Merge remote-tracking branch 'origin/3.0' into feature/dnode3

include/common/taosmsg.h

@@ -51,7 +51,7 @@ TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_QUERY, "mq-query" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_CONNECT, "mq-connect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_DISCONNECT, "mq-disconnect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_SET_CUR, "mq-set-cur" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_RSP_READY, "rsp-ready" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_RES_READY, "res-ready" )
 // message from client to mnode
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CONNECT, "connect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_ACCT, "create-acct" )	
@@ -1081,6 +1081,7 @@ typedef struct {
 } SUpdateTagValRsp;
 
 typedef struct SSchedulerQueryMsg {
+  uint64_t  schedulerId;
   uint64_t  queryId;
   uint64_t  taskId;
   uint32_t  contentLen;
@@ -1088,15 +1089,39 @@ typedef struct SSchedulerQueryMsg {
 } SSchedulerQueryMsg;
 
 typedef struct SSchedulerReadyMsg {
+  uint64_t  schedulerId;
   uint64_t  queryId;
   uint64_t  taskId;
 } SSchedulerReadyMsg;
 
 typedef struct SSchedulerFetchMsg {
+  uint64_t  schedulerId;
   uint64_t  queryId;
   uint64_t  taskId;
 } SSchedulerFetchMsg;
 
+typedef struct SSchedulerStatusMsg {
+  uint64_t  schedulerId;
+} SSchedulerStatusMsg;
+
+typedef struct STaskStatus {
+  uint64_t  queryId;
+  uint64_t  taskId;
+  int8_t    status;
+} STaskStatus;
+
+typedef struct SSchedulerStatusRsp {
+  uint32_t    num;
+  STaskStatus status[];
+} SSchedulerStatusRsp;
+
+
+typedef struct SSchedulerCancelMsg {
+  uint64_t  schedulerId;
+  uint64_t  queryId;
+  uint64_t  taskId;
+} SSchedulerCancelMsg;
+
 
 #pragma pack(pop)
 

include/dnode/qnode/qnode.h

@@ -19,7 +19,7 @@
 #ifdef __cplusplus
 extern "C" {
 #endif
-
+#include "trpc.h"
 
 typedef struct {
   uint64_t numOfStartTask;
@@ -32,48 +32,6 @@ typedef struct {
   uint64_t numOfErrors;
 } SQnodeStat;
 
-/* start Task msg */
-typedef struct {
-  uint32_t  schedulerIp;
-  uint16_t  schedulerPort;
-  int64_t   taskId;
-  int64_t   queryId;
-  uint32_t  srcIp;
-  uint16_t  srcPort;
-} SQnodeStartTaskMsg;
-
-/* stop Task msg */
-typedef struct {
-  int64_t   taskId;
-} SQnodeStopTaskMsg;
-
-/* start/stop Task msg response */
-typedef struct {
-  int64_t   taskId;
-  int32_t code;
-} SQnodeTaskRespMsg;
-
-/* Task status msg */
-typedef struct {
-  int64_t   taskId;
-  int32_t   status;
-  int64_t   queryId;
-} SQnodeTaskStatusMsg;
-
-/* Qnode/Scheduler heartbeat msg */
-typedef struct {
-  int32_t status;
-  int32_t load;
-  
-} SQnodeHeartbeatMsg;
-
-/* Qnode sent/received msg */
-typedef struct {
-  int8_t   msgType;
-  int32_t  msgLen;
-  char     msg[];
-} SQnodeMsg;
-
 
 /**
  * Start one Qnode in Dnode.

include/libs/index/index.h

@@ -44,41 +44,46 @@ typedef enum { QUERY_TERM = 0, QUERY_PREFIX = 1, QUERY_SUFFIX = 2, QUERY_REGEX =
  * @param: oper
  *
  */
-SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType oper);
-void                  indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery);
-int                   indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType type);
+SIndexMultiTermQuery* indexMultiTermQueryCreate(EIndexOperatorType oper);
+void                  indexMultiTermQueryDestroy(SIndexMultiTermQuery* pQuery);
+int                   indexMultiTermQueryAdd(SIndexMultiTermQuery* pQuery, SIndexTerm* term, EIndexQueryType type);
 /*
  * @param:
  * @param:
  */
-int  indexOpen(SIndexOpts *opt, const char *path, SIndex **index);
-void indexClose(SIndex *index);
-int  indexPut(SIndex *index, SIndexMultiTerm *terms, uint64_t uid);
-int  indexDelete(SIndex *index, SIndexMultiTermQuery *query);
-int  indexSearch(SIndex *index, SIndexMultiTermQuery *query, SArray *result);
-int  indexRebuild(SIndex *index, SIndexOpts *opt);
+int  indexOpen(SIndexOpts* opt, const char* path, SIndex** index);
+void indexClose(SIndex* index);
+int  indexPut(SIndex* index, SIndexMultiTerm* terms, uint64_t uid);
+int  indexDelete(SIndex* index, SIndexMultiTermQuery* query);
+int  indexSearch(SIndex* index, SIndexMultiTermQuery* query, SArray* result);
+int  indexRebuild(SIndex* index, SIndexOpts* opt);
 /*
  * @param
  * @param
  */
-SIndexMultiTerm *indexMultiTermCreate();
-int              indexMultiTermAdd(SIndexMultiTerm *terms, SIndexTerm *term);
-void             indexMultiTermDestroy(SIndexMultiTerm *terms);
+SIndexMultiTerm* indexMultiTermCreate();
+int              indexMultiTermAdd(SIndexMultiTerm* terms, SIndexTerm* term);
+void             indexMultiTermDestroy(SIndexMultiTerm* terms);
 /*
  * @param:
  * @param:
  */
-SIndexOpts *indexOptsCreate();
-void        indexOptsDestroy(SIndexOpts *opts);
+SIndexOpts* indexOptsCreate();
+void        indexOptsDestroy(SIndexOpts* opts);
 
 /*
  * @param:
  * @param:
  */
 
-SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn operType, uint8_t colType, const char *colName,
-    int32_t nColName, const char *colVal, int32_t nColVal);
-void        indexTermDestroy(SIndexTerm *p);
+SIndexTerm* indexTermCreate(int64_t            suid,
+                            SIndexOperOnColumn operType,
+                            uint8_t            colType,
+                            const char*        colName,
+                            int32_t            nColName,
+                            const char*        colVal,
+                            int32_t            nColVal);
+void        indexTermDestroy(SIndexTerm* p);
 
 #ifdef __cplusplus
 }

include/libs/parser/parser.h

@@ -136,6 +136,7 @@ typedef struct SParseContext {
   const char* pDbname;
   void *pRpc;
   const char* pClusterId;
+  struct SCatalog* pCatalog;
   const SEpSet* pEpSet;
   int64_t id;          // query id, generated by uuid generator
   int8_t schemaAttached; // denote if submit block is built with table schema or not

include/libs/qcom/query.h

@@ -24,6 +24,15 @@ extern "C" {
 #include "thash.h"
 #include "tlog.h"
 
+enum {
+  JOB_TASK_STATUS_NOT_START = 1,
+  JOB_TASK_STATUS_EXECUTING,
+  JOB_TASK_STATUS_SUCCEED,
+  JOB_TASK_STATUS_FAILED,
+  JOB_TASK_STATUS_CANCELLING,
+  JOB_TASK_STATUS_CANCELLED
+};
+
 typedef struct STableComInfo {
   uint8_t numOfTags;      // the number of tags in schema
   uint8_t precision;      // the number of precision

include/libs/qworker/qworker.h

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_QWORKER_H_
+#define _TD_QWORKER_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "trpc.h"
+
+typedef struct SQWorkerCfg {
+  uint32_t maxSchedulerNum;
+  uint32_t maxResCacheNum;
+  uint32_t maxSchTaskNum;
+} SQWorkerCfg;
+
+typedef struct {
+  uint64_t numOfStartTask;
+  uint64_t numOfStopTask;
+  uint64_t numOfRecvedFetch;
+  uint64_t numOfSentHb;
+  uint64_t numOfSentFetch;
+  uint64_t numOfTaskInQueue;
+  uint64_t numOfFetchInQueue;
+  uint64_t numOfErrors;
+} SQWorkerStat;
+
+
+int32_t qWorkerInit(SQWorkerCfg *cfg, void **qWorkerMgmt);
+
+int32_t qWorkerProcessQueryMsg(void *qWorkerMgmt, SSchedulerQueryMsg *msg, SRpcMsg *rsp);
+
+int32_t qWorkerProcessReadyMsg(void *qWorkerMgmt, SSchedulerReadyMsg *msg, SRpcMsg *rsp);
+
+int32_t qWorkerProcessStatusMsg(void *qWorkerMgmt, SSchedulerStatusMsg *msg, SRpcMsg *rsp);
+
+int32_t qWorkerProcessFetchMsg(void *qWorkerMgmt, SSchedulerFetchMsg *msg, SRpcMsg *rsp);
+
+int32_t qWorkerProcessCancelMsg(void *qWorkerMgmt, SSchedulerCancelMsg *msg, SRpcMsg *rsp);
+
+void qWorkerDestroy(void **qWorkerMgmt);
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_QWORKER_H_*/

include/libs/scheduler/scheduler.h

@@ -25,6 +25,7 @@ extern "C" {
 
 typedef struct SSchedulerCfg {
   int32_t clusterType;
+  int32_t maxJobNum;
 } SSchedulerCfg;
 
 typedef struct SQueryProfileSummary {

include/util/thash.h

@@ -193,11 +193,10 @@ void  taosHashCancelIterate(SHashObj *pHashObj, void *p);
 
 /**
  * Get the corresponding key information for a given data in hash table
- * @param pHashObj
  * @param data
  * @return
  */
-int32_t taosHashGetKey(SHashObj *pHashObj, void *data, void** key, size_t* keyLen);
+int32_t taosHashGetKey(void *data, void** key, size_t* keyLen);
 
 #ifdef __cplusplus
 }

source/libs/CMakeLists.txt

@@ -11,3 +11,4 @@ add_subdirectory(executor)
 add_subdirectory(planner)
 add_subdirectory(function)
 add_subdirectory(qcom)
+add_subdirectory(qworker)

source/libs/executor/src/tfilter.c

@@ -1175,7 +1175,7 @@ int32_t filterAddGroupUnitFromNode(SFilterInfo *info, tExprNode* tree, SArray *g
       void* key = NULL;
       len = 0;
 
-      taosHashGetKey((SHashObj *)data, p, &key, &len);
+      taosHashGetKey(p, &key, &len);
       void *fdata = NULL;
     
       if (IS_VAR_DATA_TYPE(type)) {

source/libs/index/inc/indexInt.h

@@ -42,11 +42,11 @@ typedef struct SIndexStat {
 
 struct SIndex {
 #ifdef USE_LUCENE
-  index_t *index;
+  index_t* index;
 #endif
-  void *    cache;
-  void *    tindex;
-  SHashObj *colObj;  // < field name, field id>
+  void*     cache;
+  void*     tindex;
+  SHashObj* colObj;  // < field name, field id>
 
   int64_t suid;      // current super table id, -1 is normal table
   int     colId;     // field id allocated to cache
@@ -58,7 +58,7 @@ struct SIndex {
 
 struct SIndexOpts {
 #ifdef USE_LUCENE
-  void *opts;
+  void* opts;
 #endif
 
 #ifdef USE_INVERTED_INDEX
@@ -69,7 +69,7 @@ struct SIndexOpts {
 
 struct SIndexMultiTermQuery {
   EIndexOperatorType opera;
-  SArray *           query;
+  SArray*            query;
 };
 
 // field and key;
@@ -77,14 +77,14 @@ typedef struct SIndexTerm {
   int64_t            suid;
   SIndexOperOnColumn operType;  // oper type, add/del/update
   uint8_t            colType;   // term data type, str/interger/json
-  char *             colName;
+  char*              colName;
   int32_t            nColName;
-  char *             colVal;
+  char*              colVal;
   int32_t            nColVal;
 } SIndexTerm;
 
 typedef struct SIndexTermQuery {
-  SIndexTerm *    term;
+  SIndexTerm*     term;
   EIndexQueryType qType;
 } SIndexTermQuery;
 

source/libs/index/inc/index_cache.h

@@ -34,19 +34,18 @@ extern "C" {
 
 typedef struct IndexCache {
   T_REF_DECLARE()
-  SSkipList *skiplist;
+  SSkipList* skiplist;
 } IndexCache;
 
 //
-IndexCache *indexCacheCreate();
+IndexCache* indexCacheCreate();
 
-void indexCacheDestroy(void *cache);
+void indexCacheDestroy(void* cache);
 
-int indexCachePut(void *cache, SIndexTerm *term, int16_t colId, int32_t version, uint64_t uid);
+int indexCachePut(void* cache, SIndexTerm* term, int16_t colId, int32_t version, uint64_t uid);
 
 // int indexCacheGet(void *cache, uint64_t *rst);
-int indexCacheSearch(
-    void *cache, SIndexTermQuery *query, int16_t colId, int32_t version, SArray *result, STermValueType *s);
+int indexCacheSearch(void* cache, SIndexTermQuery* query, int16_t colId, int32_t version, SArray* result, STermValueType* s);
 
 #ifdef __cplusplus
 }

source/libs/index/inc/index_fst.h

@@ -40,10 +40,10 @@ typedef struct FstBoundWithData {
 } FstBoundWithData;
 
 typedef struct FstStreamBuilder {
-  Fst *             fst;
-  AutomationCtx *   aut;
-  FstBoundWithData *min;
-  FstBoundWithData *max;
+  Fst*              fst;
+  AutomationCtx*    aut;
+  FstBoundWithData* min;
+  FstBoundWithData* max;
 } FstStreamBuilder, FstStreamWithStateBuilder;
 
 typedef struct FstRange {
@@ -55,10 +55,10 @@ typedef enum { GE, GT, LE, LT } RangeType;
 typedef enum { OneTransNext, OneTrans, AnyTrans, EmptyFinal } State;
 typedef enum { Ordered, OutOfOrdered, DuplicateKey } OrderType;
 
-FstBoundWithData *fstBoundStateCreate(FstBound type, FstSlice *data);
-bool              fstBoundWithDataExceededBy(FstBoundWithData *bound, FstSlice *slice);
-bool              fstBoundWithDataIsEmpty(FstBoundWithData *bound);
-bool              fstBoundWithDataIsIncluded(FstBoundWithData *bound);
+FstBoundWithData* fstBoundStateCreate(FstBound type, FstSlice* data);
+bool              fstBoundWithDataExceededBy(FstBoundWithData* bound, FstSlice* slice);
+bool              fstBoundWithDataIsEmpty(FstBoundWithData* bound);
+bool              fstBoundWithDataIsIncluded(FstBoundWithData* bound);
 
 typedef struct FstOutput {
   bool   null;
@@ -71,46 +71,46 @@ typedef struct FstOutput {
  * TODO: simple function name
  */
 typedef struct FstUnFinishedNodes {
-  SArray *stack;  // <FstBuilderNodeUnfinished> } FstUnFinishedNodes;
+  SArray* stack;  // <FstBuilderNodeUnfinished> } FstUnFinishedNodes;
 } FstUnFinishedNodes;
 
 #define FST_UNFINISHED_NODES_LEN(nodes) taosArrayGetSize(nodes->stack)
 
-FstUnFinishedNodes *fstUnFinishedNodesCreate();
-void                fstUnFinishedNodesDestroy(FstUnFinishedNodes *node);
-void                fstUnFinishedNodesPushEmpty(FstUnFinishedNodes *nodes, bool isFinal);
-void                fstUnFinishedNodesSetRootOutput(FstUnFinishedNodes *node, Output out);
-void                fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes *node, CompiledAddr addr);
-void                fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *node, FstSlice bs, Output out);
-uint64_t            fstUnFinishedNodesFindCommPrefix(FstUnFinishedNodes *node, FstSlice bs);
-FstBuilderNode *    fstUnFinishedNodesPopRoot(FstUnFinishedNodes *nodes);
-FstBuilderNode *    fstUnFinishedNodesPopFreeze(FstUnFinishedNodes *nodes, CompiledAddr addr);
-FstBuilderNode *    fstUnFinishedNodesPopEmpty(FstUnFinishedNodes *nodes);
+FstUnFinishedNodes* fstUnFinishedNodesCreate();
+void                fstUnFinishedNodesDestroy(FstUnFinishedNodes* node);
+void                fstUnFinishedNodesPushEmpty(FstUnFinishedNodes* nodes, bool isFinal);
+void                fstUnFinishedNodesSetRootOutput(FstUnFinishedNodes* node, Output out);
+void                fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes* node, CompiledAddr addr);
+void                fstUnFinishedNodesAddSuffix(FstUnFinishedNodes* node, FstSlice bs, Output out);
+uint64_t            fstUnFinishedNodesFindCommPrefix(FstUnFinishedNodes* node, FstSlice bs);
+FstBuilderNode*     fstUnFinishedNodesPopRoot(FstUnFinishedNodes* nodes);
+FstBuilderNode*     fstUnFinishedNodesPopFreeze(FstUnFinishedNodes* nodes, CompiledAddr addr);
+FstBuilderNode*     fstUnFinishedNodesPopEmpty(FstUnFinishedNodes* nodes);
 
-uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes *node, FstSlice bs, Output in, Output *out);
+uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes* node, FstSlice bs, Output in, Output* out);
 
 typedef struct FstBuilder {
-  FstCountingWriter * wrt;         // The FST raw data is written directly to `wtr`.
-  FstUnFinishedNodes *unfinished;  // The stack of unfinished nodes
-  FstRegistry *       registry;    // A map of finished nodes.
+  FstCountingWriter*  wrt;         // The FST raw data is written directly to `wtr`.
+  FstUnFinishedNodes* unfinished;  // The stack of unfinished nodes
+  FstRegistry*        registry;    // A map of finished nodes.
   FstSlice            last;        // The last word added
   CompiledAddr        lastAddr;    // The address of the last compiled node
   uint64_t            len;         // num of keys added
 } FstBuilder;
 
-FstBuilder *fstBuilderCreate(void *w, FstType ty);
+FstBuilder* fstBuilderCreate(void* w, FstType ty);
 
-void         fstBuilderDestroy(FstBuilder *b);
-void         fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in);
-bool         fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in);
-void         fstBuilderCompileFrom(FstBuilder *b, uint64_t istate);
-void *       fstBuilerIntoInner(FstBuilder *b);
-void         fstBuilderFinish(FstBuilder *b);
-OrderType    fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup);
-CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn);
+void         fstBuilderDestroy(FstBuilder* b);
+void         fstBuilderInsertOutput(FstBuilder* b, FstSlice bs, Output in);
+bool         fstBuilderInsert(FstBuilder* b, FstSlice bs, Output in);
+void         fstBuilderCompileFrom(FstBuilder* b, uint64_t istate);
+void*        fstBuilerIntoInner(FstBuilder* b);
+void         fstBuilderFinish(FstBuilder* b);
+OrderType    fstBuilderCheckLastKey(FstBuilder* b, FstSlice bs, bool ckDup);
+CompiledAddr fstBuilderCompile(FstBuilder* b, FstBuilderNode* bn);
 
 typedef struct FstTransitions {
-  FstNode *node;
+  FstNode* node;
   FstRange range;
 } FstTransitions;
 
@@ -121,56 +121,55 @@ typedef struct FstState {
   uint8_t val;
 } FstState;
 
-FstState fstStateCreateFrom(FstSlice *data, CompiledAddr addr);
+FstState fstStateCreateFrom(FstSlice* data, CompiledAddr addr);
 FstState fstStateCreate(State state);
 
 // compile
-void fstStateCompileForOneTransNext(FstCountingWriter *w, CompiledAddr addr, uint8_t inp);
-void fstStateCompileForOneTrans(FstCountingWriter *w, CompiledAddr addr, FstTransition *trn);
-void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuilderNode *node);
+void fstStateCompileForOneTransNext(FstCountingWriter* w, CompiledAddr addr, uint8_t inp);
+void fstStateCompileForOneTrans(FstCountingWriter* w, CompiledAddr addr, FstTransition* trn);
+void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuilderNode* node);
 
 // set_comm_input
-void fstStateSetCommInput(FstState *state, uint8_t inp);
+void fstStateSetCommInput(FstState* state, uint8_t inp);
 
 // comm_input
-uint8_t fstStateCommInput(FstState *state, bool *null);
+uint8_t fstStateCommInput(FstState* state, bool* null);
 
 // input_len
 
-uint64_t fstStateInputLen(FstState *state);
+uint64_t fstStateInputLen(FstState* state);
 
 // end_addr
-uint64_t fstStateEndAddrForOneTransNext(FstState *state, FstSlice *data);
-uint64_t fstStateEndAddrForOneTrans(FstState *state, FstSlice *data, PackSizes sizes);
-uint64_t fstStateEndAddrForAnyTrans(
-    FstState *state, uint64_t version, FstSlice *date, PackSizes sizes, uint64_t nTrans);
+uint64_t fstStateEndAddrForOneTransNext(FstState* state, FstSlice* data);
+uint64_t fstStateEndAddrForOneTrans(FstState* state, FstSlice* data, PackSizes sizes);
+uint64_t fstStateEndAddrForAnyTrans(FstState* state, uint64_t version, FstSlice* date, PackSizes sizes, uint64_t nTrans);
 // input
-uint8_t fstStateInput(FstState *state, FstNode *node);
-uint8_t fstStateInputForAnyTrans(FstState *state, FstNode *node, uint64_t i);
+uint8_t fstStateInput(FstState* state, FstNode* node);
+uint8_t fstStateInputForAnyTrans(FstState* state, FstNode* node, uint64_t i);
 
 // trans_addr
-CompiledAddr fstStateTransAddr(FstState *state, FstNode *node);
-CompiledAddr fstStateTransAddrForAnyTrans(FstState *state, FstNode *node, uint64_t i);
+CompiledAddr fstStateTransAddr(FstState* state, FstNode* node);
+CompiledAddr fstStateTransAddrForAnyTrans(FstState* state, FstNode* node, uint64_t i);
 
 // sizes
-PackSizes fstStateSizes(FstState *state, FstSlice *data);
+PackSizes fstStateSizes(FstState* state, FstSlice* data);
 // Output
-Output fstStateOutput(FstState *state, FstNode *node);
-Output fstStateOutputForAnyTrans(FstState *state, FstNode *node, uint64_t i);
+Output fstStateOutput(FstState* state, FstNode* node);
+Output fstStateOutputForAnyTrans(FstState* state, FstNode* node, uint64_t i);
 
 // anyTrans specify function
 
-void fstStateSetFinalState(FstState *state, bool yes);
-bool fstStateIsFinalState(FstState *state);
-void fstStateSetStateNtrans(FstState *state, uint8_t n);
+void fstStateSetFinalState(FstState* state, bool yes);
+bool fstStateIsFinalState(FstState* state);
+void fstStateSetStateNtrans(FstState* state, uint8_t n);
 // state_ntrans
-uint8_t  fstStateStateNtrans(FstState *state, bool *null);
-uint64_t fstStateTotalTransSize(FstState *state, uint64_t version, PackSizes size, uint64_t nTrans);
-uint64_t fstStateTransIndexSize(FstState *state, uint64_t version, uint64_t nTrans);
-uint64_t fstStateNtransLen(FstState *state);
-uint64_t fstStateNtrans(FstState *state, FstSlice *slice);
-Output   fstStateFinalOutput(FstState *state, uint64_t version, FstSlice *date, PackSizes sizes, uint64_t nTrans);
-uint64_t fstStateFindInput(FstState *state, FstNode *node, uint8_t b, bool *null);
+uint8_t  fstStateStateNtrans(FstState* state, bool* null);
+uint64_t fstStateTotalTransSize(FstState* state, uint64_t version, PackSizes size, uint64_t nTrans);
+uint64_t fstStateTransIndexSize(FstState* state, uint64_t version, uint64_t nTrans);
+uint64_t fstStateNtransLen(FstState* state);
+uint64_t fstStateNtrans(FstState* state, FstSlice* slice);
+Output   fstStateFinalOutput(FstState* state, uint64_t version, FstSlice* date, PackSizes sizes, uint64_t nTrans);
+uint64_t fstStateFindInput(FstState* state, FstNode* node, uint8_t b, bool* null);
 
 #define FST_STATE_ONE_TRNAS_NEXT(node) (node->state.state == OneTransNext)
 #define FST_STATE_ONE_TRNAS(node) (node->state.state == OneTrans)
@@ -187,13 +186,13 @@ typedef struct FstLastTransition {
  * TODO: simple function name
  */
 typedef struct FstBuilderNodeUnfinished {
-  FstBuilderNode *   node;
-  FstLastTransition *last;
+  FstBuilderNode*    node;
+  FstLastTransition* last;
 } FstBuilderNodeUnfinished;
 
-void fstBuilderNodeUnfinishedLastCompiled(FstBuilderNodeUnfinished *node, CompiledAddr addr);
+void fstBuilderNodeUnfinishedLastCompiled(FstBuilderNodeUnfinished* node, CompiledAddr addr);
 
-void fstBuilderNodeUnfinishedAddOutputPrefix(FstBuilderNodeUnfinished *node, Output out);
+void fstBuilderNodeUnfinishedAddOutputPrefix(FstBuilderNodeUnfinished* node, Output out);
 
 /*
  * FstNode and helper function
@@ -224,18 +223,18 @@ typedef struct FstNode {
 // Return the address of this node.
 #define FST_NODE_ADDR(node) node->start
 
-FstNode *fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice *data);
-void     fstNodeDestroy(FstNode *fstNode);
+FstNode* fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice* data);
+void     fstNodeDestroy(FstNode* fstNode);
 
-FstTransitions  fstNodeTransitionIter(FstNode *node);
-FstTransitions *fstNodeTransitions(FstNode *node);
-bool            fstNodeGetTransitionAt(FstNode *node, uint64_t i, FstTransition *res);
-bool            fstNodeGetTransitionAddrAt(FstNode *node, uint64_t i, CompiledAddr *res);
-bool            fstNodeFindInput(FstNode *node, uint8_t b, uint64_t *res);
+FstTransitions  fstNodeTransitionIter(FstNode* node);
+FstTransitions* fstNodeTransitions(FstNode* node);
+bool            fstNodeGetTransitionAt(FstNode* node, uint64_t i, FstTransition* res);
+bool            fstNodeGetTransitionAddrAt(FstNode* node, uint64_t i, CompiledAddr* res);
+bool            fstNodeFindInput(FstNode* node, uint8_t b, uint64_t* res);
 
-bool fstNodeCompile(FstNode *node, void *w, CompiledAddr lastAddr, CompiledAddr addr, FstBuilderNode *builderNode);
+bool fstNodeCompile(FstNode* node, void* w, CompiledAddr lastAddr, CompiledAddr addr, FstBuilderNode* builderNode);
 
-FstSlice fstNodeAsSlice(FstNode *node);
+FstSlice fstNodeAsSlice(FstNode* node);
 
 // ops
 
@@ -244,8 +243,8 @@ typedef struct FstIndexedValue {
   uint64_t value;
 } FstIndexedValue;
 
-FstLastTransition *fstLastTransitionCreate(uint8_t inp, Output out);
-void               fstLastTransitionDestroy(FstLastTransition *trn);
+FstLastTransition* fstLastTransitionCreate(uint8_t inp, Output out);
+void               fstLastTransitionDestroy(FstLastTransition* trn);
 
 typedef struct FstMeta {
   uint64_t     version;
@@ -256,75 +255,74 @@ typedef struct FstMeta {
 } FstMeta;
 
 typedef struct Fst {
-  FstMeta * meta;
-  FstSlice *data;  //
-  FstNode * root;  //
+  FstMeta*  meta;
+  FstSlice* data;  //
+  FstNode*  root;  //
 } Fst;
 
 // refactor simple function
 
-Fst *fstCreate(FstSlice *data);
-void fstDestroy(Fst *fst);
+Fst* fstCreate(FstSlice* data);
+void fstDestroy(Fst* fst);
 
-bool              fstGet(Fst *fst, FstSlice *b, Output *out);
-FstNode *         fstGetNode(Fst *fst, CompiledAddr);
-FstNode *         fstGetRoot(Fst *fst);
-FstType           fstGetType(Fst *fst);
-CompiledAddr      fstGetRootAddr(Fst *fst);
-Output            fstEmptyFinalOutput(Fst *fst, bool *null);
-FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx);
+bool              fstGet(Fst* fst, FstSlice* b, Output* out);
+FstNode*          fstGetNode(Fst* fst, CompiledAddr);
+FstNode*          fstGetRoot(Fst* fst);
+FstType           fstGetType(Fst* fst);
+CompiledAddr      fstGetRootAddr(Fst* fst);
+Output            fstEmptyFinalOutput(Fst* fst, bool* null);
+FstStreamBuilder* fstSearch(Fst* fst, AutomationCtx* ctx);
 
-FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx);
+FstStreamWithStateBuilder* fstSearchWithState(Fst* fst, AutomationCtx* ctx);
 // into stream to expand later
-StreamWithState *streamBuilderIntoStream(FstStreamBuilder *sb);
+StreamWithState* streamBuilderIntoStream(FstStreamBuilder* sb);
 
-bool fstVerify(Fst *fst);
+bool fstVerify(Fst* fst);
 
 // refactor this function
-bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr startAddr);
+bool fstBuilderNodeCompileTo(FstBuilderNode* b, FstCountingWriter* wrt, CompiledAddr lastAddr, CompiledAddr startAddr);
 
 typedef struct StreamState {
-  FstNode * node;
+  FstNode*  node;
   uint64_t  trans;
   FstOutput out;
-  void *    autState;
+  void*     autState;
 } StreamState;
 
-void streamStateDestroy(void *s);
+void streamStateDestroy(void* s);
 
 typedef struct StreamWithState {
-  Fst *             fst;
-  AutomationCtx *   aut;
-  SArray *          inp;
+  Fst*              fst;
+  AutomationCtx*    aut;
+  SArray*           inp;
   FstOutput         emptyOutput;
-  SArray *          stack;  // <StreamState>
-  FstBoundWithData *endAt;
+  SArray*           stack;  // <StreamState>
+  FstBoundWithData* endAt;
 } StreamWithState;
 
 typedef struct StreamWithStateResult {
   FstSlice  data;
   FstOutput out;
-  void *    state;
+  void*     state;
 } StreamWithStateResult;
 
-StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *state);
-void                   swsResultDestroy(StreamWithStateResult *result);
+StreamWithStateResult* swsResultCreate(FstSlice* data, FstOutput fOut, void* state);
+void                   swsResultDestroy(StreamWithStateResult* result);
 
-typedef void *(*StreamCallback)(void *);
-StreamWithState *streamWithStateCreate(
-    Fst *fst, AutomationCtx *automation, FstBoundWithData *min, FstBoundWithData *max);
+typedef void* (*StreamCallback)(void*);
+StreamWithState* streamWithStateCreate(Fst* fst, AutomationCtx* automation, FstBoundWithData* min, FstBoundWithData* max);
 
-void streamWithStateDestroy(StreamWithState *sws);
+void streamWithStateDestroy(StreamWithState* sws);
 
-bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min);
+bool streamWithStateSeekMin(StreamWithState* sws, FstBoundWithData* min);
 
-StreamWithStateResult *streamWithStateNextWith(StreamWithState *sws, StreamCallback callback);
+StreamWithStateResult* streamWithStateNextWith(StreamWithState* sws, StreamCallback callback);
 
-FstStreamBuilder *fstStreamBuilderCreate(Fst *fst, AutomationCtx *aut);
+FstStreamBuilder* fstStreamBuilderCreate(Fst* fst, AutomationCtx* aut);
 // set up bound range
 // refator, simple code by marco
 
-FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type);
+FstStreamBuilder* fstStreamBuilderRange(FstStreamBuilder* b, FstSlice* val, RangeType type);
 
 #ifdef __cplusplus
 }

source/libs/index/inc/index_fst_automation.h

@@ -26,18 +26,18 @@ typedef struct AutomationCtx AutomationCtx;
 typedef enum AutomationType { AUTOMATION_PREFIX, AUTMMATION_MATCH } AutomationType;
 
 typedef struct StartWith {
-  AutomationCtx *autoSelf;
+  AutomationCtx* autoSelf;
 } StartWith;
 
 typedef struct Complement {
-  AutomationCtx *autoSelf;
+  AutomationCtx* autoSelf;
 } Complement;
 
 // automation
 typedef struct AutomationCtx {
   AutomationType type;
-  void *         stdata;
-  char *         data;
+  void*          stdata;
+  char*          data;
 } AutomationCtx;
 
 typedef enum ValueType { FST_INT, FST_CHAR, FST_ARRAY } ValueType;
@@ -48,27 +48,27 @@ typedef struct StartWithStateValue {
   ValueType          type;
   union {
     int     val;
-    char *  ptr;
-    SArray *arr;
+    char*   ptr;
+    SArray* arr;
     // add more type
   };
 } StartWithStateValue;
 
-StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void *val);
-StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv);
-void                 startWithStateValueDestroy(void *sv);
+StartWithStateValue* startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void* val);
+StartWithStateValue* startWithStateValueDump(StartWithStateValue* sv);
+void                 startWithStateValueDestroy(void* sv);
 
 typedef struct AutomationFunc {
-  void *(*start)(AutomationCtx *ctx);
-  bool (*isMatch)(AutomationCtx *ctx, void *);
-  bool (*canMatch)(AutomationCtx *ctx, void *data);
-  bool (*willAlwaysMatch)(AutomationCtx *ctx, void *state);
-  void *(*accept)(AutomationCtx *ctx, void *state, uint8_t byte);
-  void *(*acceptEof)(AutomationCtx *ct, void *state);
+  void* (*start)(AutomationCtx* ctx);
+  bool (*isMatch)(AutomationCtx* ctx, void*);
+  bool (*canMatch)(AutomationCtx* ctx, void* data);
+  bool (*willAlwaysMatch)(AutomationCtx* ctx, void* state);
+  void* (*accept)(AutomationCtx* ctx, void* state, uint8_t byte);
+  void* (*acceptEof)(AutomationCtx* ct, void* state);
 } AutomationFunc;
 
-AutomationCtx *automCtxCreate(void *data, AutomationType atype);
-void           automCtxDestroy(AutomationCtx *ctx);
+AutomationCtx* automCtxCreate(void* data, AutomationType atype);
+void           automCtxDestroy(AutomationCtx* ctx);
 
 extern AutomationFunc automFuncs[];
 #ifdef __cplusplus

source/libs/index/inc/index_fst_counting_writer.h

@@ -28,9 +28,10 @@ static char tmpFile[] = "./index";
 typedef enum WriterType { TMemory, TFile } WriterType;
 
 typedef struct WriterCtx {
-  int (*write)(struct WriterCtx *ctx, uint8_t *buf, int len);
-  int (*read)(struct WriterCtx *ctx, uint8_t *buf, int len);
-  int (*flush)(struct WriterCtx *ctx);
+  int (*write)(struct WriterCtx* ctx, uint8_t* buf, int len);
+  int (*read)(struct WriterCtx* ctx, uint8_t* buf, int len);
+  int (*flush)(struct WriterCtx* ctx);
+  int (*readFrom)(struct WriterCtx* ctx, uint8_t* buf, int len, int32_t offset);
   WriterType type;
   union {
     struct {
@@ -39,41 +40,42 @@ typedef struct WriterCtx {
     } file;
     struct {
       int32_t capa;
-      char *  buf;
+      char*   buf;
     } mem;
   };
   int32_t offset;
   int32_t limit;
 } WriterCtx;
 
-static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len);
-static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len);
-static int writeCtxDoFlush(WriterCtx *ctx);
+static int writeCtxDoWrite(WriterCtx* ctx, uint8_t* buf, int len);
+static int writeCtxDoRead(WriterCtx* ctx, uint8_t* buf, int len);
+static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t offset);
+static int writeCtxDoFlush(WriterCtx* ctx);
 
-WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity);
-void       writerCtxDestroy(WriterCtx *w);
+WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity);
+void       writerCtxDestroy(WriterCtx* w);
 
 typedef uint32_t CheckSummer;
 
 typedef struct FstCountingWriter {
-  void *      wrt;  // wrap any writer that counts and checksum bytes written
+  void*       wrt;  // wrap any writer that counts and checksum bytes written
   uint64_t    count;
   CheckSummer summer;
 } FstCountingWriter;
 
-int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t len);
+int fstCountingWriterWrite(FstCountingWriter* write, uint8_t* buf, uint32_t len);
 
-int fstCountingWriterRead(FstCountingWriter *write, uint8_t *buf, uint32_t len);
+int fstCountingWriterRead(FstCountingWriter* write, uint8_t* buf, uint32_t len);
 
-int fstCountingWriterFlush(FstCountingWriter *write);
+int fstCountingWriterFlush(FstCountingWriter* write);
 
-uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write);
+uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write);
 
-FstCountingWriter *fstCountingWriterCreate(void *wtr);
-void               fstCountingWriterDestroy(FstCountingWriter *w);
+FstCountingWriter* fstCountingWriterCreate(void* wtr);
+void               fstCountingWriterDestroy(FstCountingWriter* w);
 
-void    fstCountingWriterPackUintIn(FstCountingWriter *writer, uint64_t n, uint8_t nBytes);
-uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n);
+void    fstCountingWriterPackUintIn(FstCountingWriter* writer, uint64_t n, uint8_t nBytes);
+uint8_t fstCountingWriterPackUint(FstCountingWriter* writer, uint64_t n);
 
 #define FST_WRITER_COUNT(writer) (writer->count)
 #define FST_WRITER_INTER_WRITER(writer) (writer->wtr)

source/libs/index/inc/index_fst_node.h

@@ -36,20 +36,20 @@ typedef struct FstTransition {
 typedef struct FstBuilderNode {
   bool    isFinal;
   Output  finalOutput;
-  SArray *trans;  // <FstTransition>
+  SArray* trans;  // <FstTransition>
 } FstBuilderNode;
 
-FstBuilderNode *fstBuilderNodeDefault();
+FstBuilderNode* fstBuilderNodeDefault();
 
-FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src);
+FstBuilderNode* fstBuilderNodeClone(FstBuilderNode* src);
 
-void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src);
+void fstBuilderNodeCloneFrom(FstBuilderNode* dst, FstBuilderNode* src);
 
 // bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt,
 // CompiledAddr lastAddr, CompiledAddr startAddr);
-bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2);
+bool fstBuilderNodeEqual(FstBuilderNode* n1, FstBuilderNode* n2);
 
-void fstBuilderNodeDestroy(FstBuilderNode *node);
+void fstBuilderNodeDestroy(FstBuilderNode* node);
 
 #ifdef __cplusplus
 }

source/libs/index/inc/index_fst_registry.h

@@ -25,7 +25,7 @@ extern "C" {
 
 typedef struct FstRegistryCell {
   CompiledAddr    addr;
-  FstBuilderNode *node;
+  FstBuilderNode* node;
 } FstRegistryCell;
 
 #define FST_REGISTRY_CELL_IS_EMPTY(cell) (cell->addr == NONE_ADDRESS)
@@ -45,22 +45,22 @@ typedef enum { FOUND, NOTFOUND, REJECTED } FstRegistryEntryState;
 typedef struct FstRegistryEntry {
   FstRegistryEntryState state;
   CompiledAddr          addr;
-  FstRegistryCell *     cell;
+  FstRegistryCell*      cell;
 } FstRegistryEntry;
 
 // Registry relation function
 typedef struct FstRegistry {
-  SArray * table;      //<FstRegistryCell>
+  SArray*  table;      //<FstRegistryCell>
   uint64_t tableSize;  // num of rows
   uint64_t mruSize;    // num of columns
 } FstRegistry;
 
 //
-FstRegistry *fstRegistryCreate(uint64_t tableSize, uint64_t mruSize);
-void         fstRegistryDestroy(FstRegistry *registry);
+FstRegistry* fstRegistryCreate(uint64_t tableSize, uint64_t mruSize);
+void         fstRegistryDestroy(FstRegistry* registry);
 
-FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNode);
-void              fstRegistryEntryDestroy(FstRegistryEntry *entry);
+FstRegistryEntry* fstRegistryGetEntry(FstRegistry* registry, FstBuilderNode* bNode);
+void              fstRegistryEntryDestroy(FstRegistryEntry* entry);
 
 #ifdef __cplusplus
 }

source/libs/index/inc/index_fst_util.h

@@ -69,29 +69,29 @@ extern const uint64_t TRANS_INDEX_THRESHOLD;
 // uint8_t commonIdx(uint8_t v, uint8_t max);
 
 uint8_t      packSize(uint64_t n);
-uint64_t     unpackUint64(uint8_t *ch, uint8_t sz);
+uint64_t     unpackUint64(uint8_t* ch, uint8_t sz);
 uint8_t      packDeltaSize(CompiledAddr nodeAddr, CompiledAddr transAddr);
-CompiledAddr unpackDelta(char *data, uint64_t len, uint64_t nodeAddr);
+CompiledAddr unpackDelta(char* data, uint64_t len, uint64_t nodeAddr);
 
 typedef struct FstString {
-  uint8_t *data;
+  uint8_t* data;
   uint32_t len;
   int32_t  ref;
 } FstString;
 
 typedef struct FstSlice {
-  FstString *str;
+  FstString* str;
   int32_t    start;
   int32_t    end;
 } FstSlice;
 
-FstSlice fstSliceCreate(uint8_t *data, uint64_t len);
-FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end);
-FstSlice fstSliceDeepCopy(FstSlice *s, int32_t start, int32_t end);
-bool     fstSliceIsEmpty(FstSlice *s);
-int      fstSliceCompare(FstSlice *s1, FstSlice *s2);
-void     fstSliceDestroy(FstSlice *s);
-uint8_t *fstSliceData(FstSlice *s, int32_t *sz);
+FstSlice fstSliceCreate(uint8_t* data, uint64_t len);
+FstSlice fstSliceCopy(FstSlice* s, int32_t start, int32_t end);
+FstSlice fstSliceDeepCopy(FstSlice* s, int32_t start, int32_t end);
+bool     fstSliceIsEmpty(FstSlice* s);
+int      fstSliceCompare(FstSlice* s1, FstSlice* s2);
+void     fstSliceDestroy(FstSlice* s);
+uint8_t* fstSliceData(FstSlice* s, int32_t* sz);
 
 #define FST_SLICE_LEN(s) (s->end - s->start + 1)
 

source/libs/index/inc/index_tfile.h

@@ -26,84 +26,93 @@
 extern "C" {
 #endif
 
-// tfile header
-// |<---suid--->|<---version--->|<--colLen-->|<-colName->|<---type-->|
-// |<-uint64_t->|<---int32_t--->|<--int32_t->|<-colLen-->|<-uint8_t->|
+// tfile header content
+// |<---suid--->|<---version--->|<-------colName------>|<---type-->|<--fstOffset->|
+// |<-uint64_t->|<---int32_t--->|<--TSDB_COL_NAME_LEN-->|<-uint8_t->|<---int32_t-->|
 
-typedef struct TFileReadHeader {
+#pragma pack(push, 1)
+typedef struct TFileHeader {
   uint64_t suid;
   int32_t  version;
-  char     colName[128];  //
+  char     colName[TSDB_COL_NAME_LEN];  //
   uint8_t  colType;
-} TFileReadHeader;
+  int32_t  fstOffset;
+} TFileHeader;
+#pragma pack(pop)
 
-#define TFILE_HEADER_SIZE (sizeof(TFILE_HEADER_SIZE) + sizeof(uint32_t));
-#define TFILE_HADER_PRE_SIZE (sizeof(uint64_t) + sizeof(int32_t) + sizeof(int32_t))
+#define TFILE_HEADER_SIZE (sizeof(TFileHeader))
+#define TFILE_HEADER_NO_FST (TFILE_HEADER_SIZE - sizeof(int32_t))
+//#define TFILE_HADER_PRE_SIZE (sizeof(uint64_t) + sizeof(int32_t) + sizeof(int32_t))
 
 typedef struct TFileCacheKey {
   uint64_t suid;
   uint8_t  colType;
   int32_t  version;
-  const char *colName;
+  char*    colName;
   int32_t  nColName;
 } TFileCacheKey;
 
 // table cache
 // refactor to LRU cache later
 typedef struct TFileCache {
-  SHashObj *tableCache;
+  SHashObj* tableCache;
   int16_t   capacity;
   // add more param
 } TFileCache;
 
 typedef struct TFileWriter {
-  FstBuilder *fb;
-  WriterCtx * ctx;
+  FstBuilder* fb;
+  WriterCtx*  ctx;
+  TFileHeader header;
+  uint32_t    offset;
 } TFileWriter;
 
 typedef struct TFileReader {
   T_REF_DECLARE()
-  Fst *           fst;
-  WriterCtx *     ctx;
-  TFileReadHeader header;
+  Fst*        fst;
+  WriterCtx*  ctx;
+  TFileHeader header;
 } TFileReader;
 
 typedef struct IndexTFile {
-  char *       path;
-  TFileCache * cache;
-  TFileWriter *tw;
+  char*        path;
+  TFileCache*  cache;
+  TFileWriter* tw;
 } IndexTFile;
 
 typedef struct TFileWriterOpt {
   uint64_t suid;
   int8_t   colType;
-  char *   colName;
+  char*    colName;
   int32_t  nColName;
   int32_t  version;
 } TFileWriterOpt;
 
 typedef struct TFileReaderOpt {
   uint64_t suid;
-  char *   colName;
+  char*    colName;
   int32_t  nColName;
 } TFileReaderOpt;
 
 // tfile cache, manage tindex reader
-TFileCache * tfileCacheCreate(const char *path);
-void         tfileCacheDestroy(TFileCache *tcache);
-TFileReader *tfileCacheGet(TFileCache *tcache, TFileCacheKey *key);
-void         tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader);
+TFileCache*  tfileCacheCreate(const char* path);
+void         tfileCacheDestroy(TFileCache* tcache);
+TFileReader* tfileCacheGet(TFileCache* tcache, TFileCacheKey* key);
+void         tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader* reader);
 
-TFileReader *tfileReaderCreate();
-void         TFileReaderDestroy(TFileReader *reader);
+TFileReader* tfileReaderCreate(WriterCtx* ctx);
+void         tfileReaderDestroy(TFileReader* reader);
+int          tfileReaderSearch(TFileReader* reader, SIndexTermQuery* query, SArray* result);
 
-TFileWriter *tfileWriterCreate(const char *suid, const char *colName);
-void         tfileWriterDestroy(TFileWriter *tw);
+TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header);
+void         tfileWriterDestroy(TFileWriter* tw);
+int          tfileWriterPut(TFileWriter* tw, void* data);
+int          tfileWriterFinish(TFileWriter* tw);
 
 //
-IndexTFile *indexTFileCreate(const char *path);
-int         indexTFilePut(void *tfile, SIndexTerm *term, uint64_t uid);
-int         indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result);
+IndexTFile* indexTFileCreate(const char* path);
+int         indexTFilePut(void* tfile, SIndexTerm* term, uint64_t uid);
+int         indexTFileSearch(void* tfile, SIndexTermQuery* query, SArray* result);
 
 #ifdef __cplusplus
 }

source/libs/index/inc/index_util.h

@@ -21,13 +21,13 @@ extern "C" {
 
 #define SERIALIZE_MEM_TO_BUF(buf, key, mem)                   \
   do {                                                        \
-    memcpy((void *)buf, (void *)(&key->mem), sizeof(key->mem)); \
+    memcpy((void*)buf, (void*)(&key->mem), sizeof(key->mem)); \
     buf += sizeof(key->mem);                                  \
   } while (0)
 
 #define SERIALIZE_STR_MEM_TO_BUF(buf, key, mem, len) \
   do {                                               \
-    memcpy((void *)buf, (void *)key->mem, len);      \
+    memcpy((void*)buf, (void*)key->mem, len);        \
     buf += len;                                      \
   } while (0)
 
@@ -35,13 +35,13 @@ extern "C" {
   do {                                        \
     type c = var;                             \
     assert(sizeof(var) == sizeof(type));      \
-    memcpy((void *)buf, (void *)&c, sizeof(c)); \
+    memcpy((void*)buf, (void*)&c, sizeof(c)); \
     buf += sizeof(c);                         \
   } while (0)
 
 #define SERIALIZE_STR_VAR_TO_BUF(buf, var, len) \
   do {                                          \
-    memcpy((void *)buf, (void *)var, len);      \
+    memcpy((void*)buf, (void*)var, len);        \
     buf += len;                                 \
   } while (0)
 

source/libs/index/src/index.c

@@ -23,9 +23,9 @@
 #include "lucene++/Lucene_c.h"
 #endif
 
-static int uidCompare(const void *a, const void *b) {
-  uint64_t u1 = *(uint64_t *)a;
-  uint64_t u2 = *(uint64_t *)b;
+static int uidCompare(const void* a, const void* b) {
+  uint64_t u1 = *(uint64_t*)a;
+  uint64_t u2 = *(uint64_t*)b;
   if (u1 == u2) {
     return 0;
   } else {
@@ -40,25 +40,26 @@ typedef struct SIdxColInfo {
 static pthread_once_t isInit = PTHREAD_ONCE_INIT;
 static void           indexInit();
 
-static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *term, SArray **result);
-static int indexMergeCacheIntoTindex(SIndex *sIdx);
+static int indexTermSearch(SIndex* sIdx, SIndexTermQuery* term, SArray** result);
+static int indexMergeCacheIntoTindex(SIndex* sIdx);
 
-static void indexInterResultsDestroy(SArray *results);
-static int  indexMergeFinalResults(SArray *interResults, EIndexOperatorType oType, SArray *finalResult);
+static void indexInterResultsDestroy(SArray* results);
+static int  indexMergeFinalResults(SArray* interResults, EIndexOperatorType oType, SArray* finalResult);
 
-int indexOpen(SIndexOpts *opts, const char *path, SIndex **index) {
+int indexOpen(SIndexOpts* opts, const char* path, SIndex** index) {
   pthread_once(&isInit, indexInit);
-  SIndex *sIdx = calloc(1, sizeof(SIndex));
+  SIndex* sIdx = calloc(1, sizeof(SIndex));
   if (sIdx == NULL) {
     return -1;
   }
 
 #ifdef USE_LUCENE
-  index_t *index = index_open(path);
+  index_t* index = index_open(path);
   sIdx->index = index;
 #endif
 
-  sIdx->cache = (void *)indexCacheCreate();
+#ifdef USE_INVERTED_INDEX
+  sIdx->cache = (void*)indexCacheCreate();
   sIdx->tindex = NULL;
   sIdx->colObj = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   sIdx->colId = 1;
@@ -67,9 +68,13 @@ int indexOpen(SIndexOpts *opts, const char *path, SIndex **index) {
 
   *index = sIdx;
   return 0;
+#endif
+
+  *index = NULL;
+  return -1;
 }
 
-void indexClose(SIndex *sIdx) {
+void indexClose(SIndex* sIdx) {
 #ifdef USE_LUCENE
   index_close(sIdex->index);
   sIdx->index = NULL;
@@ -84,16 +89,16 @@ void indexClose(SIndex *sIdx) {
   return;
 }
 
-int indexPut(SIndex *index, SIndexMultiTerm *fVals, uint64_t uid) {
+int indexPut(SIndex* index, SIndexMultiTerm* fVals, uint64_t uid) {
 #ifdef USE_LUCENE
-  index_document_t *doc = index_document_create();
+  index_document_t* doc = index_document_create();
 
   char buf[16] = {0};
   sprintf(buf, "%d", uid);
 
   for (int i = 0; i < taosArrayGetSize(fVals); i++) {
-    SIndexTerm *p = taosArrayGetP(fVals, i);
-    index_document_add(doc, (const char *)(p->key), p->nKey, (const char *)(p->val), p->nVal, 1);
+    SIndexTerm* p = taosArrayGetP(fVals, i);
+    index_document_add(doc, (const char*)(p->key), p->nKey, (const char*)(p->val), p->nVal, 1);
   }
   index_document_add(doc, NULL, 0, buf, strlen(buf), 0);
 
@@ -106,8 +111,8 @@ int indexPut(SIndex *index, SIndexMultiTerm *fVals, uint64_t uid) {
   // TODO(yihao): reduce the lock range
   pthread_mutex_lock(&index->mtx);
   for (int i = 0; i < taosArrayGetSize(fVals); i++) {
-    SIndexTerm * p = taosArrayGetP(fVals, i);
-    SIdxColInfo *fi = taosHashGet(index->colObj, p->colName, p->nColName);
+    SIndexTerm*  p = taosArrayGetP(fVals, i);
+    SIdxColInfo* fi = taosHashGet(index->colObj, p->colName, p->nColName);
     if (fi == NULL) {
       SIdxColInfo tfi = {.colId = index->colId};
       index->cVersion++;
@@ -120,8 +125,8 @@ int indexPut(SIndex *index, SIndexMultiTerm *fVals, uint64_t uid) {
   pthread_mutex_unlock(&index->mtx);
 
   for (int i = 0; i < taosArrayGetSize(fVals); i++) {
-    SIndexTerm * p = taosArrayGetP(fVals, i);
-    SIdxColInfo *fi = taosHashGet(index->colObj, p->colName, p->nColName);
+    SIndexTerm*  p = taosArrayGetP(fVals, i);
+    SIdxColInfo* fi = taosHashGet(index->colObj, p->colName, p->nColName);
     assert(fi != NULL);
     int32_t colId = fi->colId;
     int32_t version = index->cVersion;
@@ -134,18 +139,18 @@ int indexPut(SIndex *index, SIndexMultiTerm *fVals, uint64_t uid) {
 
   return 0;
 }
-int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result) {
+int indexSearch(SIndex* index, SIndexMultiTermQuery* multiQuerys, SArray* result) {
 #ifdef USE_LUCENE
   EIndexOperatorType opera = multiQuerys->opera;
 
   int    nQuery = taosArrayGetSize(multiQuerys->query);
-  char **fields = malloc(sizeof(char *) * nQuery);
-  char **keys = malloc(sizeof(char *) * nQuery);
-  int *  types = malloc(sizeof(int) * nQuery);
+  char** fields = malloc(sizeof(char*) * nQuery);
+  char** keys = malloc(sizeof(char*) * nQuery);
+  int*   types = malloc(sizeof(int) * nQuery);
 
   for (int i = 0; i < nQuery; i++) {
-    SIndexTermQuery *p = taosArrayGet(multiQuerys->query, i);
-    SIndexTerm *     term = p->field_value;
+    SIndexTermQuery* p = taosArrayGet(multiQuerys->query, i);
+    SIndexTerm*      term = p->field_value;
 
     fields[i] = calloc(1, term->nKey + 1);
     keys[i] = calloc(1, term->nVal + 1);
@@ -154,9 +159,9 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
     memcpy(keys[i], term->val, term->nVal);
     types[i] = (int)(p->type);
   }
-  int *tResult = NULL;
+  int* tResult = NULL;
   int  tsz = 0;
-  index_multi_search(index->index, (const char **)fields, (const char **)keys, types, nQuery, opera, &tResult, &tsz);
+  index_multi_search(index->index, (const char**)fields, (const char**)keys, types, nQuery, opera, &tResult, &tsz);
 
   for (int i = 0; i < tsz; i++) {
     taosArrayPush(result, &tResult[i]);
@@ -174,13 +179,13 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
 #ifdef USE_INVERTED_INDEX
   EIndexOperatorType opera = multiQuerys->opera;  // relation of querys
 
-  SArray *interResults = taosArrayInit(4, POINTER_BYTES);
+  SArray* interResults = taosArrayInit(4, POINTER_BYTES);
   int     nQuery = taosArrayGetSize(multiQuerys->query);
   for (size_t i = 0; i < nQuery; i++) {
-    SIndexTermQuery *qTerm = taosArrayGet(multiQuerys->query, i);
-    SArray *         tResult = NULL;
+    SIndexTermQuery* qTerm = taosArrayGet(multiQuerys->query, i);
+    SArray*          tResult = NULL;
     indexTermSearch(index, qTerm, &tResult);
-    taosArrayPush(interResults, (void *)&tResult);
+    taosArrayPush(interResults, (void*)&tResult);
   }
   indexMergeFinalResults(interResults, opera, result);
   indexInterResultsDestroy(interResults);
@@ -189,24 +194,24 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
   return 1;
 }
 
-int indexDelete(SIndex *index, SIndexMultiTermQuery *query) {
+int indexDelete(SIndex* index, SIndexMultiTermQuery* query) {
 #ifdef USE_INVERTED_INDEX
 #endif
 
   return 1;
 }
-int indexRebuild(SIndex *index, SIndexOpts *opts){
+int indexRebuild(SIndex* index, SIndexOpts* opts){
 #ifdef USE_INVERTED_INDEX
 #endif
 
 }
 
-SIndexOpts *indexOptsCreate() {
+SIndexOpts* indexOptsCreate() {
 #ifdef USE_LUCENE
 #endif
   return NULL;
 }
-void indexOptsDestroy(SIndexOpts *opts){
+void indexOptsDestroy(SIndexOpts* opts){
 #ifdef USE_LUCENE
 #endif
 } /*
@@ -214,8 +219,8 @@ void indexOptsDestroy(SIndexOpts *opts){
    *
    */
 
-SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType opera) {
-  SIndexMultiTermQuery *p = (SIndexMultiTermQuery *)malloc(sizeof(SIndexMultiTermQuery));
+SIndexMultiTermQuery* indexMultiTermQueryCreate(EIndexOperatorType opera) {
+  SIndexMultiTermQuery* p = (SIndexMultiTermQuery*)malloc(sizeof(SIndexMultiTermQuery));
   if (p == NULL) {
     return NULL;
   }
@@ -223,23 +228,28 @@ SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType opera) {
   p->query = taosArrayInit(4, sizeof(SIndexTermQuery));
   return p;
 }
-void indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery) {
+void indexMultiTermQueryDestroy(SIndexMultiTermQuery* pQuery) {
   for (int i = 0; i < taosArrayGetSize(pQuery->query); i++) {
-    SIndexTermQuery *p = (SIndexTermQuery *)taosArrayGet(pQuery->query, i);
+    SIndexTermQuery* p = (SIndexTermQuery*)taosArrayGet(pQuery->query, i);
     indexTermDestroy(p->term);
   }
   taosArrayDestroy(pQuery->query);
   free(pQuery);
 };
-int indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType qType) {
+int indexMultiTermQueryAdd(SIndexMultiTermQuery* pQuery, SIndexTerm* term, EIndexQueryType qType) {
   SIndexTermQuery q = {.qType = qType, .term = term};
   taosArrayPush(pQuery->query, &q);
   return 0;
 }
 
-SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn oper, uint8_t colType, const char *colName,
-    int32_t nColName, const char *colVal, int32_t nColVal) {
-  SIndexTerm *t = (SIndexTerm *)calloc(1, (sizeof(SIndexTerm)));
+SIndexTerm* indexTermCreate(int64_t            suid,
+                            SIndexOperOnColumn oper,
+                            uint8_t            colType,
+                            const char*        colName,
+                            int32_t            nColName,
+                            const char*        colVal,
+                            int32_t            nColVal) {
+  SIndexTerm* t = (SIndexTerm*)calloc(1, (sizeof(SIndexTerm)));
   if (t == NULL) {
     return NULL;
   }
@@ -248,30 +258,32 @@ SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn oper, uint8_t colTy
   t->operType = oper;
   t->colType = colType;
 
-  t->colName = (char *)calloc(1, nColName + 1);
+  t->colName = (char*)calloc(1, nColName + 1);
   memcpy(t->colName, colName, nColName);
   t->nColName = nColName;
 
-  t->colVal = (char *)calloc(1, nColVal + 1);
+  t->colVal = (char*)calloc(1, nColVal + 1);
   memcpy(t->colVal, colVal, nColVal);
   t->nColVal = nColVal;
   return t;
 }
-void indexTermDestroy(SIndexTerm *p) {
+void indexTermDestroy(SIndexTerm* p) {
   free(p->colName);
   free(p->colVal);
   free(p);
 }
 
-SIndexMultiTerm *indexMultiTermCreate() { return taosArrayInit(4, sizeof(SIndexTerm *)); }
+SIndexMultiTerm* indexMultiTermCreate() {
+  return taosArrayInit(4, sizeof(SIndexTerm*));
+}
 
-int indexMultiTermAdd(SIndexMultiTerm *terms, SIndexTerm *term) {
+int indexMultiTermAdd(SIndexMultiTerm* terms, SIndexTerm* term) {
   taosArrayPush(terms, &term);
   return 0;
 }
-void indexMultiTermDestroy(SIndexMultiTerm *terms) {
+void indexMultiTermDestroy(SIndexMultiTerm* terms) {
   for (int32_t i = 0; i < taosArrayGetSize(terms); i++) {
-    SIndexTerm *p = taosArrayGetP(terms, i);
+    SIndexTerm* p = taosArrayGetP(terms, i);
     indexTermDestroy(p);
   }
   taosArrayDestroy(terms);
@@ -280,13 +292,13 @@ void indexMultiTermDestroy(SIndexMultiTerm *terms) {
 void indexInit() {
   // do nothing
 }
-static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *query, SArray **result) {
+static int indexTermSearch(SIndex* sIdx, SIndexTermQuery* query, SArray** result) {
   int32_t      version = -1;
   int16_t      colId = -1;
-  SIdxColInfo *colInfo = NULL;
+  SIdxColInfo* colInfo = NULL;
 
-  SIndexTerm *term = query->term;
-  const char *colName = term->colName;
+  SIndexTerm* term = query->term;
+  const char* colName = term->colName;
   int32_t     nColName = term->nColName;
 
   pthread_mutex_lock(&sIdx->mtx);
@@ -319,23 +331,24 @@ static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *query, SArray **result
   }
   return 0;
 }
-static void indexInterResultsDestroy(SArray *results) {
+static void indexInterResultsDestroy(SArray* results) {
   if (results == NULL) {
     return;
   }
 
   size_t sz = taosArrayGetSize(results);
   for (size_t i = 0; i < sz; i++) {
-    SArray *p = taosArrayGetP(results, i);
+    SArray* p = taosArrayGetP(results, i);
     taosArrayDestroy(p);
   }
   taosArrayDestroy(results);
 }
-static int indexMergeFinalResults(SArray *interResults, EIndexOperatorType oType, SArray *fResults) {
+static int indexMergeFinalResults(SArray* interResults, EIndexOperatorType oType, SArray* fResults) {
   // refactor, merge interResults into fResults by oType
-  SArray *first = taosArrayGetP(interResults, 0);
+  SArray* first = taosArrayGetP(interResults, 0);
   taosArraySort(first, uidCompare);
   taosArrayRemoveDuplicate(first, uidCompare, NULL);
+
   if (oType == MUST) {
     // just one column index, enhance later
     taosArrayAddAll(fResults, first);
@@ -350,7 +363,7 @@ static int indexMergeFinalResults(SArray *interResults, EIndexOperatorType oType
   }
   return 0;
 }
-static int indexMergeCacheIntoTindex(SIndex *sIdx) {
+static int indexMergeCacheIntoTindex(SIndex* sIdx) {
   if (sIdx == NULL) {
     return -1;
   }

source/libs/index/src/index_cache.c

@@ -21,13 +21,14 @@
 
 // ref index_cache.h:22
 #define CACHE_KEY_LEN(p) \
-  (sizeof(int32_t) + sizeof(uint16_t) + sizeof(p->colType) + sizeof(p->nColVal) + p->nColVal + sizeof(uint64_t) + \
-      sizeof(p->operType))
+  (sizeof(int32_t) + sizeof(uint16_t) + sizeof(p->colType) + sizeof(p->nColVal) + p->nColVal + sizeof(uint64_t) + sizeof(p->operType))
 
-static char *  getIndexKey(const void *pData) { return NULL; }
-static int32_t compareKey(const void *l, const void *r) {
-  char *lp = (char *)l;
-  char *rp = (char *)r;
+static char* getIndexKey(const void* pData) {
+  return NULL;
+}
+static int32_t compareKey(const void* l, const void* r) {
+  char* lp = (char*)l;
+  char* rp = (char*)r;
 
   // skip total len, not compare
   int32_t ll, rl;  // len
@@ -40,9 +41,7 @@ static int32_t compareKey(const void *l, const void *r) {
   int16_t lf, rf;  // field id
   memcpy(&lf, lp, sizeof(lf));
   memcpy(&rf, rp, sizeof(rf));
-  if (lf != rf) {
-    return lf < rf ? -1 : 1;
-  }
+  if (lf != rf) { return lf < rf ? -1 : 1; }
   lp += sizeof(lf);
   rp += sizeof(rf);
 
@@ -89,41 +88,41 @@ static int32_t compareKey(const void *l, const void *r) {
   int32_t lv, rv;
   memcpy(&lv, lp, sizeof(lv));
   memcpy(&rv, rp, sizeof(rv));
-  if (lv != rv) {
-    return lv > rv ? -1 : 1;
-  }
+  if (lv != rv) { return lv > rv ? -1 : 1; }
+
   lp += sizeof(lv);
   rp += sizeof(rv);
   // not care item type
 
   return 0;
 }
-IndexCache *indexCacheCreate() {
-  IndexCache *cache = calloc(1, sizeof(IndexCache));
-  cache->skiplist = tSkipListCreate(
-      MAX_SKIP_LIST_LEVEL, TSDB_DATA_TYPE_BINARY, MAX_INDEX_KEY_LEN, compareKey, SL_ALLOW_DUP_KEY, getIndexKey);
+IndexCache* indexCacheCreate() {
+  IndexCache* cache = calloc(1, sizeof(IndexCache));
+  if (cache == NULL) {
+    indexError("failed to create index cache");
+    return NULL;
+  }
+  cache->skiplist =
+      tSkipListCreate(MAX_SKIP_LIST_LEVEL, TSDB_DATA_TYPE_BINARY, MAX_INDEX_KEY_LEN, compareKey, SL_ALLOW_DUP_KEY, getIndexKey);
   return cache;
 }
 
-void indexCacheDestroy(void *cache) {
-  IndexCache *pCache = cache;
-  if (pCache == NULL) {
-    return;
-  }
+void indexCacheDestroy(void* cache) {
+  IndexCache* pCache = cache;
+  if (pCache == NULL) { return; }
   tSkipListDestroy(pCache->skiplist);
   free(pCache);
 }
 
-int indexCachePut(void *cache, SIndexTerm *term, int16_t colId, int32_t version, uint64_t uid) {
-  if (cache == NULL) {
-    return -1;
-  }
+int indexCachePut(void* cache, SIndexTerm* term, int16_t colId, int32_t version, uint64_t uid) {
+  if (cache == NULL) { return -1; }
 
-  IndexCache *pCache = cache;
+  IndexCache* pCache = cache;
   // encode data
   int32_t total = CACHE_KEY_LEN(term);
-  char *  buf = calloc(1, total);
-  char *  p = buf;
+
+  char* buf = calloc(1, total);
+  char* p = buf;
 
   SERIALIZE_VAR_TO_BUF(p, total, int32_t);
   SERIALIZE_VAR_TO_BUF(p, colId, int16_t);
@@ -137,30 +136,31 @@ int indexCachePut(void *cache, SIndexTerm *term, int16_t colId, int32_t version,
 
   SERIALIZE_MEM_TO_BUF(p, term, operType);
 
-  tSkipListPut(pCache->skiplist, (void *)buf);
+  tSkipListPut(pCache->skiplist, (void*)buf);
   return 0;
   // encode end
 }
-int indexCacheDel(void *cache, int32_t fieldId, const char *fieldValue, int32_t fvlen, uint64_t uid, int8_t operType) {
-  IndexCache *pCache = cache;
+int indexCacheDel(void* cache, int32_t fieldId, const char* fieldValue, int32_t fvlen, uint64_t uid, int8_t operType) {
+  IndexCache* pCache = cache;
   return 0;
 }
-int indexCacheSearch(
-    void *cache, SIndexTermQuery *query, int16_t colId, int32_t version, SArray *result, STermValueType *s) {
-  if (cache == NULL) {
-    return -1;
-  }
-  IndexCache *    pCache = cache;
-  SIndexTerm *    term = query->term;
+int indexCacheSearch(void* cache, SIndexTermQuery* query, int16_t colId, int32_t version, SArray* result, STermValueType* s) {
+  if (cache == NULL) { return -1; }
+  IndexCache*     pCache = cache;
+  SIndexTerm*     term = query->term;
   EIndexQueryType qtype = query->qType;
 
   int32_t keyLen = CACHE_KEY_LEN(term);
 
-  char *buf = calloc(1, keyLen);
+  char* buf = calloc(1, keyLen);
   if (qtype == QUERY_TERM) {
+    //
   } else if (qtype == QUERY_PREFIX) {
+    //
   } else if (qtype == QUERY_SUFFIX) {
+    //
   } else if (qtype == QUERY_REGEX) {
+    //
   }
 
   return 0;

source/libs/index/src/index_fst_automation.c

@@ -15,19 +15,17 @@
 
 #include "index_fst_automation.h"
 
-StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void *val) {
-  StartWithStateValue *nsv = calloc(1, sizeof(StartWithStateValue));
-  if (nsv == NULL) {
-    return NULL;
-  }
+StartWithStateValue* startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void* val) {
+  StartWithStateValue* nsv = calloc(1, sizeof(StartWithStateValue));
+  if (nsv == NULL) { return NULL; }
 
   nsv->kind = kind;
   nsv->type = ty;
   if (ty == FST_INT) {
-    nsv->val = *(int *)val;
+    nsv->val = *(int*)val;
   } else if (ty == FST_CHAR) {
-    size_t len = strlen((char *)val);
-    nsv->ptr = (char *)calloc(1, len + 1);
+    size_t len = strlen((char*)val);
+    nsv->ptr = (char*)calloc(1, len + 1);
     memcpy(nsv->ptr, val, len);
   } else if (ty == FST_ARRAY) {
     // TODO,
@@ -35,11 +33,9 @@ StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueTyp
   }
   return nsv;
 }
-void startWithStateValueDestroy(void *val) {
-  StartWithStateValue *sv = (StartWithStateValue *)val;
-  if (sv == NULL) {
-    return;
-  }
+void startWithStateValueDestroy(void* val) {
+  StartWithStateValue* sv = (StartWithStateValue*)val;
+  if (sv == NULL) { return; }
 
   if (sv->type == FST_INT) {
     //
@@ -50,11 +46,9 @@ void startWithStateValueDestroy(void *val) {
   }
   free(sv);
 }
-StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv) {
-  StartWithStateValue *nsv = calloc(1, sizeof(StartWithStateValue));
-  if (nsv == NULL) {
-    return NULL;
-  }
+StartWithStateValue* startWithStateValueDump(StartWithStateValue* sv) {
+  StartWithStateValue* nsv = calloc(1, sizeof(StartWithStateValue));
+  if (nsv == NULL) { return NULL; }
 
   nsv->kind = sv->kind;
   nsv->type = sv->type;
@@ -62,41 +56,40 @@ StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv) {
     nsv->val = sv->val;
   } else if (nsv->type == FST_CHAR) {
     size_t len = strlen(sv->ptr);
-    nsv->ptr = (char *)calloc(1, len + 1);
+    nsv->ptr = (char*)calloc(1, len + 1);
     memcpy(nsv->ptr, sv->ptr, len);
   } else if (nsv->type == FST_ARRAY) {
+    //
   }
   return nsv;
 }
 
 // prefix query, impl later
 
-static void *prefixStart(AutomationCtx *ctx) {
-  StartWithStateValue *data = (StartWithStateValue *)(ctx->stdata);
+static void* prefixStart(AutomationCtx* ctx) {
+  StartWithStateValue* data = (StartWithStateValue*)(ctx->stdata);
   return startWithStateValueDump(data);
 };
-static bool prefixIsMatch(AutomationCtx *ctx, void *sv) {
-  StartWithStateValue *ssv = (StartWithStateValue *)sv;
+static bool prefixIsMatch(AutomationCtx* ctx, void* sv) {
+  StartWithStateValue* ssv = (StartWithStateValue*)sv;
   return ssv->val == strlen(ctx->data);
 }
-static bool prefixCanMatch(AutomationCtx *ctx, void *sv) {
-  StartWithStateValue *ssv = (StartWithStateValue *)sv;
+static bool prefixCanMatch(AutomationCtx* ctx, void* sv) {
+  StartWithStateValue* ssv = (StartWithStateValue*)sv;
   return ssv->val >= 0;
 }
-static bool  prefixWillAlwaysMatch(AutomationCtx *ctx, void *state) { return true; }
-static void *prefixAccept(AutomationCtx *ctx, void *state, uint8_t byte) {
-  StartWithStateValue *ssv = (StartWithStateValue *)state;
-  if (ssv == NULL || ctx == NULL) {
-    return NULL;
-  }
+static bool prefixWillAlwaysMatch(AutomationCtx* ctx, void* state) {
+  return true;
+}
+static void* prefixAccept(AutomationCtx* ctx, void* state, uint8_t byte) {
+  StartWithStateValue* ssv = (StartWithStateValue*)state;
+  if (ssv == NULL || ctx == NULL) { return NULL; }
 
-  char *data = ctx->data;
-  if (ssv->kind == Done) {
-    return startWithStateValueCreate(Done, FST_INT, &ssv->val);
-  }
+  char* data = ctx->data;
+  if (ssv->kind == Done) { return startWithStateValueCreate(Done, FST_INT, &ssv->val); }
   if ((strlen(data) > ssv->val) && data[ssv->val] == byte) {
     int                  val = ssv->val + 1;
-    StartWithStateValue *nsv = startWithStateValueCreate(Running, FST_INT, &val);
+    StartWithStateValue* nsv = startWithStateValueCreate(Running, FST_INT, &val);
     if (prefixIsMatch(ctx, nsv)) {
       nsv->kind = Done;
     } else {
@@ -106,18 +99,32 @@ static void *prefixAccept(AutomationCtx *ctx, void *state, uint8_t byte) {
   }
   return NULL;
 }
-static void *prefixAcceptEof(AutomationCtx *ctx, void *state) { return NULL; }
+static void* prefixAcceptEof(AutomationCtx* ctx, void* state) {
+  return NULL;
+}
 
 // pattern query, impl later
 
-static void *patternStart(AutomationCtx *ctx) { return NULL; }
-static bool  patternIsMatch(AutomationCtx *ctx, void *data) { return true; }
-static bool  patternCanMatch(AutomationCtx *ctx, void *data) { return true; }
-static bool  patternWillAlwaysMatch(AutomationCtx *ctx, void *state) { return true; }
+static void* patternStart(AutomationCtx* ctx) {
+  return NULL;
+}
+static bool patternIsMatch(AutomationCtx* ctx, void* data) {
+  return true;
+}
+static bool patternCanMatch(AutomationCtx* ctx, void* data) {
+  return true;
+}
+static bool patternWillAlwaysMatch(AutomationCtx* ctx, void* state) {
+  return true;
+}
 
-static void *patternAccept(AutomationCtx *ctx, void *state, uint8_t byte) { return NULL; }
+static void* patternAccept(AutomationCtx* ctx, void* state, uint8_t byte) {
+  return NULL;
+}
 
-static void *patternAcceptEof(AutomationCtx *ctx, void *state) { return NULL; }
+static void* patternAcceptEof(AutomationCtx* ctx, void* state) {
+  return NULL;
+}
 
 AutomationFunc automFuncs[] = {
     {prefixStart, prefixIsMatch, prefixCanMatch, prefixWillAlwaysMatch, prefixAccept, prefixAcceptEof},
@@ -125,34 +132,32 @@ AutomationFunc automFuncs[] = {
     // add more search type
 };
 
-AutomationCtx *automCtxCreate(void *data, AutomationType atype) {
-  AutomationCtx *ctx = calloc(1, sizeof(AutomationCtx));
-  if (ctx == NULL) {
-    return NULL;
-  }
+AutomationCtx* automCtxCreate(void* data, AutomationType atype) {
+  AutomationCtx* ctx = calloc(1, sizeof(AutomationCtx));
+  if (ctx == NULL) { return NULL; }
 
-  StartWithStateValue *sv = NULL;
+  StartWithStateValue* sv = NULL;
   if (atype == AUTOMATION_PREFIX) {
     int val = 0;
     sv = startWithStateValueCreate(Running, FST_INT, &val);
-    ctx->stdata = (void *)sv;
+    ctx->stdata = (void*)sv;
   } else if (atype == AUTMMATION_MATCH) {
   } else {
     // add more search type
   }
 
-  char * src = (char *)data;
+  char*  src = (char*)data;
   size_t len = strlen(src);
-  char * dst = (char *)malloc(len * sizeof(char) + 1);
+  char*  dst = (char*)malloc(len * sizeof(char) + 1);
   memcpy(dst, src, len);
   dst[len] = 0;
 
   ctx->data = dst;
   ctx->type = atype;
-  ctx->stdata = (void *)sv;
+  ctx->stdata = (void*)sv;
   return ctx;
 }
-void automCtxDestroy(AutomationCtx *ctx) {
+void automCtxDestroy(AutomationCtx* ctx) {
   startWithStateValueDestroy(ctx->stdata);
   free(ctx->data);
   free(ctx);

source/libs/index/src/index_fst_common.c

@@ -274,260 +274,20 @@ const uint8_t COMMON_INPUTS[] = {
 };
 
 const char COMMON_INPUTS_INV[] = {
-    't',
-    'e',
-    '/',
-    'o',
-    'a',
-    's',
-    'r',
-    'i',
-    'p',
-    'c',
-    'n',
-    'w',
-    '.',
-    'h',
-    'l',
-    'm',
-    '-',
-    'd',
-    'u',
-    '0',
-    '1',
-    '2',
-    'g',
-    '=',
-    ':',
-    'b',
-    'f',
-    '3',
-    'y',
-    '5',
-    '&',
-    '_',
-    '4',
-    'v',
-    '9',
-    '6',
-    '7',
-    '8',
-    'k',
-    '%',
-    '?',
-    'x',
-    'C',
-    'D',
-    'A',
-    'S',
-    'F',
-    'I',
-    'B',
-    'E',
-    'j',
-    'P',
-    'T',
-    'z',
-    'R',
-    'N',
-    'M',
-    '+',
-    'L',
-    'O',
-    'q',
-    'H',
-    'G',
-    'W',
-    'U',
-    'V',
-    ',',
-    'Y',
-    'K',
-    'J',
-    'Z',
-    'X',
-    'Q',
-    ';',
-    ')',
-    '(',
-    '~',
-    '[',
-    ']',
-    '$',
-    '!',
-    '\'',
-    '*',
-    '@',
-    '\x00',
-    '\x01',
-    '\x02',
-    '\x03',
-    '\x04',
-    '\x05',
-    '\x06',
-    '\x07',
-    '\x08',
-    '\t',
-    '\n',
-    '\x0b',
-    '\x0c',
-    '\r',
-    '\x0e',
-    '\x0f',
-    '\x10',
-    '\x11',
-    '\x12',
-    '\x13',
-    '\x14',
-    '\x15',
-    '\x16',
-    '\x17',
-    '\x18',
-    '\x19',
-    '\x1a',
-    '\x1b',
-    '\x1c',
-    '\x1d',
-    '\x1e',
-    '\x1f',
-    ' ',
-    '"',
-    '#',
-    '<',
-    '>',
-    '\\',
-    '^',
-    '`',
-    '{',
-    '|',
-    '}',
-    '\x7f',
-    '\x80',
-    '\x81',
-    '\x82',
-    '\x83',
-    '\x84',
-    '\x85',
-    '\x86',
-    '\x87',
-    '\x88',
-    '\x89',
-    '\x8a',
-    '\x8b',
-    '\x8c',
-    '\x8d',
-    '\x8e',
-    '\x8f',
-    '\x90',
-    '\x91',
-    '\x92',
-    '\x93',
-    '\x94',
-    '\x95',
-    '\x96',
-    '\x97',
-    '\x98',
-    '\x99',
-    '\x9a',
-    '\x9b',
-    '\x9c',
-    '\x9d',
-    '\x9e',
-    '\x9f',
-    '\xa0',
-    '\xa1',
-    '\xa2',
-    '\xa3',
-    '\xa4',
-    '\xa5',
-    '\xa6',
-    '\xa7',
-    '\xa8',
-    '\xa9',
-    '\xaa',
-    '\xab',
-    '\xac',
-    '\xad',
-    '\xae',
-    '\xaf',
-    '\xb0',
-    '\xb1',
-    '\xb2',
-    '\xb3',
-    '\xb4',
-    '\xb5',
-    '\xb6',
-    '\xb7',
-    '\xb8',
-    '\xb9',
-    '\xba',
-    '\xbb',
-    '\xbc',
-    '\xbd',
-    '\xbe',
-    '\xbf',
-    '\xc0',
-    '\xc1',
-    '\xc2',
-    '\xc3',
-    '\xc4',
-    '\xc5',
-    '\xc6',
-    '\xc7',
-    '\xc8',
-    '\xc9',
-    '\xca',
-    '\xcb',
-    '\xcc',
-    '\xcd',
-    '\xce',
-    '\xcf',
-    '\xd0',
-    '\xd1',
-    '\xd2',
-    '\xd3',
-    '\xd4',
-    '\xd5',
-    '\xd6',
-    '\xd7',
-    '\xd8',
-    '\xd9',
-    '\xda',
-    '\xdb',
-    '\xdc',
-    '\xdd',
-    '\xde',
-    '\xdf',
-    '\xe0',
-    '\xe1',
-    '\xe2',
-    '\xe3',
-    '\xe4',
-    '\xe5',
-    '\xe6',
-    '\xe7',
-    '\xe8',
-    '\xe9',
-    '\xea',
-    '\xeb',
-    '\xec',
-    '\xed',
-    '\xee',
-    '\xef',
-    '\xf0',
-    '\xf1',
-    '\xf2',
-    '\xf3',
-    '\xf4',
-    '\xf5',
-    '\xf6',
-    '\xf7',
-    '\xf8',
-    '\xf9',
-    '\xfa',
-    '\xfb',
-    '\xfc',
-    '\xfd',
-    '\xfe',
-    '\xff',
+    't',    'e',    '/',    'o',    'a',    's',    'r',    'i',    'p',    'c',    'n',    'w',    '.',    'h',    'l',    'm',
+    '-',    'd',    'u',    '0',    '1',    '2',    'g',    '=',    ':',    'b',    'f',    '3',    'y',    '5',    '&',    '_',
+    '4',    'v',    '9',    '6',    '7',    '8',    'k',    '%',    '?',    'x',    'C',    'D',    'A',    'S',    'F',    'I',
+    'B',    'E',    'j',    'P',    'T',    'z',    'R',    'N',    'M',    '+',    'L',    'O',    'q',    'H',    'G',    'W',
+    'U',    'V',    ',',    'Y',    'K',    'J',    'Z',    'X',    'Q',    ';',    ')',    '(',    '~',    '[',    ']',    '$',
+    '!',    '\'',   '*',    '@',    '\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07', '\x08', '\t',   '\n',   '\x0b',
+    '\x0c', '\r',   '\x0e', '\x0f', '\x10', '\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17', '\x18', '\x19', '\x1a', '\x1b',
+    '\x1c', '\x1d', '\x1e', '\x1f', ' ',    '"',    '#',    '<',    '>',    '\\',   '^',    '`',    '{',    '|',    '}',    '\x7f',
+    '\x80', '\x81', '\x82', '\x83', '\x84', '\x85', '\x86', '\x87', '\x88', '\x89', '\x8a', '\x8b', '\x8c', '\x8d', '\x8e', '\x8f',
+    '\x90', '\x91', '\x92', '\x93', '\x94', '\x95', '\x96', '\x97', '\x98', '\x99', '\x9a', '\x9b', '\x9c', '\x9d', '\x9e', '\x9f',
+    '\xa0', '\xa1', '\xa2', '\xa3', '\xa4', '\xa5', '\xa6', '\xa7', '\xa8', '\xa9', '\xaa', '\xab', '\xac', '\xad', '\xae', '\xaf',
+    '\xb0', '\xb1', '\xb2', '\xb3', '\xb4', '\xb5', '\xb6', '\xb7', '\xb8', '\xb9', '\xba', '\xbb', '\xbc', '\xbd', '\xbe', '\xbf',
+    '\xc0', '\xc1', '\xc2', '\xc3', '\xc4', '\xc5', '\xc6', '\xc7', '\xc8', '\xc9', '\xca', '\xcb', '\xcc', '\xcd', '\xce', '\xcf',
+    '\xd0', '\xd1', '\xd2', '\xd3', '\xd4', '\xd5', '\xd6', '\xd7', '\xd8', '\xd9', '\xda', '\xdb', '\xdc', '\xdd', '\xde', '\xdf',
+    '\xe0', '\xe1', '\xe2', '\xe3', '\xe4', '\xe5', '\xe6', '\xe7', '\xe8', '\xe9', '\xea', '\xeb', '\xec', '\xed', '\xee', '\xef',
+    '\xf0', '\xf1', '\xf2', '\xf3', '\xf4', '\xf5', '\xf6', '\xf7', '\xf8', '\xf9', '\xfa', '\xfb', '\xfc', '\xfd', '\xfe', '\xff',
 };

source/libs/index/src/index_fst_counting_writer.c

@@ -17,10 +17,8 @@
 #include "index_fst_util.h"
 #include "tutil.h"
 
-static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
-  if (ctx->offset + len > ctx->limit) {
-    return -1;
-  }
+static int writeCtxDoWrite(WriterCtx* ctx, uint8_t* buf, int len) {
+  if (ctx->offset + len > ctx->limit) { return -1; }
 
   if (ctx->type == TFile) {
     assert(len == tfWrite(ctx->file.fd, buf, len));
@@ -30,7 +28,7 @@ static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
   ctx->offset += len;
   return len;
 }
-static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
+static int writeCtxDoRead(WriterCtx* ctx, uint8_t* buf, int len) {
   int nRead = 0;
   if (ctx->type == TFile) {
     nRead = tfRead(ctx->file.fd, buf, len);
@@ -41,7 +39,18 @@ static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
 
   return nRead;
 }
-static int writeCtxDoFlush(WriterCtx *ctx) {
+static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t offset) {
+  int nRead = 0;
+  if (ctx->type == TFile) {
+    tfLseek(ctx->file.fd, offset, 0);
+    nRead = tfRead(ctx->file.fd, buf, len);
+  } else {
+    // refactor later
+    assert(0);
+  }
+  return nRead;
+}
+static int writeCtxDoFlush(WriterCtx* ctx) {
   if (ctx->type == TFile) {
     // tfFsync(ctx->fd);
     // tfFlush(ctx->file.fd);
@@ -51,11 +60,9 @@ static int writeCtxDoFlush(WriterCtx *ctx) {
   return 1;
 }
 
-WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity) {
-  WriterCtx *ctx = calloc(1, sizeof(WriterCtx));
-  if (ctx == NULL) {
-    return NULL;
-  }
+WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity) {
+  WriterCtx* ctx = calloc(1, sizeof(WriterCtx));
+  if (ctx == NULL) { return NULL; }
 
   ctx->type = type;
   if (ctx->type == TFile) {
@@ -67,8 +74,8 @@ WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int
       ctx->file.fd = tfOpenReadWrite(tmpFile);
     }
     if (ctx->file.fd < 0) {
-      goto END;
       indexError("open file error %d", errno);
+      goto END;
     }
   } else if (ctx->type == TMemory) {
     ctx->mem.buf = calloc(1, sizeof(char) * capacity);
@@ -77,18 +84,17 @@ WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int
   ctx->write = writeCtxDoWrite;
   ctx->read = writeCtxDoRead;
   ctx->flush = writeCtxDoFlush;
+  ctx->readFrom = writeCtxDoReadFrom;
 
   ctx->offset = 0;
   ctx->limit = capacity;
 
   return ctx;
 END:
-  if (ctx->type == TMemory) {
-    free(ctx->mem.buf);
-  }
+  if (ctx->type == TMemory) { free(ctx->mem.buf); }
   free(ctx);
 }
-void writerCtxDestroy(WriterCtx *ctx) {
+void writerCtxDestroy(WriterCtx* ctx) {
   if (ctx->type == TMemory) {
     free(ctx->mem.buf);
   } else {
@@ -97,57 +103,53 @@ void writerCtxDestroy(WriterCtx *ctx) {
   free(ctx);
 }
 
-FstCountingWriter *fstCountingWriterCreate(void *wrt) {
-  FstCountingWriter *cw = calloc(1, sizeof(FstCountingWriter));
-  if (cw == NULL) {
-    return NULL;
-  }
+FstCountingWriter* fstCountingWriterCreate(void* wrt) {
+  FstCountingWriter* cw = calloc(1, sizeof(FstCountingWriter));
+  if (cw == NULL) { return NULL; }
 
   cw->wrt = wrt;
   //(void *)(writerCtxCreate(TFile, readOnly));
   return cw;
 }
-void fstCountingWriterDestroy(FstCountingWriter *cw) {
+void fstCountingWriterDestroy(FstCountingWriter* cw) {
   // free wrt object: close fd or free mem
   fstCountingWriterFlush(cw);
   // writerCtxDestroy((WriterCtx *)(cw->wrt));
   free(cw);
 }
 
-int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t len) {
-  if (write == NULL) {
-    return 0;
-  }
+int fstCountingWriterWrite(FstCountingWriter* write, uint8_t* buf, uint32_t len) {
+  if (write == NULL) { return 0; }
   // update checksum
   // write data to file/socket or mem
-  WriterCtx *ctx = write->wrt;
+  WriterCtx* ctx = write->wrt;
 
   int nWrite = ctx->write(ctx, buf, len);
   assert(nWrite == len);
   write->count += len;
   return len;
 }
-int fstCountingWriterRead(FstCountingWriter *write, uint8_t *buf, uint32_t len) {
-  if (write == NULL) {
-    return 0;
-  }
-  WriterCtx *ctx = write->wrt;
+int fstCountingWriterRead(FstCountingWriter* write, uint8_t* buf, uint32_t len) {
+  if (write == NULL) { return 0; }
+  WriterCtx* ctx = write->wrt;
   int        nRead = ctx->read(ctx, buf, len);
   // assert(nRead == len);
   return nRead;
 }
 
-uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write) { return 0; }
-int      fstCountingWriterFlush(FstCountingWriter *write) {
-  WriterCtx *ctx = write->wrt;
+uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write) {
+  return 0;
+}
+int fstCountingWriterFlush(FstCountingWriter* write) {
+  WriterCtx* ctx = write->wrt;
   ctx->flush(ctx);
   // write->wtr->flush
   return 1;
 }
 
-void fstCountingWriterPackUintIn(FstCountingWriter *writer, uint64_t n, uint8_t nBytes) {
+void fstCountingWriterPackUintIn(FstCountingWriter* writer, uint64_t n, uint8_t nBytes) {
   assert(1 <= nBytes && nBytes <= 8);
-  uint8_t *buf = calloc(8, sizeof(uint8_t));
+  uint8_t* buf = calloc(8, sizeof(uint8_t));
   for (uint8_t i = 0; i < nBytes; i++) {
     buf[i] = (uint8_t)n;
     n = n >> 8;
@@ -157,7 +159,7 @@ void fstCountingWriterPackUintIn(FstCountingWriter *writer, uint64_t n, uint8_t
   return;
 }
 
-uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n) {
+uint8_t fstCountingWriterPackUint(FstCountingWriter* writer, uint64_t n) {
   uint8_t nBytes = packSize(n);
   fstCountingWriterPackUintIn(writer, n, nBytes);
   return nBytes;

source/libs/index/src/index_fst_node.c

@@ -14,60 +14,46 @@
  */
 #include "index_fst_node.h"
 
-FstBuilderNode *fstBuilderNodeDefault() {
-  FstBuilderNode *bn = malloc(sizeof(FstBuilderNode));
+FstBuilderNode* fstBuilderNodeDefault() {
+  FstBuilderNode* bn = malloc(sizeof(FstBuilderNode));
   bn->isFinal = false;
   bn->finalOutput = 0;
   bn->trans = taosArrayInit(16, sizeof(FstTransition));
   return bn;
 }
-void fstBuilderNodeDestroy(FstBuilderNode *node) {
-  if (node == NULL) {
-    return;
-  }
+void fstBuilderNodeDestroy(FstBuilderNode* node) {
+  if (node == NULL) { return; }
 
   taosArrayDestroy(node->trans);
   free(node);
 }
 
-bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2) {
-  if (n1 == n2) {
-    return true;
-  }
-  if (n1 == NULL || n2 == NULL) {
-    return false;
-  }
+bool fstBuilderNodeEqual(FstBuilderNode* n1, FstBuilderNode* n2) {
+  if (n1 == n2) { return true; }
+  if (n1 == NULL || n2 == NULL) { return false; }
 
-  if (n1->isFinal != n2->isFinal || n1->finalOutput != n2->finalOutput) {
-    return false;
-  }
+  if (n1->isFinal != n2->isFinal || n1->finalOutput != n2->finalOutput) { return false; }
   size_t s1 = n1->trans ? taosArrayGetSize(n1->trans) : 0;
   size_t s2 = n2->trans ? taosArrayGetSize(n2->trans) : 0;
-  if (s1 != s2) {
-    return false;
-  }
+  if (s1 != s2) { return false; }
   for (size_t i = 0; i < s1; i++) {
-    FstTransition *t1 = taosArrayGet(n1->trans, i);
-    FstTransition *t2 = taosArrayGet(n2->trans, i);
-    if (t1->inp != t2->inp || t1->out != t2->out || t1->addr != t2->addr) {
-      return false;
-    }
+    FstTransition* t1 = taosArrayGet(n1->trans, i);
+    FstTransition* t2 = taosArrayGet(n2->trans, i);
+    if (t1->inp != t2->inp || t1->out != t2->out || t1->addr != t2->addr) { return false; }
   }
 
   return true;
 }
-FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src) {
-  FstBuilderNode *node = malloc(sizeof(FstBuilderNode));
-  if (node == NULL) {
-    return NULL;
-  }
+FstBuilderNode* fstBuilderNodeClone(FstBuilderNode* src) {
+  FstBuilderNode* node = malloc(sizeof(FstBuilderNode));
+  if (node == NULL) { return NULL; }
 
   //
   size_t  sz = taosArrayGetSize(src->trans);
-  SArray *trans = taosArrayInit(sz, sizeof(FstTransition));
+  SArray* trans = taosArrayInit(sz, sizeof(FstTransition));
 
   for (size_t i = 0; i < sz; i++) {
-    FstTransition *tran = taosArrayGet(src->trans, i);
+    FstTransition* tran = taosArrayGet(src->trans, i);
     taosArrayPush(trans, tran);
   }
 
@@ -77,10 +63,8 @@ FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src) {
   return node;
 }
 // not destroy src, User's bussiness
-void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src) {
-  if (dst == NULL || src == NULL) {
-    return;
-  }
+void fstBuilderNodeCloneFrom(FstBuilderNode* dst, FstBuilderNode* src) {
+  if (dst == NULL || src == NULL) { return; }
 
   dst->isFinal = src->isFinal;
   dst->finalOutput = src->finalOutput;
@@ -90,7 +74,7 @@ void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src) {
   size_t sz = taosArrayGetSize(src->trans);
   dst->trans = taosArrayInit(sz, sizeof(FstTransition));
   for (size_t i = 0; i < sz; i++) {
-    FstTransition *trn = taosArrayGet(src->trans, i);
+    FstTransition* trn = taosArrayGet(src->trans, i);
     taosArrayPush(dst->trans, trn);
   }
 }

source/libs/index/src/index_fst_registry.c

@@ -15,7 +15,7 @@
 
 #include "index_fst_registry.h"
 
-uint64_t fstRegistryHash(FstRegistry *registry, FstBuilderNode *bNode) {
+uint64_t fstRegistryHash(FstRegistry* registry, FstBuilderNode* bNode) {
   // TODO(yihaoDeng): refactor later
   const uint64_t FNV_PRIME = 1099511628211;
   uint64_t       h = 14695981039346656037u;
@@ -25,21 +25,19 @@ uint64_t fstRegistryHash(FstRegistry *registry, FstBuilderNode *bNode) {
 
   uint32_t sz = (uint32_t)taosArrayGetSize(bNode->trans);
   for (uint32_t i = 0; i < sz; i++) {
-    FstTransition *trn = taosArrayGet(bNode->trans, i);
+    FstTransition* trn = taosArrayGet(bNode->trans, i);
     h = (h ^ (uint64_t)(trn->inp)) * FNV_PRIME;
     h = (h ^ (uint64_t)(trn->out)) * FNV_PRIME;
     h = (h ^ (uint64_t)(trn->addr)) * FNV_PRIME;
   }
   return h % (registry->tableSize);
 }
-static void fstRegistryCellSwap(SArray *arr, uint32_t a, uint32_t b) {
+static void fstRegistryCellSwap(SArray* arr, uint32_t a, uint32_t b) {
   size_t sz = taosArrayGetSize(arr);
-  if (a >= sz || b >= sz) {
-    return;
-  }
+  if (a >= sz || b >= sz) { return; }
 
-  FstRegistryCell *cell1 = (FstRegistryCell *)taosArrayGet(arr, a);
-  FstRegistryCell *cell2 = (FstRegistryCell *)taosArrayGet(arr, b);
+  FstRegistryCell* cell1 = (FstRegistryCell*)taosArrayGet(arr, a);
+  FstRegistryCell* cell2 = (FstRegistryCell*)taosArrayGet(arr, b);
 
   FstRegistryCell t = {.addr = cell1->addr, .node = cell1->node};
 
@@ -51,11 +49,9 @@ static void fstRegistryCellSwap(SArray *arr, uint32_t a, uint32_t b) {
   return;
 }
 
-static void fstRegistryCellPromote(SArray *arr, uint32_t start, uint32_t end) {
+static void fstRegistryCellPromote(SArray* arr, uint32_t start, uint32_t end) {
   size_t sz = taosArrayGetSize(arr);
-  if (start >= sz && end >= sz) {
-    return;
-  }
+  if (start >= sz && end >= sz) { return; }
 
   assert(start >= end);
 
@@ -67,14 +63,12 @@ static void fstRegistryCellPromote(SArray *arr, uint32_t start, uint32_t end) {
   }
 }
 
-FstRegistry *fstRegistryCreate(uint64_t tableSize, uint64_t mruSize) {
-  FstRegistry *registry = malloc(sizeof(FstRegistry));
-  if (registry == NULL) {
-    return NULL;
-  }
+FstRegistry* fstRegistryCreate(uint64_t tableSize, uint64_t mruSize) {
+  FstRegistry* registry = malloc(sizeof(FstRegistry));
+  if (registry == NULL) { return NULL; }
 
   uint64_t nCells = tableSize * mruSize;
-  SArray * tb = (SArray *)taosArrayInit(nCells, sizeof(FstRegistryCell));
+  SArray*  tb = (SArray*)taosArrayInit(nCells, sizeof(FstRegistryCell));
   if (NULL == tb) {
     free(registry);
     return NULL;
@@ -91,32 +85,28 @@ FstRegistry *fstRegistryCreate(uint64_t tableSize, uint64_t mruSize) {
   return registry;
 }
 
-void fstRegistryDestroy(FstRegistry *registry) {
-  if (registry == NULL) {
-    return;
-  }
+void fstRegistryDestroy(FstRegistry* registry) {
+  if (registry == NULL) { return; }
 
-  SArray *tb = registry->table;
+  SArray* tb = registry->table;
   size_t  sz = taosArrayGetSize(tb);
   for (size_t i = 0; i < sz; i++) {
-    FstRegistryCell *cell = taosArrayGet(tb, i);
+    FstRegistryCell* cell = taosArrayGet(tb, i);
     fstBuilderNodeDestroy(cell->node);
   }
   taosArrayDestroy(tb);
   free(registry);
 }
 
-FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNode) {
-  if (taosArrayGetSize(registry->table) <= 0) {
-    return NULL;
-  }
+FstRegistryEntry* fstRegistryGetEntry(FstRegistry* registry, FstBuilderNode* bNode) {
+  if (taosArrayGetSize(registry->table) <= 0) { return NULL; }
   uint64_t bucket = fstRegistryHash(registry, bNode);
   uint64_t start = registry->mruSize * bucket;
   uint64_t end = start + registry->mruSize;
 
-  FstRegistryEntry *entry = malloc(sizeof(FstRegistryEntry));
+  FstRegistryEntry* entry = malloc(sizeof(FstRegistryEntry));
   if (end - start == 1) {
-    FstRegistryCell *cell = taosArrayGet(registry->table, start);
+    FstRegistryCell* cell = taosArrayGet(registry->table, start);
     // cell->isNode &&
     if (cell->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell->node, bNode)) {
       entry->state = FOUND;
@@ -128,13 +118,13 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
       entry->cell = cell;  // copy or not
     }
   } else if (end - start == 2) {
-    FstRegistryCell *cell1 = taosArrayGet(registry->table, start);
+    FstRegistryCell* cell1 = taosArrayGet(registry->table, start);
     if (cell1->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell1->node, bNode)) {
       entry->state = FOUND;
       entry->addr = cell1->addr;
       return entry;
     }
-    FstRegistryCell *cell2 = taosArrayGet(registry->table, start + 1);
+    FstRegistryCell* cell2 = taosArrayGet(registry->table, start + 1);
     if (cell2->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell2->node, bNode)) {
       entry->state = FOUND;
       entry->addr = cell2->addr;
@@ -146,13 +136,13 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
     fstBuilderNodeCloneFrom(cell2->node, bNode);
 
     fstRegistryCellSwap(registry->table, start, start + 1);
-    FstRegistryCell *cCell = taosArrayGet(registry->table, start);
+    FstRegistryCell* cCell = taosArrayGet(registry->table, start);
     entry->state = NOTFOUND;
     entry->cell = cCell;
   } else {
     uint32_t i = start;
     for (; i < end; i++) {
-      FstRegistryCell *cell = (FstRegistryCell *)taosArrayGet(registry->table, i);
+      FstRegistryCell* cell = (FstRegistryCell*)taosArrayGet(registry->table, i);
       if (cell->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell->node, bNode)) {
         entry->state = FOUND;
         entry->addr = cell->addr;
@@ -162,16 +152,18 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
     }
     if (i >= end) {
       uint64_t         last = end - 1;
-      FstRegistryCell *cell = (FstRegistryCell *)taosArrayGet(registry->table, last);
+      FstRegistryCell* cell = (FstRegistryCell*)taosArrayGet(registry->table, last);
       // clone from bNode, refactor later
       fstBuilderNodeCloneFrom(cell->node, bNode);
 
       fstRegistryCellPromote(registry->table, last, start);
-      FstRegistryCell *cCell = taosArrayGet(registry->table, start);
+      FstRegistryCell* cCell = taosArrayGet(registry->table, start);
       entry->state = NOTFOUND;
       entry->cell = cCell;
     }
   }
   return entry;
 }
-void fstRegistryEntryDestroy(FstRegistryEntry *entry) { free(entry); }
+void fstRegistryEntryDestroy(FstRegistryEntry* entry) {
+  free(entry);
+}

source/libs/index/src/index_fst_util.c

@@ -61,9 +61,10 @@ uint8_t packSize(uint64_t n) {
   }
 }
 
-uint64_t unpackUint64(uint8_t *ch, uint8_t sz) {
+uint64_t unpackUint64(uint8_t* ch, uint8_t sz) {
   uint64_t n = 0;
   for (uint8_t i = 0; i < sz; i++) {
+    //
     n = n | (ch[i] << (8 * i));
   }
   return n;
@@ -75,7 +76,7 @@ uint8_t packDeltaSize(CompiledAddr nodeAddr, CompiledAddr transAddr) {
     return packSize(nodeAddr - transAddr);
   }
 }
-CompiledAddr unpackDelta(char *data, uint64_t len, uint64_t nodeAddr) {
+CompiledAddr unpackDelta(char* data, uint64_t len, uint64_t nodeAddr) {
   uint64_t delta = unpackUint64(data, len);
   // delta_add = u64_to_usize
   if (delta == EMPTY_ADDRESS) {
@@ -88,8 +89,8 @@ CompiledAddr unpackDelta(char *data, uint64_t len, uint64_t nodeAddr) {
 // fst slice func
 //
 
-FstSlice fstSliceCreate(uint8_t *data, uint64_t len) {
-  FstString *str = (FstString *)malloc(sizeof(FstString));
+FstSlice fstSliceCreate(uint8_t* data, uint64_t len) {
+  FstString* str = (FstString*)malloc(sizeof(FstString));
   str->ref = 1;
   str->len = len;
   str->data = malloc(len * sizeof(uint8_t));
@@ -99,8 +100,8 @@ FstSlice fstSliceCreate(uint8_t *data, uint64_t len) {
   return s;
 }
 // just shallow copy
-FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end) {
-  FstString *str = s->str;
+FstSlice fstSliceCopy(FstSlice* s, int32_t start, int32_t end) {
+  FstString* str = s->str;
   str->ref++;
   // uint8_t *buf = fstSliceData(s, &alen);
   // start = buf + start - (buf - s->start);
@@ -109,16 +110,16 @@ FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end) {
   FstSlice t = {.str = str, .start = start + s->start, .end = end + s->start};
   return t;
 }
-FstSlice fstSliceDeepCopy(FstSlice *s, int32_t start, int32_t end) {
+FstSlice fstSliceDeepCopy(FstSlice* s, int32_t start, int32_t end) {
   int32_t  tlen = end - start + 1;
   int32_t  slen;
-  uint8_t *data = fstSliceData(s, &slen);
+  uint8_t* data = fstSliceData(s, &slen);
   assert(tlen <= slen);
 
-  uint8_t *buf = malloc(sizeof(uint8_t) * tlen);
+  uint8_t* buf = malloc(sizeof(uint8_t) * tlen);
   memcpy(buf, data + start, tlen);
 
-  FstString *str = malloc(sizeof(FstString));
+  FstString* str = malloc(sizeof(FstString));
   str->data = buf;
   str->len = tlen;
   str->ref = 1;
@@ -129,17 +130,17 @@ FstSlice fstSliceDeepCopy(FstSlice *s, int32_t start, int32_t end) {
   ans.end = tlen - 1;
   return ans;
 }
-bool fstSliceIsEmpty(FstSlice *s) { return s->str == NULL || s->str->len == 0 || s->start < 0 || s->end < 0; }
+bool fstSliceIsEmpty(FstSlice* s) {
+  return s->str == NULL || s->str->len == 0 || s->start < 0 || s->end < 0;
+}
 
-uint8_t *fstSliceData(FstSlice *s, int32_t *size) {
-  FstString *str = s->str;
-  if (size != NULL) {
-    *size = s->end - s->start + 1;
-  }
+uint8_t* fstSliceData(FstSlice* s, int32_t* size) {
+  FstString* str = s->str;
+  if (size != NULL) { *size = s->end - s->start + 1; }
   return str->data + s->start;
 }
-void fstSliceDestroy(FstSlice *s) {
-  FstString *str = s->str;
+void fstSliceDestroy(FstSlice* s) {
+  FstString* str = s->str;
   str->ref--;
   if (str->ref <= 0) {
     free(str->data);
@@ -148,10 +149,10 @@ void fstSliceDestroy(FstSlice *s) {
   }
 }
 
-int fstSliceCompare(FstSlice *a, FstSlice *b) {
+int fstSliceCompare(FstSlice* a, FstSlice* b) {
   int32_t  alen, blen;
-  uint8_t *aBuf = fstSliceData(a, &alen);
-  uint8_t *bBuf = fstSliceData(b, &blen);
+  uint8_t* aBuf = fstSliceData(a, &alen);
+  uint8_t* bBuf = fstSliceData(b, &blen);
 
   uint32_t i, j;
   for (i = 0, j = 0; i < alen && j < blen; i++, j++) {

source/libs/index/src/index_tfile.c

@@ -21,107 +21,78 @@
 #include "index_fst_counting_writer.h"
 #include "index_util.h"
 #include "taosdef.h"
+#include "tcompare.h"
 
-static FORCE_INLINE int tfileReadLoadHeader(TFileReader *reader) {
-  // TODO simple tfile header later
-  char             buf[TFILE_HADER_PRE_SIZE];
-  char *           p = buf;
-  TFileReadHeader *header = &reader->header;
-  int64_t          nread = reader->ctx->read(reader->ctx, buf, TFILE_HADER_PRE_SIZE);
-  assert(nread == TFILE_HADER_PRE_SIZE);
+#define TF_TABLE_TATOAL_SIZE(sz) (sizeof(sz) + sz * sizeof(uint64_t))
 
-  memcpy(&header->suid, p, sizeof(header->suid));
-  p += sizeof(header->suid);
+typedef struct TFileValue {
+  char*   colVal;  // null terminated
+  SArray* tableId;
+  int32_t offset;
+} TFileValue;
 
-  memcpy(&header->version, p, sizeof(header->version));
-  p += sizeof(header->version);
+static int  tfileValueCompare(const void* a, const void* b, const void* param);
+static void tfileSerialTableIdsToBuf(char* buf, SArray* tableIds);
 
-  int32_t colLen = 0;
-  memcpy(&colLen, p, sizeof(colLen));
-  assert(colLen < sizeof(header->colName));
-  nread = reader->ctx->read(reader->ctx, header->colName, colLen);
-  assert(nread == colLen);
+static int tfileWriteHeader(TFileWriter* writer);
+static int tfileWriteFstOffset(TFileWriter* tw, int32_t offset);
+static int tfileWriteData(TFileWriter* write, TFileValue* tval);
 
-  nread = reader->ctx->read(reader->ctx, &header->colType, sizeof(header->colType));
-  return 0;
-};
-static int tfileGetFileList(const char *path, SArray *result) {
-  DIR *dir = opendir(path);
-  if (NULL == dir) {
-    return -1;
-  }
+static int tfileReadLoadHeader(TFileReader* reader);
+static int tfileReadLoadFst(TFileReader* reader);
+static int tfileReadLoadTableIds(TFileReader* reader, int32_t offset, SArray* result);
 
-  struct dirent *entry;
-  while ((entry = readdir(dir)) != NULL) {
-    size_t len = strlen(entry->d_name);
-    char * buf = calloc(1, len + 1);
-    memcpy(buf, entry->d_name, len);
-    taosArrayPush(result, &buf);
-  }
-  closedir(dir);
-  return 0;
-}
-static void tfileDestroyFileName(void *elem) {
-  char *p = *(char **)elem;
-  free(p);
-}
-static int tfileCompare(const void *a, const void *b) {
-  const char *aName = *(char **)a;
-  const char *bName = *(char **)b;
-  size_t      aLen = strlen(aName);
-  size_t      bLen = strlen(bName);
-  return strncmp(aName, bName, aLen > bLen ? aLen : bLen);
-}
-// tfile name suid-colId-version.tindex
-static int tfileParseFileName(const char *filename, uint64_t *suid, int *colId, int *version) {
-  if (3 == sscanf(filename, "%" PRIu64 "-%d-%d.tindex", suid, colId, version)) {
-    // read suid & colid & version  success
-    return 0;
-  }
-  return -1;
-}
-static void tfileSerialCacheKey(TFileCacheKey *key, char *buf) {
-  SERIALIZE_MEM_TO_BUF(buf, key, suid);
-  SERIALIZE_VAR_TO_BUF(buf, '_', char);
-  SERIALIZE_MEM_TO_BUF(buf, key, colType);
-  SERIALIZE_VAR_TO_BUF(buf, '_', char);
-  SERIALIZE_MEM_TO_BUF(buf, key, version);
-  SERIALIZE_VAR_TO_BUF(buf, '_', char);
-  SERIALIZE_STR_MEM_TO_BUF(buf, key, colName, key->nColName);
-}
+static int  tfileGetFileList(const char* path, SArray* result);
+static void tfileDestroyFileName(void* elem);
+static int  tfileCompare(const void* a, const void* b);
+static int  tfileParseFileName(const char* filename, uint64_t* suid, int* colId, int* version);
+static void tfileSerialCacheKey(TFileCacheKey* key, char* buf);
 
-TFileCache *tfileCacheCreate(const char *path) {
-  TFileCache *tcache = calloc(1, sizeof(TFileCache));
-  if (tcache == NULL) {
-    return NULL;
-  }
+TFileCache* tfileCacheCreate(const char* path) {
+  TFileCache* tcache = calloc(1, sizeof(TFileCache));
+  if (tcache == NULL) { return NULL; }
 
   tcache->tableCache = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   tcache->capacity = 64;
 
-  SArray *files = taosArrayInit(4, sizeof(void *));
+  SArray* files = taosArrayInit(4, sizeof(void*));
   tfileGetFileList(path, files);
   taosArraySort(files, tfileCompare);
-  for (size_t i = 0; i < taosArrayGetSize(files); i++) {
-    char *   file = taosArrayGetP(files, i);
   uint64_t suid;
-    int      colId, version;
-    if (0 != tfileParseFileName(file, &suid, &colId, &version)) {
-      goto End;
+  int32_t  colId, version;
+  for (size_t i = 0; i < taosArrayGetSize(files); i++) {
+    char* file = taosArrayGetP(files, i);
+    if (0 != tfileParseFileName(file, &suid, (int*)&colId, (int*)&version)) {
+      indexInfo("try parse invalid file:  %s, skip it", file);
       continue;
     }
-
-    WriterCtx *wc = writerCtxCreate(TFile, file, true, 1024 * 64);
+    WriterCtx* wc = writerCtxCreate(TFile, file, true, 1024 * 64);
     if (wc == NULL) {
       indexError("failed to open index:  %s", file);
       goto End;
     }
-    TFileReader *reader = tfileReaderCreate(wc);
+    TFileReader* reader = tfileReaderCreate(wc);
     if (0 != tfileReadLoadHeader(reader)) {
-      TFileReaderDestroy(reader);
-      indexError("failed to load index header, index Id: %s", file);
+      tfileReaderDestroy(reader);
+      indexError("failed to load index header, index file: %s", file);
       goto End;
     }
+    if (0 != tfileReadLoadFst(reader)) {
+      tfileReaderDestroy(reader);
+      indexError("failed to load index fst, index file: %s", file);
+    }
+    // loader fst and validate it
+
+    TFileHeader*  header = &reader->header;
+    TFileCacheKey key = {.suid = header->suid,
+                         .version = header->version,
+                         .colName = header->colName,
+                         .nColName = strlen(header->colName),
+                         .colType = header->colType};
+
+    char buf[128] = {0};
+    tfileSerialCacheKey(&key, buf);
+    taosHashPut(tcache->tableCache, buf, strlen(buf), &reader, sizeof(void*));
   }
   taosArrayDestroyEx(files, tfileDestroyFileName);
   return tcache;
@@ -130,78 +101,329 @@ End:
   taosArrayDestroyEx(files, tfileDestroyFileName);
   return NULL;
 }
-void tfileCacheDestroy(TFileCache *tcache) {
-  if (tcache == NULL) {
-    return;
-  }
+void tfileCacheDestroy(TFileCache* tcache) {
+  if (tcache == NULL) { return; }
 
   // free table cache
-  TFileReader **reader = taosHashIterate(tcache->tableCache, NULL);
+  TFileReader** reader = taosHashIterate(tcache->tableCache, NULL);
   while (reader) {
-    TFileReader *p = *reader;
-    indexInfo("drop table cache suid: %" PRIu64 ", colName: %s, colType: %d", p->header.suid, p->header.colName,
-        p->header.colType);
-    TFileReaderDestroy(p);
+    TFileReader* p = *reader;
+    indexInfo("drop table cache suid: %" PRIu64 ", colName: %s, colType: %d", p->header.suid, p->header.colName, p->header.colType);
+
+    tfileReaderDestroy(p);
     reader = taosHashIterate(tcache->tableCache, reader);
   }
   taosHashCleanup(tcache->tableCache);
   free(tcache);
 }
 
-TFileReader *tfileCacheGet(TFileCache *tcache, TFileCacheKey *key) {
+TFileReader* tfileCacheGet(TFileCache* tcache, TFileCacheKey* key) {
   char buf[128] = {0};
   tfileSerialCacheKey(key, buf);
-  TFileReader *reader = taosHashGet(tcache->tableCache, buf, strlen(buf));
+  TFileReader* reader = taosHashGet(tcache->tableCache, buf, strlen(buf));
   return reader;
 }
-void tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader) {
+void tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader* reader) {
   char buf[128] = {0};
   tfileSerialCacheKey(key, buf);
-  taosHashPut(tcache->tableCache, buf, strlen(buf), &reader, sizeof(void *));
+  taosHashPut(tcache->tableCache, buf, strlen(buf), &reader, sizeof(void*));
   return;
 }
 
-TFileReader *tfileReaderCreate(WriterCtx *ctx) {
-  TFileReader *reader = calloc(1, sizeof(TFileReader));
-  if (reader == NULL) {
-    return NULL;
-  }
-  reader->ctx = ctx;
+TFileReader* tfileReaderCreate(WriterCtx* ctx) {
+  TFileReader* reader = calloc(1, sizeof(TFileReader));
+  if (reader == NULL) { return NULL; }
+
   // T_REF_INC(reader);
+  reader->ctx = ctx;
   return reader;
 }
-void TFileReaderDestroy(TFileReader *reader) {
-  if (reader == NULL) {
-    return;
-  }
+void tfileReaderDestroy(TFileReader* reader) {
+  if (reader == NULL) { return; }
   // T_REF_INC(reader);
+  fstDestroy(reader->fst);
   writerCtxDestroy(reader->ctx);
   free(reader);
 }
 
-TFileWriter *tfileWriterCreate(const char *suid, const char *colName);
-void         tfileWriterDestroy(TFileWriter *tw);
+int tfileReaderSearch(TFileReader* reader, SIndexTermQuery* query, SArray* result) {
+  SIndexTerm* term = query->term;
 
-IndexTFile *indexTFileCreate(const char *path) {
-  IndexTFile *tfile = calloc(1, sizeof(IndexTFile));
-  tfile->cache = tfileCacheCreate(path);
+  // refactor to callback later
+  if (query->qType == QUERY_TERM) {
+    uint64_t offset;
+    FstSlice key = fstSliceCreate(term->colVal, term->nColVal);
+    if (fstGet(reader->fst, &key, &offset)) {
+      return tfileReadLoadTableIds(reader, offset, result);
+    } else {
+      indexInfo("index: %" PRIu64 ", col: %s, colVal: %s, not found in tindex", term->suid, term->colName, term->colVal);
+    }
+    return 0;
+  } else if (query->qType == QUERY_PREFIX) {
+    // handle later
+    //
+  } else {
+    // handle later
+  }
+  return 0;
+}
+
+TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header) {
+  // char pathBuf[128] = {0};
+  // sprintf(pathBuf, "%s/% " PRIu64 "-%d-%d.tindex", path, suid, colId, version);
+  // TFileHeader header = {.suid = suid, .version = version, .colName = {0}, colType = colType};
+  // memcpy(header.colName, );
+
+  // char buf[TFILE_HADER_PRE_SIZE];
+  // int  len = TFILE_HADER_PRE_SIZE;
+  // if (len != ctx->write(ctx, buf, len)) {
+  //  indexError("index: %" PRIu64 " failed to write header info", header->suid);
+  //  return NULL;
+  //}
+  TFileWriter* tw = calloc(1, sizeof(TFileWriter));
+  if (tw == NULL) {
+    indexError("index: %" PRIu64 " failed to alloc TFilerWriter", header->suid);
+    return NULL;
+  }
+  tw->ctx = ctx;
+  tw->header = *header;
+  tfileWriteHeader(tw);
+  return tw;
+}
+
+int tfileWriterPut(TFileWriter* tw, void* data) {
+  // sort by coltype and write to tindex
+  __compar_fn_t fn = getComparFunc(tw->header.colType, 0);
+  taosArraySortPWithExt((SArray*)(data), tfileValueCompare, &fn);
+
+  int32_t bufLimit = 4096, offset = 0;
+  char*   buf = calloc(1, sizeof(bufLimit));
+  char*   p = buf;
+  int32_t sz = taosArrayGetSize((SArray*)data);
+  int32_t fstOffset = tw->offset;
+
+  // ugly code, refactor later
+  for (size_t i = 0; i < sz; i++) {
+    TFileValue* v = taosArrayGetP((SArray*)data, i);
+
+    int32_t tbsz = taosArrayGetSize(v->tableId);
+    fstOffset += TF_TABLE_TATOAL_SIZE(tbsz);
+  }
+  // check result or not
+  tfileWriteFstOffset(tw, fstOffset);
+
+  for (size_t i = 0; i < sz; i++) {
+    TFileValue* v = taosArrayGetP((SArray*)data, i);
+
+    int32_t tbsz = taosArrayGetSize(v->tableId);
+    // check buf has enough space or not
+    int32_t ttsz = TF_TABLE_TATOAL_SIZE(tbsz);
+    if (offset + ttsz > bufLimit) {
+      // batch write
+      tw->ctx->write(tw->ctx, buf, offset);
+      offset = 0;
+      memset(buf, 0, bufLimit);
+      p = buf;
+    }
+
+    tfileSerialTableIdsToBuf(p, v->tableId);
+    offset += ttsz;
+    p = buf + offset;
+    // set up value offset
+    v->offset = tw->offset;
+    tw->offset += ttsz;
+  }
+  if (offset != 0) {
+    // write reversed data in buf to tindex
+    tw->ctx->write(tw->ctx, buf, offset);
+  }
+
+  // write fst
+  for (size_t i = 0; i < sz; i++) {
+    // TODO, fst batch write later
+    TFileValue* v = taosArrayGetP((SArray*)data, i);
+    if (tfileWriteData(tw, v) == 0) {
+      //
+      //
+    }
+  }
+
+  tfree(buf);
+  return 0;
+}
+void tfileWriterDestroy(TFileWriter* tw) {
+  if (tw == NULL) { return; }
 
+  writerCtxDestroy(tw->ctx);
+  free(tw);
+}
+
+IndexTFile* indexTFileCreate(const char* path) {
+  IndexTFile* tfile = calloc(1, sizeof(IndexTFile));
+  if (tfile == NULL) { return NULL; }
+
+  tfile->cache = tfileCacheCreate(path);
   return tfile;
 }
-void IndexTFileDestroy(IndexTFile *tfile) { free(tfile); }
+void IndexTFileDestroy(IndexTFile* tfile) {
+  free(tfile);
+}
+
+int indexTFileSearch(void* tfile, SIndexTermQuery* query, SArray* result) {
+  if (tfile == NULL) { return -1; }
+  IndexTFile* pTfile = (IndexTFile*)tfile;
+
+  SIndexTerm*   term = query->term;
+  TFileCacheKey key = {.suid = term->suid, .colType = term->colType, .version = 0, .colName = term->colName, .nColName = term->nColName};
+
+  TFileReader* reader = tfileCacheGet(pTfile->cache, &key);
+  return tfileReaderSearch(reader, query, result);
+}
+int indexTFilePut(void* tfile, SIndexTerm* term, uint64_t uid) {
+  TFileWriterOpt wOpt = {.suid = term->suid, .colType = term->colType, .colName = term->colName, .nColName = term->nColName, .version = 1};
+
+  return 0;
+}
+
+static int tfileValueCompare(const void* a, const void* b, const void* param) {
+  __compar_fn_t fn = *(__compar_fn_t*)param;
+
+  TFileValue* av = (TFileValue*)a;
+  TFileValue* bv = (TFileValue*)b;
+
+  return fn(av->colVal, bv->colVal);
+}
+static void tfileSerialTableIdsToBuf(char* buf, SArray* ids) {
+  int sz = taosArrayGetSize(ids);
+  SERIALIZE_VAR_TO_BUF(buf, sz, int32_t);
+  for (size_t i = 0; i < sz; i++) {
+    uint64_t* v = taosArrayGet(ids, i);
+    SERIALIZE_VAR_TO_BUF(buf, *v, uint64_t);
+  }
+}
+
+static int tfileWriteFstOffset(TFileWriter* tw, int32_t offset) {
+  int32_t fstOffset = offset + sizeof(tw->header.fstOffset);
+  tw->header.fstOffset = fstOffset;
+  if (sizeof(fstOffset) != tw->ctx->write(tw->ctx, (char*)&fstOffset, sizeof(fstOffset))) { return -1; }
+  return 0;
+}
+static int tfileWriteHeader(TFileWriter* writer) {
+  char  buf[TFILE_HEADER_NO_FST] = {0};
+  char* p = buf;
+
+  TFileHeader* header = &writer->header;
+  memcpy(buf, (char*)header, sizeof(buf));
+
+  int nwrite = writer->ctx->write(writer->ctx, buf, sizeof(buf));
+  if (sizeof(buf) != nwrite) { return -1; }
+  writer->offset = nwrite;
+  return 0;
+}
+static int tfileWriteData(TFileWriter* write, TFileValue* tval) {
+  TFileHeader* header = &write->header;
+  uint8_t      colType = header->colType;
+  if (colType == TSDB_DATA_TYPE_BINARY || colType == TSDB_DATA_TYPE_NCHAR) {
+    FstSlice key = fstSliceCreate((uint8_t*)(tval->colVal), (size_t)strlen(tval->colVal));
+    if (fstBuilderInsert(write->fb, key, tval->offset)) {
+      fstSliceDestroy(&key);
+      return 0;
+    }
+    fstSliceDestroy(&key);
+    return -1;
+  } else {
+    // handle other type later
+  }
+  return 0;
+}
+static int tfileReadLoadHeader(TFileReader* reader) {
+  // TODO simple tfile header later
+  char  buf[TFILE_HEADER_SIZE] = {0};
+  char* p = buf;
+
+  int64_t nread = reader->ctx->read(reader->ctx, buf, sizeof(buf));
+  assert(nread == sizeof(buf));
+  memcpy(&reader->header, buf, sizeof(buf));
+  return 0;
+}
+static int tfileReadLoadFst(TFileReader* reader) {
+  // current load fst into memory, refactor it later
+  static int FST_MAX_SIZE = 16 * 1024;
+
+  char* buf = calloc(1, sizeof(char) * FST_MAX_SIZE);
+  if (buf == NULL) { return -1; }
+
+  WriterCtx* ctx = reader->ctx;
+  int32_t    nread = ctx->readFrom(ctx, buf, FST_MAX_SIZE, reader->header.fstOffset);
+  // we assuse fst size less than FST_MAX_SIZE
+  assert(nread > 0 && nread < FST_MAX_SIZE);
+
+  FstSlice st = fstSliceCreate((uint8_t*)buf, nread);
+  reader->fst = fstCreate(&st);
+  free(buf);
+  fstSliceDestroy(&st);
+
+  return reader->fst == NULL ? 0 : -1;
+}
+static int tfileReadLoadTableIds(TFileReader* reader, int32_t offset, SArray* result) {
+  int32_t    nid;
+  WriterCtx* ctx = reader->ctx;
+  int32_t    nread = ctx->readFrom(ctx, (char*)&nid, sizeof(nid), offset);
+  assert(sizeof(nid) == nread);
 
-int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result) {
-  IndexTFile *pTfile = (IndexTFile *)tfile;
+  char* buf = calloc(1, sizeof(uint64_t) * nid);
+  if (buf == NULL) { return -1; }
 
-  SIndexTerm *  term = query->term;
-  TFileCacheKey key = {
-      .suid = term->suid, .colType = term->colType, .version = 0, .colName = term->colName, .nColName = term->nColName};
-  TFileReader *reader = tfileCacheGet(pTfile->cache, &key);
+  nread = ctx->read(ctx, buf, sizeof(uint64_t) * nid);
+  uint64_t* ids = (uint64_t*)buf;
+  for (int32_t i = 0; i < nid; i++) {
+    taosArrayPush(result, ids + i);
+  }
+  free(buf);
   return 0;
 }
-int indexTFilePut(void *tfile, SIndexTerm *term, uint64_t uid) {
-  TFileWriterOpt wOpt = {
-      .suid = term->suid, .colType = term->colType, .colName = term->colName, .nColName = term->nColName, .version = 1};
 
+static int tfileGetFileList(const char* path, SArray* result) {
+  DIR* dir = opendir(path);
+  if (NULL == dir) { return -1; }
+
+  struct dirent* entry;
+  while ((entry = readdir(dir)) != NULL) {
+    size_t len = strlen(entry->d_name);
+    char*  buf = calloc(1, len + 1);
+    memcpy(buf, entry->d_name, len);
+    taosArrayPush(result, &buf);
+  }
+  closedir(dir);
+  return 0;
+}
+static void tfileDestroyFileName(void* elem) {
+  char* p = *(char**)elem;
+  free(p);
+}
+static int tfileCompare(const void* a, const void* b) {
+  const char* aName = *(char**)a;
+  const char* bName = *(char**)b;
+
+  size_t aLen = strlen(aName);
+  size_t bLen = strlen(bName);
+
+  return strncmp(aName, bName, aLen > bLen ? aLen : bLen);
+}
+// tfile name suid-colId-version.tindex
+static int tfileParseFileName(const char* filename, uint64_t* suid, int* colId, int* version) {
+  if (3 == sscanf(filename, "%" PRIu64 "-%d-%d.tindex", suid, colId, version)) {
+    // read suid & colid & version  success
     return 0;
+  }
+  return -1;
+}
+static void tfileSerialCacheKey(TFileCacheKey* key, char* buf) {
+  SERIALIZE_MEM_TO_BUF(buf, key, suid);
+  SERIALIZE_VAR_TO_BUF(buf, '_', char);
+  SERIALIZE_MEM_TO_BUF(buf, key, colType);
+  SERIALIZE_VAR_TO_BUF(buf, '_', char);
+  SERIALIZE_MEM_TO_BUF(buf, key, version);
+  SERIALIZE_VAR_TO_BUF(buf, '_', char);
+  SERIALIZE_STR_MEM_TO_BUF(buf, key, colName, key->nColName);
 }

source/libs/parser/inc/dataBlockMgt.h

@@ -170,10 +170,11 @@ int32_t schemaIdxCompar(const void *lhs, const void *rhs);
 int32_t boundIdxCompar(const void *lhs, const void *rhs);
 void setBoundColumnInfo(SParsedDataColInfo* pColList, SSchema* pSchema, int32_t numOfCols);
 void destroyBoundColumnInfo(SParsedDataColInfo* pColList);
+void destroyBlockArrayList(SArray* pDataBlockList);
 int32_t initMemRowBuilder(SMemRowBuilder *pBuilder, uint32_t nRows, uint32_t nCols, uint32_t nBoundCols, int32_t allNullLen);
 int32_t allocateMemIfNeed(STableDataBlocks *pDataBlock, int32_t rowSize, int32_t * numOfRows);
 int32_t getDataBlockFromList(SHashObj* pHashList, int64_t id, int32_t size, int32_t startOffset, int32_t rowSize,
     const STableMeta* pTableMeta, STableDataBlocks** dataBlocks, SArray* pBlockList);
-int32_t mergeTableDataBlocks(SHashObj* pHashObj, int8_t schemaAttached, uint8_t payloadType, bool freeBlockMap);
+int32_t mergeTableDataBlocks(SHashObj* pHashObj, int8_t schemaAttached, uint8_t payloadType, SArray** pVgDataBlocks);
 
 #endif  // TDENGINE_DATABLOCKMGT_H

source/libs/parser/src/dataBlockMgt.c

@@ -269,19 +269,17 @@ void destroyDataBlock(STableDataBlocks* pDataBlock) {
   tfree(pDataBlock);
 }
 
-void* destroyBlockArrayList(SArray* pDataBlockList) {
+void destroyBlockArrayList(SArray* pDataBlockList) {
   if (pDataBlockList == NULL) {
-    return NULL;
+    return;
   }
 
   size_t size = taosArrayGetSize(pDataBlockList);
   for (int32_t i = 0; i < size; i++) {
-    void* d = taosArrayGetP(pDataBlockList, i);
-    destroyDataBlock(d);
+    destroyDataBlock(taosArrayGetP(pDataBlockList, i));
   }
 
   taosArrayDestroy(pDataBlockList);
-  return NULL;
 }
 
 // data block is disordered, sort it in ascending order
@@ -298,6 +296,7 @@ void sortRemoveDataBlockDupRowsRaw(STableDataBlocks *dataBuf) {
     int32_t i = 0;
     int32_t j = 1;
 
+    // delete rows with timestamp conflicts
     while (j < pBlocks->numOfRows) {
       TSKEY ti = *(TSKEY *)(pBlockData + dataBuf->rowSize * i);
       TSKEY tj = *(TSKEY *)(pBlockData + dataBuf->rowSize * j);
@@ -430,7 +429,7 @@ static int trimDataBlock(void* pDataBlock, STableDataBlocks* pTableDataBlock, SB
 
   char* p = pTableDataBlock->pData + sizeof(SSubmitBlk);
   pBlock->dataLen = 0;
-  int32_t numOfRows = htons(pBlock->numOfRows);
+  int32_t numOfRows = pBlock->numOfRows;
 
   if (isRawPayload) {
     for (int32_t i = 0; i < numOfRows; ++i) {
@@ -467,18 +466,10 @@ static int trimDataBlock(void* pDataBlock, STableDataBlocks* pTableDataBlock, SB
     }
   }
 
-  int32_t len = pBlock->dataLen + pBlock->schemaLen;
-  pBlock->dataLen = htonl(pBlock->dataLen);
-  pBlock->schemaLen = htonl(pBlock->schemaLen);
-
-  return len;
+  return pBlock->dataLen + pBlock->schemaLen;
 }
 
-static void extractTableNameList(SHashObj* pHashObj, bool freeBlockMap) {
-  // todo
-}
-
-int32_t mergeTableDataBlocks(SHashObj* pHashObj, int8_t schemaAttached, uint8_t payloadType, bool freeBlockMap) {
+int32_t mergeTableDataBlocks(SHashObj* pHashObj, int8_t schemaAttached, uint8_t payloadType, SArray** pVgDataBlocks) {
   const int INSERT_HEAD_SIZE = sizeof(SMsgDesc) + sizeof(SSubmitMsg);
   int       code = 0;
   bool      isRawPayload = IS_RAW_PAYLOAD(payloadType);
@@ -537,24 +528,13 @@ int32_t mergeTableDataBlocks(SHashObj* pHashObj, int8_t schemaAttached, uint8_t
                         (isRawPayload ? (pOneTableBlock->rowSize + expandSize) : getExtendedRowSize(pOneTableBlock)) +
                     sizeof(STColumn) * getNumOfColumns(pOneTableBlock->pTableMeta);
 
-      pBlocks->tid = htonl(pBlocks->tid);
-      pBlocks->uid = htobe64(pBlocks->uid);
-      pBlocks->sversion = htonl(pBlocks->sversion);
-      pBlocks->numOfRows = htons(pBlocks->numOfRows);
-      pBlocks->schemaLen = 0;
-
       // erase the empty space reserved for binary data
       int32_t finalLen = trimDataBlock(dataBuf->pData + dataBuf->size, pOneTableBlock, blkKeyInfo.pKeyTuple, schemaAttached, isRawPayload);
       assert(finalLen <= len);
 
       dataBuf->size += (finalLen + sizeof(SSubmitBlk));
       assert(dataBuf->size <= dataBuf->nAllocSize);
-
-      // the length does not include the SSubmitBlk structure
-      pBlocks->dataLen = htonl(finalLen);
       dataBuf->numOfTables += 1;
-
-      pBlocks->numOfRows = 0;
     }
 
     p = taosHashIterate(pHashObj, p);
@@ -565,12 +545,10 @@ int32_t mergeTableDataBlocks(SHashObj* pHashObj, int8_t schemaAttached, uint8_t
     pOneTableBlock = *p;
   }
 
-  extractTableNameList(pHashObj, freeBlockMap);
-
   // free the table data blocks;
   taosHashCleanup(pVnodeDataBlockHashList);
   tfree(blkKeyInfo.pKeyTuple);
-
+  *pVgDataBlocks = pVnodeDataBlockList;
   return TSDB_CODE_SUCCESS;
 }
 

source/libs/parser/src/insertParser.c

@@ -31,33 +31,16 @@
     pSql += index; \
   } while (0)
 
-#define CHECK_CODE(expr) \
+#define NEXT_TOKEN_KEEP_SQL(pSql, sToken, index) \
   do { \
-    int32_t code = expr; \
-    if (TSDB_CODE_SUCCESS != code) { \
-      terrno = code; \
-      return terrno; \
-    } \
-  } while (0)
-
-#define CHECK_CODE_1(expr, destroy) \
-  do { \
-    int32_t code = expr; \
-    if (TSDB_CODE_SUCCESS != code) { \
-      (void)destroy; \
-      terrno = code; \
-      return terrno; \
-    } \
+    sToken = tStrGetToken(pSql, &index, false); \
   } while (0)
 
-#define CHECK_CODE_2(expr, destroy1, destroy2) \
+#define CHECK_CODE(expr) \
   do { \
     int32_t code = expr; \
     if (TSDB_CODE_SUCCESS != code) { \
-      (void)destroy1; \
-      (void)destroy2; \
-      terrno = code; \
-      return terrno; \
+      return code; \
     } \
   } while (0)
 
@@ -70,9 +53,12 @@ typedef struct SInsertParseContext {
   SParseContext* pComCxt;
   const char* pSql;
   SMsgBuf msg;
-  struct SCatalog* pCatalog;
   STableMeta* pTableMeta;
-  SHashObj* pTableBlockHashObj; // data block for each table. need release
+  SParsedDataColInfo tags;
+  SKVRowBuilder tagsBuilder;
+  SHashObj* pTableBlockHashObj;
+  SArray* pTableDataBlocks;
+  SArray* pVgDataBlocks;
   int32_t totalNum;
   SInsertStmtInfo* pOutput;
 } SInsertParseContext;
@@ -151,14 +137,6 @@ static int32_t toInt64(const char* z, int16_t type, int32_t n, int64_t* value, b
   return ret;
 }
 
-static int32_t createInsertStmtInfo(SInsertStmtInfo **pInsertInfo) {
-  *pInsertInfo = calloc(1, sizeof(SQueryStmtInfo));
-  if (NULL == *pInsertInfo) {
-    return TSDB_CODE_TSC_OUT_OF_MEMORY;
-  }
-  return TSDB_CODE_SUCCESS;
-}
-
 static int32_t skipInsertInto(SInsertParseContext* pCxt) {
   SToken sToken;
   NEXT_TOKEN(pCxt->pSql, sToken);
@@ -194,7 +172,7 @@ static int32_t getTableMeta(SInsertParseContext* pCxt, SToken* pTname) {
   char fullDbName[TSDB_FULL_DB_NAME_LEN] = {0};
   char tableName[TSDB_TABLE_NAME_LEN] = {0};
   CHECK_CODE(buildName(pCxt, pTname, fullDbName, tableName));
-  CHECK_CODE(catalogGetTableMeta(pCxt->pCatalog, pCxt->pComCxt->pRpc, pCxt->pComCxt->pEpSet, fullDbName, tableName, &pCxt->pTableMeta));
+  CHECK_CODE(catalogGetTableMeta(pCxt->pComCxt->pCatalog, pCxt->pComCxt->pRpc, pCxt->pComCxt->pEpSet, fullDbName, tableName, &pCxt->pTableMeta));
   return TSDB_CODE_SUCCESS;
 }
 
@@ -209,6 +187,51 @@ static int32_t findCol(SToken* pColname, int32_t start, int32_t end, SSchema* pS
   return -1;
 }
 
+static void fillMsgHeader(SVgDataBlocks* dst) {
+    SMsgDesc* desc = (SMsgDesc*)dst->pData;
+    desc->numOfVnodes = htonl(1);
+    SSubmitMsg* submit = (SSubmitMsg*)(desc + 1);
+    submit->header.vgId    = htonl(dst->vgId);
+    submit->header.contLen = htonl(dst->size - sizeof(SMsgDesc));
+    submit->length         = submit->header.contLen;
+    submit->numOfBlocks    = htonl(dst->numOfTables);
+    SSubmitBlk* blk = (SSubmitBlk*)(submit + 1);
+    int32_t numOfBlocks = dst->numOfTables;
+    while (numOfBlocks--) {
+      int32_t dataLen = blk->dataLen;
+      blk->uid = htobe64(blk->uid);
+      blk->tid = htonl(blk->tid);
+      blk->padding = htonl(blk->padding);
+      blk->sversion = htonl(blk->sversion);
+      blk->dataLen = htonl(blk->dataLen);
+      blk->schemaLen = htonl(blk->schemaLen);
+      blk->numOfRows = htons(blk->numOfRows);
+      blk = (SSubmitBlk*)(blk->data + dataLen);
+    }
+}
+
+static int32_t buildOutput(SInsertParseContext* pCxt) {
+  size_t numOfVg = taosArrayGetSize(pCxt->pVgDataBlocks);
+  pCxt->pOutput->pDataBlocks = taosArrayInit(numOfVg, POINTER_BYTES);
+  if (NULL == pCxt->pOutput->pDataBlocks) {
+    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  }
+  for (size_t i = 0; i < numOfVg; ++i) {
+    STableDataBlocks* src = taosArrayGetP(pCxt->pVgDataBlocks, i);
+    SVgDataBlocks* dst = calloc(1, sizeof(SVgDataBlocks));
+    if (NULL == dst) {
+      return TSDB_CODE_TSC_OUT_OF_MEMORY;
+    }
+    dst->vgId = src->vgId;
+    dst->numOfTables = src->numOfTables;
+    dst->size = src->size;
+    SWAP(dst->pData, src->pData, char*);
+    fillMsgHeader(dst);
+    taosArrayPush(pCxt->pOutput->pDataBlocks, &dst);
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
 static int32_t checkTimestamp(STableDataBlocks *pDataBlocks, const char *start) {
   // once the data block is disordered, we do NOT keep previous timestamp any more
   if (!pDataBlocks->ordered) {
@@ -219,16 +242,14 @@ static int32_t checkTimestamp(STableDataBlocks *pDataBlocks, const char *start)
 
   if (k == INT64_MIN) {
     if (pDataBlocks->tsSource == TSDB_USE_CLI_TS) {
-      return -1;
-    } else if (pDataBlocks->tsSource == -1) {
-      pDataBlocks->tsSource = TSDB_USE_SERVER_TS;
+      return TSDB_CODE_FAILED; // client time/server time can not be mixed
     }
+    pDataBlocks->tsSource = TSDB_USE_SERVER_TS;
   } else {
     if (pDataBlocks->tsSource == TSDB_USE_SERVER_TS) {
-      return -1;  // client time/server time can not be mixed
-    } else if (pDataBlocks->tsSource == -1) {
-      pDataBlocks->tsSource = TSDB_USE_CLI_TS;
+      return TSDB_CODE_FAILED;  // client time/server time can not be mixed
     }
+    pDataBlocks->tsSource = TSDB_USE_CLI_TS;
   }
 
   if (k <= pDataBlocks->prevTS && (pDataBlocks->tsSource == TSDB_USE_CLI_TS)) {
@@ -608,26 +629,24 @@ static int32_t parseBoundColumns(SInsertParseContext* pCxt, SParsedDataColInfo*
 }
 
 // pSql -> tag1_value, ...)
-static int32_t parseTagsClause(SInsertParseContext* pCxt, SParsedDataColInfo* pSpd, SSchema* pTagsSchema, uint8_t precision) {
-  SKVRowBuilder kvRowBuilder = {0};
-  if (tdInitKVRowBuilder(&kvRowBuilder) < 0) {
-    destroyBoundColumnInfo(pSpd);
+static int32_t parseTagsClause(SInsertParseContext* pCxt, SSchema* pTagsSchema, uint8_t precision) {
+  if (tdInitKVRowBuilder(&pCxt->tagsBuilder) < 0) {
     return TSDB_CODE_TSC_OUT_OF_MEMORY;
   }
 
-  SKvParam param = {.builder = &kvRowBuilder};
+  SKvParam param = {.builder = &pCxt->tagsBuilder};
   SToken sToken;
   char tmpTokenBuf[TSDB_MAX_BYTES_PER_ROW] = {0};  // used for deleting Escape character: \\, \', \"
-  for (int i = 0; i < pSpd->numOfBound; ++i) {
+  for (int i = 0; i < pCxt->tags.numOfBound; ++i) {
     NEXT_TOKEN(pCxt->pSql, sToken);
-    SSchema* pSchema = &pTagsSchema[pSpd->boundedColumns[i]];
+    SSchema* pSchema = &pTagsSchema[pCxt->tags.boundedColumns[i]];
     param.schema = pSchema;
-    CHECK_CODE_2(parseOneValue(pCxt, &sToken, pSchema, precision, tmpTokenBuf, KvRowAppend, &param), tdDestroyKVRowBuilder(&kvRowBuilder), destroyBoundColumnInfo(pSpd));
+    CHECK_CODE(parseOneValue(pCxt, &sToken, pSchema, precision, tmpTokenBuf, KvRowAppend, &param));
   }
 
-  destroyBoundColumnInfo(pSpd);
-  SKVRow row = tdGetKVRowFromBuilder(&kvRowBuilder);
-  tdDestroyKVRowBuilder(&kvRowBuilder);
+  destroyBoundColumnInfo(&pCxt->tags);
+  SKVRow row = tdGetKVRowFromBuilder(&pCxt->tagsBuilder);
+  tdDestroyKVRowBuilder(&pCxt->tagsBuilder);
   if (NULL == row) {
     return buildInvalidOperationMsg(&pCxt->msg, "tag value expected");
   }
@@ -650,13 +669,12 @@ static int32_t parseUsingClause(SInsertParseContext* pCxt, SToken* pTbnameToken)
   }
 
   SSchema* pTagsSchema = getTableTagSchema(pCxt->pTableMeta);
-  SParsedDataColInfo spd = {0};
-  setBoundColumnInfo(&spd, pTagsSchema, getNumOfTags(pCxt->pTableMeta));
+  setBoundColumnInfo(&pCxt->tags, pTagsSchema, getNumOfTags(pCxt->pTableMeta));
 
   // pSql -> [(tag1_name, ...)] TAGS (tag1_value, ...)
   NEXT_TOKEN(pCxt->pSql, sToken);
   if (TK_LP == sToken.type) {
-    CHECK_CODE_1(parseBoundColumns(pCxt, &spd, pTagsSchema), destroyBoundColumnInfo(&spd));
+    CHECK_CODE(parseBoundColumns(pCxt, &pCxt->tags, pTagsSchema));
     NEXT_TOKEN(pCxt->pSql, sToken);
   }
 
@@ -668,7 +686,7 @@ static int32_t parseUsingClause(SInsertParseContext* pCxt, SToken* pTbnameToken)
   if (TK_LP != sToken.type) {
     return buildSyntaxErrMsg(&pCxt->msg, "( is expected", sToken.z);
   }
-  CHECK_CODE(parseTagsClause(pCxt, &spd, pTagsSchema, getTableInfo(pCxt->pTableMeta).precision));
+  CHECK_CODE(parseTagsClause(pCxt, pTagsSchema, getTableInfo(pCxt->pTableMeta).precision));
 
   return TSDB_CODE_SUCCESS;
 }
@@ -732,10 +750,12 @@ static int32_t parseValues(SInsertParseContext* pCxt, STableDataBlocks* pDataBlo
   char tmpTokenBuf[TSDB_MAX_BYTES_PER_ROW] = {0};  // used for deleting Escape character: \\, \', \"
   SToken sToken;
   while (1) {
-    NEXT_TOKEN(pCxt->pSql, sToken);
+    int32_t index = 0;
+    NEXT_TOKEN_KEEP_SQL(pCxt->pSql, sToken, index);
     if (TK_LP != sToken.type) {
       break;
     }
+    pCxt->pSql += index;
 
     if ((*numOfRows) >= maxRows || pDataBlock->size + extendedRowSize >= pDataBlock->nAllocSize) {
       int32_t tSize;
@@ -815,7 +835,7 @@ static int32_t parseInsertBody(SInsertParseContext* pCxt) {
 
     if (TK_LP == sToken.type) {
       // pSql -> field1_name, ...)
-      CHECK_CODE_1(parseBoundColumns(pCxt, &dataBuf->boundColumnInfo, getTableColumnSchema(pCxt->pTableMeta)), destroyBoundColumnInfo(&dataBuf->boundColumnInfo));
+      CHECK_CODE(parseBoundColumns(pCxt, &dataBuf->boundColumnInfo, getTableColumnSchema(pCxt->pTableMeta)));
       NEXT_TOKEN(pCxt->pSql, sToken);
     }
 
@@ -842,9 +862,18 @@ static int32_t parseInsertBody(SInsertParseContext* pCxt) {
   }
   // merge according to vgId
   if (!TSDB_QUERY_HAS_TYPE(pCxt->pOutput->insertType, TSDB_QUERY_TYPE_STMT_INSERT) && taosHashGetSize(pCxt->pTableBlockHashObj) > 0) {
-    CHECK_CODE(mergeTableDataBlocks(pCxt->pTableBlockHashObj, pCxt->pOutput->schemaAttache, pCxt->pOutput->payloadType, true));
+    CHECK_CODE(mergeTableDataBlocks(pCxt->pTableBlockHashObj, pCxt->pOutput->schemaAttache, pCxt->pOutput->payloadType, &pCxt->pVgDataBlocks));
   }
-  return TSDB_CODE_SUCCESS;
+  return buildOutput(pCxt);
+}
+
+static void destroyInsertParseContext(SInsertParseContext* pCxt) {
+  tfree(pCxt->pTableMeta);
+  destroyBoundColumnInfo(&pCxt->tags);
+  tdDestroyKVRowBuilder(&pCxt->tagsBuilder);
+  taosHashCleanup(pCxt->pTableBlockHashObj);
+  destroyBlockArrayList(pCxt->pTableDataBlocks);
+  destroyBlockArrayList(pCxt->pVgDataBlocks);
 }
 
 // INSERT INTO
@@ -854,26 +883,30 @@ static int32_t parseInsertBody(SInsertParseContext* pCxt) {
 //       VALUES (field1_value, ...) [(field1_value2, ...) ...] | FILE csv_file_path
 //   [...];
 int32_t parseInsertSql(SParseContext* pContext, SInsertStmtInfo** pInfo) {
-  CHECK_CODE(createInsertStmtInfo(pInfo));
-
   SInsertParseContext context = {
     .pComCxt = pContext,
     .pSql = pContext->pSql,
     .msg = {.buf = pContext->pMsg, .len = pContext->msgLen},
-    .pCatalog = NULL,
     .pTableMeta = NULL,
     .pTableBlockHashObj = taosHashInit(128, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, false),
     .totalNum = 0,
-    .pOutput = *pInfo
+    .pOutput = calloc(1, sizeof(SInsertStmtInfo))
   };
 
-  if (NULL == context.pTableBlockHashObj) {
-    return TSDB_CODE_TSC_OUT_OF_MEMORY;
+  if (NULL == context.pTableBlockHashObj || NULL == context.pOutput) {
+    terrno = TSDB_CODE_TSC_OUT_OF_MEMORY;
+    return TSDB_CODE_FAILED;
   }
 
-  CHECK_CODE(catalogGetHandle(pContext->pClusterId, &context.pCatalog));
-  CHECK_CODE(skipInsertInto(&context));
-  CHECK_CODE(parseInsertBody(&context));
+  *pInfo = context.pOutput;
+  context.pOutput->schemaAttache = pContext->schemaAttached;
+  context.pOutput->payloadType = PAYLOAD_TYPE_KV;
 
-  return TSDB_CODE_SUCCESS;
+  int32_t code = skipInsertInto(&context);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = parseInsertBody(&context);
+  }
+  destroyInsertParseContext(&context);
+  terrno = code;
+  return (TSDB_CODE_SUCCESS == code ? TSDB_CODE_SUCCESS : TSDB_CODE_FAILED);
 }

source/libs/parser/test/CMakeLists.txt

@@ -17,5 +17,3 @@ TARGET_LINK_LIBRARIES(
         parserTest
         PUBLIC os util common parser catalog transport gtest function planner qcom
 )
-
-TARGET_LINK_OPTIONS(parserTest PRIVATE -Wl,-wrap,malloc)

source/libs/parser/test/insertTest.cpp → source/libs/parser/test/insertParserTest.cpp

@@ -16,7 +16,7 @@
 #include <gtest/gtest.h>
 
 #include "insertParser.h"
-#include "mockCatalog.h"
+// #include "mockCatalog.h"
 
 using namespace std;
 using namespace testing;
@@ -27,27 +27,6 @@ namespace {
   }
 }
 
-extern "C" {
-
-#include <execinfo.h>
-
-void *__real_malloc(size_t);
-
-void *__wrap_malloc(size_t c) {  
-  // printf("My MALLOC called: %d\n", c);
-  // void *array[32]; 
-  // int size = backtrace(array, 32); 
-  // char **symbols = backtrace_symbols(array, size); 
-  // for (int i = 0; i < size; ++i) { 
-  //   cout << symbols[i] << endl;
-  // } 
-  // free(symbols); 
-
-  return __real_malloc(c);
-}
-
-}
-
 // syntax:
 // INSERT INTO
 //   tb_name
@@ -85,6 +64,51 @@ protected:
     return res_;
   }
 
+  void dumpReslut() {
+    size_t num = taosArrayGetSize(res_->pDataBlocks);
+    cout << "schemaAttache:" << (int32_t)res_->schemaAttache << ", payloadType:" << (int32_t)res_->payloadType << ", insertType:" << res_->insertType << ", numOfVgs:" << num << endl;    
+    for (size_t i = 0; i < num; ++i) {
+      SVgDataBlocks* vg = (SVgDataBlocks*)taosArrayGetP(res_->pDataBlocks, i);
+      cout << "vgId:" << vg->vgId << ", numOfTables:" << vg->numOfTables << ", dataSize:" << vg->size << endl;
+      SMsgDesc* desc = (SMsgDesc*)(vg->pData);
+      cout << "numOfVnodes:" << ntohl(desc->numOfVnodes) << endl;
+      SSubmitMsg* submit = (SSubmitMsg*)(desc + 1);
+      cout << "length:" << ntohl(submit->length) << ", numOfBlocks:" << ntohl(submit->numOfBlocks) << endl;
+      int32_t numOfBlocks = ntohl(submit->numOfBlocks);
+      SSubmitBlk* blk = (SSubmitBlk*)(submit + 1);
+      for (int32_t i = 0; i < numOfBlocks; ++i) {
+        cout << "Block:" << i << endl;
+        cout << "\tuid:" << be64toh(blk->uid) << ", tid:" << ntohl(blk->tid) << ", padding:" << ntohl(blk->padding) << ", sversion:" << ntohl(blk->sversion)
+            << ", dataLen:" << ntohl(blk->dataLen) << ", schemaLen:" << ntohl(blk->schemaLen) << ", numOfRows:" << ntohs(blk->numOfRows) << endl;
+        blk = (SSubmitBlk*)(blk->data + ntohl(blk->dataLen));
+      }
+    }
+  }
+
+  void checkReslut(int32_t numOfTables, int16_t numOfRows1, int16_t numOfRows2 = -1) {
+    ASSERT_EQ(res_->schemaAttache, 0);
+    ASSERT_EQ(res_->payloadType, PAYLOAD_TYPE_KV);
+    ASSERT_EQ(res_->insertType, TSDB_QUERY_TYPE_INSERT);
+    size_t num = taosArrayGetSize(res_->pDataBlocks);
+    ASSERT_GE(num, 0);
+    for (size_t i = 0; i < num; ++i) {
+      SVgDataBlocks* vg = (SVgDataBlocks*)taosArrayGetP(res_->pDataBlocks, i);
+      ASSERT_EQ(vg->numOfTables, numOfTables);
+      ASSERT_GE(vg->size, 0);
+      SMsgDesc* desc = (SMsgDesc*)(vg->pData);
+      ASSERT_EQ(ntohl(desc->numOfVnodes), 1);
+      SSubmitMsg* submit = (SSubmitMsg*)(desc + 1);
+      ASSERT_GE(ntohl(submit->length), 0);
+      ASSERT_GE(ntohl(submit->numOfBlocks), 0);
+      int32_t numOfBlocks = ntohl(submit->numOfBlocks);
+      SSubmitBlk* blk = (SSubmitBlk*)(submit + 1);
+      for (int32_t i = 0; i < numOfBlocks; ++i) {
+        ASSERT_EQ(ntohs(blk->numOfRows), (0 == i ? numOfRows1 : (numOfRows2 > 0 ? numOfRows2 : numOfRows1)));
+        blk = (SSubmitBlk*)(blk->data + ntohl(blk->dataLen));
+      }
+    }
+  }
+
 private:
   static const int max_err_len = 1024;
   static const int max_sql_len = 1024 * 1024;
@@ -108,15 +132,44 @@ private:
 };
 
 // INSERT INTO tb_name VALUES (field1_value, ...)
-TEST_F(InsertTest, simpleTest) {
+TEST_F(InsertTest, singleTableSingleRowTest) {
+  setDatabase("root", "test");
+
+  bind("insert into t1 values (now, 1, \"beijing\")");
+  ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
+  dumpReslut();
+  checkReslut(1, 1);
+}
+
+// INSERT INTO tb_name VALUES (field1_value, ...)(field1_value, ...)
+TEST_F(InsertTest, singleTableMultiRowTest) {
+  setDatabase("root", "test");
+
+  bind("insert into t1 values (now, 1, \"beijing\")(now+1s, 2, \"shanghai\")(now+2s, 3, \"guangzhou\")");
+  ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
+  dumpReslut();
+  checkReslut(1, 3);
+}
+
+// INSERT INTO tb1_name VALUES (field1_value, ...) tb2_name VALUES (field1_value, ...)
+TEST_F(InsertTest, multiTableSingleRowTest) {
+  setDatabase("root", "test");
+
+  bind("insert into st1s1 values (now, 1, \"beijing\") st1s2 values (now, 10, \"131028\")");
+  ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
+  dumpReslut();
+  checkReslut(2, 1);
+}
+
+// INSERT INTO tb1_name VALUES (field1_value, ...) tb2_name VALUES (field1_value, ...)
+TEST_F(InsertTest, multiTableMultiRowTest) {
   setDatabase("root", "test");
 
-  bind("insert into t1 values (now, 1, \"wxy\")");
+  bind("insert into st1s1 values (now, 1, \"beijing\")(now+1s, 2, \"shanghai\")(now+2s, 3, \"guangzhou\")"
+                  " st1s2 values (now, 10, \"131028\")(now+1s, 20, \"132028\")");
   ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
-  SInsertStmtInfo* res = reslut();
-  // todo check
-  ASSERT_EQ(res->insertType, TSDB_QUERY_TYPE_INSERT);
-  // ASSERT_EQ(taosArrayGetSize(res->pDataBlocks), 1);
+  dumpReslut();
+  checkReslut(2, 3, 2);
 }
 
 TEST_F(InsertTest, toleranceTest) {

source/libs/parser/test/mockCatalog.cpp

@@ -25,15 +25,15 @@ namespace {
 void generateTestT1(MockCatalogService* mcs) {
   ITableBuilder& builder = mcs->createTableBuilder("root.test", "t1", TSDB_NORMAL_TABLE, 3)
       .setPrecision(TSDB_TIME_PRECISION_MILLI).setVgid(1).addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
-      .addColumn("c1", TSDB_DATA_TYPE_INT).addColumn("c2", TSDB_DATA_TYPE_BINARY, 10);
+      .addColumn("c1", TSDB_DATA_TYPE_INT).addColumn("c2", TSDB_DATA_TYPE_BINARY, 20);
   builder.done();
 }
 
 void generateTestST1(MockCatalogService* mcs) {
   ITableBuilder& builder = mcs->createTableBuilder("root.test", "st1", TSDB_SUPER_TABLE, 3, 2)
       .setPrecision(TSDB_TIME_PRECISION_MILLI).addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
-      .addTag("tag1", TSDB_DATA_TYPE_INT).addTag("tag2", TSDB_DATA_TYPE_BINARY, 10)
-      .addColumn("c1", TSDB_DATA_TYPE_INT).addColumn("c2", TSDB_DATA_TYPE_BINARY, 10);
+      .addTag("tag1", TSDB_DATA_TYPE_INT).addTag("tag2", TSDB_DATA_TYPE_BINARY, 20)
+      .addColumn("c1", TSDB_DATA_TYPE_INT).addColumn("c2", TSDB_DATA_TYPE_BINARY, 20);
   builder.done();
   mcs->createSubTable("root.test", "st1", "st1s1", 1);
   mcs->createSubTable("root.test", "st1", "st1s2", 2);

source/libs/parser/test/mockCatalog.h

@@ -22,8 +22,4 @@ void initMetaDataEnv();
 void generateMetaData();
 void destroyMetaDataEnv();
 
-// mock
-// int32_t catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle);
-// int32_t catalogGetTableMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, const char* pTableName, STableMeta** pTableMeta);
-
 #endif  // MOCK_CATALOG_H

source/libs/planner/src/physicalPlan.c

@@ -221,7 +221,7 @@ static SPhyNode* createPhyNode(SPlanContext* pCxt, SQueryPlanNode* pPlanNode) {
     node->pChildren = taosArrayInit(TARRAY_MIN_SIZE, POINTER_BYTES);
     size_t size = taosArrayGetSize(pPlanNode->pChildren);
     for(int32_t i = 0; i < size; ++i) {
-      SPhyNode* child = createPhyNode(pCxt, taosArrayGet(pPlanNode->pChildren, i));
+      SPhyNode* child = createPhyNode(pCxt, taosArrayGetP(pPlanNode->pChildren, i));
       child->pParent = node;
       taosArrayPush(node->pChildren, &child);
     }