low hanging fruit in optimization

columnar/benches/bench_merge.rs

@@ -22,7 +22,7 @@ impl Display for Card {
     }
 }
 
-const NUM_DOCS: u32 = 1_000_000;
+const NUM_DOCS: u32 = 100_000;
 
 fn generate_columnar(card: Card, num_docs: u32) -> ColumnarReader {
     use tantivy_columnar::ColumnarWriter;
@@ -88,8 +88,12 @@ fn main() {
                 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);
+                let _ = black_box(merge_columnar(
+                    &columnar_readers,
+                    &[],
+                    merge_row_order.into(),
+                    &mut out,
+                ));
             },
         );
     }

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/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.

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/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/query/phrase_prefix_query/phrase_prefix_scorer.rs

@@ -2,7 +2,7 @@ use crate::docset::{DocSet, TERMINATED};
 use crate::fieldnorm::FieldNormReader;
 use crate::postings::Postings;
 use crate::query::bm25::Bm25Weight;
-use crate::query::phrase_query::{intersection_count, PhraseScorer};
+use crate::query::phrase_query::{intersection_count, intersection_exists, PhraseScorer};
 use crate::query::Scorer;
 use crate::{DocId, Score};
 
@@ -92,14 +92,17 @@ impl<TPostings: Postings> Scorer for PhraseKind<TPostings> {
     }
 }
 
-pub struct PhrasePrefixScorer<TPostings: Postings> {
+pub struct PhrasePrefixScorer<TPostings: Postings, const SCORING_ENABLED: bool> {
     phrase_scorer: PhraseKind<TPostings>,
     suffixes: Vec<TPostings>,
     suffix_offset: u32,
     phrase_count: u32,
+    suffix_position_buffer: Vec<u32>,
 }
 
-impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
+impl<TPostings: Postings, const SCORING_ENABLED: bool>
+    PhrasePrefixScorer<TPostings, SCORING_ENABLED>
+{
     // If similarity_weight is None, then scoring is disabled.
     pub fn new(
         mut term_postings: Vec<(usize, TPostings)>,
@@ -107,7 +110,7 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
         fieldnorm_reader: FieldNormReader,
         suffixes: Vec<TPostings>,
         suffix_pos: usize,
-    ) -> PhrasePrefixScorer<TPostings> {
+    ) -> PhrasePrefixScorer<TPostings, SCORING_ENABLED> {
         // correct indices so we can merge with our suffix term the PhraseScorer doesn't know about
         let max_offset = term_postings
             .iter()
@@ -140,6 +143,7 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
             suffixes,
             suffix_offset: (max_offset - suffix_pos) as u32,
             phrase_count: 0,
+            suffix_position_buffer: Vec::with_capacity(100),
         };
         if phrase_prefix_scorer.doc() != TERMINATED && !phrase_prefix_scorer.matches_prefix() {
             phrase_prefix_scorer.advance();
@@ -153,7 +157,6 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
 
     fn matches_prefix(&mut self) -> bool {
         let mut count = 0;
-        let mut positions = Vec::new();
         let current_doc = self.doc();
         let pos_matching = self.phrase_scorer.get_intersection();
         for suffix in &mut self.suffixes {
@@ -162,16 +165,27 @@ impl<TPostings: Postings> PhrasePrefixScorer<TPostings> {
             }
             let doc = suffix.seek(current_doc);
             if doc == current_doc {
-                suffix.positions_with_offset(self.suffix_offset, &mut positions);
-                count += intersection_count(pos_matching, &positions);
+                suffix.positions_with_offset(self.suffix_offset, &mut self.suffix_position_buffer);
+                if SCORING_ENABLED {
+                    count += intersection_count(pos_matching, &self.suffix_position_buffer);
+                } else {
+                    if intersection_exists(pos_matching, &self.suffix_position_buffer) {
+                        return true;
+                    }
+                }
             }
         }
+        if !SCORING_ENABLED {
+            return false;
+        }
         self.phrase_count = count as u32;
         count != 0
     }
 }
 
-impl<TPostings: Postings> DocSet for PhrasePrefixScorer<TPostings> {
+impl<TPostings: Postings, const SCORING_ENABLED: bool> DocSet
+    for PhrasePrefixScorer<TPostings, SCORING_ENABLED>
+{
     fn advance(&mut self) -> DocId {
         loop {
             let doc = self.phrase_scorer.advance();
@@ -198,9 +212,15 @@ impl<TPostings: Postings> DocSet for PhrasePrefixScorer<TPostings> {
     }
 }
 
-impl<TPostings: Postings> Scorer for PhrasePrefixScorer<TPostings> {
+impl<TPostings: Postings, const SCORING_ENABLED: bool> Scorer
+    for PhrasePrefixScorer<TPostings, SCORING_ENABLED>
+{
     fn score(&mut self) -> Score {
+        if SCORING_ENABLED {
+            self.phrase_scorer.score()
+        } else {
+            1.0f32
+        }
         // TODO modify score??
-        self.phrase_scorer.score()
     }
 }

src/query/phrase_prefix_query/phrase_prefix_weight.rs

@@ -42,11 +42,11 @@ impl PhrasePrefixWeight {
         Ok(FieldNormReader::constant(reader.max_doc(), 1))
     }
 
-    pub(crate) fn phrase_scorer(
+    pub(crate) fn phrase_prefix_scorer<const SCORING_ENABLED: bool>(
         &self,
         reader: &SegmentReader,
         boost: Score,
-    ) -> crate::Result<Option<PhrasePrefixScorer<SegmentPostings>>> {
+    ) -> crate::Result<Option<PhrasePrefixScorer<SegmentPostings, SCORING_ENABLED>>> {
         let similarity_weight_opt = self
             .similarity_weight_opt
             .as_ref()
@@ -128,15 +128,20 @@ impl PhrasePrefixWeight {
 
 impl Weight for PhrasePrefixWeight {
     fn scorer(&self, reader: &SegmentReader, boost: Score) -> crate::Result<Box<dyn Scorer>> {
-        if let Some(scorer) = self.phrase_scorer(reader, boost)? {
-            Ok(Box::new(scorer))
+        if self.similarity_weight_opt.is_some() {
+            if let Some(scorer) = self.phrase_prefix_scorer::<true>(reader, boost)? {
+                return Ok(Box::new(scorer));
+            }
         } else {
-            Ok(Box::new(EmptyScorer))
+            if let Some(scorer) = self.phrase_prefix_scorer::<false>(reader, boost)? {
+                return Ok(Box::new(scorer));
+            }
         }
+        Ok(Box::new(EmptyScorer))
     }
 
     fn explain(&self, reader: &SegmentReader, doc: DocId) -> crate::Result<Explanation> {
-        let scorer_opt = self.phrase_scorer(reader, 1.0)?;
+        let scorer_opt = self.phrase_prefix_scorer::<true>(reader, 1.0)?;
         if scorer_opt.is_none() {
             return Err(does_not_match(doc));
         }
@@ -200,7 +205,7 @@ mod tests {
             .unwrap()
             .unwrap();
         let mut phrase_scorer = phrase_weight
-            .phrase_scorer(searcher.segment_reader(0u32), 1.0)?
+            .phrase_prefix_scorer::<true>(searcher.segment_reader(0u32), 1.0)?
             .unwrap();
         assert_eq!(phrase_scorer.doc(), 1);
         assert_eq!(phrase_scorer.phrase_count(), 2);
@@ -211,6 +216,38 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    pub fn test_phrase_no_count() -> crate::Result<()> {
+        let index = create_index(&[
+            "aa bb dd cc",
+            "aa aa bb c dd aa bb cc aa bb dc",
+            " aa bb cd",
+        ])?;
+        let schema = index.schema();
+        let text_field = schema.get_field("text").unwrap();
+        let searcher = index.reader()?.searcher();
+        let phrase_query = PhrasePrefixQuery::new(vec![
+            Term::from_field_text(text_field, "aa"),
+            Term::from_field_text(text_field, "bb"),
+            Term::from_field_text(text_field, "c"),
+        ]);
+        let enable_scoring = EnableScoring::enabled_from_searcher(&searcher);
+        let phrase_weight = phrase_query
+            .phrase_prefix_query_weight(enable_scoring)
+            .unwrap()
+            .unwrap();
+        let mut phrase_scorer = phrase_weight
+            .phrase_prefix_scorer::<false>(searcher.segment_reader(0u32), 1.0)?
+            .unwrap();
+        assert_eq!(phrase_scorer.doc(), 1);
+        assert_eq!(phrase_scorer.phrase_count(), 0);
+        assert_eq!(phrase_scorer.advance(), 2);
+        assert_eq!(phrase_scorer.doc(), 2);
+        assert_eq!(phrase_scorer.phrase_count(), 0);
+        assert_eq!(phrase_scorer.advance(), TERMINATED);
+        Ok(())
+    }
+
     #[test]
     pub fn test_phrase_count_mid() -> crate::Result<()> {
         let index = create_index(&["aa dd cc", "aa aa bb c dd aa bb cc aa dc", " aa bb cd"])?;
@@ -227,7 +264,7 @@ mod tests {
             .unwrap()
             .unwrap();
         let mut phrase_scorer = phrase_weight
-            .phrase_scorer(searcher.segment_reader(0u32), 1.0)?
+            .phrase_prefix_scorer::<true>(searcher.segment_reader(0u32), 1.0)?
             .unwrap();
         assert_eq!(phrase_scorer.doc(), 1);
         assert_eq!(phrase_scorer.phrase_count(), 2);

src/query/phrase_query/mod.rs

@@ -3,8 +3,8 @@ mod phrase_scorer;
 mod phrase_weight;
 
 pub use self::phrase_query::PhraseQuery;
-pub(crate) use self::phrase_scorer::intersection_count;
 pub use self::phrase_scorer::PhraseScorer;
+pub(crate) use self::phrase_scorer::{intersection_count, intersection_exists};
 pub use self::phrase_weight::PhraseWeight;
 
 #[cfg(test)]

src/query/phrase_query/phrase_scorer.rs

@@ -58,7 +58,7 @@ pub struct PhraseScorer<TPostings: Postings> {
 }
 
 /// Returns true if and only if the two sorted arrays contain a common element
-fn intersection_exists(left: &[u32], right: &[u32]) -> bool {
+pub(crate) fn intersection_exists(left: &[u32], right: &[u32]) -> bool {
     let mut left_index = 0;
     let mut right_index = 0;
     while left_index < left.len() && right_index < right.len() {

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.