Use the cached postings lists in the query system

meilisearch-core/src/bucket_sort.rs

@@ -41,6 +41,7 @@ pub fn bucket_sort<'c, FI>(
     documents_fields_counts_store: store::DocumentsFieldsCounts,
     synonyms_store: store::Synonyms,
     prefix_documents_cache_store: store::PrefixDocumentsCache,
+    prefix_postings_lists_cache_store: store::PrefixPostingsListsCache,
 ) -> MResult<Vec<Document>>
 where
     FI: Fn(DocumentId) -> bool,
@@ -64,6 +65,7 @@ where
             documents_fields_counts_store,
             synonyms_store,
             prefix_documents_cache_store,
+            prefix_postings_lists_cache_store,
         );
     }
 
@@ -96,7 +98,14 @@ where
     let before_postings_lists_fetching = Instant::now();
     mk_arena!(arena);
     let mut bare_matches =
-        fetch_matches(reader, &automatons, &mut arena, main_store, postings_lists_store)?;
+        fetch_matches(
+            reader,
+            &automatons,
+            &mut arena,
+            main_store,
+            postings_lists_store,
+            prefix_postings_lists_cache_store,
+        )?;
     debug!("bare matches ({}) retrieved in {:.02?}",
         bare_matches.len(),
         before_postings_lists_fetching.elapsed(),
@@ -203,6 +212,7 @@ pub fn bucket_sort_with_distinct<'c, FI, FD>(
     documents_fields_counts_store: store::DocumentsFieldsCounts,
     synonyms_store: store::Synonyms,
     prefix_documents_cache_store: store::PrefixDocumentsCache,
+    prefix_postings_lists_cache_store: store::PrefixPostingsListsCache,
 ) -> MResult<Vec<Document>>
 where
     FI: Fn(DocumentId) -> bool,
@@ -213,7 +223,14 @@ where
 
     let before_postings_lists_fetching = Instant::now();
     mk_arena!(arena);
-    let mut bare_matches = fetch_matches(reader, &automatons, &mut arena, main_store, postings_lists_store)?;
+    let mut bare_matches = fetch_matches(
+        reader,
+        &automatons,
+        &mut arena,
+        main_store,
+        postings_lists_store,
+        prefix_postings_lists_cache_store,
+    )?;
     debug!("bare matches ({}) retrieved in {:.02?}",
         bare_matches.len(),
         before_postings_lists_fetching.elapsed(),
@@ -486,6 +503,7 @@ fn fetch_matches<'txn, 'tag>(
     arena: &mut SmallArena<'tag, PostingsListView<'txn>>,
     main_store: store::Main,
     postings_lists_store: store::PostingsLists,
+    pplc_store: store::PrefixPostingsListsCache,
 ) -> MResult<Vec<BareMatch<'tag>>>
 {
     let before_words_fst = Instant::now();
@@ -504,10 +522,7 @@ fn fetch_matches<'txn, 'tag>(
     let automatons_loop = Instant::now();
 
     for (query_index, automaton) in automatons.iter().enumerate() {
-        let before_dfa = Instant::now();
+        let QueryWordAutomaton { query, is_exact, is_prefix, .. } = automaton;
-        let dfa = automaton.dfa();
-        let QueryWordAutomaton { query, is_exact, .. } = automaton;
-        dfa_time += before_dfa.elapsed();
 
         let before_word_postings_lists_fetching = Instant::now();
         let mut stream_next_time = Duration::default();
@@ -515,6 +530,52 @@ fn fetch_matches<'txn, 'tag>(
         let mut postings_lists_original_length = 0;
         let mut postings_lists_length = 0;
 
+        if *is_prefix && query.len() == 1 {
+            let prefix = [query.as_bytes()[0], 0, 0, 0];
+
+            number_of_words += 1;
+
+            let before_postings_lists_fetching = Instant::now();
+            if let Some(postings_list) = pplc_store.prefix_postings_list(reader, prefix)? {
+                debug!("Found cached postings list for {:?}", query);
+                postings_lists_original_length += postings_list.len();
+
+                let input = Rc::from(&prefix[..]);
+                let postings_list = Rc::new(postings_list);
+                let postings_list_view = PostingsListView::original(input, postings_list);
+
+                let mut offset = 0;
+                for group in postings_list_view.linear_group_by_key(|di| di.document_id) {
+                    let document_id = group[0].document_id;
+
+                    if query_index != 0 {
+                        if !documents_ids.contains(&document_id) { continue }
+                    } else {
+                        documents_ids.insert(document_id);
+                    }
+
+                    postings_lists_length += group.len();
+
+                    let posting_list_index = arena.add(postings_list_view.range(offset, group.len()));
+                    let bare_match = BareMatch {
+                        document_id,
+                        query_index: query_index as u16,
+                        distance: 0,
+                        is_exact: *is_exact,
+                        postings_list: posting_list_index,
+                    };
+
+                    total_postings_lists.push(bare_match);
+                    offset += group.len();
+                }
+            }
+            postings_lists_fetching_time += before_postings_lists_fetching.elapsed();
+        }
+        else {
+            let before_dfa = Instant::now();
+            let dfa = automaton.dfa();
+            dfa_time += before_dfa.elapsed();
+
             let byte = query.as_bytes()[0];
             let mut stream = if byte == u8::max_value() {
                 words.search(&dfa).ge(&[byte]).into_stream()
@@ -550,8 +611,11 @@ fn fetch_matches<'txn, 'tag>(
                     for group in postings_list_view.linear_group_by_key(|di| di.document_id) {
                         let document_id = group[0].document_id;
 
-                    if query_index != 0 && !documents_ids.contains(&document_id) { continue }
+                        if query_index != 0 {
+                            if !documents_ids.contains(&document_id) { continue }
+                        } else {
                             documents_ids.insert(document_id);
+                        }
 
                         postings_lists_length += group.len();
 
@@ -570,6 +634,7 @@ fn fetch_matches<'txn, 'tag>(
                 }
                 postings_lists_fetching_time += before_postings_lists_fetching.elapsed();
             }
+        }
 
         debug!("{:?} gives {} words", query, number_of_words);
         debug!("{:?} gives postings lists of length {} (original was {})",

meilisearch-core/src/query_builder.rs

@@ -16,7 +16,8 @@ pub struct QueryBuilder<'c, 'f, 'd> {
     postings_lists_store: store::PostingsLists,
     documents_fields_counts_store: store::DocumentsFieldsCounts,
     synonyms_store: store::Synonyms,
-    prefix_cache_store: store::PrefixDocumentsCache,
+    prefix_documents_cache_store: store::PrefixDocumentsCache,
+    prefix_postings_lists_cache_store: store::PrefixPostingsListsCache,
 }
 
 impl<'c, 'f, 'd> QueryBuilder<'c, 'f, 'd> {
@@ -25,14 +26,16 @@ impl<'c, 'f, 'd> QueryBuilder<'c, 'f, 'd> {
         postings_lists: store::PostingsLists,
         documents_fields_counts: store::DocumentsFieldsCounts,
         synonyms: store::Synonyms,
-        prefix_cache: store::PrefixDocumentsCache,
+        prefix_documents_cache: store::PrefixDocumentsCache,
+        prefix_postings_lists_cache: store::PrefixPostingsListsCache,
     ) -> QueryBuilder<'c, 'f, 'd> {
         QueryBuilder::with_criteria(
             main,
             postings_lists,
             documents_fields_counts,
             synonyms,
-            prefix_cache,
+            prefix_documents_cache,
+            prefix_postings_lists_cache,
             Criteria::default(),
         )
     }
@@ -42,7 +45,8 @@ impl<'c, 'f, 'd> QueryBuilder<'c, 'f, 'd> {
         postings_lists: store::PostingsLists,
         documents_fields_counts: store::DocumentsFieldsCounts,
         synonyms: store::Synonyms,
-        prefix_cache: store::PrefixDocumentsCache,
+        prefix_documents_cache: store::PrefixDocumentsCache,
+        prefix_postings_lists_cache: store::PrefixPostingsListsCache,
         criteria: Criteria<'c>,
     ) -> QueryBuilder<'c, 'f, 'd> {
         QueryBuilder {
@@ -55,7 +59,8 @@ impl<'c, 'f, 'd> QueryBuilder<'c, 'f, 'd> {
             postings_lists_store: postings_lists,
             documents_fields_counts_store: documents_fields_counts,
             synonyms_store: synonyms,
-            prefix_cache_store: prefix_cache,
+            prefix_documents_cache_store: prefix_documents_cache,
+            prefix_postings_lists_cache_store: prefix_postings_lists_cache,
         }
     }
 
@@ -102,7 +107,8 @@ impl<'c, 'f, 'd> QueryBuilder<'c, 'f, 'd> {
                 self.postings_lists_store,
                 self.documents_fields_counts_store,
                 self.synonyms_store,
-                self.prefix_cache_store,
+                self.prefix_documents_cache_store,
+                self.prefix_postings_lists_cache_store,
             ),
             None => bucket_sort(
                 reader,
@@ -115,7 +121,8 @@ impl<'c, 'f, 'd> QueryBuilder<'c, 'f, 'd> {
                 self.postings_lists_store,
                 self.documents_fields_counts_store,
                 self.synonyms_store,
-                self.prefix_cache_store,
+                self.prefix_documents_cache_store,
+                self.prefix_postings_lists_cache_store,
             ),
         }
     }

meilisearch-core/src/store/mod.rs

@@ -267,6 +267,7 @@ impl Index {
             self.documents_fields_counts,
             self.synonyms,
             self.prefix_documents_cache,
+            self.prefix_postings_lists_cache,
         )
     }
 
@@ -280,6 +281,7 @@ impl Index {
             self.documents_fields_counts,
             self.synonyms,
             self.prefix_documents_cache,
+            self.prefix_postings_lists_cache,
             criteria,
         )
     }