(bm25Index)add check,bugfix

(cherry picked from commit 661811eb6c9a71dd0e4123e523eb969b0f866b29)

src/common/backend/utils/misc/guc/guc_sql.cpp

@@ -2765,7 +2765,7 @@ static void InitSqlConfigureNamesReal()
             gettext_noop("k1 factor for bm25 metrix."),
             NULL},
             &u_sess->attr.attr_sql.bm25_k1,
-            1.5,
+            1.2,
             0.0,
             3.0,
             NULL,

src/gausskernel/optimizer/commands/indexcmds.cpp

@@ -797,6 +797,9 @@ ObjectAddress DefineIndex(Oid relationId, IndexStmt* stmt, Oid indexRelationId,
      */
     lockmode = concurrent ? ShareUpdateExclusiveLock : ShareLock;
     rel = heap_open(relationId, lockmode);
+    if (RelationIsPartitioned(rel) && strcmp(stmt->accessMethod, "bm25") != 0) {
+        elog(ERROR, "%s index is not supported for partition table.", (stmt->accessMethod));
+    }
 
     bool segment = get_rel_segment(rel);
     TableCreateSupport indexCreateSupport{(int)COMPRESS_TYPE_NONE, false, false, false, false, false, true, false};

src/gausskernel/storage/access/datavec/bm25.cpp

@@ -22,6 +22,7 @@
  */
 
 #include "postgres.h"
+#include "access/multi_redo_api.h"
 #include "access/datavec/bm25.h"
 
 /*
@@ -47,6 +48,9 @@ static void bm25costestimate_internal(PlannerInfo *root, IndexPath *path, double
 PGDLLEXPORT PG_FUNCTION_INFO_V1(bm25build);
 Datum bm25build(PG_FUNCTION_ARGS)
 {
+    if (IsExtremeRedo()) {
+        elog(ERROR, "bm25 index do not support extreme rto.");
+    }
     Relation heap = (Relation)PG_GETARG_POINTER(0);
     Relation index = (Relation)PG_GETARG_POINTER(1);
     IndexInfo *indexinfo = (IndexInfo *)PG_GETARG_POINTER(2);
@@ -58,8 +62,11 @@ Datum bm25build(PG_FUNCTION_ARGS)
 PGDLLEXPORT PG_FUNCTION_INFO_V1(bm25buildempty);
 Datum bm25buildempty(PG_FUNCTION_ARGS)
 {
+    if (IsExtremeRedo()) {
+        elog(ERROR, "bm25 index do not support extreme rto.");
+    }
     Relation index = (Relation)PG_GETARG_POINTER(0);
-    //bm25buildempty_internal(index);
+    bm25buildempty_internal(index);
 
     PG_RETURN_VOID();
 }
@@ -130,6 +137,9 @@ Datum bm25costestimate(PG_FUNCTION_ARGS)
 PGDLLEXPORT PG_FUNCTION_INFO_V1(bm25insert);
 Datum bm25insert(PG_FUNCTION_ARGS)
 {
+    if (IsExtremeRedo()) {
+        elog(ERROR, "bm25 index do not support extreme rto.");
+    }
     Relation rel = (Relation)PG_GETARG_POINTER(0);
     Datum *values = (Datum *)PG_GETARG_POINTER(1);
     bool *isnull = reinterpret_cast<bool *>(PG_GETARG_POINTER(2));
@@ -160,6 +170,9 @@ Datum bm25delete(PG_FUNCTION_ARGS)
 PGDLLEXPORT PG_FUNCTION_INFO_V1(bm25bulkdelete);
 Datum bm25bulkdelete(PG_FUNCTION_ARGS)
 {
+    if (IsExtremeRedo()) {
+        elog(ERROR, "bm25 index do not support extreme rto.");
+    }
     IndexVacuumInfo *info = (IndexVacuumInfo *)PG_GETARG_POINTER(0);
     IndexBulkDeleteResult *volatile stats = (IndexBulkDeleteResult *)PG_GETARG_POINTER(1);
     PG_RETURN_POINTER(stats);
@@ -168,6 +181,9 @@ Datum bm25bulkdelete(PG_FUNCTION_ARGS)
 PGDLLEXPORT PG_FUNCTION_INFO_V1(bm25vacuumcleanup);
 Datum bm25vacuumcleanup(PG_FUNCTION_ARGS)
 {
+    if (IsExtremeRedo()) {
+        elog(ERROR, "bm25 index do not support extreme rto.");
+    }
     IndexVacuumInfo *info = (IndexVacuumInfo *)PG_GETARG_POINTER(0);
     IndexBulkDeleteResult *stats = (IndexBulkDeleteResult *)PG_GETARG_POINTER(1);
     PG_RETURN_POINTER(stats);

src/gausskernel/storage/access/datavec/bm25build.cpp

@@ -418,6 +418,7 @@ static void InsertDocForwardItem(Relation index, uint32 docId, BM25TokenizedDocD
     AllocateForwardIdxForToken(index, tokenizedDoc.tokenCount, forwardStart, forwardEnd, metaForwardPage, forkNum);
     forwardStartBlkno = metaForwardPage->startPage;
     MarkBufferDirty(metabuf);
+    UnlockReleaseBuffer(metabuf);
 
     uint64 tokenIdx = *forwardStart;
     BlockNumber curStep = tokenIdx / BM25_DOC_FORWARD_MAX_COUNT_IN_PAGE;
@@ -1053,11 +1054,24 @@ static void BuildIndex(Relation heap, Relation index, IndexInfo *indexInfo, BM25
     FreeBuildState(buildstate);
 }
 
+static void BuildIndexCheck(Relation index)
+{
+    TupleDesc tupleDesc = RelationGetDescr(index);
+    FormData_pg_attribute* attrs = tupleDesc->attrs;
+    for (int i = 0; i < tupleDesc->natts; ++i) {
+        if (attrs[i].atttypid == TEXTARRAYOID) {
+            elog(ERROR, "bm25 index is not supported currently for datatype: text array.");
+        }
+    }
+    return;
+}
+
 IndexBuildResult* bm25build_internal(Relation heap, Relation index, IndexInfo *indexInfo)
 {
     IndexBuildResult *result;
     BM25BuildState buildstate;
 
+    BuildIndexCheck(index);
     BuildIndex(heap, index, indexInfo, &buildstate, MAIN_FORKNUM);
 
     result = (IndexBuildResult *)palloc(sizeof(IndexBuildResult));
@@ -1071,6 +1085,7 @@ void bm25buildempty_internal(Relation index)
     IndexBuildResult *result;
     BM25BuildState buildstate;
 
+    BuildIndexCheck(index);
     BuildIndex(NULL, index, NULL, &buildstate, MAIN_FORKNUM);
 }
 

src/gausskernel/storage/access/datavec/bm25scan.cpp

@@ -36,6 +36,8 @@
 #include "access/datavec/bm25heap.h"
 #include "access/datavec/bm25.h"
 
+#define BM25_HEAP_DEFAULT_CAPACITY 100
+
 typedef struct BM25QueryToken {
     BlockNumber tokenPostingBlock;
     float qTokenMaxScore;
@@ -534,7 +536,7 @@ static void DocIdsGetHeapCtids(Relation index, BM25EntryPages &entryPages, BM25S
         }
         BM25DocumentItem *docItem = (BM25DocumentItem*)((char *)page + sizeof(PageHeaderData) +
             offset * BM25_DOCUMENT_ITEM_SIZE);
-        if (docItem->isActived) {
+        if (!docItem->isActived) {
             UnlockReleaseBuffer(buf);
             elog(ERROR, "Read invalid doc.");
         }
@@ -550,7 +552,9 @@ static void BM25IndexScan(Relation index, BM25QueryTokensInfo &queryTokenInfo, u
         return;
     }
     BM25Scorer scorer = BM25Scorer(u_sess->attr.attr_sql.bm25_k1, u_sess->attr.attr_sql.bm25_b, avgdl);
-    MaxMinHeap<float> heap(so->expectedCandNums);
+
+    size_t capacity = so->expectedCandNums == 0 ? BM25_HEAP_DEFAULT_CAPACITY : so->expectedCandNums;
+    MaxMinHeap<float> heap(capacity);
     if (so->expectedCandNums == 0) {
         SearchTaat(index, queryTokenInfo, heap, docNums, scorer, so->docIdMask);
     } else {
@@ -558,9 +562,9 @@ static void BM25IndexScan(Relation index, BM25QueryTokensInfo &queryTokenInfo, u
     }
 
     uint32 docId;
-    int size = heap.size();
+    int64 size = heap.size();
     so->candDocs = (BM25ScanData*)palloc0(sizeof(BM25ScanData) * size);
-    for (auto i = size - 1; i >= 0; --i) {
+    for (int64 i = size - 1; i >= 0; --i) {
         docId = heap.top().id;
         so->candDocs[i].docId = docId;
         so->candDocs[i].score = heap.top().val;

src/gausskernel/storage/access/datavec/bm25utils.cpp

@@ -169,6 +169,10 @@ uint32 BM25AllocateDocId(Relation index)
     if (unlikely(metapBuf->magicNumber != BM25_MAGIC_NUMBER))
         elog(ERROR, "bm25 index is not valid");
     docId = metapBuf->nextDocId;
+    if (unlikely(docId == BM25_INVALID_DOC_ID)) {
+        elog(ERROR, "bm25 doc id exhausted, please rebuild index.");
+    }
+
     metapBuf->nextDocId++;
     MarkBufferDirty(buf);
     UnlockReleaseBuffer(buf);
@@ -220,7 +224,7 @@ BlockNumber SeekBlocknoForDoc(Relation index, uint32 docId, BlockNumber startBlk
     Page page;
     BlockNumber docBlkno = startBlkno;
     for (int i = 0; i < step; ++i) {
-        if (unlikely(BlockNumberIsValid(docBlkno))) {
+        if (unlikely(!BlockNumberIsValid(docBlkno))) {
             elog(ERROR, "SeekBlocknoForDoc: Invalid Block Number.");
         }
         buf = ReadBuffer(index, docBlkno);

src/include/access/datavec/bm25.h

@@ -132,8 +132,8 @@ typedef struct BM25DocForwardMetaPageData {
 typedef BM25DocForwardMetaPageData *BM25DocForwardMetaPage;
 
 typedef struct BM25DocForwardItem {
-    bool tokenId;
-    bool tokenHash;
+    uint32 tokenId;
+    uint32 tokenHash;
 } BM25DocForwardItem;
 
 typedef struct BM25DocumentItem {