make Term::as_slice public

* fix doc store cache docs

addresses an issue reported in #1820

* rename doc_store_cache_size

Cargo.toml

@@ -55,6 +55,7 @@ measure_time = "0.8.2"
 async-trait = "0.1.53"
 arc-swap = "1.5.0"
 
+columnar = { version="0.1", path="./columnar", package ="tantivy-columnar" }
 sstable = { version="0.1", path="./sstable", package ="tantivy-sstable", optional = true }
 stacker = { version="0.1", path="./stacker", package ="tantivy-stacker" }
 tantivy-query-grammar = { version= "0.19.0", path="./query-grammar" }
@@ -107,7 +108,7 @@ unstable = [] # useful for benches.
 quickwit = ["sstable"]
 
 [workspace]
-members = ["query-grammar", "bitpacker", "common", "fastfield_codecs", "ownedbytes", "stacker", "sstable", "tokenizer-api"]
+members = ["query-grammar", "bitpacker", "common", "fastfield_codecs", "ownedbytes", "stacker", "sstable", "tokenizer-api", "columnar"]
 
 # Following the "fail" crate best practises, we isolate
 # tests that define specific behavior in fail check points

columnar/Cargo.toml

@@ -5,28 +5,23 @@ edition = "2021"
 license = "MIT"
 
 [dependencies]
+itertools = "0.10.5"
+log = "0.4.17"
+fnv = "1.0.7"
+fastdivide = "0.4.0"
+rand = { version = "0.8.5", optional = true }
+measure_time = { version = "0.8.2", optional = true }
+prettytable-rs = { version = "0.10.0", optional = true }
+
 stacker = { path = "../stacker", package="tantivy-stacker"}
-serde_json = "1"
-thiserror = "1"
-fnv = "1"
 sstable = { path = "../sstable", package = "tantivy-sstable" }
 common = { path = "../common", package = "tantivy-common" }
-itertools = "0.10"
-log = "0.4"
 tantivy-bitpacker = { version= "0.3", path = "../bitpacker/" }
-prettytable-rs = {version="0.10.0", optional= true}
-rand = {version="0.8.3", optional= true}
-fastdivide = "0.4"
-measure_time = { version="0.8.2", optional=true}
 
 [dev-dependencies]
-proptest = "1"
-more-asserts = "0.3.0"
-rand = "0.8.3"
-
-# temporary
-[workspace]
-members = []
+proptest = "1.0.0"
+more-asserts = "0.3.1"
+rand = "0.8.5"
 
 [features]
 unstable = []

columnar/src/column_index/optional_index/set_block/tests.rs

@@ -1,7 +1,8 @@
 use std::collections::HashMap;
 
-use crate::column_index::optional_index::set_block::dense::DENSE_BLOCK_NUM_BYTES;
-use crate::column_index::optional_index::set_block::{DenseBlockCodec, SparseBlockCodec};
+use crate::column_index::optional_index::set_block::{
+    DenseBlockCodec, SparseBlockCodec, DENSE_BLOCK_NUM_BYTES,
+};
 use crate::column_index::optional_index::{Set, SetCodec};
 
 fn test_set_helper<C: SetCodec<Item = u16>>(vals: &[u16]) -> usize {

columnar/src/columnar/column_type.rs

@@ -3,24 +3,22 @@ use std::net::Ipv6Addr;
 use crate::value::NumericalType;
 use crate::InvalidData;
 
-/// The column type represents the column type and can fit on 6-bits.
-///
-/// - bits[0..3]: Column category type.
-/// - bits[3..6]: Numerical type if necessary.
-#[derive(Hash, Eq, PartialEq, Debug, Clone, Copy)]
+/// The column type represents the column type.
+/// Any changes need to be propagated to `COLUMN_TYPES`.
+#[derive(Hash, Eq, PartialEq, Debug, Clone, Copy, Ord, PartialOrd)]
 #[repr(u8)]
 pub enum ColumnType {
     I64 = 0u8,
     U64 = 1u8,
     F64 = 2u8,
-    Bytes = 10u8,
-    Str = 14u8,
-    Bool = 18u8,
-    IpAddr = 22u8,
-    DateTime = 26u8,
+    Bytes = 3u8,
+    Str = 4u8,
+    Bool = 5u8,
+    IpAddr = 6u8,
+    DateTime = 7u8,
 }
 
-#[cfg(test)]
+// The order needs to match _exactly_ the order in the enum
 const COLUMN_TYPES: [ColumnType; 8] = [
     ColumnType::I64,
     ColumnType::U64,
@@ -38,18 +36,7 @@ impl ColumnType {
     }
 
     pub(crate) fn try_from_code(code: u8) -> Result<ColumnType, InvalidData> {
-        use ColumnType::*;
-        match code {
-            0u8 => Ok(I64),
-            1u8 => Ok(U64),
-            2u8 => Ok(F64),
-            10u8 => Ok(Bytes),
-            14u8 => Ok(Str),
-            18u8 => Ok(Bool),
-            22u8 => Ok(IpAddr),
-            26u8 => Ok(Self::DateTime),
-            _ => Err(InvalidData),
-        }
+        COLUMN_TYPES.get(code as usize).copied().ok_or(InvalidData)
     }
 }
 
@@ -64,18 +51,6 @@ impl From<NumericalType> for ColumnType {
 }
 
 impl ColumnType {
-    /// get column type category
-    pub(crate) fn column_type_category(self) -> ColumnTypeCategory {
-        match self {
-            ColumnType::I64 | ColumnType::U64 | ColumnType::F64 => ColumnTypeCategory::Numerical,
-            ColumnType::Bytes => ColumnTypeCategory::Bytes,
-            ColumnType::Str => ColumnTypeCategory::Str,
-            ColumnType::Bool => ColumnTypeCategory::Bool,
-            ColumnType::IpAddr => ColumnTypeCategory::IpAddr,
-            ColumnType::DateTime => ColumnTypeCategory::DateTime,
-        }
-    }
-
     pub fn numerical_type(&self) -> Option<NumericalType> {
         match self {
             ColumnType::I64 => Some(NumericalType::I64),
@@ -154,70 +129,20 @@ impl HasAssociatedColumnType for Ipv6Addr {
     }
 }
 
-/// Column types are grouped into different categories that
-/// corresponds to the different types of `JsonValue` types.
-///
-/// The columnar writer will apply coercion rules to make sure that
-/// at most one column exist per `ColumnTypeCategory`.
-///
-/// See also [README.md].
-#[derive(Copy, Clone, Ord, PartialOrd, Eq, PartialEq, Hash, Debug)]
-#[repr(u8)]
-pub enum ColumnTypeCategory {
-    Bool,
-    Str,
-    Numerical,
-    DateTime,
-    Bytes,
-    IpAddr,
-}
-
-impl From<ColumnType> for ColumnTypeCategory {
-    fn from(column_type: ColumnType) -> Self {
-        match column_type {
-            ColumnType::I64 => ColumnTypeCategory::Numerical,
-            ColumnType::U64 => ColumnTypeCategory::Numerical,
-            ColumnType::F64 => ColumnTypeCategory::Numerical,
-            ColumnType::Bytes => ColumnTypeCategory::Bytes,
-            ColumnType::Str => ColumnTypeCategory::Str,
-            ColumnType::Bool => ColumnTypeCategory::Bool,
-            ColumnType::IpAddr => ColumnTypeCategory::IpAddr,
-            ColumnType::DateTime => ColumnTypeCategory::DateTime,
-        }
-    }
-}
-
 #[cfg(test)]
 mod tests {
-    use std::collections::HashSet;
-
     use super::*;
     use crate::Cardinality;
 
     #[test]
     fn test_column_type_to_code() {
-        let mut column_type_set: HashSet<ColumnType> = HashSet::new();
-        for code in u8::MIN..=u8::MAX {
-            if let Ok(column_type) = ColumnType::try_from_code(code) {
-                assert_eq!(column_type.to_code(), code);
-                assert!(column_type_set.insert(column_type));
+        for (code, expected_column_type) in super::COLUMN_TYPES.iter().copied().enumerate() {
+            if let Ok(column_type) = ColumnType::try_from_code(code as u8) {
+                assert_eq!(column_type, expected_column_type);
             }
         }
-        assert_eq!(column_type_set.len(), super::COLUMN_TYPES.len());
-    }
-
-    #[test]
-    fn test_column_category_sort_consistent_with_column_type_sort() {
-        // This is a very important property because we
-        // we need to serialize colunmn in the right order.
-        let mut column_types: Vec<ColumnType> = super::COLUMN_TYPES.iter().copied().collect();
-        column_types.sort_by_key(|col| col.to_code());
-        let column_categories: Vec<ColumnTypeCategory> = column_types
-            .into_iter()
-            .map(ColumnTypeCategory::from)
-            .collect();
-        for (prev, next) in column_categories.iter().zip(column_categories.iter()) {
-            assert!(prev <= next);
+        for code in COLUMN_TYPES.len() as u8..=u8::MAX {
+            assert!(ColumnType::try_from_code(code as u8).is_err());
         }
     }
 

columnar/src/columnar/merge.rs

@@ -1,9 +1,9 @@
 use std::collections::HashMap;
 use std::io;
 
-use super::column_type::ColumnTypeCategory;
 use crate::columnar::ColumnarReader;
 use crate::dynamic_column::DynamicColumn;
+use crate::ColumnType;
 
 pub enum MergeDocOrder {
     /// Columnar tables are simply stacked one above the other.
@@ -35,7 +35,40 @@ pub fn merge_columnar(
     }
 }
 
-pub fn collect_columns(
+/// Column types are grouped into different categories.
+/// After merge, all columns belonging to the same category are coerced to
+/// the same column type.
+///
+/// In practise, today, only Numerical colummns are coerced into one type today.
+///
+/// See also [README.md].
+#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
+#[repr(u8)]
+enum ColumnTypeCategory {
+    Bool,
+    Str,
+    Numerical,
+    DateTime,
+    Bytes,
+    IpAddr,
+}
+
+impl From<ColumnType> for ColumnTypeCategory {
+    fn from(column_type: ColumnType) -> Self {
+        match column_type {
+            ColumnType::I64 => ColumnTypeCategory::Numerical,
+            ColumnType::U64 => ColumnTypeCategory::Numerical,
+            ColumnType::F64 => ColumnTypeCategory::Numerical,
+            ColumnType::Bytes => ColumnTypeCategory::Bytes,
+            ColumnType::Str => ColumnTypeCategory::Str,
+            ColumnType::Bool => ColumnTypeCategory::Bool,
+            ColumnType::IpAddr => ColumnTypeCategory::IpAddr,
+            ColumnType::DateTime => ColumnTypeCategory::DateTime,
+        }
+    }
+}
+
+fn collect_columns(
     columnar_readers: &[&ColumnarReader],
 ) -> io::Result<HashMap<String, HashMap<ColumnTypeCategory, Vec<DynamicColumn>>>> {
     // Each column name may have multiple types of column associated.
@@ -51,7 +84,7 @@ pub fn collect_columns(
                 .or_default();
 
             let columns = column_type_to_handles
-                .entry(handle.column_type().column_type_category())
+                .entry(handle.column_type().into())
                 .or_default();
             columns.push(handle.open()?);
         }
@@ -62,10 +95,9 @@ pub fn collect_columns(
     Ok(field_name_to_group)
 }
 
-/// Cast numerical type columns to the same type
-pub(crate) fn normalize_columns(
-    map: &mut HashMap<String, HashMap<ColumnTypeCategory, Vec<DynamicColumn>>>,
-) {
+/// Coerce numerical type columns to the same type
+/// TODO rename to `coerce_columns`
+fn normalize_columns(map: &mut HashMap<String, HashMap<ColumnTypeCategory, Vec<DynamicColumn>>>) {
     for (_field_name, type_category_to_columns) in map.iter_mut() {
         for (type_category, columns) in type_category_to_columns {
             if type_category == &ColumnTypeCategory::Numerical {

columnar/src/columnar/writer/column_writers.rs

@@ -184,10 +184,12 @@ impl CompatibleNumericalTypes {
 }
 
 impl NumericalColumnWriter {
-    pub fn column_type_and_cardinality(&self, num_docs: RowId) -> (NumericalType, Cardinality) {
-        let numerical_type = self.compatible_numerical_types.to_numerical_type();
-        let cardinality = self.column_writer.get_cardinality(num_docs);
-        (numerical_type, cardinality)
+    pub fn numerical_type(&self) -> NumericalType {
+        self.compatible_numerical_types.to_numerical_type()
+    }
+
+    pub fn cardinality(&self, num_docs: RowId) -> Cardinality {
+        self.column_writer.get_cardinality(num_docs)
     }
 
     pub fn record_numerical_value(

columnar/src/columnar/writer/mod.rs

@@ -15,7 +15,7 @@ use crate::column_index::SerializableColumnIndex;
 use crate::column_values::{
     ColumnValues, MonotonicallyMappableToU128, MonotonicallyMappableToU64, VecColumn,
 };
-use crate::columnar::column_type::{ColumnType, ColumnTypeCategory};
+use crate::columnar::column_type::ColumnType;
 use crate::columnar::writer::column_writers::{
     ColumnWriter, NumericalColumnWriter, StrOrBytesColumnWriter,
 };
@@ -276,35 +276,40 @@ impl ColumnarWriter {
     }
     pub fn serialize(&mut self, num_docs: RowId, wrt: &mut dyn io::Write) -> io::Result<()> {
         let mut serializer = ColumnarSerializer::new(wrt);
-        let mut columns: Vec<(&[u8], ColumnTypeCategory, Addr)> = self
+        let mut columns: Vec<(&[u8], ColumnType, Addr)> = self
             .numerical_field_hash_map
             .iter()
-            .map(|(column_name, addr, _)| (column_name, ColumnTypeCategory::Numerical, addr))
+            .map(|(column_name, addr, _)| {
+                let numerical_column_writer: NumericalColumnWriter =
+                    self.numerical_field_hash_map.read(addr);
+                let column_type = numerical_column_writer.numerical_type().into();
+                (column_name, column_type, addr)
+            })
             .collect();
         columns.extend(
             self.bytes_field_hash_map
                 .iter()
-                .map(|(term, addr, _)| (term, ColumnTypeCategory::Bytes, addr)),
+                .map(|(term, addr, _)| (term, ColumnType::Bytes, addr)),
         );
         columns.extend(
             self.str_field_hash_map
                 .iter()
-                .map(|(column_name, addr, _)| (column_name, ColumnTypeCategory::Str, addr)),
+                .map(|(column_name, addr, _)| (column_name, ColumnType::Str, addr)),
         );
         columns.extend(
             self.bool_field_hash_map
                 .iter()
-                .map(|(column_name, addr, _)| (column_name, ColumnTypeCategory::Bool, addr)),
+                .map(|(column_name, addr, _)| (column_name, ColumnType::Bool, addr)),
         );
         columns.extend(
             self.ip_addr_field_hash_map
                 .iter()
-                .map(|(column_name, addr, _)| (column_name, ColumnTypeCategory::IpAddr, addr)),
+                .map(|(column_name, addr, _)| (column_name, ColumnType::IpAddr, addr)),
         );
         columns.extend(
             self.datetime_field_hash_map
                 .iter()
-                .map(|(column_name, addr, _)| (column_name, ColumnTypeCategory::DateTime, addr)),
+                .map(|(column_name, addr, _)| (column_name, ColumnType::DateTime, addr)),
         );
         columns.sort_unstable_by_key(|(column_name, col_type, _)| (*column_name, *col_type));
 
@@ -312,8 +317,12 @@ impl ColumnarWriter {
         let mut symbol_byte_buffer: Vec<u8> = Vec::new();
         for (column_name, column_type, addr) in columns {
             match column_type {
-                ColumnTypeCategory::Bool => {
-                    let column_writer: ColumnWriter = self.bool_field_hash_map.read(addr);
+                ColumnType::Bool | ColumnType::DateTime => {
+                    let column_writer: ColumnWriter = if column_type == ColumnType::Bool {
+                        self.bool_field_hash_map.read(addr)
+                    } else {
+                        self.datetime_field_hash_map.read(addr)
+                    };
                     let cardinality = column_writer.get_cardinality(num_docs);
                     let mut column_serializer =
                         serializer.serialize_column(column_name, ColumnType::Bool);
@@ -325,7 +334,7 @@ impl ColumnarWriter {
                         &mut column_serializer,
                     )?;
                 }
-                ColumnTypeCategory::IpAddr => {
+                ColumnType::IpAddr => {
                     let column_writer: ColumnWriter = self.ip_addr_field_hash_map.read(addr);
                     let cardinality = column_writer.get_cardinality(num_docs);
                     let mut column_serializer =
@@ -338,32 +347,35 @@ impl ColumnarWriter {
                         &mut column_serializer,
                     )?;
                 }
-                ColumnTypeCategory::Bytes | ColumnTypeCategory::Str => {
-                    let (column_type, str_column_writer): (ColumnType, StrOrBytesColumnWriter) =
-                        if column_type == ColumnTypeCategory::Bytes {
-                            (ColumnType::Bytes, self.bytes_field_hash_map.read(addr))
+                ColumnType::Bytes | ColumnType::Str => {
+                    let str_or_bytes_column_writer: StrOrBytesColumnWriter =
+                        if column_type == ColumnType::Bytes {
+                            self.bytes_field_hash_map.read(addr)
                         } else {
-                            (ColumnType::Str, self.str_field_hash_map.read(addr))
+                            self.str_field_hash_map.read(addr)
                         };
                     let dictionary_builder =
-                        &dictionaries[str_column_writer.dictionary_id as usize];
-                    let cardinality = str_column_writer.column_writer.get_cardinality(num_docs);
+                        &dictionaries[str_or_bytes_column_writer.dictionary_id as usize];
+                    let cardinality = str_or_bytes_column_writer
+                        .column_writer
+                        .get_cardinality(num_docs);
                     let mut column_serializer =
                         serializer.serialize_column(column_name, column_type);
                     serialize_bytes_or_str_column(
                         cardinality,
                         num_docs,
                         dictionary_builder,
-                        str_column_writer.operation_iterator(arena, &mut symbol_byte_buffer),
+                        str_or_bytes_column_writer
+                            .operation_iterator(arena, &mut symbol_byte_buffer),
                         buffers,
                         &mut column_serializer,
                     )?;
                 }
-                ColumnTypeCategory::Numerical => {
+                ColumnType::I64 | ColumnType::F64 | ColumnType::U64 => {
                     let numerical_column_writer: NumericalColumnWriter =
                         self.numerical_field_hash_map.read(addr);
-                    let (numerical_type, cardinality) =
-                        numerical_column_writer.column_type_and_cardinality(num_docs);
+                    let numerical_type = column_type.numerical_type().unwrap();
+                    let cardinality = numerical_column_writer.cardinality(num_docs);
                     let mut column_serializer =
                         serializer.serialize_column(column_name, ColumnType::from(numerical_type));
                     serialize_numerical_column(
@@ -375,20 +387,6 @@ impl ColumnarWriter {
                         &mut column_serializer,
                     )?;
                 }
-                ColumnTypeCategory::DateTime => {
-                    let column_writer: ColumnWriter = self.datetime_field_hash_map.read(addr);
-                    let cardinality = column_writer.get_cardinality(num_docs);
-                    let mut column_serializer =
-                        serializer.serialize_column(column_name, ColumnType::DateTime);
-                    serialize_numerical_column(
-                        cardinality,
-                        num_docs,
-                        NumericalType::I64,
-                        column_writer.operation_iterator(arena, &mut symbol_byte_buffer),
-                        buffers,
-                        &mut column_serializer,
-                    )?;
-                }
             };
         }
         serializer.finalize()?;

src/core/searcher.rs

@@ -249,7 +249,7 @@ impl SearcherInner {
         index: Index,
         segment_readers: Vec<SegmentReader>,
         generation: TrackedObject<SearcherGeneration>,
-        doc_store_cache_size: usize,
+        doc_store_cache_num_blocks: usize,
     ) -> io::Result<SearcherInner> {
         assert_eq!(
             &segment_readers
@@ -261,7 +261,7 @@ impl SearcherInner {
         );
         let store_readers: Vec<StoreReader> = segment_readers
             .iter()
-            .map(|segment_reader| segment_reader.get_store_reader(doc_store_cache_size))
+            .map(|segment_reader| segment_reader.get_store_reader(doc_store_cache_num_blocks))
             .collect::<io::Result<Vec<_>>>()?;
 
         Ok(SearcherInner {

src/core/segment_reader.rs

@@ -134,9 +134,12 @@ impl SegmentReader {
         &self.fieldnorm_readers
     }
 
-    /// Accessor to the segment's `StoreReader`.
-    pub fn get_store_reader(&self, cache_size: usize) -> io::Result<StoreReader> {
-        StoreReader::open(self.store_file.clone(), cache_size)
+    /// Accessor to the segment's [`StoreReader`](crate::store::StoreReader).
+    ///
+    /// `cache_num_blocks` sets the number of decompressed blocks to be cached in an LRU.
+    /// The size of blocks is configurable, this should be reflexted in the
+    pub fn get_store_reader(&self, cache_num_blocks: usize) -> io::Result<StoreReader> {
+        StoreReader::open(self.store_file.clone(), cache_num_blocks)
     }
 
     /// Open a new segment for reading.

src/reader/mod.rs

@@ -44,7 +44,7 @@ pub struct IndexReaderBuilder {
     index: Index,
     warmers: Vec<Weak<dyn Warmer>>,
     num_warming_threads: usize,
-    doc_store_cache_size: usize,
+    doc_store_cache_num_blocks: usize,
 }
 
 impl IndexReaderBuilder {
@@ -55,7 +55,7 @@ impl IndexReaderBuilder {
             index,
             warmers: Vec::new(),
             num_warming_threads: 1,
-            doc_store_cache_size: DOCSTORE_CACHE_CAPACITY,
+            doc_store_cache_num_blocks: DOCSTORE_CACHE_CAPACITY,
         }
     }
 
@@ -72,7 +72,7 @@ impl IndexReaderBuilder {
             searcher_generation_inventory.clone(),
         )?;
         let inner_reader = InnerIndexReader::new(
-            self.doc_store_cache_size,
+            self.doc_store_cache_num_blocks,
             self.index,
             warming_state,
             searcher_generation_inventory,
@@ -119,8 +119,11 @@ impl IndexReaderBuilder {
     ///
     /// The doc store readers cache by default DOCSTORE_CACHE_CAPACITY(100) decompressed blocks.
     #[must_use]
-    pub fn doc_store_cache_size(mut self, doc_store_cache_size: usize) -> IndexReaderBuilder {
-        self.doc_store_cache_size = doc_store_cache_size;
+    pub fn doc_store_cache_num_blocks(
+        mut self,
+        doc_store_cache_num_blocks: usize,
+    ) -> IndexReaderBuilder {
+        self.doc_store_cache_num_blocks = doc_store_cache_num_blocks;
         self
     }
 
@@ -151,7 +154,7 @@ impl TryInto<IndexReader> for IndexReaderBuilder {
 }
 
 struct InnerIndexReader {
-    doc_store_cache_size: usize,
+    doc_store_cache_num_blocks: usize,
     index: Index,
     warming_state: WarmingState,
     searcher: arc_swap::ArcSwap<SearcherInner>,
@@ -161,7 +164,7 @@ struct InnerIndexReader {
 
 impl InnerIndexReader {
     fn new(
-        doc_store_cache_size: usize,
+        doc_store_cache_num_blocks: usize,
         index: Index,
         warming_state: WarmingState,
         // The searcher_generation_inventory is not used as source, but as target to track the
@@ -172,13 +175,13 @@ impl InnerIndexReader {
 
         let searcher = Self::create_searcher(
             &index,
-            doc_store_cache_size,
+            doc_store_cache_num_blocks,
             &warming_state,
             &searcher_generation_counter,
             &searcher_generation_inventory,
         )?;
         Ok(InnerIndexReader {
-            doc_store_cache_size,
+            doc_store_cache_num_blocks,
             index,
             warming_state,
             searcher: ArcSwap::from(searcher),
@@ -214,7 +217,7 @@ impl InnerIndexReader {
 
     fn create_searcher(
         index: &Index,
-        doc_store_cache_size: usize,
+        doc_store_cache_num_blocks: usize,
         warming_state: &WarmingState,
         searcher_generation_counter: &Arc<AtomicU64>,
         searcher_generation_inventory: &Inventory<SearcherGeneration>,
@@ -232,7 +235,7 @@ impl InnerIndexReader {
             index.clone(),
             segment_readers,
             searcher_generation,
-            doc_store_cache_size,
+            doc_store_cache_num_blocks,
         )?);
 
         warming_state.warm_new_searcher_generation(&searcher.clone().into())?;
@@ -242,7 +245,7 @@ impl InnerIndexReader {
     fn reload(&self) -> crate::Result<()> {
         let searcher = Self::create_searcher(
             &self.index,
-            self.doc_store_cache_size,
+            self.doc_store_cache_num_blocks,
             &self.warming_state,
             &self.searcher_generation_counter,
             &self.searcher_generation_inventory,

src/schema/term.rs

@@ -375,7 +375,7 @@ where B: AsRef<[u8]>
     ///
     /// Do NOT rely on this byte representation in the index.
     /// This value is likely to change in the future.
-    pub(crate) fn as_slice(&self) -> &[u8] {
+    pub fn as_slice(&self) -> &[u8] {
         self.0.as_ref()
     }
 }

src/store/mod.rs

@@ -4,8 +4,8 @@
 //! order to be handled in the `Store`.
 //!
 //! Internally, documents (or rather their stored fields) are serialized to a buffer.
-//! When the buffer exceeds 16K, the buffer is compressed using `brotli`, `LZ4` or `snappy`
-//! and the resulting block is written to disk.
+//! When the buffer exceeds `block_size` (defaults to 16K), the buffer is compressed using `brotli`,
+//! `LZ4` or `snappy` and the resulting block is written to disk.
 //!
 //! One can then request for a specific `DocId`.
 //! A skip list helps navigating to the right block,
@@ -28,8 +28,6 @@
 //! - at the segment level, the
 //! [`SegmentReader`'s `doc` method](../struct.SegmentReader.html#method.doc)
 //! - at the index level, the [`Searcher::doc()`](crate::Searcher::doc) method
-//!
-//! !
 
 mod compressors;
 mod decompressors;

src/store/reader.rs

@@ -114,7 +114,10 @@ impl Sum for CacheStats {
 
 impl StoreReader {
     /// Opens a store reader
-    pub fn open(store_file: FileSlice, cache_size: usize) -> io::Result<StoreReader> {
+    ///
+    /// `cache_num_blocks` sets the number of decompressed blocks to be cached in an LRU.
+    /// The size of blocks is configurable, this should be reflexted in the
+    pub fn open(store_file: FileSlice, cache_num_blocks: usize) -> io::Result<StoreReader> {
         let (footer, data_and_offset) = DocStoreFooter::extract_footer(store_file)?;
 
         let (data_file, offset_index_file) = data_and_offset.split(footer.offset as usize);
@@ -125,8 +128,8 @@ impl StoreReader {
             decompressor: footer.decompressor,
             data: data_file,
             cache: BlockCache {
-                cache: NonZeroUsize::new(cache_size)
-                    .map(|cache_size| Mutex::new(LruCache::new(cache_size))),
+                cache: NonZeroUsize::new(cache_num_blocks)
+                    .map(|cache_num_blocks| Mutex::new(LruCache::new(cache_num_blocks))),
                 cache_hits: Default::default(),
                 cache_misses: Default::default(),
             },