Revert "remove unused columnar api (#2742)" (#2748)

* Revert "remove unused columnar api (#2742)"

This reverts commit 8725594d47.

* Clippy comment + removing fill_vals

---------

Co-authored-by: Paul Masurel <paul.masurel@datadoghq.com>

columnar/benches/bench_access.rs

@@ -43,5 +43,26 @@ fn bench_group(mut runner: InputGroup<Column>) {
         }
         black_box(sum);
     });
+    runner.register("access_first_vals", |column| {
+        let mut sum = 0;
+        const BLOCK_SIZE: usize = 32;
+        let mut docs = vec![0; BLOCK_SIZE];
+        let mut buffer = vec![None; BLOCK_SIZE];
+        for i in (0..NUM_DOCS).step_by(BLOCK_SIZE) {
+            // fill docs
+            #[allow(clippy::needless_range_loop)]
+            for idx in 0..BLOCK_SIZE {
+                docs[idx] = idx as u32 + i;
+            }
+
+            column.first_vals(&docs, &mut buffer);
+            for val in buffer.iter() {
+                let Some(val) = val else { continue };
+                sum += *val;
+            }
+        }
+
+        black_box(sum);
+    });
     runner.run();
 }

columnar/src/column/mod.rs

@@ -14,7 +14,7 @@ pub use serialize::{
     serialize_column_mappable_to_u128,
 };
 
-use crate::column_index::ColumnIndex;
+use crate::column_index::{ColumnIndex, Set};
 use crate::column_values::monotonic_mapping::StrictlyMonotonicMappingToInternal;
 use crate::column_values::{ColumnValues, monotonic_map_column};
 use crate::{Cardinality, DocId, EmptyColumnValues, MonotonicallyMappableToU64, RowId};
@@ -89,6 +89,31 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
         self.values_for_doc(row_id).next()
     }
 
+    /// Load the first value for each docid in the provided slice.
+    #[inline]
+    pub fn first_vals(&self, docids: &[DocId], output: &mut [Option<T>]) {
+        match &self.index {
+            ColumnIndex::Empty { .. } => {}
+            ColumnIndex::Full => self.values.get_vals_opt(docids, output),
+            ColumnIndex::Optional(optional_index) => {
+                for (i, docid) in docids.iter().enumerate() {
+                    output[i] = optional_index
+                        .rank_if_exists(*docid)
+                        .map(|rowid| self.values.get_val(rowid));
+                }
+            }
+            ColumnIndex::Multivalued(multivalued_index) => {
+                for (i, docid) in docids.iter().enumerate() {
+                    let range = multivalued_index.range(*docid);
+                    let is_empty = range.start == range.end;
+                    if !is_empty {
+                        output[i] = Some(self.values.get_val(range.start));
+                    }
+                }
+            }
+        }
+    }
+
     /// Translates a block of docids to row_ids.
     ///
     /// returns the row_ids and the matching docids on the same index

src/fastfield/mod.rs

@@ -726,24 +726,22 @@ mod tests {
             .column_opt::<DateTime>("multi_date")
             .unwrap()
             .unwrap();
+
         {
-            let mut dates = Vec::new();
             assert_eq!(date_fast_field.get_val(0).into_timestamp_nanos(), 1i64);
-            dates.extend(dates_fast_field.values_for_doc(0));
+            let dates: Vec<DateTime> = dates_fast_field.values_for_doc(0u32).collect();
             assert_eq!(dates.len(), 2);
             assert_eq!(dates[0].into_timestamp_nanos(), 2i64);
             assert_eq!(dates[1].into_timestamp_nanos(), 3i64);
         }
         {
-            let mut dates = Vec::new();
             assert_eq!(date_fast_field.get_val(1).into_timestamp_nanos(), 4i64);
-            dates.extend(dates_fast_field.values_for_doc(1));
+            let dates: Vec<DateTime> = dates_fast_field.values_for_doc(1u32).collect();
             assert!(dates.is_empty());
         }
         {
-            let mut dates = Vec::new();
             assert_eq!(date_fast_field.get_val(2).into_timestamp_nanos(), 0i64);
-            dates.extend(dates_fast_field.values_for_doc(2));
+            let dates: Vec<DateTime> = dates_fast_field.values_for_doc(2u32).collect();
             assert_eq!(dates.len(), 2);
             assert_eq!(dates[0].into_timestamp_nanos(), 5i64);
             assert_eq!(dates[1].into_timestamp_nanos(), 6i64);

src/query/range_query/range_query.rs

@@ -268,9 +268,7 @@ mod tests {
     use crate::indexer::NoMergePolicy;
     use crate::query::range_query::fast_field_range_doc_set::RangeDocSet;
     use crate::query::range_query::range_query::InvertedIndexRangeQuery;
-    use crate::query::{
-        AllScorer, BitSetDocSet, ConstScorer, EmptyScorer, EnableScoring, Query, QueryParser,
-    };
+    use crate::query::{AllScorer, ConstScorer, EmptyScorer, EnableScoring, Query, QueryParser};
     use crate::schema::{
         Field, IntoIpv6Addr, Schema, TantivyDocument, FAST, INDEXED, STORED, TEXT,
     };

src/query/term_query/term_weight.rs

@@ -18,7 +18,7 @@ pub struct TermWeight {
 }
 
 enum TermOrEmptyOrAllScorer {
-    TermScorer(TermScorer),
+    TermScorer(Box<TermScorer>),
     Empty,
     AllMatch(AllScorer),
 }
@@ -26,7 +26,7 @@ enum TermOrEmptyOrAllScorer {
 impl TermOrEmptyOrAllScorer {
     pub fn into_boxed_scorer(self) -> Box<dyn Scorer> {
         match self {
-            TermOrEmptyOrAllScorer::TermScorer(scorer) => Box::new(scorer),
+            TermOrEmptyOrAllScorer::TermScorer(scorer) => scorer,
             TermOrEmptyOrAllScorer::Empty => Box::new(EmptyScorer),
             TermOrEmptyOrAllScorer::AllMatch(scorer) => Box::new(scorer),
         }
@@ -48,9 +48,7 @@ impl Weight for TermWeight {
                 explanation.add_context(format!("Term={:?}", self.term,));
                 Ok(explanation)
             }
-            TermOrEmptyOrAllScorer::Empty => {
-                return Err(does_not_match(doc));
-            }
+            TermOrEmptyOrAllScorer::Empty => Err(does_not_match(doc)),
             TermOrEmptyOrAllScorer::AllMatch(_) => AllWeight.explain(reader, doc),
         }
     }
@@ -75,7 +73,7 @@ impl Weight for TermWeight {
     ) -> crate::Result<()> {
         match self.specialized_scorer(reader, 1.0)? {
             TermOrEmptyOrAllScorer::TermScorer(mut term_scorer) => {
-                for_each_scorer(&mut term_scorer, callback);
+                for_each_scorer(&mut *term_scorer, callback);
             }
             TermOrEmptyOrAllScorer::Empty => {}
             TermOrEmptyOrAllScorer::AllMatch(mut all_scorer) => {
@@ -127,7 +125,7 @@ impl Weight for TermWeight {
         match specialized_scorer {
             TermOrEmptyOrAllScorer::TermScorer(term_scorer) => {
                 crate::query::boolean_query::block_wand_single_scorer(
-                    term_scorer,
+                    *term_scorer,
                     threshold,
                     callback,
                 );
@@ -173,7 +171,7 @@ impl TermWeight {
     ) -> crate::Result<Option<TermScorer>> {
         let scorer = self.specialized_scorer(reader, boost)?;
         Ok(match scorer {
-            TermOrEmptyOrAllScorer::TermScorer(scorer) => Some(scorer),
+            TermOrEmptyOrAllScorer::TermScorer(scorer) => Some(*scorer),
             _ => None,
         })
     }
@@ -203,10 +201,8 @@ impl TermWeight {
 
         let fieldnorm_reader = self.fieldnorm_reader(reader)?;
         let similarity_weight = self.similarity_weight.boost_by(boost);
-        Ok(TermOrEmptyOrAllScorer::TermScorer(TermScorer::new(
-            segment_postings,
-            fieldnorm_reader,
-            similarity_weight,
+        Ok(TermOrEmptyOrAllScorer::TermScorer(Box::new(
+            TermScorer::new(segment_postings, fieldnorm_reader, similarity_weight),
         )))
     }