xxq250
/
homework-jianmu
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

TD-13338 SELECT statement translate code

This commit is contained in:
Xiaoyu Wang 2022-02-08 05:01:30 -05:00
parent 69d196815b
commit 090250f994
6 changed files with 192 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
include/nodes/nodes.h
View File

@ -121,6 +121,14 @@ typedef struct SColumnNode {
typedef struct SValueNode {
SExprNode node; // QUERY_NODE_VALUE
char* literal;
bool isDuration;
union {
bool b;
int64_t i;
uint64_t u;
double d;
char* p;
} datum;
} SValueNode;
typedef enum EOperatorType {
@ -181,7 +189,7 @@ typedef struct SNodeListNode {
} SNodeListNode;
typedef struct SFunctionNode {
SExprNode type; // QUERY_NODE_FUNCTION
SExprNode node; // QUERY_NODE_FUNCTION
char functionName[TSDB_FUNC_NAME_LEN];
int32_t funcId;
SNodeList* pParameterList;

11
include/util/taoserror.h
View File

@ -441,10 +441,13 @@ int32_t* taosGetErrno();
#define TSDB_CODE_SCH_INTERNAL_ERROR TAOS_DEF_ERROR_CODE(0, 0x2502) //scheduler internal error
//parser
#define TSDB_CODE_PARSER_INVALID_COLUMN TAOS_DEF_ERROR_CODE(0, 0x2601) //invalid column name
#define TSDB_CODE_PARSER_TABLE_NOT_EXIST TAOS_DEF_ERROR_CODE(0, 0x2602) //table not exist
#define TSDB_CODE_PARSER_AMBIGUOUS_COLUMN TAOS_DEF_ERROR_CODE(0, 0x2603) //ambiguous column
#define TSDB_CODE_PARSER_WRONG_VALUE_TYPE TAOS_DEF_ERROR_CODE(0, 0x2604) //wrong value type
#define TSDB_CODE_PAR_INVALID_COLUMN TAOS_DEF_ERROR_CODE(0, 0x2601) //invalid column name
#define TSDB_CODE_PAR_TABLE_NOT_EXIST TAOS_DEF_ERROR_CODE(0, 0x2602) //table not exist
#define TSDB_CODE_PAR_AMBIGUOUS_COLUMN TAOS_DEF_ERROR_CODE(0, 0x2603) //ambiguous column
#define TSDB_CODE_PAR_WRONG_VALUE_TYPE TAOS_DEF_ERROR_CODE(0, 0x2604) //wrong value type
#define TSDB_CODE_PAR_FUNTION_PARA_NUM TAOS_DEF_ERROR_CODE(0, 0x2605) //invalid number of arguments
#define TSDB_CODE_PAR_FUNTION_PARA_TYPE TAOS_DEF_ERROR_CODE(0, 0x2606) //inconsistent datatypes
#define TSDB_CODE_PAR_ILLEGAL_USE_AGG_FUNCTION TAOS_DEF_ERROR_CODE(0, 0x2607) //there mustn't be aggregation
#ifdef __cplusplus
}

8
source/libs/function/src/functionMgt.c
View File

@ -78,3 +78,11 @@ FuncDef setExecFuncs(FuncDef def, FExecGetEnv getEnv, FExecInit init, FExecProce
int32_t registerFunc(FuncDef func) {
}
int32_t fmGetFuncResultType(FuncMgtHandle handle, SFunctionNode* pFunc) {
return TSDB_CODE_SUCCESS;
}
bool fmIsAggFunc(int32_t funcId) {
return false;
}

12
source/libs/parser/src/astCreateFuncs.c
View File

@ -118,14 +118,22 @@ SNode* createValueNode(SAstCreateContext* pCxt, int32_t dataType, const SToken*
val->literal = strndup(pLiteral->z, pLiteral->n);
CHECK_OUT_OF_MEM(val->literal);
val->node.resType.type = dataType;
val->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_BOOL].bytes;
val->node.resType.bytes = tDataTypes[dataType].bytes;
if (TSDB_DATA_TYPE_TIMESTAMP == dataType) {
val->node.resType.precision = TSDB_TIME_PRECISION_MILLI;
}
return (SNode*)val;
}
SNode* createDurationValueNode(SAstCreateContext* pCxt, const SToken* pLiteral) {
SValueNode* val = (SValueNode*)nodesMakeNode(QUERY_NODE_VALUE);
CHECK_OUT_OF_MEM(val);
// todo : calc, for example, 10s
val->literal = strndup(pLiteral->z, pLiteral->n);
CHECK_OUT_OF_MEM(val->literal);
val->isDuration = true;
val->node.resType.type = TSDB_DATA_TYPE_BIGINT;
val->node.resType.bytes = tDataTypes[TSDB_DATA_TYPE_BIGINT].bytes;
val->node.resType.precision = TSDB_TIME_PRECISION_MILLI;
return (SNode*)val;
}

112
source/libs/parser/src/parserImpl.c
View File

@ -16,7 +16,10 @@
#include "parserImpl.h"
#include "astCreateContext.h"
#include "functionMgt.h"
#include "parserInt.h"
#include "tglobal.h"
#include "ttime.h"
#include "ttoken.h"
typedef void* (*FMalloc)(size_t);
@ -240,6 +243,7 @@ typedef enum ESqlClause {
typedef struct STranslateContext {
SParseContext* pParseCxt;
FuncMgtHandle fmgt;
int32_t errCode;
SMsgBuf msgBuf;
SArray* pNsLevel; // element is SArray*, the element of this subarray is STableNode*
@ -251,21 +255,30 @@ static int32_t translateSubquery(STranslateContext* pCxt, SNode* pNode);
static char* getSyntaxErrFormat(int32_t errCode) {
switch (errCode) {
case TSDB_CODE_PARSER_INVALID_COLUMN:
case TSDB_CODE_PAR_INVALID_COLUMN:
return "Invalid column name : %s";
case TSDB_CODE_PARSER_TABLE_NOT_EXIST:
case TSDB_CODE_PAR_TABLE_NOT_EXIST:
return "Table does not exist : %s";
case TSDB_CODE_PARSER_AMBIGUOUS_COLUMN:
case TSDB_CODE_PAR_AMBIGUOUS_COLUMN:
return "Column ambiguously defined : %s";
case TSDB_CODE_PARSER_WRONG_VALUE_TYPE:
case TSDB_CODE_PAR_WRONG_VALUE_TYPE:
return "Invalid value type : %s";
case TSDB_CODE_PAR_FUNTION_PARA_NUM:
return "Invalid number of arguments : %s";
case TSDB_CODE_PAR_FUNTION_PARA_TYPE:
return "Inconsistent datatypes : %s";
case TSDB_CODE_PAR_ILLEGAL_USE_AGG_FUNCTION:
return "There mustn't be aggregation";
default:
return "Unknown error";
}
}
static int32_t generateSyntaxErrMsg(STranslateContext* pCxt, int32_t errCode, const char* additionalInfo) {
snprintf(pCxt->msgBuf.buf, pCxt->msgBuf.len, getSyntaxErrFormat(errCode), additionalInfo);
static int32_t generateSyntaxErrMsg(STranslateContext* pCxt, int32_t errCode, ...) {
va_list vArgList;
va_start(vArgList, errCode);
vsnprintf(pCxt->msgBuf.buf, pCxt->msgBuf.len, getSyntaxErrFormat(errCode), vArgList);
va_end(vArgList);
pCxt->errCode = errCode;
return errCode;
}
@ -394,7 +407,7 @@ static bool translateColumnWithPrefix(STranslateContext* pCxt, SColumnNode* pCol
if (findAndSetColumn(pCol, pTable)) {
break;
}
generateSyntaxErrMsg(pCxt, TSDB_CODE_PARSER_INVALID_COLUMN, pCol->colName);
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_INVALID_COLUMN, pCol->colName);
return false;
}
}
@ -409,14 +422,14 @@ static bool translateColumnWithoutPrefix(STranslateContext* pCxt, SColumnNode* p
STableNode* pTable = taosArrayGetP(pTables, i);
if (findAndSetColumn(pCol, pTable)) {
if (found) {
generateSyntaxErrMsg(pCxt, TSDB_CODE_PARSER_AMBIGUOUS_COLUMN, pCol->colName);
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_AMBIGUOUS_COLUMN, pCol->colName);
return false;
}
found = true;
}
}
if (!found) {
generateSyntaxErrMsg(pCxt, TSDB_CODE_PARSER_INVALID_COLUMN, pCol->colName);
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_INVALID_COLUMN, pCol->colName);
return false;
}
return true;
@ -429,8 +442,72 @@ static bool translateColumn(STranslateContext* pCxt, SColumnNode* pCol) {
return translateColumnWithoutPrefix(pCxt, pCol);
}
// check literal format
static int32_t trimStringCopy(const char* src, int32_t len, char* dst) {
// delete escape character: \\, \', \"
char delim = src[0];
int32_t cnt = 0;
int32_t j = 0;
for (uint32_t k = 1; k < len - 1; ++k) {
if (src[k] == '\\' || (src[k] == delim && src[k + 1] == delim)) {
dst[j] = src[k + 1];
cnt++;
j++;
k++;
continue;
}
dst[j] = src[k];
j++;
}
dst[j] = '\0';
return j;
}
static bool translateValue(STranslateContext* pCxt, SValueNode* pVal) {
if (pVal->isDuration) {
char unit = 0;
if (parseAbsoluteDuration(pVal->literal, strlen(pVal->literal), &pVal->datum.i, &unit, pVal->node.resType.precision) != TSDB_CODE_SUCCESS) {
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pVal->literal);
return false;
}
} else {
switch (pVal->node.resType.type) {
case TSDB_DATA_TYPE_NULL:
break;
case TSDB_DATA_TYPE_BOOL:
pVal->datum.b = (0 == strcasecmp(pVal->literal, "true"));
break;
case TSDB_DATA_TYPE_BIGINT: {
char* endPtr = NULL;
pVal->datum.i = strtoull(pVal->literal, &endPtr, 10);
break;
}
case TSDB_DATA_TYPE_DOUBLE: {
char* endPtr = NULL;
pVal->datum.d = strtold(pVal->literal, &endPtr);
break;
}
case TSDB_DATA_TYPE_BINARY: {
int32_t n = strlen(pVal->literal);
pVal->datum.p = calloc(1, n);
trimStringCopy(pVal->literal, n, pVal->datum.p);
break;
}
case TSDB_DATA_TYPE_TIMESTAMP: {
int32_t n = strlen(pVal->literal);
char* tmp = calloc(1, n);
int32_t len = trimStringCopy(pVal->literal, n, tmp);
if (taosParseTime(tmp, &pVal->datum.u, len, pVal->node.resType.precision, tsDaylight) != TSDB_CODE_SUCCESS) {
tfree(tmp);
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pVal->literal);
return false;
}
tfree(tmp);
break;
}
default:
break;
}
}
return true;
}
@ -440,7 +517,7 @@ static bool translateOperator(STranslateContext* pCxt, SOperatorNode* pOp) {
if (nodesIsArithmeticOp(pOp)) {
if (TSDB_DATA_TYPE_JSON == ldt.type || TSDB_DATA_TYPE_BLOB == ldt.type ||
TSDB_DATA_TYPE_JSON == rdt.type || TSDB_DATA_TYPE_BLOB == rdt.type) {
generateSyntaxErrMsg(pCxt, TSDB_CODE_PARSER_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
return false;
}
pOp->node.resType.type = TSDB_DATA_TYPE_DOUBLE;
@ -449,7 +526,7 @@ static bool translateOperator(STranslateContext* pCxt, SOperatorNode* pOp) {
} else if (nodesIsComparisonOp(pOp)) {
if (TSDB_DATA_TYPE_JSON == ldt.type || TSDB_DATA_TYPE_BLOB == ldt.type ||
TSDB_DATA_TYPE_JSON == rdt.type || TSDB_DATA_TYPE_BLOB == rdt.type) {
generateSyntaxErrMsg(pCxt, TSDB_CODE_PARSER_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)(pOp->pRight))->aliasName);
return false;
}
pOp->node.resType.type = TSDB_DATA_TYPE_BOOL;
@ -463,6 +540,15 @@ static bool translateOperator(STranslateContext* pCxt, SOperatorNode* pOp) {
}
static bool translateFunction(STranslateContext* pCxt, SFunctionNode* pFunc) {
int32_t code = fmGetFuncResultType(pCxt->fmgt, pFunc);
if (TSDB_CODE_SUCCESS != code) {
generateSyntaxErrMsg(pCxt, code, pFunc->functionName);
return false;
}
if (fmIsAggFunc(pFunc->funcId) && (SQL_CLAUSE_FROM == pCxt->currClause || SQL_CLAUSE_WHERE == pCxt->currClause)) {
generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_ILLEGAL_USE_AGG_FUNCTION);
return false;
}
return true;
}
@ -504,7 +590,7 @@ static int32_t translateTable(STranslateContext* pCxt, SNode* pTable) {
code = catalogGetTableMeta(pCxt->pParseCxt->pCatalog, pCxt->pParseCxt->pTransporter, &(pCxt->pParseCxt->mgmtEpSet),
toName(pCxt->pParseCxt->acctId, pRealTable, &name), &(pRealTable->pMeta));
if (TSDB_CODE_SUCCESS != code) {
return generateSyntaxErrMsg(pCxt, TSDB_CODE_PARSER_TABLE_NOT_EXIST, pRealTable->table.tableName);
return generateSyntaxErrMsg(pCxt, TSDB_CODE_PAR_TABLE_NOT_EXIST, pRealTable->table.tableName);
}
code = addNamespace(pCxt, pRealTable);
break;

66
source/libs/parser/test/newParserTest.cpp
View File

@ -118,6 +118,53 @@ private:
return "Unknown Data Type " + to_string(dt.type);
}
void valueNodeToStr(const SValueNode* pVal, string& str, bool isProject) {
switch (pVal->node.resType.type) {
case TSDB_DATA_TYPE_NULL:
str.append("null");
break;
case TSDB_DATA_TYPE_BOOL:
str.append(pVal->datum.b ? "true" : "false");
break;
case TSDB_DATA_TYPE_TINYINT:
case TSDB_DATA_TYPE_SMALLINT:
case TSDB_DATA_TYPE_INT:
case TSDB_DATA_TYPE_BIGINT:
str.append(to_string(pVal->datum.i));
break;
case TSDB_DATA_TYPE_FLOAT:
case TSDB_DATA_TYPE_DOUBLE:
str.append(to_string(pVal->datum.d));
break;
case TSDB_DATA_TYPE_BINARY:
case TSDB_DATA_TYPE_NCHAR:
case TSDB_DATA_TYPE_VARCHAR:
case TSDB_DATA_TYPE_VARBINARY:
str.append(pVal->datum.p);
break;
case TSDB_DATA_TYPE_TIMESTAMP:
str.append(to_string(pVal->datum.u));
break;
case TSDB_DATA_TYPE_UTINYINT:
case TSDB_DATA_TYPE_USMALLINT:
case TSDB_DATA_TYPE_UINT:
case TSDB_DATA_TYPE_UBIGINT:
str.append(to_string(pVal->datum.u));
break;
case TSDB_DATA_TYPE_JSON:
case TSDB_DATA_TYPE_DECIMAL:
case TSDB_DATA_TYPE_BLOB:
str.append("JSON or DECIMAL or BLOB");
break;
default:
break;
}
str.append(" [" + dataTypeToStr(pVal->node.resType) + "]");
if (isProject) {
str.append(" AS " + string(pVal->node.aliasName));
}
}
void nodeToStr(const SNode* node, string& str, bool isProject) {
if (nullptr == node) {
return;
@ -142,12 +189,7 @@ private:
break;
}
case QUERY_NODE_VALUE: {
SValueNode* pVal = (SValueNode*)node;
str.append(pVal->literal);
str.append(" [" + dataTypeToStr(pVal->node.resType) + "]");
if (isProject) {
str.append(" AS " + string(pVal->node.aliasName));
}
valueNodeToStr((SValueNode*)node, str, isProject);
break;
}
case QUERY_NODE_OPERATOR: {
@ -391,10 +433,20 @@ TEST_F(NewParserTest, selectSimple) {
ASSERT_TRUE(run());
}
TEST_F(NewParserTest, selectConstant) {
setDatabase("root", "test");
bind("SELECT 123, 20.4, 'abc', \"wxy\", TIMESTAMP '2022-02-09 17:30:20', true, false, 10s FROM t1");
ASSERT_TRUE(run());
bind("SELECT 1234567890123456789012345678901234567890, 20.1234567890123456789012345678901234567890, 'abc', \"wxy\", TIMESTAMP '2022-02-09 17:30:20', true, false, 15s FROM t1");
ASSERT_TRUE(run());
}
TEST_F(NewParserTest, selectExpression) {
setDatabase("root", "test");
bind("SELECT c1 + 10, c2 FROM t1");
bind("SELECT ts + 10s, c1 + 10, concat(c2, 'abc') FROM t1");
ASSERT_TRUE(run());
}