Merge pull request #15639 from taosdata/szhou/fix/join

fix: fix failed case of join.sim

include/util/taoserror.h

@@ -557,6 +557,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_INVALID_SMA_INDEX         TAOS_DEF_ERROR_CODE(0, 0x2660)
 #define TSDB_CODE_PAR_INVALID_SELECTED_EXPR     TAOS_DEF_ERROR_CODE(0, 0x2661)
 #define TSDB_CODE_PAR_GET_META_ERROR            TAOS_DEF_ERROR_CODE(0, 0x2662)
+#define TSDB_CODE_PAR_NOT_UNIQUE_TABLE_ALIAS    TAOS_DEF_ERROR_CODE(0, 0x2663)
 #define TSDB_CODE_PAR_INTERNAL_ERROR            TAOS_DEF_ERROR_CODE(0, 0x26FF)
 
 //planner

source/libs/executor/src/joinoperator.c

@@ -323,8 +323,6 @@ static void doMergeJoinImpl(struct SOperatorInfo* pOperator, SSDataBlock* pRes)
     }
 
     if (leftTs == rightTs) {
-      mergeJoinJoinLeftRight(pOperator, pRes, nrows, pJoinInfo->pLeft, pJoinInfo->leftPos, pJoinInfo->pRight,
-                             pJoinInfo->rightPos);
       mergeJoinJoinDownstreamTsRanges(pOperator, leftTs, pRes, &nrows);
     } else if (asc && leftTs < rightTs || !asc && leftTs > rightTs) {
       pJoinInfo->leftPos += 1;

source/libs/parser/src/parTranslater.c

@@ -61,16 +61,42 @@ static bool beforeHaving(ESqlClause clause) { return clause < SQL_CLAUSE_HAVING;
 
 static bool afterHaving(ESqlClause clause) { return clause > SQL_CLAUSE_HAVING; }
 
+static bool hasSameTableAlias(SArray* pTables) {
+  if (taosArrayGetSize(pTables) < 2) {
+    return false;
+  }
+  STableNode* pTable0 = taosArrayGetP(pTables, 0);
+  for (int32_t i = 1; i < taosArrayGetSize(pTables); ++i) {
+    STableNode* pTable = taosArrayGetP(pTables, i);
+    if (0 == strcmp(pTable0->tableAlias, pTable->tableAlias)) {
+      return true;
+    }
+  }
+  return false;
+}
+
 static int32_t addNamespace(STranslateContext* pCxt, void* pTable) {
   size_t currTotalLevel = taosArrayGetSize(pCxt->pNsLevel);
   if (currTotalLevel > pCxt->currLevel) {
     SArray* pTables = taosArrayGetP(pCxt->pNsLevel, pCxt->currLevel);
     taosArrayPush(pTables, &pTable);
+    if (hasSameTableAlias(pTables)) {
+      return generateSyntaxErrMsgExt(&pCxt->msgBuf,
+                              TSDB_CODE_PAR_NOT_UNIQUE_TABLE_ALIAS,
+                              "Not unique table/alias: '%s'",
+                              ((STableNode*)pTable)->tableAlias);
+    }
   } else {
     do {
       SArray* pTables = taosArrayInit(TARRAY_MIN_SIZE, POINTER_BYTES);
       if (pCxt->currLevel == currTotalLevel) {
         taosArrayPush(pTables, &pTable);
+        if (hasSameTableAlias(pTables)) {
+          return generateSyntaxErrMsgExt(&pCxt->msgBuf,
+                                  TSDB_CODE_PAR_NOT_UNIQUE_TABLE_ALIAS,
+                                  "Not unique table/alias: '%s'",
+                                  ((STableNode*)pTable)->tableAlias);
+        }
       }
       taosArrayPush(pCxt->pNsLevel, &pTables);
       ++currTotalLevel;

source/libs/scalar/src/filter.c

@@ -3929,6 +3929,26 @@ static EConditionType classifyCondition(SNode* pNode) {
                                                      : (cxt.hasTagIndexCol ? COND_TYPE_TAG_INDEX : COND_TYPE_TAG)));
 }
 
+static bool isCondColumnsFromMultiTable(SNode* pCond) {
+  SNodeList* pCondCols = nodesMakeList();
+  int32_t code = nodesCollectColumnsFromNode(pCond, NULL, COLLECT_COL_TYPE_ALL, &pCondCols);
+  if (code == TSDB_CODE_SUCCESS) {
+    if (LIST_LENGTH(pCondCols) >= 2) {
+      SColumnNode* pFirstCol = (SColumnNode*)nodesListGetNode(pCondCols, 0);
+      SNode* pColNode = NULL;
+      FOREACH(pColNode, pCondCols) {
+        if (strcmp(((SColumnNode*)pColNode)->dbName, pFirstCol->dbName) != 0 ||
+            strcmp(((SColumnNode*)pColNode)->tableAlias, pFirstCol->tableAlias) != 0) {
+          nodesDestroyList(pCondCols);
+          return true;
+        }
+      }
+    }
+    nodesDestroyList(pCondCols);
+  }
+  return false;
+}
+
 static int32_t partitionLogicCond(SNode** pCondition, SNode** pPrimaryKeyCond, SNode** pTagIndexCond, SNode** pTagCond,
                                   SNode** pOtherCond) {
   SLogicConditionNode* pLogicCond = (SLogicConditionNode*)(*pCondition);
@@ -3941,6 +3961,11 @@ static int32_t partitionLogicCond(SNode** pCondition, SNode** pPrimaryKeyCond, S
   SNodeList* pOtherConds = NULL;
   SNode*     pCond = NULL;
   FOREACH(pCond, pLogicCond->pParameterList) {
+    if (isCondColumnsFromMultiTable(pCond)) {
+      if (NULL != pOtherCond) {
+        code = nodesListMakeAppend(&pOtherConds, nodesCloneNode(pCond));
+      }
+    } else {
       switch (classifyCondition(pCond)) {
         case COND_TYPE_PRIMARY_KEY:
           if (NULL != pPrimaryKeyCond) {
@@ -3967,6 +3992,7 @@ static int32_t partitionLogicCond(SNode** pCondition, SNode** pPrimaryKeyCond, S
           }
           break;
       }
+    }
     if (TSDB_CODE_SUCCESS != code) {
       break;
     }
@@ -4026,6 +4052,12 @@ int32_t filterPartitionCond(SNode** pCondition, SNode** pPrimaryKeyCond, SNode**
   }
 
   bool needOutput = false;
+  if (isCondColumnsFromMultiTable(*pCondition)) {
+    if (NULL != pOtherCond) {
+      *pOtherCond = *pCondition;
+      needOutput = true;
+    }
+  } else {
     switch (classifyCondition(*pCondition)) {
       case COND_TYPE_PRIMARY_KEY:
         if (NULL != pPrimaryKeyCond) {
@@ -4039,7 +4071,7 @@ int32_t filterPartitionCond(SNode** pCondition, SNode** pPrimaryKeyCond, SNode**
           needOutput = true;
         }
         if (NULL != pTagCond) {
-        SNode* pTempCond = *pCondition;
+          SNode *pTempCond = *pCondition;
           if (NULL != pTagIndexCond) {
             pTempCond = nodesCloneNode(*pCondition);
             if (NULL == pTempCond) {
@@ -4064,6 +4096,7 @@ int32_t filterPartitionCond(SNode** pCondition, SNode** pPrimaryKeyCond, SNode**
         }
         break;
     }
+  }
   if (needOutput) {
     *pCondition = NULL;
   }

source/util/src/terror.c

@@ -561,6 +561,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_PAR_ONLY_SUPPORT_SINGLE_TABLE,  "Only support single
 TAOS_DEFINE_ERROR(TSDB_CODE_PAR_INVALID_SMA_INDEX,          "Invalid sma index")
 TAOS_DEFINE_ERROR(TSDB_CODE_PAR_INVALID_SELECTED_EXPR,      "Invalid SELECTed expression")
 TAOS_DEFINE_ERROR(TSDB_CODE_PAR_GET_META_ERROR,             "Fail to get table info")
+TAOS_DEFINE_ERROR(TSDB_CODE_PAR_NOT_UNIQUE_TABLE_ALIAS,     "Not unique table/alias")
 TAOS_DEFINE_ERROR(TSDB_CODE_PAR_INTERNAL_ERROR,             "Parser internal error")
 
 //planner

tests/script/jenkins/basic.txt

@@ -133,7 +133,7 @@
 ./test.sh -f tsim/parser/join_manyblocks.sim
 # TD-18018 ./test.sh -f tsim/parser/join_multitables.sim
 ./test.sh -f tsim/parser/join_multivnode.sim
-# TD-17707 ./test.sh -f tsim/parser/join.sim
+./test.sh -f tsim/parser/join.sim
 ./test.sh -f tsim/parser/last_cache.sim
 ./test.sh -f tsim/parser/last_groupby.sim
 ./test.sh -f tsim/parser/lastrow.sim

tests/script/tsim/parser/join.sim

@@ -320,7 +320,7 @@ sql_error select count(join_tb1.c3), last(join_tb1.c3) from $tb1 , $tb2 where jo
 sql_error select count(join_tb3.*) from $tb1 , $tb2 where join_tb1.ts = join_tb0.ts and join_tb1.ts <= 100002 and join_tb0.c7 = true;
 sql_error select first(join_tb1.*) from $tb1 , $tb2 where join_tb1.ts = join_tb0.ts and join_tb1.ts <= 100002 or join_tb0.c7 = true;
 sql_error select join_tb3.* from $tb1 , $tb2 where join_tb1.ts = join_tb0.ts and join_tb1.ts <= 100002 and join_tb0.c7 = true;
-sql_error select join_tb1.* from $tb1 , $tb2 where join_tb1.ts = join_tb0.ts and join_tb1.ts = join_tb0.c1;
+sql select join_tb1.* from $tb1 , $tb2 where join_tb1.ts = join_tb0.ts and join_tb1.ts = join_tb0.c1;
 sql_error select join_tb1.* from $tb1 , $tb2 where join_tb1.ts = join_tb0.c1;
 sql_error select join_tb1.* from $tb1 , $tb2 where join_tb1.c7 = join_tb0.c1;
 sql_error select join_tb1.* from $tb1 , $tb2 where join_tb1.ts > join_tb0.ts;
@@ -407,32 +407,32 @@ if $data00 != 396.000000000 then
 endi
 
 # first/last
-sql select count(join_mt0.c1), sum(join_mt1.c2), first(join_mt0.c5), last(join_mt1.c7) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts and join_mt0.t1=1 interval(10a) order by join_mt0.ts asc;
+sql select count(join_mt0.c1), sum(join_mt1.c2), first(join_mt0.c5), last(join_mt1.c7) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts and join_mt0.t1=1 interval(10a) order by _wstart asc;
 
 $val = 100
 if $rows != $val then
    print $rows
    return -1
 endi
-
+print $data00 $data01 $data02
 $val = 10
-if $data01 != $val then
+if $data00 != $val then
    return -1
 endi
 
 $val = 45.000000000
-print $data02
-if $data02 != $val then
+print $data01
+if $data01 != $val then
    return -1
 endi
 
 $val = 0
-if $data03 != 0 then
+if $data02 != 0 then
    return -1
 endi
 
 # order by first/last
-sql select count(join_mt0.c1), sum(join_mt1.c2), first(join_mt0.c5), last(join_mt1.c7) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts and join_mt0.t1=1 interval(10a) order by join_mt0.ts desc;
+sql select count(join_mt0.c1), sum(join_mt1.c2), first(join_mt0.c5), last(join_mt1.c7) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts and join_mt0.t1=1 interval(10a) order by _wstart desc;
 
 $val = 100
 if $rows != $val then
@@ -453,13 +453,13 @@ print =================>"group by not supported"
 
 #======================limit offset===================================
 # tag values not int
-sql_error select count(*) from join_mt0, join_mt1 where join_mt0.ts=join_mt1.ts and join_mt0.t2=join_mt1.t2;
+sql select count(*) from join_mt0, join_mt1 where join_mt0.ts=join_mt1.ts and join_mt0.t2=join_mt1.t2;
 
 # tag type not identical
-sql_error select count(*) from join_mt0, join_mt1 where join_mt1.t2 = join_mt0.t1 and join_mt1.ts=join_mt0.ts;
+sql select count(*) from join_mt0, join_mt1 where join_mt1.t2 = join_mt0.t1 and join_mt1.ts=join_mt0.ts;
 
 # table/super table join
-sql_error select count(join_mt0.c1) from join_mt0, join_tb1 where join_mt0.ts=join_tb1.ts
+sql select count(join_mt0.c1) from join_mt0, join_tb1 where join_mt0.ts=join_tb1.ts
 
 # multi-condition
 
@@ -471,7 +471,7 @@ sql_error select count(join_mt0.c1), count(join_mt0.c2) from join_mt0, join_mt0
 sql_error select sum(join_mt1.c2) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1;
 
 # missing tag equals
-sql_error select count(join_mt1.c3) from join_mt0, join_mt1 where join_mt0.ts=join_mt1.ts;
+sql select count(join_mt1.c3) from join_mt0, join_mt1 where join_mt0.ts=join_mt1.ts;
 
 # tag values are identical error
 sql create table m1(ts timestamp, k int) tags(a int);
@@ -486,7 +486,7 @@ sql insert into tm2 using m1 tags(1) values(1000000, 1)(2000000, 2);
 sql insert into um1 using m2 tags(1) values(1000001, 10)(2000000, 20);
 sql insert into um2 using m2 tags(9) values(1000001, 10)(2000000, 20);
 
-sql_error select count(*) from m1,m2 where m1.a=m2.a and m1.ts=m2.ts;
+sql select count(*) from m1,m2 where m1.a=m2.a and m1.ts=m2.ts;
 
 print ====> empty table/empty super-table join test, add for no result join test
 sql create database ux1;