Merge pull request #12466 from taosdata/feature/3.0_wxy

fix: some problems of parser and planner

include/libs/nodes/querynodes.h

@@ -345,6 +345,7 @@ bool nodesIsUnaryOp(const SOperatorNode* pOp);
 bool nodesIsArithmeticOp(const SOperatorNode* pOp);
 bool nodesIsComparisonOp(const SOperatorNode* pOp);
 bool nodesIsJsonOp(const SOperatorNode* pOp);
+bool nodesIsRegularOp(const SOperatorNode* pOp);
 
 bool nodesIsTimeorderQuery(const SNode* pQuery);
 bool nodesIsTimelineQuery(const SNode* pQuery);

source/libs/nodes/src/nodesUtilFuncs.c

@@ -1112,6 +1112,19 @@ bool nodesIsJsonOp(const SOperatorNode* pOp) {
   return false;
 }
 
+bool nodesIsRegularOp(const SOperatorNode* pOp) {
+  switch (pOp->opType) {
+    case OP_TYPE_LIKE:
+    case OP_TYPE_NOT_LIKE:
+    case OP_TYPE_MATCH:
+    case OP_TYPE_NMATCH:
+      return true;
+    default:
+      break;
+  }
+  return false;
+}
+
 bool nodesIsTimeorderQuery(const SNode* pQuery) { return false; }
 
 bool nodesIsTimelineQuery(const SNode* pQuery) { return false; }

source/libs/parser/inc/sql.y

@@ -868,9 +868,9 @@ query_expression_body(A) ::=
   query_expression_body(B) UNION query_expression_body(D).                        { A = createSetOperator(pCxt, SET_OP_TYPE_UNION, B, D); }
 
 query_primary(A) ::= query_specification(B).                                      { A = B; }
-//query_primary(A) ::=
+query_primary(A) ::=
-//  NK_LP query_expression_body(B) 
+  NK_LP query_expression_body(B) 
-//    order_by_clause_opt slimit_clause_opt limit_clause_opt NK_RP.                 { A = B; }
+    order_by_clause_opt slimit_clause_opt limit_clause_opt NK_RP.                 { A = B; }
 
 %type order_by_clause_opt                                                         { SNodeList* }
 %destructor order_by_clause_opt                                                   { nodesDestroyList($$); }

source/libs/parser/src/parTranslater.c

@@ -480,8 +480,8 @@ static EDealRes translateColumn(STranslateContext* pCxt, SColumnNode* pCol) {
   return res;
 }
 
-static EDealRes translateValue(STranslateContext* pCxt, SValueNode* pVal) {
+static EDealRes translateValueImpl(STranslateContext* pCxt, SValueNode* pVal, SDataType targetDt) {
-  uint8_t precision = (NULL != pCxt->pCurrStmt ? pCxt->pCurrStmt->precision : pVal->node.resType.precision);
+  uint8_t precision = (NULL != pCxt->pCurrStmt ? pCxt->pCurrStmt->precision : targetDt.precision);
   pVal->node.resType.precision = precision;
   if (pVal->placeholderNo > 0) {
     return DEAL_RES_CONTINUE;
@@ -493,7 +493,7 @@ static EDealRes translateValue(STranslateContext* pCxt, SValueNode* pVal) {
     }
     *(int64_t*)&pVal->typeData = pVal->datum.i;
   } else {
-    switch (pVal->node.resType.type) {
+    switch (targetDt.type) {
       case TSDB_DATA_TYPE_NULL:
         break;
       case TSDB_DATA_TYPE_BOOL:
@@ -562,35 +562,54 @@ static EDealRes translateValue(STranslateContext* pCxt, SValueNode* pVal) {
       }
       case TSDB_DATA_TYPE_VARCHAR:
       case TSDB_DATA_TYPE_VARBINARY: {
-        pVal->datum.p = taosMemoryCalloc(1, pVal->node.resType.bytes + VARSTR_HEADER_SIZE + 1);
+        pVal->datum.p = taosMemoryCalloc(1, targetDt.bytes + VARSTR_HEADER_SIZE + 1);
         if (NULL == pVal->datum.p) {
           return generateDealNodeErrMsg(pCxt, TSDB_CODE_OUT_OF_MEMORY);
         }
-        varDataSetLen(pVal->datum.p, pVal->node.resType.bytes);
+        varDataSetLen(pVal->datum.p, targetDt.bytes);
-        strncpy(varDataVal(pVal->datum.p), pVal->literal, pVal->node.resType.bytes);
+        strncpy(varDataVal(pVal->datum.p), pVal->literal, targetDt.bytes);
         break;
       }
       case TSDB_DATA_TYPE_TIMESTAMP: {
-        if (taosParseTime(pVal->literal, &pVal->datum.i, pVal->node.resType.bytes, precision, tsDaylight) !=
+        if (taosParseTime(pVal->literal, &pVal->datum.i, targetDt.bytes, precision, tsDaylight) != TSDB_CODE_SUCCESS) {
-            TSDB_CODE_SUCCESS) {
           return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pVal->literal);
         }
         *(int64_t*)&pVal->typeData = pVal->datum.i;
         break;
       }
-      case TSDB_DATA_TYPE_NCHAR:
+      case TSDB_DATA_TYPE_NCHAR: {
+        int32_t bytes = targetDt.bytes * TSDB_NCHAR_SIZE;
+        pVal->datum.p = taosMemoryCalloc(1, bytes + VARSTR_HEADER_SIZE + 1);
+        if (NULL == pVal->datum.p) {
+          return generateDealNodeErrMsg(pCxt, TSDB_CODE_OUT_OF_MEMORY);
+          ;
+        }
+
+        int32_t output = 0;
+        if (!taosMbsToUcs4(pVal->literal, pVal->node.resType.bytes, (TdUcs4*)varDataVal(pVal->datum.p), bytes,
+                           &output)) {
+          return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pVal->literal);
+        }
+        varDataSetLen(pVal->datum.p, output);
+        break;
+      }
       case TSDB_DATA_TYPE_JSON:
       case TSDB_DATA_TYPE_DECIMAL:
       case TSDB_DATA_TYPE_BLOB:
-        // todo
+        return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pVal->literal);
       default:
         break;
     }
   }
+  pVal->node.resType = targetDt;
   pVal->translate = true;
   return DEAL_RES_CONTINUE;
 }
 
+static EDealRes translateValue(STranslateContext* pCxt, SValueNode* pVal) {
+  return translateValueImpl(pCxt, pVal, pVal->node.resType);
+}
+
 static EDealRes translateOperator(STranslateContext* pCxt, SOperatorNode* pOp) {
   if (nodesIsUnaryOp(pOp)) {
     if (OP_TYPE_MINUS == pOp->opType) {
@@ -635,6 +654,14 @@ static EDealRes translateOperator(STranslateContext* pCxt, SOperatorNode* pOp) {
     if (OP_TYPE_IN == pOp->opType || OP_TYPE_NOT_IN == pOp->opType) {
       ((SExprNode*)pOp->pRight)->resType = ((SExprNode*)pOp->pLeft)->resType;
     }
+    if (nodesIsRegularOp(pOp)) {
+      if (!IS_STR_DATA_TYPE(((SExprNode*)(pOp->pLeft))->resType.type)) {
+        return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pLeft))->aliasName);
+      }
+      if (QUERY_NODE_VALUE != nodeType(pOp->pRight) || !IS_STR_DATA_TYPE(((SExprNode*)(pOp->pRight))->resType.type)) {
+        return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
+      }
+    }
     pOp->node.resType.type = TSDB_DATA_TYPE_BOOL;
     pOp->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_BOOL].bytes;
   } else if (nodesIsJsonOp(pOp)) {
@@ -3858,11 +3885,23 @@ static int32_t createValueFromFunction(STranslateContext* pCxt, SFunctionNode* p
   return scalarCalculateConstants((SNode*)pFunc, (SNode**)pVal);
 }
 
-static int32_t translateTagVal(STranslateContext* pCxt, SNode* pNode, SValueNode** pVal) {
+static SDataType schemaToDataType(SSchema* pSchema) {
+  SDataType dt = {.type = pSchema->type, .bytes = pSchema->bytes, .precision = 0, .scale = 0};
+  if (TSDB_DATA_TYPE_VARCHAR == dt.type || TSDB_DATA_TYPE_BINARY == dt.type || TSDB_DATA_TYPE_VARBINARY == dt.type) {
+    dt.bytes -= VARSTR_HEADER_SIZE;
+  } else if (TSDB_DATA_TYPE_NCHAR == dt.type) {
+    dt.bytes = (dt.bytes - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE;
+  }
+  return dt;
+}
+
+static int32_t translateTagVal(STranslateContext* pCxt, SSchema* pSchema, SNode* pNode, SValueNode** pVal) {
   if (QUERY_NODE_FUNCTION == nodeType(pNode)) {
     return createValueFromFunction(pCxt, (SFunctionNode*)pNode, pVal);
   } else if (QUERY_NODE_VALUE == nodeType(pNode)) {
-    return (DEAL_RES_ERROR == translateValue(pCxt, (SValueNode*)pNode) ? pCxt->errCode : TSDB_CODE_SUCCESS);
+    return (DEAL_RES_ERROR == translateValueImpl(pCxt, (SValueNode*)pNode, schemaToDataType(pSchema))
+                ? pCxt->errCode
+                : TSDB_CODE_SUCCESS);
   } else {
     return TSDB_CODE_FAILED;
   }
@@ -3891,7 +3930,7 @@ static int32_t buildKVRowForBindTags(STranslateContext* pCxt, SCreateSubTableCla
       return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_TAG_NAME, pCol->colName);
     }
     SValueNode* pVal = NULL;
-    int32_t     code = translateTagVal(pCxt, pNode, &pVal);
+    int32_t     code = translateTagVal(pCxt, pSchema, pNode, &pVal);
     if (TSDB_CODE_SUCCESS == code) {
       if (NULL == pVal) {
         pVal = (SValueNode*)pNode;
@@ -3921,7 +3960,7 @@ static int32_t buildKVRowForAllTags(STranslateContext* pCxt, SCreateSubTableClau
   int32_t  index = 0;
   FOREACH(pNode, pStmt->pValsOfTags) {
     SValueNode* pVal = NULL;
-    int32_t     code = translateTagVal(pCxt, pNode, &pVal);
+    int32_t     code = translateTagVal(pCxt, pTagSchema + index, pNode, &pVal);
     if (TSDB_CODE_SUCCESS == code) {
       if (NULL == pVal) {
         pVal = (SValueNode*)pNode;

source/libs/parser/test/parSelectTest.cpp

@@ -232,4 +232,12 @@ TEST_F(ParserSelectTest, semanticError) {
       PARSER_STAGE_TRANSLATE);
 }
 
+TEST_F(ParserSelectTest, setOperator) {
+  useDb("root", "test");
+
+  run("SELECT * FROM t1 UNION ALL SELECT * FROM t1");
+
+  run("(SELECT * FROM t1) UNION ALL (SELECT * FROM t1)");
+}
+
 }  // namespace ParserTest

source/libs/planner/src/planOptimizer.c

@@ -723,7 +723,10 @@ static int32_t opkGetScanNodesImpl(SLogicNode* pNode, bool* pNotOptimize, SNodeL
 
   switch (nodeType(pNode)) {
     case QUERY_NODE_LOGIC_PLAN_SCAN:
+      if (TSDB_SUPER_TABLE != ((SScanLogicNode*)pNode)->pMeta->tableType) {
         return nodesListMakeAppend(pScanNodes, pNode);
+      }
+      break;
     case QUERY_NODE_LOGIC_PLAN_JOIN:
       code = opkGetScanNodesImpl(nodesListGetNode(pNode->pChildren, 0), pNotOptimize, pScanNodes);
       if (TSDB_CODE_SUCCESS == code) {
@@ -739,6 +742,7 @@ static int32_t opkGetScanNodesImpl(SLogicNode* pNode, bool* pNotOptimize, SNodeL
 
   if (1 != LIST_LENGTH(pNode->pChildren)) {
     *pNotOptimize = true;
+    return TSDB_CODE_SUCCESS;
   }
 
   return opkGetScanNodesImpl(nodesListGetNode(pNode->pChildren, 0), pNotOptimize, pScanNodes);

source/libs/planner/test/planSubqueryTest.cpp

@@ -23,12 +23,15 @@ class PlanSubqeuryTest : public PlannerTestBase {};
 TEST_F(PlanSubqeuryTest, basic) {
   useDb("root", "test");
 
-  run("select * from (select * from t1)");
+  run("SELECT * FROM (SELECT * FROM t1)");
+
+  // run("SELECT LAST(c1) FROM ( SELECT * FROM t1)");
 }
 
 TEST_F(PlanSubqeuryTest, doubleGroupBy) {
   useDb("root", "test");
 
-  run("select count(*) from (select c1 + c3 a, c1 + count(*) b from t1 where c2 = 'abc' group by c1, c3) where a > 100 "
+  run("SELECT COUNT(*) FROM ("
-      "group by b");
+      "SELECT c1 + c3 a, c1 + COUNT(*) b FROM t1 WHERE c2 = 'abc' GROUP BY c1, c3) "
+      "WHERE a > 100 GROUP BY b");
 }