 5);
-        assert_eq!(&buf, b"abcde");
-        assert_eq!(bytes.as_slice(), b"");
-        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 0);
-        assert_eq!(&buf, b"abcde");
-        Ok(())
-    }
-    #[test]
-    fn test_owned_bytes_read_incomplete() -> io::Result<()> {
-        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
-        let mut buf = [0u8; 7];
-        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 5);
-        assert_eq!(&buf[..5], b"abcde");
-        assert_eq!(bytes.read(&mut buf[..]).unwrap(), 0);
-        Ok(())
-    }
-
-    #[test]
-    fn test_owned_bytes_read_to_end() -> io::Result<()> {
-        let mut bytes = OwnedBytes::new(b"abcde".as_ref());
-        let mut buf = Vec::new();
-        bytes.read_to_end(&mut buf)?;
-        assert_eq!(buf.as_slice(), b"abcde".as_ref());
-        Ok(())
-    }
-
-    #[test]
-    fn test_owned_bytes_read_u8() -> io::Result<()> {
-        let mut bytes = OwnedBytes::new(b"\xFF".as_ref());
-        assert_eq!(bytes.read_u8(), 255);
-        assert_eq!(bytes.len(), 0);
-        Ok(())
-    }
-
-    #[test]
-    fn test_owned_bytes_read_u64() -> io::Result<()> {
-        let mut bytes = OwnedBytes::new(b"\0\xFF\xFF\xFF\xFF\xFF\xFF\xFF".as_ref());
-        assert_eq!(bytes.read_u64(), u64::MAX - 255);
-        assert_eq!(bytes.len(), 0);
-        Ok(())
-    }
-
-    #[test]
-    fn test_owned_bytes_split() {
-        let bytes = OwnedBytes::new(b"abcdefghi".as_ref());
-        let (left, right) = bytes.split(3);
-        assert_eq!(left.as_slice(), b"abc");
-        assert_eq!(right.as_slice(), b"defghi");
-    }
-
-    #[test]
-    fn test_owned_bytes_split_boundary() {
-        let bytes = OwnedBytes::new(b"abcdefghi".as_ref());
-        {
-            let (left, right) = bytes.clone().split(0);
-            assert_eq!(left.as_slice(), b"");
-            assert_eq!(right.as_slice(), b"abcdefghi");
-        }
-        {
-            let (left, right) = bytes.split(9);
-            assert_eq!(left.as_slice(), b"abcdefghi");
-            assert_eq!(right.as_slice(), b"");
-        }
-    }
-}

src/directory/ram_directory.rs

@@ -46,7 +46,7 @@ impl Drop for VecWriter {
     fn drop(&mut self) {
         if !self.is_flushed {
             panic!(
-                "You forgot to flush {:?} before its writter got Drop. Do not rely on drop.",
+                "You forgot to flush {:?} before its writter got Drop. Do not rely on drop. This also occurs when the indexer crashed, so you may want to check the logs for the root cause.",
                 self.path
             )
         }

src/directory/tests.rs

@@ -166,26 +166,26 @@ fn test_write_create_the_file(directory: &dyn Directory) {
 fn test_directory_delete(directory: &dyn Directory) -> crate::Result<()> {
     let test_path: &'static Path = Path::new("some_path_for_test");
     assert!(directory.open_read(test_path).is_err());
-    let mut write_file = directory.open_write(&test_path)?;
+    let mut write_file = directory.open_write(test_path)?;
     write_file.write_all(&[1, 2, 3, 4])?;
     write_file.flush()?;
     {
-        let read_handle = directory.open_read(&test_path)?.read_bytes()?;
+        let read_handle = directory.open_read(test_path)?.read_bytes()?;
         assert_eq!(read_handle.as_slice(), &[1u8, 2u8, 3u8, 4u8]);
         // Mapped files can't be deleted on Windows
         if !cfg!(windows) {
-            assert!(directory.delete(&test_path).is_ok());
+            assert!(directory.delete(test_path).is_ok());
             assert_eq!(read_handle.as_slice(), &[1u8, 2u8, 3u8, 4u8]);
         }
         assert!(directory.delete(Path::new("SomeOtherPath")).is_err());
     }
 
     if cfg!(windows) {
-        assert!(directory.delete(&test_path).is_ok());
+        assert!(directory.delete(test_path).is_ok());
     }
 
-    assert!(directory.open_read(&test_path).is_err());
-    assert!(directory.delete(&test_path).is_err());
+    assert!(directory.open_read(test_path).is_err());
+    assert!(directory.delete(test_path).is_err());
     Ok(())
 }
 

src/fastfield/bytes/writer.rs

@@ -1,6 +1,5 @@
 use std::io;
 
-use crate::fastfield::serializer::FastFieldSerializer;
 use crate::schema::{Document, Field, Value};
 use crate::DocId;
 use crate::{
@@ -84,11 +83,11 @@ impl BytesFastFieldWriter {
         &'a self,
         doc_id_map: Option<&'b DocIdMapping>,
     ) -> impl Iterator<Item = &'b [u8]> {
-        let doc_id_iter = if let Some(doc_id_map) = doc_id_map {
-            Box::new(doc_id_map.iter_old_doc_ids().cloned()) as Box<dyn Iterator<Item = u32>>
+        let doc_id_iter: Box<dyn Iterator<Item = u32>> = if let Some(doc_id_map) = doc_id_map {
+            Box::new(doc_id_map.iter_old_doc_ids())
         } else {
-            Box::new(self.doc_index.iter().enumerate().map(|el| el.0 as u32))
-                as Box<dyn Iterator<Item = u32>>
+            let max_doc = self.doc_index.len() as u32;
+            Box::new(0..max_doc)
         };
         doc_id_iter.map(move |doc_id| self.get_values_for_doc_id(doc_id))
     }

src/fastfield/delete.rs

@@ -91,6 +91,10 @@ impl DeleteBitSet {
         b & (1u8 << shift) != 0
     }
 
+    /// The number of deleted docs
+    pub fn num_deleted(&self) -> usize {
+        self.num_deleted
+    }
     /// Summarize total space usage of this bitset.
     pub fn space_usage(&self) -> ByteCount {
         self.data.len()

src/fastfield/mod.rs

@@ -29,12 +29,13 @@ 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(crate) 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::serializer::FastFieldDataAccess;
+pub use self::serializer::FastFieldStats;
 pub use self::writer::{FastFieldsWriter, IntFastFieldWriter};
 use crate::schema::Cardinality;
 use crate::schema::FieldType;
@@ -213,15 +214,14 @@ mod tests {
 
     use super::*;
     use crate::common::CompositeFile;
+    use crate::common::HasLen;
     use crate::directory::{Directory, RamDirectory, WritePtr};
-    use crate::fastfield::BitpackedFastFieldReader;
     use crate::merge_policy::NoMergePolicy;
     use crate::schema::Field;
     use crate::schema::Schema;
     use crate::schema::FAST;
     use crate::schema::{Document, IntOptions};
     use crate::{Index, SegmentId, SegmentReader};
-    use common::HasLen;
     use once_cell::sync::Lazy;
     use rand::prelude::SliceRandom;
     use rand::rngs::StdRng;
@@ -239,7 +239,7 @@ mod tests {
 
     #[test]
     pub fn test_fastfield() {
-        let test_fastfield = BitpackedFastFieldReader::<u64>::from(vec![100, 200, 300]);
+        let test_fastfield = DynamicFastFieldReader::<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);
@@ -267,11 +267,11 @@ mod tests {
                 .unwrap();
             serializer.close().unwrap();
         }
-        let file = directory.open_read(&path).unwrap();
-        assert_eq!(file.len(), 36 as usize);
+        let file = directory.open_read(path).unwrap();
+        assert_eq!(file.len(), 37);
         let composite_file = CompositeFile::open(&file)?;
         let file = composite_file.open_read(*FIELD).unwrap();
-        let fast_field_reader = BitpackedFastFieldReader::<u64>::open(file)?;
+        let fast_field_reader = DynamicFastFieldReader::<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);
@@ -298,12 +298,12 @@ mod tests {
             fast_field_writers.serialize(&mut serializer, &HashMap::new(), None)?;
             serializer.close()?;
         }
-        let file = directory.open_read(&path)?;
-        assert_eq!(file.len(), 61 as usize);
+        let file = directory.open_read(path)?;
+        assert_eq!(file.len(), 62);
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = DynamicFastFieldReader::<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);
@@ -334,12 +334,12 @@ mod tests {
                 .unwrap();
             serializer.close().unwrap();
         }
-        let file = directory.open_read(&path).unwrap();
-        assert_eq!(file.len(), 34 as usize);
+        let file = directory.open_read(path).unwrap();
+        assert_eq!(file.len(), 35);
         {
             let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
             for doc in 0..10_000 {
                 assert_eq!(fast_field_reader.get(doc), 100_000u64);
             }
@@ -366,12 +366,12 @@ mod tests {
                 .unwrap();
             serializer.close().unwrap();
         }
-        let file = directory.open_read(&path).unwrap();
-        assert_eq!(file.len(), 80042 as usize);
+        let file = directory.open_read(path).unwrap();
+        assert_eq!(file.len(), 80043);
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
             assert_eq!(fast_field_reader.get(0), 0u64);
             for doc in 1..10_001 {
                 assert_eq!(
@@ -405,12 +405,13 @@ mod tests {
                 .unwrap();
             serializer.close().unwrap();
         }
-        let file = directory.open_read(&path).unwrap();
-        assert_eq!(file.len(), 17709 as usize);
+        let file = directory.open_read(path).unwrap();
+        //assert_eq!(file.len(), 17710 as usize); //bitpacked size
+        assert_eq!(file.len(), 10175_usize); // linear interpol size
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(i64_field).unwrap();
-            let fast_field_reader = BitpackedFastFieldReader::<i64>::open(data)?;
+            let fast_field_reader = DynamicFastFieldReader::<i64>::open(data)?;
 
             assert_eq!(fast_field_reader.min_value(), -100i64);
             assert_eq!(fast_field_reader.max_value(), 9_999i64);
@@ -446,11 +447,11 @@ mod tests {
             serializer.close().unwrap();
         }
 
-        let file = directory.open_read(&path).unwrap();
+        let file = directory.open_read(path).unwrap();
         {
             let fast_fields_composite = CompositeFile::open(&file).unwrap();
             let data = fast_fields_composite.open_read(i64_field).unwrap();
-            let fast_field_reader = BitpackedFastFieldReader::<i64>::open(data)?;
+            let fast_field_reader = DynamicFastFieldReader::<i64>::open(data)?;
             assert_eq!(fast_field_reader.get(0u32), 0i64);
         }
         Ok(())
@@ -479,11 +480,11 @@ mod tests {
             fast_field_writers.serialize(&mut serializer, &HashMap::new(), None)?;
             serializer.close()?;
         }
-        let file = directory.open_read(&path)?;
+        let file = directory.open_read(path)?;
         {
             let fast_fields_composite = CompositeFile::open(&file)?;
             let data = fast_fields_composite.open_read(*FIELD).unwrap();
-            let fast_field_reader = BitpackedFastFieldReader::<u64>::open(data)?;
+            let fast_field_reader = DynamicFastFieldReader::<u64>::open(data)?;
 
             let mut a = 0u64;
             for _ in 0..n {
@@ -627,7 +628,7 @@ mod bench {
         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);
             for &x in &permutation {
                 fast_field_writers.add_document(&doc!(*FIELD=>x));
@@ -641,7 +642,7 @@ mod bench {
         {
             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).unwrap();
+            let fast_field_reader = DynamicFastFieldReader::<u64>::open(data).unwrap();
 
             b.iter(|| {
                 let n = test::black_box(7000u32);
@@ -661,7 +662,7 @@ mod bench {
         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);
             for &x in &permutation {
                 fast_field_writers.add_document(&doc!(*FIELD=>x));
@@ -675,7 +676,7 @@ mod bench {
         {
             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).unwrap();
+            let fast_field_reader = DynamicFastFieldReader::<u64>::open(data).unwrap();
 
             b.iter(|| {
                 let n = test::black_box(1000u32);

src/fastfield/multivalued/mod.rs

@@ -90,7 +90,7 @@ mod tests {
         {
             let parser = QueryParser::for_index(&index, vec![date_field]);
             let query = parser
-                .parse_query(&format!("\"{}\"", first_time_stamp.to_rfc3339()).to_string())
+                .parse_query(&format!("\"{}\"", first_time_stamp.to_rfc3339()))
                 .expect("could not parse query");
             let results = searcher
                 .search(&query, &TopDocs::with_limit(5))
@@ -121,7 +121,7 @@ mod tests {
         {
             let parser = QueryParser::for_index(&index, vec![date_field]);
             let query = parser
-                .parse_query(&format!("\"{}\"", two_secs_ahead.to_rfc3339()).to_string())
+                .parse_query(&format!("\"{}\"", two_secs_ahead.to_rfc3339()))
                 .expect("could not parse query");
             let results = searcher
                 .search(&query, &TopDocs::with_limit(5))

src/fastfield/multivalued/reader.rs

@@ -1,6 +1,6 @@
 use std::ops::Range;
 
-use crate::fastfield::{BitpackedFastFieldReader, FastFieldReader, FastValue, MultiValueLength};
+use crate::fastfield::{DynamicFastFieldReader, 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: BitpackedFastFieldReader<u64>,
-    vals_reader: BitpackedFastFieldReader<Item>,
+    idx_reader: DynamicFastFieldReader<u64>,
+    vals_reader: DynamicFastFieldReader<Item>,
 }
 
 impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
     pub(crate) fn open(
-        idx_reader: BitpackedFastFieldReader<u64>,
-        vals_reader: BitpackedFastFieldReader<Item>,
+        idx_reader: DynamicFastFieldReader<u64>,
+        vals_reader: DynamicFastFieldReader<Item>,
     ) -> MultiValuedFastFieldReader<Item> {
         MultiValuedFastFieldReader {
             idx_reader,
@@ -30,6 +30,7 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
 
     /// Returns `(start, stop)`, such that the values associated
     /// to the given document are `start..stop`.
+    #[inline]
     fn range(&self, doc: DocId) -> Range<u64> {
         let start = self.idx_reader.get(doc);
         let stop = self.idx_reader.get(doc + 1);
@@ -37,20 +38,41 @@ impl<Item: FastValue> MultiValuedFastFieldReader<Item> {
     }
 
     /// Returns the array of values associated to the given `doc`.
+    #[inline]
     pub fn get_vals(&self, doc: DocId, vals: &mut Vec<Item>) {
         let range = self.range(doc);
         let len = (range.end - range.start) as usize;
         vals.resize(len, Item::make_zero());
-        self.vals_reader.get_range_u64(range.start, &mut vals[..]);
+        self.vals_reader.get_range(range.start, &mut vals[..]);
+    }
+
+    /// Returns the minimum value for this fast field.
+    ///
+    /// The min value does not take in account of possible
+    /// deleted document, and should be considered as a lower bound
+    /// of the actual mimimum value.
+    pub fn min_value(&self) -> Item {
+        self.vals_reader.min_value()
+    }
+
+    /// 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.
+    pub fn max_value(&self) -> Item {
+        self.vals_reader.max_value()
     }
 
     /// Returns the number of values associated with the document `DocId`.
+    #[inline]
     pub fn num_vals(&self, doc: DocId) -> usize {
         let range = self.range(doc);
         (range.end - range.start) as usize
     }
 
     /// Returns the overall number of values in this field  .
+    #[inline]
     pub fn total_num_vals(&self) -> u64 {
         self.idx_reader.max_value()
     }
@@ -69,7 +91,7 @@ impl<Item: FastValue> MultiValueLength for MultiValuedFastFieldReader<Item> {
 mod tests {
 
     use crate::core::Index;
-    use crate::schema::{Facet, Schema, INDEXED};
+    use crate::schema::{Cardinality, Facet, IntOptions, Schema, INDEXED};
 
     #[test]
     fn test_multifastfield_reader() {
@@ -124,4 +146,32 @@ mod tests {
             assert_eq!(&vals[..], &[4]);
         }
     }
+
+    #[test]
+    fn test_multifastfield_reader_min_max() {
+        let mut schema_builder = Schema::builder();
+        let field_options = IntOptions::default()
+            .set_indexed()
+            .set_fast(Cardinality::MultiValues);
+        let item_field = schema_builder.add_i64_field("items", field_options);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index
+            .writer_for_tests()
+            .expect("Failed to create index writer.");
+        index_writer.add_document(doc!(
+            item_field => 2i64,
+            item_field => 3i64,
+            item_field => -2i64,
+        ));
+        index_writer.add_document(doc!(item_field => 6i64, item_field => 3i64));
+        index_writer.add_document(doc!(item_field => 4i64));
+        index_writer.commit().expect("Commit failed");
+        let searcher = index.reader().unwrap().searcher();
+        let segment_reader = searcher.segment_reader(0);
+        let field_reader = segment_reader.fast_fields().i64s(item_field).unwrap();
+
+        assert_eq!(field_reader.min_value(), -2);
+        assert_eq!(field_reader.max_value(), 6);
+    }
 }

src/fastfield/multivalued/writer.rs

@@ -1,5 +1,4 @@
-use crate::fastfield::serializer::DynamicFastFieldSerializer;
-use crate::fastfield::serializer::FastFieldSerializer;
+use crate::fastfield::serializer::BitpackedFastFieldSerializerLegacy;
 use crate::fastfield::CompositeFastFieldSerializer;
 use crate::postings::UnorderedTermId;
 use crate::schema::{Document, Field};
@@ -103,11 +102,11 @@ impl MultiValuedFastFieldWriter {
         &'a self,
         doc_id_map: Option<&'b DocIdMapping>,
     ) -> impl Iterator<Item = &'b [u64]> {
-        let doc_id_iter = if let Some(doc_id_map) = doc_id_map {
-            Box::new(doc_id_map.iter_old_doc_ids().cloned()) as Box<dyn Iterator<Item = u32>>
+        let doc_id_iter: Box<dyn Iterator<Item = u32>> = if let Some(doc_id_map) = doc_id_map {
+            Box::new(doc_id_map.iter_old_doc_ids())
         } else {
-            Box::new(self.doc_index.iter().enumerate().map(|el| el.0 as u32))
-                as Box<dyn Iterator<Item = u32>>
+            let max_doc = self.doc_index.len() as DocId;
+            Box::new(0..max_doc)
         };
         doc_id_iter.map(move |doc_id| self.get_values_for_doc_id(doc_id))
     }
@@ -155,7 +154,7 @@ impl MultiValuedFastFieldWriter {
         }
         {
             // writing the values themselves.
-            let mut value_serializer: DynamicFastFieldSerializer<'_, _>;
+            let mut value_serializer: BitpackedFastFieldSerializerLegacy<'_, _>;
             match mapping_opt {
                 Some(mapping) => {
                     value_serializer = serializer.new_u64_fast_field_with_idx(

src/fastfield/reader.rs

@@ -8,11 +8,17 @@ use crate::fastfield::{CompositeFastFieldSerializer, FastFieldsWriter};
 use crate::schema::Schema;
 use crate::schema::FAST;
 use crate::DocId;
+use fastfield_codecs::bitpacked::BitpackedFastFieldReader as BitpackedReader;
+use fastfield_codecs::bitpacked::BitpackedFastFieldSerializer;
+use fastfield_codecs::linearinterpol::LinearInterpolFastFieldReader;
+use fastfield_codecs::linearinterpol::LinearInterpolFastFieldSerializer;
+use fastfield_codecs::multilinearinterpol::MultiLinearInterpolFastFieldReader;
+use fastfield_codecs::multilinearinterpol::MultiLinearInterpolFastFieldSerializer;
+use fastfield_codecs::FastFieldCodecReader;
+use fastfield_codecs::FastFieldCodecSerializer;
 use std::collections::HashMap;
 use std::marker::PhantomData;
 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 {
@@ -38,13 +44,13 @@ pub trait FastFieldReader<Item: FastValue>: Clone {
     ///
     /// May panic if `start + output.len()` is greater than
     /// the segment's `maxdoc`.
-    fn get_range(&self, start: DocId, output: &mut [Item]);
+    fn get_range(&self, start: u64, 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.
+    /// The min value does not take in account of possible
+    /// deleted document, and should be considered as a lower bound
+    /// of the actual mimimum value.
     fn min_value(&self) -> Item;
 
     /// Returns the maximum value for this fast field.
@@ -61,15 +67,48 @@ pub trait FastFieldReader<Item: FastValue>: Clone {
 ///
 pub enum DynamicFastFieldReader<Item: FastValue> {
     /// Bitpacked compressed fastfield data.
-    Bitpacked(BitpackedFastFieldReader<Item>),
+    Bitpacked(FastFieldReaderCodecWrapper<Item, BitpackedReader>),
+    /// Linear interpolated values + bitpacked
+    LinearInterpol(FastFieldReaderCodecWrapper<Item, LinearInterpolFastFieldReader>),
+    /// Blockwise linear interpolated values + bitpacked
+    MultiLinearInterpol(FastFieldReaderCodecWrapper<Item, MultiLinearInterpolFastFieldReader>),
 }
 
 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)?,
-        ))
+        let mut bytes = file.read_bytes()?;
+        let id = bytes.read_u8();
+
+        let reader = match id {
+            BitpackedFastFieldSerializer::ID => {
+                DynamicFastFieldReader::Bitpacked(FastFieldReaderCodecWrapper::<
+                    Item,
+                    BitpackedReader,
+                >::open_from_bytes(bytes)?)
+            }
+            LinearInterpolFastFieldSerializer::ID => {
+                DynamicFastFieldReader::LinearInterpol(FastFieldReaderCodecWrapper::<
+                    Item,
+                    LinearInterpolFastFieldReader,
+                >::open_from_bytes(bytes)?)
+            }
+            MultiLinearInterpolFastFieldSerializer::ID => {
+                DynamicFastFieldReader::MultiLinearInterpol(FastFieldReaderCodecWrapper::<
+                    Item,
+                    MultiLinearInterpolFastFieldReader,
+                >::open_from_bytes(
+                    bytes
+                )?)
+            }
+            _ => {
+                panic!(
+                    "unknown fastfield id {:?}. Data corrupted or using old tantivy version.",
+                    id
+                )
+            }
+        };
+        Ok(reader)
     }
 }
 
@@ -77,57 +116,67 @@ impl<Item: FastValue> FastFieldReader<Item> for DynamicFastFieldReader<Item> {
     fn get(&self, doc: DocId) -> Item {
         match self {
             Self::Bitpacked(reader) => reader.get(doc),
+            Self::LinearInterpol(reader) => reader.get(doc),
+            Self::MultiLinearInterpol(reader) => reader.get(doc),
         }
     }
-    fn get_range(&self, start: DocId, output: &mut [Item]) {
+    fn get_range(&self, start: u64, output: &mut [Item]) {
         match self {
             Self::Bitpacked(reader) => reader.get_range(start, output),
+            Self::LinearInterpol(reader) => reader.get_range(start, output),
+            Self::MultiLinearInterpol(reader) => reader.get_range(start, output),
         }
     }
     fn min_value(&self) -> Item {
         match self {
             Self::Bitpacked(reader) => reader.min_value(),
+            Self::LinearInterpol(reader) => reader.min_value(),
+            Self::MultiLinearInterpol(reader) => reader.min_value(),
         }
     }
     fn max_value(&self) -> Item {
         match self {
             Self::Bitpacked(reader) => reader.max_value(),
+            Self::LinearInterpol(reader) => reader.max_value(),
+            Self::MultiLinearInterpol(reader) => reader.max_value(),
         }
     }
 }
 
-/// Trait for accessing a fastfield.
+/// Wrapper for accessing a fastfield.
+///
+/// Holds the data and the codec to the read the data.
 ///
-/// Depending on the field type, a different
-/// fast field is required.
 #[derive(Clone)]
-pub struct BitpackedFastFieldReader<Item: FastValue> {
+pub struct FastFieldReaderCodecWrapper<Item: FastValue, CodecReader> {
+    reader: CodecReader,
     bytes: OwnedBytes,
-    bit_unpacker: BitUnpacker,
-    min_value_u64: u64,
-    max_value_u64: u64,
     _phantom: PhantomData<Item>,
 }
 
-impl<Item: FastValue> BitpackedFastFieldReader<Item> {
+impl<Item: FastValue, C: FastFieldCodecReader> FastFieldReaderCodecWrapper<Item, C> {
     /// Opens a fast field given a file.
     pub fn open(file: FileSlice) -> crate::Result<Self> {
         let mut bytes = file.read_bytes()?;
-        let min_value = u64::deserialize(&mut bytes)?;
-        let amplitude = u64::deserialize(&mut bytes)?;
-        let max_value = min_value + amplitude;
-        let num_bits = compute_num_bits(amplitude);
-        let bit_unpacker = BitUnpacker::new(num_bits);
-        Ok(BitpackedFastFieldReader {
+        let id = u8::deserialize(&mut bytes)?;
+        assert_eq!(
+            BitpackedFastFieldSerializer::ID,
+            id,
+            "Tried to open fast field as bitpacked encoded (id=1), but got serializer with different id"
+        );
+        Self::open_from_bytes(bytes)
+    }
+    /// Opens a fast field given the bytes.
+    pub fn open_from_bytes(bytes: OwnedBytes) -> crate::Result<Self> {
+        let reader = C::open_from_bytes(bytes.as_slice())?;
+        Ok(FastFieldReaderCodecWrapper {
+            reader,
             bytes,
-            min_value_u64: min_value,
-            max_value_u64: max_value,
-            bit_unpacker,
             _phantom: PhantomData,
         })
     }
     pub(crate) fn get_u64(&self, doc: u64) -> Item {
-        Item::from_u64(self.min_value_u64 + self.bit_unpacker.get(doc, &self.bytes))
+        Item::from_u64(self.reader.get_u64(doc, self.bytes.as_slice()))
     }
 
     /// Internally `multivalued` also use SingleValue Fast fields.
@@ -149,7 +198,9 @@ impl<Item: FastValue> BitpackedFastFieldReader<Item> {
     }
 }
 
-impl<Item: FastValue> FastFieldReader<Item> for BitpackedFastFieldReader<Item> {
+impl<Item: FastValue, C: FastFieldCodecReader + Clone> FastFieldReader<Item>
+    for FastFieldReaderCodecWrapper<Item, C>
+{
     /// Return the value associated to the given document.
     ///
     /// This accessor should return as fast as possible.
@@ -175,8 +226,8 @@ impl<Item: FastValue> FastFieldReader<Item> for BitpackedFastFieldReader<Item> {
     ///
     /// May panic if `start + output.len()` is greater than
     /// the segment's `maxdoc`.
-    fn get_range(&self, start: DocId, output: &mut [Item]) {
-        self.get_range_u64(u64::from(start), output);
+    fn get_range(&self, start: u64, output: &mut [Item]) {
+        self.get_range_u64(start, output);
     }
 
     /// Returns the minimum value for this fast field.
@@ -185,7 +236,7 @@ impl<Item: FastValue> FastFieldReader<Item> for BitpackedFastFieldReader<Item> {
     /// deleted document, and should be considered as an upper bound
     /// of the actual maximum value.
     fn min_value(&self) -> Item {
-        Item::from_u64(self.min_value_u64)
+        Item::from_u64(self.reader.min_value())
     }
 
     /// Returns the maximum value for this fast field.
@@ -194,12 +245,14 @@ impl<Item: FastValue> FastFieldReader<Item> for BitpackedFastFieldReader<Item> {
     /// deleted document, and should be considered as an upper bound
     /// of the actual maximum value.
     fn max_value(&self) -> Item {
-        Item::from_u64(self.max_value_u64)
+        Item::from_u64(self.reader.max_value())
     }
 }
 
-impl<Item: FastValue> From<Vec<Item>> for BitpackedFastFieldReader<Item> {
-    fn from(vals: Vec<Item>) -> BitpackedFastFieldReader<Item> {
+pub(crate) type BitpackedFastFieldReader<Item> = FastFieldReaderCodecWrapper<Item, BitpackedReader>;
+
+impl<Item: FastValue> From<Vec<Item>> for DynamicFastFieldReader<Item> {
+    fn from(vals: Vec<Item>) -> DynamicFastFieldReader<Item> {
         let mut schema_builder = Schema::builder();
         let field = schema_builder.add_u64_field("field", FAST);
         let schema = schema_builder.build();
@@ -231,6 +284,6 @@ impl<Item: FastValue> From<Vec<Item>> for BitpackedFastFieldReader<Item> {
         let field_file = composite_file
             .open_read(field)
             .expect("File component not found");
-        BitpackedFastFieldReader::open(field_file).unwrap()
+        DynamicFastFieldReader::open(field_file).unwrap()
     }
 }

src/fastfield/readers.rs

@@ -8,7 +8,6 @@ use crate::space_usage::PerFieldSpaceUsage;
 use crate::TantivyError;
 
 use super::reader::DynamicFastFieldReader;
-use super::FastFieldReader;
 
 /// Provides access to all of the BitpackedFastFieldReader.
 ///
@@ -100,23 +99,27 @@ impl FastFieldReaders {
         Ok(())
     }
 
+    pub(crate) fn typed_fast_field_reader_with_idx<TFastValue: FastValue>(
+        &self,
+        field: Field,
+        index: usize,
+    ) -> crate::Result<DynamicFastFieldReader<TFastValue>> {
+        let fast_field_slice = self.fast_field_data(field, index)?;
+        DynamicFastFieldReader::open(fast_field_slice)
+    }
     pub(crate) fn typed_fast_field_reader<TFastValue: FastValue>(
         &self,
         field: Field,
     ) -> crate::Result<DynamicFastFieldReader<TFastValue>> {
-        let fast_field_slice = self.fast_field_data(field, 0)?;
-        DynamicFastFieldReader::open(fast_field_slice)
+        self.typed_fast_field_reader_with_idx(field, 0)
     }
 
     pub(crate) fn typed_fast_field_multi_reader<TFastValue: FastValue>(
         &self,
         field: Field,
     ) -> 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 = BitpackedFastFieldReader::open(fast_field_slice_idx)?;
-        let vals_reader: BitpackedFastFieldReader<TFastValue> =
-            BitpackedFastFieldReader::open(fast_field_slice_vals)?;
+        let idx_reader = self.typed_fast_field_reader(field)?;
+        let vals_reader = self.typed_fast_field_reader_with_idx(field, 1)?;
         Ok(MultiValuedFastFieldReader::open(idx_reader, vals_reader))
     }
 
@@ -139,7 +142,7 @@ 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 i64(&self, field: Field) -> crate::Result<impl FastFieldReader<i64>> {
+    pub fn i64(&self, field: Field) -> crate::Result<DynamicFastFieldReader<i64>> {
         self.check_type(field, FastType::I64, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }
@@ -147,7 +150,7 @@ 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<impl FastFieldReader<crate::DateTime>> {
+    pub fn date(&self, field: Field) -> crate::Result<DynamicFastFieldReader<crate::DateTime>> {
         self.check_type(field, FastType::Date, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }
@@ -155,7 +158,7 @@ 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<impl FastFieldReader<f64>> {
+    pub fn f64(&self, field: Field) -> crate::Result<DynamicFastFieldReader<f64>> {
         self.check_type(field, FastType::F64, Cardinality::SingleValue)?;
         self.typed_fast_field_reader(field)
     }

src/fastfield/serializer.rs

@@ -1,256 +0,0 @@
-use crate::common::BinarySerializable;
-use crate::common::CompositeWrite;
-use crate::common::CountingWriter;
-use crate::directory::WritePtr;
-use crate::schema::Field;
-use std::io::{self, Write};
-use tantivy_bitpacker::compute_num_bits;
-use tantivy_bitpacker::BitPacker;
-
-/// `CompositeFastFieldSerializer` is in charge of serializing
-/// fastfields on disk.
-///
-/// Fast fields have differnt encodings like bit-packing.
-///
-/// `FastFieldWriter`s are in charge of pushing the data to
-/// the serializer.
-/// The serializer expects to receive the following calls.
-///
-/// * `new_u64_fast_field(...)`
-/// * `add_val(...)`
-/// * `add_val(...)`
-/// * `add_val(...)`
-/// * ...
-/// * `close_field()`
-/// * `new_u64_fast_field(...)`
-/// * `add_val(...)`
-/// * ...
-/// * `close_field()`
-/// * `close()`
-pub struct CompositeFastFieldSerializer {
-    composite_write: CompositeWrite<WritePtr>,
-}
-
-impl CompositeFastFieldSerializer {
-    /// Constructor
-    pub fn from_write(write: WritePtr) -> io::Result<CompositeFastFieldSerializer> {
-        // just making room for the pointer to header.
-        let composite_write = CompositeWrite::wrap(write);
-        Ok(CompositeFastFieldSerializer { composite_write })
-    }
-
-    /// Start serializing a new u64 fast field
-    pub fn new_u64_fast_field(
-        &mut self,
-        field: Field,
-        min_value: u64,
-        max_value: u64,
-    ) -> io::Result<DynamicFastFieldSerializer<'_, CountingWriter<WritePtr>>> {
-        self.new_u64_fast_field_with_idx(field, min_value, max_value, 0)
-    }
-
-    /// Start serializing a new u64 fast field
-    pub fn new_u64_fast_field_with_idx(
-        &mut self,
-        field: Field,
-        min_value: u64,
-        max_value: u64,
-        idx: usize,
-    ) -> io::Result<DynamicFastFieldSerializer<'_, CountingWriter<WritePtr>>> {
-        let field_write = self.composite_write.for_field_with_idx(field, idx);
-        DynamicFastFieldSerializer::open(field_write, min_value, max_value)
-    }
-
-    /// Start serializing a new [u8] fast field
-    pub fn new_bytes_fast_field_with_idx(
-        &mut self,
-        field: Field,
-        idx: usize,
-    ) -> FastBytesFieldSerializer<'_, CountingWriter<WritePtr>> {
-        let field_write = self.composite_write.for_field_with_idx(field, idx);
-        FastBytesFieldSerializer { write: field_write }
-    }
-
-    /// Closes the serializer
-    ///
-    /// After this call the data must be persistently save on disk.
-    pub fn close(self) -> io::Result<()> {
-        self.composite_write.close()
-    }
-}
-
-#[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> BitpackedFastFieldSerializer<'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.
-    fn open(
-        write: &'a mut W,
-        min_value: u64,
-        max_value: u64,
-    ) -> 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(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.
-    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(())
-    }
-    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,
-}
-
-impl<'a, W: Write> FastBytesFieldSerializer<'a, W> {
-    pub fn write_all(&mut self, vals: &[u8]) -> io::Result<()> {
-        self.write.write_all(vals)
-    }
-
-    pub fn flush(&mut self) -> io::Result<()> {
-        self.write.flush()
-    }
-}

src/fastfield/serializer/mod.rs

@@ -0,0 +1,218 @@
+use crate::common::BinarySerializable;
+use crate::common::CompositeWrite;
+use crate::common::CountingWriter;
+use crate::directory::WritePtr;
+use crate::schema::Field;
+pub use fastfield_codecs::bitpacked::BitpackedFastFieldSerializer;
+pub use fastfield_codecs::bitpacked::BitpackedFastFieldSerializerLegacy;
+use fastfield_codecs::linearinterpol::LinearInterpolFastFieldSerializer;
+use fastfield_codecs::multilinearinterpol::MultiLinearInterpolFastFieldSerializer;
+pub use fastfield_codecs::FastFieldCodecSerializer;
+pub use fastfield_codecs::FastFieldDataAccess;
+pub use fastfield_codecs::FastFieldStats;
+use std::io::{self, Write};
+
+/// `CompositeFastFieldSerializer` is in charge of serializing
+/// fastfields on disk.
+///
+/// Fast fields have different encodings like bit-packing.
+///
+/// `FastFieldWriter`s are in charge of pushing the data to
+/// the serializer.
+/// The serializer expects to receive the following calls.
+///
+/// * `new_u64_fast_field(...)`
+/// * `add_val(...)`
+/// * `add_val(...)`
+/// * `add_val(...)`
+/// * ...
+/// * `close_field()`
+/// * `new_u64_fast_field(...)`
+/// * `add_val(...)`
+/// * ...
+/// * `close_field()`
+/// * `close()`
+pub struct CompositeFastFieldSerializer {
+    composite_write: CompositeWrite<WritePtr>,
+}
+
+// use this, when this is merged and stabilized explicit_generic_args_with_impl_trait
+// https://github.com/rust-lang/rust/pull/86176
+fn codec_estimation<T: FastFieldCodecSerializer, A: FastFieldDataAccess>(
+    stats: FastFieldStats,
+    fastfield_accessor: &A,
+    estimations: &mut Vec<(f32, &str, u8)>,
+) {
+    if !T::is_applicable(fastfield_accessor, stats.clone()) {
+        return;
+    }
+    let (ratio, name, id) = (T::estimate(fastfield_accessor, stats), T::NAME, T::ID);
+    estimations.push((ratio, name, id));
+}
+
+impl CompositeFastFieldSerializer {
+    /// Constructor
+    pub fn from_write(write: WritePtr) -> io::Result<CompositeFastFieldSerializer> {
+        // just making room for the pointer to header.
+        let composite_write = CompositeWrite::wrap(write);
+        Ok(CompositeFastFieldSerializer { composite_write })
+    }
+
+    /// Serialize data into a new u64 fast field. The best compression codec will be chosen automatically.
+    pub fn create_auto_detect_u64_fast_field(
+        &mut self,
+        field: Field,
+        stats: FastFieldStats,
+        fastfield_accessor: impl FastFieldDataAccess,
+        data_iter_1: impl Iterator<Item = u64>,
+        data_iter_2: impl Iterator<Item = u64>,
+    ) -> io::Result<()> {
+        self.create_auto_detect_u64_fast_field_with_idx(
+            field,
+            stats,
+            fastfield_accessor,
+            data_iter_1,
+            data_iter_2,
+            0,
+        )
+    }
+    /// Serialize data into a new u64 fast field. The best compression codec will be chosen automatically.
+    pub fn create_auto_detect_u64_fast_field_with_idx(
+        &mut self,
+        field: Field,
+        stats: FastFieldStats,
+        fastfield_accessor: impl FastFieldDataAccess,
+        data_iter_1: impl Iterator<Item = u64>,
+        data_iter_2: impl Iterator<Item = u64>,
+        idx: usize,
+    ) -> io::Result<()> {
+        let field_write = self.composite_write.for_field_with_idx(field, idx);
+
+        let mut estimations = vec![];
+
+        codec_estimation::<BitpackedFastFieldSerializer, _>(
+            stats.clone(),
+            &fastfield_accessor,
+            &mut estimations,
+        );
+        codec_estimation::<LinearInterpolFastFieldSerializer, _>(
+            stats.clone(),
+            &fastfield_accessor,
+            &mut estimations,
+        );
+        codec_estimation::<MultiLinearInterpolFastFieldSerializer, _>(
+            stats.clone(),
+            &fastfield_accessor,
+            &mut estimations,
+        );
+        if let Some(broken_estimation) = estimations
+            .iter()
+            .find(|estimation| estimation.0 == f32::NAN)
+        {
+            warn!(
+                "broken estimation for fast field codec {}",
+                broken_estimation.1
+            );
+        }
+        // removing nan values for codecs with broken calculations, and max values which disables codecs
+        estimations.retain(|estimation| !estimation.0.is_nan() && estimation.0 != f32::MAX);
+        estimations.sort_by(|a, b| a.0.partial_cmp(&b.0).unwrap());
+        let (_ratio, name, id) = estimations[0];
+        debug!(
+            "choosing fast field codec {} for field_id {:?}",
+            name, field
+        ); // todo print actual field name
+        id.serialize(field_write)?;
+        match name {
+            BitpackedFastFieldSerializer::NAME => {
+                BitpackedFastFieldSerializer::serialize(
+                    field_write,
+                    &fastfield_accessor,
+                    stats,
+                    data_iter_1,
+                    data_iter_2,
+                )?;
+            }
+            LinearInterpolFastFieldSerializer::NAME => {
+                LinearInterpolFastFieldSerializer::serialize(
+                    field_write,
+                    &fastfield_accessor,
+                    stats,
+                    data_iter_1,
+                    data_iter_2,
+                )?;
+            }
+            MultiLinearInterpolFastFieldSerializer::NAME => {
+                MultiLinearInterpolFastFieldSerializer::serialize(
+                    field_write,
+                    &fastfield_accessor,
+                    stats,
+                    data_iter_1,
+                    data_iter_2,
+                )?;
+            }
+            _ => {
+                panic!("unknown fastfield serializer {}", name)
+            }
+        };
+        field_write.flush()?;
+
+        Ok(())
+    }
+
+    /// Start serializing a new u64 fast field
+    pub fn new_u64_fast_field(
+        &mut self,
+        field: Field,
+        min_value: u64,
+        max_value: u64,
+    ) -> io::Result<BitpackedFastFieldSerializerLegacy<'_, CountingWriter<WritePtr>>> {
+        self.new_u64_fast_field_with_idx(field, min_value, max_value, 0)
+    }
+
+    /// Start serializing a new u64 fast field
+    pub fn new_u64_fast_field_with_idx(
+        &mut self,
+        field: Field,
+        min_value: u64,
+        max_value: u64,
+        idx: usize,
+    ) -> io::Result<BitpackedFastFieldSerializerLegacy<'_, CountingWriter<WritePtr>>> {
+        let field_write = self.composite_write.for_field_with_idx(field, idx);
+        // Prepend codec id to field data for compatibility with DynamicFastFieldReader.
+        let id = BitpackedFastFieldSerializer::ID;
+        id.serialize(field_write)?;
+        BitpackedFastFieldSerializerLegacy::open(field_write, min_value, max_value)
+    }
+
+    /// Start serializing a new [u8] fast field
+    pub fn new_bytes_fast_field_with_idx(
+        &mut self,
+        field: Field,
+        idx: usize,
+    ) -> FastBytesFieldSerializer<'_, CountingWriter<WritePtr>> {
+        let field_write = self.composite_write.for_field_with_idx(field, idx);
+        FastBytesFieldSerializer { write: field_write }
+    }
+
+    /// Closes the serializer
+    ///
+    /// After this call the data must be persistently save on disk.
+    pub fn close(self) -> io::Result<()> {
+        self.composite_write.close()
+    }
+}
+
+pub struct FastBytesFieldSerializer<'a, W: Write> {
+    write: &'a mut W,
+}
+
+impl<'a, W: Write> FastBytesFieldSerializer<'a, W> {
+    pub fn write_all(&mut self, vals: &[u8]) -> io::Result<()> {
+        self.write.write_all(vals)
+    }
+
+    pub fn flush(&mut self) -> io::Result<()> {
+        self.write.flush()
+    }
+}

src/fastfield/writer.rs

@@ -1,6 +1,7 @@
 use super::multivalued::MultiValuedFastFieldWriter;
+use super::serializer::FastFieldStats;
+use super::FastFieldDataAccess;
 use crate::common;
-use crate::fastfield::serializer::FastFieldSerializer;
 use crate::fastfield::{BytesFastFieldWriter, CompositeFastFieldSerializer};
 use crate::indexer::doc_id_mapping::DocIdMapping;
 use crate::postings::UnorderedTermId;
@@ -265,9 +266,9 @@ impl IntFastFieldWriter {
         self.add_val(val);
     }
 
-    /// Extract the stored data
-    pub(crate) fn get_data(&self) -> Vec<u64> {
-        self.vals.iter().collect::<Vec<u64>>()
+    /// get iterator over the data
+    pub(crate) fn iter(&self) -> impl Iterator<Item = u64> + '_ {
+        self.vals.iter()
     }
 
     /// Push the fast fields value to the `FastFieldWriter`.
@@ -281,17 +282,59 @@ impl IntFastFieldWriter {
         } else {
             (self.val_min, self.val_max)
         };
-        let mut single_field_serializer = serializer.new_u64_fast_field(self.field, min, max)?;
-        if let Some(doc_id_map) = doc_id_map {
-            for doc_id in doc_id_map.iter_old_doc_ids() {
-                single_field_serializer.add_val(self.vals.get(*doc_id as usize))?;
-            }
-        } else {
-            for val in self.vals.iter() {
-                single_field_serializer.add_val(val)?;
-            }
+        let fastfield_accessor = WriterFastFieldAccessProvider {
+            doc_id_map,
+            vals: &self.vals,
+        };
+        let stats = FastFieldStats {
+            min_value: min,
+            max_value: max,
+            num_vals: self.val_count as u64,
         };
 
-        single_field_serializer.close_field()
+        if let Some(doc_id_map) = doc_id_map {
+            let iter = doc_id_map
+                .iter_old_doc_ids()
+                .map(|doc_id| self.vals.get(doc_id as usize));
+            serializer.create_auto_detect_u64_fast_field(
+                self.field,
+                stats,
+                fastfield_accessor,
+                iter.clone(),
+                iter,
+            )?;
+        } else {
+            serializer.create_auto_detect_u64_fast_field(
+                self.field,
+                stats,
+                fastfield_accessor,
+                self.vals.iter(),
+                self.vals.iter(),
+            )?;
+        };
+        Ok(())
+    }
+}
+
+#[derive(Clone)]
+struct WriterFastFieldAccessProvider<'map, 'bitp> {
+    doc_id_map: Option<&'map DocIdMapping>,
+    vals: &'bitp BlockedBitpacker,
+}
+impl<'map, 'bitp> FastFieldDataAccess for WriterFastFieldAccessProvider<'map, 'bitp> {
+    /// Return the value associated to the given doc.
+    ///
+    /// Whenever possible use the Iterator passed to the fastfield creation instead, for performance reasons.
+    ///
+    /// # Panics
+    ///
+    /// May panic if `doc` is greater than the index.
+    fn get_val(&self, doc: u64) -> u64 {
+        if let Some(doc_id_map) = self.doc_id_map {
+            self.vals
+                .get(doc_id_map.get_old_doc_id(doc as u32) as usize) // consider extra FastFieldReader wrapper for non doc_id_map
+        } else {
+            self.vals.get(doc as usize)
+        }
     }
 }

src/fieldnorm/writer.rs

@@ -98,7 +98,7 @@ impl FieldNormsWriter {
                 let mut mapped_fieldnorm_values = vec![];
                 mapped_fieldnorm_values.resize(fieldnorm_values.len(), 0u8);
                 for (new_doc_id, old_doc_id) in doc_id_map.iter_old_doc_ids().enumerate() {
-                    mapped_fieldnorm_values[new_doc_id] = fieldnorm_values[*old_doc_id as usize];
+                    mapped_fieldnorm_values[new_doc_id] = fieldnorm_values[old_doc_id as usize];
                 }
                 fieldnorms_serializer.serialize_field(field, &mapped_fieldnorm_values)?;
             } else {

src/functional_test.rs

@@ -38,7 +38,7 @@ fn test_functional_store() -> crate::Result<()> {
     for iteration in 0..500 {
         dbg!(iteration);
         let num_docs: usize = rng.gen_range(0..4);
-        if doc_set.len() >= 1 {
+        if !doc_set.is_empty() {
             let doc_to_remove_id = rng.gen_range(0..doc_set.len());
             let removed_doc_id = doc_set.swap_remove(doc_to_remove_id);
             index_writer.delete_term(Term::from_field_u64(id_field, removed_doc_id));
@@ -88,19 +88,17 @@ fn test_functional_indexing() -> crate::Result<()> {
                 &searcher,
                 &committed_docs.iter().cloned().collect::<Vec<u64>>(),
             )?;
+        } else if committed_docs.remove(&random_val) || uncommitted_docs.remove(&random_val) {
+            let doc_id_term = Term::from_field_u64(id_field, random_val);
+            index_writer.delete_term(doc_id_term);
         } else {
-            if committed_docs.remove(&random_val) || uncommitted_docs.remove(&random_val) {
-                let doc_id_term = Term::from_field_u64(id_field, random_val);
-                index_writer.delete_term(doc_id_term);
-            } else {
-                uncommitted_docs.insert(random_val);
-                let mut doc = Document::new();
-                doc.add_u64(id_field, random_val);
-                for i in 1u64..10u64 {
-                    doc.add_u64(multiples_field, random_val * i);
-                }
-                index_writer.add_document(doc);
+            uncommitted_docs.insert(random_val);
+            let mut doc = Document::new();
+            doc.add_u64(id_field, random_val);
+            for i in 1u64..10u64 {
+                doc.add_u64(multiples_field, random_val * i);
             }
+            index_writer.add_document(doc);
         }
     }
     Ok(())

src/indexer/delete_queue.rs

@@ -1,6 +1,6 @@
 use super::operation::DeleteOperation;
 use crate::Opstamp;
-use std::mem;
+
 use std::ops::DerefMut;
 use std::sync::{Arc, RwLock, Weak};
 
@@ -105,7 +105,7 @@ impl DeleteQueue {
             return None;
         }
 
-        let delete_operations = mem::replace(&mut self_wlock.writer, vec![]);
+        let delete_operations = std::mem::take(&mut self_wlock.writer);
 
         let new_block = Arc::new(Block {
             operations: Arc::from(delete_operations.into_boxed_slice()),
@@ -286,7 +286,7 @@ mod tests {
             operations_it.advance();
         }
         {
-            let mut operations_it = snapshot.clone();
+            let mut operations_it = snapshot;
             assert_eq!(operations_it.get().unwrap().opstamp, 1);
             operations_it.advance();
             assert_eq!(operations_it.get().unwrap().opstamp, 2);

src/indexer/doc_id_mapping.rs

@@ -2,13 +2,61 @@
 //! to get mappings from old doc_id to new doc_id and vice versa, after sorting
 //!
 
-use super::SegmentWriter;
+use super::{merger::SegmentReaderWithOrdinal, SegmentWriter};
 use crate::{
     schema::{Field, Schema},
     DocId, IndexSortByField, Order, TantivyError,
 };
-use std::cmp::Reverse;
-/// Struct to provide mapping from old doc_id to new doc_id and vice versa
+use std::{cmp::Reverse, ops::Index};
+
+/// Struct to provide mapping from new doc_id to old doc_id and segment.
+#[derive(Clone)]
+pub(crate) struct SegmentDocidMapping<'a> {
+    new_doc_id_to_old_and_segment: Vec<(DocId, SegmentReaderWithOrdinal<'a>)>,
+    is_trivial: bool,
+}
+
+impl<'a> SegmentDocidMapping<'a> {
+    pub(crate) fn new(
+        new_doc_id_to_old_and_segment: Vec<(DocId, SegmentReaderWithOrdinal<'a>)>,
+        is_trivial: bool,
+    ) -> Self {
+        Self {
+            new_doc_id_to_old_and_segment,
+            is_trivial,
+        }
+    }
+    pub(crate) fn iter(&self) -> impl Iterator<Item = &(DocId, SegmentReaderWithOrdinal)> {
+        self.new_doc_id_to_old_and_segment.iter()
+    }
+    pub(crate) fn len(&self) -> usize {
+        self.new_doc_id_to_old_and_segment.len()
+    }
+    /// This flags means the segments are simply stacked in the order of their ordinal.
+    /// e.g. [(0, 1), .. (n, 1), (0, 2)..., (m, 2)]
+    ///
+    /// This allows for some optimization.
+    pub(crate) fn is_trivial(&self) -> bool {
+        self.is_trivial
+    }
+}
+impl<'a> Index<usize> for SegmentDocidMapping<'a> {
+    type Output = (DocId, SegmentReaderWithOrdinal<'a>);
+
+    fn index(&self, idx: usize) -> &Self::Output {
+        &self.new_doc_id_to_old_and_segment[idx]
+    }
+}
+impl<'a> IntoIterator for SegmentDocidMapping<'a> {
+    type Item = (DocId, SegmentReaderWithOrdinal<'a>);
+    type IntoIter = std::vec::IntoIter<Self::Item>;
+
+    fn into_iter(self) -> Self::IntoIter {
+        self.new_doc_id_to_old_and_segment.into_iter()
+    }
+}
+
+/// Struct to provide mapping from old doc_id to new doc_id and vice versa within a segment.
 pub struct DocIdMapping {
     new_doc_id_to_old: Vec<DocId>,
     old_doc_id_to_new: Vec<DocId>,
@@ -24,8 +72,8 @@ impl DocIdMapping {
         self.new_doc_id_to_old[doc_id as usize]
     }
     /// iterate over old doc_ids in order of the new doc_ids
-    pub fn iter_old_doc_ids(&self) -> std::slice::Iter<'_, DocId> {
-        self.new_doc_id_to_old.iter()
+    pub fn iter_old_doc_ids(&self) -> impl Iterator<Item = DocId> + Clone + '_ {
+        self.new_doc_id_to_old.iter().cloned()
     }
 }
 
@@ -60,9 +108,8 @@ pub(crate) fn get_doc_id_mapping_from_field(
         })?;
 
     // create new doc_id to old doc_id index (used in fast_field_writers)
-    let data = fast_field.get_data();
-    let mut doc_id_and_data = data
-        .into_iter()
+    let mut doc_id_and_data = fast_field
+        .iter()
         .enumerate()
         .map(|el| (el.0 as DocId, el.1))
         .collect::<Vec<_>>();

src/indexer/doc_opstamp_mapping.rs

@@ -14,35 +14,27 @@ use crate::Opstamp;
 // The doc to opstamp mapping stores precisely an array
 // indexed by doc id and storing the opstamp of the document.
 //
-// This mapping is (for the moment) stricly increasing
-// because of the way document id are allocated.
+// This mapping is NOT necessarily increasing, because
+// we might be sorting documents according to a fast field.
 #[derive(Clone)]
 pub enum DocToOpstampMapping<'a> {
     WithMap(&'a [Opstamp]),
     None,
 }
 
-impl<'a> From<&'a [u64]> for DocToOpstampMapping<'a> {
-    fn from(opstamps: &[Opstamp]) -> DocToOpstampMapping {
-        DocToOpstampMapping::WithMap(opstamps)
-    }
-}
-
 impl<'a> DocToOpstampMapping<'a> {
-    /// Given an opstamp return the limit doc id L
-    /// such that all doc id D such that
-    // D >= L iff opstamp(D) >= than `target_opstamp`.
-    //
-    // The edge case opstamp = some doc opstamp is in practise
-    // never called.
-    pub fn compute_doc_limit(&self, target_opstamp: Opstamp) -> DocId {
-        match *self {
-            DocToOpstampMapping::WithMap(ref doc_opstamps) => {
-                match doc_opstamps.binary_search(&target_opstamp) {
-                    Ok(doc_id) | Err(doc_id) => doc_id as DocId,
-                }
+    /// Assess whether a document should be considered deleted given that it contains
+    /// a deleted term that was deleted at the opstamp: `delete_opstamp`.
+    ///
+    /// This function returns true if the `DocToOpstamp` mapping is none or if
+    /// the `doc_opstamp` is anterior to the delete opstamp.
+    pub fn is_deleted(&self, doc_id: DocId, delete_opstamp: Opstamp) -> bool {
+        match self {
+            Self::WithMap(doc_opstamps) => {
+                let doc_opstamp = doc_opstamps[doc_id as usize];
+                doc_opstamp < delete_opstamp
             }
-            DocToOpstampMapping::None => DocId::max_value(),
+            Self::None => true,
         }
     }
 }
@@ -55,40 +47,17 @@ mod tests {
     #[test]
     fn test_doc_to_opstamp_mapping_none() {
         let doc_to_opstamp_mapping = DocToOpstampMapping::None;
-        assert_eq!(
-            doc_to_opstamp_mapping.compute_doc_limit(1),
-            u32::max_value()
-        );
+        assert!(doc_to_opstamp_mapping.is_deleted(1u32, 0u64));
+        assert!(doc_to_opstamp_mapping.is_deleted(1u32, 2u64));
     }
 
     #[test]
-    fn test_doc_to_opstamp_mapping_complex() {
-        {
-            let doc_to_opstamp_mapping = DocToOpstampMapping::from(&[][..]);
-            assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(0u64), 0);
-            assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(2u64), 0);
-        }
-        {
-            let doc_to_opstamp_mapping = DocToOpstampMapping::from(&[1u64][..]);
-            assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(0u64), 0);
-            assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(2u64), 1);
-        }
-        {
-            let doc_to_opstamp_mapping =
-                DocToOpstampMapping::from(&[1u64, 12u64, 17u64, 23u64][..]);
-            assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(0u64), 0);
-            for i in 2u64..13u64 {
-                assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(i), 1);
-            }
-            for i in 13u64..18u64 {
-                assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(i), 2);
-            }
-            for i in 18u64..24u64 {
-                assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(i), 3);
-            }
-            for i in 24u64..30u64 {
-                assert_eq!(doc_to_opstamp_mapping.compute_doc_limit(i), 4);
-            }
-        }
+    fn test_doc_to_opstamp_mapping_with_map() {
+        let doc_to_opstamp_mapping = DocToOpstampMapping::WithMap(&[5u64, 1u64, 0u64, 4u64, 3u64]);
+        assert_eq!(doc_to_opstamp_mapping.is_deleted(0u32, 2u64), false);
+        assert_eq!(doc_to_opstamp_mapping.is_deleted(1u32, 2u64), true);
+        assert_eq!(doc_to_opstamp_mapping.is_deleted(2u32, 2u64), true);
+        assert_eq!(doc_to_opstamp_mapping.is_deleted(3u32, 2u64), false);
+        assert_eq!(doc_to_opstamp_mapping.is_deleted(4u32, 2u64), false);
     }
 }

src/indexer/index_writer.rs

@@ -106,22 +106,18 @@ fn compute_deleted_bitset(
         }
 
         // A delete operation should only affect
-        // document that were inserted after it.
-        //
-        // Limit doc helps identify the first document
-        // that may be affected by the delete operation.
-        let limit_doc = doc_opstamps.compute_doc_limit(delete_op.opstamp);
+        // document that were inserted before it.
         let inverted_index = segment_reader.inverted_index(delete_op.term.field())?;
         if let Some(mut docset) =
             inverted_index.read_postings(&delete_op.term, IndexRecordOption::Basic)?
         {
-            let mut deleted_doc = docset.doc();
-            while deleted_doc != TERMINATED {
-                if deleted_doc < limit_doc {
-                    delete_bitset.insert(deleted_doc);
+            let mut doc_matching_deleted_term = docset.doc();
+            while doc_matching_deleted_term != TERMINATED {
+                if doc_opstamps.is_deleted(doc_matching_deleted_term, delete_op.opstamp) {
+                    delete_bitset.insert(doc_matching_deleted_term);
                     might_have_changed = true;
                 }
-                deleted_doc = docset.advance();
+                doc_matching_deleted_term = docset.advance();
             }
         }
         delete_cursor.advance();
@@ -230,14 +226,8 @@ fn index_documents(
 
     let segment_with_max_doc = segment.with_max_doc(max_doc);
 
-    let last_docstamp: Opstamp = *(doc_opstamps.last().unwrap());
-
-    let delete_bitset_opt = apply_deletes(
-        &segment_with_max_doc,
-        &mut delete_cursor,
-        &doc_opstamps,
-        last_docstamp,
-    )?;
+    let delete_bitset_opt =
+        apply_deletes(&segment_with_max_doc, &mut delete_cursor, &doc_opstamps)?;
 
     let meta = segment_with_max_doc.meta().clone();
     meta.untrack_temp_docstore();
@@ -247,19 +237,26 @@ fn index_documents(
     Ok(true)
 }
 
+/// `doc_opstamps` is required to be non-empty.
 fn apply_deletes(
     segment: &Segment,
     mut delete_cursor: &mut DeleteCursor,
     doc_opstamps: &[Opstamp],
-    last_docstamp: Opstamp,
 ) -> crate::Result<Option<BitSet>> {
     if delete_cursor.get().is_none() {
         // if there are no delete operation in the queue, no need
         // to even open the segment.
         return Ok(None);
     }
+
+    let max_doc_opstamp: Opstamp = doc_opstamps
+        .iter()
+        .cloned()
+        .max()
+        .expect("Empty DocOpstamp is forbidden");
+
     let segment_reader = SegmentReader::open(segment)?;
-    let doc_to_opstamps = DocToOpstampMapping::from(doc_opstamps);
+    let doc_to_opstamps = DocToOpstampMapping::WithMap(doc_opstamps);
 
     let max_doc = segment.meta().max_doc();
     let mut deleted_bitset = BitSet::with_max_value(max_doc);
@@ -268,7 +265,7 @@ fn apply_deletes(
         &segment_reader,
         &mut delete_cursor,
         &doc_to_opstamps,
-        last_docstamp,
+        max_doc_opstamp,
     )?;
     Ok(if may_have_deletes {
         Some(deleted_bitset)
@@ -358,7 +355,7 @@ impl IndexWriter {
         // dropping the last reference to the segment_updater.
         self.drop_sender();
 
-        let former_workers_handles = mem::replace(&mut self.workers_join_handle, vec![]);
+        let former_workers_handles = std::mem::take(&mut self.workers_join_handle);
         for join_handle in former_workers_handles {
             join_handle
                 .join()
@@ -628,7 +625,7 @@ impl IndexWriter {
         // and recreate a new one.
         self.recreate_document_channel();
 
-        let former_workers_join_handle = mem::replace(&mut self.workers_join_handle, Vec::new());
+        let former_workers_join_handle = std::mem::take(&mut self.workers_join_handle);
 
         for worker_handle in former_workers_join_handle {
             let indexing_worker_result = worker_handle
@@ -784,17 +781,44 @@ impl Drop for IndexWriter {
 
 #[cfg(test)]
 mod tests {
+    use std::collections::HashMap;
+    use std::collections::HashSet;
+
+    use futures::executor::block_on;
+    use proptest::prelude::*;
+    use proptest::prop_oneof;
+    use proptest::strategy::Strategy;
 
     use super::super::operation::UserOperation;
     use crate::collector::TopDocs;
     use crate::directory::error::LockError;
     use crate::error::*;
+    use crate::fastfield::FastFieldReader;
     use crate::indexer::NoMergePolicy;
+    use crate::query::QueryParser;
     use crate::query::TermQuery;
-    use crate::schema::{self, IndexRecordOption, STRING};
+    use crate::schema::Cardinality;
+    use crate::schema::Facet;
+    use crate::schema::IntOptions;
+    use crate::schema::TextFieldIndexing;
+    use crate::schema::TextOptions;
+    use crate::schema::STORED;
+    use crate::schema::TEXT;
+    use crate::schema::{self, IndexRecordOption, FAST, INDEXED, STRING};
+    use crate::DocAddress;
     use crate::Index;
     use crate::ReloadPolicy;
     use crate::Term;
+    use crate::{IndexSettings, IndexSortByField, Order};
+
+    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.";
 
     #[test]
     fn test_operations_group() {
@@ -1282,6 +1306,330 @@ mod tests {
         assert!(commit_again.is_ok());
     }
 
+    #[test]
+    fn test_delete_with_sort_by_field() -> crate::Result<()> {
+        let mut schema_builder = schema::Schema::builder();
+        let id_field =
+            schema_builder.add_u64_field("id", schema::INDEXED | schema::STORED | schema::FAST);
+        let schema = schema_builder.build();
+
+        let settings = IndexSettings {
+            sort_by_field: Some(IndexSortByField {
+                field: "id".to_string(),
+                order: Order::Desc,
+            }),
+            ..Default::default()
+        };
+
+        let index = Index::builder()
+            .schema(schema)
+            .settings(settings)
+            .create_in_ram()?;
+        let index_reader = index.reader()?;
+        let mut index_writer = index.writer_for_tests()?;
+
+        // create and delete docs in same commit
+        for id in 0u64..5u64 {
+            index_writer.add_document(doc!(id_field => id));
+        }
+        for id in 2u64..4u64 {
+            index_writer.delete_term(Term::from_field_u64(id_field, id));
+        }
+        for id in 5u64..10u64 {
+            index_writer.add_document(doc!(id_field => id));
+        }
+        index_writer.commit()?;
+        index_reader.reload()?;
+
+        let searcher = index_reader.searcher();
+        assert_eq!(searcher.segment_readers().len(), 1);
+
+        let segment_reader = searcher.segment_reader(0);
+        assert_eq!(segment_reader.num_docs(), 8);
+        assert_eq!(segment_reader.max_doc(), 10);
+        let fast_field_reader = segment_reader.fast_fields().u64(id_field)?;
+        let in_order_alive_ids: Vec<u64> = segment_reader
+            .doc_ids_alive()
+            .map(|doc| fast_field_reader.get(doc))
+            .collect();
+        assert_eq!(&in_order_alive_ids[..], &[9, 8, 7, 6, 5, 4, 1, 0]);
+        Ok(())
+    }
+
+    #[derive(Debug, Clone, Copy)]
+    enum IndexingOp {
+        AddDoc { id: u64 },
+        DeleteDoc { id: u64 },
+        Commit,
+    }
+
+    fn operation_strategy() -> impl Strategy<Value = IndexingOp> {
+        prop_oneof![
+            (0u64..10u64).prop_map(|id| IndexingOp::DeleteDoc { id }),
+            (0u64..10u64).prop_map(|id| IndexingOp::AddDoc { id }),
+            (0u64..2u64).prop_map(|_| IndexingOp::Commit),
+        ]
+    }
+
+    fn expected_ids(ops: &[IndexingOp]) -> (HashMap<u64, u64>, HashSet<u64>) {
+        let mut existing_ids = HashMap::new();
+        let mut deleted_ids = HashSet::new();
+        for &op in ops {
+            match op {
+                IndexingOp::AddDoc { id } => {
+                    *existing_ids.entry(id).or_insert(0) += 1;
+                    deleted_ids.remove(&id);
+                }
+                IndexingOp::DeleteDoc { id } => {
+                    existing_ids.remove(&id);
+                    deleted_ids.insert(id);
+                }
+                _ => {}
+            }
+        }
+        (existing_ids, deleted_ids)
+    }
+
+    fn test_operation_strategy(
+        ops: &[IndexingOp],
+        sort_index: bool,
+        force_merge: bool,
+    ) -> crate::Result<()> {
+        let mut schema_builder = schema::Schema::builder();
+        let id_field = schema_builder.add_u64_field("id", FAST | INDEXED | STORED);
+        let text_field = schema_builder.add_text_field(
+            "text_field",
+            TextOptions::default()
+                .set_indexing_options(
+                    TextFieldIndexing::default()
+                        .set_index_option(schema::IndexRecordOption::WithFreqsAndPositions),
+                )
+                .set_stored(),
+        );
+
+        let large_text_field = schema_builder.add_text_field("large_text_field", TEXT | STORED);
+
+        let multi_numbers = schema_builder.add_u64_field(
+            "multi_numbers",
+            IntOptions::default()
+                .set_fast(Cardinality::MultiValues)
+                .set_stored(),
+        );
+        let facet_field = schema_builder.add_facet_field("facet", INDEXED);
+        let schema = schema_builder.build();
+        let settings = if sort_index {
+            IndexSettings {
+                sort_by_field: Some(IndexSortByField {
+                    field: "id".to_string(),
+                    order: Order::Asc,
+                }),
+                ..Default::default()
+            }
+        } else {
+            IndexSettings {
+                ..Default::default()
+            }
+        };
+        let index = Index::builder()
+            .schema(schema)
+            .settings(settings)
+            .create_in_ram()?;
+        let mut index_writer = index.writer_for_tests()?;
+        for &op in ops {
+            match op {
+                IndexingOp::AddDoc { id } => {
+                    let facet = Facet::from(&("/cola/".to_string() + &id.to_string()));
+                    index_writer
+                        .add_document(doc!(id_field=>id, multi_numbers=> id, multi_numbers => id, text_field => id.to_string(), facet_field => facet, large_text_field=> LOREM));
+                }
+                IndexingOp::DeleteDoc { id } => {
+                    index_writer.delete_term(Term::from_field_u64(id_field, id));
+                }
+                IndexingOp::Commit => {
+                    index_writer.commit()?;
+                }
+            }
+        }
+        index_writer.commit()?;
+
+        let searcher = index.reader()?.searcher();
+        if force_merge {
+            index_writer.wait_merging_threads()?;
+            let mut index_writer = index.writer_for_tests()?;
+            let segment_ids = index
+                .searchable_segment_ids()
+                .expect("Searchable segments failed.");
+            if segment_ids.len() >= 2 {
+                block_on(index_writer.merge(&segment_ids)).unwrap();
+                assert!(index_writer.wait_merging_threads().is_ok());
+            }
+        }
+        let ids: HashSet<u64> = searcher
+            .segment_readers()
+            .iter()
+            .flat_map(|segment_reader| {
+                let ff_reader = segment_reader.fast_fields().u64(id_field).unwrap();
+                segment_reader
+                    .doc_ids_alive()
+                    .map(move |doc| ff_reader.get(doc))
+            })
+            .collect();
+
+        let (expected_ids_and_num_occurences, deleted_ids) = expected_ids(ops);
+        assert_eq!(
+            ids,
+            expected_ids_and_num_occurences
+                .keys()
+                .cloned()
+                .collect::<HashSet<_>>()
+        );
+
+        // multivalue fast field tests
+        for segment_reader in searcher.segment_readers().iter() {
+            let ff_reader = segment_reader.fast_fields().u64s(multi_numbers).unwrap();
+            for doc in segment_reader.doc_ids_alive() {
+                let mut vals = vec![];
+                ff_reader.get_vals(doc, &mut vals);
+                assert_eq!(vals.len(), 2);
+                assert_eq!(vals[0], vals[1]);
+                assert!(expected_ids_and_num_occurences.contains_key(&vals[0]));
+            }
+        }
+
+        // doc store tests
+        for segment_reader in searcher.segment_readers().iter() {
+            let store_reader = segment_reader.get_store_reader().unwrap();
+            // test store iterator
+            for doc in store_reader.iter(segment_reader.delete_bitset()) {
+                let id = doc
+                    .unwrap()
+                    .get_first(id_field)
+                    .unwrap()
+                    .u64_value()
+                    .unwrap();
+                assert!(expected_ids_and_num_occurences.contains_key(&id));
+            }
+            // test store random access
+            for doc_id in segment_reader.doc_ids_alive() {
+                let id = store_reader
+                    .get(doc_id)
+                    .unwrap()
+                    .get_first(id_field)
+                    .unwrap()
+                    .u64_value()
+                    .unwrap();
+                assert!(expected_ids_and_num_occurences.contains_key(&id));
+                let id2 = store_reader
+                    .get(doc_id)
+                    .unwrap()
+                    .get_first(multi_numbers)
+                    .unwrap()
+                    .u64_value()
+                    .unwrap();
+                assert_eq!(id, id2);
+            }
+        }
+        // test search
+        let my_text_field = index.schema().get_field("text_field").unwrap();
+
+        let do_search = |term: &str| {
+            let query = QueryParser::for_index(&index, vec![my_text_field])
+                .parse_query(term)
+                .unwrap();
+            let top_docs: Vec<(f32, DocAddress)> =
+                searcher.search(&query, &TopDocs::with_limit(1000)).unwrap();
+
+            top_docs.iter().map(|el| el.1).collect::<Vec<_>>()
+        };
+
+        for (existing_id, count) in expected_ids_and_num_occurences {
+            assert_eq!(do_search(&existing_id.to_string()).len() as u64, count);
+        }
+        for existing_id in deleted_ids {
+            assert_eq!(do_search(&existing_id.to_string()).len(), 0);
+        }
+        // test facets
+        for segment_reader in searcher.segment_readers().iter() {
+            let mut facet_reader = segment_reader.facet_reader(facet_field).unwrap();
+            let ff_reader = segment_reader.fast_fields().u64(id_field).unwrap();
+            for doc_id in segment_reader.doc_ids_alive() {
+                let mut facet_ords = Vec::new();
+                facet_reader.facet_ords(doc_id, &mut facet_ords);
+                assert_eq!(facet_ords.len(), 1);
+                let mut facet = Facet::default();
+                facet_reader
+                    .facet_from_ord(facet_ords[0], &mut facet)
+                    .unwrap();
+                let id = ff_reader.get(doc_id);
+                let facet_expected = Facet::from(&("/cola/".to_string() + &id.to_string()));
+
+                assert_eq!(facet, facet_expected);
+            }
+        }
+        Ok(())
+    }
+
+    proptest! {
+        #[test]
+        fn test_delete_with_sort_proptest(ops in proptest::collection::vec(operation_strategy(), 1..10)) {
+            assert!(test_operation_strategy(&ops[..], true, false).is_ok());
+        }
+        #[test]
+        fn test_delete_without_sort_proptest(ops in proptest::collection::vec(operation_strategy(), 1..10)) {
+            assert!(test_operation_strategy(&ops[..], false, false).is_ok());
+        }
+        #[test]
+        fn test_delete_with_sort_proptest_with_merge(ops in proptest::collection::vec(operation_strategy(), 1..10)) {
+            assert!(test_operation_strategy(&ops[..], true, true).is_ok());
+        }
+        #[test]
+        fn test_delete_without_sort_proptest_with_merge(ops in proptest::collection::vec(operation_strategy(), 1..10)) {
+            assert!(test_operation_strategy(&ops[..], false, true).is_ok());
+        }
+    }
+
+    #[test]
+    fn test_delete_with_sort_by_field_last_opstamp_is_not_max() -> crate::Result<()> {
+        let mut schema_builder = schema::Schema::builder();
+        let sort_by_field = schema_builder.add_u64_field("sort_by", FAST);
+        let id_field = schema_builder.add_u64_field("id", INDEXED);
+        let schema = schema_builder.build();
+
+        let settings = IndexSettings {
+            sort_by_field: Some(IndexSortByField {
+                field: "sort_by".to_string(),
+                order: Order::Asc,
+            }),
+            ..Default::default()
+        };
+
+        let index = Index::builder()
+            .schema(schema)
+            .settings(settings)
+            .create_in_ram()?;
+        let mut index_writer = index.writer_for_tests()?;
+
+        // We add a doc...
+        index_writer.add_document(doc!(sort_by_field => 2u64, id_field => 0u64));
+        // And remove it.
+        index_writer.delete_term(Term::from_field_u64(id_field, 0u64));
+        // We add another doc.
+        index_writer.add_document(doc!(sort_by_field=>1u64, id_field => 0u64));
+
+        // The expected result is a segment with
+        // maxdoc = 2
+        // numdoc = 1.
+        index_writer.commit()?;
+
+        let searcher = index.reader()?.searcher();
+        assert_eq!(searcher.segment_readers().len(), 1);
+
+        let segment_reader = searcher.segment_reader(0);
+        assert_eq!(segment_reader.max_doc(), 2);
+        assert_eq!(segment_reader.num_deleted_docs(), 1);
+        Ok(())
+    }
+
     #[test]
     fn test_index_doc_missing_field() {
         let mut schema_builder = schema::Schema::builder();

src/indexer/merger.rs

@@ -1,14 +1,15 @@
-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::FastFieldDataAccess;
 use crate::fastfield::FastFieldReader;
-use crate::fastfield::FastFieldSerializer;
+use crate::fastfield::FastFieldStats;
 use crate::fastfield::MultiValuedFastFieldReader;
 use crate::fieldnorm::FieldNormsSerializer;
 use crate::fieldnorm::FieldNormsWriter;
 use crate::fieldnorm::{FieldNormReader, FieldNormReaders};
+use crate::indexer::doc_id_mapping::SegmentDocidMapping;
 use crate::indexer::SegmentSerializer;
 use crate::postings::Postings;
 use crate::postings::{InvertedIndexSerializer, SegmentPostings};
@@ -18,17 +19,18 @@ use crate::schema::{Field, Schema};
 use crate::store::StoreWriter;
 use crate::termdict::TermMerger;
 use crate::termdict::TermOrdinal;
+use crate::IndexSortByField;
 use crate::{common::HasLen, fastfield::MultiValueLength};
 use crate::{common::MAX_DOC_LIMIT, IndexSettings};
 use crate::{core::Segment, indexer::doc_id_mapping::expect_field_id_for_sort_field};
 use crate::{core::SegmentReader, Order};
-use crate::{core::SerializableSegment, IndexSortByField};
 use crate::{
     docset::{DocSet, TERMINATED},
     SegmentOrdinal,
 };
 use crate::{DocId, InvertedIndexReader, SegmentComponent};
 use itertools::Itertools;
+use measure_time::debug_time;
 use std::cmp;
 use std::collections::HashMap;
 use std::sync::Arc;
@@ -214,7 +216,7 @@ impl IndexMerger {
         let mut readers_with_min_sort_values = readers
             .into_iter()
             .map(|reader| {
-                let accessor = Self::get_sort_field_accessor(&reader, &sort_by_field)?;
+                let accessor = Self::get_sort_field_accessor(&reader, sort_by_field)?;
                 Ok((reader, accessor.min_value()))
             })
             .collect::<crate::Result<Vec<_>>>()?;
@@ -232,33 +234,24 @@ impl IndexMerger {
     fn write_fieldnorms(
         &self,
         mut fieldnorms_serializer: FieldNormsSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
         let fields = FieldNormsWriter::fields_with_fieldnorm(&self.schema);
         let mut fieldnorms_data = Vec::with_capacity(self.max_doc as usize);
         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 =
-                        &fieldnorms_readers[reader_with_ordinal.ordinal as usize];
-                    let fieldnorm_id = fieldnorms_reader.fieldnorm_id(*doc_id);
-                    fieldnorms_data.push(fieldnorm_id);
-                }
-            } else {
-                for reader in &self.readers {
-                    let fieldnorms_reader = reader.get_fieldnorms_reader(field)?;
-                    for doc_id in reader.doc_ids_alive() {
-                        let fieldnorm_id = fieldnorms_reader.fieldnorm_id(doc_id);
-                        fieldnorms_data.push(fieldnorm_id);
-                    }
-                }
+
+            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.iter() {
+                let fieldnorms_reader = &fieldnorms_readers[reader_with_ordinal.ordinal as usize];
+                let fieldnorm_id = fieldnorms_reader.fieldnorm_id(*doc_id);
+                fieldnorms_data.push(fieldnorm_id);
             }
+
             fieldnorms_serializer.serialize_field(field, &fieldnorms_data[..])?;
         }
         fieldnorms_serializer.close()?;
@@ -269,8 +262,10 @@ impl IndexMerger {
         &self,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
         mut term_ord_mappings: HashMap<Field, TermOrdinalMapping>,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
+        debug_time!("write_fast_fields");
+
         for (field, field_entry) in self.schema.fields() {
             let field_type = field_entry.field_type();
             match field_type {
@@ -318,7 +313,7 @@ impl IndexMerger {
         &self,
         field: Field,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
         let (min_value, max_value) = self.readers.iter().map(|reader|{
                 let u64_reader: DynamicFastFieldReader<u64> = reader
@@ -327,7 +322,7 @@ impl IndexMerger {
                 .expect("Failed to find a reader for single fast field. This is a tantivy bug and it should never happen.");
                 compute_min_max_val(&u64_reader, reader.max_doc(), reader.delete_bitset())
             })
-            .filter_map(|x| x)
+            .flatten()
             .reduce(|a, b| {
                 (a.0.min(b.0), a.1.max(b.1))
             }).expect("Unexpected error, empty readers in IndexMerger");
@@ -343,51 +338,44 @@ impl IndexMerger {
                 u64_reader
             })
             .collect::<Vec<_>>();
-        if let Some(doc_id_mapping) = doc_id_mapping {
-            let sorted_doc_ids = doc_id_mapping.iter().map(|(doc_id, reader_with_ordinal)| {
-                (
-                    doc_id,
-                    &fast_field_readers[reader_with_ordinal.ordinal as usize],
-                )
-            });
-            // add values in order of the new doc_ids
-            let mut fast_single_field_serializer =
-                fast_field_serializer.new_u64_fast_field(field, min_value, max_value)?;
-            for (doc_id, field_reader) in sorted_doc_ids {
-                let val = field_reader.get(*doc_id);
-                fast_single_field_serializer.add_val(val)?;
-            }
 
-            fast_single_field_serializer.close_field()?;
-            Ok(())
-        } else {
-            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: 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.");
-                (reader.max_doc(), u64_reader, reader.delete_bitset())
-            }).collect::<Vec<_>>();
-
-            let mut fast_single_field_serializer =
-                fast_field_serializer.new_u64_fast_field(field, min_value, max_value)?;
-            for (max_doc, u64_reader, delete_bitset_opt) in u64_readers {
-                for doc_id in 0u32..max_doc {
-                    let is_deleted = delete_bitset_opt
-                        .map(|delete_bitset| delete_bitset.is_deleted(doc_id))
-                        .unwrap_or(false);
-                    if !is_deleted {
-                        let val = u64_reader.get(doc_id);
-                        fast_single_field_serializer.add_val(val)?;
-                    }
-                }
-            }
-
-            fast_single_field_serializer.close_field()?;
-            Ok(())
+        let stats = FastFieldStats {
+            min_value,
+            max_value,
+            num_vals: doc_id_mapping.len() as u64,
+        };
+        #[derive(Clone)]
+        struct SortedDocidFieldAccessProvider<'a> {
+            doc_id_mapping: &'a SegmentDocidMapping<'a>,
+            fast_field_readers: &'a Vec<DynamicFastFieldReader<u64>>,
         }
+        impl<'a> FastFieldDataAccess for SortedDocidFieldAccessProvider<'a> {
+            fn get_val(&self, doc: u64) -> u64 {
+                let (doc_id, reader_with_ordinal) = self.doc_id_mapping[doc as usize];
+                self.fast_field_readers[reader_with_ordinal.ordinal as usize].get(doc_id)
+            }
+        }
+        let fastfield_accessor = SortedDocidFieldAccessProvider {
+            doc_id_mapping,
+            fast_field_readers: &fast_field_readers,
+        };
+        let iter1 = doc_id_mapping.iter().map(|(doc_id, reader_with_ordinal)| {
+            let fast_field_reader = &fast_field_readers[reader_with_ordinal.ordinal as usize];
+            fast_field_reader.get(*doc_id)
+        });
+        let iter2 = doc_id_mapping.iter().map(|(doc_id, reader_with_ordinal)| {
+            let fast_field_reader = &fast_field_readers[reader_with_ordinal.ordinal as usize];
+            fast_field_reader.get(*doc_id)
+        });
+        fast_field_serializer.create_auto_detect_u64_fast_field(
+            field,
+            stats,
+            fastfield_accessor,
+            iter1,
+            iter2,
+        )?;
+
+        Ok(())
     }
 
     /// Checks if the readers are disjunct for their sort property and in the correct order to be
@@ -416,7 +404,7 @@ impl IndexMerger {
         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 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)
     }
@@ -451,7 +439,7 @@ impl IndexMerger {
     pub(crate) fn generate_doc_id_mapping(
         &self,
         sort_by_field: &IndexSortByField,
-    ) -> crate::Result<Vec<(DocId, SegmentReaderWithOrdinal)>> {
+    ) -> crate::Result<SegmentDocidMapping> {
         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
 
@@ -487,7 +475,7 @@ impl IndexMerger {
             })
             .map(|(doc_id, reader_with_id, _)| (doc_id, reader_with_id))
             .collect::<Vec<_>>();
-        Ok(sorted_doc_ids)
+        Ok(SegmentDocidMapping::new(sorted_doc_ids, false))
     }
 
     // Creating the index file to point into the data, generic over `BytesFastFieldReader` and
@@ -496,19 +484,21 @@ impl IndexMerger {
     // Important: reader_and_field_accessor needs
     // to have the same order as self.readers since ReaderWithOrdinal
     // is used to index the reader_and_field_accessors vec.
-    fn write_1_n_fast_field_idx_generic(
+    fn write_1_n_fast_field_idx_generic<T: MultiValueLength>(
         field: Field,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
-        reader_and_field_accessors: &[(&SegmentReader, impl MultiValueLength)],
-    ) -> crate::Result<()> {
+        doc_id_mapping: &SegmentDocidMapping,
+        reader_and_field_accessors: &[(&SegmentReader, T)],
+    ) -> crate::Result<Vec<u64>> {
         let mut total_num_vals = 0u64;
         // In the first pass, we compute the total number of vals.
         //
         // This is required by the bitpacker, as it needs to know
         // what should be the bit length use for bitpacking.
+        let mut idx_num_vals = 0;
         for (reader, u64s_reader) in reader_and_field_accessors.iter() {
             if let Some(delete_bitset) = reader.delete_bitset() {
+                idx_num_vals += reader.max_doc() as u64 - delete_bitset.len() as u64;
                 for doc in 0u32..reader.max_doc() {
                     if delete_bitset.is_alive(doc) {
                         let num_vals = u64s_reader.get_len(doc) as u64;
@@ -516,46 +506,49 @@ impl IndexMerger {
                     }
                 }
             } else {
+                idx_num_vals += reader.max_doc() as u64;
                 total_num_vals += u64s_reader.get_total_len();
             }
         }
 
+        let stats = FastFieldStats {
+            max_value: total_num_vals,
+            num_vals: idx_num_vals,
+            min_value: 0,
+        };
         // We can now create our `idx` serializer, and in a second pass,
         // can effectively push the different indexes.
-        if let Some(doc_id_mapping) = doc_id_mapping {
-            let mut serialize_idx =
-                fast_field_serializer.new_u64_fast_field_with_idx(field, 0, total_num_vals, 0)?;
 
-            let mut offset = 0;
-            for (doc_id, reader) in doc_id_mapping {
-                let reader = &reader_and_field_accessors[reader.ordinal as usize].1;
-                serialize_idx.add_val(offset)?;
-                offset += reader.get_len(*doc_id) as u64;
-            }
-            serialize_idx.add_val(offset as u64)?;
+        // copying into a temp vec is not ideal, but the fast field codec api requires random
+        // access, which is used in the estimation. It's possible to 1. calculate random
+        // acccess on the fly or 2. change the codec api to make random access optional, but
+        // they both have also major drawbacks.
 
-            serialize_idx.close_field()?;
-        } else {
-            let mut serialize_idx =
-                fast_field_serializer.new_u64_fast_field_with_idx(field, 0, total_num_vals, 0)?;
-            let mut idx = 0;
-            for (segment_reader, u64s_reader) in reader_and_field_accessors.iter() {
-                for doc in segment_reader.doc_ids_alive() {
-                    serialize_idx.add_val(idx)?;
-                    idx += u64s_reader.get_len(doc) as u64;
-                }
-            }
-            serialize_idx.add_val(idx)?;
-            serialize_idx.close_field()?;
+        let mut offsets = vec![];
+        let mut offset = 0;
+        for (doc_id, reader) in doc_id_mapping.iter() {
+            let reader = &reader_and_field_accessors[reader.ordinal as usize].1;
+            offsets.push(offset);
+            offset += reader.get_len(*doc_id) as u64;
         }
-        Ok(())
+        offsets.push(offset);
+
+        fast_field_serializer.create_auto_detect_u64_fast_field(
+            field,
+            stats,
+            &offsets[..],
+            offsets.iter().cloned(),
+            offsets.iter().cloned(),
+        )?;
+        Ok(offsets)
     }
+    /// Returns the fastfield index (index for the data, not the data).
     fn write_multi_value_fast_field_idx(
         &self,
         field: Field,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
-    ) -> crate::Result<()> {
+        doc_id_mapping: &SegmentDocidMapping,
+    ) -> crate::Result<Vec<u64>> {
         let reader_and_field_accessors = self.readers.iter().map(|reader|{
             let u64s_reader: MultiValuedFastFieldReader<u64> = reader.fast_fields()
                 .typed_fast_field_multi_reader(field)
@@ -576,9 +569,11 @@ impl IndexMerger {
         field: Field,
         term_ordinal_mappings: &TermOrdinalMapping,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
-        // Multifastfield consists in 2 fastfields.
+        debug_time!("write_hierarchical_facet_field");
+
+        // Multifastfield consists of 2 fastfields.
         // The first serves as an index into the second one and is stricly increasing.
         // The second contains the actual values.
 
@@ -603,53 +598,61 @@ impl IndexMerger {
             let mut serialize_vals =
                 fast_field_serializer.new_u64_fast_field_with_idx(field, 0u64, max_term_ord, 1)?;
             let mut vals = Vec::with_capacity(100);
-            if let Some(doc_id_mapping) = doc_id_mapping {
-                for (old_doc_id, reader_with_ordinal) in doc_id_mapping {
-                    let term_ordinal_mapping: &[TermOrdinal] =
-                        term_ordinal_mappings.get_segment(reader_with_ordinal.ordinal as usize);
 
-                    let ff_reader = &fast_field_reader[reader_with_ordinal.ordinal as usize];
-                    ff_reader.get_vals(*old_doc_id, &mut vals);
-                    for &prev_term_ord in &vals {
-                        let new_term_ord = term_ordinal_mapping[prev_term_ord as usize];
-                        serialize_vals.add_val(new_term_ord)?;
-                    }
-                }
-            } else {
-                for (segment_ord, segment_reader) in self.readers.iter().enumerate() {
-                    let term_ordinal_mapping: &[TermOrdinal] =
-                        term_ordinal_mappings.get_segment(segment_ord);
-                    let ff_reader = &fast_field_reader[segment_ord as usize];
-                    // TODO optimize if no deletes
-                    for doc in segment_reader.doc_ids_alive() {
-                        ff_reader.get_vals(doc, &mut vals);
-                        for &prev_term_ord in &vals {
-                            let new_term_ord = term_ordinal_mapping[prev_term_ord as usize];
-                            serialize_vals.add_val(new_term_ord)?;
-                        }
-                    }
+            for (old_doc_id, reader_with_ordinal) in doc_id_mapping.iter() {
+                let term_ordinal_mapping: &[TermOrdinal] =
+                    term_ordinal_mappings.get_segment(reader_with_ordinal.ordinal as usize);
+
+                let ff_reader = &fast_field_reader[reader_with_ordinal.ordinal as usize];
+                ff_reader.get_vals(*old_doc_id, &mut vals);
+                for &prev_term_ord in &vals {
+                    let new_term_ord = term_ordinal_mapping[prev_term_ord as usize];
+                    serialize_vals.add_val(new_term_ord)?;
                 }
             }
+
             serialize_vals.close_field()?;
         }
         Ok(())
     }
 
+    /// Creates a mapping if the segments are stacked. this is helpful to merge codelines between index
+    /// sorting and the others
+    pub(crate) fn get_doc_id_from_concatenated_data(&self) -> crate::Result<SegmentDocidMapping> {
+        let mapping: Vec<_> = self
+            .readers
+            .iter()
+            .enumerate()
+            .map(|(ordinal, reader)| {
+                let reader_with_ordinal = SegmentReaderWithOrdinal {
+                    ordinal: ordinal as u32,
+                    reader,
+                };
+                reader
+                    .doc_ids_alive()
+                    .map(move |doc_id| (doc_id, reader_with_ordinal))
+            })
+            .flatten()
+            .collect();
+        Ok(SegmentDocidMapping::new(mapping, true))
+    }
     fn write_multi_fast_field(
         &self,
         field: Field,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
         // Multifastfield consists in 2 fastfields.
         // The first serves as an index into the second one and is stricly increasing.
         // The second contains the actual values.
 
         // First we merge the idx fast field.
-        self.write_multi_value_fast_field_idx(field, fast_field_serializer, doc_id_mapping)?;
+        let offsets =
+            self.write_multi_value_fast_field_idx(field, fast_field_serializer, doc_id_mapping)?;
 
         let mut min_value = u64::max_value();
         let mut max_value = u64::min_value();
+        let mut num_vals = 0;
 
         let mut vals = Vec::with_capacity(100);
 
@@ -675,6 +678,7 @@ impl IndexMerger {
                     min_value = cmp::min(val, min_value);
                     max_value = cmp::max(val, max_value);
                 }
+                num_vals += vals.len();
             }
             ff_readers.push(ff_reader);
             // TODO optimize when no deletes
@@ -685,40 +689,76 @@ impl IndexMerger {
             max_value = 0;
         }
 
-        let fast_field_reader = self
-            .readers
-            .iter()
-            .map(|reader| {
-                let ff_reader : MultiValuedFastFieldReader<u64> = reader.fast_fields()
-                .typed_fast_field_multi_reader(field)
-                .expect("Failed to find index for multivalued field. This is a bug in tantivy, please report.");
-                ff_reader
-            })
-            .collect::<Vec<_>>();
-
         // We can now initialize our serializer, and push it the different values
-        let mut serialize_vals =
-            fast_field_serializer.new_u64_fast_field_with_idx(field, min_value, max_value, 1)?;
-        if let Some(doc_id_mapping) = doc_id_mapping {
-            for (doc_id, reader_with_ordinal) in doc_id_mapping {
-                let ff_reader = &fast_field_reader[reader_with_ordinal.ordinal as usize];
-                ff_reader.get_vals(*doc_id, &mut vals);
-                for &val in &vals {
-                    serialize_vals.add_val(val)?;
-                }
-            }
-        } else {
-            for (reader, ff_reader) in self.readers.iter().zip(ff_readers) {
-                // TODO optimize if no deletes
-                for doc in reader.doc_ids_alive() {
-                    ff_reader.get_vals(doc, &mut vals);
-                    for &val in &vals {
-                        serialize_vals.add_val(val)?;
-                    }
-                }
+        let stats = FastFieldStats {
+            max_value,
+            num_vals: num_vals as u64,
+            min_value,
+        };
+
+        struct SortedDocidMultiValueAccessProvider<'a> {
+            doc_id_mapping: &'a SegmentDocidMapping<'a>,
+            fast_field_readers: &'a Vec<MultiValuedFastFieldReader<u64>>,
+            offsets: Vec<u64>,
+        }
+        impl<'a> FastFieldDataAccess for SortedDocidMultiValueAccessProvider<'a> {
+            fn get_val(&self, pos: u64) -> u64 {
+                // use the offsets index to find the doc_id which will contain the position.
+                // the offsets are stricly increasing so we can do a simple search on it.
+                let new_docid = self
+                    .offsets
+                    .iter()
+                    .position(|&offset| offset > pos)
+                    .expect("pos is out of bounds")
+                    - 1;
+
+                // now we need to find the position of `pos` in the multivalued bucket
+                let num_pos_covered_until_now = self.offsets[new_docid];
+                let pos_in_values = pos - num_pos_covered_until_now;
+
+                let (old_doc_id, reader_with_ordinal) = self.doc_id_mapping[new_docid as usize];
+                let num_vals = self.fast_field_readers[reader_with_ordinal.ordinal as usize]
+                    .get_len(old_doc_id);
+                assert!(num_vals >= pos_in_values);
+                let mut vals = vec![];
+                self.fast_field_readers[reader_with_ordinal.ordinal as usize]
+                    .get_vals(old_doc_id, &mut vals);
+
+                vals[pos_in_values as usize]
             }
         }
-        serialize_vals.close_field()?;
+        let fastfield_accessor = SortedDocidMultiValueAccessProvider {
+            doc_id_mapping,
+            fast_field_readers: &ff_readers,
+            offsets,
+        };
+        let iter1 = doc_id_mapping
+            .iter()
+            .map(|(doc_id, reader_with_ordinal)| {
+                let ff_reader = &ff_readers[reader_with_ordinal.ordinal as usize];
+                let mut vals = vec![];
+                ff_reader.get_vals(*doc_id, &mut vals);
+                vals.into_iter()
+            })
+            .flatten();
+        let iter2 = doc_id_mapping
+            .iter()
+            .map(|(doc_id, reader_with_ordinal)| {
+                let ff_reader = &ff_readers[reader_with_ordinal.ordinal as usize];
+                let mut vals = vec![];
+                ff_reader.get_vals(*doc_id, &mut vals);
+                vals.into_iter()
+            })
+            .flatten();
+        fast_field_serializer.create_auto_detect_u64_fast_field_with_idx(
+            field,
+            stats,
+            fastfield_accessor,
+            iter1,
+            iter2,
+            1,
+        )?;
+
         Ok(())
     }
 
@@ -726,7 +766,7 @@ impl IndexMerger {
         &self,
         field: Field,
         fast_field_serializer: &mut CompositeFastFieldSerializer,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
         let reader_and_field_accessors = self
             .readers
@@ -745,24 +785,13 @@ impl IndexMerger {
             &reader_and_field_accessors,
         )?;
         let mut serialize_vals = fast_field_serializer.new_bytes_fast_field_with_idx(field, 1);
-        if let Some(doc_id_mapping) = doc_id_mapping {
-            for (doc_id, reader_with_ordinal) in doc_id_mapping {
-                let bytes_reader =
-                    &reader_and_field_accessors[reader_with_ordinal.ordinal as usize].1;
-                let val = bytes_reader.get_bytes(*doc_id);
-                serialize_vals.write_all(val)?;
-            }
-        } else {
-            for segment_reader in &self.readers {
-                let bytes_reader = segment_reader.fast_fields().bytes(field)
-                .expect("Failed to find bytes field in fast field reader. This is a bug in tantivy. Please report.");
-                // TODO: optimize if no deletes
-                for doc in segment_reader.doc_ids_alive() {
-                    let val = bytes_reader.get_bytes(doc);
-                    serialize_vals.write_all(val)?;
-                }
-            }
+
+        for (doc_id, reader_with_ordinal) in doc_id_mapping.iter() {
+            let bytes_reader = &reader_and_field_accessors[reader_with_ordinal.ordinal as usize].1;
+            let val = bytes_reader.get_bytes(*doc_id);
+            serialize_vals.write_all(val)?;
         }
+
         serialize_vals.flush()?;
         Ok(())
     }
@@ -773,8 +802,9 @@ impl IndexMerger {
         field_type: &FieldType,
         serializer: &mut InvertedIndexSerializer,
         fieldnorm_reader: Option<FieldNormReader>,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<Option<TermOrdinalMapping>> {
+        debug_time!("write_postings_for_field");
         let mut positions_buffer: Vec<u32> = Vec::with_capacity(1_000);
         let mut delta_computer = DeltaComputer::new();
 
@@ -799,40 +829,23 @@ impl IndexMerger {
         };
 
         let mut merged_terms = TermMerger::new(field_term_streams);
-        let mut max_doc = 0;
 
         // map from segment doc ids to the resulting merged segment doc id.
-        let mut merged_doc_id_map: Vec<Vec<Option<DocId>>> = Vec::with_capacity(self.readers.len());
 
-        if let Some(doc_id_mapping) = doc_id_mapping {
-            merged_doc_id_map = self
-                .readers
-                .iter()
-                .map(|reader| {
-                    let mut segment_local_map = vec![];
-                    segment_local_map.resize(reader.max_doc() as usize, None);
-                    segment_local_map
-                })
-                .collect();
-            for (new_doc_id, (old_doc_id, segment_and_ordinal)) in doc_id_mapping.iter().enumerate()
-            {
-                let segment_map = &mut merged_doc_id_map[segment_and_ordinal.ordinal as usize];
-                segment_map[*old_doc_id as usize] = Some(new_doc_id as DocId);
-            }
-        } else {
-            for reader in &self.readers {
-                let mut segment_local_map = Vec::with_capacity(reader.max_doc() as usize);
-                for doc_id in 0..reader.max_doc() {
-                    if reader.is_deleted(doc_id) {
-                        segment_local_map.push(None);
-                    } else {
-                        segment_local_map.push(Some(max_doc));
-                        max_doc += 1u32;
-                    }
-                }
-                merged_doc_id_map.push(segment_local_map);
-            }
+        let mut merged_doc_id_map: Vec<Vec<Option<DocId>>> = self
+            .readers
+            .iter()
+            .map(|reader| {
+                let mut segment_local_map = vec![];
+                segment_local_map.resize(reader.max_doc() as usize, None);
+                segment_local_map
+            })
+            .collect();
+        for (new_doc_id, (old_doc_id, segment_and_ordinal)) in doc_id_mapping.iter().enumerate() {
+            let segment_map = &mut merged_doc_id_map[segment_and_ordinal.ordinal as usize];
+            segment_map[*old_doc_id as usize] = Some(new_doc_id as DocId);
         }
+
         // The total number of tokens will only be exact when there has been no deletes.
         //
         // Otherwise, we approximate by removing deleted documents proportionally.
@@ -928,7 +941,7 @@ impl IndexMerger {
                         // I think this is not strictly necessary, it would be possible to
                         // avoid the loading into a vec via some form of kmerge, but then the merge
                         // logic would deviate much more from the stacking case (unsorted index)
-                        if doc_id_mapping.is_some() {
+                        if !doc_id_mapping.is_trivial() {
                             doc_id_and_positions.push((
                                 remapped_doc_id,
                                 term_freq,
@@ -943,7 +956,7 @@ impl IndexMerger {
                     doc = segment_postings.advance();
                 }
             }
-            if doc_id_mapping.is_some() {
+            if !doc_id_mapping.is_trivial() {
                 doc_id_and_positions.sort_unstable_by_key(|&(doc_id, _, _)| doc_id);
                 for (doc_id, term_freq, positions) in &doc_id_and_positions {
                     field_serializer.write_doc(*doc_id, *term_freq, positions);
@@ -962,7 +975,7 @@ impl IndexMerger {
         &self,
         serializer: &mut InvertedIndexSerializer,
         fieldnorm_readers: FieldNormReaders,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<HashMap<Field, TermOrdinalMapping>> {
         let mut term_ordinal_mappings = HashMap::new();
         for (field, field_entry) in self.schema.fields() {
@@ -985,8 +998,10 @@ impl IndexMerger {
     fn write_storable_fields(
         &self,
         store_writer: &mut StoreWriter,
-        doc_id_mapping: &Option<Vec<(DocId, SegmentReaderWithOrdinal)>>,
+        doc_id_mapping: &SegmentDocidMapping,
     ) -> crate::Result<()> {
+        debug_time!("write_storable_fields");
+
         let store_readers: Vec<_> = self
             .readers
             .iter()
@@ -997,8 +1012,8 @@ impl IndexMerger {
             .enumerate()
             .map(|(i, store)| store.iter_raw(self.readers[i].delete_bitset()))
             .collect();
-        if let Some(doc_id_mapping) = doc_id_mapping {
-            for (old_doc_id, reader_with_ordinal) in doc_id_mapping {
+        if !doc_id_mapping.is_trivial() {
+            for (old_doc_id, reader_with_ordinal) in doc_id_mapping.iter() {
                 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?;
@@ -1042,25 +1057,24 @@ impl IndexMerger {
         }
         Ok(())
     }
-}
 
-impl SerializableSegment for IndexMerger {
-    fn write(
-        &self,
-        mut serializer: SegmentSerializer,
-        _: Option<&DocIdMapping>,
-    ) -> crate::Result<u32> {
+    /// Writes the merged segment by pushing information
+    /// to the `SegmentSerializer`.
+    ///
+    /// # Returns
+    /// The number of documents in the resulting segment.
+    pub fn write(&self, mut serializer: SegmentSerializer) -> crate::Result<u32> {
         let doc_id_mapping = if let Some(sort_by_field) = self.index_settings.sort_by_field.as_ref()
         {
             // 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
+                self.get_doc_id_from_concatenated_data()?
             } else {
-                Some(self.generate_doc_id_mapping(sort_by_field)?)
+                self.generate_doc_id_mapping(sort_by_field)?
             }
         } else {
-            None
+            self.get_doc_id_from_concatenated_data()?
         };
 
         if let Some(fieldnorms_serializer) = serializer.extract_fieldnorms_serializer() {

src/indexer/merger_sorted_index_test.rs

@@ -362,6 +362,7 @@ mod bench_sorted_index_merge {
 
     use crate::core::Index;
     //use cratedoc_id, readerdoc_id_mappinglet vals = reader.fate::schema;
+    use crate::fastfield::DynamicFastFieldReader;
     use crate::fastfield::FastFieldReader;
     use crate::indexer::merger::IndexMerger;
     use crate::schema::Cardinality;
@@ -422,7 +423,7 @@ 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.reader
+            let u64_reader: DynamicFastFieldReader<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.");

src/indexer/segment_manager.rs

@@ -32,6 +32,12 @@ impl SegmentRegisters {
         } else if self.committed.contains_all(segment_ids) {
             Some(SegmentsStatus::Committed)
         } else {
+            warn!(
+                "segment_ids: {:?}, committed_ids: {:?}, uncommitted_ids {:?}",
+                segment_ids,
+                self.committed.segment_ids(),
+                self.uncommitted.segment_ids()
+            );
             None
         }
     }
@@ -58,21 +64,6 @@ impl Debug for SegmentManager {
     }
 }
 
-pub fn get_mergeable_segments(
-    in_merge_segment_ids: &HashSet<SegmentId>,
-    segment_manager: &SegmentManager,
-) -> (Vec<SegmentMeta>, Vec<SegmentMeta>) {
-    let registers_lock = segment_manager.read();
-    (
-        registers_lock
-            .committed
-            .get_mergeable_segments(in_merge_segment_ids),
-        registers_lock
-            .uncommitted
-            .get_mergeable_segments(in_merge_segment_ids),
-    )
-}
-
 impl SegmentManager {
     pub fn from_segments(
         segment_metas: Vec<SegmentMeta>,
@@ -86,6 +77,20 @@ impl SegmentManager {
         }
     }
 
+    pub fn get_mergeable_segments(
+        &self,
+        in_merge_segment_ids: &HashSet<SegmentId>,
+    ) -> (Vec<SegmentMeta>, Vec<SegmentMeta>) {
+        let registers_lock = self.read();
+        (
+            registers_lock
+                .committed
+                .get_mergeable_segments(in_merge_segment_ids),
+            registers_lock
+                .uncommitted
+                .get_mergeable_segments(in_merge_segment_ids),
+        )
+    }
     /// Returns all of the segment entries (committed or uncommitted)
     pub fn segment_entries(&self) -> Vec<SegmentEntry> {
         let registers_lock = self.read();

src/indexer/segment_register.rs

@@ -4,6 +4,7 @@ use crate::indexer::delete_queue::DeleteCursor;
 use crate::indexer::segment_entry::SegmentEntry;
 use std::collections::HashMap;
 use std::collections::HashSet;
+use std::fmt::Display;
 use std::fmt::{self, Debug, Formatter};
 
 /// The segment register keeps track
@@ -29,6 +30,16 @@ impl Debug for SegmentRegister {
         Ok(())
     }
 }
+impl Display for SegmentRegister {
+    fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), fmt::Error> {
+        write!(f, "SegmentRegister(")?;
+        for k in self.segment_states.keys() {
+            write!(f, "{}, ", k.short_uuid_string())?;
+        }
+        write!(f, ")")?;
+        Ok(())
+    }
+}
 
 impl SegmentRegister {
     pub fn clear(&mut self) {
@@ -46,6 +57,10 @@ impl SegmentRegister {
             .collect()
     }
 
+    pub fn segment_ids(&self) -> Vec<SegmentId> {
+        self.segment_states.keys().cloned().collect()
+    }
+
     pub fn segment_entries(&self) -> Vec<SegmentEntry> {
         self.segment_states.values().cloned().collect()
     }

src/indexer/segment_updater.rs

@@ -1,11 +1,10 @@
-use super::segment_manager::{get_mergeable_segments, SegmentManager};
+use super::segment_manager::SegmentManager;
 use crate::core::Index;
 use crate::core::IndexMeta;
 use crate::core::IndexSettings;
 use crate::core::Segment;
 use crate::core::SegmentId;
 use crate::core::SegmentMeta;
-use crate::core::SerializableSegment;
 use crate::core::META_FILEPATH;
 use crate::directory::{Directory, DirectoryClone, GarbageCollectionResult};
 use crate::indexer::delete_queue::DeleteCursor;
@@ -140,7 +139,7 @@ fn merge(
     // ... we just serialize this index merger in our new segment to merge the segments.
     let segment_serializer = SegmentSerializer::for_segment(merged_segment.clone(), true)?;
 
-    let num_docs = merger.write(segment_serializer, None)?;
+    let num_docs = merger.write(segment_serializer)?;
 
     let merged_segment_id = merged_segment.id();
 
@@ -209,7 +208,7 @@ pub fn merge_segments<Dir: Directory>(
         &segments[..],
     )?;
     let segment_serializer = SegmentSerializer::for_segment(merged_segment, true)?;
-    let num_docs = merger.write(segment_serializer, None)?;
+    let num_docs = merger.write(segment_serializer)?;
 
     let segment_meta = merged_index.new_segment_meta(merged_segment_id, num_docs);
 
@@ -528,10 +527,14 @@ impl SegmentUpdater {
         }))
     }
 
-    async fn consider_merge_options(&self) {
+    pub(crate) fn get_mergeable_segments(&self) -> (Vec<SegmentMeta>, Vec<SegmentMeta>) {
         let merge_segment_ids: HashSet<SegmentId> = self.merge_operations.segment_in_merge();
-        let (committed_segments, uncommitted_segments) =
-            get_mergeable_segments(&merge_segment_ids, &self.segment_manager);
+        self.segment_manager
+            .get_mergeable_segments(&merge_segment_ids)
+    }
+
+    async fn consider_merge_options(&self) {
+        let (committed_segments, uncommitted_segments) = self.get_mergeable_segments();
 
         // Committed segments cannot be merged with uncommitted_segments.
         // We therefore consider merges using these two sets of segments independently.
@@ -717,7 +720,7 @@ mod tests {
 
         let seg_ids = index.searchable_segment_ids()?;
         // docs exist, should have at least 1 segment
-        assert!(seg_ids.len() > 0);
+        assert!(!seg_ids.is_empty());
 
         let term_vals = vec!["a", "b", "c", "d", "e", "f"];
         for term_val in term_vals {

src/indexer/segment_writer.rs

@@ -12,12 +12,12 @@ use crate::schema::Schema;
 use crate::schema::Term;
 use crate::schema::Value;
 use crate::schema::{Field, FieldEntry};
+use crate::store::StoreReader;
 use crate::tokenizer::{BoxTokenStream, PreTokenizedStream};
 use crate::tokenizer::{FacetTokenizer, TextAnalyzer};
 use crate::tokenizer::{TokenStreamChain, Tokenizer};
 use crate::Opstamp;
 use crate::{core::Segment, store::StoreWriter};
-use crate::{core::SerializableSegment, store::StoreReader};
 use crate::{DocId, SegmentComponent};
 
 /// Computes the initial size of the hash table.
@@ -36,6 +36,20 @@ fn initial_table_size(per_thread_memory_budget: usize) -> crate::Result<usize> {
     }
 }
 
+fn remap_doc_opstamps(
+    opstamps: Vec<Opstamp>,
+    doc_id_mapping_opt: Option<&DocIdMapping>,
+) -> Vec<Opstamp> {
+    if let Some(doc_id_mapping_opt) = doc_id_mapping_opt {
+        doc_id_mapping_opt
+            .iter_old_doc_ids()
+            .map(|doc| opstamps[doc as usize])
+            .collect()
+    } else {
+        opstamps
+    }
+}
+
 /// A `SegmentWriter` is in charge of creating segment index from a
 /// set of documents.
 ///
@@ -112,14 +126,15 @@ impl SegmentWriter {
             .clone()
             .map(|sort_by_field| get_doc_id_mapping_from_field(sort_by_field, &self))
             .transpose()?;
-        write(
+        remap_and_write(
             &self.multifield_postings,
             &self.fast_field_writers,
             &self.fieldnorms_writer,
             self.segment_serializer,
             mapping.as_ref(),
         )?;
-        Ok(self.doc_opstamps)
+        let doc_opstamps = remap_doc_opstamps(self.doc_opstamps, mapping.as_ref());
+        Ok(doc_opstamps)
     }
 
     pub fn mem_usage(&self) -> usize {
@@ -176,7 +191,7 @@ impl SegmentWriter {
                             .process(&mut |token| {
                                 term_buffer.set_text(&token.text);
                                 let unordered_term_id =
-                                    multifield_postings.subscribe(doc_id, &term_buffer);
+                                    multifield_postings.subscribe(doc_id, term_buffer);
                                 unordered_term_id_opt = Some(unordered_term_id);
                             });
                         if let Some(unordered_term_id) = unordered_term_id_opt {
@@ -237,7 +252,7 @@ impl SegmentWriter {
                             .u64_value()
                             .ok_or_else(make_schema_error)?;
                         term_buffer.set_u64(u64_val);
-                        multifield_postings.subscribe(doc_id, &term_buffer);
+                        multifield_postings.subscribe(doc_id, term_buffer);
                     }
                 }
                 FieldType::Date(_) => {
@@ -248,7 +263,7 @@ impl SegmentWriter {
                             .date_value()
                             .ok_or_else(make_schema_error)?;
                         term_buffer.set_i64(date_val.timestamp());
-                        multifield_postings.subscribe(doc_id, &term_buffer);
+                        multifield_postings.subscribe(doc_id, term_buffer);
                     }
                 }
                 FieldType::I64(_) => {
@@ -259,7 +274,7 @@ impl SegmentWriter {
                             .i64_value()
                             .ok_or_else(make_schema_error)?;
                         term_buffer.set_i64(i64_val);
-                        multifield_postings.subscribe(doc_id, &term_buffer);
+                        multifield_postings.subscribe(doc_id, term_buffer);
                     }
                 }
                 FieldType::F64(_) => {
@@ -270,7 +285,7 @@ impl SegmentWriter {
                             .f64_value()
                             .ok_or_else(make_schema_error)?;
                         term_buffer.set_f64(f64_val);
-                        multifield_postings.subscribe(doc_id, &term_buffer);
+                        multifield_postings.subscribe(doc_id, term_buffer);
                     }
                 }
                 FieldType::Bytes(_) => {
@@ -281,7 +296,7 @@ impl SegmentWriter {
                             .bytes_value()
                             .ok_or_else(make_schema_error)?;
                         term_buffer.set_bytes(bytes);
-                        self.multifield_postings.subscribe(doc_id, &term_buffer);
+                        self.multifield_postings.subscribe(doc_id, term_buffer);
                     }
                 }
             }
@@ -315,8 +330,12 @@ impl SegmentWriter {
     }
 }
 
-// This method is used as a trick to workaround the borrow checker
-fn write(
+/// This method is used as a trick to workaround the borrow checker
+/// Writes a view of a segment by pushing information
+/// to the `SegmentSerializer`.
+///
+/// `doc_id_map` is used to map to the new doc_id order.
+fn remap_and_write(
     multifield_postings: &MultiFieldPostingsWriter,
     fast_field_writers: &FastFieldsWriter,
     fieldnorms_writer: &FieldNormsWriter,
@@ -340,6 +359,7 @@ fn write(
         &term_ord_map,
         doc_id_map,
     )?;
+
     // finalize temp docstore and create version, which reflects the doc_id_map
     if let Some(doc_id_map) = doc_id_map {
         let store_write = serializer
@@ -356,31 +376,16 @@ fn write(
                 .segment()
                 .open_read(SegmentComponent::TempStore)?,
         )?;
+
         for old_doc_id in doc_id_map.iter_old_doc_ids() {
-            let doc_bytes = store_read.get_document_bytes(*old_doc_id)?;
+            let doc_bytes = store_read.get_document_bytes(old_doc_id)?;
             serializer.get_store_writer().store_bytes(&doc_bytes)?;
         }
     }
-    serializer.close()?;
-    Ok(())
-}
 
-impl SerializableSegment for SegmentWriter {
-    fn write(
-        &self,
-        serializer: SegmentSerializer,
-        doc_id_map: Option<&DocIdMapping>,
-    ) -> crate::Result<u32> {
-        let max_doc = self.max_doc;
-        write(
-            &self.multifield_postings,
-            &self.fast_field_writers,
-            &self.fieldnorms_writer,
-            serializer,
-            doc_id_map,
-        )?;
-        Ok(max_doc)
-    }
+    serializer.close()?;
+
+    Ok(())
 }
 
 #[cfg(test)]

src/lib.rs

@@ -1,6 +1,13 @@
 #![doc(html_logo_url = "http://fulmicoton.com/tantivy-logo/tantivy-logo.png")]
 #![cfg_attr(all(feature = "unstable", test), feature(test))]
-#![cfg_attr(feature = "cargo-clippy", allow(clippy::module_inception))]
+#![cfg_attr(
+    feature = "cargo-clippy",
+    allow(
+        clippy::module_inception,
+        clippy::needless_range_loop,
+        clippy::bool_assert_comparison
+    )
+)]
 #![doc(test(attr(allow(unused_variables), deny(warnings))))]
 #![warn(missing_docs)]
 
@@ -100,7 +107,6 @@
 
 #[cfg_attr(test, macro_use)]
 extern crate serde_json;
-
 #[macro_use]
 extern crate log;
 
@@ -934,7 +940,7 @@ mod tests {
         let id = schema_builder.add_u64_field("id", INDEXED);
         let schema = schema_builder.build();
 
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
         let index_reader = index.reader()?;
 
         let mut index_writer = index.writer_for_tests()?;
@@ -973,7 +979,7 @@ mod tests {
         let searcher = index_reader.searcher();
         let segment_ids: Vec<SegmentId> = searcher
             .segment_readers()
-            .into_iter()
+            .iter()
             .map(|reader| reader.segment_id())
             .collect();
         block_on(index_writer.merge(&segment_ids)).unwrap();

src/positions/mod.rs

@@ -46,7 +46,7 @@ pub mod tests {
     fn create_positions_data(vals: &[u32]) -> crate::Result<OwnedBytes> {
         let mut positions_buffer = vec![];
         let mut serializer = PositionSerializer::new(&mut positions_buffer);
-        serializer.write_positions_delta(&vals);
+        serializer.write_positions_delta(vals);
         serializer.close_term()?;
         serializer.close()?;
         Ok(OwnedBytes::new(positions_buffer))
@@ -169,7 +169,7 @@ pub mod tests {
         let positions_delta: Vec<u32> = (0..2_000_000).collect();
         let positions_data = create_positions_data(&positions_delta[..])?;
         assert_eq!(positions_data.len(), 5003499);
-        let mut position_reader = PositionReader::open(positions_data.clone())?;
+        let mut position_reader = PositionReader::open(positions_data)?;
         let mut buf = [0u32; 256];
         position_reader.read(128, &mut buf);
         for i in 0..256 {

src/postings/block_search.rs

@@ -57,7 +57,7 @@ mod sse2 {
         fn test_linear_search_sse2_128_u32() {
             let mut block = [0u32; COMPRESSION_BLOCK_SIZE];
             for el in 0u32..128u32 {
-                block[el as usize] = el * 2 + 1 << 18;
+                block[el as usize] = (el * 2 + 1) << 18;
             }
             let target = block[64] + 1;
             assert_eq!(linear_search_sse2_128(&AlignedBuffer(block), target), 65);
@@ -91,7 +91,7 @@ fn exponential_search(arr: &[u32], target: u32) -> Range<usize> {
 
 #[inline(never)]
 fn galloping(block_docs: &[u32], target: u32) -> usize {
-    let range = exponential_search(&block_docs, target);
+    let range = exponential_search(block_docs, target);
     range.start + linear_search(&block_docs[range], target)
 }
 

src/postings/block_segment_postings.rs

@@ -13,11 +13,7 @@ use crate::schema::IndexRecordOption;
 use crate::{DocId, Score, TERMINATED};
 
 fn max_score<I: Iterator<Item = Score>>(mut it: I) -> Option<Score> {
-    if let Some(first) = it.next() {
-        Some(it.fold(first, Score::max))
-    } else {
-        None
-    }
+    it.next().map(|first| it.fold(first, Score::max))
 }
 
 /// `BlockSegmentPostings` is a cursor iterating over blocks
@@ -482,11 +478,11 @@ mod tests {
             docs.push((i * i / 100) + i);
         }
         let mut block_postings = build_block_postings(&docs[..]);
-        for i in vec![0, 424, 10000] {
-            block_postings.seek(i);
+        for i in &[0, 424, 10000] {
+            block_postings.seek(*i);
             let docs = block_postings.docs();
-            assert!(docs[0] <= i);
-            assert!(docs.last().cloned().unwrap_or(0u32) >= i);
+            assert!(docs[0] <= *i);
+            assert!(docs.last().cloned().unwrap_or(0u32) >= *i);
         }
         block_postings.seek(100_000);
         assert_eq!(block_postings.doc(COMPRESSION_BLOCK_SIZE - 1), TERMINATED);

src/postings/compression/mod.rs

@@ -85,13 +85,13 @@ impl BlockDecoder {
     ) -> usize {
         self.output_len = COMPRESSION_BLOCK_SIZE;
         self.bitpacker
-            .decompress_sorted(offset, &compressed_data, &mut self.output.0, num_bits)
+            .decompress_sorted(offset, compressed_data, &mut self.output.0, num_bits)
     }
 
     pub fn uncompress_block_unsorted(&mut self, compressed_data: &[u8], num_bits: u8) -> usize {
         self.output_len = COMPRESSION_BLOCK_SIZE;
         self.bitpacker
-            .decompress(&compressed_data, &mut self.output.0, num_bits)
+            .decompress(compressed_data, &mut self.output.0, num_bits)
     }
 
     #[inline]
@@ -303,7 +303,7 @@ pub mod tests {
             assert!(encoded_data.len() <= expected_length);
             let mut decoder = BlockDecoder::default();
             let consumed_num_bytes =
-                decoder.uncompress_vint_sorted(&encoded_data, *offset, input.len(), PADDING_VALUE);
+                decoder.uncompress_vint_sorted(encoded_data, *offset, input.len(), PADDING_VALUE);
             assert_eq!(consumed_num_bytes, encoded_data.len());
             assert_eq!(input, decoder.output_array());
             for i in input.len()..COMPRESSION_BLOCK_SIZE {
@@ -354,18 +354,18 @@ mod bench {
         });
     }
 
-    #[test]
-    fn test_all_docs_compression_numbits() {
-        for expected_num_bits in 0u8.. {
-            let mut data = [0u32; 128];
-            if expected_num_bits > 0 {
-                data[0] = (1u64 << (expected_num_bits as usize) - 1) as u32;
-            }
-            let mut encoder = BlockEncoder::new();
-            let (num_bits, compressed) = encoder.compress_block_unsorted(&data);
-            assert_eq!(compressed.len(), compressed_block_size(num_bits));
-        }
-    }
+    //#[test]
+    //fn test_all_docs_compression_numbits() {
+    //for expected_num_bits in 0u8.. {
+    //let mut data = [0u32; 128];
+    //if expected_num_bits > 0 {
+    //data[0] = (1u64 << (expected_num_bits as usize) - 1) as u32;
+    //}
+    //let mut encoder = BlockEncoder::new();
+    //let (num_bits, compressed) = encoder.compress_block_unsorted(&data);
+    //assert_eq!(compressed.len(), compressed_block_size(num_bits));
+    //}
+    //}
 
     const NUM_INTS_BENCH_VINT: usize = 10;
 

src/postings/mod.rs

@@ -54,7 +54,7 @@ pub mod tests {
     use crate::DocId;
     use crate::HasLen;
     use crate::Score;
-    use std::{iter, mem};
+    use std::mem;
 
     #[test]
     pub fn test_position_write() -> crate::Result<()> {
@@ -153,8 +153,8 @@ pub mod tests {
 
     #[test]
     pub fn test_drop_token_that_are_too_long() -> crate::Result<()> {
-        let ok_token_text: String = iter::repeat('A').take(MAX_TOKEN_LEN).collect();
-        let mut exceeding_token_text: String = iter::repeat('A').take(MAX_TOKEN_LEN + 1).collect();
+        let ok_token_text: String = "A".repeat(MAX_TOKEN_LEN);
+        let mut exceeding_token_text: String = "A".repeat(MAX_TOKEN_LEN + 1);
         exceeding_token_text.push_str(" hello");
         let mut schema_builder = Schema::builder();
         let text_options = TextOptions::default().set_indexing_options(
@@ -164,7 +164,7 @@ pub mod tests {
         );
         let text_field = schema_builder.add_text_field("text", text_options);
         let schema = schema_builder.build();
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
         index
             .tokenizers()
             .register("simple_no_truncation", SimpleTokenizer);
@@ -229,7 +229,7 @@ pub mod tests {
                 segment_writer.add_document(op, &schema).unwrap();
             }
             for i in 2..1000 {
-                let mut text: String = iter::repeat("e ").take(i).collect();
+                let mut text: String = "e ".repeat(i);
                 text.push_str(" a");
                 let op = AddOperation {
                     opstamp: 2u64,
@@ -409,7 +409,7 @@ pub mod tests {
                 .unwrap();
 
             for i in 0..num_docs / 2 - 1 {
-                assert!(segment_postings.seek(i * 2 + 1) > (i * 1) * 2);
+                assert!(segment_postings.seek(i * 2 + 1) > i * 2);
                 assert_eq!(segment_postings.doc(), (i + 1) * 2);
             }
         }
@@ -431,13 +431,10 @@ pub mod tests {
                 .read_postings(&term_2, IndexRecordOption::Basic)?
                 .unwrap();
 
+            assert_eq!(segment_postings.seek(i), i);
+            assert_eq!(segment_postings.doc(), i);
             if i % 2 == 0 {
-                assert_eq!(segment_postings.seek(i), i);
-                assert_eq!(segment_postings.doc(), i);
                 assert!(segment_reader.is_deleted(i));
-            } else {
-                assert_eq!(segment_postings.seek(i), i);
-                assert_eq!(segment_postings.doc(), i);
             }
         }
 

src/postings/postings_writer.rs

@@ -231,13 +231,7 @@ pub trait PostingsWriter {
             // We skip all tokens with a len greater than u16.
             if token.text.len() <= MAX_TOKEN_LEN {
                 term_buffer.set_text(token.text.as_str());
-                self.subscribe(
-                    term_index,
-                    doc_id,
-                    token.position as u32,
-                    &term_buffer,
-                    heap,
-                );
+                self.subscribe(term_index, doc_id, token.position as u32, term_buffer, heap);
             } else {
                 warn!(
                     "A token exceeding MAX_TOKEN_LEN ({}>{}) was dropped. Search for \

src/postings/recorder.rs

@@ -282,7 +282,7 @@ impl Recorder for TfAndPositionRecorder {
                 doc_id_and_positions
                     .push((doc_id_map.get_new_doc_id(doc), buffer_positions.to_vec()));
             } else {
-                serializer.write_doc(doc, buffer_positions.len() as u32, &buffer_positions);
+                serializer.write_doc(doc, buffer_positions.len() as u32, buffer_positions);
             }
         }
         if doc_id_map.is_some() {

src/postings/segment_postings.rs

@@ -107,7 +107,7 @@ impl SegmentPostings {
         let fieldnorm_reader = fieldnorms.map(FieldNormReader::for_test);
         let average_field_norm = fieldnorms
             .map(|fieldnorms| {
-                if fieldnorms.len() == 0 {
+                if fieldnorms.is_empty() {
                     return 0.0;
                 }
                 let total_num_tokens: u64 = fieldnorms
@@ -184,7 +184,7 @@ impl DocSet for SegmentPostings {
 
         // At this point we are on the block, that might contain our document.
         let output = self.block_cursor.docs_aligned();
-        self.cur = self.block_searcher.search_in_block(&output, target);
+        self.cur = self.block_searcher.search_in_block(output, target);
 
         // The last block is not full and padded with the value TERMINATED,
         // so that we are guaranteed to have at least doc in the block (a real one or the padding)

src/postings/serializer.rs

@@ -356,7 +356,7 @@ impl<W: Write> PostingsSerializer<W> {
             // encode the doc ids
             let (num_bits, block_encoded): (u8, &[u8]) = self
                 .block_encoder
-                .compress_block_sorted(&self.block.doc_ids(), self.last_doc_id_encoded);
+                .compress_block_sorted(self.block.doc_ids(), self.last_doc_id_encoded);
             self.last_doc_id_encoded = self.block.last_doc();
             self.skip_write
                 .write_doc(self.last_doc_id_encoded, num_bits);
@@ -366,7 +366,7 @@ impl<W: Write> PostingsSerializer<W> {
         if self.mode.has_freq() {
             let (num_bits, block_encoded): (u8, &[u8]) = self
                 .block_encoder
-                .compress_block_unsorted(&self.block.term_freqs());
+                .compress_block_unsorted(self.block.term_freqs());
             self.postings_write.extend(block_encoded);
             self.skip_write.write_term_freq(num_bits);
             if self.mode.has_positions() {
@@ -426,7 +426,7 @@ impl<W: Write> PostingsSerializer<W> {
             {
                 let block_encoded = self
                     .block_encoder
-                    .compress_vint_sorted(&self.block.doc_ids(), self.last_doc_id_encoded);
+                    .compress_vint_sorted(self.block.doc_ids(), self.last_doc_id_encoded);
                 self.postings_write.write_all(block_encoded)?;
             }
             // ... Idem for term frequencies

src/postings/stacker/expull.rs

@@ -280,11 +280,8 @@ mod tests {
             if v.len() >= 10 {
                 break;
             }
-            match len_to_capacity(i) {
-                CapacityResult::NeedAlloc(cap) => {
-                    v.push((i, cap));
-                }
-                _ => {}
+            if let CapacityResult::NeedAlloc(cap) = len_to_capacity(i) {
+                v.push((i, cap));
             }
         }
         assert_eq!(

src/postings/stacker/term_hashmap.rs

@@ -151,7 +151,7 @@ impl TermHashMap {
     pub fn iter(&self) -> Iter<'_> {
         Iter {
             inner: self.occupied.iter(),
-            hashmap: &self,
+            hashmap: self,
         }
     }
 
@@ -261,8 +261,8 @@ mod tests {
         }
 
         let mut vanilla_hash_map = HashMap::new();
-        let mut iter_values = hash_map.iter();
-        while let Some((key, addr, _)) = iter_values.next() {
+        let iter_values = hash_map.iter();
+        for (key, addr, _) in iter_values {
             let val: u32 = hash_map.heap.read(addr);
             vanilla_hash_map.insert(key.to_owned(), val);
         }

src/query/boolean_query/block_wand.rs

@@ -251,7 +251,7 @@ mod tests {
 
     impl PartialEq for Float {
         fn eq(&self, other: &Self) -> bool {
-            self.cmp(&other) == Ordering::Equal
+            self.cmp(other) == Ordering::Equal
         }
     }
 
@@ -289,7 +289,7 @@ mod tests {
             if !nearly_equals(score, limit) {
                 checkpoints.push((doc, score));
             }
-            return limit;
+            limit
         });
         checkpoints
     }
@@ -368,10 +368,10 @@ mod tests {
             .iter()
             .map(|posting_list| {
                 posting_list
-                    .into_iter()
+                    .iter()
                     .cloned()
                     .flat_map(|(doc, term_freq)| {
-                        (0 as u32..REPEAT as u32).map(move |offset| {
+                        (0_u32..REPEAT as u32).map(move |offset| {
                             (
                                 doc * (REPEAT as u32) + offset,
                                 if offset == 0 { term_freq } else { 1 },

src/query/more_like_this/more_like_this.rs

@@ -329,7 +329,7 @@ impl MoreLikeThis {
                 continue;
             }
 
-            let doc_freq = searcher.doc_freq(&term)?;
+            let doc_freq = searcher.doc_freq(term)?;
 
             // ignore terms with less than min_doc_frequency
             if self

src/query/more_like_this/query.rs

@@ -260,7 +260,7 @@ mod tests {
             .with_document(DocAddress::new(0, 0));
         let top_docs = searcher.search(&query, &TopDocs::with_limit(5)).unwrap();
         let mut doc_ids: Vec<_> = top_docs.iter().map(|item| item.1.doc_id).collect();
-        doc_ids.sort();
+        doc_ids.sort_unstable();
 
         assert_eq!(doc_ids.len(), 3);
         assert_eq!(doc_ids, vec![0, 1, 3]);
@@ -276,7 +276,7 @@ mod tests {
             .with_document(DocAddress::new(0, 4));
         let top_docs = searcher.search(&query, &TopDocs::with_limit(5)).unwrap();
         let mut doc_ids: Vec<_> = top_docs.iter().map(|item| item.1.doc_id).collect();
-        doc_ids.sort();
+        doc_ids.sort_unstable();
 
         assert_eq!(doc_ids.len(), 2);
         assert_eq!(doc_ids, vec![3, 4]);

src/query/phrase_query/phrase_weight.rs

@@ -48,43 +48,36 @@ impl PhraseWeight {
     ) -> crate::Result<Option<PhraseScorer<SegmentPostings>>> {
         let similarity_weight = self.similarity_weight.boost_by(boost);
         let fieldnorm_reader = self.fieldnorm_reader(reader)?;
+        let mut term_postings_list = Vec::new();
         if reader.has_deletes() {
-            let mut term_postings_list = Vec::new();
             for &(offset, ref term) in &self.phrase_terms {
                 if let Some(postings) = reader
                     .inverted_index(term.field())?
-                    .read_postings(&term, IndexRecordOption::WithFreqsAndPositions)?
+                    .read_postings(term, IndexRecordOption::WithFreqsAndPositions)?
                 {
                     term_postings_list.push((offset, postings));
                 } else {
                     return Ok(None);
                 }
             }
-            Ok(Some(PhraseScorer::new(
-                term_postings_list,
-                similarity_weight,
-                fieldnorm_reader,
-                self.scoring_enabled,
-            )))
         } else {
-            let mut term_postings_list = Vec::new();
             for &(offset, ref term) in &self.phrase_terms {
                 if let Some(postings) = reader
                     .inverted_index(term.field())?
-                    .read_postings_no_deletes(&term, IndexRecordOption::WithFreqsAndPositions)?
+                    .read_postings_no_deletes(term, IndexRecordOption::WithFreqsAndPositions)?
                 {
                     term_postings_list.push((offset, postings));
                 } else {
                     return Ok(None);
                 }
             }
-            Ok(Some(PhraseScorer::new(
-                term_postings_list,
-                similarity_weight,
-                fieldnorm_reader,
-                self.scoring_enabled,
-            )))
         }
+        Ok(Some(PhraseScorer::new(
+            term_postings_list,
+            similarity_weight,
+            fieldnorm_reader,
+            self.scoring_enabled,
+        )))
     }
 }
 

src/query/query_parser/query_parser.rs

@@ -722,7 +722,7 @@ mod test {
         let is_not_indexed_err = |query: &str| {
             let result: Result<Box<dyn Query>, QueryParserError> = query_parser.parse_query(query);
             if let Err(QueryParserError::FieldNotIndexed(field_name)) = result {
-                Some(field_name.clone())
+                Some(field_name)
             } else {
                 None
             }

src/query/range_query.rs

@@ -91,8 +91,8 @@ impl RangeQuery {
         RangeQuery {
             field,
             value_type,
-            left_bound: map_bound(&left_bound, &verify_and_unwrap_term),
-            right_bound: map_bound(&right_bound, &verify_and_unwrap_term),
+            left_bound: map_bound(left_bound, &verify_and_unwrap_term),
+            right_bound: map_bound(right_bound, &verify_and_unwrap_term),
         }
     }
 
@@ -493,7 +493,7 @@ mod tests {
         let year = schema.get_field("year").unwrap();
         index_writer.add_document(doc!(
           title => "hemoglobin blood",
-          year => 1990 as i64
+          year => 1990_i64
         ));
         index_writer.commit()?;
         let reader = index.reader()?;

src/query/regex_query.rs

@@ -58,7 +58,7 @@ pub struct RegexQuery {
 impl RegexQuery {
     /// Creates a new RegexQuery from a given pattern
     pub fn from_pattern(regex_pattern: &str, field: Field) -> crate::Result<Self> {
-        let regex = Regex::new(&regex_pattern)
+        let regex = Regex::new(regex_pattern)
             .map_err(|_| TantivyError::InvalidArgument(regex_pattern.to_string()))?;
         Ok(RegexQuery::from_regex(regex, field))
     }
@@ -169,9 +169,9 @@ mod test {
 
         verify_regex_query(matching_one, matching_zero, reader.clone());
 
-        let matching_one = RegexQuery::from_regex(r1.clone(), field);
-        let matching_zero = RegexQuery::from_regex(r2.clone(), field);
+        let matching_one = RegexQuery::from_regex(r1, field);
+        let matching_zero = RegexQuery::from_regex(r2, field);
 
-        verify_regex_query(matching_one, matching_zero, reader.clone());
+        verify_regex_query(matching_one, matching_zero, reader);
     }
 }

src/query/term_query/term_scorer.rs

@@ -194,8 +194,8 @@ mod tests {
                   .collect();
 
          let mut fieldnorms: Vec<u32> = vec![];
-         for i in 0..term_doc_freq {
-             let (tf, num_extra_terms) = term_freqs_fieldnorms[i];
+         for item in term_freqs_fieldnorms.iter().take(term_doc_freq) {
+             let (tf, num_extra_terms) = item;
              fieldnorms.push(tf + num_extra_terms);
          }
          let average_fieldnorm = fieldnorms
@@ -253,7 +253,7 @@ mod tests {
     }
 
     fn test_block_wand_aux(term_query: &TermQuery, searcher: &Searcher) -> crate::Result<()> {
-        let term_weight = term_query.specialized_weight(&searcher, true)?;
+        let term_weight = term_query.specialized_weight(searcher, true)?;
         for reader in searcher.segment_readers() {
             let mut block_max_scores = vec![];
             let mut block_max_scores_b = vec![];
@@ -309,7 +309,7 @@ mod tests {
         }
         writer.commit()?;
         let term_query = TermQuery::new(
-            Term::from_field_text(text_field, &"bbbb"),
+            Term::from_field_text(text_field, "bbbb"),
             IndexRecordOption::WithFreqs,
         );
         let segment_ids: Vec<SegmentId>;

src/query/union.rs

@@ -319,7 +319,7 @@ mod tests {
             let res: Box<dyn DocSet> = Box::new(Union::<_, DoNothingCombiner>::from(
                 docs_list
                     .iter()
-                    .map(|docs| docs.clone())
+                    .cloned()
                     .map(VecDocSet::from)
                     .map(|docset| ConstScorer::new(docset, 1.0))
                     .collect::<Vec<_>>(),

src/reader/pool.rs

@@ -253,7 +253,7 @@ mod tests {
 
         threads.push(thread::spawn(move || {
             assert_eq!(start_1_recv.recv(), Ok("start"));
-            let _leased_searcher = &pool1.acquire();
+            let _leased_searcher = pool1.acquire();
             assert!(event_send1.send("1 acquired").is_ok());
             assert_eq!(start_1_recv.recv(), Ok("stop"));
             assert!(event_send1.send("1 stopped").is_ok());
@@ -262,7 +262,7 @@ mod tests {
 
         threads.push(thread::spawn(move || {
             assert_eq!(start_2_recv.recv(), Ok("start"));
-            let _leased_searcher = &pool2.acquire();
+            let _leased_searcher = pool2.acquire();
             assert!(event_send2.send("2 acquired").is_ok());
             assert_eq!(start_2_recv.recv(), Ok("stop"));
             mem::drop(_leased_searcher);
@@ -271,7 +271,7 @@ mod tests {
 
         threads.push(thread::spawn(move || {
             assert_eq!(start_3_recv.recv(), Ok("start"));
-            let _leased_searcher = &pool3.acquire();
+            let _leased_searcher = pool3.acquire();
             assert!(event_send3.send("3 acquired").is_ok());
             assert_eq!(start_3_recv.recv(), Ok("stop"));
             mem::drop(_leased_searcher);

src/schema/facet.rs

@@ -39,7 +39,7 @@ pub enum FacetParseError {
 /// belonging to a facet also belongs to the ancestor of
 /// its facet. In the example above, `/electronics/tv_and_video/`
 /// and `/electronics`.
-#[derive(Clone, Eq, Hash, PartialEq, Ord, PartialOrd)]
+#[derive(Clone, Default, Eq, Hash, PartialEq, Ord, PartialOrd)]
 pub struct Facet(String);
 
 impl Facet {

src/schema/field_entry.rs

@@ -23,6 +23,15 @@ pub struct FieldEntry {
 }
 
 impl FieldEntry {
+    /// Creates a new field entry given a name and a field type
+    pub fn new(field_name: String, field_type: FieldType) -> FieldEntry {
+        assert!(is_valid_field_name(&field_name));
+        FieldEntry {
+            name: field_name,
+            field_type,
+        }
+    }
+
     /// Creates a new u64 field entry in the schema, given
     /// a name, and some options.
     pub fn new_text(field_name: String, text_options: TextOptions) -> FieldEntry {
@@ -308,7 +317,7 @@ mod tests {
         assert_eq!("title", field_value.name);
 
         match field_value.field_type {
-            FieldType::Str(_) => assert!(true),
+            FieldType::Str(_) => {}
             _ => panic!("expected FieldType::Str"),
         }
     }

src/schema/mod.rs

@@ -151,20 +151,13 @@ pub use self::flags::{FAST, INDEXED, STORED};
 pub use self::int_options::Cardinality;
 pub use self::int_options::IntOptions;
 
-use once_cell::sync::Lazy;
-use regex::Regex;
-
-/// Regular expression representing the restriction on a valid field names.
-pub const FIELD_NAME_PATTERN: &str = r#"^[_a-zA-Z][_\-a-zA-Z0-9]*$"#;
-
 /// Validator for a potential `field_name`.
-/// Returns true iff the name can be use for a field name.
+/// Returns true if the name can be use for a field name.
 ///
-/// A field name must start by a letter `[a-zA-Z]`.
-/// The other characters can be any alphanumic character `[a-ZA-Z0-9]` or `_`.
+/// A field name can be any character, must have at least one character
+/// and must not start with a `-`.
 pub fn is_valid_field_name(field_name: &str) -> bool {
-    static FIELD_NAME_PTN: Lazy<Regex> = Lazy::new(|| Regex::new(FIELD_NAME_PATTERN).unwrap());
-    FIELD_NAME_PTN.is_match(field_name)
+    field_name.len() > 0 && !field_name.starts_with('-')
 }
 
 #[cfg(test)]
@@ -174,16 +167,8 @@ mod tests {
 
     #[test]
     fn test_is_valid_name() {
-        assert!(is_valid_field_name("text"));
-        assert!(is_valid_field_name("text0"));
-        assert!(!is_valid_field_name("0text"));
-        assert!(is_valid_field_name("field-name"));
-        assert!(is_valid_field_name("field_name"));
-        assert!(!is_valid_field_name("field!name"));
+        assert!(is_valid_field_name("シャボン玉"));
         assert!(!is_valid_field_name("-fieldname"));
-        assert!(is_valid_field_name("_fieldname"));
         assert!(!is_valid_field_name(""));
-        assert!(!is_valid_field_name("シャボン玉"));
-        assert!(is_valid_field_name("my_text_field"));
     }
 }

src/schema/schema.rs

@@ -779,7 +779,7 @@ mod tests {
   }
 ]"#;
         let tmp_schema: Schema =
-            serde_json::from_str(&schema_content).expect("error while reading json");
+            serde_json::from_str(schema_content).expect("error while reading json");
         for (_field, field_entry) in tmp_schema.fields() {
             schema_builder.add_field(field_entry.clone());
         }

src/snippet/mod.rs

@@ -137,7 +137,7 @@ fn search_fragments<'a>(
             };
             fragment = FragmentCandidate::new(next.offset_from);
         }
-        fragment.try_add_token(next, &terms);
+        fragment.try_add_token(next, terms);
     }
     if fragment.score > 0.0 {
         fragments.push(fragment)
@@ -286,8 +286,8 @@ impl SnippetGenerator {
     /// Generates a snippet for the given text.
     pub fn snippet(&self, text: &str) -> Snippet {
         let fragment_candidates =
-            search_fragments(&self.tokenizer, &text, &self.terms_text, self.max_num_chars);
-        select_best_fragment_combination(&fragment_candidates[..], &text)
+            search_fragments(&self.tokenizer, text, &self.terms_text, self.max_num_chars);
+        select_best_fragment_combination(&fragment_candidates[..], text)
     }
 }
 
@@ -301,9 +301,8 @@ mod tests {
     use crate::SnippetGenerator;
     use maplit::btreemap;
     use std::collections::BTreeMap;
-    use std::iter::Iterator;
 
-    const TEST_TEXT: &'static str = r#"Rust is a systems programming language sponsored by
+    const TEST_TEXT: &str = r#"Rust is a systems programming language sponsored by
 Mozilla which describes it as a "safe, concurrent, practical language", supporting functional and
 imperative-procedural paradigms. Rust is syntactically similar to C++[according to whom?],
 but its designers intend it to provide better memory safety while still maintaining
@@ -330,7 +329,7 @@ Survey in 2016, 2017, and 2018."#;
             assert_eq!(first.score, 1.9);
             assert_eq!(first.stop_offset, 89);
         }
-        let snippet = select_best_fragment_combination(&fragments[..], &TEST_TEXT);
+        let snippet = select_best_fragment_combination(&fragments[..], TEST_TEXT);
         assert_eq!(
             snippet.fragments,
             "Rust is a systems programming language sponsored by\n\
@@ -356,7 +355,7 @@ Survey in 2016, 2017, and 2018."#;
                 assert_eq!(first.score, 1.0);
                 assert_eq!(first.stop_offset, 17);
             }
-            let snippet = select_best_fragment_combination(&fragments[..], &TEST_TEXT);
+            let snippet = select_best_fragment_combination(&fragments[..], TEST_TEXT);
             assert_eq!(snippet.to_html(), "<b>Rust</b> is a systems")
         }
         {
@@ -371,7 +370,7 @@ Survey in 2016, 2017, and 2018."#;
                 assert_eq!(first.score, 0.9);
                 assert_eq!(first.stop_offset, 17);
             }
-            let snippet = select_best_fragment_combination(&fragments[..], &TEST_TEXT);
+            let snippet = select_best_fragment_combination(&fragments[..], TEST_TEXT);
             assert_eq!(snippet.to_html(), "programming <b>language</b>")
         }
     }
@@ -383,17 +382,17 @@ Survey in 2016, 2017, and 2018."#;
         let mut terms = BTreeMap::new();
         terms.insert(String::from("c"), 1.0);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), text, &terms, 3);
 
         assert_eq!(fragments.len(), 1);
         {
-            let first = fragments.iter().nth(0).unwrap();
+            let first = &fragments[0];
             assert_eq!(first.score, 1.0);
             assert_eq!(first.start_offset, 4);
             assert_eq!(first.stop_offset, 7);
         }
 
-        let snippet = select_best_fragment_combination(&fragments[..], &text);
+        let snippet = select_best_fragment_combination(&fragments[..], text);
         assert_eq!(snippet.fragments, "c d");
         assert_eq!(snippet.to_html(), "<b>c</b> d");
     }
@@ -405,17 +404,17 @@ Survey in 2016, 2017, and 2018."#;
         let mut terms = BTreeMap::new();
         terms.insert(String::from("f"), 1.0);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), text, &terms, 3);
 
         assert_eq!(fragments.len(), 2);
         {
-            let first = fragments.iter().nth(0).unwrap();
+            let first = &fragments[0];
             assert_eq!(first.score, 1.0);
             assert_eq!(first.stop_offset, 11);
             assert_eq!(first.start_offset, 8);
         }
 
-        let snippet = select_best_fragment_combination(&fragments[..], &text);
+        let snippet = select_best_fragment_combination(&fragments[..], text);
         assert_eq!(snippet.fragments, "e f");
         assert_eq!(snippet.to_html(), "e <b>f</b>");
     }
@@ -428,17 +427,17 @@ Survey in 2016, 2017, and 2018."#;
         terms.insert(String::from("f"), 1.0);
         terms.insert(String::from("a"), 0.9);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 7);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), text, &terms, 7);
 
         assert_eq!(fragments.len(), 2);
         {
-            let first = fragments.iter().nth(0).unwrap();
+            let first = &fragments[0];
             assert_eq!(first.score, 0.9);
             assert_eq!(first.stop_offset, 7);
             assert_eq!(first.start_offset, 0);
         }
 
-        let snippet = select_best_fragment_combination(&fragments[..], &text);
+        let snippet = select_best_fragment_combination(&fragments[..], text);
         assert_eq!(snippet.fragments, "e f g");
         assert_eq!(snippet.to_html(), "e <b>f</b> g");
     }
@@ -450,11 +449,11 @@ Survey in 2016, 2017, and 2018."#;
         let mut terms = BTreeMap::new();
         terms.insert(String::from("z"), 1.0);
 
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), text, &terms, 3);
 
         assert_eq!(fragments.len(), 0);
 
-        let snippet = select_best_fragment_combination(&fragments[..], &text);
+        let snippet = select_best_fragment_combination(&fragments[..], text);
         assert_eq!(snippet.fragments, "");
         assert_eq!(snippet.to_html(), "");
     }
@@ -464,10 +463,10 @@ Survey in 2016, 2017, and 2018."#;
         let text = "a b c d";
 
         let terms = BTreeMap::new();
-        let fragments = search_fragments(&From::from(SimpleTokenizer), &text, &terms, 3);
+        let fragments = search_fragments(&From::from(SimpleTokenizer), text, &terms, 3);
         assert_eq!(fragments.len(), 0);
 
-        let snippet = select_best_fragment_combination(&fragments[..], &text);
+        let snippet = select_best_fragment_combination(&fragments[..], text);
         assert_eq!(snippet.fragments, "");
         assert_eq!(snippet.to_html(), "");
     }

src/space_usage/mod.rs

@@ -296,7 +296,7 @@ mod test {
     #[test]
     fn test_empty() {
         let schema = Schema::builder().build();
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
         let reader = index.reader().unwrap();
         let searcher = reader.searcher();
         let searcher_space_usage = searcher.space_usage().unwrap();
@@ -325,7 +325,7 @@ mod test {
         let mut schema_builder = Schema::builder();
         let name = schema_builder.add_u64_field("name", FAST | INDEXED);
         let schema = schema_builder.build();
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
 
         {
             let mut index_writer = index.writer_for_tests().unwrap();
@@ -362,7 +362,7 @@ mod test {
         let mut schema_builder = Schema::builder();
         let name = schema_builder.add_text_field("name", TEXT);
         let schema = schema_builder.build();
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
 
         {
             let mut index_writer = index.writer_for_tests().unwrap();
@@ -401,7 +401,7 @@ mod test {
         let mut schema_builder = Schema::builder();
         let name = schema_builder.add_text_field("name", STORED);
         let schema = schema_builder.build();
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
 
         {
             let mut index_writer = index.writer_for_tests().unwrap();
@@ -439,7 +439,7 @@ mod test {
         let mut schema_builder = Schema::builder();
         let name = schema_builder.add_u64_field("name", INDEXED);
         let schema = schema_builder.build();
-        let index = Index::create_in_ram(schema.clone());
+        let index = Index::create_in_ram(schema);
 
         {
             let mut index_writer = index.writer_for_tests()?;

src/store/index/mod.rs

@@ -40,7 +40,7 @@ impl fmt::Debug for Checkpoint {
 #[cfg(test)]
 mod tests {
 
-    use std::{io, iter};
+    use std::io;
 
     use futures::executor::block_on;
     use proptest::strategy::{BoxedStrategy, Strategy};
@@ -134,9 +134,7 @@ mod tests {
         let index = Index::create_in_ram(schema);
         let mut index_writer = index.writer_for_tests()?;
         index_writer.set_merge_policy(Box::new(NoMergePolicy));
-        let long_text: String = iter::repeat("abcdefghijklmnopqrstuvwxyz")
-            .take(1_000)
-            .collect();
+        let long_text: String = "abcdefghijklmnopqrstuvwxyz".repeat(1_000);
         for _ in 0..20 {
             index_writer.add_document(doc!(body=>long_text.clone()));
         }
@@ -219,8 +217,7 @@ mod tests {
     ) -> Option<Checkpoint> {
         checkpoints
             .into_iter()
-            .filter(|checkpoint| checkpoint.doc_range.end > target)
-            .next()
+            .find(|checkpoint| checkpoint.doc_range.end > target)
     }
 
     fn test_skip_index_aux(skip_index: SkipIndex, checkpoints: &[Checkpoint]) {

src/store/reader.rs

@@ -163,7 +163,7 @@ 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_err(|e| e.kind())); // map error in order to enable cloning
+            .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;
@@ -177,7 +177,7 @@ impl StoreReader {
                     curr_checkpoint = checkpoint_block_iter.next();
                     curr_block = curr_checkpoint
                         .as_ref()
-                        .map(|checkpoint| self.read_block(&checkpoint).map_err(|e| e.kind()));
+                        .map(|checkpoint| self.read_block(checkpoint).map_err(|e| e.kind()));
                     reset_block_pos = true;
                     num_skipped = 0;
                 }