refactor:modify schemaless function to speed

source/client/test/smlTest.cpp

@@ -190,17 +190,21 @@ TEST(testCase, smlParseCols_Error_Test) {
     "c=-3.402823466e+39u64",
     "c=-339u64",
     "c=18446744073709551616u64",
+    "c=1,c=2"
   };
 
+  SHashObj *dumplicateKey = taosHashInit(32, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
   for(int i = 0; i < sizeof(data)/sizeof(data[0]); i++){
     char       msg[256] = {0};
     SSmlMsgBuf msgBuf;
     msgBuf.buf = msg;
     msgBuf.len = 256;
     int32_t len = strlen(data[i]);
-    int32_t ret = smlParseCols(data[i], len, NULL, false, &msgBuf);
+    int32_t ret = smlParseCols(data[i], len, NULL, false, dumplicateKey, &msgBuf);
     ASSERT_NE(ret, TSDB_CODE_SUCCESS);
+    taosHashClear(dumplicateKey);
   }
+  taosHashCleanup(dumplicateKey);
 }
 
 TEST(testCase, smlParseCols_tag_Test) {
@@ -211,11 +215,12 @@ TEST(testCase, smlParseCols_tag_Test) {
 
   SArray *cols = taosArrayInit(16, POINTER_BYTES);
   ASSERT_NE(cols, NULL);
+  SHashObj *dumplicateKey = taosHashInit(32, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
 
   const char *data =
       "cbin=\"passit hello,c=2\",cnch=L\"iisdfsf\",cbool=false,cf64=4.31f64,cf32_=8.32,cf32=8.23f32,ci8=-34i8,cu8=89u8,ci16=233i16,cu16=898u16,ci32=98289i32,cu32=12323u32,ci64=-89238i64,ci=989i,cu64=8989323u64,cbooltrue=true,cboolt=t,cboolf=f,cnch_=l\"iuwq\"";
   int32_t len = strlen(data);
-  int32_t ret = smlParseCols(data, len, cols, true, &msgBuf);
+  int32_t ret = smlParseCols(data, len, cols, true, dumplicateKey, &msgBuf);
   ASSERT_EQ(ret, TSDB_CODE_SUCCESS);
   int32_t size = taosArrayGetSize(cols);
   ASSERT_EQ(size, 19);
@@ -239,10 +244,14 @@ TEST(testCase, smlParseCols_tag_Test) {
   taosMemoryFree(kv);
 
   taosArrayClear(cols);
+
+
+  // test tag is null
   data = "t=3e";
   len = 0;
   memset(msgBuf.buf, 0, msgBuf.len);
-  ret = smlParseCols(data, len, cols, true, &msgBuf);
+  taosHashClear(dumplicateKey);
+  ret = smlParseCols(data, len, cols, true, dumplicateKey, &msgBuf);
   ASSERT_EQ(ret, TSDB_CODE_SUCCESS);
   size = taosArrayGetSize(cols);
   ASSERT_EQ(size, 1);
@@ -255,6 +264,9 @@ TEST(testCase, smlParseCols_tag_Test) {
   ASSERT_EQ(kv->valueLen, strlen(TAG));
   ASSERT_EQ(strncasecmp(kv->value, TAG, strlen(TAG)), 0);
   taosMemoryFree(kv);
+
+  taosArrayDestroy(cols);
+  taosHashCleanup(dumplicateKey);
 }
 
 TEST(testCase, smlParseCols_Test) {
@@ -266,9 +278,11 @@ TEST(testCase, smlParseCols_Test) {
   SArray *cols = taosArrayInit(16, POINTER_BYTES);
   ASSERT_NE(cols, NULL);
 
+  SHashObj *dumplicateKey = taosHashInit(32, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
+
   const char *data = "cbin=\"passit hello,c=2\",cnch=L\"iisdfsf\",cbool=false,cf64=4.31f64,cf32_=8.32,cf32=8.23f32,ci8=-34i8,cu8=89u8,ci16=233i16,cu16=898u16,ci32=98289i32,cu32=12323u32,ci64=-89238i64,ci=989i,cu64=8989323u64,cbooltrue=true,cboolt=t,cboolf=f,cnch_=l\"iuwq\"";
   int32_t len = strlen(data);
-  int32_t ret = smlParseCols(data, len, cols, false, &msgBuf);
+  int32_t ret = smlParseCols(data, len, cols, false, dumplicateKey, &msgBuf);
   ASSERT_EQ(ret, TSDB_CODE_SUCCESS);
   int32_t size = taosArrayGetSize(cols);
   ASSERT_EQ(size, 19);
@@ -450,6 +464,7 @@ TEST(testCase, smlParseCols_Test) {
   taosMemoryFree(kv);
 
   taosArrayDestroy(cols);
+  taosHashCleanup(dumplicateKey);
 }
 
 TEST(testCase, smlParseLine_Test) {
@@ -485,6 +500,30 @@ TEST(testCase, smlParseLine_Test) {
 //  }
 }
 
+TEST(testCase, smlParseLine_error_Test) {
+  TAOS *taos = taos_connect("localhost", "root", "taosdata", NULL, 0);
+  ASSERT_NE(taos, NULL);
+
+  TAOS_RES* pRes = taos_query(taos, "create database if not exists sml_db");
+  taos_free_result(pRes);
+
+  pRes = taos_query(taos, "use sml_db");
+  taos_free_result(pRes);
+
+  SRequestObj *request = createRequest(taos, NULL, NULL, TSDB_SQL_INSERT);
+  ASSERT_NE(request, NULL);
+
+  SSmlHandle *info = smlBuildSmlInfo(taos, request, TSDB_SML_LINE_PROTOCOL, TSDB_SML_TIMESTAMP_NANO_SECONDS, true);
+  ASSERT_NE(info, NULL);
+
+  const char *sql[2] = {
+      "measure,t1=3 c1=8",
+      "measure,t2=3 c1=8u8"
+  };
+  int ret = smlInsertLines(info, sql, 2);
+  ASSERT_NE(ret, 0);
+}
+
 // TEST(testCase, smlParseTS_Test) {
 //   char       msg[256] = {0};
 //   SSmlMsgBuf msgBuf;

source/libs/parser/src/parInsert.c

@@ -1549,7 +1549,7 @@ typedef struct SmlExecHandle {
   SQuery* pQuery;
 } SSmlExecHandle;
 
-static int32_t smlBoundColumns(SArray *cols, SParsedDataColInfo* pColList, SSchema* pSchema) {
+static int32_t smlBoundColumnData(SArray *cols, SParsedDataColInfo* pColList, SSchema* pSchema) {
   col_id_t nCols = pColList->numOfCols;
 
   pColList->numOfBound = 0;
@@ -1620,7 +1620,7 @@ static int32_t smlBoundColumns(SArray *cols, SParsedDataColInfo* pColList, SSche
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t smlBoundTags(SArray *cols, SKVRowBuilder *tagsBuilder, SParsedDataColInfo* tags, SSchema* pSchema, SKVRow *row, SMsgBuf *msg) {
+static int32_t smlBuildTagRow(SArray *cols, SKVRowBuilder *tagsBuilder, SParsedDataColInfo* tags, SSchema* pSchema, SKVRow *row, SMsgBuf *msg) {
   if (tdInitKVRowBuilder(tagsBuilder) < 0) {
     return TSDB_CODE_TSC_OUT_OF_MEMORY;
   }
@@ -1649,13 +1649,13 @@ int32_t smlBindData(void *handle, SArray *tags, SArray *colsFormat, SArray *cols
   SSmlExecHandle *smlHandle = (SSmlExecHandle *)handle;
   SSchema* pTagsSchema = getTableTagSchema(pTableMeta);
   setBoundColumnInfo(&smlHandle->tags, pTagsSchema, getNumOfTags(pTableMeta));
-  int ret = smlBoundColumns(tags, &smlHandle->tags, pTagsSchema);
+  int ret = smlBoundColumnData(tags, &smlHandle->tags, pTagsSchema);
   if(ret != TSDB_CODE_SUCCESS){
     buildInvalidOperationMsg(&pBuf, "bound tags error");
     return ret;
   }
   SKVRow row = NULL;
-  ret = smlBoundTags(tags, &smlHandle->tagsBuilder, &smlHandle->tags, pTagsSchema, &row, &pBuf);
+  ret = smlBuildTagRow(tags, &smlHandle->tagsBuilder, &smlHandle->tags, pTagsSchema, &row, &pBuf);
   if(ret != TSDB_CODE_SUCCESS){
     return ret;
   }
@@ -1673,7 +1673,7 @@ int32_t smlBindData(void *handle, SArray *tags, SArray *colsFormat, SArray *cols
 
   SSchema* pSchema = getTableColumnSchema(pTableMeta);
 
-  ret = smlBoundColumns(colsSchema, &pDataBlock->boundColumnInfo, pSchema);
+  ret = smlBoundColumnData(colsSchema, &pDataBlock->boundColumnInfo, pSchema);
   if(ret != TSDB_CODE_SUCCESS){
     buildInvalidOperationMsg(&pBuf, "bound cols error");
     return ret;
@@ -1698,10 +1698,10 @@ int32_t smlBindData(void *handle, SArray *tags, SArray *colsFormat, SArray *cols
     STSRow* row = (STSRow*)(pDataBlock->pData + pDataBlock->size);  // skip the SSubmitBlk header
     tdSRowResetBuf(pBuilder, row);
     void *rowData = NULL;
-    bool eleEqual = false;
+    size_t rowDataSize = 0;
     if(format){
       rowData = taosArrayGetP(colsFormat, r);
-      eleEqual = (taosArrayGetSize(rowData) == spd->numOfBound);
+      rowDataSize = taosArrayGetSize(rowData);
     }else{
       rowData = taosArrayGetP(cols, r);
     }
@@ -1715,9 +1715,12 @@ int32_t smlBindData(void *handle, SArray *tags, SArray *colsFormat, SArray *cols
 
       SSmlKv *kv = NULL;
       if(format){
-        kv = taosArrayGetP(rowData, c);
         do{
-          if (!eleEqual && kv && (kv->keyLen != strlen(pColSchema->name) || strncmp(kv->key, pColSchema->name, kv->keyLen) != 0)){
+          if(rowDataSize >= c){
+            break;
+          }
+          kv = taosArrayGetP(rowData, c);
+          if (rowDataSize != spd->numOfBound && kv && (kv->keyLen != strlen(pColSchema->name) || strncmp(kv->key, pColSchema->name, kv->keyLen) != 0)){
             MemRowAppend(&pBuf, NULL, 0, &param);
             c++;
             if(c >= spd->numOfBound) break;