update basic_search example

Cargo.toml

@@ -15,7 +15,8 @@ rust-version = "1.63"
 exclude = ["benches/*.json", "benches/*.txt"]
 
 [dependencies]
-oneshot = "0.1.7"
+# Switch back to the non-forked oneshot crate once https://github.com/faern/oneshot/pull/35 is merged
+oneshot = { git = "https://github.com/fulmicoton/oneshot.git", rev = "b208f49" }
 base64 = "0.22.0"
 byteorder = "1.4.3"
 crc32fast = "1.3.2"
@@ -63,7 +64,7 @@ query-grammar = { version = "0.22.0", path = "./query-grammar", package = "tanti
 tantivy-bitpacker = { version = "0.6", path = "./bitpacker" }
 common = { version = "0.7", path = "./common/", package = "tantivy-common" }
 tokenizer-api = { version = "0.3", path = "./tokenizer-api", package = "tantivy-tokenizer-api" }
-sketches-ddsketch = { version = "0.3.0", features = ["use_serde"] }
+sketches-ddsketch = { version = "0.2.1", features = ["use_serde"] }
 futures-util = { version = "0.3.28", optional = true }
 fnv = "1.0.7"
 

benches/agg_bench.rs

@@ -47,7 +47,6 @@ fn bench_agg(mut group: InputGroup<Index>) {
     register!(group, average_f64);
     register!(group, average_f64_u64);
     register!(group, stats_f64);
-    register!(group, extendedstats_f64);
     register!(group, percentiles_f64);
     register!(group, terms_few);
     register!(group, terms_many);
@@ -106,12 +105,7 @@ fn stats_f64(index: &Index) {
     });
     exec_term_with_agg(index, agg_req)
 }
-fn extendedstats_f64(index: &Index) {
-    let agg_req = json!({
-        "extendedstats_f64": { "extended_stats": { "field": "score_f64", } }
-    });
-    exec_term_with_agg(index, agg_req)
-}
+
 fn percentiles_f64(index: &Index) {
     let agg_req = json!({
       "mypercentiles": {

columnar/Cargo.toml

@@ -23,12 +23,6 @@ downcast-rs = "1.2.0"
 proptest = "1"
 more-asserts = "0.3.1"
 rand = "0.8"
-binggan = "0.8.1"
-
-[[bench]]
-name = "bench_merge"
-harness = false
-
 
 [features]
 unstable = []

columnar/benches/bench_merge.rs

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

columnar/src/column_index/merge/mod.rs

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

columnar/src/column_index/merge/shuffled.rs

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

columnar/src/column_index/merge/stacked.rs

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

columnar/src/column_index/multivalued_index.rs

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

columnar/src/column_index/optional_index/mod.rs

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

columnar/src/column_index/serialize.rs

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

columnar/src/column_values/mod.rs

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

columnar/src/column_values/u64_based/bitpacked.rs

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

columnar/src/column_values/u64_based/blockwise_linear.rs

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

columnar/src/column_values/u64_based/linear.rs

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

columnar/src/column_values/u64_based/mod.rs

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

columnar/src/columnar/merge/mod.rs

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

columnar/src/columnar/writer/mod.rs

@@ -644,10 +644,7 @@ fn send_to_serialize_column_mappable_to_u128<
             let multivalued_index_builder = value_index_builders.borrow_multivalued_index_builder();
             consume_operation_iterator(op_iterator, multivalued_index_builder, values);
             let multivalued_index = multivalued_index_builder.finish(num_rows);
-            SerializableColumnIndex::Multivalued {
-                indices: Box::new(multivalued_index),
-                stats: Default::default(), // TODO: implement stats for u128
-            }
+            SerializableColumnIndex::Multivalued(Box::new(multivalued_index))
         }
     };
     crate::column::serialize_column_mappable_to_u128(
@@ -702,10 +699,7 @@ fn send_to_serialize_column_mappable_to_u64(
             if sort_values_within_row {
                 sort_values_within_row_in_place(multivalued_index, values);
             }
-            SerializableColumnIndex::Multivalued {
-                indices: Box::new(multivalued_index),
-                stats: None,
-            }
+            SerializableColumnIndex::Multivalued(Box::new(multivalued_index))
         }
     };
     crate::column::serialize_column_mappable_to_u64(

columnar/src/tests.rs

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

common/Cargo.toml

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

examples/basic_search.rs

@@ -19,13 +19,14 @@ use tantivy::{doc, Index, IndexWriter, ReloadPolicy};
 use tempfile::TempDir;
 
 fn main() -> tantivy::Result<()> {
-    // Let's create a temporary directory for the
-    // sake of this example
+    // Normally you would use `MMapDirectory` instead to persist data on disk.
+    // https://docs.rs/tantivy/latest/tantivy/directory/struct.MmapDirectory.html
+    // But for this example, we will use a temporary directory `TempDir`.
     let index_path = TempDir::new()?;
 
     // # Defining the schema
     //
-    // The Tantivy index requires a very strict schema.
+    // The Tantivy index requires a schema.
     // The schema declares which fields are in the index,
     // and for each field, its type and "the way it should
     // be indexed".

query-grammar/src/query_grammar.rs

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

src/aggregation/agg_req.rs

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

src/aggregation/agg_req_with_accessor.rs

@@ -11,8 +11,8 @@ use super::bucket::{
     DateHistogramAggregationReq, HistogramAggregation, RangeAggregation, TermsAggregation,
 };
 use super::metric::{
-    AverageAggregation, CountAggregation, ExtendedStatsAggregation, MaxAggregation, MinAggregation,
-    StatsAggregation, SumAggregation,
+    AverageAggregation, CountAggregation, MaxAggregation, MinAggregation, StatsAggregation,
+    SumAggregation,
 };
 use super::segment_agg_result::AggregationLimits;
 use super::VecWithNames;
@@ -276,10 +276,6 @@ impl AggregationWithAccessor {
                 field: ref field_name,
                 ..
             })
-            | ExtendedStats(ExtendedStatsAggregation {
-                field: ref field_name,
-                ..
-            })
             | Sum(SumAggregation {
                 field: ref field_name,
                 ..

src/aggregation/agg_result.rs

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

src/aggregation/intermediate_agg_result.rs

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

src/aggregation/metric/mod.rs

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

src/aggregation/metric/stats.rs

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

src/aggregation/metric/top_hits.rs

@@ -1,7 +1,7 @@
 use std::collections::HashMap;
 use std::net::Ipv6Addr;
 
-use columnar::{Column, ColumnType, ColumnarReader, DynamicColumn};
+use columnar::{ColumnarReader, DynamicColumn};
 use common::json_path_writer::JSON_PATH_SEGMENT_SEP_STR;
 use common::DateTime;
 use regex::Regex;
@@ -443,10 +443,10 @@ impl std::cmp::PartialEq for TopHitsTopNComputer {
 
 impl TopHitsTopNComputer {
     /// Create a new TopHitsCollector
-    pub fn new(req: &TopHitsAggregation) -> Self {
+    pub fn new(req: TopHitsAggregation) -> Self {
         Self {
             top_n: TopNComputer::new(req.size + req.from.unwrap_or(0)),
-            req: req.clone(),
+            req,
         }
     }
 
@@ -491,6 +491,7 @@ impl TopHitsTopNComputer {
 pub(crate) struct TopHitsSegmentCollector {
     segment_ordinal: SegmentOrdinal,
     accessor_idx: usize,
+    req: TopHitsAggregation,
     top_n: TopNComputer<Vec<DocValueAndOrder>, DocAddress, false>,
 }
 
@@ -501,6 +502,7 @@ impl TopHitsSegmentCollector {
         segment_ordinal: SegmentOrdinal,
     ) -> Self {
         Self {
+            req: req.clone(),
             top_n: TopNComputer::new(req.size + req.from.unwrap_or(0)),
             segment_ordinal,
             accessor_idx,
@@ -509,13 +511,14 @@ impl TopHitsSegmentCollector {
     fn into_top_hits_collector(
         self,
         value_accessors: &HashMap<String, Vec<DynamicColumn>>,
-        req: &TopHitsAggregation,
     ) -> TopHitsTopNComputer {
-        let mut top_hits_computer = TopHitsTopNComputer::new(req);
+        let mut top_hits_computer = TopHitsTopNComputer::new(self.req.clone());
         let top_results = self.top_n.into_vec();
 
         for res in top_results {
-            let doc_value_fields = req.get_document_field_data(value_accessors, res.doc.doc_id);
+            let doc_value_fields = self
+                .req
+                .get_document_field_data(value_accessors, res.doc.doc_id);
             top_hits_computer.collect(
                 DocSortValuesAndFields {
                     sorts: res.feature,
@@ -527,15 +530,34 @@ impl TopHitsSegmentCollector {
 
         top_hits_computer
     }
+}
 
-    /// TODO add a specialized variant for a single sort field
-    fn collect_with(
+impl SegmentAggregationCollector for TopHitsSegmentCollector {
+    fn add_intermediate_aggregation_result(
+        self: Box<Self>,
+        agg_with_accessor: &crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
+        results: &mut crate::aggregation::intermediate_agg_result::IntermediateAggregationResults,
+    ) -> crate::Result<()> {
+        let name = agg_with_accessor.aggs.keys[self.accessor_idx].to_string();
+
+        let value_accessors = &agg_with_accessor.aggs.values[self.accessor_idx].value_accessors;
+
+        let intermediate_result =
+            IntermediateMetricResult::TopHits(self.into_top_hits_collector(value_accessors));
+        results.push(
+            name,
+            IntermediateAggregationResult::Metric(intermediate_result),
+        )
+    }
+
+    fn collect(
         &mut self,
         doc_id: crate::DocId,
-        req: &TopHitsAggregation,
-        accessors: &[(Column<u64>, ColumnType)],
+        agg_with_accessor: &mut crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let sorts: Vec<DocValueAndOrder> = req
+        let accessors = &agg_with_accessor.aggs.values[self.accessor_idx].accessors;
+        let sorts: Vec<DocValueAndOrder> = self
+            .req
             .sort
             .iter()
             .enumerate()
@@ -560,62 +582,15 @@ impl TopHitsSegmentCollector {
         );
         Ok(())
     }
-}
-
-impl SegmentAggregationCollector for TopHitsSegmentCollector {
-    fn add_intermediate_aggregation_result(
-        self: Box<Self>,
-        agg_with_accessor: &crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
-        results: &mut crate::aggregation::intermediate_agg_result::IntermediateAggregationResults,
-    ) -> crate::Result<()> {
-        let name = agg_with_accessor.aggs.keys[self.accessor_idx].to_string();
-
-        let value_accessors = &agg_with_accessor.aggs.values[self.accessor_idx].value_accessors;
-        let tophits_req = &agg_with_accessor.aggs.values[self.accessor_idx]
-            .agg
-            .agg
-            .as_top_hits()
-            .expect("aggregation request must be of type top hits");
-
-        let intermediate_result = IntermediateMetricResult::TopHits(
-            self.into_top_hits_collector(value_accessors, tophits_req),
-        );
-        results.push(
-            name,
-            IntermediateAggregationResult::Metric(intermediate_result),
-        )
-    }
-
-    /// TODO: Consider a caching layer to reduce the call overhead
-    fn collect(
-        &mut self,
-        doc_id: crate::DocId,
-        agg_with_accessor: &mut crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
-    ) -> crate::Result<()> {
-        let tophits_req = &agg_with_accessor.aggs.values[self.accessor_idx]
-            .agg
-            .agg
-            .as_top_hits()
-            .expect("aggregation request must be of type top hits");
-        let accessors = &agg_with_accessor.aggs.values[self.accessor_idx].accessors;
-        self.collect_with(doc_id, tophits_req, accessors)?;
-        Ok(())
-    }
 
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
         agg_with_accessor: &mut crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let tophits_req = &agg_with_accessor.aggs.values[self.accessor_idx]
-            .agg
-            .agg
-            .as_top_hits()
-            .expect("aggregation request must be of type top hits");
-        let accessors = &agg_with_accessor.aggs.values[self.accessor_idx].accessors;
         // TODO: Consider getting fields with the column block accessor.
         for doc in docs {
-            self.collect_with(*doc, tophits_req, accessors)?;
+            self.collect(*doc, agg_with_accessor)?;
         }
         Ok(())
     }

src/aggregation/segment_agg_result.rs

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

src/collector/top_score_collector.rs

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

src/core/executor.rs

@@ -195,7 +195,7 @@ mod tests {
         let (tx, rx) = crossbeam_channel::bounded::<()>(0);
         let rx = Arc::new(rx);
         let executor = Executor::multi_thread(3, "search-test").unwrap();
-        for _ in 0..1000 {
+        for i in 0..1000 {
             let counter_clone: Arc<AtomicU64> = counter.clone();
             let other_counter_clone: Arc<AtomicU64> = other_counter.clone();
 
@@ -203,18 +203,18 @@ mod tests {
             let rx_clone2 = rx.clone();
             let fut = executor.spawn_blocking(move || {
                 counter_clone.fetch_add(1, Ordering::SeqCst);
-                let _ = rx_clone.recv();
+                let () = rx_clone.recv().unwrap();
             });
             futures.push(fut);
             let other_fut = executor.spawn_blocking(move || {
                 other_counter_clone.fetch_add(1, Ordering::SeqCst);
-                let _ = rx_clone2.recv();
+                let () = rx_clone2.recv().unwrap();
             });
             other_futures.push(other_fut);
         }
 
         // We execute 100 futures.
-        for _ in 0..100 {
+        for i in 0..100 {
             tx.send(()).unwrap();
         }
 
@@ -226,7 +226,7 @@ mod tests {
         drop(other_futures);
 
         // We execute 100 futures.
-        for _ in 0..100 {
+        for i in 0..100 {
             tx.send(()).unwrap();
         }
 

src/indexer/index_writer.rs

@@ -808,7 +808,7 @@ mod tests {
     use proptest::prop_oneof;
 
     use super::super::operation::UserOperation;
-    use crate::collector::{Count, TopDocs};
+    use crate::collector::TopDocs;
     use crate::directory::error::LockError;
     use crate::error::*;
     use crate::indexer::index_writer::MEMORY_BUDGET_NUM_BYTES_MIN;
@@ -1572,74 +1572,20 @@ mod tests {
         Ok(())
     }
 
-    #[derive(Debug, Clone)]
+    #[derive(Debug, Clone, Copy)]
     enum IndexingOp {
-        AddMultipleDoc {
-            id: u64,
-            num_docs: u64,
-            value: IndexValue,
-        },
-        AddDoc {
-            id: u64,
-            value: IndexValue,
-        },
-        DeleteDoc {
-            id: u64,
-        },
-        DeleteDocQuery {
-            id: u64,
-        },
+        AddDoc { id: u64 },
+        DeleteDoc { id: u64 },
+        DeleteDocQuery { id: u64 },
         Commit,
         Merge,
     }
-    impl IndexingOp {
-        fn add(id: u64) -> Self {
-            IndexingOp::AddDoc {
-                id,
-                value: IndexValue::F64(id as f64),
-            }
-        }
-    }
-
-    use serde::Serialize;
-    #[derive(Debug, Clone, Serialize)]
-    #[serde(untagged)]
-    enum IndexValue {
-        Str(String),
-        F64(f64),
-        U64(u64),
-        I64(i64),
-    }
-    impl Default for IndexValue {
-        fn default() -> Self {
-            IndexValue::F64(0.0)
-        }
-    }
-
-    fn value_strategy() -> impl Strategy<Value = IndexValue> {
-        prop_oneof![
-            any::<f64>().prop_map(IndexValue::F64),
-            any::<u64>().prop_map(IndexValue::U64),
-            any::<i64>().prop_map(IndexValue::I64),
-            any::<String>().prop_map(IndexValue::Str),
-        ]
-    }
 
     fn balanced_operation_strategy() -> impl Strategy<Value = IndexingOp> {
         prop_oneof![
             (0u64..20u64).prop_map(|id| IndexingOp::DeleteDoc { id }),
             (0u64..20u64).prop_map(|id| IndexingOp::DeleteDocQuery { id }),
-            (0u64..20u64, value_strategy())
-                .prop_map(move |(id, value)| IndexingOp::AddDoc { id, value }),
-            ((0u64..20u64), (1u64..100), value_strategy()).prop_map(
-                move |(id, num_docs, value)| {
-                    IndexingOp::AddMultipleDoc {
-                        id,
-                        num_docs,
-                        value,
-                    }
-                }
-            ),
+            (0u64..20u64).prop_map(|id| IndexingOp::AddDoc { id }),
             (0u64..1u64).prop_map(|_| IndexingOp::Commit),
             (0u64..1u64).prop_map(|_| IndexingOp::Merge),
         ]
@@ -1649,17 +1595,7 @@ mod tests {
         prop_oneof![
             5 => (0u64..100u64).prop_map(|id| IndexingOp::DeleteDoc { id }),
             5 => (0u64..100u64).prop_map(|id| IndexingOp::DeleteDocQuery { id }),
-            50 => (0u64..100u64, value_strategy())
-                .prop_map(move |(id, value)| IndexingOp::AddDoc { id, value }),
-            50 => (0u64..100u64, (1u64..100), value_strategy()).prop_map(
-                move |(id, num_docs, value)| {
-                    IndexingOp::AddMultipleDoc {
-                        id,
-                        num_docs,
-                        value,
-                    }
-                }
-            ),
+            50 => (0u64..100u64).prop_map(|id| IndexingOp::AddDoc { id }),
             2 => (0u64..1u64).prop_map(|_| IndexingOp::Commit),
             1 => (0u64..1u64).prop_map(|_| IndexingOp::Merge),
         ]
@@ -1668,27 +1604,19 @@ mod tests {
     fn expected_ids(ops: &[IndexingOp]) -> (HashMap<u64, u64>, HashSet<u64>) {
         let mut existing_ids = HashMap::new();
         let mut deleted_ids = HashSet::new();
-        for op in ops {
+        for &op in ops {
             match op {
-                IndexingOp::AddDoc { id, value: _ } => {
-                    *existing_ids.entry(*id).or_insert(0) += 1;
-                    deleted_ids.remove(id);
-                }
-                IndexingOp::AddMultipleDoc {
-                    id,
-                    num_docs,
-                    value: _,
-                } => {
-                    *existing_ids.entry(*id).or_insert(0) += num_docs;
-                    deleted_ids.remove(id);
+                IndexingOp::AddDoc { id } => {
+                    *existing_ids.entry(id).or_insert(0) += 1;
+                    deleted_ids.remove(&id);
                 }
                 IndexingOp::DeleteDoc { id } => {
                     existing_ids.remove(&id);
-                    deleted_ids.insert(*id);
+                    deleted_ids.insert(id);
                 }
                 IndexingOp::DeleteDocQuery { id } => {
                     existing_ids.remove(&id);
-                    deleted_ids.insert(*id);
+                    deleted_ids.insert(id);
                 }
                 _ => {}
             }
@@ -1698,19 +1626,16 @@ mod tests {
 
     fn get_id_list(ops: &[IndexingOp]) -> Vec<u64> {
         let mut id_list = Vec::new();
-        for op in ops {
+        for &op in ops {
             match op {
-                IndexingOp::AddDoc { id, value: _ } => {
-                    id_list.push(*id);
-                }
-                IndexingOp::AddMultipleDoc { id, .. } => {
-                    id_list.push(*id);
+                IndexingOp::AddDoc { id } => {
+                    id_list.push(id);
                 }
                 IndexingOp::DeleteDoc { id } => {
-                    id_list.retain(|el| el != id);
+                    id_list.retain(|el| *el != id);
                 }
                 IndexingOp::DeleteDocQuery { id } => {
-                    id_list.retain(|el| el != id);
+                    id_list.retain(|el| *el != id);
                 }
                 _ => {}
             }
@@ -1791,59 +1716,42 @@ mod tests {
 
         let ip_from_id = |id| Ipv6Addr::from_u128(id as u128);
 
-        let add_docs = |index_writer: &mut IndexWriter,
-                        id: u64,
-                        value: IndexValue,
-                        num: u64|
-         -> crate::Result<()> {
-            let facet = Facet::from(&("/cola/".to_string() + &id.to_string()));
-            let ip = ip_from_id(id);
-            let doc = if !id_is_full_doc(id) {
-                // every 3rd doc has no ip field
-                doc!(
-                    id_field=>id,
-                )
-            } else {
-                let json = json!({"date1": format!("2022-{id}-01T00:00:01Z"), "date2": format!("{id}-05-01T00:00:01Z"), "id": id, "ip": ip.to_string(), "val": value});
-                doc!(id_field=>id,
-                        json_field=>json,
-                        bytes_field => id.to_le_bytes().as_slice(),
-                        id_opt_field => id,
-                        ip_field => ip,
-                        ips_field => ip,
-                        ips_field => ip,
-                        multi_numbers=> id,
-                        multi_numbers => id,
-                        bool_field => (id % 2u64) != 0,
-                        i64_field => id as i64,
-                        f64_field => id as f64,
-                        date_field => DateTime::from_timestamp_secs(id as i64),
-                        multi_bools => (id % 2u64) != 0,
-                        multi_bools => (id % 2u64) == 0,
-                        text_field => id.to_string(),
-                        facet_field => facet,
-                        large_text_field => LOREM,
-                        multi_text_fields => multi_text_field_text1,
-                        multi_text_fields => multi_text_field_text2,
-                        multi_text_fields => multi_text_field_text3,
-                )
-            };
-            for _ in 0..num {
-                index_writer.add_document(doc.clone())?;
-            }
-            Ok(())
-        };
-        for op in ops {
-            match op.clone() {
-                IndexingOp::AddMultipleDoc {
-                    id,
-                    num_docs,
-                    value,
-                } => {
-                    add_docs(&mut index_writer, id, value, num_docs)?;
-                }
-                IndexingOp::AddDoc { id, value } => {
-                    add_docs(&mut index_writer, id, value, 1)?;
+        for &op in ops {
+            match op {
+                IndexingOp::AddDoc { id } => {
+                    let facet = Facet::from(&("/cola/".to_string() + &id.to_string()));
+                    let ip = ip_from_id(id);
+
+                    if !id_is_full_doc(id) {
+                        // every 3rd doc has no ip field
+                        index_writer.add_document(doc!(
+                            id_field=>id,
+                        ))?;
+                    } else {
+                        let json = json!({"date1": format!("2022-{id}-01T00:00:01Z"), "date2": format!("{id}-05-01T00:00:01Z"), "id": id, "ip": ip.to_string()});
+                        index_writer.add_document(doc!(id_field=>id,
+                                json_field=>json,
+                                bytes_field => id.to_le_bytes().as_slice(),
+                                id_opt_field => id,
+                                ip_field => ip,
+                                ips_field => ip,
+                                ips_field => ip,
+                                multi_numbers=> id,
+                                multi_numbers => id,
+                                bool_field => (id % 2u64) != 0,
+                                i64_field => id as i64,
+                                f64_field => id as f64,
+                                date_field => DateTime::from_timestamp_secs(id as i64),
+                                multi_bools => (id % 2u64) != 0,
+                                multi_bools => (id % 2u64) == 0,
+                                text_field => id.to_string(),
+                                facet_field => facet,
+                                large_text_field => LOREM,
+                                multi_text_fields => multi_text_field_text1,
+                                multi_text_fields => multi_text_field_text2,
+                                multi_text_fields => multi_text_field_text3,
+                        ))?;
+                    }
                 }
                 IndexingOp::DeleteDoc { id } => {
                     index_writer.delete_term(Term::from_field_u64(id_field, id));
@@ -2124,22 +2032,18 @@ mod tests {
 
             top_docs.iter().map(|el| el.1).collect::<Vec<_>>()
         };
-        let count_search = |term: &str, field| {
-            let query = QueryParser::for_index(&index, vec![field])
-                .parse_query(term)
-                .unwrap();
-            searcher.search(&query, &Count).unwrap()
-        };
 
-        let count_search2 = |term: Term| {
+        let do_search2 = |term: Term| {
             let query = TermQuery::new(term, IndexRecordOption::Basic);
-            searcher.search(&query, &Count).unwrap()
+            let top_docs: Vec<(f32, DocAddress)> =
+                searcher.search(&query, &TopDocs::with_limit(1000)).unwrap();
+
+            top_docs.iter().map(|el| el.1).collect::<Vec<_>>()
         };
 
         for (id, count) in &expected_ids_and_num_occurrences {
-            // skip expensive queries
             let (existing_id, count) = (*id, *count);
-            let get_num_hits = |field| count_search(&existing_id.to_string(), field) as u64;
+            let get_num_hits = |field| do_search(&existing_id.to_string(), field).len() as u64;
             assert_eq!(get_num_hits(id_field), count);
             if !id_is_full_doc(existing_id) {
                 continue;
@@ -2149,31 +2053,29 @@ mod tests {
             assert_eq!(get_num_hits(f64_field), count);
 
             // Test multi text
-            if num_docs_with_values < 1000 {
-                assert_eq!(
-                    do_search("\"test1 test2\"", multi_text_fields).len(),
-                    num_docs_with_values
-                );
-                assert_eq!(
-                    do_search("\"test2 test3\"", multi_text_fields).len(),
-                    num_docs_with_values
-                );
-            }
+            assert_eq!(
+                do_search("\"test1 test2\"", multi_text_fields).len(),
+                num_docs_with_values
+            );
+            assert_eq!(
+                do_search("\"test2 test3\"", multi_text_fields).len(),
+                num_docs_with_values
+            );
 
             // Test bytes
             let term = Term::from_field_bytes(bytes_field, existing_id.to_le_bytes().as_slice());
-            assert_eq!(count_search2(term) as u64, count);
+            assert_eq!(do_search2(term).len() as u64, count);
 
             // Test date
             let term = Term::from_field_date(
                 date_field,
                 DateTime::from_timestamp_secs(existing_id as i64),
             );
-            assert_eq!(count_search2(term) as u64, count);
+            assert_eq!(do_search2(term).len() as u64, count);
         }
         for deleted_id in deleted_ids {
             let assert_field = |field| {
-                assert_eq!(count_search(&deleted_id.to_string(), field) as u64, 0);
+                assert_eq!(do_search(&deleted_id.to_string(), field).len() as u64, 0);
             };
             assert_field(text_field);
             assert_field(f64_field);
@@ -2182,12 +2084,12 @@ mod tests {
 
             // Test bytes
             let term = Term::from_field_bytes(bytes_field, deleted_id.to_le_bytes().as_slice());
-            assert_eq!(count_search2(term), 0);
+            assert_eq!(do_search2(term).len() as u64, 0);
 
             // Test date
             let term =
                 Term::from_field_date(date_field, DateTime::from_timestamp_secs(deleted_id as i64));
-            assert_eq!(count_search2(term), 0);
+            assert_eq!(do_search2(term).len() as u64, 0);
         }
         // search ip address
         //
@@ -2196,13 +2098,13 @@ mod tests {
             if !id_is_full_doc(existing_id) {
                 continue;
             }
-            let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+            let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
             let ip_addr = Ipv6Addr::from_u128(existing_id as u128);
             // Test incoming ip as ipv6
             assert_eq!(do_search_ip_field(&format!("\"{ip_addr}\"")), count);
 
             let term = Term::from_field_ip_addr(ip_field, ip_addr);
-            assert_eq!(count_search2(term) as u64, count);
+            assert_eq!(do_search2(term).len() as u64, count);
 
             // Test incoming ip as ipv4
             if let Some(ip_addr) = ip_addr.to_ipv4_mapped() {
@@ -2219,7 +2121,7 @@ mod tests {
         if !sample.is_empty() {
             let (left_sample, right_sample) = sample.split_at(sample.len() / 2);
 
-            let calc_expected_count = |sample: &[(&u64, &u64)]| {
+            let expected_count = |sample: &[(&u64, &u64)]| {
                 sample
                     .iter()
                     .filter(|(id, _)| id_is_full_doc(**id))
@@ -2235,17 +2137,18 @@ mod tests {
             }
 
             // Query first half
-            let expected_count = calc_expected_count(left_sample);
-            if !left_sample.is_empty() && expected_count < 1000 {
+            if !left_sample.is_empty() {
+                let expected_count = expected_count(left_sample);
+
                 let start_range = *left_sample[0].0;
                 let end_range = *left_sample.last().unwrap().0;
                 let query = gen_query_inclusive("id_opt", start_range, end_range);
-                assert_eq!(count_search(&query, id_opt_field) as u64, expected_count);
+                assert_eq!(do_search(&query, id_opt_field).len() as u64, expected_count);
 
                 // Range query on ip field
                 let ip1 = ip_from_id(start_range);
                 let ip2 = ip_from_id(end_range);
-                let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+                let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
                 let query = gen_query_inclusive("ip", ip1, ip2);
                 assert_eq!(do_search_ip_field(&query), expected_count);
                 let query = gen_query_inclusive("ip", "*", ip2);
@@ -2257,19 +2160,19 @@ mod tests {
                 assert_eq!(do_search_ip_field(&query), expected_count);
             }
             // Query second half
-            let expected_count = calc_expected_count(right_sample);
-            if !right_sample.is_empty() && expected_count < 1000 {
+            if !right_sample.is_empty() {
+                let expected_count = expected_count(right_sample);
                 let start_range = *right_sample[0].0;
                 let end_range = *right_sample.last().unwrap().0;
                 // Range query on id opt field
                 let query =
                     gen_query_inclusive("id_opt", start_range.to_string(), end_range.to_string());
-                assert_eq!(count_search(&query, id_opt_field) as u64, expected_count);
+                assert_eq!(do_search(&query, id_opt_field).len() as u64, expected_count);
 
                 // Range query on ip field
                 let ip1 = ip_from_id(start_range);
                 let ip2 = ip_from_id(end_range);
-                let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+                let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
                 let query = gen_query_inclusive("ip", ip1, ip2);
                 assert_eq!(do_search_ip_field(&query), expected_count);
                 let query = gen_query_inclusive("ip", ip1, "*");
@@ -2294,7 +2197,7 @@ mod tests {
             };
             let ip = ip_from_id(existing_id);
 
-            let do_search_ip_field = |term: &str| count_search(term, ip_field) as u64;
+            let do_search_ip_field = |term: &str| do_search(term, ip_field).len() as u64;
             // Range query on single value field
             let query = gen_query_inclusive("ip", ip, ip);
             assert_eq!(do_search_ip_field(&query), count);
@@ -2354,7 +2257,7 @@ mod tests {
 
     #[test]
     fn test_fast_field_range() {
-        let ops: Vec<_> = (0..1000).map(|id| IndexingOp::add(id)).collect();
+        let ops: Vec<_> = (0..1000).map(|id| IndexingOp::AddDoc { id }).collect();
         assert!(test_operation_strategy(&ops, false, true).is_ok());
     }
 
@@ -2362,8 +2265,8 @@ mod tests {
     fn test_sort_index_on_opt_field_regression() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(81),
-                IndexingOp::add(70),
+                IndexingOp::AddDoc { id: 81 },
+                IndexingOp::AddDoc { id: 70 },
                 IndexingOp::DeleteDoc { id: 70 }
             ],
             true,
@@ -2372,45 +2275,14 @@ mod tests {
         .is_ok());
     }
 
-    #[test]
-    fn test_simple_multiple_doc() {
-        assert!(test_operation_strategy(
-            &[
-                IndexingOp::AddMultipleDoc {
-                    id: 7,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-                IndexingOp::AddMultipleDoc {
-                    id: 92,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-                IndexingOp::AddMultipleDoc {
-                    id: 30,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-                IndexingOp::AddMultipleDoc {
-                    id: 33,
-                    num_docs: 800,
-                    value: IndexValue::U64(0),
-                },
-            ],
-            true,
-            false
-        )
-        .is_ok());
-    }
-
     #[test]
     fn test_ip_range_query_multivalue_bug() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(2),
+                IndexingOp::AddDoc { id: 2 },
                 IndexingOp::Commit,
-                IndexingOp::add(1),
-                IndexingOp::add(1),
+                IndexingOp::AddDoc { id: 1 },
+                IndexingOp::AddDoc { id: 1 },
                 IndexingOp::Commit,
                 IndexingOp::Merge
             ],
@@ -2424,11 +2296,11 @@ mod tests {
     fn test_ff_num_ips_regression() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(13),
-                IndexingOp::add(1),
+                IndexingOp::AddDoc { id: 13 },
+                IndexingOp::AddDoc { id: 1 },
                 IndexingOp::Commit,
                 IndexingOp::DeleteDocQuery { id: 13 },
-                IndexingOp::add(1),
+                IndexingOp::AddDoc { id: 1 },
                 IndexingOp::Commit,
             ],
             false,
@@ -2440,7 +2312,7 @@ mod tests {
     #[test]
     fn test_minimal_sort_force_end_merge() {
         assert!(test_operation_strategy(
-            &[IndexingOp::add(23), IndexingOp::add(13),],
+            &[IndexingOp::AddDoc { id: 23 }, IndexingOp::AddDoc { id: 13 },],
             false,
             false
         )
@@ -2501,8 +2373,8 @@ mod tests {
     fn test_minimal_sort_force_end_merge_with_delete() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(23),
-                IndexingOp::add(13),
+                IndexingOp::AddDoc { id: 23 },
+                IndexingOp::AddDoc { id: 13 },
                 IndexingOp::DeleteDoc { id: 13 }
             ],
             true,
@@ -2515,8 +2387,8 @@ mod tests {
     fn test_minimal_no_sort_no_force_end_merge() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::add(23),
-                IndexingOp::add(13),
+                IndexingOp::AddDoc { id: 23 },
+                IndexingOp::AddDoc { id: 13 },
                 IndexingOp::DeleteDoc { id: 13 }
             ],
             false,
@@ -2527,7 +2399,7 @@ mod tests {
 
     #[test]
     fn test_minimal_sort_merge() {
-        assert!(test_operation_strategy(&[IndexingOp::add(3),], true, true).is_ok());
+        assert!(test_operation_strategy(&[IndexingOp::AddDoc { id: 3 },], true, true).is_ok());
     }
 
     use proptest::prelude::*;
@@ -2623,14 +2495,14 @@ mod tests {
     fn test_delete_bug_reproduction_ip_addr() {
         use IndexingOp::*;
         let ops = &[
-            IndexingOp::add(1),
-            IndexingOp::add(2),
+            AddDoc { id: 1 },
+            AddDoc { id: 2 },
             Commit,
-            IndexingOp::add(3),
+            AddDoc { id: 3 },
             DeleteDoc { id: 1 },
             Commit,
             Merge,
-            IndexingOp::add(4),
+            AddDoc { id: 4 },
             Commit,
         ];
         test_operation_strategy(&ops[..], false, true).unwrap();
@@ -2639,13 +2511,7 @@ mod tests {
     #[test]
     fn test_merge_regression_1() {
         use IndexingOp::*;
-        let ops = &[
-            IndexingOp::add(15),
-            Commit,
-            IndexingOp::add(9),
-            Commit,
-            Merge,
-        ];
+        let ops = &[AddDoc { id: 15 }, Commit, AddDoc { id: 9 }, Commit, Merge];
         test_operation_strategy(&ops[..], false, true).unwrap();
     }
 
@@ -2653,9 +2519,9 @@ mod tests {
     fn test_range_query_bug_1() {
         use IndexingOp::*;
         let ops = &[
-            IndexingOp::add(9),
-            IndexingOp::add(0),
-            IndexingOp::add(13),
+            AddDoc { id: 9 },
+            AddDoc { id: 0 },
+            AddDoc { id: 13 },
             Commit,
         ];
         test_operation_strategy(&ops[..], false, true).unwrap();
@@ -2663,11 +2529,12 @@ mod tests {
 
     #[test]
     fn test_range_query_bug_2() {
+        use IndexingOp::*;
         let ops = &[
-            IndexingOp::add(3),
-            IndexingOp::add(6),
-            IndexingOp::add(9),
-            IndexingOp::add(10),
+            AddDoc { id: 3 },
+            AddDoc { id: 6 },
+            AddDoc { id: 9 },
+            AddDoc { id: 10 },
         ];
         test_operation_strategy(&ops[..], false, false).unwrap();
     }
@@ -2689,7 +2556,7 @@ mod tests {
         assert!(test_operation_strategy(
             &[
                 IndexingOp::DeleteDoc { id: 0 },
-                IndexingOp::add(6),
+                IndexingOp::AddDoc { id: 6 },
                 IndexingOp::DeleteDocQuery { id: 11 },
                 IndexingOp::Commit,
                 IndexingOp::Merge,
@@ -2706,13 +2573,10 @@ mod tests {
     fn test_bug_1617_2() {
         assert!(test_operation_strategy(
             &[
-                IndexingOp::AddDoc {
-                    id: 13,
-                    value: Default::default()
-                },
+                IndexingOp::AddDoc { id: 13 },
                 IndexingOp::DeleteDoc { id: 13 },
                 IndexingOp::Commit,
-                IndexingOp::add(30),
+                IndexingOp::AddDoc { id: 30 },
                 IndexingOp::Commit,
                 IndexingOp::Merge,
             ],

src/indexer/merger.rs

@@ -787,8 +787,6 @@ impl IndexMerger {
 mod tests {
 
     use columnar::Column;
-    use proptest::prop_oneof;
-    use proptest::strategy::Strategy;
     use schema::FAST;
 
     use crate::collector::tests::{
@@ -796,7 +794,6 @@ mod tests {
     };
     use crate::collector::{Count, FacetCollector};
     use crate::index::{Index, SegmentId};
-    use crate::indexer::NoMergePolicy;
     use crate::query::{AllQuery, BooleanQuery, EnableScoring, Scorer, TermQuery};
     use crate::schema::{
         Facet, FacetOptions, IndexRecordOption, NumericOptions, TantivyDocument, Term,
@@ -1534,112 +1531,6 @@ mod tests {
         Ok(())
     }
 
-    #[derive(Debug, Clone, Copy, Eq, PartialEq)]
-    enum IndexingOp {
-        ZeroVal,
-        OneVal { val: u64 },
-        TwoVal { val: u64 },
-        Commit,
-    }
-
-    fn balanced_operation_strategy() -> impl Strategy<Value = IndexingOp> {
-        prop_oneof![
-            (0u64..1u64).prop_map(|_| IndexingOp::ZeroVal),
-            (0u64..1u64).prop_map(|val| IndexingOp::OneVal { val }),
-            (0u64..1u64).prop_map(|val| IndexingOp::TwoVal { val }),
-            (0u64..1u64).prop_map(|_| IndexingOp::Commit),
-        ]
-    }
-
-    use proptest::prelude::*;
-    proptest! {
-        #[test]
-        fn test_merge_columnar_int_proptest(ops in proptest::collection::vec(balanced_operation_strategy(), 1..20)) {
-            assert!(test_merge_int_fields(&ops[..]).is_ok());
-        }
-    }
-    fn test_merge_int_fields(ops: &[IndexingOp]) -> crate::Result<()> {
-        if ops.iter().all(|op| *op == IndexingOp::Commit) {
-            return Ok(());
-        }
-        let expected_doc_and_vals: Vec<(u32, Vec<u64>)> = ops
-            .iter()
-            .filter(|op| *op != &IndexingOp::Commit)
-            .map(|op| match op {
-                IndexingOp::ZeroVal => vec![],
-                IndexingOp::OneVal { val } => vec![*val],
-                IndexingOp::TwoVal { val } => vec![*val, *val],
-                IndexingOp::Commit => unreachable!(),
-            })
-            .enumerate()
-            .map(|(id, val)| (id as u32, val))
-            .collect();
-
-        let mut schema_builder = schema::Schema::builder();
-        let int_options = NumericOptions::default().set_fast().set_indexed();
-        let int_field = schema_builder.add_u64_field("intvals", int_options);
-        let index = Index::create_in_ram(schema_builder.build());
-        {
-            let mut index_writer = index.writer_for_tests()?;
-            index_writer.set_merge_policy(Box::new(NoMergePolicy));
-            let index_doc = |index_writer: &mut IndexWriter, int_vals: &[u64]| {
-                let mut doc = TantivyDocument::default();
-                for &val in int_vals {
-                    doc.add_u64(int_field, val);
-                }
-                index_writer.add_document(doc).unwrap();
-            };
-
-            for op in ops {
-                match op {
-                    IndexingOp::ZeroVal => index_doc(&mut index_writer, &[]),
-                    IndexingOp::OneVal { val } => index_doc(&mut index_writer, &[*val]),
-                    IndexingOp::TwoVal { val } => index_doc(&mut index_writer, &[*val, *val]),
-                    IndexingOp::Commit => {
-                        index_writer.commit().expect("commit failed");
-                    }
-                }
-            }
-            index_writer.commit().expect("commit failed");
-        }
-        {
-            let mut segment_ids = index.searchable_segment_ids()?;
-            segment_ids.sort();
-            let mut index_writer: IndexWriter = index.writer_for_tests()?;
-            index_writer.merge(&segment_ids).wait()?;
-            index_writer.wait_merging_threads()?;
-        }
-        let reader = index.reader()?;
-        reader.reload()?;
-
-        let mut vals: Vec<u64> = Vec::new();
-        let mut test_vals = move |col: &Column<u64>, doc: DocId, expected: &[u64]| {
-            vals.clear();
-            vals.extend(col.values_for_doc(doc));
-            assert_eq!(&vals[..], expected);
-        };
-
-        let mut test_col = move |col: &Column<u64>, column_expected: &[(u32, Vec<u64>)]| {
-            for (doc_id, vals) in column_expected.iter() {
-                test_vals(col, *doc_id, vals);
-            }
-        };
-
-        {
-            let searcher = reader.searcher();
-            let segment = searcher.segment_reader(0u32);
-            let col = segment
-                .fast_fields()
-                .column_opt::<u64>("intvals")
-                .unwrap()
-                .unwrap();
-
-            test_col(&col, &expected_doc_and_vals);
-        }
-
-        Ok(())
-    }
-
     #[test]
     fn test_merge_multivalued_int_fields_simple() -> crate::Result<()> {
         let mut schema_builder = schema::Schema::builder();

src/indexer/segment_writer.rs

@@ -202,8 +202,9 @@ impl SegmentWriter {
             match field_entry.field_type() {
                 FieldType::Facet(_) => {
                     let mut facet_tokenizer = FacetTokenizer::default(); // this can be global
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         let facet_str = value.as_facet().ok_or_else(make_schema_error)?;
                         let mut facet_tokenizer = facet_tokenizer.token_stream(facet_str);
@@ -219,14 +220,15 @@ impl SegmentWriter {
                 }
                 FieldType::Str(_) => {
                     let mut indexing_position = IndexingPosition::default();
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         let mut token_stream = if let Some(text) = value.as_str() {
                             let text_analyzer =
                                 &mut self.per_field_text_analyzers[field.field_id() as usize];
                             text_analyzer.token_stream(text)
-                        } else if let Some(tok_str) = value.into_pre_tokenized_text() {
+                        } else if let Some(tok_str) = value.as_pre_tokenized_text() {
                             BoxTokenStream::new(PreTokenizedStream::from(*tok_str.clone()))
                         } else {
                             continue;
@@ -248,8 +250,9 @@ impl SegmentWriter {
                 }
                 FieldType::U64(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         num_vals += 1;
                         let u64_val = value.as_u64().ok_or_else(make_schema_error)?;
@@ -262,8 +265,10 @@ impl SegmentWriter {
                 }
                 FieldType::Date(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value_access = value_access as D::Value<'_>;
+                        let value = value_access.as_value();
 
                         num_vals += 1;
                         let date_val = value.as_datetime().ok_or_else(make_schema_error)?;
@@ -277,8 +282,9 @@ impl SegmentWriter {
                 }
                 FieldType::I64(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         num_vals += 1;
                         let i64_val = value.as_i64().ok_or_else(make_schema_error)?;
@@ -291,8 +297,10 @@ impl SegmentWriter {
                 }
                 FieldType::F64(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
+
                         num_vals += 1;
                         let f64_val = value.as_f64().ok_or_else(make_schema_error)?;
                         term_buffer.set_f64(f64_val);
@@ -304,8 +312,10 @@ impl SegmentWriter {
                 }
                 FieldType::Bool(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
+
                         num_vals += 1;
                         let bool_val = value.as_bool().ok_or_else(make_schema_error)?;
                         term_buffer.set_bool(bool_val);
@@ -317,8 +327,10 @@ impl SegmentWriter {
                 }
                 FieldType::Bytes(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
+
                         num_vals += 1;
                         let bytes = value.as_bytes().ok_or_else(make_schema_error)?;
                         term_buffer.set_bytes(bytes);
@@ -352,8 +364,9 @@ impl SegmentWriter {
                 }
                 FieldType::IpAddr(_) => {
                     let mut num_vals = 0;
-                    for value in values {
-                        let value = value.as_value();
+                    for value_access in values {
+                        // Used to help with linting and type checking.
+                        let value = value_access as D::Value<'_>;
 
                         num_vals += 1;
                         let ip_addr = value.as_ip_addr().ok_or_else(make_schema_error)?;

src/lib.rs

@@ -397,20 +397,16 @@ pub mod tests {
     #[macro_export]
     macro_rules! assert_nearly_equals {
         ($left:expr, $right:expr) => {{
-            assert_nearly_equals!($left, $right, 0.0005);
-        }};
-        ($left:expr, $right:expr, $epsilon:expr) => {{
-            match (&$left, &$right, &$epsilon) {
-                (left_val, right_val, epsilon_val) => {
+            match (&$left, &$right) {
+                (left_val, right_val) => {
                     let diff = (left_val - right_val).abs();
-
-                    if diff > *epsilon_val {
+                    let add = left_val.abs() + right_val.abs();
+                    if diff > 0.0005 * add {
                         panic!(
-                            r#"assertion failed: `abs(left-right)>epsilon`
-    left: `{:?}`,
-    right: `{:?}`,
-    epsilon: `{:?}`"#,
-                            &*left_val, &*right_val, &*epsilon_val
+                            r#"assertion failed: `(left ~= right)`
+  left: `{:?}`,
+ right: `{:?}`"#,
+                            &*left_val, &*right_val
                         )
                     }
                 }

src/query/query_parser/logical_ast.rs

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

src/query/query_parser/query_parser.rs

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

src/schema/document/value.rs

@@ -17,6 +17,15 @@ pub trait Value<'a>: Send + Sync + Debug {
     /// Returns the field value represented by an enum which borrows it's data.
     fn as_value(&self) -> ReferenceValue<'a, Self>;
 
+    #[inline]
+    /// Returns if the value is `null` or not.
+    fn is_null(&self) -> bool {
+        matches!(
+            self.as_value(),
+            ReferenceValue::Leaf(ReferenceValueLeaf::Null)
+        )
+    }
+
     #[inline]
     /// If the Value is a leaf, returns the associated leaf. Returns None otherwise.
     fn as_leaf(&self) -> Option<ReferenceValueLeaf<'a>> {
@@ -108,6 +117,18 @@ pub trait Value<'a>: Send + Sync + Debug {
             None
         }
     }
+
+    #[inline]
+    /// Returns true if the Value is an array.
+    fn is_array(&self) -> bool {
+        matches!(self.as_value(), ReferenceValue::Object(_))
+    }
+
+    #[inline]
+    /// Returns true if the Value is an object.
+    fn is_object(&self) -> bool {
+        matches!(self.as_value(), ReferenceValue::Object(_))
+    }
 }
 
 /// A enum representing a leaf value for tantivy to index.