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

Merge pull request #9264 from taosdata/feature/index_cache 

update index TFile write
This commit is contained in:
Shengliang Guan 2021-12-22 13:24:43 +08:00 committed by GitHub
parent a630d82cf2 ebcb9be39f
commit 1e32e1f86a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
21 changed files with 1058 additions and 1217 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
include/libs/index/index.h
View File

@ -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
}

18
source/libs/index/inc/indexInt.h
View File

@ -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;

11
source/libs/index/inc/index_cache.h
View File

@ -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
}

220
source/libs/index/inc/index_fst.h
View File

@ -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
}

34
source/libs/index/inc/index_fst_automation.h
View File

@ -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

36
source/libs/index/inc/index_fst_counting_writer.h
View File

@ -28,9 +28,9 @@ 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);
WriterType type;
union {
struct {
@ -39,41 +39,41 @@ 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 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)

12
source/libs/index/inc/index_fst_node.h
View File

@ -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
}

14
source/libs/index/inc/index_fst_registry.h
View File

@ -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
}

22
source/libs/index/inc/index_fst_util.h
View File

@ -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)

59
source/libs/index/inc/index_tfile.h
View File

@ -26,84 +26,89 @@
extern "C" {
#endif
// tfile header
// tfile header content
// |<---suid--->|<---version--->|<--colLen-->|<-colName->|<---type-->|
// |<-uint64_t->|<---int32_t--->|<--int32_t->|<-colLen-->|<-uint8_t->|
typedef struct TFileReadHeader {
typedef struct TFileHeader {
uint64_t suid;
int32_t version;
char colName[128]; //
uint8_t colType;
} TFileReadHeader;
} TFileHeader;
#define TFILE_HEADER_SIZE (sizeof(TFILE_HEADER_SIZE) + sizeof(uint32_t));
#define TFILE_HEADER_SIZE (sizeof(TFileHeader) + sizeof(uint32_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
}

8
source/libs/index/inc/index_util.h
View File

@ -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)

133
source/libs/index/src/index.c
View File

@ -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;
}

78
source/libs/index/src/index_cache.c
View File

@ -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;

626
source/libs/index/src/index_fst.c
View File

File diff suppressed because it is too large Load Diff

113
source/libs/index/src/index_fst_automation.c
View File

@ -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);

272
source/libs/index/src/index_fst_common.c
View File

@ -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',
};

66
source/libs/index/src/index_fst_counting_writer.c
View File

@ -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 +39,7 @@ static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
return nRead;
}
static int writeCtxDoFlush(WriterCtx *ctx) {
static int writeCtxDoFlush(WriterCtx* ctx) {
if (ctx->type == TFile) {
// tfFsync(ctx->fd);
// tfFlush(ctx->file.fd);
@ -51,11 +49,9 @@ 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 +63,8 @@ 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);
@ -83,12 +79,10 @@ WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int
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 +91,53 @@ 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 +147,7 @@ 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;

56
source/libs/index/src/index_fst_node.c
View File

@ -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);
}
}

66
source/libs/index/src/index_fst_registry.c
View File

@ -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);
}

43
source/libs/index/src/index_fst_util.c
View File

@ -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++) {

309
source/libs/index/src/index_tfile.c
View File

@ -21,15 +21,74 @@
#include "index_fst_counting_writer.h"
#include "index_util.h"
#include "taosdef.h"
#include "tcompare.h"
static FORCE_INLINE int tfileReadLoadHeader(TFileReader *reader) {
#define TF_TABLE_TATOAL_SIZE(sz) (sizeof(sz) + sz * sizeof(uint64_t))
typedef struct TFileValue {
char* colVal; // null terminated
SArray* tableId;
int32_t offset;
} TFileValue;
// static tfileGetCompareFunc(uint8_t byte) {}
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* tableIds) {
int tbSz = taosArrayGetSize(tableIds);
SERIALIZE_VAR_TO_BUF(buf, tbSz, int32_t);
for (size_t i = 0; i < tbSz; i++) {
uint64_t* v = taosArrayGet(tableIds, i);
SERIALIZE_VAR_TO_BUF(buf, *v, uint64_t);
}
}
static FORCE_INLINE int tfileWriteHeader(TFileWriter* writer) {
char buf[TFILE_HEADER_SIZE] = {0};
char* p = buf;
TFileHeader* header = &writer->header;
SERIALIZE_MEM_TO_BUF(p, header, suid);
SERIALIZE_MEM_TO_BUF(p, header, version);
SERIALIZE_VAR_TO_BUF(p, strlen(header->colName), int32_t);
SERIALIZE_STR_MEM_TO_BUF(p, header, colName, strlen(header->colName));
SERIALIZE_MEM_TO_BUF(p, header, colType);
int offset = p - buf;
int nwrite = writer->ctx->write(writer->ctx, buf, offset);
if (offset != nwrite) { return -1; }
writer->offset = offset;
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
}
}
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;
char* p = buf;
int64_t nread = reader->ctx->read(reader->ctx, buf, TFILE_HADER_PRE_SIZE);
assert(nread == TFILE_HADER_PRE_SIZE);
TFileHeader* header = &reader->header;
memcpy(&header->suid, p, sizeof(header->suid));
p += sizeof(header->suid);
@ -44,43 +103,44 @@ static FORCE_INLINE int tfileReadLoadHeader(TFileReader *reader) {
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;
}
}
struct dirent *entry;
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);
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;
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;
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) {
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) {
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);
@ -90,38 +150,47 @@ static void tfileSerialCacheKey(TFileCacheKey *key, char *buf) {
SERIALIZE_STR_MEM_TO_BUF(buf, key, colName, key->nColName);
}
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);
tfileReaderDestroy(reader);
indexError("failed to load index header, index Id: %s", file);
goto End;
}
// 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 +199,170 @@ 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);
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;
// 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)) {
//
} 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) {
//
//
}
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);
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 and
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; }
IndexTFile *indexTFileCreate(const char *path) {
IndexTFile *tfile = calloc(1, sizeof(IndexTFile));
tfile->cache = tfileCacheCreate(path);
return tfile;
}
void IndexTFileDestroy(IndexTFile *tfile) { free(tfile); }
int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result) {
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 0;
void IndexTFileDestroy(IndexTFile* tfile) {
free(tfile);
}
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};
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;
}