[fix](inverted index) Fix for Inaccurate match_phrase_prefix Cache in…

be/src/olap/rowset/segment_v2/inverted_index_reader.cpp

@@ -227,7 +227,7 @@ Status InvertedIndexReader::match_index_search(
         InvertedIndexQueryType query_type, const InvertedIndexQueryInfo& query_info,
         const FulltextIndexSearcherPtr& index_searcher,
         const std::shared_ptr<roaring::Roaring>& term_match_bitmap) {
-    TQueryOptions queryOptions = runtime_state->query_options();
+    const auto& queryOptions = runtime_state->query_options();
     try {
         SCOPED_RAW_TIMER(&stats->inverted_index_searcher_search_timer);
         auto query = QueryFactory::create(query_type, index_searcher, queryOptions, io_ctx);
@@ -262,13 +262,14 @@ Status FullTextIndexReader::query(const io::IOContext* io_ctx, OlapReaderStatist
     VLOG_DEBUG << column_name << " begin to search the fulltext index from clucene, query_str ["
                << search_str << "]";
 
+    const auto& queryOptions = runtime_state->query_options();
     try {
         InvertedIndexQueryInfo query_info;
         InvertedIndexQueryCache::CacheKey cache_key;
         auto index_file_key = _inverted_index_file_reader->get_index_file_cache_key(&_index_meta);
 
+        // terms
         if (query_type == InvertedIndexQueryType::MATCH_REGEXP_QUERY) {
-            cache_key = {index_file_key, column_name, query_type, search_str};
             query_info.terms.emplace_back(search_str);
         } else {
             if (query_type == InvertedIndexQueryType::MATCH_PHRASE_QUERY) {
@@ -291,6 +292,7 @@ Status FullTextIndexReader::query(const io::IOContext* io_ctx, OlapReaderStatist
             query_info.terms = inverted_index::InvertedIndexAnalyzer::get_analyse_result(
                     reader.get(), analyzer.get(), column_name, query_type);
         }
+
         if (query_info.terms.empty()) {
             auto msg = fmt::format(
                     "token parser result is empty for query, "
@@ -304,22 +306,20 @@ Status FullTextIndexReader::query(const io::IOContext* io_ctx, OlapReaderStatist
             }
         }
 
-        std::unique_ptr<lucene::search::Query> query;
+        // field_name
         query_info.field_name = StringUtil::string_to_wstring(column_name);
 
-        if (query_type == InvertedIndexQueryType::MATCH_PHRASE_QUERY ||
-            query_type == InvertedIndexQueryType::MATCH_PHRASE_PREFIX_QUERY ||
-            query_type == InvertedIndexQueryType::MATCH_PHRASE_EDGE_QUERY ||
-            query_type == InvertedIndexQueryType::MATCH_ALL_QUERY ||
-            query_type == InvertedIndexQueryType::EQUAL_QUERY ||
-            query_type == InvertedIndexQueryType::MATCH_ANY_QUERY) {
-            std::string str_tokens = join(query_info.terms, " ");
-            if (query_type == InvertedIndexQueryType::MATCH_PHRASE_QUERY) {
-                str_tokens += " " + std::to_string(query_info.slop);
-                str_tokens += " " + std::to_string(query_info.ordered);
-            }
-            cache_key = {index_file_key, column_name, query_type, str_tokens};
+        // cache_key
+        std::string str_tokens = join(query_info.terms, " ");
+        if (query_type == InvertedIndexQueryType::MATCH_PHRASE_QUERY) {
+            str_tokens += " " + std::to_string(query_info.slop);
+            str_tokens += " " + std::to_string(query_info.ordered);
+        } else if (query_type == InvertedIndexQueryType::MATCH_PHRASE_PREFIX_QUERY ||
+                   query_type == InvertedIndexQueryType::MATCH_REGEXP_QUERY) {
+            str_tokens += " " + std::to_string(queryOptions.inverted_index_max_expansions);
         }
+        cache_key = {index_file_key, column_name, query_type, std::move(str_tokens)};
+
         auto* cache = InvertedIndexQueryCache::instance();
         InvertedIndexQueryCacheHandle cache_handler;
 
@@ -328,12 +328,11 @@ Status FullTextIndexReader::query(const io::IOContext* io_ctx, OlapReaderStatist
         if (cache_status.ok()) {
             return Status::OK();
         }
-        FulltextIndexSearcherPtr* searcher_ptr = nullptr;
 
         InvertedIndexCacheHandle inverted_index_cache_handle;
         RETURN_IF_ERROR(handle_searcher_cache(&inverted_index_cache_handle, io_ctx, stats));
         auto searcher_variant = inverted_index_cache_handle.get_index_searcher();
-        searcher_ptr = std::get_if<FulltextIndexSearcherPtr>(&searcher_variant);
+        auto* searcher_ptr = std::get_if<FulltextIndexSearcherPtr>(&searcher_variant);
         if (searcher_ptr != nullptr) {
             term_match_bitmap = std::make_shared<roaring::Roaring>();
             RETURN_IF_ERROR(match_index_search(io_ctx, stats, runtime_state, query_type, query_info,

regression-test/data/inverted_index_p0/test_index_match_phrase_prefix_1.out

@@ -11,3 +11,9 @@
 -- !sql --
 6
 
+-- !sql --
+0
+
+-- !sql --
+1
+

regression-test/data/inverted_index_p0/test_index_match_regexp.out

@@ -20,3 +20,9 @@
 -- !sql --
 0
 
+-- !sql --
+4
+
+-- !sql --
+377
+

regression-test/suites/inverted_index_p0/test_index_match_phrase_prefix_1.groovy

@@ -47,6 +47,8 @@ suite("test_index_match_phrase_prefix_1", "nonConcurrent"){
     sql """ INSERT INTO ${indexTbName1} VALUES (6, "O1704361998540E2Cemx9S 123456789", "O1704361998540E2Cemx9S 123456789", "O1704361998540E2Cemx9S 123456789"); """
     sql """ INSERT INTO ${indexTbName1} VALUES (7, "O1704361998540E2Cemx9S*123456789", "O1704361998540E2Cemx9S*123456789", "O1704361998540E2Cemx9S*123456789"); """
 
+    sql """ INSERT INTO ${indexTbName1} VALUES (1, "", "s1", ""), (2, "", "s2", ""), (3, "", "s3", ""), (4, "", "s4", ""), (5, "", "tv s5", ""); """
+
     try {
         sql "sync"
         sql """ set enable_common_expr_pushdown = true; """
@@ -58,7 +60,14 @@ suite("test_index_match_phrase_prefix_1", "nonConcurrent"){
         qt_sql """ select count() from ${indexTbName1} where c match_phrase_prefix 'O1704361998540E2Cemx9S=123456789'; """
         qt_sql """ select count() from ${indexTbName1} where d match_phrase_prefix 'O1704361998540E2Cemx9S=123456789'; """
 
+        sql """ set inverted_index_max_expansions = 3; """
+        qt_sql """ select count() from ${indexTbName1} where c match_phrase_prefix 'tv s'; """
+
+        sql """ set inverted_index_max_expansions = 5; """
+        qt_sql """ select count() from ${indexTbName1} where c match_phrase_prefix 'tv s'; """
+
     } finally {
+        sql """ set inverted_index_max_expansions = 50; """
         GetDebugPoint().disableDebugPointForAllBEs("VMatchPredicate.execute")
     }
 }

regression-test/suites/inverted_index_p0/test_index_match_regexp.groovy

@@ -90,7 +90,14 @@ suite("test_index_match_regexp", "nonConcurrent"){
         qt_sql """ select count() from test_index_match_regexp where request match_regexp '.*tickets.*'; """
         qt_sql """ select count() from test_index_match_regexp where request match_regexp 'nonexistence'; """
 
+        sql """ set inverted_index_max_expansions = 1; """
+        qt_sql """ select count() from test_index_match_regexp where request match_regexp 'b'; """
+
+        sql """ set inverted_index_max_expansions = 50; """
+        qt_sql """ select count() from test_index_match_regexp where request match_regexp 'b'; """
+
     } finally {
+        sql """ set inverted_index_max_expansions = 50; """
         GetDebugPoint().disableDebugPointForAllBEs("VMatchPredicate.execute")
     }
 }