Merge pull request #20603 from taosdata/enh/3.0_planner_optimize

fix: udf error check

source/libs/parser/src/parInsertSql.c

@@ -1492,6 +1492,7 @@ static void resetEnvPreTable(SInsertParseContext* pCxt, SVnodeModifyOpStmt* pStm
   pStmt->pBoundCols = NULL;
   pStmt->usingTableProcessing = false;
   pStmt->fileProcessing = false;
+  pStmt->usingTableName.type = 0;
 }
 
 // input pStmt->pSql: [(field1_name, ...)] [ USING ... ] VALUES ... | FILE ...
@@ -1539,6 +1540,10 @@ static int32_t checkTableClauseFirstToken(SInsertParseContext* pCxt, SVnodeModif
     }
   }
 
+  if (TK_NK_ID != pTbName->type && TK_NK_STRING != pTbName->type && TK_NK_QUESTION != pTbName->type) {
+    return buildSyntaxErrMsg(&pCxt->msg, "table_name is expected", pTbName->z);
+  }
+
   *pHasData = true;
   return TSDB_CODE_SUCCESS;
 }

source/libs/parser/src/parTranslater.c

@@ -6422,14 +6422,16 @@ static int32_t translateCreateFunction(STranslateContext* pCxt, SCreateFunctionS
     return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_FUNCTION_NAME);
   }
 
-  if (TSDB_DATA_TYPE_JSON == pStmt->outputDt.type ||
-      TSDB_DATA_TYPE_VARBINARY == pStmt->outputDt.type ||
-      TSDB_DATA_TYPE_DECIMAL == pStmt->outputDt.type ||
-      TSDB_DATA_TYPE_BLOB == pStmt->outputDt.type ||
+  if (TSDB_DATA_TYPE_JSON == pStmt->outputDt.type || TSDB_DATA_TYPE_VARBINARY == pStmt->outputDt.type ||
+      TSDB_DATA_TYPE_DECIMAL == pStmt->outputDt.type || TSDB_DATA_TYPE_BLOB == pStmt->outputDt.type ||
       TSDB_DATA_TYPE_MEDIUMBLOB == pStmt->outputDt.type) {
     return generateSyntaxErrMsgExt(&pCxt->msgBuf, TSDB_CODE_PAR_SYNTAX_ERROR, "Unsupported output type for UDF");
   }
 
+  if (!pStmt->isAgg && pStmt->bufSize > 0) {
+    return generateSyntaxErrMsgExt(&pCxt->msgBuf, TSDB_CODE_PAR_SYNTAX_ERROR, "BUFSIZE can only be used with UDAF");
+  }
+
   SCreateFuncReq req = {0};
   strcpy(req.name, pStmt->funcName);
   req.igExists = pStmt->ignoreExists;

tests/script/tsim/query/udf.sim

@@ -23,10 +23,10 @@ if $system_content == Windows_NT then
   return 0;
 endi
 if $system_content == Windows_NT then
-  sql create function bit_and as 'C:\\Windows\\Temp\\bitand.dll' outputtype int bufSize 8;
+  sql create function bit_and as 'C:\\Windows\\Temp\\bitand.dll' outputtype int;
   sql create aggregate function l2norm as 'C:\\Windows\\Temp\\l2norm.dll' outputtype double bufSize 8;
 else
-  sql create function bit_and as '/tmp/udf/libbitand.so' outputtype int bufSize 8;
+  sql create function bit_and as '/tmp/udf/libbitand.so' outputtype int;
   sql create aggregate function l2norm as '/tmp/udf/libl2norm.so' outputtype double bufSize 8;
 endi
 

tests/script/tsim/query/udf_with_const.sim

@@ -23,9 +23,9 @@ if $system_content == Windows_NT then
 endi
 
 if $system_content == Windows_NT then
-  sql create function gpd as 'C:\\Windows\\Temp\\gpd.dll' outputtype int bufSize 8;
+  sql create function gpd as 'C:\\Windows\\Temp\\gpd.dll' outputtype int;
 else
-  sql create function gpd as '/tmp/udf/libgpd.so' outputtype int bufSize 8;
+  sql create function gpd as '/tmp/udf/libgpd.so' outputtype int;
 endi
 sql show functions;
 if $rows != 1 then

tests/script/tsim/query/udfpy.sim

@@ -29,10 +29,10 @@ if $system_content == Windows_NT then
   return 0;
 endi
 if $system_content == Windows_NT then
-  sql create function bit_and as 'C:\\Windows\\Temp\\bitand.dll' outputtype int bufSize 8;
+  sql create function bit_and as 'C:\\Windows\\Temp\\bitand.dll' outputtype int;
   sql create aggregate function l2norm as 'C:\\Windows\\Temp\\l2norm.dll' outputtype double bufSize 8;
 else
-  sql create function bit_and as '/tmp/udf/libbitand.so' outputtype int bufSize 8;
+  sql create function bit_and as '/tmp/udf/libbitand.so' outputtype int;
   sql create aggregate function l2norm as '/tmp/udf/libl2norm.so' outputtype double bufSize 8;
 endi
 sql create function pybitand as '/tmp/pyudf/pybitand.py' outputtype int language 'python';

tests/script/tsim/valgrind/checkUdf.sim

@@ -15,10 +15,10 @@ sql insert into t values(now, 1)(now+1s, 2);
 
 system_content printf %OS%
 if $system_content == Windows_NT then
-  sql create function udf1 as 'C:\\Windows\\Temp\\udf1.dll' outputtype int bufSize 8;
+  sql create function udf1 as 'C:\\Windows\\Temp\\udf1.dll' outputtype int;
   sql create aggregate function udf2 as 'C:\\Windows\\Temp\\udf2.dll' outputtype double bufSize 8;
 else
-  sql create function udf1 as '/tmp/udf/libudf1.so' outputtype int bufSize 8;
+  sql create function udf1 as '/tmp/udf/libudf1.so' outputtype int;
   sql create aggregate function udf2 as '/tmp/udf/libudf2.so' outputtype double bufSize 8;
 endi
 sql show functions;

tests/system-test/0-others/udfTest.py

@@ -146,7 +146,7 @@ class TDTestCase:
 
         for i in range(5):
             # create  scalar functions
-            tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+            tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
             # create aggregate functions
 
@@ -171,7 +171,7 @@ class TDTestCase:
                 tdLog.info("drop two udf functions success ")
 
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         # create aggregate functions
 
@@ -562,7 +562,7 @@ class TDTestCase:
 
         # create function without buffer
         tdSql.execute("create  aggregate function udf1 as '%s' outputtype int bufSize 8 "%self.libudf1)
-        tdSql.execute("create  function udf2 as '%s' outputtype double bufSize 8"%self.libudf2)
+        tdSql.execute("create  function udf2 as '%s' outputtype double"%self.libudf2)
         udf1_sqls ,udf2_sqls = self.try_query_sql()
 
         for scalar_sql in udf1_sqls:
@@ -570,7 +570,7 @@ class TDTestCase:
         for aggregate_sql in udf2_sqls:
             tdSql.error(aggregate_sql)
 
-        tdSql.execute(" create function db as '%s' outputtype int bufSize 8 "%self.libudf1)
+        tdSql.execute(" create function db as '%s' outputtype int "%self.libudf1)
         tdSql.execute(" create aggregate function test as '%s' outputtype int bufSize 8 "%self.libudf1)
         tdSql.error(" select db(c1) from stb1 ")
         tdSql.error(" select db(c1,c6), db(c6) from stb1 ")
@@ -619,9 +619,9 @@ class TDTestCase:
         tdLog.info(" create function name is not build_in functions ")
         tdSql.execute(" drop function udf1 ")
         tdSql.execute(" drop function udf2 ")
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int "%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int "%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function tbname as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function function as '%s' outputtype double bufSize 8"%self.libudf2)

tests/system-test/0-others/udf_cfg1.py

@@ -148,7 +148,7 @@ class TDTestCase:
 
         for i in range(5):
             # create  scalar functions
-            tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+            tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
             # create aggregate functions
 
@@ -173,7 +173,7 @@ class TDTestCase:
                 tdLog.info("drop two udf functions success ")
 
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         # create aggregate functions
 

tests/system-test/0-others/udf_cfg2.py

@@ -148,7 +148,7 @@ class TDTestCase:
 
         for i in range(5):
             # create  scalar functions
-            tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+            tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
             # create aggregate functions
 
@@ -173,7 +173,7 @@ class TDTestCase:
                 tdLog.info("drop two udf functions success ")
 
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         # create aggregate functions
 
@@ -564,7 +564,7 @@ class TDTestCase:
 
         # create function without buffer
         tdSql.execute("create  aggregate function udf1 as '%s' outputtype int bufSize 8 "%self.libudf1)
-        tdSql.execute("create  function udf2 as '%s' outputtype double bufSize 8"%self.libudf2)
+        tdSql.execute("create  function udf2 as '%s' outputtype double"%self.libudf2)
         udf1_sqls ,udf2_sqls = self.try_query_sql()
 
         for scalar_sql in udf1_sqls:
@@ -572,7 +572,7 @@ class TDTestCase:
         for aggregate_sql in udf2_sqls:
             tdSql.error(aggregate_sql)
 
-        tdSql.execute(" create function db as '%s' outputtype int bufSize 8 "%self.libudf1)
+        tdSql.execute(" create function db as '%s' outputtype int "%self.libudf1)
         tdSql.execute(" create aggregate function test as '%s' outputtype int bufSize 8 "%self.libudf1)
         tdSql.error(" select db(c1) from stb1 ")
         tdSql.error(" select db(c1,c6), db(c6) from stb1 ")
@@ -621,9 +621,9 @@ class TDTestCase:
         tdLog.info(" create function name is not build_in functions ")
         tdSql.execute(" drop function udf1 ")
         tdSql.execute(" drop function udf2 ")
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int"%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int"%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function tbname as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function function as '%s' outputtype double bufSize 8"%self.libudf2)

tests/system-test/0-others/udf_create.py

@@ -148,7 +148,7 @@ class TDTestCase:
 
         for i in range(5):
             # create  scalar functions
-            tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+            tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
             # create aggregate functions
 
@@ -173,7 +173,7 @@ class TDTestCase:
                 tdLog.info("drop two udf functions success ")
 
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         # create aggregate functions
 
@@ -574,7 +574,7 @@ class TDTestCase:
 
         # create function without buffer
         tdSql.execute("create  aggregate function udf1 as '%s' outputtype int bufSize 8 "%self.libudf1)
-        tdSql.execute("create  function udf2 as '%s' outputtype double bufSize 8"%self.libudf2)
+        tdSql.execute("create  function udf2 as '%s' outputtype double "%self.libudf2)
         udf1_sqls ,udf2_sqls = self.try_query_sql()
 
         for scalar_sql in udf1_sqls:
@@ -582,7 +582,7 @@ class TDTestCase:
         for aggregate_sql in udf2_sqls:
             tdSql.error(aggregate_sql)
 
-        tdSql.execute(" create function db as '%s' outputtype int bufSize 8 "%self.libudf1)
+        tdSql.execute(" create function db as '%s' outputtype int "%self.libudf1)
         tdSql.execute(" create aggregate function test as '%s' outputtype int bufSize 8 "%self.libudf1)
         tdSql.error(" select db(c1) from stb1 ")
         tdSql.error(" select db(c1,c6), db(c6) from stb1 ")
@@ -631,9 +631,9 @@ class TDTestCase:
         tdLog.info(" create function name is not build_in functions ")
         tdSql.execute(" drop function udf1 ")
         tdSql.execute(" drop function udf2 ")
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int"%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int"%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function tbname as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function function as '%s' outputtype double bufSize 8"%self.libudf2)

tests/system-test/0-others/udf_restart_taosd.py

@@ -145,7 +145,7 @@ class TDTestCase:
 
         for i in range(5):
             # create  scalar functions
-            tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+            tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
             # create aggregate functions
 
@@ -170,7 +170,7 @@ class TDTestCase:
                 tdLog.info("drop two udf functions success ")
 
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         # create aggregate functions
 
@@ -561,7 +561,7 @@ class TDTestCase:
 
         # create function without buffer
         tdSql.execute("create  aggregate function udf1 as '%s' outputtype int bufSize 8 "%self.libudf1)
-        tdSql.execute("create  function udf2 as '%s' outputtype double bufSize 8"%self.libudf2)
+        tdSql.execute("create  function udf2 as '%s' outputtype double"%self.libudf2)
         udf1_sqls ,udf2_sqls = self.try_query_sql()
 
         for scalar_sql in udf1_sqls:
@@ -569,7 +569,7 @@ class TDTestCase:
         for aggregate_sql in udf2_sqls:
             tdSql.error(aggregate_sql)
 
-        tdSql.execute(" create function db as '%s' outputtype int bufSize 8 "%self.libudf1)
+        tdSql.execute(" create function db as '%s' outputtype int "%self.libudf1)
         tdSql.execute(" create aggregate function test as '%s' outputtype int bufSize 8 "%self.libudf1)
         tdSql.error(" select db(c1) from stb1 ")
         tdSql.error(" select db(c1,c6), db(c6) from stb1 ")
@@ -618,9 +618,9 @@ class TDTestCase:
         tdLog.info(" create function name is not build_in functions ")
         tdSql.execute(" drop function udf1 ")
         tdSql.execute(" drop function udf2 ")
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int"%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
-        tdSql.error("create function max as '%s' outputtype int bufSize 8"%self.libudf1)
+        tdSql.error("create function max as '%s' outputtype int"%self.libudf1)
         tdSql.error("create aggregate function sum as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function tbname as '%s' outputtype double bufSize 8"%self.libudf2)
         tdSql.error("create aggregate function function as '%s' outputtype double bufSize 8"%self.libudf2)

tests/system-test/7-tmq/tmqUdf-multCtb-snapshot0.py

@@ -50,7 +50,7 @@ class TDTestCase:
 
     def create_udf_function(self):
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         functions = tdSql.getResult("show functions")
         function_nums = len(functions)

tests/system-test/7-tmq/tmqUdf-multCtb-snapshot1.py

@@ -50,7 +50,7 @@ class TDTestCase:
 
     def create_udf_function(self):
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         functions = tdSql.getResult("show functions")
         function_nums = len(functions)

tests/system-test/7-tmq/tmqUdf.py

@@ -50,7 +50,7 @@ class TDTestCase:
 
     def create_udf_function(self):
         # create  scalar functions
-        tdSql.execute("create function udf1 as '%s' outputtype int bufSize 8;"%self.libudf1)
+        tdSql.execute("create function udf1 as '%s' outputtype int;"%self.libudf1)
 
         functions = tdSql.getResult("show functions")
         function_nums = len(functions)