Added Empty ColumnIndex (#1910)

columnar/src/column/mod.rs

@@ -45,6 +45,7 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
 
     pub fn num_docs(&self) -> RowId {
         match &self.idx {
+            ColumnIndex::Empty { num_docs } => *num_docs,
             ColumnIndex::Full => self.values.num_vals(),
             ColumnIndex::Optional(optional_index) => optional_index.num_docs(),
             ColumnIndex::Multivalued(col_index) => {
@@ -78,17 +79,17 @@ impl<T: PartialOrd + Copy + Debug + Send + Sync + 'static> Column<T> {
         &self,
         value_range: RangeInclusive<T>,
         selected_docid_range: Range<u32>,
-        docids: &mut Vec<u32>,
+        doc_ids: &mut Vec<u32>,
     ) {
         // convert passed docid range to row id range
         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, docids);
+            .get_row_ids_for_value_range(value_range, rowid_range, doc_ids);
         // Convert rows to docids
         self.idx
-            .select_batch_in_place(docids, selected_docid_range.start);
+            .select_batch_in_place(selected_docid_range.start, doc_ids);
     }
 
     /// Fils the output vector with the (possibly multiple values that are associated_with
@@ -151,6 +152,7 @@ impl<T: PartialOrd + Debug + Send + Sync + Copy + 'static> ColumnValues<T>
 
     fn num_vals(&self) -> u32 {
         match &self.column.idx {
+            ColumnIndex::Empty { .. } => 0u32,
             ColumnIndex::Full => self.column.values.num_vals(),
             ColumnIndex::Optional(optional_idx) => optional_idx.num_docs(),
             ColumnIndex::Multivalued(multivalue_idx) => multivalue_idx.num_docs(),

columnar/src/column_index/merge/shuffled.rs

@@ -91,6 +91,7 @@ fn iter_num_values<'a>(
                     return 0u32;
                 };
         match column_index {
+            ColumnIndex::Empty { .. } => 0u32,
             ColumnIndex::Full => 1,
             ColumnIndex::Optional(optional_index) => {
                 u32::from(optional_index.contains(row_addr.row_id))

columnar/src/column_index/merge/stacked.rs

@@ -55,7 +55,7 @@ impl<'a> Iterable<RowId> for StackedOptionalIndex<'a> {
                         Some(ColumnIndex::Multivalued(_)) => {
                             panic!("No multivalued index is allowed when stacking column index");
                         }
-                        None => Box::new(std::iter::empty()),
+                        None | Some(ColumnIndex::Empty { .. }) => Box::new(std::iter::empty()),
                     };
                     rows_it
                 }),
@@ -74,7 +74,9 @@ fn convert_column_opt_to_multivalued_index<'a>(
     num_rows: RowId,
 ) -> Box<dyn Iterator<Item = RowId> + 'a> {
     match column_index_opt {
-        None => Box::new(iter::repeat(0u32).take(num_rows as usize + 1)),
+        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(

columnar/src/column_index/mod.rs

@@ -14,6 +14,9 @@ use crate::{Cardinality, DocId, RowId};
 
 #[derive(Clone)]
 pub enum ColumnIndex {
+    Empty {
+        num_docs: u32,
+    },
     Full,
     Optional(OptionalIndex),
     /// In addition, at index num_rows, an extra value is added
@@ -37,6 +40,7 @@ impl ColumnIndex {
     #[inline]
     pub fn get_cardinality(&self) -> Cardinality {
         match self {
+            ColumnIndex::Empty { .. } => Cardinality::Optional,
             ColumnIndex::Full => Cardinality::Full,
             ColumnIndex::Optional(_) => Cardinality::Optional,
             ColumnIndex::Multivalued(_) => Cardinality::Multivalued,
@@ -46,6 +50,7 @@ impl ColumnIndex {
     /// Returns true if and only if there are at least one value associated to the row.
     pub fn has_value(&self, doc_id: DocId) -> bool {
         match self {
+            ColumnIndex::Empty { .. } => false,
             ColumnIndex::Full => true,
             ColumnIndex::Optional(optional_index) => optional_index.contains(doc_id),
             ColumnIndex::Multivalued(multivalued_index) => {
@@ -56,6 +61,7 @@ impl ColumnIndex {
 
     pub fn value_row_ids(&self, doc_id: DocId) -> Range<RowId> {
         match self {
+            ColumnIndex::Empty { .. } => 0..0,
             ColumnIndex::Full => doc_id..doc_id + 1,
             ColumnIndex::Optional(optional_index) => {
                 if let Some(val) = optional_index.rank_if_exists(doc_id) {
@@ -70,6 +76,7 @@ impl ColumnIndex {
 
     pub fn docid_range_to_rowids(&self, doc_id: Range<DocId>) -> Range<RowId> {
         match self {
+            ColumnIndex::Empty { .. } => 0..0,
             ColumnIndex::Full => doc_id,
             ColumnIndex::Optional(optional_index) => {
                 let row_start = optional_index.rank(doc_id.start);
@@ -88,8 +95,9 @@ impl ColumnIndex {
         }
     }
 
-    pub fn select_batch_in_place(&self, rank_ids: &mut Vec<RowId>, doc_id_start: DocId) {
+    pub fn select_batch_in_place(&self, doc_id_start: DocId, rank_ids: &mut Vec<RowId>) {
         match self {
+            ColumnIndex::Empty { .. } => {}
             ColumnIndex::Full => {
                 // No need to do anything:
                 // value_idx and row_idx are the same.