Kouhei Sutou
null+****@clear*****
Sat Dec 24 00:10:36 JST 2016
Kouhei Sutou 2016-12-24 00:10:36 +0900 (Sat, 24 Dec 2016) New Revision: 106fe049f5886f818345f23f4264dd1519134834 https://github.com/pgroonga/pgroonga/commit/106fe049f5886f818345f23f4264dd1519134834 Message: pgroonga_score: improve performance for one column primary key case GitHub: fix #31 100K rows case: 9sec -> 300msec Reported by yongxianggao-chanjet. Thanks!!! Modified files: src/pgroonga.c Modified: src/pgroonga.c (+139 -36) =================================================================== --- src/pgroonga.c 2016-12-23 22:27:04 +0900 (db9da15) +++ src/pgroonga.c 2016-12-24 00:10:36 +0900 (a804bb2) @@ -906,24 +906,129 @@ PGrnSetSources(Relation index, grn_obj *sourcesTable) #ifdef PGRN_SUPPORT_SCORE static double -PGrnCollectScoreScanOpaque(Relation table, HeapTuple tuple, PGrnScanOpaque so) +PGrnCollectScoreScanOpaqueGetScore(Relation table, + PGrnScanOpaque so, + grn_id recordID) +{ + double score = 0.0; + grn_id id; + ItemPointerData ctid; + + id = grn_table_get(ctx, so->searched, &recordID, sizeof(grn_id)); + if (id == GRN_ID_NIL) + return 0.0; + + GRN_BULK_REWIND(&(buffers->ctid)); + grn_obj_get_value(ctx, so->ctidAccessor, id, &(buffers->ctid)); + ctid = PGrnCtidUnpack(GRN_UINT64_VALUE(&(buffers->ctid))); + + if (!PGrnCtidIsAlive(table, &ctid)) + return 0.0; + + GRN_BULK_REWIND(&(buffers->score)); + grn_obj_get_value(ctx, so->scoreAccessor, id, &(buffers->score)); + if (buffers->score.header.domain == GRN_DB_FLOAT) + { + score = GRN_FLOAT_VALUE(&(buffers->score)); + } + else + { + score = GRN_INT32_VALUE(&(buffers->score)); + } + + return score; +} + +static double +PGrnCollectScoreScanOpaqueOneColumnPrimaryKey(Relation table, + HeapTuple tuple, + PGrnScanOpaque so) { double score = 0.0; TupleDesc desc; - grn_obj *expression; - grn_obj *variable; - slist_iter iter; - unsigned int nPrimaryKeyColumns = 0; + PGrnPrimaryKeyColumn *primaryKeyColumn; + grn_index_datum indexDatum; + grn_obj *lexicon; + grn_id termID; + grn_ii_cursor *iiCursor; + int iiNElements = 2; + grn_posting *posting; - if (so->dataTableID != tuple->t_tableOid) + desc = RelationGetDescr(table); + primaryKeyColumn = slist_container(PGrnPrimaryKeyColumn, + node, + so->primaryKeyColumns.head.next); + + { + unsigned int nIndexData; + + nIndexData = grn_column_find_index_data(ctx, + primaryKeyColumn->column, + GRN_OP_EQUAL, + &indexDatum, + 1); + if (nIndexData == 0) + return 0.0; + } + + lexicon = grn_ctx_at(ctx, indexDatum.index->header.domain); + if (!lexicon) return 0.0; - if (!so->scoreAccessor) + { + bool isNULL; + Datum primaryKeyValue; + + grn_obj_reinit(ctx, + &(buffers->general), + primaryKeyColumn->domain, + primaryKeyColumn->flags); + primaryKeyValue = heap_getattr(tuple, + primaryKeyColumn->number, + desc, + &isNULL); + PGrnConvertFromData(primaryKeyValue, + primaryKeyColumn->type, + &(buffers->general)); + } + termID = grn_table_get(ctx, + lexicon, + GRN_BULK_HEAD(&(buffers->general)), + GRN_BULK_VSIZE(&(buffers->general))); + if (termID == GRN_ID_NIL) return 0.0; - if (slist_is_empty(&(so->primaryKeyColumns))) + iiCursor = grn_ii_cursor_open(ctx, + (grn_ii *)(indexDatum.index), + termID, + GRN_ID_NIL, + GRN_ID_NIL, + iiNElements, + 0); + if (!iiCursor) return 0.0; + while ((posting = grn_ii_cursor_next(ctx, iiCursor))) + { + score += PGrnCollectScoreScanOpaqueGetScore(table, so, posting->rid); + } + grn_ii_cursor_close(ctx, iiCursor); + + return score; +} + +static double +PGrnCollectScoreScanOpaqueMultiColumnPrimaryKey(Relation table, + HeapTuple tuple, + PGrnScanOpaque so) +{ + double score = 0.0; + TupleDesc desc; + grn_obj *expression; + grn_obj *variable; + slist_iter iter; + unsigned int nPrimaryKeyColumns = 0; + desc = RelationGetDescr(table); if (!so->scoreTargetRecords) @@ -984,38 +1089,14 @@ PGrnCollectScoreScanOpaque(Relation table, HeapTuple tuple, PGrnScanOpaque so) 0, -1, GRN_CURSOR_ASCENDING); while (grn_table_cursor_next(ctx, tableCursor) != GRN_ID_NIL) { + void *key; grn_id recordID; - grn_id id; - ItemPointerData ctid; - { - void *key; - grn_table_cursor_get_key(ctx, tableCursor, &key); - recordID = *((grn_id *) key); - } + grn_table_cursor_get_key(ctx, tableCursor, &key); + recordID = *((grn_id *) key); grn_table_cursor_delete(ctx, tableCursor); - id = grn_table_get(ctx, so->searched, &recordID, sizeof(grn_id)); - if (id == GRN_ID_NIL) - continue; - - GRN_BULK_REWIND(&(buffers->ctid)); - grn_obj_get_value(ctx, so->ctidAccessor, id, &(buffers->ctid)); - ctid = PGrnCtidUnpack(GRN_UINT64_VALUE(&(buffers->ctid))); - - if (!PGrnCtidIsAlive(table, &ctid)) - continue; - - GRN_BULK_REWIND(&(buffers->score)); - grn_obj_get_value(ctx, so->scoreAccessor, id, &(buffers->score)); - if (buffers->score.header.domain == GRN_DB_FLOAT) - { - score += GRN_FLOAT_VALUE(&(buffers->score)); - } - else - { - score += GRN_INT32_VALUE(&(buffers->score)); - } + score += PGrnCollectScoreScanOpaqueGetScore(table, so, recordID); } grn_obj_unlink(ctx, tableCursor); } @@ -1024,6 +1105,28 @@ PGrnCollectScoreScanOpaque(Relation table, HeapTuple tuple, PGrnScanOpaque so) } static double +PGrnCollectScoreScanOpaque(Relation table, HeapTuple tuple, PGrnScanOpaque so) +{ + if (so->dataTableID != tuple->t_tableOid) + return 0.0; + + if (!so->scoreAccessor) + return 0.0; + + if (slist_is_empty(&(so->primaryKeyColumns))) + return 0.0; + + if (so->primaryKeyColumns.head.next->next) + { + return PGrnCollectScoreScanOpaqueMultiColumnPrimaryKey(table, tuple, so); + } + else + { + return PGrnCollectScoreScanOpaqueOneColumnPrimaryKey(table, tuple, so); + } +} + +static double PGrnCollectScore(Relation table, HeapTuple tuple) { double score = 0.0; -------------- next part -------------- HTML����������������������������...다운로드
