Removed azure stuff

.gitignore

@@ -13,5 +13,3 @@ benchmark
 .idea
 trace.dat
 cargo-timing*
-control
-variable

Cargo.toml

@@ -16,7 +16,6 @@ rust-version = "1.62"
 [dependencies]
 oneshot = "0.1.5"
 base64 = "0.21.0"
-byteorder = "1.4.3"
 crc32fast = "1.3.2"
 once_cell = "1.10.0"
 regex = { version = "1.5.5", default-features = false, features = ["std", "unicode"] }
@@ -32,7 +31,7 @@ log = "0.4.16"
 serde = { version = "1.0.136", features = ["derive"] }
 serde_json = "1.0.79"
 num_cpus = "1.13.1"
-fs4 = { version = "0.6.3", optional = true }
+fs2 = { version = "0.4.3", optional = true }
 levenshtein_automata = "0.2.1"
 uuid = { version = "1.0.0", features = ["v4", "serde"] }
 crossbeam-channel = "0.5.4"
@@ -48,7 +47,7 @@ murmurhash32 = "0.3.0"
 time = { version = "0.3.10", features = ["serde-well-known"] }
 smallvec = "1.8.0"
 rayon = "1.5.2"
-lru = "0.10.0"
+lru = "0.9.0"
 fastdivide = "0.4.0"
 itertools = "0.10.3"
 measure_time = "0.8.2"
@@ -94,7 +93,7 @@ overflow-checks = true
 
 [features]
 default = ["mmap", "stopwords", "lz4-compression"]
-mmap = ["fs4", "tempfile", "memmap2"]
+mmap = ["fs2", "tempfile", "memmap2"]
 stopwords = []
 
 brotli-compression = ["brotli"]

columnar/Cargo.toml

@@ -23,7 +23,6 @@ serde = "1.0.152"
 proptest = "1"
 more-asserts = "0.3.1"
 rand = "0.8.5"
-serde_json  = "1"
 
 [features]
 unstable = []

columnar/src/block_accessor.rs

@@ -1,36 +0,0 @@
-use crate::{Column, DocId, RowId};
-
-#[derive(Debug, Default, Clone)]
-pub struct ColumnBlockAccessor<T> {
-    val_cache: Vec<T>,
-    docid_cache: Vec<DocId>,
-    row_id_cache: Vec<RowId>,
-}
-
-impl<T: PartialOrd + Copy + std::fmt::Debug + Send + Sync + 'static + Default>
-    ColumnBlockAccessor<T>
-{
-    #[inline]
-    pub fn fetch_block(&mut self, docs: &[u32], accessor: &Column<T>) {
-        self.docid_cache.clear();
-        self.row_id_cache.clear();
-        accessor.row_ids_for_docs(docs, &mut self.docid_cache, &mut self.row_id_cache);
-        self.val_cache.resize(self.row_id_cache.len(), T::default());
-        accessor
-            .values
-            .get_vals(&self.row_id_cache, &mut self.val_cache);
-    }
-
-    #[inline]
-    pub fn iter_vals(&self) -> impl Iterator<Item = T> + '_ {
-        self.val_cache.iter().cloned()
-    }
-
-    #[inline]
-    pub fn iter_docid_vals(&self) -> impl Iterator<Item = (DocId, T)> + '_ {
-        self.docid_cache
-            .iter()
-            .cloned()
-            .zip(self.val_cache.iter().cloned())
-    }
-}

columnar/src/column/dictionary_encoded.rs

@@ -1,6 +1,6 @@
+use std::io;
 use std::ops::Deref;
 use std::sync::Arc;
-use std::{fmt, io};
 
 use sstable::{Dictionary, VoidSSTable};
 
@@ -21,14 +21,6 @@ pub struct BytesColumn {
     pub(crate) term_ord_column: Column<u64>,
 }
 
-impl fmt::Debug for BytesColumn {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        f.debug_struct("BytesColumn")
-            .field("term_ord_column", &self.term_ord_column)
-            .finish()
-    }
-}
-
 impl BytesColumn {
     /// Fills the given `output` buffer with the term associated to the ordinal `ord`.
     ///
@@ -64,12 +56,6 @@ impl BytesColumn {
 #[derive(Clone)]
 pub struct StrColumn(BytesColumn);
 
-impl fmt::Debug for StrColumn {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        write!(f, "{:?}", self.term_ord_column)
-    }
-}
-
 impl From<StrColumn> for BytesColumn {
     fn from(str_column: StrColumn) -> BytesColumn {
         str_column.0

columnar/src/column/mod.rs

@@ -1,7 +1,6 @@
 mod dictionary_encoded;
 mod serialize;
 
-use core::fmt;
 use std::fmt::Debug;
 use std::io::Write;
 use std::ops::{Deref, Range, RangeInclusive};
@@ -17,33 +16,14 @@ pub use serialize::{
 use crate::column_index::ColumnIndex;
 use crate::column_values::monotonic_mapping::StrictlyMonotonicMappingToInternal;
 use crate::column_values::{monotonic_map_column, ColumnValues};
-use crate::{Cardinality, DocId, EmptyColumnValues, MonotonicallyMappableToU64, RowId};
+use crate::{Cardinality, MonotonicallyMappableToU64, RowId};
 
 #[derive(Clone)]
 pub struct Column<T = u64> {
-    pub index: ColumnIndex,
+    pub idx: ColumnIndex,
     pub values: Arc<dyn ColumnValues<T>>,
 }
 
-impl<T: fmt::Debug + PartialOrd + Send + Sync + Copy + 'static> fmt::Debug for Column<T> {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        let num_docs = self.num_docs();
-        let entries = (0..num_docs)
-            .map(|i| (i, self.values_for_doc(i).collect::<Vec<_>>()))
-            .filter(|(_, vals)| !vals.is_empty());
-        f.debug_map().entries(entries).finish()
-    }
-}
-
-impl<T: PartialOrd + Default> Column<T> {
-    pub fn build_empty_column(num_docs: u32) -> Column<T> {
-        Column {
-            index: ColumnIndex::Empty { num_docs },
-            values: Arc::new(EmptyColumnValues),
-        }
-    }
-}
-
 impl<T: MonotonicallyMappableToU64> Column<T> {
     pub fn to_u64_monotonic(self) -> Column<u64> {
         let values = Arc::new(monotonic_map_column(
@@ -51,7 +31,7 @@ impl<T: MonotonicallyMappableToU64> Column<T> {
             StrictlyMonotonicMappingToInternal::<T>::new(),
         ));
         Column {
-            index: self.index,
+            idx: self.idx,
             values,
         }
     }
@@ -60,11 +40,11 @@ impl<T: MonotonicallyMappableToU64> Column<T> {
 impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
     #[inline]
     pub fn get_cardinality(&self) -> Cardinality {
-        self.index.get_cardinality()
+        self.idx.get_cardinality()
     }
 
     pub fn num_docs(&self) -> RowId {
-        match &self.index {
+        match &self.idx {
             ColumnIndex::Empty { num_docs } => *num_docs,
             ColumnIndex::Full => self.values.num_vals(),
             ColumnIndex::Optional(optional_index) => optional_index.num_docs(),
@@ -88,25 +68,8 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
         self.values_for_doc(row_id).next()
     }
 
-    /// Translates a block of docis to row_ids.
-    ///
-    /// returns the row_ids and the matching docids on the same index
-    /// e.g.
-    /// DocId In:  [0, 5, 6]
-    /// DocId Out: [0, 0, 6, 6]
-    /// RowId Out: [0, 1, 2, 3]
-    #[inline]
-    pub fn row_ids_for_docs(
-        &self,
-        doc_ids: &[DocId],
-        doc_ids_out: &mut Vec<DocId>,
-        row_ids: &mut Vec<RowId>,
-    ) {
-        self.index.docids_to_rowids(doc_ids, doc_ids_out, row_ids)
-    }
-
-    pub fn values_for_doc(&self, doc_id: DocId) -> impl Iterator<Item = T> + '_ {
-        self.value_row_ids(doc_id)
+    pub fn values_for_doc(&self, row_id: RowId) -> impl Iterator<Item = T> + '_ {
+        self.value_row_ids(row_id)
             .map(|value_row_id: RowId| self.values.get_val(value_row_id))
     }
 
@@ -119,15 +82,13 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
         doc_ids: &mut Vec<u32>,
     ) {
         // convert passed docid range to row id range
-        let rowid_range = self
-            .index
-            .docid_range_to_rowids(selected_docid_range.clone());
+        let rowid_range = self.idx.docid_range_to_rowids(selected_docid_range.clone());
 
         // Load rows
         self.values
             .get_row_ids_for_value_range(value_range, rowid_range, doc_ids);
         // Convert rows to docids
-        self.index
+        self.idx
             .select_batch_in_place(selected_docid_range.start, doc_ids);
     }
 
@@ -152,7 +113,7 @@ impl<T> Deref for Column<T> {
     type Target = ColumnIndex;
 
     fn deref(&self) -> &Self::Target {
-        &self.index
+        &self.idx
     }
 }
 
@@ -190,7 +151,7 @@ impl<T: PartialOrd + Debug + Send + Sync + Copy + 'static> ColumnValues<T>
     }
 
     fn num_vals(&self) -> u32 {
-        match &self.column.index {
+        match &self.column.idx {
             ColumnIndex::Empty { .. } => 0u32,
             ColumnIndex::Full => self.column.values.num_vals(),
             ColumnIndex::Optional(optional_idx) => optional_idx.num_docs(),

columnar/src/column/serialize.rs

@@ -52,7 +52,7 @@ pub fn open_column_u64<T: MonotonicallyMappableToU64>(bytes: OwnedBytes) -> io::
     let column_index = crate::column_index::open_column_index(column_index_data)?;
     let column_values = load_u64_based_column_values(column_values_data)?;
     Ok(Column {
-        index: column_index,
+        idx: column_index,
         values: column_values,
     })
 }
@@ -71,7 +71,7 @@ pub fn open_column_u128<T: MonotonicallyMappableToU128>(
     let column_index = crate::column_index::open_column_index(column_index_data)?;
     let column_values = crate::column_values::open_u128_mapped(column_values_data)?;
     Ok(Column {
-        index: column_index,
+        idx: column_index,
         values: column_values,
     })
 }

columnar/src/column_index/merge/mod.rs

@@ -8,16 +8,17 @@ use crate::column_index::SerializableColumnIndex;
 use crate::{Cardinality, ColumnIndex, MergeRowOrder};
 
 // For simplification, we never have cardinality go down due to deletes.
-fn detect_cardinality(columns: &[ColumnIndex]) -> Cardinality {
+fn detect_cardinality(columns: &[Option<ColumnIndex>]) -> Cardinality {
     columns
         .iter()
+        .flatten()
         .map(ColumnIndex::get_cardinality)
         .max()
         .unwrap_or(Cardinality::Full)
 }
 
 pub fn merge_column_index<'a>(
-    columns: &'a [ColumnIndex],
+    columns: &'a [Option<ColumnIndex>],
     merge_row_order: &'a MergeRowOrder,
 ) -> SerializableColumnIndex<'a> {
     // For simplification, we do not try to detect whether the cardinality could be
@@ -52,33 +53,34 @@ mod tests {
         let optional_index: ColumnIndex = OptionalIndex::for_test(1, &[]).into();
         let multivalued_index: ColumnIndex = MultiValueIndex::for_test(&[0, 1]).into();
         assert_eq!(
-            detect_cardinality(&[optional_index.clone(), ColumnIndex::Empty { num_docs: 0 }]),
+            detect_cardinality(&[Some(optional_index.clone()), None]),
             Cardinality::Optional
         );
         assert_eq!(
-            detect_cardinality(&[optional_index.clone(), ColumnIndex::Full]),
+            detect_cardinality(&[Some(optional_index.clone()), Some(ColumnIndex::Full)]),
             Cardinality::Optional
         );
+        assert_eq!(
+            detect_cardinality(&[Some(multivalued_index.clone()), None]),
+            Cardinality::Multivalued
+        );
         assert_eq!(
             detect_cardinality(&[
-                multivalued_index.clone(),
-                ColumnIndex::Empty { num_docs: 0 }
+                Some(multivalued_index.clone()),
+                Some(optional_index.clone())
             ]),
             Cardinality::Multivalued
         );
         assert_eq!(
-            detect_cardinality(&[multivalued_index.clone(), optional_index.clone()]),
-            Cardinality::Multivalued
-        );
-        assert_eq!(
-            detect_cardinality(&[optional_index, multivalued_index]),
+            detect_cardinality(&[Some(optional_index), Some(multivalued_index)]),
             Cardinality::Multivalued
         );
     }
 
     #[test]
     fn test_merge_index_multivalued_sorted() {
-        let column_indexes: Vec<ColumnIndex> = vec![MultiValueIndex::for_test(&[0, 2, 5]).into()];
+        let column_indexes: Vec<Option<ColumnIndex>> =
+            vec![Some(MultiValueIndex::for_test(&[0, 2, 5]).into())];
         let merge_row_order: MergeRowOrder = ShuffleMergeOrder::for_test(
             &[2],
             vec![
@@ -102,10 +104,10 @@ mod tests {
 
     #[test]
     fn test_merge_index_multivalued_sorted_several_segment() {
-        let column_indexes: Vec<ColumnIndex> = vec![
-            MultiValueIndex::for_test(&[0, 2, 5]).into(),
-            ColumnIndex::Empty { num_docs: 0 },
-            MultiValueIndex::for_test(&[0, 1, 4]).into(),
+        let column_indexes: Vec<Option<ColumnIndex>> = vec![
+            Some(MultiValueIndex::for_test(&[0, 2, 5]).into()),
+            None,
+            Some(MultiValueIndex::for_test(&[0, 1, 4]).into()),
         ];
         let merge_row_order: MergeRowOrder = ShuffleMergeOrder::for_test(
             &[2, 0, 2],

columnar/src/column_index/merge/shuffled.rs

@@ -5,7 +5,7 @@ use crate::iterable::Iterable;
 use crate::{Cardinality, ColumnIndex, RowId, ShuffleMergeOrder};
 
 pub fn merge_column_index_shuffled<'a>(
-    column_indexes: &'a [ColumnIndex],
+    column_indexes: &'a [Option<ColumnIndex>],
     cardinality_after_merge: Cardinality,
     shuffle_merge_order: &'a ShuffleMergeOrder,
 ) -> SerializableColumnIndex<'a> {
@@ -33,41 +33,41 @@ pub fn merge_column_index_shuffled<'a>(
 ///
 /// In other words the column_indexes passed as argument may NOT be multivalued.
 fn merge_column_index_shuffled_optional<'a>(
-    column_indexes: &'a [ColumnIndex],
+    column_indexes: &'a [Option<ColumnIndex>],
     merge_order: &'a ShuffleMergeOrder,
 ) -> Box<dyn Iterable<RowId> + 'a> {
-    Box::new(ShuffledIndex {
+    Box::new(ShuffledOptionalIndex {
         column_indexes,
         merge_order,
     })
 }
 
-struct ShuffledIndex<'a> {
-    column_indexes: &'a [ColumnIndex],
+struct ShuffledOptionalIndex<'a> {
+    column_indexes: &'a [Option<ColumnIndex>],
     merge_order: &'a ShuffleMergeOrder,
 }
 
-impl<'a> Iterable<u32> for ShuffledIndex<'a> {
+impl<'a> Iterable<u32> for ShuffledOptionalIndex<'a> {
     fn boxed_iter(&self) -> Box<dyn Iterator<Item = u32> + '_> {
-        Box::new(
-            self.merge_order
-                .iter_new_to_old_row_addrs()
-                .enumerate()
-                .filter_map(|(new_row_id, old_row_addr)| {
-                    let column_index = &self.column_indexes[old_row_addr.segment_ord as usize];
-                    let row_id = new_row_id as u32;
-                    if column_index.has_value(old_row_addr.row_id) {
-                        Some(row_id)
-                    } else {
-                        None
-                    }
-                }),
-        )
+        Box::new(self.merge_order
+        .iter_new_to_old_row_addrs()
+        .enumerate()
+        .filter_map(|(new_row_id, old_row_addr)| {
+            let Some(column_index) = &self.column_indexes[old_row_addr.segment_ord as usize] else {
+                return None;
+            };
+            let row_id = new_row_id as u32;
+            if column_index.has_value(old_row_addr.row_id) {
+                Some(row_id)
+            } else {
+                None
+            }
+        }))
     }
 }
 
 fn merge_column_index_shuffled_multivalued<'a>(
-    column_indexes: &'a [ColumnIndex],
+    column_indexes: &'a [Option<ColumnIndex>],
     merge_order: &'a ShuffleMergeOrder,
 ) -> Box<dyn Iterable<RowId> + 'a> {
     Box::new(ShuffledMultivaluedIndex {
@@ -77,16 +77,19 @@ fn merge_column_index_shuffled_multivalued<'a>(
 }
 
 struct ShuffledMultivaluedIndex<'a> {
-    column_indexes: &'a [ColumnIndex],
+    column_indexes: &'a [Option<ColumnIndex>],
     merge_order: &'a ShuffleMergeOrder,
 }
 
 fn iter_num_values<'a>(
-    column_indexes: &'a [ColumnIndex],
+    column_indexes: &'a [Option<ColumnIndex>],
     merge_order: &'a ShuffleMergeOrder,
 ) -> impl Iterator<Item = u32> + 'a {
     merge_order.iter_new_to_old_row_addrs().map(|row_addr| {
-        let column_index = &column_indexes[row_addr.segment_ord as usize];
+        let Some(column_index) = &column_indexes[row_addr.segment_ord as usize] else {
+                    // No values in the entire column. It surely means there are 0 values associated to this row.
+                    return 0u32;
+                };
         match column_index {
             ColumnIndex::Empty { .. } => 0u32,
             ColumnIndex::Full => 1,
@@ -140,7 +143,7 @@ mod tests {
     #[test]
     fn test_merge_column_index_optional_shuffle() {
         let optional_index: ColumnIndex = OptionalIndex::for_test(2, &[0]).into();
-        let column_indexes = vec![optional_index, ColumnIndex::Full];
+        let column_indexes = vec![Some(optional_index), Some(ColumnIndex::Full)];
         let row_addrs = vec![
             RowAddr {
                 segment_ord: 0u32,

columnar/src/column_index/merge/stacked.rs

@@ -9,7 +9,7 @@ use crate::{Cardinality, ColumnIndex, RowId, StackMergeOrder};
 ///
 /// There are no sort nor deletes involved.
 pub fn merge_column_index_stacked<'a>(
-    columns: &'a [ColumnIndex],
+    columns: &'a [Option<ColumnIndex>],
     cardinality_after_merge: Cardinality,
     stack_merge_order: &'a StackMergeOrder,
 ) -> SerializableColumnIndex<'a> {
@@ -33,7 +33,7 @@ pub fn merge_column_index_stacked<'a>(
 }
 
 struct StackedOptionalIndex<'a> {
-    columns: &'a [ColumnIndex],
+    columns: &'a [Option<ColumnIndex>],
     stack_merge_order: &'a StackMergeOrder,
 }
 
@@ -46,16 +46,16 @@ impl<'a> Iterable<RowId> for StackedOptionalIndex<'a> {
                 .flat_map(|(columnar_id, column_index_opt)| {
                     let columnar_row_range = self.stack_merge_order.columnar_range(columnar_id);
                     let rows_it: Box<dyn Iterator<Item = RowId>> = match column_index_opt {
-                        ColumnIndex::Full => Box::new(columnar_row_range),
-                        ColumnIndex::Optional(optional_index) => Box::new(
+                        Some(ColumnIndex::Full) => Box::new(columnar_row_range),
+                        Some(ColumnIndex::Optional(optional_index)) => Box::new(
                             optional_index
                                 .iter_rows()
                                 .map(move |row_id: RowId| columnar_row_range.start + row_id),
                         ),
-                        ColumnIndex::Multivalued(_) => {
+                        Some(ColumnIndex::Multivalued(_)) => {
                             panic!("No multivalued index is allowed when stacking column index");
                         }
-                        ColumnIndex::Empty { .. } => Box::new(std::iter::empty()),
+                        None | Some(ColumnIndex::Empty { .. }) => Box::new(std::iter::empty()),
                     };
                     rows_it
                 }),
@@ -65,18 +65,20 @@ impl<'a> Iterable<RowId> for StackedOptionalIndex<'a> {
 
 #[derive(Clone, Copy)]
 struct StackedMultivaluedIndex<'a> {
-    columns: &'a [ColumnIndex],
+    columns: &'a [Option<ColumnIndex>],
     stack_merge_order: &'a StackMergeOrder,
 }
 
 fn convert_column_opt_to_multivalued_index<'a>(
-    column_index_opt: &'a ColumnIndex,
+    column_index_opt: Option<&'a ColumnIndex>,
     num_rows: RowId,
 ) -> Box<dyn Iterator<Item = RowId> + 'a> {
     match column_index_opt {
-        ColumnIndex::Empty { .. } => Box::new(iter::repeat(0u32).take(num_rows as usize + 1)),
-        ColumnIndex::Full => Box::new(0..num_rows + 1),
-        ColumnIndex::Optional(optional_index) => {
+        None | Some(ColumnIndex::Empty { .. }) => {
+            Box::new(iter::repeat(0u32).take(num_rows as usize + 1))
+        }
+        Some(ColumnIndex::Full) => Box::new(0..num_rows + 1),
+        Some(ColumnIndex::Optional(optional_index)) => {
             Box::new(
                 (0..num_rows)
                     // TODO optimize
@@ -84,7 +86,9 @@ fn convert_column_opt_to_multivalued_index<'a>(
                     .chain(std::iter::once(optional_index.num_non_nulls())),
             )
         }
-        ColumnIndex::Multivalued(multivalued_index) => multivalued_index.start_index_column.iter(),
+        Some(ColumnIndex::Multivalued(multivalued_index)) => {
+            multivalued_index.start_index_column.iter()
+        }
     }
 }
 
@@ -93,6 +97,7 @@ impl<'a> Iterable<RowId> for StackedMultivaluedIndex<'a> {
         let multivalued_indexes =
             self.columns
                 .iter()
+                .map(Option::as_ref)
                 .enumerate()
                 .map(|(columnar_id, column_opt)| {
                     let num_rows =

columnar/src/column_index/mod.rs

@@ -12,7 +12,7 @@ pub use serialize::{open_column_index, serialize_column_index, SerializableColum
 use crate::column_index::multivalued_index::MultiValueIndex;
 use crate::{Cardinality, DocId, RowId};
 
-#[derive(Clone, Debug)]
+#[derive(Clone)]
 pub enum ColumnIndex {
     Empty {
         num_docs: u32,
@@ -37,15 +37,11 @@ impl From<MultiValueIndex> for ColumnIndex {
 }
 
 impl ColumnIndex {
-    // Returns the cardinality of the column index.
-    //
-    // By convention, if the column contains no docs, we consider that it is
-    // full.
     #[inline]
     pub fn get_cardinality(&self) -> Cardinality {
         match self {
-            ColumnIndex::Empty { num_docs: 0 } | ColumnIndex::Full => Cardinality::Full,
             ColumnIndex::Empty { .. } => Cardinality::Optional,
+            ColumnIndex::Full => Cardinality::Full,
             ColumnIndex::Optional(_) => Cardinality::Optional,
             ColumnIndex::Multivalued(_) => Cardinality::Multivalued,
         }
@@ -78,45 +74,6 @@ impl ColumnIndex {
         }
     }
 
-    /// Translates a block of docis to row_ids.
-    ///
-    /// returns the row_ids and the matching docids on the same index
-    /// e.g.
-    /// DocId In:  [0, 5, 6]
-    /// DocId Out: [0, 0, 6, 6]
-    /// RowId Out: [0, 1, 2, 3]
-    #[inline]
-    pub fn docids_to_rowids(
-        &self,
-        doc_ids: &[DocId],
-        doc_ids_out: &mut Vec<DocId>,
-        row_ids: &mut Vec<RowId>,
-    ) {
-        match self {
-            ColumnIndex::Empty { .. } => {}
-            ColumnIndex::Full => {
-                doc_ids_out.extend_from_slice(doc_ids);
-                row_ids.extend_from_slice(doc_ids);
-            }
-            ColumnIndex::Optional(optional_index) => {
-                for doc_id in doc_ids {
-                    if let Some(row_id) = optional_index.rank_if_exists(*doc_id) {
-                        doc_ids_out.push(*doc_id);
-                        row_ids.push(row_id);
-                    }
-                }
-            }
-            ColumnIndex::Multivalued(multivalued_index) => {
-                for doc_id in doc_ids {
-                    for row_id in multivalued_index.range(*doc_id) {
-                        doc_ids_out.push(*doc_id);
-                        row_ids.push(row_id);
-                    }
-                }
-            }
-        }
-    }
-
     pub fn docid_range_to_rowids(&self, doc_id: Range<DocId>) -> Range<RowId> {
         match self {
             ColumnIndex::Empty { .. } => 0..0,
@@ -156,21 +113,3 @@ impl ColumnIndex {
         }
     }
 }
-
-#[cfg(test)]
-mod tests {
-    use crate::{Cardinality, ColumnIndex};
-
-    #[test]
-    fn test_column_index_get_cardinality() {
-        assert_eq!(
-            ColumnIndex::Empty { num_docs: 0 }.get_cardinality(),
-            Cardinality::Full
-        );
-        assert_eq!(ColumnIndex::Full.get_cardinality(), Cardinality::Full);
-        assert_eq!(
-            ColumnIndex::Empty { num_docs: 1 }.get_cardinality(),
-            Cardinality::Optional
-        );
-    }
-}

columnar/src/column_index/multivalued_index.rs

@@ -35,14 +35,6 @@ pub struct MultiValueIndex {
     pub start_index_column: Arc<dyn crate::ColumnValues<RowId>>,
 }
 
-impl std::fmt::Debug for MultiValueIndex {
-    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
-        f.debug_struct("MultiValuedIndex")
-            .field("num_rows", &self.start_index_column.num_vals())
-            .finish_non_exhaustive()
-    }
-}
-
 impl From<Arc<dyn ColumnValues<RowId>>> for MultiValueIndex {
     fn from(start_index_column: Arc<dyn ColumnValues<RowId>>) -> Self {
         MultiValueIndex { start_index_column }
@@ -114,8 +106,11 @@ impl MultiValueIndex {
 #[cfg(test)]
 mod tests {
     use std::ops::Range;
+    use std::sync::Arc;
 
     use super::MultiValueIndex;
+    use crate::column_values::IterColumn;
+    use crate::{ColumnValues, RowId};
 
     fn index_to_pos_helper(
         index: &MultiValueIndex,
@@ -129,7 +124,9 @@ mod tests {
 
     #[test]
     fn test_positions_to_docid() {
-        let index = MultiValueIndex::for_test(&[0, 10, 12, 15, 22, 23]);
+        let offsets: Vec<RowId> = vec![0, 10, 12, 15, 22, 23]; // docid values are [0..10, 10..12, 12..15, etc.]
+        let column: Arc<dyn ColumnValues<RowId>> = Arc::new(IterColumn::from(offsets.into_iter()));
+        let index = MultiValueIndex::from(column);
         assert_eq!(index.num_docs(), 5);
         let positions = &[10u32, 11, 15, 20, 21, 22];
         assert_eq!(index_to_pos_helper(&index, 0..5, positions), vec![1, 3, 4]);

columnar/src/column_index/optional_index/mod.rs

@@ -88,15 +88,6 @@ pub struct OptionalIndex {
     block_metas: Arc<[BlockMeta]>,
 }
 
-impl std::fmt::Debug for OptionalIndex {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.debug_struct("OptionalIndex")
-            .field("num_rows", &self.num_rows)
-            .field("num_non_null_rows", &self.num_non_null_rows)
-            .finish_non_exhaustive()
-    }
-}
-
 /// Splits a value address into lower and upper 16bits.
 /// The lower 16 bits are the value in the block
 /// The upper 16 bits are the block index

columnar/src/column_values/merge.rs

@@ -5,7 +5,7 @@ use crate::iterable::Iterable;
 use crate::{ColumnIndex, ColumnValues, MergeRowOrder};
 
 pub(crate) struct MergedColumnValues<'a, T> {
-    pub(crate) column_indexes: &'a [ColumnIndex],
+    pub(crate) column_indexes: &'a [Option<ColumnIndex>],
     pub(crate) column_values: &'a [Option<Arc<dyn ColumnValues<T>>>],
     pub(crate) merge_row_order: &'a MergeRowOrder,
 }
@@ -23,7 +23,8 @@ impl<'a, T: Copy + PartialOrd + Debug> Iterable<T> for MergedColumnValues<'a, T>
                 shuffle_merge_order
                     .iter_new_to_old_row_addrs()
                     .flat_map(|row_addr| {
-                        let column_index = &self.column_indexes[row_addr.segment_ord as usize];
+                        let column_index =
+                            self.column_indexes[row_addr.segment_ord as usize].as_ref()?;
                         let column_values =
                             self.column_values[row_addr.segment_ord as usize].as_ref()?;
                         let value_range = column_index.value_row_ids(row_addr.row_id);

columnar/src/column_values/mod.rs

@@ -58,21 +58,10 @@ pub trait ColumnValues<T: PartialOrd = u64>: Send + Sync {
     /// # Panics
     ///
     /// May panic if `idx` is greater than the column length.
-    fn get_vals(&self, indexes: &[u32], output: &mut [T]) {
-        assert!(indexes.len() == output.len());
-        let out_and_idx_chunks = output.chunks_exact_mut(4).zip(indexes.chunks_exact(4));
-        for (out_x4, idx_x4) in out_and_idx_chunks {
-            out_x4[0] = self.get_val(idx_x4[0]);
-            out_x4[1] = self.get_val(idx_x4[1]);
-            out_x4[2] = self.get_val(idx_x4[2]);
-            out_x4[3] = self.get_val(idx_x4[3]);
-        }
-
-        let step_size = 4;
-        let cutoff = indexes.len() - indexes.len() % step_size;
-
-        for idx in cutoff..indexes.len() {
-            output[idx] = self.get_val(indexes[idx]);
+    fn get_vals(&self, idx: &[u32], output: &mut [T]) {
+        assert!(idx.len() == output.len());
+        for (out, idx) in output.iter_mut().zip(idx.iter()) {
+            *out = self.get_val(*idx as u32);
         }
     }
 
@@ -110,26 +99,20 @@ pub trait ColumnValues<T: PartialOrd = u64>: Send + Sync {
         }
     }
 
-    /// Returns a lower bound for this column of values.
+    /// Returns the minimum value for this fast field.
     ///
-    /// All values are guaranteed to be higher than `.min_value()`
-    /// but this value is not necessary the best boundary value.
-    ///
-    /// We have
-    /// ∀i < self.num_vals(), self.get_val(i) >= self.min_value()
-    /// But we don't have necessarily
-    /// ∃i < self.num_vals(), self.get_val(i) == self.min_value()
+    /// This min_value may not be exact.
+    /// For instance, the min value does not take in account of possible
+    /// deleted document. All values are however guaranteed to be higher than
+    /// `.min_value()`.
     fn min_value(&self) -> T;
 
-    /// Returns an upper bound for this column of values.
+    /// Returns the maximum value for this fast field.
     ///
-    /// All values are guaranteed to be lower than `.max_value()`
-    /// but this value is not necessary the best boundary value.
-    ///
-    /// We have
-    /// ∀i < self.num_vals(), self.get_val(i) <= self.max_value()
-    /// But we don't have necessarily
-    /// ∃i < self.num_vals(), self.get_val(i) == self.max_value()
+    /// This max_value may not be exact.
+    /// For instance, the max value does not take in account of possible
+    /// deleted document. All values are however guaranteed to be higher than
+    /// `.max_value()`.
     fn max_value(&self) -> T;
 
     /// The number of values in the column.
@@ -141,27 +124,6 @@ pub trait ColumnValues<T: PartialOrd = u64>: Send + Sync {
     }
 }
 
-/// Empty column of values.
-pub struct EmptyColumnValues;
-
-impl<T: PartialOrd + Default> ColumnValues<T> for EmptyColumnValues {
-    fn get_val(&self, _idx: u32) -> T {
-        panic!("Internal Error: Called get_val of empty column.")
-    }
-
-    fn min_value(&self) -> T {
-        T::default()
-    }
-
-    fn max_value(&self) -> T {
-        T::default()
-    }
-
-    fn num_vals(&self) -> u32 {
-        0
-    }
-}
-
 impl<T: Copy + PartialOrd + Debug> ColumnValues<T> for Arc<dyn ColumnValues<T>> {
     #[inline(always)]
     fn get_val(&self, idx: u32) -> T {
@@ -205,5 +167,54 @@ impl<T: Copy + PartialOrd + Debug> ColumnValues<T> for Arc<dyn ColumnValues<T>>
     }
 }
 
+/// Wraps an cloneable iterator into a `Column`.
+pub struct IterColumn<T>(T);
+
+impl<T> From<T> for IterColumn<T>
+where T: Iterator + Clone + ExactSizeIterator
+{
+    fn from(iter: T) -> Self {
+        IterColumn(iter)
+    }
+}
+
+impl<T> ColumnValues<T::Item> for IterColumn<T>
+where
+    T: Iterator + Clone + ExactSizeIterator + Send + Sync,
+    T::Item: PartialOrd + Debug,
+{
+    fn get_val(&self, idx: u32) -> T::Item {
+        self.0.clone().nth(idx as usize).unwrap()
+    }
+
+    fn min_value(&self) -> T::Item {
+        self.0.clone().next().unwrap()
+    }
+
+    fn max_value(&self) -> T::Item {
+        self.0.clone().last().unwrap()
+    }
+
+    fn num_vals(&self) -> u32 {
+        self.0.len() as u32
+    }
+
+    fn iter(&self) -> Box<dyn Iterator<Item = T::Item> + '_> {
+        Box::new(self.0.clone())
+    }
+}
+
 #[cfg(all(test, feature = "unstable"))]
 mod bench;
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_range_as_col() {
+        let col = IterColumn::from(10..100);
+        assert_eq!(col.num_vals(), 90);
+        assert_eq!(col.max_value(), 99);
+    }
+}

columnar/src/column_values/monotonic_column.rs

@@ -50,7 +50,7 @@ where
     Input: PartialOrd + Send + Debug + Sync + Clone,
     Output: PartialOrd + Send + Debug + Sync + Clone,
 {
-    #[inline(always)]
+    #[inline]
     fn get_val(&self, idx: u32) -> Output {
         let from_val = self.from_column.get_val(idx);
         self.monotonic_mapping.mapping(from_val)

columnar/src/column_values/u64_based/tests.rs

@@ -1,6 +1,6 @@
 use proptest::prelude::*;
 use proptest::strategy::Strategy;
-use proptest::{prop_oneof, proptest};
+use proptest::{num, prop_oneof, proptest};
 
 #[test]
 fn test_serialize_and_load_simple() {
@@ -99,28 +99,14 @@ pub(crate) fn create_and_validate<TColumnCodec: ColumnCodec>(
 
     let reader = TColumnCodec::load(OwnedBytes::new(buffer)).unwrap();
     assert_eq!(reader.num_vals(), vals.len() as u32);
-    let mut buffer = Vec::new();
     for (doc, orig_val) in vals.iter().copied().enumerate() {
         let val = reader.get_val(doc as u32);
         assert_eq!(
             val, orig_val,
             "val `{val}` does not match orig_val {orig_val:?}, in data set {name}, data `{vals:?}`",
         );
-
-        buffer.resize(1, 0);
-        reader.get_vals(&[doc as u32], &mut buffer);
-        let val = buffer[0];
-        assert_eq!(
-            val, orig_val,
-            "val `{val}` does not match orig_val {orig_val:?}, in data set {name}, data `{vals:?}`",
-        );
     }
 
-    let all_docs: Vec<u32> = (0..vals.len() as u32).collect();
-    buffer.resize(all_docs.len(), 0);
-    reader.get_vals(&all_docs, &mut buffer);
-    assert_eq!(vals, buffer);
-
     if !vals.is_empty() {
         let test_rand_idx = rand::thread_rng().gen_range(0..=vals.len() - 1);
         let expected_positions: Vec<u32> = vals

columnar/src/columnar/column_type.rs

@@ -1,4 +1,3 @@
-use std::fmt;
 use std::fmt::Debug;
 use std::net::Ipv6Addr;
 
@@ -22,22 +21,6 @@ pub enum ColumnType {
     DateTime = 7u8,
 }
 
-impl fmt::Display for ColumnType {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        let short_str = match self {
-            ColumnType::I64 => "i64",
-            ColumnType::U64 => "u64",
-            ColumnType::F64 => "f64",
-            ColumnType::Bytes => "bytes",
-            ColumnType::Str => "str",
-            ColumnType::Bool => "bool",
-            ColumnType::IpAddr => "ip",
-            ColumnType::DateTime => "datetime",
-        };
-        write!(f, "{}", short_str)
-    }
-}
-
 // The order needs to match _exactly_ the order in the enum
 const COLUMN_TYPES: [ColumnType; 8] = [
     ColumnType::I64,

columnar/src/columnar/merge/merge_dict_column.rs

@@ -52,18 +52,21 @@ impl<'a> Iterable for RemappedTermOrdinalsValues<'a> {
 
 impl<'a> RemappedTermOrdinalsValues<'a> {
     fn boxed_iter_stacked(&self) -> Box<dyn Iterator<Item = u64> + '_> {
-        let iter = self.bytes_columns.iter().flatten().enumerate().flat_map(
-            move |(seg_ord_with_column, bytes_column)| {
-                let term_ord_after_merge_mapping = self
-                    .term_ord_mapping
-                    .get_segment(seg_ord_with_column as u32);
-                bytes_column
-                    .ords()
-                    .values
-                    .iter()
-                    .map(move |term_ord| term_ord_after_merge_mapping[term_ord as usize])
-            },
-        );
+        let iter = self
+            .bytes_columns
+            .iter()
+            .enumerate()
+            .flat_map(|(segment_ord, byte_column)| {
+                let segment_ord = self.term_ord_mapping.get_segment(segment_ord as u32);
+                byte_column.iter().flat_map(move |bytes_column| {
+                    bytes_column
+                        .ords()
+                        .values
+                        .iter()
+                        .map(move |term_ord| segment_ord[term_ord as usize])
+                })
+            });
+        // TODO see if we can better decompose the mapping / and the stacking
         Box::new(iter)
     }
 
@@ -130,6 +133,7 @@ fn serialize_merged_dict(
     let mut merged_terms = TermMerger::new(field_term_streams);
     let mut sstable_builder = sstable::VoidSSTable::writer(output);
 
+    // TODO support complex `merge_row_order`.
     match merge_row_order {
         MergeRowOrder::Stack(_) => {
             let mut current_term_ord = 0;

columnar/src/columnar/merge/mod.rs

@@ -28,7 +28,7 @@ use crate::{
 ///
 /// See also [README.md].
 #[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
-pub(crate) enum ColumnTypeCategory {
+enum ColumnTypeCategory {
     Bool,
     Str,
     Numerical,
@@ -78,10 +78,6 @@ pub fn merge_columnar(
     output: &mut impl io::Write,
 ) -> io::Result<()> {
     let mut serializer = ColumnarSerializer::new(output);
-    let num_rows_per_columnar = columnar_readers
-        .iter()
-        .map(|reader| reader.num_rows())
-        .collect::<Vec<u32>>();
 
     let columns_to_merge = group_columns_for_merge(columnar_readers, required_columns)?;
     for ((column_name, column_type), columns) in columns_to_merge {
@@ -89,7 +85,6 @@ pub fn merge_columnar(
             serializer.serialize_column(column_name.as_bytes(), column_type);
         merge_column(
             column_type,
-            &num_rows_per_columnar,
             columns,
             &merge_row_order,
             &mut column_serializer,
@@ -113,7 +108,6 @@ fn dynamic_column_to_u64_monotonic(dynamic_column: DynamicColumn) -> Option<Colu
 
 fn merge_column(
     column_type: ColumnType,
-    num_docs_per_column: &[u32],
     columns: Vec<Option<DynamicColumn>>,
     merge_row_order: &MergeRowOrder,
     wrt: &mut impl io::Write,
@@ -124,19 +118,17 @@ fn merge_column(
         | ColumnType::F64
         | ColumnType::DateTime
         | ColumnType::Bool => {
-            let mut column_indexes: Vec<ColumnIndex> = Vec::with_capacity(columns.len());
+            let mut column_indexes: Vec<Option<ColumnIndex>> = Vec::with_capacity(columns.len());
             let mut column_values: Vec<Option<Arc<dyn ColumnValues>>> =
                 Vec::with_capacity(columns.len());
-            for (i, dynamic_column_opt) in columns.into_iter().enumerate() {
-                if let Some(Column { index: idx, values }) =
+            for dynamic_column_opt in columns {
+                if let Some(Column { idx, values }) =
                     dynamic_column_opt.and_then(dynamic_column_to_u64_monotonic)
                 {
-                    column_indexes.push(idx);
+                    column_indexes.push(Some(idx));
                     column_values.push(Some(values));
                 } else {
-                    column_indexes.push(ColumnIndex::Empty {
-                        num_docs: num_docs_per_column[i],
-                    });
+                    column_indexes.push(None);
                     column_values.push(None);
                 }
             }
@@ -150,19 +142,15 @@ fn merge_column(
             serialize_column_mappable_to_u64(merged_column_index, &merge_column_values, wrt)?;
         }
         ColumnType::IpAddr => {
-            let mut column_indexes: Vec<ColumnIndex> = Vec::with_capacity(columns.len());
+            let mut column_indexes: Vec<Option<ColumnIndex>> = Vec::with_capacity(columns.len());
             let mut column_values: Vec<Option<Arc<dyn ColumnValues<Ipv6Addr>>>> =
                 Vec::with_capacity(columns.len());
-            for (i, dynamic_column_opt) in columns.into_iter().enumerate() {
-                if let Some(DynamicColumn::IpAddr(Column { index: idx, values })) =
-                    dynamic_column_opt
-                {
-                    column_indexes.push(idx);
+            for dynamic_column_opt in columns {
+                if let Some(DynamicColumn::IpAddr(Column { idx, values })) = dynamic_column_opt {
+                    column_indexes.push(Some(idx));
                     column_values.push(Some(values));
                 } else {
-                    column_indexes.push(ColumnIndex::Empty {
-                        num_docs: num_docs_per_column[i],
-                    });
+                    column_indexes.push(None);
                     column_values.push(None);
                 }
             }
@@ -178,22 +166,20 @@ fn merge_column(
             serialize_column_mappable_to_u128(merged_column_index, &merge_column_values, wrt)?;
         }
         ColumnType::Bytes | ColumnType::Str => {
-            let mut column_indexes: Vec<ColumnIndex> = Vec::with_capacity(columns.len());
+            let mut column_indexes: Vec<Option<ColumnIndex>> = Vec::with_capacity(columns.len());
             let mut bytes_columns: Vec<Option<BytesColumn>> = Vec::with_capacity(columns.len());
-            for (i, dynamic_column_opt) in columns.into_iter().enumerate() {
+            for dynamic_column_opt in columns {
                 match dynamic_column_opt {
                     Some(DynamicColumn::Str(str_column)) => {
-                        column_indexes.push(str_column.term_ord_column.index.clone());
+                        column_indexes.push(Some(str_column.term_ord_column.idx.clone()));
                         bytes_columns.push(Some(str_column.into()));
                     }
                     Some(DynamicColumn::Bytes(bytes_column)) => {
-                        column_indexes.push(bytes_column.term_ord_column.index.clone());
+                        column_indexes.push(Some(bytes_column.term_ord_column.idx.clone()));
                         bytes_columns.push(Some(bytes_column));
                     }
                     _ => {
-                        column_indexes.push(ColumnIndex::Empty {
-                            num_docs: num_docs_per_column[i],
-                        });
+                        column_indexes.push(None);
                         bytes_columns.push(None);
                     }
                 }

columnar/src/columnar/merge/tests.rs

@@ -1,5 +1,3 @@
-use itertools::Itertools;
-
 use super::*;
 use crate::{Cardinality, ColumnarWriter, HasAssociatedColumnType, RowId};
 
@@ -155,24 +153,20 @@ fn make_numerical_columnar_multiple_columns(
     ColumnarReader::open(buffer).unwrap()
 }
 
-#[track_caller]
-fn make_byte_columnar_multiple_columns(
-    columns: &[(&str, &[&[&[u8]]])],
-    num_rows: u32,
-) -> ColumnarReader {
+fn make_byte_columnar_multiple_columns(columns: &[(&str, &[&[&[u8]]])]) -> ColumnarReader {
     let mut dataframe_writer = ColumnarWriter::default();
     for (column_name, column_values) in columns {
-        assert_eq!(
-            column_values.len(),
-            num_rows as usize,
-            "All columns must have `{num_rows}` rows"
-        );
         for (row_id, vals) in column_values.iter().enumerate() {
             for val in vals.iter() {
                 dataframe_writer.record_bytes(row_id as u32, column_name, val);
             }
         }
     }
+    let num_rows = columns
+        .iter()
+        .map(|(_, val_rows)| val_rows.len() as RowId)
+        .max()
+        .unwrap_or(0u32);
     let mut buffer: Vec<u8> = Vec::new();
     dataframe_writer
         .serialize(num_rows, None, &mut buffer)
@@ -251,8 +245,6 @@ fn test_merge_columnar_texts() {
     let cols = columnar_reader.read_columns("texts").unwrap();
     let dynamic_column = cols[0].open().unwrap();
     let DynamicColumn::Str(vals) = dynamic_column else { panic!() };
-    assert_eq!(vals.ords().get_cardinality(), Cardinality::Optional);
-
     let get_str_for_ord = |ord| {
         let mut out = String::new();
         vals.ord_to_str(ord, &mut out).unwrap();
@@ -280,8 +272,8 @@ fn test_merge_columnar_texts() {
 
 #[test]
 fn test_merge_columnar_byte() {
-    let columnar1 = make_byte_columnar_multiple_columns(&[("bytes", &[&[b"bbbb"], &[b"baaa"]])], 2);
-    let columnar2 = make_byte_columnar_multiple_columns(&[("bytes", &[&[], &[b"a"]])], 2);
+    let columnar1 = make_byte_columnar_multiple_columns(&[("bytes", &[&[b"bbbb"], &[b"baaa"]])]);
+    let columnar2 = make_byte_columnar_multiple_columns(&[("bytes", &[&[], &[b"a"]])]);
     let mut buffer = Vec::new();
     let columnars = &[&columnar1, &columnar2];
     let stack_merge_order = StackMergeOrder::stack(columnars);
@@ -324,149 +316,3 @@ fn test_merge_columnar_byte() {
     assert_eq!(get_bytes_for_row(2), b"");
     assert_eq!(get_bytes_for_row(3), b"a");
 }
-
-#[test]
-fn test_merge_columnar_byte_with_missing() {
-    let columnar1 = make_byte_columnar_multiple_columns(&[], 3);
-    let columnar2 = make_byte_columnar_multiple_columns(&[("col", &[&[b"b"], &[]])], 2);
-    let columnar3 = make_byte_columnar_multiple_columns(
-        &[
-            ("col", &[&[], &[b"b"], &[b"a", b"b"]]),
-            ("col2", &[&[b"hello"], &[], &[b"a", b"b"]]),
-        ],
-        3,
-    );
-    let mut buffer = Vec::new();
-    let columnars = &[&columnar1, &columnar2, &columnar3];
-    let stack_merge_order = StackMergeOrder::stack(columnars);
-    crate::columnar::merge_columnar(
-        columnars,
-        &[],
-        MergeRowOrder::Stack(stack_merge_order),
-        &mut buffer,
-    )
-    .unwrap();
-    let columnar_reader = ColumnarReader::open(buffer).unwrap();
-    assert_eq!(columnar_reader.num_rows(), 3 + 2 + 3);
-    assert_eq!(columnar_reader.num_columns(), 2);
-    let cols = columnar_reader.read_columns("col").unwrap();
-    let dynamic_column = cols[0].open().unwrap();
-    let DynamicColumn::Bytes(vals) = dynamic_column else { panic!() };
-    let get_bytes_for_ord = |ord| {
-        let mut out = Vec::new();
-        vals.ord_to_bytes(ord, &mut out).unwrap();
-        out
-    };
-    assert_eq!(vals.dictionary.num_terms(), 2);
-    assert_eq!(get_bytes_for_ord(0), b"a");
-    assert_eq!(get_bytes_for_ord(1), b"b");
-    let get_bytes_for_row = |row_id| {
-        let terms: Vec<Vec<u8>> = vals
-            .term_ords(row_id)
-            .map(|term_ord| {
-                let mut out = Vec::new();
-                vals.ord_to_bytes(term_ord, &mut out).unwrap();
-                out
-            })
-            .collect();
-        terms
-    };
-    assert!(get_bytes_for_row(0).is_empty());
-    assert!(get_bytes_for_row(1).is_empty());
-    assert!(get_bytes_for_row(2).is_empty());
-    assert_eq!(get_bytes_for_row(3), vec![b"b".to_vec()]);
-    assert!(get_bytes_for_row(4).is_empty());
-    assert!(get_bytes_for_row(5).is_empty());
-    assert_eq!(get_bytes_for_row(6), vec![b"b".to_vec()]);
-    assert_eq!(get_bytes_for_row(7), vec![b"a".to_vec(), b"b".to_vec()]);
-}
-
-#[test]
-fn test_merge_columnar_different_types() {
-    let columnar1 = make_text_columnar_multiple_columns(&[("mixed", &[&["a"]])]);
-    let columnar2 = make_text_columnar_multiple_columns(&[("mixed", &[&[], &["b"]])]);
-    let columnar3 = make_columnar("mixed", &[1i64]);
-    let mut buffer = Vec::new();
-    let columnars = &[&columnar1, &columnar2, &columnar3];
-    let stack_merge_order = StackMergeOrder::stack(columnars);
-    crate::columnar::merge_columnar(
-        columnars,
-        &[],
-        MergeRowOrder::Stack(stack_merge_order),
-        &mut buffer,
-    )
-    .unwrap();
-    let columnar_reader = ColumnarReader::open(buffer).unwrap();
-    assert_eq!(columnar_reader.num_rows(), 4);
-    assert_eq!(columnar_reader.num_columns(), 2);
-    let cols = columnar_reader.read_columns("mixed").unwrap();
-
-    // numeric column
-    let dynamic_column = cols[0].open().unwrap();
-    let DynamicColumn::I64(vals) = dynamic_column else { panic!() };
-    assert_eq!(vals.get_cardinality(), Cardinality::Optional);
-    assert_eq!(vals.values_for_doc(0).collect_vec(), vec![]);
-    assert_eq!(vals.values_for_doc(1).collect_vec(), vec![]);
-    assert_eq!(vals.values_for_doc(2).collect_vec(), vec![]);
-    assert_eq!(vals.values_for_doc(3).collect_vec(), vec![1]);
-    assert_eq!(vals.values_for_doc(4).collect_vec(), vec![]);
-
-    // text column
-    let dynamic_column = cols[1].open().unwrap();
-    let DynamicColumn::Str(vals) = dynamic_column else { panic!() };
-    assert_eq!(vals.ords().get_cardinality(), Cardinality::Optional);
-    let get_str_for_ord = |ord| {
-        let mut out = String::new();
-        vals.ord_to_str(ord, &mut out).unwrap();
-        out
-    };
-
-    assert_eq!(vals.dictionary.num_terms(), 2);
-    assert_eq!(get_str_for_ord(0), "a");
-    assert_eq!(get_str_for_ord(1), "b");
-
-    let get_str_for_row = |row_id| {
-        let term_ords: Vec<String> = vals
-            .term_ords(row_id)
-            .map(|el| {
-                let mut out = String::new();
-                vals.ord_to_str(el, &mut out).unwrap();
-                out
-            })
-            .collect();
-        term_ords
-    };
-
-    assert_eq!(get_str_for_row(0), vec!["a".to_string()]);
-    assert_eq!(get_str_for_row(1), Vec::<String>::new());
-    assert_eq!(get_str_for_row(2), vec!["b".to_string()]);
-    assert_eq!(get_str_for_row(3), Vec::<String>::new());
-}
-
-#[test]
-fn test_merge_columnar_different_empty_cardinality() {
-    let columnar1 = make_text_columnar_multiple_columns(&[("mixed", &[&["a"]])]);
-    let columnar2 = make_columnar("mixed", &[1i64]);
-    let mut buffer = Vec::new();
-    let columnars = &[&columnar1, &columnar2];
-    let stack_merge_order = StackMergeOrder::stack(columnars);
-    crate::columnar::merge_columnar(
-        columnars,
-        &[],
-        MergeRowOrder::Stack(stack_merge_order),
-        &mut buffer,
-    )
-    .unwrap();
-    let columnar_reader = ColumnarReader::open(buffer).unwrap();
-    assert_eq!(columnar_reader.num_rows(), 2);
-    assert_eq!(columnar_reader.num_columns(), 2);
-    let cols = columnar_reader.read_columns("mixed").unwrap();
-
-    // numeric column
-    let dynamic_column = cols[0].open().unwrap();
-    assert_eq!(dynamic_column.get_cardinality(), Cardinality::Optional);
-
-    // text column
-    let dynamic_column = cols[1].open().unwrap();
-    assert_eq!(dynamic_column.get_cardinality(), Cardinality::Optional);
-}

columnar/src/columnar/mod.rs

@@ -5,8 +5,6 @@ mod reader;
 mod writer;
 
 pub use column_type::{ColumnType, HasAssociatedColumnType};
-#[cfg(test)]
-pub(crate) use merge::ColumnTypeCategory;
 pub use merge::{merge_columnar, MergeRowOrder, ShuffleMergeOrder, StackMergeOrder};
 pub use reader::ColumnarReader;
 pub use writer::ColumnarWriter;

columnar/src/columnar/reader/mod.rs

@@ -1,4 +1,4 @@
-use std::{fmt, io, mem};
+use std::{io, mem};
 
 use common::file_slice::FileSlice;
 use common::BinarySerializable;
@@ -21,32 +21,6 @@ pub struct ColumnarReader {
     num_rows: RowId,
 }
 
-impl fmt::Debug for ColumnarReader {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        let num_rows = self.num_rows();
-        let columns = self.list_columns().unwrap();
-        let num_cols = columns.len();
-        let mut debug_struct = f.debug_struct("Columnar");
-        debug_struct
-            .field("num_rows", &num_rows)
-            .field("num_cols", &num_cols);
-        for (col_name, dynamic_column_handle) in columns.into_iter().take(5) {
-            let col = dynamic_column_handle.open().unwrap();
-            if col.num_values() > 10 {
-                debug_struct.field(&col_name, &"..");
-            } else {
-                debug_struct.field(&col_name, &col);
-            }
-        }
-        if num_cols > 5 {
-            debug_struct.finish_non_exhaustive()?;
-        } else {
-            debug_struct.finish()?;
-        }
-        Ok(())
-    }
-}
-
 /// Functions by both the async/sync code listing columns.
 /// It takes a stream from the column sstable and return the list of
 /// `DynamicColumn` available in it.

columnar/src/dynamic_column.rs

@@ -1,14 +1,14 @@
+use std::io;
 use std::net::Ipv6Addr;
 use std::sync::Arc;
-use std::{fmt, io};
 
 use common::file_slice::FileSlice;
-use common::{ByteCount, DateTime, HasLen, OwnedBytes};
+use common::{DateTime, HasLen, OwnedBytes};
 
 use crate::column::{BytesColumn, Column, StrColumn};
 use crate::column_values::{monotonic_map_column, StrictlyMonotonicFn};
 use crate::columnar::ColumnType;
-use crate::{Cardinality, ColumnIndex, NumericalType};
+use crate::{Cardinality, NumericalType};
 
 #[derive(Clone)]
 pub enum DynamicColumn {
@@ -22,54 +22,19 @@ pub enum DynamicColumn {
     Str(StrColumn),
 }
 
-impl fmt::Debug for DynamicColumn {
-    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        write!(f, "[{} {} |", self.get_cardinality(), self.column_type())?;
-        match self {
-            DynamicColumn::Bool(col) => write!(f, " {:?}", col)?,
-            DynamicColumn::I64(col) => write!(f, " {:?}", col)?,
-            DynamicColumn::U64(col) => write!(f, " {:?}", col)?,
-            DynamicColumn::F64(col) => write!(f, "{:?}", col)?,
-            DynamicColumn::IpAddr(col) => write!(f, "{:?}", col)?,
-            DynamicColumn::DateTime(col) => write!(f, "{:?}", col)?,
-            DynamicColumn::Bytes(col) => write!(f, "{:?}", col)?,
-            DynamicColumn::Str(col) => write!(f, "{:?}", col)?,
-        }
-        write!(f, "]")
-    }
-}
-
 impl DynamicColumn {
-    pub fn column_index(&self) -> &ColumnIndex {
-        match self {
-            DynamicColumn::Bool(c) => &c.index,
-            DynamicColumn::I64(c) => &c.index,
-            DynamicColumn::U64(c) => &c.index,
-            DynamicColumn::F64(c) => &c.index,
-            DynamicColumn::IpAddr(c) => &c.index,
-            DynamicColumn::DateTime(c) => &c.index,
-            DynamicColumn::Bytes(c) => &c.ords().index,
-            DynamicColumn::Str(c) => &c.ords().index,
-        }
-    }
-
     pub fn get_cardinality(&self) -> Cardinality {
-        self.column_index().get_cardinality()
-    }
-
-    pub fn num_values(&self) -> u32 {
         match self {
-            DynamicColumn::Bool(c) => c.values.num_vals(),
-            DynamicColumn::I64(c) => c.values.num_vals(),
-            DynamicColumn::U64(c) => c.values.num_vals(),
-            DynamicColumn::F64(c) => c.values.num_vals(),
-            DynamicColumn::IpAddr(c) => c.values.num_vals(),
-            DynamicColumn::DateTime(c) => c.values.num_vals(),
-            DynamicColumn::Bytes(c) => c.ords().values.num_vals(),
-            DynamicColumn::Str(c) => c.ords().values.num_vals(),
+            DynamicColumn::Bool(c) => c.get_cardinality(),
+            DynamicColumn::I64(c) => c.get_cardinality(),
+            DynamicColumn::U64(c) => c.get_cardinality(),
+            DynamicColumn::F64(c) => c.get_cardinality(),
+            DynamicColumn::IpAddr(c) => c.get_cardinality(),
+            DynamicColumn::DateTime(c) => c.get_cardinality(),
+            DynamicColumn::Bytes(c) => c.ords().get_cardinality(),
+            DynamicColumn::Str(c) => c.ords().get_cardinality(),
         }
     }
-
     pub fn column_type(&self) -> ColumnType {
         match self {
             DynamicColumn::Bool(_) => ColumnType::Bool,
@@ -108,11 +73,11 @@ impl DynamicColumn {
     fn coerce_to_f64(self) -> Option<DynamicColumn> {
         match self {
             DynamicColumn::I64(column) => Some(DynamicColumn::F64(Column {
-                index: column.index,
+                idx: column.idx,
                 values: Arc::new(monotonic_map_column(column.values, MapI64ToF64)),
             })),
             DynamicColumn::U64(column) => Some(DynamicColumn::F64(Column {
-                index: column.index,
+                idx: column.idx,
                 values: Arc::new(monotonic_map_column(column.values, MapU64ToF64)),
             })),
             DynamicColumn::F64(_) => Some(self),
@@ -126,7 +91,7 @@ impl DynamicColumn {
                     return None;
                 }
                 Some(DynamicColumn::I64(Column {
-                    index: column.index,
+                    idx: column.idx,
                     values: Arc::new(monotonic_map_column(column.values, MapU64ToI64)),
                 }))
             }
@@ -141,7 +106,7 @@ impl DynamicColumn {
                     return None;
                 }
                 Some(DynamicColumn::U64(Column {
-                    index: column.index,
+                    idx: column.idx,
                     values: Arc::new(monotonic_map_column(column.values, MapI64ToU64)),
                 }))
             }
@@ -283,8 +248,8 @@ impl DynamicColumnHandle {
         Ok(dynamic_column)
     }
 
-    pub fn num_bytes(&self) -> ByteCount {
-        self.file_slice.len().into()
+    pub fn num_bytes(&self) -> usize {
+        self.file_slice.len()
     }
 
     pub fn column_type(&self) -> ColumnType {

columnar/src/lib.rs

@@ -7,10 +7,8 @@ extern crate more_asserts;
 #[cfg(all(test, feature = "unstable"))]
 extern crate test;
 
-use std::fmt::Display;
 use std::io;
 
-mod block_accessor;
 mod column;
 mod column_index;
 pub mod column_values;
@@ -21,12 +19,9 @@ mod iterable;
 pub(crate) mod utils;
 mod value;
 
-pub use block_accessor::ColumnBlockAccessor;
 pub use column::{BytesColumn, Column, StrColumn};
 pub use column_index::ColumnIndex;
-pub use column_values::{
-    ColumnValues, EmptyColumnValues, MonotonicallyMappableToU128, MonotonicallyMappableToU64,
-};
+pub use column_values::{ColumnValues, MonotonicallyMappableToU128, MonotonicallyMappableToU64};
 pub use columnar::{
     merge_columnar, ColumnType, ColumnarReader, ColumnarWriter, HasAssociatedColumnType,
     MergeRowOrder, ShuffleMergeOrder, StackMergeOrder,
@@ -76,17 +71,6 @@ pub enum Cardinality {
     Multivalued = 2,
 }
 
-impl Display for Cardinality {
-    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
-        let short_str = match self {
-            Cardinality::Full => "full",
-            Cardinality::Optional => "opt",
-            Cardinality::Multivalued => "mult",
-        };
-        write!(f, "{short_str}")
-    }
-}
-
 impl Cardinality {
     pub fn is_optional(&self) -> bool {
         matches!(self, Cardinality::Optional)
@@ -97,6 +81,7 @@ impl Cardinality {
     pub(crate) fn to_code(self) -> u8 {
         self as u8
     }
+
     pub(crate) fn try_from_code(code: u8) -> Result<Cardinality, InvalidData> {
         match code {
             0 => Ok(Cardinality::Full),

columnar/src/tests.rs

@@ -1,17 +1,10 @@
-use std::collections::HashMap;
-use std::fmt::Debug;
 use std::net::Ipv6Addr;
 
-use common::DateTime;
-use proptest::prelude::*;
-
 use crate::column_values::MonotonicallyMappableToU128;
 use crate::columnar::ColumnType;
 use crate::dynamic_column::{DynamicColumn, DynamicColumnHandle};
-use crate::value::{Coerce, NumericalValue};
-use crate::{
-    BytesColumn, Cardinality, Column, ColumnarReader, ColumnarWriter, RowId, StackMergeOrder,
-};
+use crate::value::NumericalValue;
+use crate::{Cardinality, ColumnarReader, ColumnarWriter};
 
 #[test]
 fn test_dataframe_writer_str() {
@@ -24,7 +17,7 @@ fn test_dataframe_writer_str() {
     assert_eq!(columnar.num_columns(), 1);
     let cols: Vec<DynamicColumnHandle> = columnar.read_columns("my_string").unwrap();
     assert_eq!(cols.len(), 1);
-    assert_eq!(cols[0].num_bytes(), 89);
+    assert_eq!(cols[0].num_bytes(), 158);
 }
 
 #[test]
@@ -38,7 +31,7 @@ fn test_dataframe_writer_bytes() {
     assert_eq!(columnar.num_columns(), 1);
     let cols: Vec<DynamicColumnHandle> = columnar.read_columns("my_string").unwrap();
     assert_eq!(cols.len(), 1);
-    assert_eq!(cols[0].num_bytes(), 89);
+    assert_eq!(cols[0].num_bytes(), 158);
 }
 
 #[test]
@@ -133,7 +126,7 @@ fn test_dataframe_writer_numerical() {
     assert_eq!(cols[0].num_bytes(), 33);
     let column = cols[0].open().unwrap();
     let DynamicColumn::I64(column_i64) = column else { panic!(); };
-    assert_eq!(column_i64.index.get_cardinality(), Cardinality::Optional);
+    assert_eq!(column_i64.idx.get_cardinality(), Cardinality::Optional);
     assert_eq!(column_i64.first(0), None);
     assert_eq!(column_i64.first(1), Some(12i64));
     assert_eq!(column_i64.first(2), Some(13i64));
@@ -217,422 +210,3 @@ fn test_dictionary_encoded_bytes() {
         .unwrap();
     assert_eq!(term_buffer, b"b");
 }
-
-fn num_strategy() -> impl Strategy<Value = NumericalValue> {
-    prop_oneof![
-        Just(NumericalValue::U64(0u64)),
-        Just(NumericalValue::U64(u64::MAX)),
-        Just(NumericalValue::I64(0i64)),
-        Just(NumericalValue::I64(i64::MIN)),
-        Just(NumericalValue::I64(i64::MAX)),
-        Just(NumericalValue::F64(1.2f64)),
-    ]
-}
-
-#[derive(Debug, Clone, Copy)]
-enum ColumnValue {
-    Str(&'static str),
-    Bytes(&'static [u8]),
-    Numerical(NumericalValue),
-    IpAddr(Ipv6Addr),
-    Bool(bool),
-    DateTime(DateTime),
-}
-
-impl ColumnValue {
-    pub(crate) fn column_type_category(&self) -> ColumnTypeCategory {
-        match self {
-            ColumnValue::Str(_) => ColumnTypeCategory::Str,
-            ColumnValue::Bytes(_) => ColumnTypeCategory::Bytes,
-            ColumnValue::Numerical(numerical_val) => ColumnTypeCategory::Numerical,
-            ColumnValue::IpAddr(_) => ColumnTypeCategory::IpAddr,
-            ColumnValue::Bool(_) => ColumnTypeCategory::Bool,
-            ColumnValue::DateTime(_) => ColumnTypeCategory::DateTime,
-        }
-    }
-}
-
-fn column_name_strategy() -> impl Strategy<Value = &'static str> {
-    prop_oneof![Just("c1"), Just("c2")]
-}
-
-fn string_strategy() -> impl Strategy<Value = &'static str> {
-    prop_oneof![Just("a"), Just("b")]
-}
-
-fn bytes_strategy() -> impl Strategy<Value = &'static [u8]> {
-    prop_oneof![Just(&[0u8][..]), Just(&[1u8][..])]
-}
-
-// A random column value
-fn column_value_strategy() -> impl Strategy<Value = ColumnValue> {
-    prop_oneof![
-        string_strategy().prop_map(|s| ColumnValue::Str(s)),
-        bytes_strategy().prop_map(|b| ColumnValue::Bytes(b)),
-        num_strategy().prop_map(|n| ColumnValue::Numerical(n)),
-        (1u16..3u16).prop_map(|ip_addr_byte| ColumnValue::IpAddr(Ipv6Addr::new(
-            127,
-            0,
-            0,
-            0,
-            0,
-            0,
-            0,
-            ip_addr_byte
-        ))),
-        any::<bool>().prop_map(|b| ColumnValue::Bool(b)),
-        (0_679_723_993i64..1_679_723_995i64)
-            .prop_map(|val| { ColumnValue::DateTime(DateTime::from_timestamp_secs(val)) })
-    ]
-}
-
-// A document contains up to 4 values.
-fn doc_strategy() -> impl Strategy<Value = Vec<(&'static str, ColumnValue)>> {
-    proptest::collection::vec((column_name_strategy(), column_value_strategy()), 0..4)
-}
-
-// A columnar contains up to 2 docs.
-fn columnar_docs_strategy() -> impl Strategy<Value = Vec<Vec<(&'static str, ColumnValue)>>> {
-    proptest::collection::vec(doc_strategy(), 0..=2)
-}
-
-fn columnar_docs_and_mapping_strategy(
-) -> impl Strategy<Value = (Vec<Vec<(&'static str, ColumnValue)>>, Vec<RowId>)> {
-    columnar_docs_strategy().prop_flat_map(|docs| {
-        permutation_strategy(docs.len()).prop_map(move |permutation| (docs.clone(), permutation))
-    })
-}
-
-fn permutation_strategy(n: usize) -> impl Strategy<Value = Vec<RowId>> {
-    Just((0u32..n as RowId).collect()).prop_shuffle()
-}
-
-fn build_columnar_with_mapping(
-    docs: &[Vec<(&'static str, ColumnValue)>],
-    old_to_new_row_ids_opt: Option<&[RowId]>,
-) -> ColumnarReader {
-    let num_docs = docs.len() as u32;
-    let mut buffer = Vec::new();
-    let mut columnar_writer = ColumnarWriter::default();
-    for (doc_id, vals) in docs.iter().enumerate() {
-        for (column_name, col_val) in vals {
-            match *col_val {
-                ColumnValue::Str(str_val) => {
-                    columnar_writer.record_str(doc_id as u32, column_name, str_val);
-                }
-                ColumnValue::Bytes(bytes) => {
-                    columnar_writer.record_bytes(doc_id as u32, column_name, bytes)
-                }
-                ColumnValue::Numerical(num) => {
-                    columnar_writer.record_numerical(doc_id as u32, column_name, num);
-                }
-                ColumnValue::IpAddr(ip_addr) => {
-                    columnar_writer.record_ip_addr(doc_id as u32, column_name, ip_addr);
-                }
-                ColumnValue::Bool(bool_val) => {
-                    columnar_writer.record_bool(doc_id as u32, column_name, bool_val);
-                }
-                ColumnValue::DateTime(date_time) => {
-                    columnar_writer.record_datetime(doc_id as u32, column_name, date_time);
-                }
-            }
-        }
-    }
-    columnar_writer
-        .serialize(num_docs, old_to_new_row_ids_opt, &mut buffer)
-        .unwrap();
-    let columnar_reader = ColumnarReader::open(buffer).unwrap();
-    columnar_reader
-}
-
-fn build_columnar(docs: &[Vec<(&'static str, ColumnValue)>]) -> ColumnarReader {
-    build_columnar_with_mapping(docs, None)
-}
-
-fn assert_columnar_eq(left: &ColumnarReader, right: &ColumnarReader) {
-    assert_eq!(left.num_rows(), right.num_rows());
-    let left_columns = left.list_columns().unwrap();
-    let right_columns = right.list_columns().unwrap();
-    assert_eq!(left_columns.len(), right_columns.len());
-    for i in 0..left_columns.len() {
-        assert_eq!(left_columns[i].0, right_columns[i].0);
-        let left_column = left_columns[i].1.open().unwrap();
-        let right_column = right_columns[i].1.open().unwrap();
-        assert_dyn_column_eq(&left_column, &right_column);
-    }
-}
-
-fn assert_column_eq<T: PartialEq + Copy>(left: &Column<T>, right: &Column<T>) {}
-
-fn assert_bytes_column_eq(left: &BytesColumn, right: &BytesColumn) {}
-
-fn assert_dyn_column_eq(left_dyn_column: &DynamicColumn, right_dyn_column: &DynamicColumn) {
-    assert_eq!(
-        &left_dyn_column.column_type(),
-        &right_dyn_column.column_type()
-    );
-    assert_eq!(
-        &left_dyn_column.get_cardinality(),
-        &right_dyn_column.get_cardinality()
-    );
-    match &(left_dyn_column, right_dyn_column) {
-        (DynamicColumn::Bool(left_col), DynamicColumn::Bool(right_col)) => {
-            assert_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::I64(left_col), DynamicColumn::I64(right_col)) => {
-            assert_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::U64(left_col), DynamicColumn::U64(right_col)) => {
-            assert_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::F64(left_col), DynamicColumn::F64(right_col)) => {
-            assert_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::DateTime(left_col), DynamicColumn::DateTime(right_col)) => {
-            assert_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::IpAddr(left_col), DynamicColumn::IpAddr(right_col)) => {
-            assert_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::Bytes(left_col), DynamicColumn::Bytes(right_col)) => {
-            assert_bytes_column_eq(left_col, right_col);
-        }
-        (DynamicColumn::Str(left_col), DynamicColumn::Str(right_col)) => {
-            assert_bytes_column_eq(left_col, right_col);
-        }
-        _ => {
-            unreachable!()
-        }
-    }
-}
-
-trait AssertEqualToColumnValue {
-    fn assert_equal_to_column_value(&self, column_value: &ColumnValue);
-}
-
-use crate::columnar::ColumnTypeCategory;
-
-impl AssertEqualToColumnValue for bool {
-    fn assert_equal_to_column_value(&self, column_value: &ColumnValue) {
-        let ColumnValue::Bool(val) = column_value else { panic!() };
-        assert_eq!(self, val);
-    }
-}
-
-impl AssertEqualToColumnValue for Ipv6Addr {
-    fn assert_equal_to_column_value(&self, column_value: &ColumnValue) {
-        let ColumnValue::IpAddr(val) = column_value else { panic!() };
-        assert_eq!(self, val);
-    }
-}
-
-impl<T: Coerce + PartialEq + Debug + Into<NumericalValue>> AssertEqualToColumnValue for T {
-    fn assert_equal_to_column_value(&self, column_value: &ColumnValue) {
-        let ColumnValue::Numerical(num) = column_value else { panic!() };
-        assert_eq!(self, &T::coerce(*num));
-    }
-}
-
-impl AssertEqualToColumnValue for DateTime {
-    fn assert_equal_to_column_value(&self, column_value: &ColumnValue) {
-        let ColumnValue::DateTime(dt) = column_value else { panic!() };
-        assert_eq!(self, dt);
-    }
-}
-
-fn assert_column_values<
-    T: AssertEqualToColumnValue + PartialEq + Copy + PartialOrd + Debug + Send + Sync + 'static,
->(
-    col: &Column<T>,
-    expected: &HashMap<u32, Vec<&ColumnValue>>,
-) {
-    let mut num_non_empty_rows = 0;
-    for doc in 0..col.num_docs() {
-        let doc_vals: Vec<T> = col.values_for_doc(doc).collect();
-        if doc_vals.is_empty() {
-            continue;
-        }
-        num_non_empty_rows += 1;
-        let expected_vals = expected.get(&doc).unwrap();
-        assert_eq!(doc_vals.len(), expected_vals.len());
-        for (val, &expected) in doc_vals.iter().zip(expected_vals.iter()) {
-            val.assert_equal_to_column_value(expected)
-        }
-    }
-    assert_eq!(num_non_empty_rows, expected.len());
-}
-
-fn assert_bytes_column_values(
-    col: &BytesColumn,
-    expected: &HashMap<u32, Vec<&ColumnValue>>,
-    is_str: bool,
-) {
-    let mut num_non_empty_rows = 0;
-    let mut buffer = Vec::new();
-    for doc in 0..col.term_ord_column.num_docs() {
-        let doc_vals: Vec<u64> = col.term_ords(doc).collect();
-        if doc_vals.is_empty() {
-            continue;
-        }
-        let expected_vals = expected.get(&doc).unwrap();
-        assert_eq!(doc_vals.len(), expected_vals.len());
-        for (&expected_col_val, &ord) in expected_vals.iter().zip(&doc_vals) {
-            col.ord_to_bytes(ord, &mut buffer).unwrap();
-            match expected_col_val {
-                ColumnValue::Str(str_val) => {
-                    assert!(is_str);
-                    assert_eq!(str_val.as_bytes(), &buffer);
-                }
-                ColumnValue::Bytes(bytes_val) => {
-                    assert!(!is_str);
-                    assert_eq!(bytes_val, &buffer);
-                }
-                _ => {
-                    panic!();
-                }
-            }
-        }
-        num_non_empty_rows += 1;
-    }
-    assert_eq!(num_non_empty_rows, expected.len());
-}
-
-proptest! {
-    /// This proptest attempts to create a tiny columnar based of up to 3 rows, and checks that the resulting
-    /// columnar matches the row data.
-    #[test]
-    fn test_single_columnar_builder_proptest(docs in columnar_docs_strategy()) {
-        let columnar = build_columnar(&docs[..]);
-        assert_eq!(columnar.num_rows() as usize, docs.len());
-        let mut expected_columns: HashMap<(&str, ColumnTypeCategory), HashMap<u32, Vec<&ColumnValue>> > = Default::default();
-        for (doc_id, doc_vals) in docs.iter().enumerate() {
-            for (col_name, col_val) in doc_vals {
-                expected_columns
-                    .entry((col_name, col_val.column_type_category()))
-                    .or_default()
-                    .entry(doc_id as u32)
-                    .or_default()
-                    .push(col_val);
-            }
-        }
-        let column_list = columnar.list_columns().unwrap();
-        assert_eq!(expected_columns.len(), column_list.len());
-        for (column_name, column) in column_list {
-            let dynamic_column = column.open().unwrap();
-            let col_category: ColumnTypeCategory = dynamic_column.column_type().into();
-            let expected_col_values: &HashMap<u32, Vec<&ColumnValue>> = expected_columns.get(&(column_name.as_str(), col_category)).unwrap();
-            match &dynamic_column {
-                DynamicColumn::Bool(col) =>
-                    assert_column_values(col, expected_col_values),
-                DynamicColumn::I64(col) =>
-                    assert_column_values(col, expected_col_values),
-                DynamicColumn::U64(col) =>
-                    assert_column_values(col, expected_col_values),
-                DynamicColumn::F64(col) =>
-                    assert_column_values(col, expected_col_values),
-                DynamicColumn::IpAddr(col) =>
-                    assert_column_values(col, expected_col_values),
-                DynamicColumn::DateTime(col) =>
-                    assert_column_values(col, expected_col_values),
-                DynamicColumn::Bytes(col) =>
-                    assert_bytes_column_values(col, expected_col_values, false),
-                DynamicColumn::Str(col) =>
-                    assert_bytes_column_values(col, expected_col_values, true),
-            }
-        }
-    }
-
-    /// Same as `test_single_columnar_builder_proptest` but with a shuffling mapping.
-    #[test]
-    fn test_single_columnar_builder_with_shuffle_proptest((docs, mapping) in columnar_docs_and_mapping_strategy()) {
-        let columnar = build_columnar_with_mapping(&docs[..], Some(&mapping));
-        assert_eq!(columnar.num_rows() as usize, docs.len());
-        let mut expected_columns: HashMap<(&str, ColumnTypeCategory), HashMap<u32, Vec<&ColumnValue>> > = Default::default();
-        for (doc_id, doc_vals) in docs.iter().enumerate() {
-            for (col_name, col_val) in doc_vals {
-                expected_columns
-                    .entry((col_name, col_val.column_type_category()))
-                    .or_default()
-                    .entry(mapping[doc_id])
-                    .or_default()
-                    .push(col_val);
-            }
-        }
-        let column_list = columnar.list_columns().unwrap();
-        assert_eq!(expected_columns.len(), column_list.len());
-        for (column_name, column) in column_list {
-            let dynamic_column = column.open().unwrap();
-            let col_category: ColumnTypeCategory = dynamic_column.column_type().into();
-            let expected_col_values: &HashMap<u32, Vec<&ColumnValue>> = expected_columns.get(&(column_name.as_str(), col_category)).unwrap();
-            for doc_id in 0..columnar.num_rows() {
-                match &dynamic_column {
-                    DynamicColumn::Bool(col) =>
-                        assert_column_values(col, expected_col_values),
-                    DynamicColumn::I64(col) =>
-                        assert_column_values(col, expected_col_values),
-                    DynamicColumn::U64(col) =>
-                        assert_column_values(col, expected_col_values),
-                    DynamicColumn::F64(col) =>
-                        assert_column_values(col, expected_col_values),
-                    DynamicColumn::IpAddr(col) =>
-                        assert_column_values(col, expected_col_values),
-                    DynamicColumn::DateTime(col) =>
-                        assert_column_values(col, expected_col_values),
-                    DynamicColumn::Bytes(col) =>
-                        assert_bytes_column_values(col, expected_col_values, false),
-                    DynamicColumn::Str(col) =>
-                        assert_bytes_column_values(col, expected_col_values, true),
-                }
-            }
-        }
-    }
-
-    /// This tests create 2 or 3 random small columnar and attempts to merge them.
-    /// It compares the resulting merged dataframe with what would have been obtained by building the
-    /// dataframe from the concatenated rows to begin with.
-    #[test]
-    fn test_columnar_merge_proptest(columnar_docs in proptest::collection::vec(columnar_docs_strategy(), 2..=3)) {
-        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().cloned().flatten().collect();
-        let expected_merged_columnar = build_columnar(&concat_rows[..]);
-        assert_columnar_eq(&merged_columnar, &expected_merged_columnar);
-    }
-
-
-}
-
-#[test]
-fn test_columnar_failing_test() {
-    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[..]);
-    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().cloned().flatten().collect();
-    let expected_merged_columnar = build_columnar(&concat_rows[..]);
-    assert_columnar_eq(&merged_columnar, &expected_merged_columnar);
-}
-
-// TODO add non trivial remap and merge
-// TODO test required
-// TODO add support for empty columnar.

common/Cargo.toml

@@ -13,7 +13,6 @@ repository = "https://github.com/quickwit-oss/tantivy"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-byteorder = "1.4.3"
 ownedbytes = { version= "0.5", path="../ownedbytes" }
 async-trait = "0.1"
 time = { version = "0.3.10", features = ["serde-well-known"] }

common/src/bitset.rs

@@ -4,8 +4,6 @@ use std::{fmt, io, u64};
 
 use ownedbytes::OwnedBytes;
 
-use crate::ByteCount;
-
 #[derive(Clone, Copy, Eq, PartialEq)]
 pub struct TinySet(u64);
 
@@ -388,8 +386,8 @@ impl ReadOnlyBitSet {
     }
 
     /// Number of bytes used in the bitset representation.
-    pub fn num_bytes(&self) -> ByteCount {
-        self.data.len().into()
+    pub fn num_bytes(&self) -> usize {
+        self.data.len()
     }
 }
 

common/src/byte_count.rs

@@ -1,114 +0,0 @@
-use std::iter::Sum;
-use std::ops::{Add, AddAssign};
-
-use serde::{Deserialize, Serialize};
-
-/// Indicates space usage in bytes
-#[derive(Copy, Clone, Default, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
-pub struct ByteCount(u64);
-
-impl std::fmt::Debug for ByteCount {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.write_str(&self.human_readable())
-    }
-}
-
-impl std::fmt::Display for ByteCount {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.write_str(&self.human_readable())
-    }
-}
-
-const SUFFIX_AND_THRESHOLD: [(&str, u64); 5] = [
-    ("KB", 1_000),
-    ("MB", 1_000_000),
-    ("GB", 1_000_000_000),
-    ("TB", 1_000_000_000_000),
-    ("PB", 1_000_000_000_000_000),
-];
-
-impl ByteCount {
-    #[inline]
-    pub fn get_bytes(&self) -> u64 {
-        self.0
-    }
-
-    pub fn human_readable(&self) -> String {
-        for (suffix, threshold) in SUFFIX_AND_THRESHOLD.iter().rev() {
-            if self.get_bytes() >= *threshold {
-                let unit_num = self.get_bytes() as f64 / *threshold as f64;
-                return format!("{:.2} {}", unit_num, suffix);
-            }
-        }
-        format!("{:.2} B", self.get_bytes())
-    }
-}
-
-impl From<u64> for ByteCount {
-    fn from(value: u64) -> Self {
-        ByteCount(value)
-    }
-}
-impl From<usize> for ByteCount {
-    fn from(value: usize) -> Self {
-        ByteCount(value as u64)
-    }
-}
-
-impl Sum for ByteCount {
-    #[inline]
-    fn sum<I: Iterator<Item = Self>>(iter: I) -> Self {
-        iter.fold(ByteCount::default(), |acc, x| acc + x)
-    }
-}
-
-impl PartialEq<u64> for ByteCount {
-    #[inline]
-    fn eq(&self, other: &u64) -> bool {
-        self.get_bytes() == *other
-    }
-}
-
-impl PartialOrd<u64> for ByteCount {
-    #[inline]
-    fn partial_cmp(&self, other: &u64) -> Option<std::cmp::Ordering> {
-        self.get_bytes().partial_cmp(other)
-    }
-}
-
-impl Add for ByteCount {
-    type Output = Self;
-
-    #[inline]
-    fn add(self, other: Self) -> Self {
-        Self(self.get_bytes() + other.get_bytes())
-    }
-}
-
-impl AddAssign for ByteCount {
-    #[inline]
-    fn add_assign(&mut self, other: Self) {
-        *self = Self(self.get_bytes() + other.get_bytes());
-    }
-}
-
-#[cfg(test)]
-mod test {
-    use crate::ByteCount;
-
-    #[test]
-    fn test_bytes() {
-        assert_eq!(ByteCount::from(0u64).human_readable(), "0 B");
-        assert_eq!(ByteCount::from(300u64).human_readable(), "300 B");
-        assert_eq!(ByteCount::from(1_000_000u64).human_readable(), "1.00 MB");
-        assert_eq!(ByteCount::from(1_500_000u64).human_readable(), "1.50 MB");
-        assert_eq!(
-            ByteCount::from(1_500_000_000u64).human_readable(),
-            "1.50 GB"
-        );
-        assert_eq!(
-            ByteCount::from(3_213_000_000_000u64).human_readable(),
-            "3.21 TB"
-        );
-    }
-}

common/src/dictionary_footer.rs

@@ -1,63 +0,0 @@
-use std::io::{self, Read, Write};
-
-use crate::BinarySerializable;
-
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
-#[repr(u32)]
-pub enum DictionaryKind {
-    Fst = 1,
-    SSTable = 2,
-}
-
-#[derive(Debug, Clone, PartialEq)]
-pub struct DictionaryFooter {
-    pub kind: DictionaryKind,
-    pub version: u32,
-}
-
-impl DictionaryFooter {
-    pub fn verify_equal(&self, other: &DictionaryFooter) -> io::Result<()> {
-        if self.kind != other.kind {
-            return Err(io::Error::new(
-                io::ErrorKind::Other,
-                format!(
-                    "Invalid dictionary type, expected {:?}, found {:?}",
-                    self.kind, other.kind
-                ),
-            ));
-        }
-        if self.version != other.version {
-            return Err(io::Error::new(
-                io::ErrorKind::Other,
-                format!(
-                    "Unsuported dictionary version, expected {}, found {}",
-                    self.version, other.version
-                ),
-            ));
-        }
-        Ok(())
-    }
-}
-
-impl BinarySerializable for DictionaryFooter {
-    fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        self.version.serialize(writer)?;
-        (self.kind as u32).serialize(writer)
-    }
-    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        let version = u32::deserialize(reader)?;
-        let kind = u32::deserialize(reader)?;
-        let kind = match kind {
-            1 => DictionaryKind::Fst,
-            2 => DictionaryKind::SSTable,
-            _ => {
-                return Err(io::Error::new(
-                    io::ErrorKind::Other,
-                    format!("invalid dictionary kind: {kind}"),
-                ))
-            }
-        };
-
-        Ok(DictionaryFooter { kind, version })
-    }
-}

common/src/file_slice.rs

@@ -5,7 +5,7 @@ use std::{fmt, io};
 use async_trait::async_trait;
 use ownedbytes::{OwnedBytes, StableDeref};
 
-use crate::{ByteCount, HasLen};
+use crate::HasLen;
 
 /// Objects that represents files sections in tantivy.
 ///
@@ -216,11 +216,6 @@ impl FileSlice {
     pub fn slice_to(&self, to_offset: usize) -> FileSlice {
         self.slice(0..to_offset)
     }
-
-    /// Returns the byte count of the FileSlice.
-    pub fn num_bytes(&self) -> ByteCount {
-        self.range.len().into()
-    }
 }
 
 #[async_trait]

common/src/lib.rs

@@ -2,21 +2,15 @@
 
 use std::ops::Deref;
 
-pub use byteorder::LittleEndian as Endianness;
-
 mod bitset;
-mod byte_count;
 mod datetime;
-mod dictionary_footer;
 pub mod file_slice;
 mod group_by;
 mod serialize;
 mod vint;
 mod writer;
 pub use bitset::*;
-pub use byte_count::ByteCount;
 pub use datetime::{DatePrecision, DateTime};
-pub use dictionary_footer::*;
 pub use group_by::GroupByIteratorExtended;
 pub use ownedbytes::{OwnedBytes, StableDeref};
 pub use serialize::{BinarySerializable, DeserializeFrom, FixedSize};

common/src/serialize.rs

@@ -1,9 +1,7 @@
 use std::io::{Read, Write};
 use std::{fmt, io};
 
-use byteorder::{ReadBytesExt, WriteBytesExt};
-
-use crate::{Endianness, VInt};
+use crate::VInt;
 
 #[derive(Default)]
 struct Counter(u64);
@@ -107,11 +105,13 @@ impl<Left: BinarySerializable + FixedSize, Right: BinarySerializable + FixedSize
 
 impl BinarySerializable for u32 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_u32::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
 
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<u32> {
-        reader.read_u32::<Endianness>()
+        let mut buf = [0u8; 4];
+        reader.read_exact(&mut buf)?;
+        Ok(u32::from_le_bytes(buf))
     }
 }
 
@@ -121,11 +121,13 @@ impl FixedSize for u32 {
 
 impl BinarySerializable for u16 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_u16::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
 
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<u16> {
-        reader.read_u16::<Endianness>()
+        let mut buf = [0u8; 2];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -135,10 +137,12 @@ impl FixedSize for u16 {
 
 impl BinarySerializable for u64 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_u64::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        reader.read_u64::<Endianness>()
+        let mut buf = [0u8; 8];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -148,10 +152,12 @@ impl FixedSize for u64 {
 
 impl BinarySerializable for u128 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_u128::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        reader.read_u128::<Endianness>()
+        let mut buf = [0u8; 16];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -161,10 +167,12 @@ impl FixedSize for u128 {
 
 impl BinarySerializable for f32 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_f32::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        reader.read_f32::<Endianness>()
+        let mut buf = [0u8; 4];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -174,10 +182,12 @@ impl FixedSize for f32 {
 
 impl BinarySerializable for i64 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_i64::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        reader.read_i64::<Endianness>()
+        let mut buf = [0u8; Self::SIZE_IN_BYTES];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -187,10 +197,12 @@ impl FixedSize for i64 {
 
 impl BinarySerializable for f64 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_f64::<Endianness>(*self)
+        writer.write_all(&self.to_le_bytes())
     }
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
-        reader.read_f64::<Endianness>()
+        let mut buf = [0u8; Self::SIZE_IN_BYTES];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -200,10 +212,12 @@ impl FixedSize for f64 {
 
 impl BinarySerializable for u8 {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_u8(*self)
+        writer.write_all(&self.to_le_bytes())
     }
-    fn deserialize<R: Read>(reader: &mut R) -> io::Result<u8> {
-        reader.read_u8()
+    fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
+        let mut buf = [0u8; Self::SIZE_IN_BYTES];
+        reader.read_exact(&mut buf)?;
+        Ok(Self::from_le_bytes(buf))
     }
 }
 
@@ -213,10 +227,10 @@ impl FixedSize for u8 {
 
 impl BinarySerializable for bool {
     fn serialize<W: Write + ?Sized>(&self, writer: &mut W) -> io::Result<()> {
-        writer.write_u8(u8::from(*self))
+        (*self as u8).serialize(writer)
     }
     fn deserialize<R: Read>(reader: &mut R) -> io::Result<bool> {
-        let val = reader.read_u8()?;
+        let val = u8::deserialize(reader)?;
         match val {
             0 => Ok(false),
             1 => Ok(true),

common/src/vint.rs

@@ -1,8 +1,6 @@
 use std::io;
 use std::io::{Read, Write};
 
-use byteorder::{ByteOrder, LittleEndian};
-
 use super::BinarySerializable;
 
 /// Variable int serializes a u128 number
@@ -127,7 +125,7 @@ pub fn serialize_vint_u32(val: u32, buf: &mut [u8; 8]) -> &[u8] {
             5,
         ),
     };
-    LittleEndian::write_u64(&mut buf[..], res);
+    buf.copy_from_slice(&res.to_le_bytes());
     &buf[0..num_bytes]
 }
 

examples/aggregation.rs

@@ -192,7 +192,7 @@ fn main() -> tantivy::Result<()> {
     //
 
     let agg_req: Aggregations = serde_json::from_str(agg_req_str)?;
-    let collector = AggregationCollector::from_aggs(agg_req, Default::default());
+    let collector = AggregationCollector::from_aggs(agg_req, None);
 
     let agg_res: AggregationResults = searcher.search(&AllQuery, &collector).unwrap();
     let res2: Value = serde_json::to_value(agg_res)?;
@@ -204,7 +204,7 @@ fn main() -> tantivy::Result<()> {
 
     let agg_req: Aggregations = vec![(
         "group_by_stock".to_string(),
-        Aggregation::Bucket(Box::new(BucketAggregation {
+        Aggregation::Bucket(BucketAggregation {
             bucket_agg: BucketAggregationType::Range(RangeAggregation {
                 field: "stock".to_string(),
                 ranges: vec![
@@ -234,12 +234,12 @@ fn main() -> tantivy::Result<()> {
             )]
             .into_iter()
             .collect(),
-        })),
+        }),
     )]
     .into_iter()
     .collect();
 
-    let collector = AggregationCollector::from_aggs(agg_req, Default::default());
+    let collector = AggregationCollector::from_aggs(agg_req, None);
     // We use the `AllQuery` which will pass all documents to the AggregationCollector.
     let agg_res: AggregationResults = searcher.search(&AllQuery, &collector).unwrap();
 
@@ -287,7 +287,7 @@ fn main() -> tantivy::Result<()> {
 
     let agg_req: Aggregations = serde_json::from_str(agg_req_str)?;
 
-    let collector = AggregationCollector::from_aggs(agg_req, Default::default());
+    let collector = AggregationCollector::from_aggs(agg_req, None);
 
     let agg_res: AggregationResults = searcher.search(&AllQuery, &collector).unwrap();
     let res: Value = serde_json::to_value(agg_res)?;

examples/pre_tokenized_text.rs

@@ -12,7 +12,7 @@
 use tantivy::collector::{Count, TopDocs};
 use tantivy::query::TermQuery;
 use tantivy::schema::*;
-use tantivy::tokenizer::{PreTokenizedString, SimpleTokenizer, Token, TokenStream, Tokenizer};
+use tantivy::tokenizer::{PreTokenizedString, SimpleTokenizer, Token, Tokenizer};
 use tantivy::{doc, Index, ReloadPolicy};
 use tempfile::TempDir;
 

examples/stop_words.rs

@@ -50,13 +50,12 @@ fn main() -> tantivy::Result<()> {
 
     // This tokenizer lowers all of the text (to help with stop word matching)
     // then removes all instances of `the` and `and` from the corpus
-    let tokenizer = TextAnalyzer::builder(SimpleTokenizer)
+    let tokenizer = TextAnalyzer::from(SimpleTokenizer)
         .filter(LowerCaser)
         .filter(StopWordFilter::remove(vec![
             "the".to_string(),
             "and".to_string(),
-        ]))
-        .build();
+        ]));
 
     index.tokenizers().register("stoppy", tokenizer);
 

src/aggregation/agg_limits.rs

@@ -1,95 +0,0 @@
-use std::collections::HashMap;
-use std::sync::atomic::AtomicU64;
-use std::sync::Arc;
-
-use common::ByteCount;
-
-use super::collector::DEFAULT_MEMORY_LIMIT;
-use super::{AggregationError, DEFAULT_BUCKET_LIMIT};
-use crate::TantivyError;
-
-/// An estimate for memory consumption. Non recursive
-pub trait MemoryConsumption {
-    fn memory_consumption(&self) -> usize;
-}
-
-impl<K, V, S> MemoryConsumption for HashMap<K, V, S> {
-    fn memory_consumption(&self) -> usize {
-        let num_items = self.capacity();
-        (std::mem::size_of::<K>() + std::mem::size_of::<V>()) * num_items
-    }
-}
-
-/// Aggregation memory limit after which the request fails. Defaults to DEFAULT_MEMORY_LIMIT
-/// (500MB). The limit is shared by all SegmentCollectors
-pub struct AggregationLimits {
-    /// The counter which is shared between the aggregations for one request.
-    memory_consumption: Arc<AtomicU64>,
-    /// The memory_limit in bytes
-    memory_limit: ByteCount,
-    /// The maximum number of buckets _returned_
-    /// This is not counting intermediate buckets.
-    bucket_limit: u32,
-}
-impl Clone for AggregationLimits {
-    fn clone(&self) -> Self {
-        Self {
-            memory_consumption: Arc::clone(&self.memory_consumption),
-            memory_limit: self.memory_limit,
-            bucket_limit: self.bucket_limit,
-        }
-    }
-}
-
-impl Default for AggregationLimits {
-    fn default() -> Self {
-        Self {
-            memory_consumption: Default::default(),
-            memory_limit: DEFAULT_MEMORY_LIMIT.into(),
-            bucket_limit: DEFAULT_BUCKET_LIMIT,
-        }
-    }
-}
-
-impl AggregationLimits {
-    /// *memory_limit*
-    /// memory_limit is defined in bytes.
-    /// Aggregation fails when the estimated memory consumption of the aggregation is higher than
-    /// memory_limit.     
-    /// memory_limit will default to `DEFAULT_MEMORY_LIMIT` (500MB)
-    ///
-    /// *bucket_limit*
-    /// Limits the maximum number of buckets returned from an aggregation request.
-    /// bucket_limit will default to `DEFAULT_BUCKET_LIMIT` (65000)
-    pub fn new(memory_limit: Option<u64>, bucket_limit: Option<u32>) -> Self {
-        Self {
-            memory_consumption: Default::default(),
-            memory_limit: memory_limit.unwrap_or(DEFAULT_MEMORY_LIMIT).into(),
-            bucket_limit: bucket_limit.unwrap_or(DEFAULT_BUCKET_LIMIT),
-        }
-    }
-    pub(crate) fn validate_memory_consumption(&self) -> crate::Result<()> {
-        if self.get_memory_consumed() > self.memory_limit {
-            return Err(TantivyError::AggregationError(
-                AggregationError::MemoryExceeded {
-                    limit: self.memory_limit,
-                    current: self.get_memory_consumed(),
-                },
-            ));
-        }
-        Ok(())
-    }
-    pub(crate) fn add_memory_consumed(&self, num_bytes: u64) {
-        self.memory_consumption
-            .fetch_add(num_bytes, std::sync::atomic::Ordering::Relaxed);
-    }
-    /// Returns the estimated memory consumed by the aggregations
-    pub fn get_memory_consumed(&self) -> ByteCount {
-        self.memory_consumption
-            .load(std::sync::atomic::Ordering::Relaxed)
-            .into()
-    }
-    pub(crate) fn get_bucket_limit(&self) -> u32 {
-        self.bucket_limit
-    }
-}

src/aggregation/agg_req.rs

@@ -16,14 +16,14 @@
 //! let agg_req1: Aggregations = vec![
 //!     (
 //!         "range".to_string(),
-//!         Aggregation::Bucket(Box::new(BucketAggregation {
+//!         Aggregation::Bucket(BucketAggregation {
 //!             bucket_agg: BucketAggregationType::Range(RangeAggregation{
 //!                 field: "score".to_string(),
 //!                 ranges: vec![(3f64..7f64).into(), (7f64..20f64).into()],
 //!                 keyed: false,
 //!             }),
 //!             sub_aggregation: Default::default(),
-//!         })),
+//!         }),
 //!     ),
 //! ]
 //! .into_iter()
@@ -143,7 +143,7 @@ pub fn get_fast_field_names(aggs: &Aggregations) -> HashSet<String> {
 #[serde(untagged)]
 pub enum Aggregation {
     /// Bucket aggregation, see [`BucketAggregation`] for details.
-    Bucket(Box<BucketAggregation>),
+    Bucket(BucketAggregation),
     /// Metric aggregation, see [`MetricAggregation`] for details.
     Metric(MetricAggregation),
 }
@@ -301,7 +301,7 @@ mod tests {
     fn serialize_to_json_test() {
         let agg_req1: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Range(RangeAggregation {
                     field: "score".to_string(),
                     ranges: vec![
@@ -313,7 +313,7 @@ mod tests {
                     keyed: true,
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -351,7 +351,7 @@ mod tests {
         let agg_req2: Aggregations = vec![
             (
                 "range".to_string(),
-                Aggregation::Bucket(Box::new(BucketAggregation {
+                Aggregation::Bucket(BucketAggregation {
                     bucket_agg: BucketAggregationType::Range(RangeAggregation {
                         field: "score2".to_string(),
                         ranges: vec![
@@ -363,7 +363,7 @@ mod tests {
                         ..Default::default()
                     }),
                     sub_aggregation: Default::default(),
-                })),
+                }),
             ),
             (
                 "metric".to_string(),
@@ -377,7 +377,7 @@ mod tests {
 
         let agg_req1: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Range(RangeAggregation {
                     field: "score".to_string(),
                     ranges: vec![
@@ -389,7 +389,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: agg_req2,
-            })),
+            }),
         )]
         .into_iter()
         .collect();

src/aggregation/agg_req_with_accessor.rs

@@ -1,6 +1,9 @@
 //! This will enhance the request tree with access to the fastfield and metadata.
 
-use columnar::{Column, ColumnBlockAccessor, ColumnType, StrColumn};
+use std::rc::Rc;
+use std::sync::atomic::AtomicU32;
+
+use columnar::{Column, ColumnType, StrColumn};
 
 use super::agg_req::{Aggregation, Aggregations, BucketAggregationType, MetricAggregation};
 use super::bucket::{
@@ -10,9 +13,9 @@ use super::metric::{
     AverageAggregation, CountAggregation, MaxAggregation, MinAggregation, StatsAggregation,
     SumAggregation,
 };
-use super::segment_agg_result::AggregationLimits;
+use super::segment_agg_result::BucketCount;
 use super::VecWithNames;
-use crate::SegmentReader;
+use crate::{SegmentReader, TantivyError};
 
 #[derive(Clone, Default)]
 pub(crate) struct AggregationsWithAccessor {
@@ -42,8 +45,7 @@ pub struct BucketAggregationWithAccessor {
     pub(crate) field_type: ColumnType,
     pub(crate) bucket_agg: BucketAggregationType,
     pub(crate) sub_aggregation: AggregationsWithAccessor,
-    pub(crate) limits: AggregationLimits,
-    pub(crate) column_block_accessor: ColumnBlockAccessor<u64>,
+    pub(crate) bucket_count: BucketCount,
 }
 
 impl BucketAggregationWithAccessor {
@@ -51,7 +53,8 @@ impl BucketAggregationWithAccessor {
         bucket: &BucketAggregationType,
         sub_aggregation: &Aggregations,
         reader: &SegmentReader,
-        limits: AggregationLimits,
+        bucket_count: Rc<AtomicU32>,
+        max_bucket_count: u32,
     ) -> crate::Result<BucketAggregationWithAccessor> {
         let mut str_dict_column = None;
         let (accessor, field_type) = match &bucket {
@@ -79,12 +82,15 @@ impl BucketAggregationWithAccessor {
             sub_aggregation: get_aggs_with_accessor_and_validate(
                 &sub_aggregation,
                 reader,
-                &limits.clone(),
+                bucket_count.clone(),
+                max_bucket_count,
             )?,
             bucket_agg: bucket.clone(),
             str_dict_column,
-            limits,
-            column_block_accessor: Default::default(),
+            bucket_count: BucketCount {
+                bucket_count,
+                max_bucket_count,
+            },
         })
     }
 }
@@ -95,7 +101,6 @@ pub struct MetricAggregationWithAccessor {
     pub metric: MetricAggregation,
     pub field_type: ColumnType,
     pub accessor: Column<u64>,
-    pub column_block_accessor: ColumnBlockAccessor<u64>,
 }
 
 impl MetricAggregationWithAccessor {
@@ -116,7 +121,6 @@ impl MetricAggregationWithAccessor {
                     accessor,
                     field_type,
                     metric: metric.clone(),
-                    column_block_accessor: Default::default(),
                 })
             }
         }
@@ -126,7 +130,8 @@ impl MetricAggregationWithAccessor {
 pub(crate) fn get_aggs_with_accessor_and_validate(
     aggs: &Aggregations,
     reader: &SegmentReader,
-    limits: &AggregationLimits,
+    bucket_count: Rc<AtomicU32>,
+    max_bucket_count: u32,
 ) -> crate::Result<AggregationsWithAccessor> {
     let mut metrics = vec![];
     let mut buckets = vec![];
@@ -138,7 +143,8 @@ pub(crate) fn get_aggs_with_accessor_and_validate(
                     &bucket.bucket_agg,
                     &bucket.sub_aggregation,
                     reader,
-                    limits.clone(),
+                    Rc::clone(&bucket_count),
+                    max_bucket_count,
                 )?,
             )),
             Aggregation::Metric(metric) => metrics.push((
@@ -161,11 +167,8 @@ fn get_ff_reader_and_validate(
     let ff_fields = reader.fast_fields();
     let ff_field_with_type = ff_fields
         .u64_lenient_with_type(field_name)?
-        .unwrap_or_else(|| {
-            (
-                Column::build_empty_column(reader.num_docs()),
-                ColumnType::U64,
-            )
-        });
+        .ok_or_else(|| {
+            TantivyError::InvalidArgument(format!("No fast field found for field: {}", field_name))
+        })?;
     Ok(ff_field_with_type)
 }

src/aggregation/agg_result.rs

@@ -11,7 +11,6 @@ use super::agg_req::BucketAggregationInternal;
 use super::bucket::GetDocCount;
 use super::intermediate_agg_result::{IntermediateBucketResult, IntermediateMetricResult};
 use super::metric::{SingleMetricResult, Stats};
-use super::segment_agg_result::AggregationLimits;
 use super::Key;
 use crate::TantivyError;
 
@@ -20,13 +19,6 @@ use crate::TantivyError;
 pub struct AggregationResults(pub FxHashMap<String, AggregationResult>);
 
 impl AggregationResults {
-    pub(crate) fn get_bucket_count(&self) -> u64 {
-        self.0
-            .values()
-            .map(|agg| agg.get_bucket_count())
-            .sum::<u64>()
-    }
-
     pub(crate) fn get_value_from_aggregation(
         &self,
         name: &str,
@@ -55,13 +47,6 @@ pub enum AggregationResult {
 }
 
 impl AggregationResult {
-    pub(crate) fn get_bucket_count(&self) -> u64 {
-        match self {
-            AggregationResult::BucketResult(bucket) => bucket.get_bucket_count(),
-            AggregationResult::MetricResult(_) => 0,
-        }
-    }
-
     pub(crate) fn get_value_from_aggregation(
         &self,
         _name: &str,
@@ -168,28 +153,9 @@ pub enum BucketResult {
 }
 
 impl BucketResult {
-    pub(crate) fn get_bucket_count(&self) -> u64 {
-        match self {
-            BucketResult::Range { buckets } => {
-                buckets.iter().map(|bucket| bucket.get_bucket_count()).sum()
-            }
-            BucketResult::Histogram { buckets } => {
-                buckets.iter().map(|bucket| bucket.get_bucket_count()).sum()
-            }
-            BucketResult::Terms {
-                buckets,
-                sum_other_doc_count: _,
-                doc_count_error_upper_bound: _,
-            } => buckets.iter().map(|bucket| bucket.get_bucket_count()).sum(),
-        }
-    }
-
-    pub(crate) fn empty_from_req(
-        req: &BucketAggregationInternal,
-        limits: &AggregationLimits,
-    ) -> crate::Result<Self> {
+    pub(crate) fn empty_from_req(req: &BucketAggregationInternal) -> crate::Result<Self> {
         let empty_bucket = IntermediateBucketResult::empty_from_req(&req.bucket_agg);
-        empty_bucket.into_final_bucket_result(req, limits)
+        empty_bucket.into_final_bucket_result(req)
     }
 }
 
@@ -204,15 +170,6 @@ pub enum BucketEntries<T> {
     HashMap(FxHashMap<String, T>),
 }
 
-impl<T> BucketEntries<T> {
-    fn iter<'a>(&'a self) -> Box<dyn Iterator<Item = &T> + 'a> {
-        match self {
-            BucketEntries::Vec(vec) => Box::new(vec.iter()),
-            BucketEntries::HashMap(map) => Box::new(map.values()),
-        }
-    }
-}
-
 /// This is the default entry for a bucket, which contains a key, count, and optionally
 /// sub-aggregations.
 ///
@@ -252,11 +209,6 @@ pub struct BucketEntry {
     /// Sub-aggregations in this bucket.
     pub sub_aggregation: AggregationResults,
 }
-impl BucketEntry {
-    pub(crate) fn get_bucket_count(&self) -> u64 {
-        1 + self.sub_aggregation.get_bucket_count()
-    }
-}
 impl GetDocCount for &BucketEntry {
     fn doc_count(&self) -> u64 {
         self.doc_count
@@ -320,8 +272,3 @@ pub struct RangeBucketEntry {
     #[serde(skip_serializing_if = "Option::is_none")]
     pub to_as_string: Option<String>,
 }
-impl RangeBucketEntry {
-    pub(crate) fn get_bucket_count(&self) -> u64 {
-        1 + self.sub_aggregation.get_bucket_count()
-    }
-}

src/aggregation/agg_tests.rs

@@ -9,12 +9,11 @@ use crate::aggregation::buf_collector::DOC_BLOCK_SIZE;
 use crate::aggregation::collector::AggregationCollector;
 use crate::aggregation::intermediate_agg_result::IntermediateAggregationResults;
 use crate::aggregation::metric::AverageAggregation;
-use crate::aggregation::segment_agg_result::AggregationLimits;
 use crate::aggregation::tests::{get_test_index_2_segments, get_test_index_from_values_and_terms};
 use crate::aggregation::DistributedAggregationCollector;
 use crate::query::{AllQuery, TermQuery};
-use crate::schema::{IndexRecordOption, Schema, FAST};
-use crate::{Index, Term};
+use crate::schema::IndexRecordOption;
+use crate::Term;
 
 fn get_avg_req(field_name: &str) -> Aggregation {
     Aggregation::Metric(MetricAggregation::Average(
@@ -22,10 +21,6 @@ fn get_avg_req(field_name: &str) -> Aggregation {
     ))
 }
 
-fn get_collector(agg_req: Aggregations) -> AggregationCollector {
-    AggregationCollector::from_aggs(agg_req, Default::default())
-}
-
 // *** EVERY BUCKET-TYPE SHOULD BE TESTED HERE ***
 fn test_aggregation_flushing(
     merge_segments: bool,
@@ -103,18 +98,15 @@ fn test_aggregation_flushing(
             .unwrap();
 
     let agg_res: AggregationResults = if use_distributed_collector {
-        let collector = DistributedAggregationCollector::from_aggs(
-            agg_req.clone(),
-            AggregationLimits::default(),
-        );
+        let collector = DistributedAggregationCollector::from_aggs(agg_req.clone(), None);
 
         let searcher = reader.searcher();
         let intermediate_agg_result = searcher.search(&AllQuery, &collector).unwrap();
         intermediate_agg_result
-            .into_final_bucket_result(agg_req, &Default::default())
+            .into_final_bucket_result(agg_req)
             .unwrap()
     } else {
-        let collector = get_collector(agg_req);
+        let collector = AggregationCollector::from_aggs(agg_req, None);
 
         let searcher = reader.searcher();
         searcher.search(&AllQuery, &collector).unwrap()
@@ -216,42 +208,42 @@ fn test_aggregation_level1() -> crate::Result<()> {
         ("average".to_string(), get_avg_req("score")),
         (
             "range".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Range(RangeAggregation {
                     field: "score".to_string(),
                     ranges: vec![(3f64..7f64).into(), (7f64..20f64).into()],
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         ),
         (
             "rangef64".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Range(RangeAggregation {
                     field: "score_f64".to_string(),
                     ranges: vec![(3f64..7f64).into(), (7f64..20f64).into()],
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         ),
         (
             "rangei64".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Range(RangeAggregation {
                     field: "score_i64".to_string(),
                     ranges: vec![(3f64..7f64).into(), (7f64..20f64).into()],
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         ),
     ]
     .into_iter()
     .collect();
 
-    let collector = get_collector(agg_req_1);
+    let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
     let searcher = reader.searcher();
     let agg_res: AggregationResults = searcher.search(&term_query, &collector).unwrap();
@@ -316,13 +308,13 @@ fn test_aggregation_level2(
         ("average_in_range".to_string(), get_avg_req("score")),
         (
             "term_agg".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Terms(TermsAggregation {
                     field: "text".to_string(),
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         ),
     ]
     .into_iter()
@@ -390,7 +382,7 @@ fn test_aggregation_level2(
             ("average".to_string(), get_avg_req("score")),
             (
                 "range".to_string(),
-                Aggregation::Bucket(Box::new(BucketAggregation {
+                Aggregation::Bucket(BucketAggregation {
                     bucket_agg: BucketAggregationType::Range(RangeAggregation {
                         field: "score".to_string(),
                         ranges: vec![
@@ -401,11 +393,11 @@ fn test_aggregation_level2(
                         ..Default::default()
                     }),
                     sub_aggregation: sub_agg_req.clone(),
-                })),
+                }),
             ),
             (
                 "rangef64".to_string(),
-                Aggregation::Bucket(Box::new(BucketAggregation {
+                Aggregation::Bucket(BucketAggregation {
                     bucket_agg: BucketAggregationType::Range(RangeAggregation {
                         field: "score_f64".to_string(),
                         ranges: vec![
@@ -416,11 +408,11 @@ fn test_aggregation_level2(
                         ..Default::default()
                     }),
                     sub_aggregation: sub_agg_req.clone(),
-                })),
+                }),
             ),
             (
                 "rangei64".to_string(),
-                Aggregation::Bucket(Box::new(BucketAggregation {
+                Aggregation::Bucket(BucketAggregation {
                     bucket_agg: BucketAggregationType::Range(RangeAggregation {
                         field: "score_i64".to_string(),
                         ranges: vec![
@@ -431,7 +423,7 @@ fn test_aggregation_level2(
                         ..Default::default()
                     }),
                     sub_aggregation: sub_agg_req,
-                })),
+                }),
             ),
         ]
         .into_iter()
@@ -440,18 +432,16 @@ fn test_aggregation_level2(
     };
 
     let agg_res: AggregationResults = if use_distributed_collector {
-        let collector =
-            DistributedAggregationCollector::from_aggs(agg_req.clone(), Default::default());
+        let collector = DistributedAggregationCollector::from_aggs(agg_req.clone(), None);
 
         let searcher = reader.searcher();
         let res = searcher.search(&term_query, &collector).unwrap();
         // Test de/serialization roundtrip on intermediate_agg_result
         let res: IntermediateAggregationResults =
             serde_json::from_str(&serde_json::to_string(&res).unwrap()).unwrap();
-        res.into_final_bucket_result(agg_req.clone(), &Default::default())
-            .unwrap()
+        res.into_final_bucket_result(agg_req.clone()).unwrap()
     } else {
-        let collector = get_collector(agg_req.clone());
+        let collector = AggregationCollector::from_aggs(agg_req.clone(), None);
 
         let searcher = reader.searcher();
         searcher.search(&term_query, &collector).unwrap()
@@ -509,7 +499,7 @@ fn test_aggregation_level2(
     );
 
     // Test empty result set
-    let collector = get_collector(agg_req);
+    let collector = AggregationCollector::from_aggs(agg_req, None);
     let searcher = reader.searcher();
     searcher.search(&query_with_no_hits, &collector).unwrap();
 
@@ -572,203 +562,34 @@ fn test_aggregation_invalid_requests() -> crate::Result<()> {
         .into_iter()
         .collect();
 
-        let collector = get_collector(agg_req_1);
+        let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
         let searcher = reader.searcher();
 
-        searcher.search(&AllQuery, &collector)
+        searcher.search(&AllQuery, &collector).unwrap_err()
     };
 
-    let agg_res = avg_on_field("dummy_text").unwrap_err();
+    let agg_res = avg_on_field("dummy_text");
     assert_eq!(
         format!("{:?}", agg_res),
-        r#"InvalidArgument("Field \"dummy_text\" is not configured as fast field")"#
+        r#"InvalidArgument("No fast field found for field: dummy_text")"#
     );
 
-    // TODO: This should return an error
-    // let agg_res = avg_on_field("not_exist_field").unwrap_err();
-    // assert_eq!(
-    // format!("{:?}", agg_res),
-    // r#"InvalidArgument("No fast field found for field: not_exist_field")"#
-    //);
+    let agg_res = avg_on_field("not_exist_field");
+    assert_eq!(
+        format!("{:?}", agg_res),
+        r#"InvalidArgument("No fast field found for field: not_exist_field")"#
+    );
 
-    // TODO: This should return an error
-    // let agg_res = avg_on_field("ip_addr").unwrap_err();
-    // assert_eq!(
-    // format!("{:?}", agg_res),
-    // r#"InvalidArgument("No fast field found for field: ip_addr")"#
-    //);
+    let agg_res = avg_on_field("ip_addr");
+    assert_eq!(
+        format!("{:?}", agg_res),
+        r#"InvalidArgument("No fast field found for field: ip_addr")"#
+    );
 
     Ok(())
 }
 
-#[test]
-fn test_aggregation_on_json_object() {
-    let mut schema_builder = Schema::builder();
-    let json = schema_builder.add_json_field("json", FAST);
-    let schema = schema_builder.build();
-    let index = Index::create_in_ram(schema);
-    let mut index_writer = index.writer_for_tests().unwrap();
-    index_writer
-        .add_document(doc!(json => json!({"color": "red"})))
-        .unwrap();
-    index_writer
-        .add_document(doc!(json => json!({"color": "blue"})))
-        .unwrap();
-    index_writer.commit().unwrap();
-    let reader = index.reader().unwrap();
-    let searcher = reader.searcher();
-    let agg: Aggregations = vec![(
-        "jsonagg".to_string(),
-        Aggregation::Bucket(Box::new(BucketAggregation {
-            bucket_agg: BucketAggregationType::Terms(TermsAggregation {
-                field: "json.color".to_string(),
-                ..Default::default()
-            }),
-            sub_aggregation: Default::default(),
-        })),
-    )]
-    .into_iter()
-    .collect();
-    let aggregation_collector = get_collector(agg);
-    let aggregation_results = searcher.search(&AllQuery, &aggregation_collector).unwrap();
-    let aggregation_res_json = serde_json::to_value(aggregation_results).unwrap();
-    assert_eq!(
-        &aggregation_res_json,
-        &serde_json::json!({
-            "jsonagg": {
-                "buckets": [
-                    {"doc_count": 1, "key": "blue"},
-                    {"doc_count": 1, "key": "red"}
-                ],
-                "doc_count_error_upper_bound": 0,
-                "sum_other_doc_count": 0
-            }
-        })
-    );
-}
-
-#[test]
-fn test_aggregation_on_json_object_empty_columns() {
-    let mut schema_builder = Schema::builder();
-    let json = schema_builder.add_json_field("json", FAST);
-    let schema = schema_builder.build();
-    let index = Index::create_in_ram(schema);
-    let mut index_writer = index.writer_for_tests().unwrap();
-    // => Empty column when accessing color
-    index_writer
-        .add_document(doc!(json => json!({"price": 10.0})))
-        .unwrap();
-    index_writer.commit().unwrap();
-    // => Empty column when accessing price
-    index_writer
-        .add_document(doc!(json => json!({"color": "blue"})))
-        .unwrap();
-    index_writer.commit().unwrap();
-
-    // => Non Empty columns
-    index_writer
-        .add_document(doc!(json => json!({"color": "red", "price": 10.0})))
-        .unwrap();
-    index_writer
-        .add_document(doc!(json => json!({"color": "red", "price": 10.0})))
-        .unwrap();
-    index_writer
-        .add_document(doc!(json => json!({"color": "green", "price": 20.0})))
-        .unwrap();
-    index_writer
-        .add_document(doc!(json => json!({"color": "green", "price": 20.0})))
-        .unwrap();
-    index_writer
-        .add_document(doc!(json => json!({"color": "green", "price": 20.0})))
-        .unwrap();
-
-    index_writer.commit().unwrap();
-
-    let reader = index.reader().unwrap();
-    let searcher = reader.searcher();
-    let agg: Aggregations = vec![(
-        "jsonagg".to_string(),
-        Aggregation::Bucket(Box::new(BucketAggregation {
-            bucket_agg: BucketAggregationType::Terms(TermsAggregation {
-                field: "json.color".to_string(),
-                ..Default::default()
-            }),
-            sub_aggregation: Default::default(),
-        })),
-    )]
-    .into_iter()
-    .collect();
-
-    let aggregation_collector = get_collector(agg);
-    let aggregation_results = searcher.search(&AllQuery, &aggregation_collector).unwrap();
-    let aggregation_res_json = serde_json::to_value(aggregation_results).unwrap();
-    assert_eq!(
-        &aggregation_res_json,
-        &serde_json::json!({
-            "jsonagg": {
-                "buckets": [
-                    {"doc_count": 3, "key": "green"},
-                    {"doc_count": 2, "key": "red"},
-                    {"doc_count": 1, "key": "blue"}
-                ],
-                "doc_count_error_upper_bound": 0,
-                "sum_other_doc_count": 0
-            }
-        })
-    );
-
-    let agg_req_str = r#"
-    {
-      "jsonagg": {
-        "aggs": {
-          "min_price": { "min": { "field": "json.price" } }
-        },
-        "terms": {
-          "field": "json.color",
-          "order": { "min_price": "desc" }
-        }
-      }
-    } "#;
-    let agg: Aggregations = serde_json::from_str(agg_req_str).unwrap();
-    let aggregation_collector = get_collector(agg);
-    let aggregation_results = searcher.search(&AllQuery, &aggregation_collector).unwrap();
-    let aggregation_res_json = serde_json::to_value(aggregation_results).unwrap();
-    assert_eq!(
-        &aggregation_res_json,
-        &serde_json::json!(
-            {
-              "jsonagg": {
-                "buckets": [
-                  {
-                    "key": "green",
-                    "doc_count": 3,
-                    "min_price": {
-                      "value": 20.0
-                    }
-                  },
-                  {
-                    "key": "red",
-                    "doc_count": 2,
-                    "min_price": {
-                      "value": 10.0
-                    }
-                  },
-                  {
-                    "key": "blue",
-                    "doc_count": 1,
-                    "min_price": {
-                      "value": null
-                    }
-                  }
-                ],
-                "sum_other_doc_count": 0
-              }
-            }
-        )
-    );
-}
-
 #[cfg(all(test, feature = "unstable"))]
 mod bench {
 
@@ -892,7 +713,7 @@ mod bench {
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&term_query, &collector).unwrap()
@@ -921,7 +742,7 @@ mod bench {
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&term_query, &collector).unwrap()
@@ -950,7 +771,7 @@ mod bench {
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&term_query, &collector).unwrap()
@@ -987,7 +808,7 @@ mod bench {
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&term_query, &collector).unwrap()
@@ -1003,21 +824,18 @@ mod bench {
         b.iter(|| {
             let agg_req: Aggregations = vec![(
                 "my_texts".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Terms(TermsAggregation {
-                            field: "text_few_terms".to_string(),
-                            ..Default::default()
-                        }),
-                        sub_aggregation: Default::default(),
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Terms(TermsAggregation {
+                        field: "text_few_terms".to_string(),
+                        ..Default::default()
+                    }),
+                    sub_aggregation: Default::default(),
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req);
+            let collector = AggregationCollector::from_aggs(agg_req, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1042,21 +860,18 @@ mod bench {
 
             let agg_req: Aggregations = vec![(
                 "my_texts".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Terms(TermsAggregation {
-                            field: "text_many_terms".to_string(),
-                            ..Default::default()
-                        }),
-                        sub_aggregation: sub_agg_req,
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Terms(TermsAggregation {
+                        field: "text_many_terms".to_string(),
+                        ..Default::default()
+                    }),
+                    sub_aggregation: sub_agg_req,
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req);
+            let collector = AggregationCollector::from_aggs(agg_req, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1072,21 +887,18 @@ mod bench {
         b.iter(|| {
             let agg_req: Aggregations = vec![(
                 "my_texts".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Terms(TermsAggregation {
-                            field: "text_many_terms".to_string(),
-                            ..Default::default()
-                        }),
-                        sub_aggregation: Default::default(),
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Terms(TermsAggregation {
+                        field: "text_many_terms".to_string(),
+                        ..Default::default()
+                    }),
+                    sub_aggregation: Default::default(),
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req);
+            let collector = AggregationCollector::from_aggs(agg_req, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1102,25 +914,22 @@ mod bench {
         b.iter(|| {
             let agg_req: Aggregations = vec![(
                 "my_texts".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Terms(TermsAggregation {
-                            field: "text_many_terms".to_string(),
-                            order: Some(CustomOrder {
-                                order: Order::Desc,
-                                target: OrderTarget::Key,
-                            }),
-                            ..Default::default()
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Terms(TermsAggregation {
+                        field: "text_many_terms".to_string(),
+                        order: Some(CustomOrder {
+                            order: Order::Desc,
+                            target: OrderTarget::Key,
                         }),
-                        sub_aggregation: Default::default(),
-                    }
-                    .into(),
-                ),
+                        ..Default::default()
+                    }),
+                    sub_aggregation: Default::default(),
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req);
+            let collector = AggregationCollector::from_aggs(agg_req, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1136,29 +945,26 @@ mod bench {
         b.iter(|| {
             let agg_req_1: Aggregations = vec![(
                 "rangef64".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                            field: "score_f64".to_string(),
-                            ranges: vec![
-                                (3f64..7000f64).into(),
-                                (7000f64..20000f64).into(),
-                                (20000f64..30000f64).into(),
-                                (30000f64..40000f64).into(),
-                                (40000f64..50000f64).into(),
-                                (50000f64..60000f64).into(),
-                            ],
-                            ..Default::default()
-                        }),
-                        sub_aggregation: Default::default(),
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                        field: "score_f64".to_string(),
+                        ranges: vec![
+                            (3f64..7000f64).into(),
+                            (7000f64..20000f64).into(),
+                            (20000f64..30000f64).into(),
+                            (30000f64..40000f64).into(),
+                            (40000f64..50000f64).into(),
+                            (50000f64..60000f64).into(),
+                        ],
+                        ..Default::default()
+                    }),
+                    sub_aggregation: Default::default(),
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1183,29 +989,26 @@ mod bench {
 
             let agg_req_1: Aggregations = vec![(
                 "rangef64".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                            field: "score_f64".to_string(),
-                            ranges: vec![
-                                (3f64..7000f64).into(),
-                                (7000f64..20000f64).into(),
-                                (20000f64..30000f64).into(),
-                                (30000f64..40000f64).into(),
-                                (40000f64..50000f64).into(),
-                                (50000f64..60000f64).into(),
-                            ],
-                            ..Default::default()
-                        }),
-                        sub_aggregation: sub_agg_req,
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                        field: "score_f64".to_string(),
+                        ranges: vec![
+                            (3f64..7000f64).into(),
+                            (7000f64..20000f64).into(),
+                            (20000f64..30000f64).into(),
+                            (30000f64..40000f64).into(),
+                            (40000f64..50000f64).into(),
+                            (50000f64..60000f64).into(),
+                        ],
+                        ..Default::default()
+                    }),
+                    sub_aggregation: sub_agg_req,
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1226,26 +1029,24 @@ mod bench {
         b.iter(|| {
             let agg_req_1: Aggregations = vec![(
                 "rangef64".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                            field: "score_f64".to_string(),
-                            interval: 100f64,
-                            hard_bounds: Some(HistogramBounds {
-                                min: 1000.0,
-                                max: 300_000.0,
-                            }),
-                            ..Default::default()
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                        field: "score_f64".to_string(),
+                        interval: 100f64,
+                        hard_bounds: Some(HistogramBounds {
+                            min: 1000.0,
+                            max: 300_000.0,
                         }),
-                        sub_aggregation: Default::default(),
-                    }
-                    .into(),
-                ),
+                        ..Default::default()
+                    }),
+                    sub_aggregation: Default::default(),
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
+
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
         });
@@ -1269,22 +1070,19 @@ mod bench {
 
             let agg_req_1: Aggregations = vec![(
                 "rangef64".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                            field: "score_f64".to_string(),
-                            interval: 100f64, // 1000 buckets
-                            ..Default::default()
-                        }),
-                        sub_aggregation: sub_agg_req,
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                        field: "score_f64".to_string(),
+                        interval: 100f64, // 1000 buckets
+                        ..Default::default()
+                    }),
+                    sub_aggregation: sub_agg_req,
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1300,22 +1098,19 @@ mod bench {
         b.iter(|| {
             let agg_req_1: Aggregations = vec![(
                 "rangef64".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                            field: "score_f64".to_string(),
-                            interval: 100f64, // 1000 buckets
-                            ..Default::default()
-                        }),
-                        sub_aggregation: Default::default(),
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                        field: "score_f64".to_string(),
+                        interval: 100f64, // 1000 buckets
+                        ..Default::default()
+                    }),
+                    sub_aggregation: Default::default(),
+                }),
             )]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&AllQuery, &collector).unwrap()
@@ -1353,27 +1148,24 @@ mod bench {
                 ),
                 (
                     "rangef64".to_string(),
-                    Aggregation::Bucket(
-                        BucketAggregation {
-                            bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                                field: "score_f64".to_string(),
-                                ranges: vec![
-                                    (3f64..7000f64).into(),
-                                    (7000f64..20000f64).into(),
-                                    (20000f64..60000f64).into(),
-                                ],
-                                ..Default::default()
-                            }),
-                            sub_aggregation: sub_agg_req_1,
-                        }
-                        .into(),
-                    ),
+                    Aggregation::Bucket(BucketAggregation {
+                        bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                            field: "score_f64".to_string(),
+                            ranges: vec![
+                                (3f64..7000f64).into(),
+                                (7000f64..20000f64).into(),
+                                (20000f64..60000f64).into(),
+                            ],
+                            ..Default::default()
+                        }),
+                        sub_aggregation: sub_agg_req_1,
+                    }),
                 ),
             ]
             .into_iter()
             .collect();
 
-            let collector = get_collector(agg_req_1);
+            let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
             let searcher = reader.searcher();
             searcher.search(&term_query, &collector).unwrap()

src/aggregation/bucket/histogram/date_histogram.rs

@@ -62,9 +62,7 @@ pub struct DateHistogramAggregationReq {
     ///
     /// Fractional time values are not supported, but you can address this by shifting to another
     /// time unit (e.g., `1.5h` could instead be specified as `90m`).
-    ///
-    /// `Option` for validation, the parameter is not optional
-    pub fixed_interval: Option<String>,
+    pub fixed_interval: String,
     /// Intervals implicitly defines an absolute grid of buckets `[interval * k, interval * (k +
     /// 1))`.
     pub offset: Option<String>,
@@ -114,7 +112,7 @@ impl DateHistogramAggregationReq {
         self.validate()?;
         Ok(HistogramAggregation {
             field: self.field.to_string(),
-            interval: parse_into_microseconds(self.fixed_interval.as_ref().unwrap())? as f64,
+            interval: parse_into_microseconds(&self.fixed_interval)? as f64,
             offset: self
                 .offset
                 .as_ref()
@@ -129,18 +127,11 @@ impl DateHistogramAggregationReq {
     }
 
     fn validate(&self) -> crate::Result<()> {
-        if let Some(interval) = self.interval.as_ref() {
+        if self.interval.is_some() {
             return Err(crate::TantivyError::InvalidArgument(format!(
                 "`interval` parameter {:?} in date histogram is unsupported, only \
                  `fixed_interval` is supported",
-                interval
-            )));
-        }
-        if let Some(interval) = self.date_interval.as_ref() {
-            return Err(crate::TantivyError::InvalidArgument(format!(
-                "`date_interval` parameter {:?} in date histogram is unsupported, only \
-                 `fixed_interval` is supported",
-                interval
+                self.interval
             )));
         }
         if self.format.is_some() {
@@ -149,13 +140,15 @@ impl DateHistogramAggregationReq {
             ));
         }
 
-        if self.fixed_interval.is_none() {
+        if self.date_interval.is_some() {
             return Err(crate::TantivyError::InvalidArgument(
-                "fixed_interval in date histogram is missing".to_string(),
+                "date_interval in date histogram is unsupported, only `fixed_interval` is \
+                 supported"
+                    .to_string(),
             ));
         }
 
-        parse_into_microseconds(self.fixed_interval.as_ref().unwrap())?;
+        parse_into_microseconds(&self.fixed_interval)?;
 
         Ok(())
     }
@@ -477,34 +470,6 @@ mod tests {
         });
         assert_eq!(res, expected_res);
 
-        Ok(())
-    }
-    #[test]
-    fn histogram_test_invalid_req() -> crate::Result<()> {
-        let docs = vec![];
-
-        let index = get_test_index_from_docs(false, &docs)?;
-        let elasticsearch_compatible_json = json!(
-            {
-              "sales_over_time": {
-                "date_histogram": {
-                  "field": "date",
-                  "interval": "30d",
-                  "offset": "-4d"
-                }
-              }
-            }
-        );
-
-        let agg_req: Aggregations =
-            serde_json::from_str(&serde_json::to_string(&elasticsearch_compatible_json).unwrap())
-                .unwrap();
-        let err = exec_request(agg_req, &index).unwrap_err();
-        assert_eq!(
-            err.to_string(),
-            r#"An invalid argument was passed: '`interval` parameter "30d" in date histogram is unsupported, only `fixed_interval` is supported'"#
-        );
-
         Ok(())
     }
 }

src/aggregation/bucket/histogram/histogram.rs

@@ -7,7 +7,6 @@ use rustc_hash::FxHashMap;
 use serde::{Deserialize, Serialize};
 use tantivy_bitpacker::minmax;
 
-use crate::aggregation::agg_limits::MemoryConsumption;
 use crate::aggregation::agg_req::AggregationsInternal;
 use crate::aggregation::agg_req_with_accessor::{
     AggregationsWithAccessor, BucketAggregationWithAccessor,
@@ -17,10 +16,10 @@ use crate::aggregation::intermediate_agg_result::{
     IntermediateAggregationResults, IntermediateBucketResult, IntermediateHistogramBucketEntry,
 };
 use crate::aggregation::segment_agg_result::{
-    build_segment_agg_collector, AggregationLimits, SegmentAggregationCollector,
+    build_segment_agg_collector, SegmentAggregationCollector,
 };
 use crate::aggregation::{f64_from_fastfield_u64, format_date, VecWithNames};
-use crate::TantivyError;
+use crate::{DocId, TantivyError};
 
 /// Histogram is a bucket aggregation, where buckets are created dynamically for given `interval`.
 /// Each document value is rounded down to its bucket.
@@ -231,65 +230,51 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
         })
     }
 
-    #[inline]
     fn collect(
         &mut self,
         doc: crate::DocId,
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
         self.collect_block(&[doc], agg_with_accessor)
     }
 
-    #[inline]
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let bucket_agg_accessor = &mut agg_with_accessor.buckets.values[self.accessor_idx];
-
-        let mem_pre = self.get_memory_consumption();
+        let accessor = &agg_with_accessor.buckets.values[self.accessor_idx].accessor;
+        let sub_aggregation_accessor =
+            &agg_with_accessor.buckets.values[self.accessor_idx].sub_aggregation;
 
         let bounds = self.bounds;
         let interval = self.interval;
         let offset = self.offset;
         let get_bucket_pos = |val| (get_bucket_pos_f64(val, interval, offset) as i64);
 
-        bucket_agg_accessor
-            .column_block_accessor
-            .fetch_block(docs, &bucket_agg_accessor.accessor);
+        for doc in docs {
+            for val in accessor.values_for_doc(*doc) {
+                let val = self.f64_from_fastfield_u64(val);
 
-        for (doc, val) in bucket_agg_accessor.column_block_accessor.iter_docid_vals() {
-            let val = self.f64_from_fastfield_u64(val);
+                let bucket_pos = get_bucket_pos(val);
 
-            let bucket_pos = get_bucket_pos(val);
-
-            if bounds.contains(val) {
-                let bucket = self.buckets.entry(bucket_pos).or_insert_with(|| {
-                    let key = get_bucket_key_from_pos(bucket_pos as f64, interval, offset);
-                    SegmentHistogramBucketEntry { key, doc_count: 0 }
-                });
-                bucket.doc_count += 1;
-                if let Some(sub_aggregation_blueprint) = self.sub_aggregation_blueprint.as_mut() {
-                    self.sub_aggregations
-                        .entry(bucket_pos)
-                        .or_insert_with(|| sub_aggregation_blueprint.clone())
-                        .collect(doc, &mut bucket_agg_accessor.sub_aggregation)?;
+                if bounds.contains(val) {
+                    self.increment_bucket(
+                        bucket_pos,
+                        *doc,
+                        sub_aggregation_accessor,
+                        interval,
+                        offset,
+                    )?;
                 }
             }
         }
-
-        let mem_delta = self.get_memory_consumption() - mem_pre;
-        let limits = &agg_with_accessor.buckets.values[self.accessor_idx].limits;
-        limits.add_memory_consumed(mem_delta as u64);
-        limits.validate_memory_consumption()?;
-
         Ok(())
     }
 
-    fn flush(&mut self, agg_with_accessor: &mut AggregationsWithAccessor) -> crate::Result<()> {
+    fn flush(&mut self, agg_with_accessor: &AggregationsWithAccessor) -> crate::Result<()> {
         let sub_aggregation_accessor =
-            &mut agg_with_accessor.buckets.values[self.accessor_idx].sub_aggregation;
+            &agg_with_accessor.buckets.values[self.accessor_idx].sub_aggregation;
 
         for sub_aggregation in self.sub_aggregations.values_mut() {
             sub_aggregation.flush(sub_aggregation_accessor)?;
@@ -300,12 +285,6 @@ impl SegmentAggregationCollector for SegmentHistogramCollector {
 }
 
 impl SegmentHistogramCollector {
-    fn get_memory_consumption(&self) -> usize {
-        let self_mem = std::mem::size_of::<Self>();
-        let sub_aggs_mem = self.sub_aggregations.memory_consumption();
-        let buckets_mem = self.buckets.memory_consumption();
-        self_mem + sub_aggs_mem + buckets_mem
-    }
     pub fn into_intermediate_bucket_result(
         self,
         agg_with_accessor: &BucketAggregationWithAccessor,
@@ -364,6 +343,29 @@ impl SegmentHistogramCollector {
         })
     }
 
+    #[inline]
+    fn increment_bucket(
+        &mut self,
+        bucket_pos: i64,
+        doc: DocId,
+        bucket_with_accessor: &AggregationsWithAccessor,
+        interval: f64,
+        offset: f64,
+    ) -> crate::Result<()> {
+        let bucket = self.buckets.entry(bucket_pos).or_insert_with(|| {
+            let key = get_bucket_key_from_pos(bucket_pos as f64, interval, offset);
+            SegmentHistogramBucketEntry { key, doc_count: 0 }
+        });
+        bucket.doc_count += 1;
+        if let Some(sub_aggregation_blueprint) = self.sub_aggregation_blueprint.as_mut() {
+            self.sub_aggregations
+                .entry(bucket_pos)
+                .or_insert_with(|| sub_aggregation_blueprint.clone())
+                .collect(doc, bucket_with_accessor)?;
+        }
+        Ok(())
+    }
+
     #[inline]
     fn f64_from_fastfield_u64(&self, val: u64) -> f64 {
         f64_from_fastfield_u64(val, &self.column_type)
@@ -385,7 +387,6 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
     buckets: Vec<IntermediateHistogramBucketEntry>,
     histogram_req: &HistogramAggregation,
     sub_aggregation: &AggregationsInternal,
-    limits: &AggregationLimits,
 ) -> crate::Result<Vec<BucketEntry>> {
     // Generate the full list of buckets without gaps.
     //
@@ -393,17 +394,7 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
     // extended_bounds from the request
     let min_max = minmax(buckets.iter().map(|bucket| bucket.key));
 
-    // memory check upfront
-    let (_, first_bucket_num, last_bucket_num) =
-        generate_bucket_pos_with_opt_minmax(histogram_req, min_max);
-    let added_buckets = (first_bucket_num..=last_bucket_num)
-        .count()
-        .saturating_sub(buckets.len());
-    limits.add_memory_consumed(
-        added_buckets as u64 * std::mem::size_of::<IntermediateHistogramBucketEntry>() as u64,
-    );
-    limits.validate_memory_consumption()?;
-    // create buckets
+    // TODO add memory check
     let fill_gaps_buckets = generate_buckets_with_opt_minmax(histogram_req, min_max);
 
     let empty_sub_aggregation = IntermediateAggregationResults::empty_from_req(sub_aggregation);
@@ -432,9 +423,7 @@ fn intermediate_buckets_to_final_buckets_fill_gaps(
                 sub_aggregation: empty_sub_aggregation.clone(),
             },
         })
-        .map(|intermediate_bucket| {
-            intermediate_bucket.into_final_bucket_entry(sub_aggregation, limits)
-        })
+        .map(|intermediate_bucket| intermediate_bucket.into_final_bucket_entry(sub_aggregation))
         .collect::<crate::Result<Vec<_>>>()
 }
 
@@ -444,26 +433,18 @@ pub(crate) fn intermediate_histogram_buckets_to_final_buckets(
     column_type: Option<ColumnType>,
     histogram_req: &HistogramAggregation,
     sub_aggregation: &AggregationsInternal,
-    limits: &AggregationLimits,
 ) -> crate::Result<Vec<BucketEntry>> {
     let mut buckets = if histogram_req.min_doc_count() == 0 {
         // With min_doc_count != 0, we may need to add buckets, so that there are no
         // gaps, since intermediate result does not contain empty buckets (filtered to
         // reduce serialization size).
 
-        intermediate_buckets_to_final_buckets_fill_gaps(
-            buckets,
-            histogram_req,
-            sub_aggregation,
-            limits,
-        )?
+        intermediate_buckets_to_final_buckets_fill_gaps(buckets, histogram_req, sub_aggregation)?
     } else {
         buckets
             .into_iter()
             .filter(|histogram_bucket| histogram_bucket.doc_count >= histogram_req.min_doc_count())
-            .map(|histogram_bucket| {
-                histogram_bucket.into_final_bucket_entry(sub_aggregation, limits)
-            })
+            .map(|histogram_bucket| histogram_bucket.into_final_bucket_entry(sub_aggregation))
             .collect::<crate::Result<Vec<_>>>()?
     };
 
@@ -502,27 +483,15 @@ fn get_req_min_max(req: &HistogramAggregation, min_max: Option<(f64, f64)>) -> (
 /// Generates buckets with req.interval
 /// Range is computed for provided min_max and request extended_bounds/hard_bounds
 /// returns empty vec when there is no range to span
-pub(crate) fn generate_bucket_pos_with_opt_minmax(
+pub(crate) fn generate_buckets_with_opt_minmax(
     req: &HistogramAggregation,
     min_max: Option<(f64, f64)>,
-) -> (f64, i64, i64) {
+) -> Vec<f64> {
     let (min, max) = get_req_min_max(req, min_max);
 
     let offset = req.offset.unwrap_or(0.0);
     let first_bucket_num = get_bucket_pos_f64(min, req.interval, offset) as i64;
     let last_bucket_num = get_bucket_pos_f64(max, req.interval, offset) as i64;
-    (offset, first_bucket_num, last_bucket_num)
-}
-
-/// Generates buckets with req.interval
-/// Range is computed for provided min_max and request extended_bounds/hard_bounds
-/// returns empty vec when there is no range to span
-pub(crate) fn generate_buckets_with_opt_minmax(
-    req: &HistogramAggregation,
-    min_max: Option<(f64, f64)>,
-) -> Vec<f64> {
-    let (offset, first_bucket_num, last_bucket_num) =
-        generate_bucket_pos_with_opt_minmax(req, min_max);
     let mut buckets = Vec::with_capacity((first_bucket_num..=last_bucket_num).count());
     for bucket_pos in first_bucket_num..=last_bucket_num {
         let bucket_key = bucket_pos as f64 * req.interval + offset;
@@ -544,8 +513,8 @@ mod tests {
     };
     use crate::aggregation::metric::{AverageAggregation, StatsAggregation};
     use crate::aggregation::tests::{
-        exec_request, exec_request_with_query, exec_request_with_query_and_memory_limit,
-        get_test_index_2_segments, get_test_index_from_values, get_test_index_with_num_docs,
+        exec_request, exec_request_with_query, get_test_index_2_segments,
+        get_test_index_from_values, get_test_index_with_num_docs,
     };
 
     #[test]
@@ -556,7 +525,7 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "my_interval".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 3.5,
@@ -564,7 +533,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -582,7 +551,7 @@ mod tests {
         // With offset
         let agg_req: Aggregations = vec![(
             "my_interval".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 3.5,
@@ -590,7 +559,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -631,14 +600,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "my_interval".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -666,14 +635,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -690,40 +659,6 @@ mod tests {
         Ok(())
     }
 
-    #[test]
-    fn histogram_memory_limit() -> crate::Result<()> {
-        let index = get_test_index_with_num_docs(true, 100)?;
-
-        let agg_req: Aggregations = vec![(
-            "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
-                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                    field: "score_f64".to_string(),
-                    interval: 0.1,
-                    ..Default::default()
-                }),
-                sub_aggregation: Default::default(),
-            })),
-        )]
-        .into_iter()
-        .collect();
-
-        let res = exec_request_with_query_and_memory_limit(
-            agg_req,
-            &index,
-            None,
-            AggregationLimits::new(Some(5_000), None),
-        )
-        .unwrap_err();
-        assert_eq!(
-            res.to_string(),
-            "Aborting aggregation because memory limit was exceeded. Limit: 5.00 KB, Current: \
-             102.48 KB"
-        );
-
-        Ok(())
-    }
-
     #[test]
     fn histogram_merge_test() -> crate::Result<()> {
         // Merge buckets counts from different segments
@@ -733,14 +668,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -773,7 +708,7 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
@@ -781,7 +716,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -811,7 +746,7 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
@@ -822,7 +757,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -843,7 +778,7 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
@@ -851,7 +786,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -874,7 +809,7 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(Box::new(BucketAggregation {
+            Aggregation::Bucket(BucketAggregation {
                 bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
                     field: "score_f64".to_string(),
                     interval: 1.0,
@@ -883,7 +818,7 @@ mod tests {
                     ..Default::default()
                 }),
                 sub_aggregation: Default::default(),
-            })),
+            }),
         )]
         .into_iter()
         .collect();
@@ -918,21 +853,18 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        hard_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    hard_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -952,25 +884,22 @@ mod tests {
         //
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        hard_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        extended_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    hard_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+                    extended_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
+                    }),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -989,25 +918,22 @@ mod tests {
         // Invalid request
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        hard_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        extended_bounds: Some(HistogramBounds {
-                            min: 1.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    hard_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+                    extended_bounds: Some(HistogramBounds {
+                        min: 1.0,
+                        max: 12.0,
+                    }),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1037,17 +963,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        ..Default::default()
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1088,21 +1011,18 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        extended_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    extended_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1119,22 +1039,19 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        extended_bounds: Some(HistogramBounds { min: 2.0, max: 5.0 }),
-                        hard_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    extended_bounds: Some(HistogramBounds { min: 2.0, max: 5.0 }),
+                    hard_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1151,21 +1068,18 @@ mod tests {
         // hard_bounds will not extend the result
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        hard_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    hard_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1200,21 +1114,18 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 1.0,
-                        extended_bounds: Some(HistogramBounds {
-                            min: 2.0,
-                            max: 12.0,
-                        }),
-                        ..Default::default()
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 1.0,
+                    extended_bounds: Some(HistogramBounds {
+                        min: 2.0,
+                        max: 12.0,
                     }),
-                    sub_aggregation: agg_req,
-                }
-                .into(),
-            ),
+                    ..Default::default()
+                }),
+                sub_aggregation: agg_req,
+            }),
         )]
         .into_iter()
         .collect();
@@ -1264,17 +1175,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 100000.0,
-                        ..Default::default()
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 100000.0,
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1305,17 +1213,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "date".to_string(),
-                        interval: 86400000000.0, // one day in microseconds
-                        ..Default::default()
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "date".to_string(),
+                    interval: 86400000000.0, // one day in microseconds
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1356,17 +1261,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 0.0,
-                        ..Default::default()
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 0.0,
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -1384,18 +1286,15 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "histogram".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
-                        field: "score_f64".to_string(),
-                        interval: 50.0,
-                        keyed: true,
-                        ..Default::default()
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Histogram(HistogramAggregation {
+                    field: "score_f64".to_string(),
+                    interval: 50.0,
+                    keyed: true,
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();

src/aggregation/bucket/range.rs

@@ -11,7 +11,7 @@ use crate::aggregation::intermediate_agg_result::{
     IntermediateRangeBucketResult,
 };
 use crate::aggregation::segment_agg_result::{
-    build_segment_agg_collector, AggregationLimits, SegmentAggregationCollector,
+    build_segment_agg_collector, BucketCount, SegmentAggregationCollector,
 };
 use crate::aggregation::{
     f64_from_fastfield_u64, f64_to_fastfield_u64, format_date, Key, SerializedKey, VecWithNames,
@@ -208,44 +208,41 @@ impl SegmentAggregationCollector for SegmentRangeCollector {
         })
     }
 
-    #[inline]
     fn collect(
         &mut self,
         doc: crate::DocId,
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
         self.collect_block(&[doc], agg_with_accessor)
     }
 
-    #[inline]
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let bucket_agg_accessor = &mut agg_with_accessor.buckets.values[self.accessor_idx];
+        let accessor = &agg_with_accessor.buckets.values[self.accessor_idx].accessor;
+        let sub_aggregation_accessor =
+            &agg_with_accessor.buckets.values[self.accessor_idx].sub_aggregation;
+        for doc in docs {
+            for val in accessor.values_for_doc(*doc) {
+                let bucket_pos = self.get_bucket_pos(val);
 
-        bucket_agg_accessor
-            .column_block_accessor
-            .fetch_block(docs, &bucket_agg_accessor.accessor);
+                let bucket = &mut self.buckets[bucket_pos];
 
-        for (doc, val) in bucket_agg_accessor.column_block_accessor.iter_docid_vals() {
-            let bucket_pos = self.get_bucket_pos(val);
-
-            let bucket = &mut self.buckets[bucket_pos];
-
-            bucket.bucket.doc_count += 1;
-            if let Some(sub_aggregation) = &mut bucket.bucket.sub_aggregation {
-                sub_aggregation.collect(doc, &mut bucket_agg_accessor.sub_aggregation)?;
+                bucket.bucket.doc_count += 1;
+                if let Some(sub_aggregation) = &mut bucket.bucket.sub_aggregation {
+                    sub_aggregation.collect(*doc, sub_aggregation_accessor)?;
+                }
             }
         }
 
         Ok(())
     }
 
-    fn flush(&mut self, agg_with_accessor: &mut AggregationsWithAccessor) -> crate::Result<()> {
+    fn flush(&mut self, agg_with_accessor: &AggregationsWithAccessor) -> crate::Result<()> {
         let sub_aggregation_accessor =
-            &mut agg_with_accessor.buckets.values[self.accessor_idx].sub_aggregation;
+            &agg_with_accessor.buckets.values[self.accessor_idx].sub_aggregation;
 
         for bucket in self.buckets.iter_mut() {
             if let Some(sub_agg) = bucket.bucket.sub_aggregation.as_mut() {
@@ -261,7 +258,7 @@ impl SegmentRangeCollector {
     pub(crate) fn from_req_and_validate(
         req: &RangeAggregation,
         sub_aggregation: &AggregationsWithAccessor,
-        limits: &AggregationLimits,
+        bucket_count: &BucketCount,
         field_type: ColumnType,
         accessor_idx: usize,
     ) -> crate::Result<Self> {
@@ -305,10 +302,8 @@ impl SegmentRangeCollector {
             })
             .collect::<crate::Result<_>>()?;
 
-        limits.add_memory_consumed(
-            buckets.len() as u64 * std::mem::size_of::<SegmentRangeAndBucketEntry>() as u64,
-        );
-        limits.validate_memory_consumption()?;
+        bucket_count.add_count(buckets.len() as u32);
+        bucket_count.validate_bucket_count()?;
 
         Ok(SegmentRangeCollector {
             buckets,
@@ -480,17 +475,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                        field: "fraction_f64".to_string(),
-                        ranges: vec![(0f64..0.1f64).into(), (0.1f64..0.2f64).into()],
-                        ..Default::default()
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                    field: "fraction_f64".to_string(),
+                    ranges: vec![(0f64..0.1f64).into(), (0.1f64..0.2f64).into()],
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -524,17 +516,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                        field: "fraction_f64".to_string(),
-                        ranges: vec![(0f64..0.1f64).into(), (0.1f64..0.2f64).into()],
-                        ..Default::default()
-                    }),
-                    sub_aggregation: sub_agg_req,
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                    field: "fraction_f64".to_string(),
+                    ranges: vec![(0f64..0.1f64).into(), (0.1f64..0.2f64).into()],
+                    ..Default::default()
+                }),
+                sub_aggregation: sub_agg_req,
+            }),
         )]
         .into_iter()
         .collect();
@@ -559,17 +548,14 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                        field: "fraction_f64".to_string(),
-                        ranges: vec![(0f64..0.1f64).into(), (0.1f64..0.2f64).into()],
-                        keyed: true,
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                    field: "fraction_f64".to_string(),
+                    ranges: vec![(0f64..0.1f64).into(), (0.1f64..0.2f64).into()],
+                    keyed: true,
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -599,28 +585,25 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                        field: "fraction_f64".to_string(),
-                        ranges: vec![
-                            RangeAggregationRange {
-                                key: Some("custom-key-0-to-0.1".to_string()),
-                                from: Some(0f64),
-                                to: Some(0.1f64),
-                            },
-                            RangeAggregationRange {
-                                key: None,
-                                from: Some(0.1f64),
-                                to: Some(0.2f64),
-                            },
-                        ],
-                        keyed: false,
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                    field: "fraction_f64".to_string(),
+                    ranges: vec![
+                        RangeAggregationRange {
+                            key: Some("custom-key-0-to-0.1".to_string()),
+                            from: Some(0f64),
+                            to: Some(0.1f64),
+                        },
+                        RangeAggregationRange {
+                            key: None,
+                            from: Some(0.1f64),
+                            to: Some(0.2f64),
+                        },
+                    ],
+                    keyed: false,
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -659,28 +642,25 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "date_ranges".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                        field: "date".to_string(),
-                        ranges: vec![
-                            RangeAggregationRange {
-                                key: None,
-                                from: None,
-                                to: Some(1546300800000000.0f64),
-                            },
-                            RangeAggregationRange {
-                                key: None,
-                                from: Some(1546300800000000.0f64),
-                                to: Some(1546387200000000.0f64),
-                            },
-                        ],
-                        keyed: false,
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                    field: "date".to_string(),
+                    ranges: vec![
+                        RangeAggregationRange {
+                            key: None,
+                            from: None,
+                            to: Some(1546300800000000.0f64),
+                        },
+                        RangeAggregationRange {
+                            key: None,
+                            from: Some(1546300800000000.0f64),
+                            to: Some(1546387200000000.0f64),
+                        },
+                    ],
+                    keyed: false,
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();
@@ -724,21 +704,18 @@ mod tests {
 
         let agg_req: Aggregations = vec![(
             "range".to_string(),
-            Aggregation::Bucket(
-                BucketAggregation {
-                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                        field: "fraction_f64".to_string(),
-                        ranges: vec![RangeAggregationRange {
-                            key: Some("custom-key-0-to-0.1".to_string()),
-                            from: Some(0f64),
-                            to: Some(0.1f64),
-                        }],
-                        keyed: true,
-                    }),
-                    sub_aggregation: Default::default(),
-                }
-                .into(),
-            ),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                    field: "fraction_f64".to_string(),
+                    ranges: vec![RangeAggregationRange {
+                        key: Some("custom-key-0-to-0.1".to_string()),
+                        from: Some(0f64),
+                        to: Some(0.1f64),
+                    }],
+                    keyed: true,
+                }),
+                sub_aggregation: Default::default(),
+            }),
         )]
         .into_iter()
         .collect();

src/aggregation/buf_collector.rs

@@ -34,7 +34,6 @@ impl BufAggregationCollector {
 }
 
 impl SegmentAggregationCollector for BufAggregationCollector {
-    #[inline]
     fn into_intermediate_aggregations_result(
         self: Box<Self>,
         agg_with_accessor: &AggregationsWithAccessor,
@@ -42,11 +41,10 @@ impl SegmentAggregationCollector for BufAggregationCollector {
         Box::new(self.collector).into_intermediate_aggregations_result(agg_with_accessor)
     }
 
-    #[inline]
     fn collect(
         &mut self,
         doc: crate::DocId,
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
         self.staged_docs[self.num_staged_docs] = doc;
         self.num_staged_docs += 1;
@@ -58,19 +56,18 @@ impl SegmentAggregationCollector for BufAggregationCollector {
         Ok(())
     }
 
-    #[inline]
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        self.collector.collect_block(docs, agg_with_accessor)?;
-
+        for doc in docs {
+            self.collect(*doc, agg_with_accessor)?;
+        }
         Ok(())
     }
 
-    #[inline]
-    fn flush(&mut self, agg_with_accessor: &mut AggregationsWithAccessor) -> crate::Result<()> {
+    fn flush(&mut self, agg_with_accessor: &AggregationsWithAccessor) -> crate::Result<()> {
         self.collector
             .collect_block(&self.staged_docs[..self.num_staged_docs], agg_with_accessor)?;
         self.num_staged_docs = 0;

src/aggregation/collector.rs

@@ -1,36 +1,36 @@
+use std::rc::Rc;
+
 use super::agg_req::Aggregations;
 use super::agg_req_with_accessor::AggregationsWithAccessor;
 use super::agg_result::AggregationResults;
 use super::buf_collector::BufAggregationCollector;
 use super::intermediate_agg_result::IntermediateAggregationResults;
-use super::segment_agg_result::{
-    build_segment_agg_collector, AggregationLimits, SegmentAggregationCollector,
-};
+use super::segment_agg_result::{build_segment_agg_collector, SegmentAggregationCollector};
 use crate::aggregation::agg_req_with_accessor::get_aggs_with_accessor_and_validate;
 use crate::collector::{Collector, SegmentCollector};
-use crate::{DocId, SegmentReader, TantivyError};
+use crate::{SegmentReader, TantivyError};
 
 /// The default max bucket count, before the aggregation fails.
-pub const DEFAULT_BUCKET_LIMIT: u32 = 65000;
-
-/// The default memory limit in bytes before the aggregation fails. 500MB
-pub const DEFAULT_MEMORY_LIMIT: u64 = 500_000_000;
+pub const MAX_BUCKET_COUNT: u32 = 65000;
 
 /// Collector for aggregations.
 ///
 /// The collector collects all aggregations by the underlying aggregation request.
 pub struct AggregationCollector {
     agg: Aggregations,
-    limits: AggregationLimits,
+    max_bucket_count: u32,
 }
 
 impl AggregationCollector {
     /// Create collector from aggregation request.
     ///
-    /// Aggregation fails when the limits in `AggregationLimits` is exceeded. (memory limit and
-    /// bucket limit)
-    pub fn from_aggs(agg: Aggregations, limits: AggregationLimits) -> Self {
-        Self { agg, limits }
+    /// Aggregation fails when the total bucket count is higher than max_bucket_count.
+    /// max_bucket_count will default to `MAX_BUCKET_COUNT` (65000) when unset
+    pub fn from_aggs(agg: Aggregations, max_bucket_count: Option<u32>) -> Self {
+        Self {
+            agg,
+            max_bucket_count: max_bucket_count.unwrap_or(MAX_BUCKET_COUNT),
+        }
     }
 }
 
@@ -44,16 +44,18 @@ impl AggregationCollector {
 /// into the final `AggregationResults` via the `into_final_result()` method.
 pub struct DistributedAggregationCollector {
     agg: Aggregations,
-    limits: AggregationLimits,
+    max_bucket_count: u32,
 }
 
 impl DistributedAggregationCollector {
     /// Create collector from aggregation request.
     ///
-    /// Aggregation fails when the limits in `AggregationLimits` is exceeded. (memory limit and
-    /// bucket limit)
-    pub fn from_aggs(agg: Aggregations, limits: AggregationLimits) -> Self {
-        Self { agg, limits }
+    /// max_bucket_count will default to `MAX_BUCKET_COUNT` (65000) when unset
+    pub fn from_aggs(agg: Aggregations, max_bucket_count: Option<u32>) -> Self {
+        Self {
+            agg,
+            max_bucket_count: max_bucket_count.unwrap_or(MAX_BUCKET_COUNT),
+        }
     }
 }
 
@@ -67,7 +69,11 @@ impl Collector for DistributedAggregationCollector {
         _segment_local_id: crate::SegmentOrdinal,
         reader: &crate::SegmentReader,
     ) -> crate::Result<Self::Child> {
-        AggregationSegmentCollector::from_agg_req_and_reader(&self.agg, reader, &self.limits)
+        AggregationSegmentCollector::from_agg_req_and_reader(
+            &self.agg,
+            reader,
+            self.max_bucket_count,
+        )
     }
 
     fn requires_scoring(&self) -> bool {
@@ -92,7 +98,11 @@ impl Collector for AggregationCollector {
         _segment_local_id: crate::SegmentOrdinal,
         reader: &crate::SegmentReader,
     ) -> crate::Result<Self::Child> {
-        AggregationSegmentCollector::from_agg_req_and_reader(&self.agg, reader, &self.limits)
+        AggregationSegmentCollector::from_agg_req_and_reader(
+            &self.agg,
+            reader,
+            self.max_bucket_count,
+        )
     }
 
     fn requires_scoring(&self) -> bool {
@@ -104,7 +114,7 @@ impl Collector for AggregationCollector {
         segment_fruits: Vec<<Self::Child as SegmentCollector>::Fruit>,
     ) -> crate::Result<Self::Fruit> {
         let res = merge_fruits(segment_fruits)?;
-        res.into_final_bucket_result(self.agg.clone(), &self.limits)
+        res.into_final_bucket_result(self.agg.clone())
     }
 }
 
@@ -125,7 +135,7 @@ fn merge_fruits(
 /// `AggregationSegmentCollector` does the aggregation collection on a segment.
 pub struct AggregationSegmentCollector {
     aggs_with_accessor: AggregationsWithAccessor,
-    agg_collector: BufAggregationCollector,
+    result: BufAggregationCollector,
     error: Option<TantivyError>,
 }
 
@@ -135,14 +145,15 @@ impl AggregationSegmentCollector {
     pub fn from_agg_req_and_reader(
         agg: &Aggregations,
         reader: &SegmentReader,
-        limits: &AggregationLimits,
+        max_bucket_count: u32,
     ) -> crate::Result<Self> {
-        let aggs_with_accessor = get_aggs_with_accessor_and_validate(agg, reader, limits)?;
+        let aggs_with_accessor =
+            get_aggs_with_accessor_and_validate(agg, reader, Rc::default(), max_bucket_count)?;
         let result =
             BufAggregationCollector::new(build_segment_agg_collector(&aggs_with_accessor)?);
         Ok(AggregationSegmentCollector {
             aggs_with_accessor,
-            agg_collector: result,
+            result,
             error: None,
         })
     }
@@ -152,29 +163,11 @@ impl SegmentCollector for AggregationSegmentCollector {
     type Fruit = crate::Result<IntermediateAggregationResults>;
 
     #[inline]
-    fn collect(&mut self, doc: DocId, _score: crate::Score) {
+    fn collect(&mut self, doc: crate::DocId, _score: crate::Score) {
         if self.error.is_some() {
             return;
         }
-        if let Err(err) = self
-            .agg_collector
-            .collect(doc, &mut self.aggs_with_accessor)
-        {
-            self.error = Some(err);
-        }
-    }
-
-    /// The query pushes the documents to the collector via this method.
-    ///
-    /// Only valid for Collectors that ignore docs
-    fn collect_block(&mut self, docs: &[DocId]) {
-        if self.error.is_some() {
-            return;
-        }
-        if let Err(err) = self
-            .agg_collector
-            .collect_block(docs, &mut self.aggs_with_accessor)
-        {
+        if let Err(err) = self.result.collect(doc, &self.aggs_with_accessor) {
             self.error = Some(err);
         }
     }
@@ -183,7 +176,7 @@ impl SegmentCollector for AggregationSegmentCollector {
         if let Some(err) = self.error {
             return Err(err);
         }
-        self.agg_collector.flush(&mut self.aggs_with_accessor)?;
-        Box::new(self.agg_collector).into_intermediate_aggregations_result(&self.aggs_with_accessor)
+        self.result.flush(&self.aggs_with_accessor)?;
+        Box::new(self.result).into_intermediate_aggregations_result(&self.aggs_with_accessor)
     }
 }

src/aggregation/error.rs

@@ -1,33 +1,9 @@
-use common::ByteCount;
-
 use super::bucket::DateHistogramParseError;
 
 /// Error that may occur when opening a directory
 #[derive(Debug, Clone, PartialEq, Eq, Error)]
 pub enum AggregationError {
-    /// Date histogram parse error
+    /// Failed to open the directory.
     #[error("Date histogram parse error: {0:?}")]
     DateHistogramParseError(#[from] DateHistogramParseError),
-    /// Memory limit exceeded
-    #[error(
-        "Aborting aggregation because memory limit was exceeded. Limit: {limit:?}, Current: \
-         {current:?}"
-    )]
-    MemoryExceeded {
-        /// Memory consumption limit
-        limit: ByteCount,
-        /// Current memory consumption
-        current: ByteCount,
-    },
-    /// Bucket limit exceeded
-    #[error(
-        "Aborting aggregation because bucket limit was exceeded. Limit: {limit:?}, Current: \
-         {current:?}"
-    )]
-    BucketLimitExceeded {
-        /// Bucket limit
-        limit: u32,
-        /// Current num buckets
-        current: u32,
-    },
 }

src/aggregation/intermediate_agg_result.rs

@@ -22,11 +22,9 @@ use super::metric::{
     IntermediateAverage, IntermediateCount, IntermediateMax, IntermediateMin, IntermediateStats,
     IntermediateSum,
 };
-use super::segment_agg_result::AggregationLimits;
-use super::{format_date, AggregationError, Key, SerializedKey, VecWithNames};
+use super::{format_date, Key, SerializedKey, VecWithNames};
 use crate::aggregation::agg_result::{AggregationResults, BucketEntries, BucketEntry};
 use crate::aggregation::bucket::TermsAggregationInternal;
-use crate::TantivyError;
 
 /// Contains the intermediate aggregation result, which is optimized to be merged with other
 /// intermediate results.
@@ -40,22 +38,8 @@ pub struct IntermediateAggregationResults {
 
 impl IntermediateAggregationResults {
     /// Convert intermediate result and its aggregation request to the final result.
-    pub fn into_final_bucket_result(
-        self,
-        req: Aggregations,
-        limits: &AggregationLimits,
-    ) -> crate::Result<AggregationResults> {
-        let res = self.into_final_bucket_result_internal(&(req.into()), limits)?;
-        let bucket_count = res.get_bucket_count() as u32;
-        if bucket_count > limits.get_bucket_limit() {
-            return Err(TantivyError::AggregationError(
-                AggregationError::BucketLimitExceeded {
-                    limit: limits.get_bucket_limit(),
-                    current: bucket_count,
-                },
-            ));
-        }
-        Ok(res)
+    pub fn into_final_bucket_result(self, req: Aggregations) -> crate::Result<AggregationResults> {
+        self.into_final_bucket_result_internal(&(req.into()))
     }
 
     /// Convert intermediate result and its aggregation request to the final result.
@@ -65,7 +49,6 @@ impl IntermediateAggregationResults {
     pub(crate) fn into_final_bucket_result_internal(
         self,
         req: &AggregationsInternal,
-        limits: &AggregationLimits,
     ) -> crate::Result<AggregationResults> {
         // Important assumption:
         // When the tree contains buckets/metric, we expect it to have all buckets/metrics from the
@@ -73,11 +56,11 @@ impl IntermediateAggregationResults {
         let mut results: FxHashMap<String, AggregationResult> = FxHashMap::default();
 
         if let Some(buckets) = self.buckets {
-            convert_and_add_final_buckets_to_result(&mut results, buckets, &req.buckets, limits)?
+            convert_and_add_final_buckets_to_result(&mut results, buckets, &req.buckets)?
         } else {
             // When there are no buckets, we create empty buckets, so that the serialized json
             // format is constant
-            add_empty_final_buckets_to_result(&mut results, &req.buckets, limits)?
+            add_empty_final_buckets_to_result(&mut results, &req.buckets)?
         };
 
         if let Some(metrics) = self.metrics {
@@ -178,12 +161,10 @@ fn add_empty_final_metrics_to_result(
 fn add_empty_final_buckets_to_result(
     results: &mut FxHashMap<String, AggregationResult>,
     req_buckets: &VecWithNames<BucketAggregationInternal>,
-    limits: &AggregationLimits,
 ) -> crate::Result<()> {
     let requested_buckets = req_buckets.iter();
     for (key, req) in requested_buckets {
-        let empty_bucket =
-            AggregationResult::BucketResult(BucketResult::empty_from_req(req, limits)?);
+        let empty_bucket = AggregationResult::BucketResult(BucketResult::empty_from_req(req)?);
         results.insert(key.to_string(), empty_bucket);
     }
     Ok(())
@@ -193,13 +174,12 @@ fn convert_and_add_final_buckets_to_result(
     results: &mut FxHashMap<String, AggregationResult>,
     buckets: VecWithNames<IntermediateBucketResult>,
     req_buckets: &VecWithNames<BucketAggregationInternal>,
-    limits: &AggregationLimits,
 ) -> crate::Result<()> {
     assert_eq!(buckets.len(), req_buckets.len());
 
     let buckets_with_request = buckets.into_iter().zip(req_buckets.values());
     for ((key, bucket), req) in buckets_with_request {
-        let result = AggregationResult::BucketResult(bucket.into_final_bucket_result(req, limits)?);
+        let result = AggregationResult::BucketResult(bucket.into_final_bucket_result(req)?);
         results.insert(key, result);
     }
     Ok(())
@@ -307,7 +287,6 @@ impl IntermediateBucketResult {
     pub(crate) fn into_final_bucket_result(
         self,
         req: &BucketAggregationInternal,
-        limits: &AggregationLimits,
     ) -> crate::Result<BucketResult> {
         match self {
             IntermediateBucketResult::Range(range_res) => {
@@ -320,7 +299,6 @@ impl IntermediateBucketResult {
                             req.as_range()
                                 .expect("unexpected aggregation, expected histogram aggregation"),
                             range_res.column_type,
-                            limits,
                         )
                     })
                     .collect::<crate::Result<Vec<_>>>()?;
@@ -359,7 +337,6 @@ impl IntermediateBucketResult {
                     column_type,
                     histogram_req,
                     &req.sub_aggregation,
-                    limits,
                 )?;
 
                 let buckets = if histogram_req.keyed {
@@ -378,7 +355,6 @@ impl IntermediateBucketResult {
                 req.as_term()
                     .expect("unexpected aggregation, expected term aggregation"),
                 &req.sub_aggregation,
-                limits,
             ),
         }
     }
@@ -473,7 +449,6 @@ impl IntermediateTermBucketResult {
         self,
         req: &TermsAggregation,
         sub_aggregation_req: &AggregationsInternal,
-        limits: &AggregationLimits,
     ) -> crate::Result<BucketResult> {
         let req = TermsAggregationInternal::from_req(req);
         let mut buckets: Vec<BucketEntry> = self
@@ -487,7 +462,7 @@ impl IntermediateTermBucketResult {
                     doc_count: entry.doc_count,
                     sub_aggregation: entry
                         .sub_aggregation
-                        .into_final_bucket_result_internal(sub_aggregation_req, limits)?,
+                        .into_final_bucket_result_internal(sub_aggregation_req)?,
                 })
             })
             .collect::<crate::Result<_>>()?;
@@ -519,7 +494,7 @@ impl IntermediateTermBucketResult {
                         let val = bucket
                             .sub_aggregation
                             .get_value_from_aggregation(agg_name, agg_property)?
-                            .unwrap_or(f64::MIN);
+                            .unwrap_or(f64::NAN);
                         Ok((bucket, val))
                     })
                     .collect::<crate::Result<Vec<_>>>()?;
@@ -607,7 +582,6 @@ impl IntermediateHistogramBucketEntry {
     pub(crate) fn into_final_bucket_entry(
         self,
         req: &AggregationsInternal,
-        limits: &AggregationLimits,
     ) -> crate::Result<BucketEntry> {
         Ok(BucketEntry {
             key_as_string: None,
@@ -615,7 +589,7 @@ impl IntermediateHistogramBucketEntry {
             doc_count: self.doc_count,
             sub_aggregation: self
                 .sub_aggregation
-                .into_final_bucket_result_internal(req, limits)?,
+                .into_final_bucket_result_internal(req)?,
         })
     }
 }
@@ -654,14 +628,13 @@ impl IntermediateRangeBucketEntry {
         req: &AggregationsInternal,
         _range_req: &RangeAggregation,
         column_type: Option<ColumnType>,
-        limits: &AggregationLimits,
     ) -> crate::Result<RangeBucketEntry> {
         let mut range_bucket_entry = RangeBucketEntry {
             key: self.key,
             doc_count: self.doc_count,
             sub_aggregation: self
                 .sub_aggregation
-                .into_final_bucket_result_internal(req, limits)?,
+                .into_final_bucket_result_internal(req)?,
             to: self.to,
             from: self.from,
             to_as_string: None,

src/aggregation/metric/mod.rs

@@ -81,7 +81,7 @@ mod tests {
             "price_sum": { "sum": { "field": "price" } }
         }"#;
         let aggregations: Aggregations = serde_json::from_str(aggregations_json).unwrap();
-        let collector = AggregationCollector::from_aggs(aggregations, Default::default());
+        let collector = AggregationCollector::from_aggs(aggregations, None);
         let reader = index.reader().unwrap();
         let searcher = reader.searcher();
         let aggregations_res: AggregationResults = searcher.search(&AllQuery, &collector).unwrap();

src/aggregation/metric/stats.rs

@@ -1,10 +1,8 @@
-use columnar::ColumnType;
+use columnar::{Cardinality, Column, ColumnType};
 use serde::{Deserialize, Serialize};
 
 use super::*;
-use crate::aggregation::agg_req_with_accessor::{
-    AggregationsWithAccessor, MetricAggregationWithAccessor,
-};
+use crate::aggregation::agg_req_with_accessor::AggregationsWithAccessor;
 use crate::aggregation::intermediate_agg_result::{
     IntermediateAggregationResults, IntermediateMetricResult,
 };
@@ -158,7 +156,6 @@ pub(crate) struct SegmentStatsCollector {
     pub(crate) collecting_for: SegmentStatsType,
     pub(crate) stats: IntermediateStats,
     pub(crate) accessor_idx: usize,
-    val_cache: Vec<u64>,
 }
 
 impl SegmentStatsCollector {
@@ -172,28 +169,28 @@ impl SegmentStatsCollector {
             collecting_for,
             stats: IntermediateStats::default(),
             accessor_idx,
-            val_cache: Default::default(),
         }
     }
     #[inline]
-    pub(crate) fn collect_block_with_field(
-        &mut self,
-        docs: &[DocId],
-        agg_accessor: &mut MetricAggregationWithAccessor,
-    ) {
-        agg_accessor
-            .column_block_accessor
-            .fetch_block(docs, &agg_accessor.accessor);
-
-        for val in agg_accessor.column_block_accessor.iter_vals() {
-            let val1 = f64_from_fastfield_u64(val, &self.field_type);
-            self.stats.collect(val1);
+    pub(crate) fn collect_block_with_field(&mut self, docs: &[DocId], field: &Column<u64>) {
+        if field.get_cardinality() == Cardinality::Full {
+            for doc in docs {
+                let val = field.values.get_val(*doc);
+                let val1 = f64_from_fastfield_u64(val, &self.field_type);
+                self.stats.collect(val1);
+            }
+        } else {
+            for doc in docs {
+                for val in field.values_for_doc(*doc) {
+                    let val1 = f64_from_fastfield_u64(val, &self.field_type);
+                    self.stats.collect(val1);
+                }
+            }
         }
     }
 }
 
 impl SegmentAggregationCollector for SegmentStatsCollector {
-    #[inline]
     fn into_intermediate_aggregations_result(
         self: Box<Self>,
         agg_with_accessor: &AggregationsWithAccessor,
@@ -230,11 +227,10 @@ impl SegmentAggregationCollector for SegmentStatsCollector {
         })
     }
 
-    #[inline]
     fn collect(
         &mut self,
         doc: crate::DocId,
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
         let field = &agg_with_accessor.metrics.values[self.accessor_idx].accessor;
 
@@ -250,9 +246,9 @@ impl SegmentAggregationCollector for SegmentStatsCollector {
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
-        let field = &mut agg_with_accessor.metrics.values[self.accessor_idx];
+        let field = &agg_with_accessor.metrics.values[self.accessor_idx].accessor;
         self.collect_block_with_field(docs, field);
         Ok(())
     }
@@ -293,7 +289,7 @@ mod tests {
         .into_iter()
         .collect();
 
-        let collector = AggregationCollector::from_aggs(agg_req_1, Default::default());
+        let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
         let reader = index.reader()?;
         let searcher = reader.searcher();
@@ -330,7 +326,7 @@ mod tests {
         .into_iter()
         .collect();
 
-        let collector = AggregationCollector::from_aggs(agg_req_1, Default::default());
+        let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
         let reader = index.reader()?;
         let searcher = reader.searcher();
@@ -384,33 +380,30 @@ mod tests {
             ),
             (
                 "range".to_string(),
-                Aggregation::Bucket(
-                    BucketAggregation {
-                        bucket_agg: BucketAggregationType::Range(RangeAggregation {
-                            field: "score".to_string(),
-                            ranges: vec![
-                                (3f64..7f64).into(),
-                                (7f64..19f64).into(),
-                                (19f64..20f64).into(),
-                            ],
-                            ..Default::default()
-                        }),
-                        sub_aggregation: iter::once((
-                            "stats".to_string(),
-                            Aggregation::Metric(MetricAggregation::Stats(
-                                StatsAggregation::from_field_name("score".to_string()),
-                            )),
-                        ))
-                        .collect(),
-                    }
-                    .into(),
-                ),
+                Aggregation::Bucket(BucketAggregation {
+                    bucket_agg: BucketAggregationType::Range(RangeAggregation {
+                        field: "score".to_string(),
+                        ranges: vec![
+                            (3f64..7f64).into(),
+                            (7f64..19f64).into(),
+                            (19f64..20f64).into(),
+                        ],
+                        ..Default::default()
+                    }),
+                    sub_aggregation: iter::once((
+                        "stats".to_string(),
+                        Aggregation::Metric(MetricAggregation::Stats(
+                            StatsAggregation::from_field_name("score".to_string()),
+                        )),
+                    ))
+                    .collect(),
+                }),
             ),
         ]
         .into_iter()
         .collect();
 
-        let collector = AggregationCollector::from_aggs(agg_req_1, Default::default());
+        let collector = AggregationCollector::from_aggs(agg_req_1, None);
 
         let searcher = reader.searcher();
         let agg_res: AggregationResults = searcher.search(&term_query, &collector).unwrap();

src/aggregation/mod.rs

@@ -70,7 +70,7 @@
 //!     .into_iter()
 //!     .collect();
 //!
-//!     let collector = AggregationCollector::from_aggs(agg_req, Default::default());
+//!     let collector = AggregationCollector::from_aggs(agg_req, None);
 //!
 //!     let searcher = reader.searcher();
 //!     let agg_res: AggregationResults = searcher.search(&AllQuery, &collector).unwrap();
@@ -130,14 +130,14 @@
 //! let agg_req_1: Aggregations = vec![
 //!     (
 //!         "range".to_string(),
-//!         Aggregation::Bucket(Box::new(BucketAggregation {
+//!         Aggregation::Bucket(BucketAggregation {
 //!             bucket_agg: BucketAggregationType::Range(RangeAggregation{
 //!                 field: "score".to_string(),
 //!                 ranges: vec![(3f64..7f64).into(), (7f64..20f64).into()],
 //!                 keyed: false,
 //!             }),
 //!             sub_aggregation: sub_agg_req_1.clone(),
-//!         })),
+//!         }),
 //!     ),
 //! ]
 //! .into_iter()
@@ -155,7 +155,6 @@
 //! [`AggregationResults`](agg_result::AggregationResults) via the
 //! [`into_final_bucket_result`](intermediate_agg_result::IntermediateAggregationResults::into_final_bucket_result) method.
 
-mod agg_limits;
 pub mod agg_req;
 mod agg_req_with_accessor;
 pub mod agg_result;
@@ -166,7 +165,6 @@ mod date;
 mod error;
 pub mod intermediate_agg_result;
 pub mod metric;
-
 mod segment_agg_result;
 use std::collections::HashMap;
 use std::fmt::Display;
@@ -174,10 +172,9 @@ use std::fmt::Display;
 #[cfg(test)]
 mod agg_tests;
 
-pub use agg_limits::AggregationLimits;
 pub use collector::{
     AggregationCollector, AggregationSegmentCollector, DistributedAggregationCollector,
-    DEFAULT_BUCKET_LIMIT,
+    MAX_BUCKET_COUNT,
 };
 use columnar::{ColumnType, MonotonicallyMappableToU64};
 pub(crate) use date::format_date;
@@ -348,8 +345,9 @@ mod tests {
     use time::OffsetDateTime;
 
     use super::agg_req::Aggregations;
-    use super::segment_agg_result::AggregationLimits;
     use super::*;
+    use crate::aggregation::agg_req::{Aggregation, BucketAggregation, BucketAggregationType};
+    use crate::aggregation::bucket::TermsAggregation;
     use crate::indexer::NoMergePolicy;
     use crate::query::{AllQuery, TermQuery};
     use crate::schema::{IndexRecordOption, Schema, TextFieldIndexing, FAST, STRING};
@@ -373,16 +371,7 @@ mod tests {
         index: &Index,
         query: Option<(&str, &str)>,
     ) -> crate::Result<Value> {
-        exec_request_with_query_and_memory_limit(agg_req, index, query, Default::default())
-    }
-
-    pub fn exec_request_with_query_and_memory_limit(
-        agg_req: Aggregations,
-        index: &Index,
-        query: Option<(&str, &str)>,
-        limits: AggregationLimits,
-    ) -> crate::Result<Value> {
-        let collector = AggregationCollector::from_aggs(agg_req, limits);
+        let collector = AggregationCollector::from_aggs(agg_req, None);
 
         let reader = index.reader()?;
         let searcher = reader.searcher();
@@ -606,4 +595,50 @@ mod tests {
 
         Ok(index)
     }
+
+    #[test]
+    fn test_aggregation_on_json_object() {
+        let mut schema_builder = Schema::builder();
+        let json = schema_builder.add_json_field("json", FAST);
+        let schema = schema_builder.build();
+        let index = Index::create_in_ram(schema);
+        let mut index_writer = index.writer_for_tests().unwrap();
+        index_writer
+            .add_document(doc!(json => json!({"color": "red"})))
+            .unwrap();
+        index_writer
+            .add_document(doc!(json => json!({"color": "blue"})))
+            .unwrap();
+        index_writer.commit().unwrap();
+        let reader = index.reader().unwrap();
+        let searcher = reader.searcher();
+        let agg: Aggregations = vec![(
+            "jsonagg".to_string(),
+            Aggregation::Bucket(BucketAggregation {
+                bucket_agg: BucketAggregationType::Terms(TermsAggregation {
+                    field: "json.color".to_string(),
+                    ..Default::default()
+                }),
+                sub_aggregation: Default::default(),
+            }),
+        )]
+        .into_iter()
+        .collect();
+        let aggregation_collector = AggregationCollector::from_aggs(agg, None);
+        let aggregation_results = searcher.search(&AllQuery, &aggregation_collector).unwrap();
+        let aggregation_res_json = serde_json::to_value(aggregation_results).unwrap();
+        assert_eq!(
+            &aggregation_res_json,
+            &serde_json::json!({
+                "jsonagg": {
+                    "buckets": [
+                        {"doc_count": 1, "key": "blue"},
+                        {"doc_count": 1, "key": "red"}
+                    ],
+                    "doc_count_error_upper_bound": 0,
+                    "sum_other_doc_count": 0
+                }
+            })
+        );
+    }
 }

src/aggregation/segment_agg_result.rs

@@ -4,13 +4,15 @@
 //! merging.
 
 use std::fmt::Debug;
+use std::rc::Rc;
+use std::sync::atomic::AtomicU32;
 
-pub(crate) use super::agg_limits::AggregationLimits;
 use super::agg_req::MetricAggregation;
 use super::agg_req_with_accessor::{
     AggregationsWithAccessor, BucketAggregationWithAccessor, MetricAggregationWithAccessor,
 };
 use super::bucket::{SegmentHistogramCollector, SegmentRangeCollector, SegmentTermCollector};
+use super::collector::MAX_BUCKET_COUNT;
 use super::intermediate_agg_result::IntermediateAggregationResults;
 use super::metric::{
     AverageAggregation, CountAggregation, MaxAggregation, MinAggregation, SegmentStatsCollector,
@@ -18,6 +20,7 @@ use super::metric::{
 };
 use super::VecWithNames;
 use crate::aggregation::agg_req::BucketAggregationType;
+use crate::TantivyError;
 
 pub(crate) trait SegmentAggregationCollector: CollectorClone + Debug {
     fn into_intermediate_aggregations_result(
@@ -28,18 +31,18 @@ pub(crate) trait SegmentAggregationCollector: CollectorClone + Debug {
     fn collect(
         &mut self,
         doc: crate::DocId,
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()>;
 
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()>;
 
     /// Finalize method. Some Aggregator collect blocks of docs before calling `collect_block`.
     /// This method ensures those staged docs will be collected.
-    fn flush(&mut self, _agg_with_accessor: &mut AggregationsWithAccessor) -> crate::Result<()> {
+    fn flush(&mut self, _agg_with_accessor: &AggregationsWithAccessor) -> crate::Result<()> {
         Ok(())
     }
 }
@@ -128,7 +131,7 @@ pub(crate) fn build_bucket_segment_agg_collector(
             Ok(Box::new(SegmentRangeCollector::from_req_and_validate(
                 range_req,
                 &req.sub_aggregation,
-                &req.limits,
+                &req.bucket_count,
                 req.field_type,
                 accessor_idx,
             )?))
@@ -206,7 +209,7 @@ impl SegmentAggregationCollector for GenericSegmentAggregationResultsCollector {
     fn collect(
         &mut self,
         doc: crate::DocId,
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
         self.collect_block(&[doc], agg_with_accessor)?;
 
@@ -216,7 +219,7 @@ impl SegmentAggregationCollector for GenericSegmentAggregationResultsCollector {
     fn collect_block(
         &mut self,
         docs: &[crate::DocId],
-        agg_with_accessor: &mut AggregationsWithAccessor,
+        agg_with_accessor: &AggregationsWithAccessor,
     ) -> crate::Result<()> {
         if let Some(metrics) = self.metrics.as_mut() {
             for collector in metrics {
@@ -233,7 +236,7 @@ impl SegmentAggregationCollector for GenericSegmentAggregationResultsCollector {
         Ok(())
     }
 
-    fn flush(&mut self, agg_with_accessor: &mut AggregationsWithAccessor) -> crate::Result<()> {
+    fn flush(&mut self, agg_with_accessor: &AggregationsWithAccessor) -> crate::Result<()> {
         if let Some(metrics) = &mut self.metrics {
             for collector in metrics {
                 collector.flush(agg_with_accessor)?;
@@ -281,3 +284,37 @@ impl GenericSegmentAggregationResultsCollector {
         Ok(GenericSegmentAggregationResultsCollector { metrics, buckets })
     }
 }
+
+#[derive(Clone)]
+pub(crate) struct BucketCount {
+    /// The counter which is shared between the aggregations for one request.
+    pub(crate) bucket_count: Rc<AtomicU32>,
+    pub(crate) max_bucket_count: u32,
+}
+
+impl Default for BucketCount {
+    fn default() -> Self {
+        Self {
+            bucket_count: Default::default(),
+            max_bucket_count: MAX_BUCKET_COUNT,
+        }
+    }
+}
+
+impl BucketCount {
+    pub(crate) fn validate_bucket_count(&self) -> crate::Result<()> {
+        if self.get_count() > self.max_bucket_count {
+            return Err(TantivyError::InvalidArgument(
+                "Aborting aggregation because too many buckets were created".to_string(),
+            ));
+        }
+        Ok(())
+    }
+    pub(crate) fn add_count(&self, count: u32) {
+        self.bucket_count
+            .fetch_add(count, std::sync::atomic::Ordering::Relaxed);
+    }
+    pub(crate) fn get_count(&self) -> u32 {
+        self.bucket_count.load(std::sync::atomic::Ordering::Relaxed)
+    }
+}

src/collector/mod.rs

@@ -180,11 +180,9 @@ pub trait Collector: Sync + Send {
                 })?;
             }
             (Some(alive_bitset), false) => {
-                weight.for_each_no_score(reader, &mut |docs| {
-                    for doc in docs.iter().cloned() {
-                        if alive_bitset.is_alive(doc) {
-                            segment_collector.collect(doc, 0.0);
-                        }
+                weight.for_each_no_score(reader, &mut |doc| {
+                    if alive_bitset.is_alive(doc) {
+                        segment_collector.collect(doc, 0.0);
                     }
                 })?;
             }
@@ -194,8 +192,8 @@ pub trait Collector: Sync + Send {
                 })?;
             }
             (None, false) => {
-                weight.for_each_no_score(reader, &mut |docs| {
-                    segment_collector.collect_block(docs);
+                weight.for_each_no_score(reader, &mut |doc| {
+                    segment_collector.collect(doc, 0.0);
                 })?;
             }
         }
@@ -272,13 +270,6 @@ pub trait SegmentCollector: 'static {
     /// The query pushes the scored document to the collector via this method.
     fn collect(&mut self, doc: DocId, score: Score);
 
-    /// The query pushes the scored document to the collector via this method.
-    fn collect_block(&mut self, docs: &[DocId]) {
-        for doc in docs {
-            self.collect(*doc, 0.0);
-        }
-    }
-
     /// Extract the fruit of the collection from the `SegmentCollector`.
     fn harvest(self) -> Self::Fruit;
 }

src/collector/top_score_collector.rs

@@ -1030,7 +1030,7 @@ mod tests {
         let segment = searcher.segment_reader(0);
         let top_collector = TopDocs::with_limit(4).order_by_u64_field(SIZE);
         let err = top_collector.for_segment(0, segment).err().unwrap();
-        assert!(matches!(err, crate::TantivyError::InvalidArgument(_)));
+        assert!(matches!(err, crate::TantivyError::SchemaError(_)));
         Ok(())
     }
 

src/core/segment_reader.rs

@@ -327,7 +327,7 @@ impl SegmentReader {
             self.alive_bitset_opt
                 .as_ref()
                 .map(AliveBitSet::space_usage)
-                .unwrap_or_default(),
+                .unwrap_or(0),
         ))
     }
 }

src/directory/composite_file.rs

@@ -172,7 +172,7 @@ impl CompositeFile {
         let mut fields = Vec::new();
         for (&field_addr, byte_range) in &self.offsets_index {
             let mut field_usage = FieldUsage::empty(field_addr.field);
-            field_usage.add_field_idx(field_addr.idx, byte_range.len().into());
+            field_usage.add_field_idx(field_addr.idx, byte_range.len());
             fields.push(field_usage);
         }
         PerFieldSpaceUsage::new(fields)

src/directory/mmap_directory.rs

@@ -7,7 +7,7 @@ use std::sync::{Arc, RwLock, Weak};
 use std::{fmt, result};
 
 use common::StableDeref;
-use fs4::FileExt;
+use fs2::FileExt;
 use memmap2::Mmap;
 use serde::{Deserialize, Serialize};
 use tempfile::TempDir;

src/docset.rs

@@ -9,8 +9,6 @@ use crate::DocId;
 /// to compare `[u32; 4]`.
 pub const TERMINATED: DocId = i32::MAX as u32;
 
-pub const BUFFER_LEN: usize = 64;
-
 /// Represents an iterable set of sorted doc ids.
 pub trait DocSet: Send {
     /// Goes to the next element.
@@ -61,7 +59,7 @@ pub trait DocSet: Send {
     /// This method is only here for specific high-performance
     /// use case where batching. The normal way to
     /// go through the `DocId`'s is to call `.advance()`.
-    fn fill_buffer(&mut self, buffer: &mut [DocId; BUFFER_LEN]) -> usize {
+    fn fill_buffer(&mut self, buffer: &mut [DocId]) -> usize {
         if self.doc() == TERMINATED {
             return 0;
         }
@@ -151,11 +149,6 @@ impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
         unboxed.seek(target)
     }
 
-    fn fill_buffer(&mut self, buffer: &mut [DocId; BUFFER_LEN]) -> usize {
-        let unboxed: &mut TDocSet = self.borrow_mut();
-        unboxed.fill_buffer(buffer)
-    }
-
     fn doc(&self) -> DocId {
         let unboxed: &TDocSet = self.borrow();
         unboxed.doc()

src/error.rs

@@ -55,7 +55,7 @@ impl fmt::Debug for DataCorruption {
 #[derive(Debug, Clone, Error)]
 pub enum TantivyError {
     /// Error when handling aggregations.
-    #[error(transparent)]
+    #[error("AggregationError {0:?}")]
     AggregationError(#[from] AggregationError),
     /// Failed to open the directory.
     #[error("Failed to open the directory: '{0:?}'")]

src/fastfield/alive_bitset.rs

@@ -1,8 +1,9 @@
 use std::io;
 use std::io::Write;
 
-use common::{intersect_bitsets, BitSet, ByteCount, OwnedBytes, ReadOnlyBitSet};
+use common::{intersect_bitsets, BitSet, OwnedBytes, ReadOnlyBitSet};
 
+use crate::space_usage::ByteCount;
 use crate::DocId;
 
 /// Write an alive `BitSet`

src/fastfield/mod.rs

@@ -80,7 +80,7 @@ mod tests {
     use std::path::Path;
 
     use columnar::{Column, MonotonicallyMappableToU64, StrColumn};
-    use common::{ByteCount, HasLen, TerminatingWrite};
+    use common::{HasLen, TerminatingWrite};
     use once_cell::sync::Lazy;
     use rand::prelude::SliceRandom;
     use rand::rngs::StdRng;
@@ -130,7 +130,7 @@ mod tests {
         }
         let file = directory.open_read(path).unwrap();
 
-        assert_eq!(file.len(), 95);
+        assert_eq!(file.len(), 161);
         let fast_field_readers = FastFieldReaders::open(file, SCHEMA.clone()).unwrap();
         let column = fast_field_readers
             .u64("field")
@@ -180,7 +180,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 123);
+        assert_eq!(file.len(), 189);
         let fast_field_readers = FastFieldReaders::open(file, SCHEMA.clone()).unwrap();
         let col = fast_field_readers
             .u64("field")
@@ -213,7 +213,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 96);
+        assert_eq!(file.len(), 162);
         let fast_field_readers = FastFieldReaders::open(file, SCHEMA.clone()).unwrap();
         let fast_field_reader = fast_field_readers
             .u64("field")
@@ -245,7 +245,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 4491);
+        assert_eq!(file.len(), 4557);
         {
             let fast_field_readers = FastFieldReaders::open(file, SCHEMA.clone()).unwrap();
             let col = fast_field_readers
@@ -278,7 +278,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 267);
+        assert_eq!(file.len(), 333_usize);
 
         {
             let fast_field_readers = FastFieldReaders::open(file, schema).unwrap();
@@ -772,7 +772,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 104);
+        assert_eq!(file.len(), 175);
         let fast_field_readers = FastFieldReaders::open(file, schema).unwrap();
         let bool_col = fast_field_readers.bool("field_bool").unwrap();
         assert_eq!(bool_col.first(0), Some(true));
@@ -804,7 +804,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 116);
+        assert_eq!(file.len(), 187);
         let readers = FastFieldReaders::open(file, schema).unwrap();
         let bool_col = readers.bool("field_bool").unwrap();
         for i in 0..25 {
@@ -829,7 +829,7 @@ mod tests {
             write.terminate().unwrap();
         }
         let file = directory.open_read(path).unwrap();
-        assert_eq!(file.len(), 106);
+        assert_eq!(file.len(), 177);
         let fastfield_readers = FastFieldReaders::open(file, schema).unwrap();
         let col = fastfield_readers.bool("field_bool").unwrap();
         assert_eq!(col.first(0), None);
@@ -862,16 +862,16 @@ mod tests {
     #[test]
     pub fn test_gcd_date() {
         let size_prec_sec = test_gcd_date_with_codec(DatePrecision::Seconds);
-        assert!((1000 * 13 / 8..100 + 1000 * 13 / 8).contains(&size_prec_sec.get_bytes())); // 13 bits per val = ceil(log_2(number of seconds in 2hours);
+        assert!((1000 * 13 / 8..100 + 1000 * 13 / 8).contains(&size_prec_sec)); // 13 bits per val = ceil(log_2(number of seconds in 2hours);
         let size_prec_micros = test_gcd_date_with_codec(DatePrecision::Microseconds);
-        assert!((1000 * 33 / 8..100 + 1000 * 33 / 8).contains(&size_prec_micros.get_bytes()));
+        assert!((1000 * 33 / 8..100 + 1000 * 33 / 8).contains(&size_prec_micros));
         // 33 bits per
         // val = ceil(log_2(number
         // of microsecsseconds
         // in 2hours);
     }
 
-    fn test_gcd_date_with_codec(precision: DatePrecision) -> ByteCount {
+    fn test_gcd_date_with_codec(precision: DatePrecision) -> usize {
         let mut rng = StdRng::seed_from_u64(2u64);
         const T0: i64 = 1_662_345_825_012_529i64;
         const ONE_HOUR_IN_MICROSECS: i64 = 3_600 * 1_000_000;
@@ -1068,8 +1068,8 @@ mod tests {
         let searcher = index.reader().unwrap().searcher();
         let segment_reader = searcher.segment_reader(0u32);
         let fast_fields = segment_reader.fast_fields();
-        let column_without_opt = fast_fields.str("without.hello");
-        assert!(column_without_opt.is_err());
+        let column_without_opt: Option<StrColumn> = fast_fields.str("without.hello").unwrap();
+        assert!(column_without_opt.is_none());
         let column_with_opt: Option<StrColumn> = fast_fields.str("with.hello").unwrap();
         let column_with: StrColumn = column_with_opt.unwrap();
         assert!(column_with.term_ords(0).next().is_none());
@@ -1166,7 +1166,7 @@ mod tests {
         let searcher = index.reader().unwrap().searcher();
         let fast_field_reader = searcher.segment_reader(0u32).fast_fields();
         let column = fast_field_reader
-            .column_opt::<i64>("jsonfield.attr.age")
+            .column_opt::<i64>(&"jsonfield.attr.age")
             .unwrap()
             .unwrap();
         let vals: Vec<i64> = column.values_for_doc(0u32).collect();
@@ -1191,7 +1191,7 @@ mod tests {
         let searcher = index.reader().unwrap().searcher();
         let fast_field_reader = searcher.segment_reader(0u32).fast_fields();
         let column = fast_field_reader
-            .column_opt::<i64>("jsonfield.attr.age")
+            .column_opt::<i64>(&"jsonfield.attr.age")
             .unwrap()
             .unwrap();
         let vals: Vec<i64> = column.values_for_doc(0u32).collect();

src/fastfield/readers.rs

@@ -6,13 +6,11 @@ use columnar::{
     BytesColumn, Column, ColumnType, ColumnValues, ColumnarReader, DynamicColumn,
     DynamicColumnHandle, HasAssociatedColumnType, StrColumn,
 };
-use common::ByteCount;
 
 use crate::core::json_utils::encode_column_name;
 use crate::directory::FileSlice;
 use crate::schema::{Field, FieldEntry, FieldType, Schema};
 use crate::space_usage::{FieldUsage, PerFieldSpaceUsage};
-use crate::TantivyError;
 
 /// Provides access to all of the BitpackedFastFieldReader.
 ///
@@ -30,7 +28,7 @@ impl FastFieldReaders {
         Ok(FastFieldReaders { columnar, schema })
     }
 
-    fn resolve_field(&self, column_name: &str) -> crate::Result<Option<String>> {
+    fn resolve_field(&self, column_name: &str) -> Option<String> {
         let default_field_opt: Option<Field> = if cfg!(feature = "quickwit") {
             self.schema.get_field("_dynamic").ok()
         } else {
@@ -43,7 +41,7 @@ impl FastFieldReaders {
         let mut per_field_usages: Vec<FieldUsage> = Default::default();
         for (field, field_entry) in schema.fields() {
             let column_handles = self.columnar.read_columns(field_entry.name())?;
-            let num_bytes: ByteCount = column_handles
+            let num_bytes: usize = column_handles
                 .iter()
                 .map(|column_handle| column_handle.num_bytes())
                 .sum();
@@ -84,35 +82,27 @@ impl FastFieldReaders {
         &'a self,
         field_name: &'a str,
         default_field_opt: Option<Field>,
-    ) -> crate::Result<Option<String>> {
-        let Some((field, path)): Option<(Field, &str)> = self
+    ) -> Option<String> {
+        let (field, path): (Field, &str) = self
             .schema
             .find_field(field_name)
-            .or_else(|| default_field_opt.map(|default_field| (default_field, field_name)))
-        else{
-            return Ok(None);
-        };
-        let field_entry: &FieldEntry = self.schema.get_field_entry(field);
-        if !field_entry.is_fast() {
-            return Err(TantivyError::InvalidArgument(format!(
-                "Field {field_name:?} is not configured as fast field"
-            )));
-        }
+            .or_else(|| default_field_opt.map(|default_field| (default_field, field_name)))?;
         let field_name = self.schema.get_field_name(field);
         if path.is_empty() {
-            return Ok(Some(field_name.to_string()));
+            return Some(field_name.to_string());
         }
+        let field_entry: &FieldEntry = self.schema.get_field_entry(field);
         let field_type = field_entry.field_type();
         match (field_type, path) {
             (FieldType::JsonObject(json_options), path) if !path.is_empty() => {
-                Ok(Some(encode_column_name(
+                Some(encode_column_name(
                     field_entry.name(),
                     path,
                     json_options.is_expand_dots_enabled(),
-                )))
+                ))
             }
-            (_, "") => Ok(Some(field_entry.name().to_string())),
-            _ => Ok(None),
+            (_, "") => Some(field_entry.name().to_string()),
+            _ => None,
         }
     }
 
@@ -137,9 +127,9 @@ impl FastFieldReaders {
     /// Returns the number of `bytes` associated with a column.
     ///
     /// Returns 0 if the column does not exist.
-    pub fn column_num_bytes(&self, field: &str) -> crate::Result<ByteCount> {
-        let Some(resolved_field_name) = self.resolve_field(field)? else {
-            return Ok(0u64.into());
+    pub fn column_num_bytes(&self, field: &str) -> crate::Result<usize> {
+        let Some(resolved_field_name) = self.resolve_field(field) else {
+            return Ok(0);
         };
         Ok(self
             .columnar
@@ -226,7 +216,7 @@ impl FastFieldReaders {
         field_name: &str,
         column_type: ColumnType,
     ) -> crate::Result<Option<DynamicColumnHandle>> {
-        let Some(resolved_field_name) = self.resolve_field(field_name)? else {
+        let Some(resolved_field_name) = self.resolve_field(field_name) else {
             return Ok(None);
         };
         let dynamic_column_handle_opt = self
@@ -242,7 +232,7 @@ impl FastFieldReaders {
         &self,
         field_name: &str,
     ) -> crate::Result<Vec<DynamicColumnHandle>> {
-        let Some(resolved_field_name) = self.resolve_field(field_name)? else {
+        let Some(resolved_field_name) = self.resolve_field(field_name) else {
             return Ok(Vec::new());
         };
         let columns = self
@@ -261,14 +251,12 @@ impl FastFieldReaders {
     }
 
     /// Returns the `u64` column used to represent any `u64`-mapped typed (i64, u64, f64, DateTime).
-    ///
-    /// Returns Ok(None) for empty columns
     #[doc(hidden)]
     pub fn u64_lenient_with_type(
         &self,
         field_name: &str,
     ) -> crate::Result<Option<(Column<u64>, ColumnType)>> {
-        let Some(resolved_field_name) = self.resolve_field(field_name)? else {
+        let Some(resolved_field_name) = self.resolve_field(field_name) else {
             return Ok(None);
         };
         for col in self.columnar.read_columns(&resolved_field_name)? {
@@ -328,57 +316,44 @@ mod tests {
         let reader = searcher.segment_reader(0u32);
         let fast_field_readers = reader.fast_fields();
         assert_eq!(
-            fast_field_readers
-                .resolve_column_name_given_default_field("age", None)
-                .unwrap(),
+            fast_field_readers.resolve_column_name_given_default_field("age", None),
             Some("age".to_string())
         );
         assert_eq!(
-            fast_field_readers
-                .resolve_column_name_given_default_field("age", Some(dynamic_field))
-                .unwrap(),
+            fast_field_readers.resolve_column_name_given_default_field("age", Some(dynamic_field)),
             Some("age".to_string())
         );
         assert_eq!(
-            fast_field_readers
-                .resolve_column_name_given_default_field(
-                    "json_expand_dots_disabled.attr.color",
-                    None
-                )
-                .unwrap(),
+            fast_field_readers.resolve_column_name_given_default_field(
+                "json_expand_dots_disabled.attr.color",
+                None
+            ),
             Some("json_expand_dots_disabled\u{1}attr\u{1}color".to_string())
         );
         assert_eq!(
-            fast_field_readers
-                .resolve_column_name_given_default_field(
-                    "json_expand_dots_disabled.attr\\.color",
-                    Some(dynamic_field)
-                )
-                .unwrap(),
+            fast_field_readers.resolve_column_name_given_default_field(
+                "json_expand_dots_disabled.attr\\.color",
+                Some(dynamic_field)
+            ),
             Some("json_expand_dots_disabled\u{1}attr.color".to_string())
         );
         assert_eq!(
-            fast_field_readers
-                .resolve_column_name_given_default_field(
-                    "json_expand_dots_enabled.attr\\.color",
-                    Some(dynamic_field)
-                )
-                .unwrap(),
+            fast_field_readers.resolve_column_name_given_default_field(
+                "json_expand_dots_enabled.attr\\.color",
+                Some(dynamic_field)
+            ),
             Some("json_expand_dots_enabled\u{1}attr\u{1}color".to_string())
         );
         assert_eq!(
             fast_field_readers
-                .resolve_column_name_given_default_field("notinschema.attr.color", None)
-                .unwrap(),
+                .resolve_column_name_given_default_field("notinschema.attr.color", None),
             None
         );
         assert_eq!(
-            fast_field_readers
-                .resolve_column_name_given_default_field(
-                    "notinschema.attr.color",
-                    Some(dynamic_field)
-                )
-                .unwrap(),
+            fast_field_readers.resolve_column_name_given_default_field(
+                "notinschema.attr.color",
+                Some(dynamic_field)
+            ),
             Some("_dyna\u{1}notinschema\u{1}attr\u{1}color".to_string())
         );
     }

src/indexer/index_writer.rs

@@ -94,12 +94,10 @@ fn compute_deleted_bitset(
         // document that were inserted before it.
         delete_op
             .target
-            .for_each_no_score(segment_reader, &mut |docs_matching_delete_query| {
-                for doc_matching_delete_query in docs_matching_delete_query.iter().cloned() {
-                    if doc_opstamps.is_deleted(doc_matching_delete_query, delete_op.opstamp) {
-                        alive_bitset.remove(doc_matching_delete_query);
-                        might_have_changed = true;
-                    }
+            .for_each_no_score(segment_reader, &mut |doc_matching_delete_query| {
+                if doc_opstamps.is_deleted(doc_matching_delete_query, delete_op.opstamp) {
+                    alive_bitset.remove(doc_matching_delete_query);
+                    might_have_changed = true;
                 }
             })?;
         delete_cursor.advance();

src/indexer/segment_writer.rs

@@ -188,7 +188,7 @@ impl SegmentWriter {
                         let mut indexing_position = IndexingPosition::default();
                         postings_writer.index_text(
                             doc_id,
-                            &mut facet_tokenizer,
+                            &mut *facet_tokenizer,
                             term_buffer,
                             ctx,
                             &mut indexing_position,

src/query/all_query.rs

@@ -1,5 +1,5 @@
 use crate::core::SegmentReader;
-use crate::docset::{DocSet, BUFFER_LEN, TERMINATED};
+use crate::docset::{DocSet, TERMINATED};
 use crate::query::boost_query::BoostScorer;
 use crate::query::explanation::does_not_match;
 use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
@@ -44,7 +44,6 @@ pub struct AllScorer {
 }
 
 impl DocSet for AllScorer {
-    #[inline(always)]
     fn advance(&mut self) -> DocId {
         if self.doc + 1 >= self.max_doc {
             self.doc = TERMINATED;
@@ -54,30 +53,6 @@ impl DocSet for AllScorer {
         self.doc
     }
 
-    fn fill_buffer(&mut self, buffer: &mut [DocId; BUFFER_LEN]) -> usize {
-        if self.doc() == TERMINATED {
-            return 0;
-        }
-        let is_safe_distance = self.doc() + (buffer.len() as u32) < self.max_doc;
-        if is_safe_distance {
-            let num_items = buffer.len();
-            for buffer_val in buffer {
-                *buffer_val = self.doc();
-                self.doc += 1;
-            }
-            num_items
-        } else {
-            for (i, buffer_val) in buffer.iter_mut().enumerate() {
-                *buffer_val = self.doc();
-                if self.advance() == TERMINATED {
-                    return i + 1;
-                }
-            }
-            buffer.len()
-        }
-    }
-
-    #[inline(always)]
     fn doc(&self) -> DocId {
         self.doc
     }
@@ -96,8 +71,8 @@ impl Scorer for AllScorer {
 #[cfg(test)]
 mod tests {
     use super::AllQuery;
-    use crate::docset::{DocSet, BUFFER_LEN, TERMINATED};
-    use crate::query::{AllScorer, EnableScoring, Query};
+    use crate::docset::TERMINATED;
+    use crate::query::{EnableScoring, Query};
     use crate::schema::{Schema, TEXT};
     use crate::Index;
 
@@ -157,22 +132,4 @@ mod tests {
         }
         Ok(())
     }
-
-    #[test]
-    pub fn test_fill_buffer() {
-        let mut postings = AllScorer {
-            doc: 0u32,
-            max_doc: BUFFER_LEN as u32 * 2 + 9,
-        };
-        let mut buffer = [0u32; BUFFER_LEN];
-        assert_eq!(postings.fill_buffer(&mut buffer), BUFFER_LEN);
-        for i in 0u32..BUFFER_LEN as u32 {
-            assert_eq!(buffer[i as usize], i);
-        }
-        assert_eq!(postings.fill_buffer(&mut buffer), BUFFER_LEN);
-        for i in 0u32..BUFFER_LEN as u32 {
-            assert_eq!(buffer[i as usize], i + BUFFER_LEN as u32);
-        }
-        assert_eq!(postings.fill_buffer(&mut buffer), 9);
-    }
 }

src/query/bitset/mod.rs

@@ -45,7 +45,6 @@ impl From<BitSet> for BitSetDocSet {
 }
 
 impl DocSet for BitSetDocSet {
-    #[inline]
     fn advance(&mut self) -> DocId {
         if let Some(lower) = self.cursor_tinybitset.pop_lowest() {
             self.doc = (self.cursor_bucket * 64u32) | lower;

src/query/boolean_query/boolean_weight.rs

@@ -1,12 +1,11 @@
 use std::collections::HashMap;
 
 use crate::core::SegmentReader;
-use crate::docset::BUFFER_LEN;
 use crate::postings::FreqReadingOption;
 use crate::query::explanation::does_not_match;
 use crate::query::score_combiner::{DoNothingCombiner, ScoreCombiner};
 use crate::query::term_query::TermScorer;
-use crate::query::weight::{for_each_docset_buffered, for_each_pruning_scorer, for_each_scorer};
+use crate::query::weight::{for_each_docset, for_each_pruning_scorer, for_each_scorer};
 use crate::query::{
     intersect_scorers, EmptyScorer, Exclude, Explanation, Occur, RequiredOptionalScorer, Scorer,
     Union, Weight,
@@ -223,18 +222,16 @@ impl<TScoreCombiner: ScoreCombiner + Sync> Weight for BooleanWeight<TScoreCombin
     fn for_each_no_score(
         &self,
         reader: &SegmentReader,
-        callback: &mut dyn FnMut(&[DocId]),
+        callback: &mut dyn FnMut(DocId),
     ) -> crate::Result<()> {
         let scorer = self.complex_scorer(reader, 1.0, || DoNothingCombiner)?;
-        let mut buffer = [0u32; BUFFER_LEN];
-
         match scorer {
             SpecializedScorer::TermUnion(term_scorers) => {
                 let mut union_scorer = Union::build(term_scorers, &self.score_combiner_fn);
-                for_each_docset_buffered(&mut union_scorer, &mut buffer, callback);
+                for_each_docset(&mut union_scorer, callback);
             }
             SpecializedScorer::Other(mut scorer) => {
-                for_each_docset_buffered(scorer.as_mut(), &mut buffer, callback);
+                for_each_docset(scorer.as_mut(), callback);
             }
         }
         Ok(())

src/query/boost_query.rs

@@ -1,6 +1,5 @@
 use std::fmt;
 
-use crate::docset::BUFFER_LEN;
 use crate::fastfield::AliveBitSet;
 use crate::query::explanation::does_not_match;
 use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
@@ -107,7 +106,7 @@ impl<S: Scorer> DocSet for BoostScorer<S> {
         self.underlying.seek(target)
     }
 
-    fn fill_buffer(&mut self, buffer: &mut [DocId; BUFFER_LEN]) -> usize {
+    fn fill_buffer(&mut self, buffer: &mut [DocId]) -> usize {
         self.underlying.fill_buffer(buffer)
     }
 

src/query/const_score_query.rs

@@ -1,6 +1,5 @@
 use std::fmt;
 
-use crate::docset::BUFFER_LEN;
 use crate::query::{EnableScoring, Explanation, Query, Scorer, Weight};
 use crate::{DocId, DocSet, Score, SegmentReader, TantivyError, Term};
 
@@ -120,7 +119,7 @@ impl<TDocSet: DocSet> DocSet for ConstScorer<TDocSet> {
         self.docset.seek(target)
     }
 
-    fn fill_buffer(&mut self, buffer: &mut [DocId; BUFFER_LEN]) -> usize {
+    fn fill_buffer(&mut self, buffer: &mut [DocId]) -> usize {
         self.docset.fill_buffer(buffer)
     }
 

src/query/more_like_this/more_like_this.rs

@@ -4,9 +4,7 @@ use std::collections::{BinaryHeap, HashMap};
 use crate::query::bm25::idf;
 use crate::query::{BooleanQuery, BoostQuery, Occur, Query, TermQuery};
 use crate::schema::{Field, FieldType, IndexRecordOption, Term, Value};
-use crate::tokenizer::{
-    BoxTokenStream, FacetTokenizer, PreTokenizedStream, TokenStream, Tokenizer,
-};
+use crate::tokenizer::{BoxTokenStream, FacetTokenizer, PreTokenizedStream, Tokenizer};
 use crate::{DocAddress, Result, Searcher, TantivyError};
 
 #[derive(Debug, PartialEq)]

src/query/query_parser/query_parser.rs

@@ -913,10 +913,9 @@ mod test {
         let tokenizer_manager = TokenizerManager::default();
         tokenizer_manager.register(
             "en_with_stop_words",
-            TextAnalyzer::builder(SimpleTokenizer)
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(LowerCaser)
-                .filter(StopWordFilter::remove(vec!["the".to_string()]))
-                .build(),
+                .filter(StopWordFilter::remove(vec!["the".to_string()])),
         );
         QueryParser::new(schema, default_fields, tokenizer_manager)
     }

src/query/term_query/term_weight.rs

@@ -1,11 +1,11 @@
 use super::term_scorer::TermScorer;
 use crate::core::SegmentReader;
-use crate::docset::{DocSet, BUFFER_LEN};
+use crate::docset::DocSet;
 use crate::fieldnorm::FieldNormReader;
 use crate::postings::SegmentPostings;
 use crate::query::bm25::Bm25Weight;
 use crate::query::explanation::does_not_match;
-use crate::query::weight::{for_each_docset_buffered, for_each_scorer};
+use crate::query::weight::{for_each_docset, for_each_scorer};
 use crate::query::{Explanation, Scorer, Weight};
 use crate::schema::IndexRecordOption;
 use crate::{DocId, Score, Term};
@@ -61,11 +61,10 @@ impl Weight for TermWeight {
     fn for_each_no_score(
         &self,
         reader: &SegmentReader,
-        callback: &mut dyn FnMut(&[DocId]),
+        callback: &mut dyn FnMut(DocId),
     ) -> crate::Result<()> {
         let mut scorer = self.specialized_scorer(reader, 1.0)?;
-        let mut buffer = [0u32; BUFFER_LEN];
-        for_each_docset_buffered(&mut scorer, &mut buffer, callback);
+        for_each_docset(&mut scorer, callback);
         Ok(())
     }
 

src/query/vec_docset.rs

@@ -53,7 +53,7 @@ impl HasLen for VecDocSet {
 pub mod tests {
 
     use super::*;
-    use crate::docset::{DocSet, BUFFER_LEN};
+    use crate::docset::DocSet;
     use crate::DocId;
 
     #[test]
@@ -72,17 +72,17 @@ pub mod tests {
 
     #[test]
     pub fn test_fill_buffer() {
-        let doc_ids: Vec<DocId> = (1u32..=(BUFFER_LEN as u32 * 2 + 9)).collect();
+        let doc_ids: Vec<DocId> = (1u32..210u32).collect();
         let mut postings = VecDocSet::from(doc_ids);
-        let mut buffer = [0u32; BUFFER_LEN];
-        assert_eq!(postings.fill_buffer(&mut buffer), BUFFER_LEN);
-        for i in 0u32..BUFFER_LEN as u32 {
+        let mut buffer = vec![1000u32; 100];
+        assert_eq!(postings.fill_buffer(&mut buffer[..]), 100);
+        for i in 0u32..100u32 {
             assert_eq!(buffer[i as usize], i + 1);
         }
-        assert_eq!(postings.fill_buffer(&mut buffer), BUFFER_LEN);
-        for i in 0u32..BUFFER_LEN as u32 {
-            assert_eq!(buffer[i as usize], i + 1 + BUFFER_LEN as u32);
+        assert_eq!(postings.fill_buffer(&mut buffer[..]), 100);
+        for i in 0u32..100u32 {
+            assert_eq!(buffer[i as usize], i + 101);
         }
-        assert_eq!(postings.fill_buffer(&mut buffer), 9);
+        assert_eq!(postings.fill_buffer(&mut buffer[..]), 9);
     }
 }

src/query/weight.rs

@@ -1,6 +1,5 @@
 use super::Scorer;
 use crate::core::SegmentReader;
-use crate::docset::BUFFER_LEN;
 use crate::query::Explanation;
 use crate::{DocId, DocSet, Score, TERMINATED};
 
@@ -19,18 +18,11 @@ pub(crate) fn for_each_scorer<TScorer: Scorer + ?Sized>(
 
 /// Iterates through all of the documents matched by the DocSet
 /// `DocSet`.
-#[inline]
-pub(crate) fn for_each_docset_buffered<T: DocSet + ?Sized>(
-    docset: &mut T,
-    buffer: &mut [DocId; BUFFER_LEN],
-    mut callback: impl FnMut(&[DocId]),
-) {
-    loop {
-        let num_items = docset.fill_buffer(buffer);
-        callback(&buffer[..num_items]);
-        if num_items != buffer.len() {
-            break;
-        }
+pub(crate) fn for_each_docset<T: DocSet + ?Sized>(docset: &mut T, callback: &mut dyn FnMut(DocId)) {
+    let mut doc = docset.doc();
+    while doc != TERMINATED {
+        callback(doc);
+        doc = docset.advance();
     }
 }
 
@@ -101,12 +93,10 @@ pub trait Weight: Send + Sync + 'static {
     fn for_each_no_score(
         &self,
         reader: &SegmentReader,
-        callback: &mut dyn FnMut(&[DocId]),
+        callback: &mut dyn FnMut(DocId),
     ) -> crate::Result<()> {
         let mut docset = self.scorer(reader, 1.0)?;
-
-        let mut buffer = [0u32; BUFFER_LEN];
-        for_each_docset_buffered(&mut docset, &mut buffer, callback);
+        for_each_docset(docset.as_mut(), callback);
         Ok(())
     }
 

src/schema/term.rs

@@ -364,8 +364,7 @@ where B: AsRef<[u8]>
     /// (this does not include the field.)
     ///
     /// If the term is a string, its value is utf-8 encoded.
-    /// If the term is a u64, its value is encoded according
-    /// to `byteorder::BigEndian`.
+    /// If the term is a u64, its value is encoded in big endian.
     pub fn value_bytes(&self) -> &[u8] {
         &self.0.as_ref()[TERM_METADATA_LENGTH..]
     }

src/space_usage/mod.rs

@@ -9,12 +9,14 @@
 
 use std::collections::HashMap;
 
-use common::ByteCount;
 use serde::{Deserialize, Serialize};
 
 use crate::schema::Field;
 use crate::SegmentComponent;
 
+/// Indicates space usage in bytes
+pub type ByteCount = usize;
+
 /// Enum containing any of the possible space usage results for segment components.
 pub enum ComponentSpaceUsage {
     /// Data is stored per field in a uniform way
@@ -36,7 +38,7 @@ impl SearcherSpaceUsage {
     pub(crate) fn new() -> SearcherSpaceUsage {
         SearcherSpaceUsage {
             segments: Vec::new(),
-            total: Default::default(),
+            total: 0,
         }
     }
 
@@ -258,7 +260,7 @@ impl FieldUsage {
     pub(crate) fn empty(field: Field) -> FieldUsage {
         FieldUsage {
             field,
-            num_bytes: Default::default(),
+            num_bytes: 0,
             sub_num_bytes: Vec::new(),
         }
     }
@@ -292,7 +294,7 @@ impl FieldUsage {
 mod test {
     use crate::core::Index;
     use crate::schema::{Field, Schema, FAST, INDEXED, STORED, TEXT};
-    use crate::space_usage::PerFieldSpaceUsage;
+    use crate::space_usage::{ByteCount, PerFieldSpaceUsage};
     use crate::Term;
 
     #[test]
@@ -302,14 +304,14 @@ mod test {
         let reader = index.reader().unwrap();
         let searcher = reader.searcher();
         let searcher_space_usage = searcher.space_usage().unwrap();
-        assert_eq!(searcher_space_usage.total(), 0u64);
+        assert_eq!(0, searcher_space_usage.total());
     }
 
     fn expect_single_field(
         field_space: &PerFieldSpaceUsage,
         field: &Field,
-        min_size: u64,
-        max_size: u64,
+        min_size: ByteCount,
+        max_size: ByteCount,
     ) {
         assert!(field_space.total() >= min_size);
         assert!(field_space.total() <= max_size);
@@ -351,12 +353,12 @@ mod test {
 
         expect_single_field(segment.termdict(), &name, 1, 512);
         expect_single_field(segment.postings(), &name, 1, 512);
-        assert_eq!(segment.positions().total(), 0);
+        assert_eq!(0, segment.positions().total());
         expect_single_field(segment.fast_fields(), &name, 1, 512);
         expect_single_field(segment.fieldnorms(), &name, 1, 512);
         // TODO: understand why the following fails
         //        assert_eq!(0, segment.store().total());
-        assert_eq!(segment.deletes(), 0);
+        assert_eq!(0, segment.deletes());
         Ok(())
     }
 
@@ -392,11 +394,11 @@ mod test {
         expect_single_field(segment.termdict(), &name, 1, 512);
         expect_single_field(segment.postings(), &name, 1, 512);
         expect_single_field(segment.positions(), &name, 1, 512);
-        assert_eq!(segment.fast_fields().total(), 0);
+        assert_eq!(0, segment.fast_fields().total());
         expect_single_field(segment.fieldnorms(), &name, 1, 512);
         // TODO: understand why the following fails
         //        assert_eq!(0, segment.store().total());
-        assert_eq!(segment.deletes(), 0);
+        assert_eq!(0, segment.deletes());
         Ok(())
     }
 
@@ -428,14 +430,14 @@ mod test {
 
         assert_eq!(4, segment.num_docs());
 
-        assert_eq!(segment.termdict().total(), 0);
-        assert_eq!(segment.postings().total(), 0);
-        assert_eq!(segment.positions().total(), 0);
-        assert_eq!(segment.fast_fields().total(), 0);
-        assert_eq!(segment.fieldnorms().total(), 0);
+        assert_eq!(0, segment.termdict().total());
+        assert_eq!(0, segment.postings().total());
+        assert_eq!(0, segment.positions().total());
+        assert_eq!(0, segment.fast_fields().total());
+        assert_eq!(0, segment.fieldnorms().total());
         assert!(segment.store().total() > 0);
         assert!(segment.store().total() < 512);
-        assert_eq!(segment.deletes(), 0);
+        assert_eq!(0, segment.deletes());
         Ok(())
     }
 
@@ -476,8 +478,8 @@ mod test {
 
         expect_single_field(segment_space_usage.termdict(), &name, 1, 512);
         expect_single_field(segment_space_usage.postings(), &name, 1, 512);
-        assert_eq!(segment_space_usage.positions().total(), 0u64);
-        assert_eq!(segment_space_usage.fast_fields().total(), 0u64);
+        assert_eq!(0, segment_space_usage.positions().total());
+        assert_eq!(0, segment_space_usage.fast_fields().total());
         expect_single_field(segment_space_usage.fieldnorms(), &name, 1, 512);
         assert!(segment_space_usage.deletes() > 0);
         Ok(())

src/store/reader.rs

@@ -5,7 +5,7 @@ use std::ops::{AddAssign, Range};
 use std::sync::atomic::{AtomicUsize, Ordering};
 use std::sync::{Arc, Mutex};
 
-use common::{BinarySerializable, OwnedBytes};
+use common::{BinarySerializable, HasLen, OwnedBytes};
 use lru::LruCache;
 
 use super::footer::DocStoreFooter;
@@ -122,8 +122,7 @@ impl StoreReader {
 
         let (data_file, offset_index_file) = data_and_offset.split(footer.offset as usize);
         let index_data = offset_index_file.read_bytes()?;
-        let space_usage =
-            StoreSpaceUsage::new(data_file.num_bytes(), offset_index_file.num_bytes());
+        let space_usage = StoreSpaceUsage::new(data_file.len(), offset_index_file.len());
         let skip_index = SkipIndex::open(index_data);
         Ok(StoreReader {
             decompressor: footer.decompressor,

src/termdict/fst_termdict/term_info_store.rs

@@ -1,7 +1,6 @@
 use std::cmp;
 use std::io::{self, Read, Write};
 
-use byteorder::{ByteOrder, LittleEndian};
 use common::{BinarySerializable, FixedSize};
 use tantivy_bitpacker::{compute_num_bits, BitPacker};
 
@@ -104,7 +103,7 @@ fn extract_bits(data: &[u8], addr_bits: usize, num_bits: u8) -> u64 {
     let addr_byte = addr_bits / 8;
     let bit_shift = (addr_bits % 8) as u64;
     let val_unshifted_unmasked: u64 = if data.len() >= addr_byte + 8 {
-        LittleEndian::read_u64(&data[addr_byte..][..8])
+        u64::from_le_bytes(data[addr_byte..][..8].try_into().unwrap())
     } else {
         // the buffer is not large enough.
         // Let's copy the few remaining bytes to a 8 byte buffer
@@ -113,7 +112,7 @@ fn extract_bits(data: &[u8], addr_bits: usize, num_bits: u8) -> u64 {
         let data_to_copy = &data[addr_byte..];
         let nbytes = data_to_copy.len();
         buf[..nbytes].copy_from_slice(data_to_copy);
-        LittleEndian::read_u64(&buf)
+        u64::from_le_bytes(buf)
     };
     let val_shifted_unmasked = val_unshifted_unmasked >> bit_shift;
     let mask = (1u64 << u64::from(num_bits)) - 1;

src/tokenizer/alphanum_only.rs

@@ -2,18 +2,16 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = TextAnalyzer::builder(RawTokenizer)
-//!   .filter(AlphaNumOnlyFilter)
-//!   .build();
+//! let tokenizer = TextAnalyzer::from(RawTokenizer)
+//!   .filter(AlphaNumOnlyFilter);
 //!
 //! let mut stream = tokenizer.token_stream("hello there");
 //! // is none because the raw filter emits one token that
 //! // contains a space
 //! assert!(stream.next().is_none());
 //!
-//! let tokenizer = TextAnalyzer::builder(SimpleTokenizer)
-//!   .filter(AlphaNumOnlyFilter)
-//!   .build();
+//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
+//!   .filter(AlphaNumOnlyFilter);
 //!
 //! let mut stream = tokenizer.token_stream("hello there 💣");
 //! assert!(stream.next().is_some());
@@ -21,45 +19,30 @@
 //! // the "emoji" is dropped because its not an alphanum
 //! assert!(stream.next().is_none());
 //! ```
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
 
 /// `TokenFilter` that removes all tokens that contain non
 /// ascii alphanumeric characters.
 #[derive(Clone)]
 pub struct AlphaNumOnlyFilter;
 
-pub struct AlphaNumOnlyFilterStream<T> {
-    tail: T,
+pub struct AlphaNumOnlyFilterStream<'a> {
+    tail: BoxTokenStream<'a>,
 }
 
-impl<T> AlphaNumOnlyFilterStream<T> {
+impl<'a> AlphaNumOnlyFilterStream<'a> {
     fn predicate(&self, token: &Token) -> bool {
         token.text.chars().all(|c| c.is_ascii_alphanumeric())
     }
 }
 
 impl TokenFilter for AlphaNumOnlyFilter {
-    type Tokenizer<T: Tokenizer> = AlphaNumOnlyFilterWrapper<T>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> AlphaNumOnlyFilterWrapper<T> {
-        AlphaNumOnlyFilterWrapper(tokenizer)
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(AlphaNumOnlyFilterStream { tail: token_stream })
     }
 }
 
-#[derive(Clone)]
-pub struct AlphaNumOnlyFilterWrapper<T>(T);
-
-impl<T: Tokenizer> Tokenizer for AlphaNumOnlyFilterWrapper<T> {
-    type TokenStream<'a> = AlphaNumOnlyFilterStream<T::TokenStream<'a>>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        AlphaNumOnlyFilterStream {
-            tail: self.0.token_stream(text),
-        }
-    }
-}
-
-impl<T: TokenStream> TokenStream for AlphaNumOnlyFilterStream<T> {
+impl<'a> TokenStream for AlphaNumOnlyFilterStream<'a> {
     fn advance(&mut self) -> bool {
         while self.tail.advance() {
             if self.predicate(self.tail.token()) {
@@ -96,9 +79,7 @@ mod tests {
     }
 
     fn token_stream_helper(text: &str) -> Vec<Token> {
-        let a = TextAnalyzer::builder(SimpleTokenizer)
-            .filter(AlphaNumOnlyFilter)
-            .build();
+        let a = TextAnalyzer::from(SimpleTokenizer).filter(AlphaNumOnlyFilter);
         let mut token_stream = a.token_stream(text);
         let mut tokens: Vec<Token> = vec![];
         let mut add_token = |token: &Token| {

src/tokenizer/ascii_folding_filter.rs

@@ -1,6 +1,6 @@
 use std::mem;
 
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
 
 /// This class converts alphabetic, numeric, and symbolic Unicode characters
 /// which are not in the first 127 ASCII characters (the "Basic Latin" Unicode
@@ -9,33 +9,20 @@ use super::{Token, TokenFilter, TokenStream, Tokenizer};
 pub struct AsciiFoldingFilter;
 
 impl TokenFilter for AsciiFoldingFilter {
-    type Tokenizer<T: Tokenizer> = AsciiFoldingFilterWrapper<T>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> AsciiFoldingFilterWrapper<T> {
-        AsciiFoldingFilterWrapper(tokenizer)
-    }
-}
-
-#[derive(Clone)]
-pub struct AsciiFoldingFilterWrapper<T>(T);
-
-impl<T: Tokenizer> Tokenizer for AsciiFoldingFilterWrapper<T> {
-    type TokenStream<'a> = AsciiFoldingFilterTokenStream<T::TokenStream<'a>>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        AsciiFoldingFilterTokenStream {
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        From::from(AsciiFoldingFilterTokenStream {
+            tail: token_stream,
             buffer: String::with_capacity(100),
-            tail: self.0.token_stream(text),
-        }
+        })
     }
 }
 
-pub struct AsciiFoldingFilterTokenStream<T> {
+pub struct AsciiFoldingFilterTokenStream<'a> {
     buffer: String,
-    tail: T,
+    tail: BoxTokenStream<'a>,
 }
 
-impl<T: TokenStream> TokenStream for AsciiFoldingFilterTokenStream<T> {
+impl<'a> TokenStream for AsciiFoldingFilterTokenStream<'a> {
     fn advance(&mut self) -> bool {
         if !self.tail.advance() {
             return false;
@@ -1573,9 +1560,8 @@ mod tests {
 
     fn folding_helper(text: &str) -> Vec<String> {
         let mut tokens = Vec::new();
-        TextAnalyzer::builder(SimpleTokenizer)
+        TextAnalyzer::from(SimpleTokenizer)
             .filter(AsciiFoldingFilter)
-            .build()
             .token_stream(text)
             .process(&mut |token| {
                 tokens.push(token.text.clone());
@@ -1584,9 +1570,8 @@ mod tests {
     }
 
     fn folding_using_raw_tokenizer_helper(text: &str) -> String {
-        let mut token_stream = TextAnalyzer::builder(RawTokenizer)
+        let mut token_stream = TextAnalyzer::from(RawTokenizer)
             .filter(AsciiFoldingFilter)
-            .build()
             .token_stream(text);
         token_stream.advance();
         token_stream.token().text.clone()

src/tokenizer/empty_tokenizer.rs

@@ -1,17 +1,16 @@
-use crate::tokenizer::{Token, TokenStream, Tokenizer};
+use crate::tokenizer::{BoxTokenStream, Token, TokenStream, Tokenizer};
 
 #[derive(Clone)]
 pub(crate) struct EmptyTokenizer;
 
 impl Tokenizer for EmptyTokenizer {
-    type TokenStream<'a> = EmptyTokenStream;
-    fn token_stream(&self, _text: &str) -> EmptyTokenStream {
-        EmptyTokenStream::default()
+    fn token_stream<'a>(&self, _text: &'a str) -> BoxTokenStream<'a> {
+        EmptyTokenStream::default().into()
     }
 }
 
 #[derive(Default)]
-pub struct EmptyTokenStream {
+struct EmptyTokenStream {
     token: Token,
 }
 
@@ -31,7 +30,7 @@ impl TokenStream for EmptyTokenStream {
 
 #[cfg(test)]
 mod tests {
-    use crate::tokenizer::{TokenStream, Tokenizer};
+    use crate::tokenizer::Tokenizer;
 
     #[test]
     fn test_empty_tokenizer() {

src/tokenizer/facet_tokenizer.rs

@@ -1,4 +1,4 @@
-use super::{Token, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenStream, Tokenizer};
 use crate::schema::FACET_SEP_BYTE;
 
 /// The `FacetTokenizer` process a `Facet` binary representation
@@ -26,8 +26,7 @@ pub struct FacetTokenStream<'a> {
 }
 
 impl Tokenizer for FacetTokenizer {
-    type TokenStream<'a> = FacetTokenStream<'a>;
-    fn token_stream<'a>(&self, text: &'a str) -> FacetTokenStream<'a> {
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
         let token = Token {
             position: 0,
             ..Default::default()
@@ -37,6 +36,7 @@ impl Tokenizer for FacetTokenizer {
             state: State::RootFacetNotEmitted, //< pos is the first char that has not been processed yet.
             token,
         }
+        .into()
     }
 }
 
@@ -87,7 +87,7 @@ mod tests {
 
     use super::FacetTokenizer;
     use crate::schema::Facet;
-    use crate::tokenizer::{Token, TokenStream, Tokenizer};
+    use crate::tokenizer::{Token, Tokenizer};
 
     #[test]
     fn test_facet_tokenizer() {

src/tokenizer/lower_caser.rs

@@ -1,42 +1,29 @@
 use std::mem;
 
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
+
+impl TokenFilter for LowerCaser {
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(LowerCaserTokenStream {
+            tail: token_stream,
+            buffer: String::with_capacity(100),
+        })
+    }
+}
 
 /// Token filter that lowercase terms.
 #[derive(Clone)]
 pub struct LowerCaser;
 
-impl TokenFilter for LowerCaser {
-    type Tokenizer<T: Tokenizer> = LowerCaserFilter<T>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> Self::Tokenizer<T> {
-        LowerCaserFilter(tokenizer)
-    }
-}
-
-#[derive(Clone)]
-pub struct LowerCaserFilter<T>(T);
-
-impl<T: Tokenizer> Tokenizer for LowerCaserFilter<T> {
-    type TokenStream<'a> = LowerCaserTokenStream<T::TokenStream<'a>>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        LowerCaserTokenStream {
-            tail: self.0.token_stream(text),
-            buffer: String::new(),
-        }
-    }
-}
-
-pub struct LowerCaserTokenStream<T> {
+pub struct LowerCaserTokenStream<'a> {
     buffer: String,
-    tail: T,
+    tail: BoxTokenStream<'a>,
 }
 
 // writes a lowercased version of text into output.
 fn to_lowercase_unicode(text: &str, output: &mut String) {
     output.clear();
-    output.reserve(50);
     for c in text.chars() {
         // Contrary to the std, we do not take care of sigma special case.
         // This will have an normalizationo effect, which is ok for search.
@@ -44,7 +31,7 @@ fn to_lowercase_unicode(text: &str, output: &mut String) {
     }
 }
 
-impl<T: TokenStream> TokenStream for LowerCaserTokenStream<T> {
+impl<'a> TokenStream for LowerCaserTokenStream<'a> {
     fn advance(&mut self) -> bool {
         if !self.tail.advance() {
             return false;
@@ -86,9 +73,8 @@ mod tests {
     }
 
     fn token_stream_helper(text: &str) -> Vec<Token> {
-        let mut token_stream = TextAnalyzer::builder(SimpleTokenizer)
+        let mut token_stream = TextAnalyzer::from(SimpleTokenizer)
             .filter(LowerCaser)
-            .build()
             .token_stream(text);
         let mut tokens = vec![];
         let mut add_token = |token: &Token| {

src/tokenizer/mod.rs

@@ -66,11 +66,10 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let en_stem = TextAnalyzer::builder(SimpleTokenizer)
+//! let en_stem = TextAnalyzer::from(SimpleTokenizer)
 //!     .filter(RemoveLongFilter::limit(40))
 //!     .filter(LowerCaser)
-//!     .filter(Stemmer::new(Language::English))
-//!     .build();
+//!     .filter(Stemmer::new(Language::English));
 //! ```
 //!
 //! Once your tokenizer is defined, you need to
@@ -113,10 +112,9 @@
 //! let index = Index::create_in_ram(schema);
 //!
 //! // We need to register our tokenizer :
-//! let custom_en_tokenizer = TextAnalyzer::builder(SimpleTokenizer)
+//! let custom_en_tokenizer = TextAnalyzer::from(SimpleTokenizer)
 //!     .filter(RemoveLongFilter::limit(40))
-//!     .filter(LowerCaser)
-//!     .build();
+//!     .filter(LowerCaser);
 //! index
 //!     .tokenizers()
 //!     .register("custom_en", custom_en_tokenizer);
@@ -139,7 +137,9 @@ mod tokenizer;
 mod tokenizer_manager;
 mod whitespace_tokenizer;
 
-pub use tokenizer_api::{BoxTokenStream, Token, TokenFilter, TokenStream, Tokenizer};
+pub use tokenizer_api::{
+    BoxTokenFilter, BoxTokenStream, Token, TokenFilter, TokenStream, Tokenizer,
+};
 
 pub use self::alphanum_only::AlphaNumOnlyFilter;
 pub use self::ascii_folding_filter::AsciiFoldingFilter;
@@ -237,11 +237,10 @@ pub mod tests {
         let tokenizer_manager = TokenizerManager::default();
         tokenizer_manager.register(
             "el_stem",
-            TextAnalyzer::builder(SimpleTokenizer)
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
                 .filter(LowerCaser)
-                .filter(Stemmer::new(Language::Greek))
-                .build(),
+                .filter(Stemmer::new(Language::Greek)),
         );
         let en_tokenizer = tokenizer_manager.get("el_stem").unwrap();
         let mut tokens: Vec<Token> = vec![];

src/tokenizer/ngram_tokenizer.rs

@@ -1,4 +1,5 @@
 use super::{Token, TokenStream, Tokenizer};
+use crate::tokenizer::BoxTokenStream;
 
 /// Tokenize the text by splitting words into n-grams of the given size(s)
 ///
@@ -131,9 +132,8 @@ pub struct NgramTokenStream<'a> {
 }
 
 impl Tokenizer for NgramTokenizer {
-    type TokenStream<'a> = NgramTokenStream<'a>;
-    fn token_stream<'a>(&self, text: &'a str) -> NgramTokenStream<'a> {
-        NgramTokenStream {
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+        From::from(NgramTokenStream {
             ngram_charidx_iterator: StutteringIterator::new(
                 CodepointFrontiers::for_str(text),
                 self.min_gram,
@@ -142,7 +142,7 @@ impl Tokenizer for NgramTokenizer {
             prefix_only: self.prefix_only,
             text,
             token: Token::default(),
-        }
+        })
     }
 }
 
@@ -303,9 +303,9 @@ mod tests {
 
     use super::{utf8_codepoint_width, CodepointFrontiers, NgramTokenizer, StutteringIterator};
     use crate::tokenizer::tests::assert_token;
-    use crate::tokenizer::{Token, TokenStream, Tokenizer};
+    use crate::tokenizer::{BoxTokenStream, Token, Tokenizer};
 
-    fn test_helper<T: TokenStream>(mut tokenizer: T) -> Vec<Token> {
+    fn test_helper(mut tokenizer: BoxTokenStream) -> Vec<Token> {
         let mut tokens: Vec<Token> = vec![];
         tokenizer.process(&mut |token: &Token| tokens.push(token.clone()));
         tokens

src/tokenizer/raw_tokenizer.rs

@@ -1,4 +1,5 @@
 use super::{Token, TokenStream, Tokenizer};
+use crate::tokenizer::BoxTokenStream;
 
 /// For each value of the field, emit a single unprocessed token.
 #[derive(Clone)]
@@ -10,8 +11,7 @@ pub struct RawTokenStream {
 }
 
 impl Tokenizer for RawTokenizer {
-    type TokenStream<'a> = RawTokenStream;
-    fn token_stream(&self, text: &str) -> RawTokenStream {
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
         let token = Token {
             offset_from: 0,
             offset_to: text.len(),
@@ -23,6 +23,7 @@ impl Tokenizer for RawTokenizer {
             token,
             has_token: true,
         }
+        .into()
     }
 }
 

src/tokenizer/regex_tokenizer.rs

@@ -1,6 +1,6 @@
 use regex::Regex;
 
-use super::{Token, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenStream, Tokenizer};
 use crate::TantivyError;
 
 /// Tokenize the text by using a regex pattern to split.
@@ -60,14 +60,13 @@ impl RegexTokenizer {
 }
 
 impl Tokenizer for RegexTokenizer {
-    type TokenStream<'a> = RegexTokenStream<'a>;
-    fn token_stream<'a>(&self, text: &'a str) -> RegexTokenStream<'a> {
-        RegexTokenStream {
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(RegexTokenStream {
             regex: self.regex.clone(),
             text,
             token: Token::default(),
             cursor: 0,
-        }
+        })
     }
 }
 

src/tokenizer/remove_long.rs

@@ -2,9 +2,8 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = TextAnalyzer::builder(SimpleTokenizer)
-//!   .filter(RemoveLongFilter::limit(5))
-//!   .build();
+//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
+//!   .filter(RemoveLongFilter::limit(5));
 //!
 //! let mut stream = tokenizer.token_stream("toolong nice");
 //! // because `toolong` is more than 5 characters, it is filtered
@@ -12,7 +11,8 @@
 //! assert_eq!(stream.next().unwrap().text, "nice");
 //! assert!(stream.next().is_none());
 //! ```
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
 
 /// `RemoveLongFilter` removes tokens that are longer
 /// than a given number of bytes (in UTF-8 representation).
@@ -31,46 +31,27 @@ impl RemoveLongFilter {
     }
 }
 
-impl<T> RemoveLongFilterStream<T> {
+impl<'a> RemoveLongFilterStream<'a> {
     fn predicate(&self, token: &Token) -> bool {
         token.text.len() < self.token_length_limit
     }
 }
 
 impl TokenFilter for RemoveLongFilter {
-    type Tokenizer<T: Tokenizer> = RemoveLongFilterWrapper<T>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> RemoveLongFilterWrapper<T> {
-        RemoveLongFilterWrapper {
-            length_limit: self.length_limit,
-            inner: tokenizer,
-        }
-    }
-}
-
-#[derive(Clone)]
-pub struct RemoveLongFilterWrapper<T: Tokenizer> {
-    length_limit: usize,
-    inner: T,
-}
-
-impl<T: Tokenizer> Tokenizer for RemoveLongFilterWrapper<T> {
-    type TokenStream<'a> = RemoveLongFilterStream<T::TokenStream<'a>>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        RemoveLongFilterStream {
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(RemoveLongFilterStream {
             token_length_limit: self.length_limit,
-            tail: self.inner.token_stream(text),
-        }
+            tail: token_stream,
+        })
     }
 }
 
-pub struct RemoveLongFilterStream<T> {
+pub struct RemoveLongFilterStream<'a> {
     token_length_limit: usize,
-    tail: T,
+    tail: BoxTokenStream<'a>,
 }
 
-impl<T: TokenStream> TokenStream for RemoveLongFilterStream<T> {
+impl<'a> TokenStream for RemoveLongFilterStream<'a> {
     fn advance(&mut self) -> bool {
         while self.tail.advance() {
             if self.predicate(self.tail.token()) {
@@ -103,9 +84,7 @@ mod tests {
     }
 
     fn token_stream_helper(text: &str) -> Vec<Token> {
-        let a = TextAnalyzer::builder(SimpleTokenizer)
-            .filter(RemoveLongFilter::limit(6))
-            .build();
+        let a = TextAnalyzer::from(SimpleTokenizer).filter(RemoveLongFilter::limit(6));
         let mut token_stream = a.token_stream(text);
         let mut tokens: Vec<Token> = vec![];
         let mut add_token = |token: &Token| {

src/tokenizer/simple_tokenizer.rs

@@ -1,6 +1,6 @@
 use std::str::CharIndices;
 
-use super::{Token, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenStream, Tokenizer};
 
 /// Tokenize the text by splitting on whitespaces and punctuation.
 #[derive(Clone)]
@@ -13,13 +13,12 @@ pub struct SimpleTokenStream<'a> {
 }
 
 impl Tokenizer for SimpleTokenizer {
-    type TokenStream<'a> = SimpleTokenStream<'a>;
-    fn token_stream<'a>(&self, text: &'a str) -> SimpleTokenStream<'a> {
-        SimpleTokenStream {
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(SimpleTokenStream {
             text,
             chars: text.char_indices(),
             token: Token::default(),
-        }
+        })
     }
 }
 

src/tokenizer/split_compound_words.rs

@@ -2,7 +2,7 @@ use std::sync::Arc;
 
 use aho_corasick::{AhoCorasick, AhoCorasickBuilder, MatchKind, StateID};
 
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
 
 /// A [`TokenFilter`] which splits compound words into their parts
 /// based on a given dictionary.
@@ -23,11 +23,9 @@ use super::{Token, TokenFilter, TokenStream, Tokenizer};
 /// use tantivy::tokenizer::{SimpleTokenizer, SplitCompoundWords, TextAnalyzer};
 ///
 /// let tokenizer =
-///        TextAnalyzer::builder(SimpleTokenizer)
-///        .filter(SplitCompoundWords::from_dictionary([
+///        TextAnalyzer::from(SimpleTokenizer).filter(SplitCompoundWords::from_dictionary([
 ///            "dampf", "schiff", "fahrt", "brot", "backen", "automat",
-///        ]))
-///        .build();
+///        ]));
 ///
 /// let mut stream = tokenizer.token_stream("dampfschifffahrt");
 /// assert_eq!(stream.next().unwrap().text, "dampf");
@@ -78,45 +76,24 @@ impl<S: StateID> SplitCompoundWords<S> {
 }
 
 impl<S: StateID + Send + Sync + 'static> TokenFilter for SplitCompoundWords<S> {
-    type Tokenizer<T: Tokenizer> = SplitCompoundWordsFilter<T, S>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> SplitCompoundWordsFilter<T, S> {
-        SplitCompoundWordsFilter {
-            dict: self.dict,
-            inner: tokenizer,
-        }
-    }
-}
-
-#[derive(Clone)]
-pub struct SplitCompoundWordsFilter<T, S: StateID> {
-    dict: Arc<AhoCorasick<S>>,
-    inner: T,
-}
-
-impl<T: Tokenizer, S: StateID + Send + Sync + 'static> Tokenizer
-    for SplitCompoundWordsFilter<T, S>
-{
-    type TokenStream<'a> = SplitCompoundWordsTokenStream<T::TokenStream<'a>, S>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        SplitCompoundWordsTokenStream {
+    fn transform<'a>(&self, stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(SplitCompoundWordsTokenStream {
             dict: self.dict.clone(),
-            tail: self.inner.token_stream(text),
+            tail: stream,
             cuts: Vec::new(),
             parts: Vec::new(),
-        }
+        })
     }
 }
 
-pub struct SplitCompoundWordsTokenStream<T, S: StateID> {
+struct SplitCompoundWordsTokenStream<'a, S: StateID> {
     dict: Arc<AhoCorasick<S>>,
-    tail: T,
+    tail: BoxTokenStream<'a>,
     cuts: Vec<usize>,
     parts: Vec<Token>,
 }
 
-impl<T: TokenStream, S: StateID> SplitCompoundWordsTokenStream<T, S> {
+impl<'a, S: StateID> SplitCompoundWordsTokenStream<'a, S> {
     // Will use `self.cuts` to fill `self.parts` if `self.tail.token()`
     // can fully be split into consecutive matches against `self.dict`.
     fn split(&mut self) {
@@ -152,7 +129,7 @@ impl<T: TokenStream, S: StateID> SplitCompoundWordsTokenStream<T, S> {
     }
 }
 
-impl<T: TokenStream, S: StateID> TokenStream for SplitCompoundWordsTokenStream<T, S> {
+impl<'a, S: StateID> TokenStream for SplitCompoundWordsTokenStream<'a, S> {
     fn advance(&mut self) -> bool {
         self.parts.pop();
 
@@ -188,9 +165,8 @@ mod tests {
 
     #[test]
     fn splitting_compound_words_works() {
-        let tokenizer = TextAnalyzer::builder(SimpleTokenizer)
-            .filter(SplitCompoundWords::from_dictionary(["foo", "bar"]))
-            .build();
+        let tokenizer = TextAnalyzer::from(SimpleTokenizer)
+            .filter(SplitCompoundWords::from_dictionary(["foo", "bar"]));
 
         {
             let mut stream = tokenizer.token_stream("");

src/tokenizer/stemmer.rs

@@ -4,7 +4,8 @@ use std::mem;
 use rust_stemmers::{self, Algorithm};
 use serde::{Deserialize, Serialize};
 
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{Token, TokenFilter, TokenStream};
+use crate::tokenizer::BoxTokenStream;
 
 /// Available stemmer languages.
 #[derive(Debug, Serialize, Deserialize, Eq, PartialEq, Copy, Clone)]
@@ -81,42 +82,23 @@ impl Default for Stemmer {
 }
 
 impl TokenFilter for Stemmer {
-    type Tokenizer<T: Tokenizer> = StemmerFilter<T>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> StemmerFilter<T> {
-        StemmerFilter {
-            stemmer_algorithm: self.stemmer_algorithm,
-            inner: tokenizer,
-        }
-    }
-}
-
-#[derive(Clone)]
-pub struct StemmerFilter<T> {
-    stemmer_algorithm: Algorithm,
-    inner: T,
-}
-
-impl<T: Tokenizer> Tokenizer for StemmerFilter<T> {
-    type TokenStream<'a> = StemmerTokenStream<T::TokenStream<'a>>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        let stemmer = rust_stemmers::Stemmer::create(self.stemmer_algorithm);
-        StemmerTokenStream {
-            tail: self.inner.token_stream(text),
-            stemmer,
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        let inner_stemmer = rust_stemmers::Stemmer::create(self.stemmer_algorithm);
+        BoxTokenStream::from(StemmerTokenStream {
+            tail: token_stream,
+            stemmer: inner_stemmer,
             buffer: String::new(),
-        }
+        })
     }
 }
 
-pub struct StemmerTokenStream<T> {
-    tail: T,
+pub struct StemmerTokenStream<'a> {
+    tail: BoxTokenStream<'a>,
     stemmer: rust_stemmers::Stemmer,
     buffer: String,
 }
 
-impl<T: TokenStream> TokenStream for StemmerTokenStream<T> {
+impl<'a> TokenStream for StemmerTokenStream<'a> {
     fn advance(&mut self) -> bool {
         if !self.tail.advance() {
             return false;

src/tokenizer/stop_word_filter/mod.rs

@@ -2,9 +2,8 @@
 //! ```rust
 //! use tantivy::tokenizer::*;
 //!
-//! let tokenizer = TextAnalyzer::builder(SimpleTokenizer)
-//!   .filter(StopWordFilter::remove(vec!["the".to_string(), "is".to_string()]))
-//!   .build();
+//! let tokenizer = TextAnalyzer::from(SimpleTokenizer)
+//!   .filter(StopWordFilter::remove(vec!["the".to_string(), "is".to_string()]));
 //!
 //! let mut stream = tokenizer.token_stream("the fox is crafty");
 //! assert_eq!(stream.next().unwrap().text, "fox");
@@ -21,7 +20,7 @@ use rustc_hash::FxHashSet;
 
 #[cfg(feature = "stopwords")]
 use super::Language;
-use super::{Token, TokenFilter, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenFilter, TokenStream};
 
 /// `TokenFilter` that removes stop words from a token stream
 #[derive(Clone)]
@@ -70,46 +69,27 @@ impl StopWordFilter {
     }
 }
 
+pub struct StopWordFilterStream<'a> {
+    words: Arc<FxHashSet<String>>,
+    tail: BoxTokenStream<'a>,
+}
+
 impl TokenFilter for StopWordFilter {
-    type Tokenizer<T: Tokenizer> = StopWordFilterWrapper<T>;
-
-    fn transform<T: Tokenizer>(self, tokenizer: T) -> StopWordFilterWrapper<T> {
-        StopWordFilterWrapper {
-            words: self.words,
-            inner: tokenizer,
-        }
-    }
-}
-
-#[derive(Clone)]
-pub struct StopWordFilterWrapper<T> {
-    words: Arc<FxHashSet<String>>,
-    inner: T,
-}
-
-impl<T: Tokenizer> Tokenizer for StopWordFilterWrapper<T> {
-    type TokenStream<'a> = StopWordFilterStream<T::TokenStream<'a>>;
-
-    fn token_stream<'a>(&self, text: &'a str) -> Self::TokenStream<'a> {
-        StopWordFilterStream {
+    fn transform<'a>(&self, token_stream: BoxTokenStream<'a>) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(StopWordFilterStream {
             words: self.words.clone(),
-            tail: self.inner.token_stream(text),
-        }
+            tail: token_stream,
+        })
     }
 }
 
-pub struct StopWordFilterStream<T> {
-    words: Arc<FxHashSet<String>>,
-    tail: T,
-}
-
-impl<T> StopWordFilterStream<T> {
+impl<'a> StopWordFilterStream<'a> {
     fn predicate(&self, token: &Token) -> bool {
         !self.words.contains(&token.text)
     }
 }
 
-impl<T: TokenStream> TokenStream for StopWordFilterStream<T> {
+impl<'a> TokenStream for StopWordFilterStream<'a> {
     fn advance(&mut self) -> bool {
         while self.tail.advance() {
             if self.predicate(self.tail.token()) {
@@ -151,9 +131,7 @@ mod tests {
             "am".to_string(),
             "i".to_string(),
         ];
-        let a = TextAnalyzer::builder(SimpleTokenizer)
-            .filter(StopWordFilter::remove(stops))
-            .build();
+        let a = TextAnalyzer::from(SimpleTokenizer).filter(StopWordFilter::remove(stops));
         let mut token_stream = a.token_stream(text);
         let mut tokens: Vec<Token> = vec![];
         let mut add_token = |token: &Token| {

src/tokenizer/tokenizer.rs

@@ -1,12 +1,15 @@
 /// The tokenizer module contains all of the tools used to process
 /// text in `tantivy`.
-use tokenizer_api::{BoxTokenStream, BoxableTokenizer, TokenFilter, Tokenizer};
+use tokenizer_api::{BoxTokenFilter, BoxTokenStream, Tokenizer};
 
 use crate::tokenizer::empty_tokenizer::EmptyTokenizer;
 
 /// `TextAnalyzer` tokenizes an input text into tokens and modifies the resulting `TokenStream`.
+///
+/// It simply wraps a `Tokenizer` and a list of `TokenFilter` that are applied sequentially.
 pub struct TextAnalyzer {
-    tokenizer: Box<dyn BoxableTokenizer>,
+    tokenizer: Box<dyn Tokenizer>,
+    token_filters: Vec<BoxTokenFilter>,
 }
 
 impl Default for TextAnalyzer {
@@ -15,21 +18,52 @@ impl Default for TextAnalyzer {
     }
 }
 
-impl<T: Tokenizer + Clone> From<T> for TextAnalyzer {
+impl<T: Tokenizer> From<T> for TextAnalyzer {
     fn from(tokenizer: T) -> Self {
-        TextAnalyzer::builder(tokenizer).build()
+        TextAnalyzer::new(tokenizer, Vec::new())
     }
 }
 
 impl TextAnalyzer {
-    /// Create a new TextAnalyzerBuilder
-    pub fn builder<T: Tokenizer>(tokenizer: T) -> TextAnalyzerBuilder<T> {
-        TextAnalyzerBuilder { tokenizer }
+    /// Creates a new `TextAnalyzer` given a tokenizer and a vector of `BoxTokenFilter`.
+    ///
+    /// When creating a `TextAnalyzer` from a `Tokenizer` alone, prefer using
+    /// `TextAnalyzer::from(tokenizer)`.
+    pub fn new<T: Tokenizer>(tokenizer: T, token_filters: Vec<BoxTokenFilter>) -> TextAnalyzer {
+        TextAnalyzer {
+            tokenizer: Box::new(tokenizer),
+            token_filters,
+        }
+    }
+
+    /// Appends a token filter to the current tokenizer.
+    ///
+    /// The method consumes the current `TokenStream` and returns a
+    /// new one.
+    ///
+    /// # Example
+    ///
+    /// ```rust
+    /// use tantivy::tokenizer::*;
+    ///
+    /// let en_stem = TextAnalyzer::from(SimpleTokenizer)
+    ///     .filter(RemoveLongFilter::limit(40))
+    ///     .filter(LowerCaser)
+    ///     .filter(Stemmer::default());
+    /// ```
+    #[must_use]
+    pub fn filter<F: Into<BoxTokenFilter>>(mut self, token_filter: F) -> Self {
+        self.token_filters.push(token_filter.into());
+        self
     }
 
     /// Creates a token stream for a given `str`.
     pub fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
-        self.tokenizer.box_token_stream(text)
+        let mut token_stream = self.tokenizer.token_stream(text);
+        for token_filter in &self.token_filters {
+            token_stream = token_filter.transform(token_stream);
+        }
+        token_stream
     }
 }
 
@@ -37,39 +71,11 @@ impl Clone for TextAnalyzer {
     fn clone(&self) -> Self {
         TextAnalyzer {
             tokenizer: self.tokenizer.box_clone(),
-        }
-    }
-}
-
-/// Builder helper for [`TextAnalyzer`]
-pub struct TextAnalyzerBuilder<T> {
-    tokenizer: T,
-}
-
-impl<T: Tokenizer> TextAnalyzerBuilder<T> {
-    /// Appends a token filter to the current builder.
-    ///
-    /// # Example
-    ///
-    /// ```rust
-    /// use tantivy::tokenizer::*;
-    ///
-    /// let en_stem = TextAnalyzer::builder(SimpleTokenizer)
-    ///     .filter(RemoveLongFilter::limit(40))
-    ///     .filter(LowerCaser)
-    ///     .filter(Stemmer::default())
-    ///     .build();
-    /// ```
-    pub fn filter<F: TokenFilter>(self, token_filter: F) -> TextAnalyzerBuilder<F::Tokenizer<T>> {
-        TextAnalyzerBuilder {
-            tokenizer: token_filter.transform(self.tokenizer),
-        }
-    }
-
-    /// Finalize building the TextAnalyzer
-    pub fn build(self) -> TextAnalyzer {
-        TextAnalyzer {
-            tokenizer: Box::new(self.tokenizer),
+            token_filters: self
+                .token_filters
+                .iter()
+                .map(|token_filter| token_filter.box_clone())
+                .collect(),
         }
     }
 }

src/tokenizer/tokenizer_manager.rs

@@ -61,18 +61,16 @@ impl Default for TokenizerManager {
         manager.register("raw", RawTokenizer);
         manager.register(
             "default",
-            TextAnalyzer::builder(SimpleTokenizer)
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
-                .filter(LowerCaser)
-                .build(),
+                .filter(LowerCaser),
         );
         manager.register(
             "en_stem",
-            TextAnalyzer::builder(SimpleTokenizer)
+            TextAnalyzer::from(SimpleTokenizer)
                 .filter(RemoveLongFilter::limit(40))
                 .filter(LowerCaser)
-                .filter(Stemmer::new(Language::English))
-                .build(),
+                .filter(Stemmer::new(Language::English)),
         );
         manager.register("whitespace", WhitespaceTokenizer);
         manager

src/tokenizer/whitespace_tokenizer.rs

@@ -1,6 +1,6 @@
 use std::str::CharIndices;
 
-use super::{Token, TokenStream, Tokenizer};
+use super::{BoxTokenStream, Token, TokenStream, Tokenizer};
 
 /// Tokenize the text by splitting on whitespaces.
 #[derive(Clone)]
@@ -13,13 +13,12 @@ pub struct WhitespaceTokenStream<'a> {
 }
 
 impl Tokenizer for WhitespaceTokenizer {
-    type TokenStream<'a> = WhitespaceTokenStream<'a>;
-    fn token_stream<'a>(&self, text: &'a str) -> WhitespaceTokenStream<'a> {
-        WhitespaceTokenStream {
+    fn token_stream<'a>(&self, text: &'a str) -> BoxTokenStream<'a> {
+        BoxTokenStream::from(WhitespaceTokenStream {
             text,
             chars: text.char_indices(),
             token: Token::default(),
-        }
+        })
     }
 }
 

sstable/Cargo.toml

@@ -6,6 +6,8 @@ license = "MIT"
 
 [dependencies]
 common = {path="../common", package="tantivy-common"}
+ciborium = "0.2"
+serde = "1"
 tantivy-fst = "0.4"
 
 [dev-dependencies]

sstable/README.md

@@ -26,95 +26,3 @@ possible.
 - it allows incremental encoding of the keys
 - the front compression is leveraged to optimize
 the intersection with an automaton
-
-# On disk format
-
-Overview of the SSTable format. Unless noted otherwise, numbers are little-endian.
-
-### SSTable
-```
-+-------+-------+-----+--------+
-| Block | Block | ... | Footer |
-+-------+-------+-----+--------+
-|----( # of blocks)---|
-```
-- Block(`SSTBlock`): list of independent block, terminated by a single empty block.
-- Footer(`SSTFooter`)
-
-### SSTBlock
-```
-+----------+--------+-------+-------+-----+
-| BlockLen | Values | Delta | Delta | ... |
-+----------+--------+-------+-------+-----+
-                    |----( # of deltas)---|
-```
-- BlockLen(u32): length of the block
-- Values: an application defined format storing a sequence of value, capable of determining it own length
-- Delta
-
-### Delta
-```
-+---------+--------+
-| KeepAdd | Suffix |
-+---------+--------+
-```
-- KeepAdd
-- Suffix: KeepAdd.add bytes of key suffix
-
-### KeepAdd
-KeepAdd can be represented in two different representation, a very compact 1byte one which is enough for most usage, and a longer variable-len one when required
-
-When keep < 16 and add < 16
-```
-+-----+------+
-| Add | Keep |
-+-----+------+
-```
-- Add(u4): number of bytes to push
-- Keep(u4): number of bytes to pop
-
-Otherwise:
-```
-+------+------+-----+
-| 0x01 | Keep | Add |
-+------+------+-----+
-```
-- Add(VInt): number of bytes to push
-- Keep(VInt): number of bytes to pop
-
-
-Note: there is no ambiguity between both representation as Add is always guarantee to be non-zero, except for the very first key of an SSTable, where Keep is guaranteed to be zero.
-
-### SSTFooter
-```
-+-------+-------+-----+-------------+---------+---------+------+
-| Block | Block | ... | IndexOffset | NumTerm | Version | Type |
-+-------+-------+-----+-------------+---------+---------+------+
-|----( # of blocks)---|
-```
-- Block(SSTBlock): uses IndexValue for its Values format
-- IndexOffset(u64): Offset to the start of the SSTFooter
-- NumTerm(u64): number of terms in the sstable
-- Version(u32): Currently defined to 0x00\_00\_00\_01
-- Type(u32): Defined to 0x00\_00\_00\_02
-
-### IndexValue
-```
-+------------+----------+-------+-------+-----+
-| EntryCount | StartPos | Entry | Entry | ... |
-+------------+----------+-------+-------+-----+
-                        |---( # of entries)---|
-```
-
-- EntryCount(VInt): number of entries
-- StartPos(VInt): the start pos of the first (data) block referenced by this (index) block
-- Entry (IndexEntry)
-
-### Entry
-```
-+----------+--------------+
-| BlockLen | FirstOrdinal |
-+----------+--------------+
-```
-- BlockLen(VInt): length of the block
-- FirstOrdinal(VInt): ordinal of the first element in the given block

sstable/src/delta.rs

@@ -18,7 +18,6 @@ where W: io::Write
     value_writer: TValueWriter,
     // Only here to avoid allocations.
     stateless_buffer: Vec<u8>,
-    block_len: usize,
 }
 
 impl<W, TValueWriter> DeltaWriter<W, TValueWriter>
@@ -32,14 +31,15 @@ where
             write: CountingWriter::wrap(BufWriter::new(wrt)),
             value_writer: TValueWriter::default(),
             stateless_buffer: Vec::new(),
-            block_len: BLOCK_LEN,
         }
     }
+}
 
-    pub fn set_block_len(&mut self, block_len: usize) {
-        self.block_len = block_len
-    }
-
+impl<W, TValueWriter> DeltaWriter<W, TValueWriter>
+where
+    W: io::Write,
+    TValueWriter: value::ValueWriter,
+{
     pub fn flush_block(&mut self) -> io::Result<Option<Range<usize>>> {
         if self.block.is_empty() {
             return Ok(None);
@@ -82,7 +82,7 @@ where
     }
 
     pub fn flush_block_if_required(&mut self) -> io::Result<Option<Range<usize>>> {
-        if self.block.len() > self.block_len {
+        if self.block.len() > BLOCK_LEN {
             return self.flush_block();
         }
         Ok(None)