From 7cb90161378fcc3c527e381a26776c40492e038b Mon Sep 17 00:00:00 2001
From: wangmm0220 <wangmm0220@gmail.com>
Date: Fri, 15 Oct 2021 10:53:14 +0800
Subject: [PATCH] TD-6129<feature> match/nmatch support for nchar

---
 src/client/src/tscSQLParser.c |  8 +-------
 src/query/src/qFilter.c       | 23 +++++++++++++++++++++--
 src/tsdb/src/tsdbRead.c       |  3 ++-
 3 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/src/client/src/tscSQLParser.c b/src/client/src/tscSQLParser.c
index 942ff603dc..0948d013ac 100644
--- a/src/client/src/tscSQLParser.c
+++ b/src/client/src/tscSQLParser.c
@@ -4453,7 +4453,6 @@ static int32_t validateMatchExpr(tSqlExpr* pExpr, STableMeta* pTableMeta, int32_
   const char* msg1 = "regular expression string should be less than %d characters";
   const char* msg2 = "illegal column type for match/nmatch";
   const char* msg3 = "invalid regular expression";
-  const char* msg4 = "json column type must encode by binary for match/nmatch";
 
   tSqlExpr* pLeft  = pExpr->pLeft;
   tSqlExpr* pRight = pExpr->pRight;
@@ -4466,15 +4465,10 @@ static int32_t validateMatchExpr(tSqlExpr* pExpr, STableMeta* pTableMeta, int32_
     }
 
     SSchema* pSchema = tscGetTableSchema(pTableMeta);
-    if ((pLeft->tokenId != TK_ARROW) && (!isTablenameToken(&pLeft->columnName))
-        && (pSchema[index].type != TSDB_DATA_TYPE_BINARY)) {
+    if ((!isTablenameToken(&pLeft->columnName)) && (pSchema[index].type != TSDB_DATA_TYPE_BINARY)) {
       return invalidOperationMsg(msgBuf, msg2);
     }
 
-    if(pLeft->tokenId == TK_ARROW && pSchema[index].type == TSDB_DATA_TYPE_JSON){
-      return invalidOperationMsg(msgBuf, msg4);
-    }
-
     if (!(pRight->type == SQL_NODE_VALUE && pRight->value.nType == TSDB_DATA_TYPE_BINARY)) {
       return invalidOperationMsg(msgBuf, msg3);
     }
diff --git a/src/query/src/qFilter.c b/src/query/src/qFilter.c
index fdd9ef59cb..a132ffe536 100644
--- a/src/query/src/qFilter.c
+++ b/src/query/src/qFilter.c
@@ -2989,8 +2989,18 @@ bool filterExecuteImplMisc(void *pinfo, int32_t numOfRows, int8_t** p, SDataStat
       all = false;
       continue;
     }
+    // match nchar
 
-    (*p)[i] = filterDoCompare(gDataCompare[info->cunits[uidx].func], info->cunits[uidx].optr, colData, info->cunits[uidx].valData);
+    if(info->cunits[uidx].dataType == TSDB_DATA_TYPE_NCHAR && (info->cunits[uidx].optr == TSDB_RELATION_MATCH || info->cunits[uidx].optr == TSDB_RELATION_NMATCH)){
+      char newColData[TSDB_REGEX_STRING_DEFAULT_LEN * TSDB_NCHAR_SIZE] = {0};
+      char newValData[TSDB_REGEX_STRING_DEFAULT_LEN * TSDB_NCHAR_SIZE] = {0};
+      taosUcs4ToMbs(varDataVal(colData), varDataLen(colData), newColData);
+      taosUcs4ToMbs(varDataVal(info->cunits[uidx].valData), varDataLen(info->cunits[uidx].valData), newValData);
+
+      (*p)[i] = filterDoCompare(gDataCompare[info->cunits[uidx].func], info->cunits[uidx].optr, newColData, newValData);
+    }else{
+      (*p)[i] = filterDoCompare(gDataCompare[info->cunits[uidx].func], info->cunits[uidx].optr, colData, info->cunits[uidx].valData);
+    }
             
     if ((*p)[i] == 0) {
       all = false;
@@ -3038,7 +3048,16 @@ bool filterExecuteImpl(void *pinfo, int32_t numOfRows, int8_t** p, SDataStatis *
             } else if (cunit->rfunc >= 0) {
               (*p)[i] = (*gRangeCompare[cunit->rfunc])(colData, colData, cunit->valData, cunit->valData2, gDataCompare[cunit->func]);
             } else {
-              (*p)[i] = filterDoCompare(gDataCompare[cunit->func], cunit->optr, colData, cunit->valData);
+              if(cunit->dataType == TSDB_DATA_TYPE_NCHAR && (cunit->optr == TSDB_RELATION_MATCH || cunit->optr == TSDB_RELATION_NMATCH)){
+                char newColData[TSDB_REGEX_STRING_DEFAULT_LEN * TSDB_NCHAR_SIZE] = {0};
+                char newValData[TSDB_REGEX_STRING_DEFAULT_LEN * TSDB_NCHAR_SIZE] = {0};
+                taosUcs4ToMbs(varDataVal(colData), varDataLen(colData), newColData);
+                taosUcs4ToMbs(varDataVal(cunit->valData), varDataLen(cunit->valData), newValData);
+
+                (*p)[i] = filterDoCompare(gDataCompare[cunit->func], cunit->optr, newColData, newValData);
+              }else{
+                (*p)[i] = filterDoCompare(gDataCompare[cunit->func], cunit->optr, colData, cunit->valData);
+              }
             }
           
           //FILTER_UNIT_SET_R(info, uidx, p[i]);
diff --git a/src/tsdb/src/tsdbRead.c b/src/tsdb/src/tsdbRead.c
index 32ce603366..1e230dad49 100644
--- a/src/tsdb/src/tsdbRead.c
+++ b/src/tsdb/src/tsdbRead.c
@@ -4088,7 +4088,8 @@ static FORCE_INLINE int32_t tsdbGetJsonTagDataFromId(void *param, int32_t id, ch
     *data = TABLE_NAME(pTable);
   } else {
     void* jsonData = getJsonTagValue(pTable, name, TSDB_MAX_JSON_KEY_MD5_LEN, NULL);
-    if (jsonData != NULL) jsonData += CHAR_BYTES;   // jump type
+    assert(jsonData != NULL);
+    jsonData += CHAR_BYTES;   // jump type
     *data = jsonData;
   }
 
-- 
GitLab