add concat_ws and concat/length test case

src/common/inc/texpr.h

@@ -62,7 +62,8 @@ struct SSchema;
 #define TSDB_FUNC_SCALAR_ROUND        (TSDB_FUNC_FLAG_SCALAR | 0x000C)
 #define TSDB_FUNC_SCALAR_CONCAT       (TSDB_FUNC_FLAG_SCALAR | 0x000D)
 #define TSDB_FUNC_SCALAR_LENGTH       (TSDB_FUNC_FLAG_SCALAR | 0x000E)
-#define TSDB_FUNC_SCALAR_MAX_NUM      15
+#define TSDB_FUNC_SCALAR_CONCAT_WS    (TSDB_FUNC_FLAG_SCALAR | 0x000F)
+#define TSDB_FUNC_SCALAR_MAX_NUM      16
 
 #define TSDB_FUNC_SCALAR_NAME_MAX_LEN 16
 

src/common/src/texpr.c

@@ -30,6 +30,7 @@
 
 static int32_t exprValidateMathNode(tExprNode *pExpr);
 static int32_t exprValidateStringConcatNode(tExprNode *pExpr);
+static int32_t exprValidateStringConcatWsNode(tExprNode *pExpr);
 static int32_t exprValidateStringLengthNode(tExprNode *pExpr);
 
 int32_t exprTreeValidateFunctionNode(tExprNode *pExpr) {
@@ -57,6 +58,9 @@ int32_t exprTreeValidateFunctionNode(tExprNode *pExpr) {
     case TSDB_FUNC_SCALAR_LENGTH: {
       return exprValidateStringLengthNode(pExpr);
     }
+    case TSDB_FUNC_SCALAR_CONCAT_WS: {
+      return exprValidateStringConcatWsNode(pExpr);
+    }
 
     default:
       break;
@@ -871,7 +875,7 @@ tExprNode* exprdup(tExprNode* pNode) {
 ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
 // scalar functions
 int32_t exprValidateStringConcatNode(tExprNode *pExpr) {
-  if (pExpr->_func.numChildren < 2) {
+  if (pExpr->_func.numChildren < 2 || pExpr->_func.numChildren > 8) {
     return TSDB_CODE_TSC_INVALID_OPERATION;
   }
 
@@ -942,6 +946,82 @@ int32_t exprValidateStringConcatNode(tExprNode *pExpr) {
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t exprValidateStringConcatWsNode(tExprNode *pExpr) {
+  if (pExpr->_func.numChildren < 3 || pExpr->_func.numChildren > 9) {
+    return TSDB_CODE_TSC_INVALID_OPERATION;
+  }
+
+  int16_t prevResultType = TSDB_DATA_TYPE_NULL;
+  int16_t resultType = TSDB_DATA_TYPE_NULL;
+  bool resultTypeDeduced = false;
+  for (int32_t i = 0; i < pExpr->_func.numChildren; ++i) {
+    tExprNode *child = pExpr->_func.pChildren[i];
+    if (child->nodeType != TSQL_NODE_VALUE) {
+      resultType = child->resultType;
+      if (!IS_VAR_DATA_TYPE(resultType)) {
+        return TSDB_CODE_TSC_INVALID_OPERATION;
+      }
+      if (!resultTypeDeduced) {
+        resultTypeDeduced = true;
+      } else {
+        if (resultType != prevResultType) {
+          return TSDB_CODE_TSC_INVALID_OPERATION;
+        }
+      }
+      prevResultType = child->resultType;
+    } else {
+      if (!IS_VAR_DATA_TYPE(child->resultType)) {
+        return TSDB_CODE_TSC_INVALID_OPERATION;
+      }
+    }
+  }
+
+  if (resultTypeDeduced) {
+    for (int32_t i = 0; i < pExpr->_func.numChildren; ++i) {
+      tExprNode *child = pExpr->_func.pChildren[i];
+      if (child->nodeType == TSQL_NODE_VALUE) {
+        if (!IS_VAR_DATA_TYPE(child->pVal->nType)) {
+          return TSDB_CODE_TSC_INVALID_OPERATION;
+        }
+        char* payload = malloc(child->pVal->nLen * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
+        tVariantDump(child->pVal, payload, resultType, true);
+        int16_t resultBytes = varDataTLen(payload);
+        free(payload);
+        child->resultType = resultType;
+        child->resultBytes = (int16_t)(resultBytes);
+      }
+    }
+  } else {
+    for (int32_t i = 0; i < pExpr->_func.numChildren; ++i) {
+      tExprNode *child = pExpr->_func.pChildren[i];
+      assert(child->nodeType == TSQL_NODE_VALUE) ;
+      resultType = child->resultType;
+      for (int j = i+1; j < pExpr->_func.numChildren; ++j) {
+        if (pExpr->_func.pChildren[j]->resultType != resultType) {
+          return TSDB_CODE_TSC_INVALID_OPERATION;
+        }
+      }
+    }
+  }
+
+  pExpr->resultType = resultType;
+  int16_t resultBytes = 0;
+  for (int32_t i = 1; i < pExpr->_func.numChildren; ++i) {
+    tExprNode *child = pExpr->_func.pChildren[i];
+    if (resultBytes <= resultBytes + child->resultBytes - VARSTR_HEADER_SIZE) {
+      resultBytes += child->resultBytes - VARSTR_HEADER_SIZE;
+    } else {
+      return TSDB_CODE_TSC_INVALID_OPERATION;
+    }
+  }
+  tExprNode* wsNode = pExpr->_func.pChildren[0];
+  int16_t wsResultBytes = wsNode->resultBytes - VARSTR_HEADER_SIZE;
+  resultBytes += wsResultBytes * (pExpr->_func.numChildren - 2);
+  pExpr->resultBytes = resultBytes + VARSTR_HEADER_SIZE;
+  return TSDB_CODE_SUCCESS;
+}
+
+
 int32_t exprValidateStringLengthNode(tExprNode *pExpr) {
   if (pExpr->_func.numChildren != 1) {
     return TSDB_CODE_TSC_INVALID_OPERATION;
@@ -1080,6 +1160,50 @@ void vectorConcat(int16_t functionId, tExprOperandInfo* pInputs, int32_t numInpu
   free(inputData);
 }
 
+void vectorConcatWs(int16_t functionId, tExprOperandInfo* pInputs, int32_t numInputs, tExprOperandInfo* pOutput, int32_t order) {
+  assert(functionId == TSDB_FUNC_SCALAR_CONCAT_WS && numInputs >=3 && order == TSDB_ORDER_ASC);
+  for (int i = 0; i < numInputs; ++i) {
+    assert(pInputs[i].numOfRows == 1 || pInputs[i].numOfRows == pOutput->numOfRows);
+  }
+
+  char* outputData = NULL;
+  char** inputData = calloc(numInputs, sizeof(char*));
+  for (int i = 0; i < pOutput->numOfRows; ++i) {
+    for (int j = 0; j < numInputs; ++j) {
+      if (pInputs[j].numOfRows == 1) {
+        inputData[j] = pInputs[j].data;
+      } else {
+        inputData[j] = pInputs[j].data + i * pInputs[j].bytes;
+      }
+    }
+
+    outputData = pOutput->data + i * pOutput->bytes;
+
+    bool hasNullInputs = false;
+    for (int j = 0; j < numInputs; ++j) {
+      if (isNull(inputData[j], pInputs[j].type)) {
+        hasNullInputs = true;
+        setNull(outputData, pOutput->type, pOutput->bytes);
+      }
+    }
+
+    if (!hasNullInputs) {
+      int16_t dataLen = 0;
+      for (int j = 1; j < numInputs; ++j) {
+        memcpy(((char*)varDataVal(outputData))+dataLen, varDataVal(inputData[j]), varDataLen(inputData[j]));
+        dataLen += varDataLen(inputData[j]);
+        if (j < numInputs - 1) {
+          memcpy(((char*)varDataVal(outputData))+dataLen, varDataVal(inputData[0]), varDataLen(inputData[0]));
+          dataLen += varDataLen(inputData[0]);
+        }
+      }
+      varDataSetLen(outputData, dataLen);
+    }
+  }
+
+  free(inputData);
+}
+
 void vectorLength(int16_t functionId, tExprOperandInfo *pInputs, int32_t numInputs, tExprOperandInfo* pOutput, int32_t order) {
   assert(functionId == TSDB_FUNC_SCALAR_LENGTH && numInputs == 1 && order == TSDB_ORDER_ASC);
 
@@ -1398,5 +1522,10 @@ tScalarFunctionInfo aScalarFunctions[] = {
         TSDB_FUNC_SCALAR_LENGTH,
         "length",
         vectorLength
+    },
+    {
+        TSDB_FUNC_SCALAR_CONCAT_WS,
+        "concat_ws",
+        vectorConcatWs
     }
 };

tests/script/fullGeneralSuite.sim

@@ -40,6 +40,7 @@ run general/compute/top.sim
 run general/compute/block_dist.sim
 run general/compute/scalar_pow.sim
 run general/compute/scalar_triangle.sim
+run general/compute/scalar_str_concat_len.sim
 run general/db/alter_option.sim
 run general/db/alter_tables_d2.sim
 run general/db/alter_tables_v1.sim

tests/script/general/compute/scalar_str_concat_len.sim

+system sh/stop_dnodes.sh
+
+system sh/deploy.sh -n dnode1 -i 1
+system sh/cfg.sh -n dnode1 -c walLevel -v 1
+system sh/exec.sh -n dnode1 -s start
+sleep 500
+sql connect
+
+$dbPrefix = db
+$tbPrefix = ct
+$mtPrefix = st
+$quote = '
+$tbNum = 2
+$rowNum = 50
+
+print =============== step1 create stable/table
+$i = 0
+$db = $dbPrefix . $i
+$mt = $mtPrefix . $i
+
+sql drop database $db -x step1
+step1:
+sql create database $db
+sql use $db
+sql create table $mt (ts timestamp, c1 int, c2 binary(10), c3 binary(30), c4 binary(40), c5 binary(50), c6 nchar(10), c7 nchar(20), c8 nchar(30), c9 nchar(40)) TAGS (tgcol int)
+$i = 0
+while $i < $tbNum
+  $tb = $tbPrefix . $i
+  sql create table $tb using $mt tags( $i )
+
+  $x = 0
+
+  $z2 = $x . 2 
+  $y2 = $quote . $z2 
+  $y2 = $y2 . $quote
+
+  $z3 = $x . 3 
+  $y3 = $quote . $z3 
+  $y3 = $y3 . $quote
+
+  $z4 = $x . 4 
+  $y4 = $quote . $z4
+  $y4 = $y4 . $quote
+
+  $z5 = $x . 5 
+  $y5 = $quote . $z5
+  $y5 = $y5 . $quote
+
+  $z6 = $x . 6 
+  $y6 = $quote . $z6
+  $y6 = $y6 . $quote
+
+  $z7 = $x . 7 
+  $y7 = $quote . $z7
+  $y7 = $y7 . $quote
+
+  $z8 = $x . 8 
+  $y8 = $quote . $z8
+  $y8 = $y8 . $quote
+
+  $z9 = $x . 9 
+  $y9 = $quote . $z9
+  $y9 = $y9 . $quote
+
+  while $x < $rowNum
+    $cc = $x * 60000
+    $ms = 1601481600000 + $cc
+    
+    sql insert into $tb values ($ms , $x , $y2 , $y3 , $y4 , $y5 , $y6 , $y7 , $y8 , $y9 )
+    $x = $x + 1
+					$z2 = $x . 2 
+					$y2 = $quote . $z2 
+					$y2 = $y2 . $quote
+
+					$z3 = $x . 3 
+					$y3 = $quote . $z3 
+					$y3 = $y3 . $quote
+
+					$z4 = $x . 4 
+					$y4 = $quote . $z4
+					$y4 = $y4 . $quote
+
+					$z5 = $x . 5 
+					$y5 = $quote . $z5
+					$y5 = $y5 . $quote
+
+					$z6 = $x . 6 
+					$y6 = $quote . $z6
+					$y6 = $y6 . $quote
+
+					$z7 = $x . 7 
+					$y7 = $quote . $z7
+					$y7 = $y7 . $quote
+
+					$z8 = $x . 8 
+					$y8 = $quote . $z8
+					$y8 = $y8 . $quote
+
+					$z9 = $x . 9 
+					$y9 = $quote . $z9
+					$y9 = $y9 . $quote
+  endw
+
+  $i = $i + 1
+endw
+
+print ================= step2
+$i = 1
+$tb = $tbPrefix . $i
+$stb = $mtPrefix . 0
+
+print sql select concat(c2, c3, c4, c5) from $tb
+sql select concat(c2, c3, c4, c5) from $tb
+print $data00
+if $data00 != 02030405 then
+  return -1
+endi
+
+print sql select concat_ws('data',c2,c3,c4,c5) from $tb
+sql select concat_ws('data',c2,c3,c4,c5) from $tb
+print $data00
+
+if $data00 != 02data03data04data05 then
+  return -1
+endi
+print sql select concat(c6, c7, c8, c9) from $tb
+sql select concat(c6, c7, c8, c9) from $tb
+print $data00
+if $data00 != 06070809 then
+  return -1
+endi
+
+print sql select concat_ws('data' ,c6,c7,c8,c9) from $tb
+sql select concat_ws('data' ,c6,c7,c8,c9) from $tb
+print $data00
+
+if $data00 != 06data07data08data09 then
+  return -1
+endi
+
+print sql select length(concat(c2, c3, c4, c5)) from $tb
+sql select length(concat(c2, c3, c4, c5)) from $tb
+print $data00
+if $data00 != 8 then
+  return -1
+endi
+
+print sql select length(concat_ws('data',c2,c3,c4,c5)) from $tb
+sql select length(concat_ws('data',c2,c3,c4,c5)) from $tb
+print $data00
+
+if $data00 != 20 then
+  return -1
+endi
+print sql select length(concat(c6, c7, c8, c9)) from $tb
+sql select length(concat(c6, c7, c8, c9)) from $tb
+print $data00
+if $data00 != 32 then
+  return -1
+endi
+
+print sql select length(concat_ws('data' ,c6,c7,c8,c9)) from $tb
+sql select length(concat_ws('data' ,c6,c7,c8,c9)) from $tb
+print $data00
+
+if $data00 != 80 then
+  return -1
+endi
+
+print sql_error select concat(c1, c2, c3, c4, c5) from $tb
+sql_error select concat(c1, c2, c3, c4, c5) from $tb
+print sql_error select concat_ws('data',c1,c2,c3,c4,c5) from $tb
+sql_error select concat_ws('data',c1,c2,c3,c4,c5) from $tb
+
+print ===============> step 3 sql_error stable, group by, window
+sql_error select concat(c2) from $stb group by tbname;
+
+sql_error select concat(c2) from $stb group by tgcol;
+
+sql_error select concat(c2) from $stb group by c3;
+
+sql_error select concat(c2) from $stb interval(1m);
+
+sql_error select concat(c2) from $stb state_window(c7);
+
+sql_error select concat(c2) from $tb state_window(c7);
+
+sql_error select concat(c2) from $stb session(ts, 30s);
+
+sql_error select concat(c2) from $tb session(ts, 30s);
+
+sql_error select concat(c2) from $stb slimit 2;
+
+sql_error select concat(c2) from $stb interval(1m) slimit 2;
+
+sql_error select length(c2) from $stb group by tbname;
+
+sql_error select length(c2) from $stb group by tgcol;
+
+sql_error select length(c2) from $stb group by c3;
+
+sql_error select length(c2) from $stb interval(1m);
+
+sql_error select length(c2) from $stb state_window(c7);
+
+sql_error select length(c2) from $tb state_window(c7);
+
+sql_error select length(c2) from $stb session(ts, 30s);
+
+sql_error select length(c2) from $tb session(ts, 30s);
+
+sql_error select length(c2) from $stb slimit 2;
+
+sql_error select length(c2) from $stb interval(1m) slimit 2;
+
+sql_error select concat_ws(c2) from $stb group by tbname;
+
+sql_error select concat_ws(c2) from $stb group by tgcol;
+
+sql_error select concat_ws(c2) from $stb group by c3;
+
+sql_error select concat_ws(c2) from $stb interval(1m);
+
+sql_error select concat_ws(c2) from $stb state_window(c7);
+
+sql_error select concat_ws(c2) from $tb state_window(c7);
+
+sql_error select concat_ws(c2) from $stb session(ts, 30s);
+
+sql_error select concat_ws(c2) from $tb session(ts, 30s);
+
+sql_error select concat_ws(c2) from $stb slimit 2;
+
+sql_error select concat_ws(c2) from $stb interval(1m) slimit 2;
+print =============== clear
+#sql drop database $db
+#sql show databases
+#if $rows != 0 then 
+#  return -1
+#endi
+
+#system sh/exec.sh -n dnode1 -s stop -x SIGINT
+

tests/script/general/compute/testSuite.sim

@@ -22,3 +22,4 @@ run general/compute/top.sim
 run general/compute/block_dist.sim
 run general/compute/scalar_pow.sim
 run general/compute/scalar_triangle.sim
+run general/compute/scalar_str_concat_len.sim

tests/script/regressionSuite.sim

@@ -40,6 +40,7 @@ run general/compute/top.sim
 run general/compute/block_dist.sim
 run general/compute/scalar_pow.sim
 run general/compute/scalar_triangle.sim
+run general/compute/scalar_str_concat_len.sim
 run general/db/alter_option.sim
 run general/db/alter_tables_d2.sim
 run general/db/alter_tables_v1.sim