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

update index TFile manage and format code

This commit is contained in:
yihaoDeng 2021-12-21 16:36:31 +08:00
parent 6fe118c516
commit 5d4d7b47a0
22 changed files with 2205 additions and 2023 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
include/libs/index/index.h
View File

@ -39,11 +39,11 @@ typedef enum {
} SIndexOperOnColumn; } SIndexOperOnColumn;
typedef enum { MUST = 0, SHOULD = 1, NOT = 2 } EIndexOperatorType; typedef enum { MUST = 0, SHOULD = 1, NOT = 2 } EIndexOperatorType;
typedef enum { QUERY_TERM = 0, QUERY_PREFIX = 1, QUERY_SUFFIX = 2,QUERY_REGEX = 3} EIndexQueryType; typedef enum { QUERY_TERM = 0, QUERY_PREFIX = 1, QUERY_SUFFIX = 2, QUERY_REGEX = 3 } EIndexQueryType;
/* /*
* @param: oper * @param: oper
* *
*/ */
SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType oper); SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType oper);
void indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery); void indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery);
int indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType type); int indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType type);
@ -76,11 +76,10 @@ void indexOptsDestroy(SIndexOpts *opts);
* @param: * @param:
*/ */
SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn operType, uint8_t colType, SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn operType, uint8_t colType, const char *colName,
const char *colName, int32_t nColName, const char *colVal, int32_t nColVal); int32_t nColName, const char *colVal, int32_t nColVal);
void indexTermDestroy(SIndexTerm *p); void indexTermDestroy(SIndexTerm *p);
#ifdef __cplusplus #ifdef __cplusplus
} }
#endif #endif

67
source/libs/index/inc/indexInt.h
View File

@ -18,20 +18,19 @@
#include "index.h" #include "index.h"
#include "index_fst.h" #include "index_fst.h"
#include "tlog.h"
#include "thash.h"
#include "taos.h" #include "taos.h"
#include "thash.h"
#include "tlog.h"
#ifdef USE_LUCENE #ifdef USE_LUCENE
#include <lucene++/Lucene_c.h> #include <lucene++/Lucene_c.h>
#endif #endif
#ifdef __cplusplus #ifdef __cplusplus
extern "C" { extern "C" {
#endif #endif
typedef enum {kTypeValue, kTypeDeletion} STermValueType ; typedef enum { kTypeValue, kTypeDeletion } STermValueType;
typedef struct SIndexStat { typedef struct SIndexStat {
int32_t totalAdded; // int32_t totalAdded; //
@ -45,9 +44,9 @@ struct SIndex {
#ifdef USE_LUCENE #ifdef USE_LUCENE
index_t *index; index_t *index;
#endif #endif
void *cache; void * cache;
void *tindex; void * tindex;
SHashObj *colObj;// < field name, field id> SHashObj *colObj; // < field name, field id>
int64_t suid; // current super table id, -1 is normal table int64_t suid; // current super table id, -1 is normal table
int colId; // field id allocated to cache int colId; // field id allocated to cache
@ -66,12 +65,11 @@ struct SIndexOpts {
int32_t cacheSize; // MB int32_t cacheSize; // MB
// add cache module later // add cache module later
#endif #endif
}; };
struct SIndexMultiTermQuery { struct SIndexMultiTermQuery {
EIndexOperatorType opera; EIndexOperatorType opera;
SArray *query; SArray * query;
}; };
// field and key; // field and key;
@ -79,26 +77,53 @@ typedef struct SIndexTerm {
int64_t suid; int64_t suid;
SIndexOperOnColumn operType; // oper type, add/del/update SIndexOperOnColumn operType; // oper type, add/del/update
uint8_t colType; // term data type, str/interger/json uint8_t colType; // term data type, str/interger/json
char *colName; char * colName;
int32_t nColName; int32_t nColName;
char *colVal; char * colVal;
int32_t nColVal; int32_t nColVal;
} SIndexTerm; } SIndexTerm;
typedef struct SIndexTermQuery { typedef struct SIndexTermQuery {
SIndexTerm* term; SIndexTerm * term;
EIndexQueryType qType; EIndexQueryType qType;
} SIndexTermQuery; } SIndexTermQuery;
#define indexFatal(...) \
do { \
#define indexFatal(...) do { if (sDebugFlag & DEBUG_FATAL) { taosPrintLog("index FATAL ", 255, __VA_ARGS__); }} while(0) if (sDebugFlag & DEBUG_FATAL) { \
#define indexError(...) do { if (sDebugFlag & DEBUG_ERROR) { taosPrintLog("index ERROR ", 255, __VA_ARGS__); }} while(0) taosPrintLog("index FATAL ", 255, __VA_ARGS__); \
#define indexWarn(...) do { if (sDebugFlag & DEBUG_WARN) { taosPrintLog("index WARN ", 255, __VA_ARGS__); }} while(0) } \
#define indexInfo(...) do { if (sDebugFlag & DEBUG_INFO) { taosPrintLog("index ", 255, __VA_ARGS__); }} while(0) } while (0)
#define indexDebug(...) do { if (sDebugFlag & DEBUG_DEBUG) { taosPrintLog("index ", sDebugFlag, __VA_ARGS__); }} while(0) #define indexError(...) \
#define indexTrace(...) do { if (sDebugFlag & DEBUG_TRACE) { taosPrintLog("index ", sDebugFlag, __VA_ARGS__); }} while(0) do { \
if (sDebugFlag & DEBUG_ERROR) { \
taosPrintLog("index ERROR ", 255, __VA_ARGS__); \
} \
} while (0)
#define indexWarn(...) \
do { \
if (sDebugFlag & DEBUG_WARN) { \
taosPrintLog("index WARN ", 255, __VA_ARGS__); \
} \
} while (0)
#define indexInfo(...) \
do { \
if (sDebugFlag & DEBUG_INFO) { \
taosPrintLog("index ", 255, __VA_ARGS__); \
} \
} while (0)
#define indexDebug(...) \
do { \
if (sDebugFlag & DEBUG_DEBUG) { \
taosPrintLog("index ", sDebugFlag, __VA_ARGS__); \
} \
} while (0)
#define indexTrace(...) \
do { \
if (sDebugFlag & DEBUG_TRACE) { \
taosPrintLog("index ", sDebugFlag, __VA_ARGS__); \
} \
} while (0)
#ifdef __cplusplus #ifdef __cplusplus
} }

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

@ -22,8 +22,10 @@
// ----------------- key structure in skiplist --------------------- // ----------------- key structure in skiplist ---------------------
/* A data row, the format is like below: /* A data row, the format is like below:
* content: |<--totalLen-->|<-- fieldid-->|<--field type -->|<-- value len--->|<-- value -->|<-- uid -->|<--version--->|<-- itermType -->| * content: |<--totalLen-->|<-- fieldid-->|<--field type-->|<-- value len--->|
* len : |<--int32_t -->|<-- int16_t-->|<-- int8_t --->|<--- int32_t --->|<--valuelen->|<--uint64_t->|<-- int32_t-->|<-- int8_t --->| * |<-- value -->|<--uid -->|<--version--->|<-- itermType -->|
* len : |<--int32_t -->|<-- int16_t-->|<-- int8_t --->|<--- int32_t --->|
* <--valuelen->|<--uint64_t->| * <-- int32_t-->|<-- int8_t --->|
*/ */
#ifdef __cplusplus #ifdef __cplusplus
@ -35,7 +37,6 @@ typedef struct IndexCache {
SSkipList *skiplist; SSkipList *skiplist;
} IndexCache; } IndexCache;
// //
IndexCache *indexCacheCreate(); IndexCache *indexCacheCreate();
@ -43,13 +44,12 @@ 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 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 #ifdef __cplusplus
} }
#endif #endif
#endif #endif

132
source/libs/index/inc/index_fst.h
View File

@ -20,11 +20,11 @@
extern "C" { extern "C" {
#endif #endif
#include "tarray.h"
#include "index_fst_util.h"
#include "index_fst_registry.h"
#include "index_fst_counting_writer.h"
#include "index_fst_automation.h" #include "index_fst_automation.h"
#include "index_fst_counting_writer.h"
#include "index_fst_registry.h"
#include "index_fst_util.h"
#include "tarray.h"
#define OUTPUT_PREFIX(a, b) ((a) > (b) ? (b) : (a) #define OUTPUT_PREFIX(a, b) ((a) > (b) ? (b) : (a)
@ -32,7 +32,7 @@ typedef struct Fst Fst;
typedef struct FstNode FstNode; typedef struct FstNode FstNode;
typedef struct StreamWithState StreamWithState; typedef struct StreamWithState StreamWithState;
typedef enum { Included, Excluded, Unbounded} FstBound; typedef enum { Included, Excluded, Unbounded } FstBound;
typedef struct FstBoundWithData { typedef struct FstBoundWithData {
FstSlice data; FstSlice data;
@ -40,8 +40,8 @@ typedef struct FstBoundWithData {
} FstBoundWithData; } FstBoundWithData;
typedef struct FstStreamBuilder { typedef struct FstStreamBuilder {
Fst *fst; Fst * fst;
AutomationCtx *aut; AutomationCtx * aut;
FstBoundWithData *min; FstBoundWithData *min;
FstBoundWithData *max; FstBoundWithData *max;
} FstStreamBuilder, FstStreamWithStateBuilder; } FstStreamBuilder, FstStreamWithStateBuilder;
@ -51,18 +51,15 @@ typedef struct FstRange {
uint64_t end; uint64_t end;
} FstRange; } FstRange;
typedef enum { GE, GT, LE, LT } RangeType;
typedef enum { OneTransNext, OneTrans, AnyTrans, EmptyFinal } State;
typedef enum { Ordered, OutOfOrdered, DuplicateKey } OrderType;
typedef enum {GE, GT, LE, LT} RangeType; FstBoundWithData *fstBoundStateCreate(FstBound type, FstSlice *data);
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 fstBoundWithDataExceededBy(FstBoundWithData *bound, FstSlice *slice);
bool fstBoundWithDataIsEmpty(FstBoundWithData *bound); bool fstBoundWithDataIsEmpty(FstBoundWithData *bound);
bool fstBoundWithDataIsIncluded(FstBoundWithData *bound); bool fstBoundWithDataIsIncluded(FstBoundWithData *bound);
typedef struct FstOutput { typedef struct FstOutput {
bool null; bool null;
Output out; Output out;
@ -82,76 +79,71 @@ typedef struct FstUnFinishedNodes {
FstUnFinishedNodes *fstUnFinishedNodesCreate(); FstUnFinishedNodes *fstUnFinishedNodesCreate();
void fstUnFinishedNodesDestroy(FstUnFinishedNodes *node); void fstUnFinishedNodesDestroy(FstUnFinishedNodes *node);
void fstUnFinishedNodesPushEmpty(FstUnFinishedNodes *nodes, bool isFinal); void fstUnFinishedNodesPushEmpty(FstUnFinishedNodes *nodes, bool isFinal);
FstBuilderNode *fstUnFinishedNodesPopRoot(FstUnFinishedNodes *nodes);
FstBuilderNode *fstUnFinishedNodesPopFreeze(FstUnFinishedNodes *nodes, CompiledAddr addr);
FstBuilderNode *fstUnFinishedNodesPopEmpty(FstUnFinishedNodes *nodes);
void fstUnFinishedNodesSetRootOutput(FstUnFinishedNodes *node, Output out); void fstUnFinishedNodesSetRootOutput(FstUnFinishedNodes *node, Output out);
void fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes *node, CompiledAddr addr); void fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes *node, CompiledAddr addr);
void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *node, FstSlice bs, Output out); void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *node, FstSlice bs, Output out);
uint64_t fstUnFinishedNodesFindCommPrefix(FstUnFinishedNodes *node, FstSlice bs); 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 { typedef struct FstBuilder {
FstCountingWriter *wrt; // The FST raw data is written directly to `wtr`. FstCountingWriter * wrt; // The FST raw data is written directly to `wtr`.
FstUnFinishedNodes *unfinished; // The stack of unfinished nodes FstUnFinishedNodes *unfinished; // The stack of unfinished nodes
FstRegistry* registry; // A map of finished nodes. FstRegistry * registry; // A map of finished nodes.
FstSlice last; // The last word added FstSlice last; // The last word added
CompiledAddr lastAddr; // The address of the last compiled node CompiledAddr lastAddr; // The address of the last compiled node
uint64_t len; // num of keys added uint64_t len; // num of keys added
} FstBuilder; } FstBuilder;
FstBuilder *fstBuilderCreate(void *w, FstType ty); FstBuilder *fstBuilderCreate(void *w, FstType ty);
void fstBuilderDestroy(FstBuilder *b); void fstBuilderDestroy(FstBuilder *b);
void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in); void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in);
bool fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in); bool fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in);
OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup);
void fstBuilderCompileFrom(FstBuilder *b, uint64_t istate); void fstBuilderCompileFrom(FstBuilder *b, uint64_t istate);
CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn); void * fstBuilerIntoInner(FstBuilder *b);
void* fstBuilerIntoInner(FstBuilder *b);
void fstBuilderFinish(FstBuilder *b); void fstBuilderFinish(FstBuilder *b);
OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup);
CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn);
typedef struct FstTransitions { typedef struct FstTransitions {
FstNode *node; FstNode *node;
FstRange range; FstRange range;
} FstTransitions; } FstTransitions;
//FstState and relation function // FstState and relation function
typedef struct FstState { typedef struct FstState {
State state; State state;
uint8_t val; uint8_t val;
} FstState; } FstState;
FstState fstStateCreateFrom(FstSlice* data, CompiledAddr addr); FstState fstStateCreateFrom(FstSlice *data, CompiledAddr addr);
FstState fstStateCreate(State state); FstState fstStateCreate(State state);
//compile // compile
void fstStateCompileForOneTransNext(FstCountingWriter *w, CompiledAddr addr, uint8_t inp); void fstStateCompileForOneTransNext(FstCountingWriter *w, CompiledAddr addr, uint8_t inp);
void fstStateCompileForOneTrans(FstCountingWriter *w, CompiledAddr addr, FstTransition *trn); void fstStateCompileForOneTrans(FstCountingWriter *w, CompiledAddr addr, FstTransition *trn);
void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuilderNode *node); void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuilderNode *node);
// set_comm_input // set_comm_input
void fstStateSetCommInput(FstState* state, uint8_t inp); void fstStateSetCommInput(FstState *state, uint8_t inp);
// comm_input // comm_input
uint8_t fstStateCommInput(FstState* state, bool *null); uint8_t fstStateCommInput(FstState *state, bool *null);
// input_len // input_len
uint64_t fstStateInputLen(FstState* state); uint64_t fstStateInputLen(FstState *state);
// end_addr // end_addr
uint64_t fstStateEndAddrForOneTransNext(FstState* state, FstSlice *data); uint64_t fstStateEndAddrForOneTransNext(FstState *state, FstSlice *data);
uint64_t fstStateEndAddrForOneTrans(FstState *state, FstSlice *data, PackSizes sizes); 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 fstStateEndAddrForAnyTrans(
FstState *state, uint64_t version, FstSlice *date, PackSizes sizes, uint64_t nTrans);
// input // input
uint8_t fstStateInput(FstState *state, FstNode *node); uint8_t fstStateInput(FstState *state, FstNode *node);
uint8_t fstStateInputForAnyTrans(FstState *state, FstNode *node, uint64_t i); uint8_t fstStateInputForAnyTrans(FstState *state, FstNode *node, uint64_t i);
@ -180,14 +172,11 @@ uint64_t fstStateNtrans(FstState *state, FstSlice *slice);
Output fstStateFinalOutput(FstState *state, uint64_t version, FstSlice *date, PackSizes sizes, uint64_t nTrans); 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); 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_NEXT(node) (node->state.state == OneTransNext)
#define FST_STATE_ONE_TRNAS(node) (node->state.state == OneTrans) #define FST_STATE_ONE_TRNAS(node) (node->state.state == OneTrans)
#define FST_STATE_ANY_TRANS(node) (node->state.state == AnyTrans) #define FST_STATE_ANY_TRANS(node) (node->state.state == AnyTrans)
#define FST_STATE_EMPTY_FINAL(node) (node->state.state == EmptyFinal) #define FST_STATE_EMPTY_FINAL(node) (node->state.state == EmptyFinal)
typedef struct FstLastTransition { typedef struct FstLastTransition {
uint8_t inp; uint8_t inp;
Output out; Output out;
@ -198,13 +187,12 @@ typedef struct FstLastTransition {
* TODO: simple function name * TODO: simple function name
*/ */
typedef struct FstBuilderNodeUnfinished { typedef struct FstBuilderNodeUnfinished {
FstBuilderNode *node; FstBuilderNode * node;
FstLastTransition* last; FstLastTransition *last;
} FstBuilderNodeUnfinished; } 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);
/* /*
@ -222,27 +210,31 @@ typedef struct FstNode {
Output finalOutput; Output finalOutput;
} FstNode; } FstNode;
// If this node is final and has a terminal output value, then it is, returned. Otherwise, a zero output is returned // If this node is final and has a terminal output value, then it is, returned.
// Otherwise, a zero output is returned
#define FST_NODE_FINAL_OUTPUT(node) node->finalOutput #define FST_NODE_FINAL_OUTPUT(node) node->finalOutput
// Returns true if and only if this node corresponds to a final or "match", state in the finite state transducer. // Returns true if and only if this node corresponds to a final or "match",
// state in the finite state transducer.
#define FST_NODE_IS_FINAL(node) node->isFinal #define FST_NODE_IS_FINAL(node) node->isFinal
// Returns the number of transitions in this node, The maximum number of transitions is 256. // Returns the number of transitions in this node, The maximum number of
// transitions is 256.
#define FST_NODE_LEN(node) node->nTrans #define FST_NODE_LEN(node) node->nTrans
// Returns true if and only if this node has zero transitions. // Returns true if and only if this node has zero transitions.
#define FST_NODE_IS_EMPTYE(node) (node->nTrans == 0) #define FST_NODE_IS_EMPTYE(node) (node->nTrans == 0)
// Return the address of this node. // Return the address of this node.
#define FST_NODE_ADDR(node) node->start #define FST_NODE_ADDR(node) node->start
FstNode *fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice *data); FstNode *fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice *data);
void fstNodeDestroy(FstNode *fstNode); void fstNodeDestroy(FstNode *fstNode);
FstTransitions fstNodeTransitionIter(FstNode *node); FstTransitions fstNodeTransitionIter(FstNode *node);
FstTransitions* fstNodeTransitions(FstNode *node); FstTransitions *fstNodeTransitions(FstNode *node);
bool fstNodeGetTransitionAt(FstNode *node, uint64_t i, FstTransition *res); bool fstNodeGetTransitionAt(FstNode *node, uint64_t i, FstTransition *res);
bool fstNodeGetTransitionAddrAt(FstNode *node, uint64_t i, CompiledAddr *res); bool fstNodeGetTransitionAddrAt(FstNode *node, uint64_t i, CompiledAddr *res);
bool fstNodeFindInput(FstNode *node, uint8_t b, uint64_t *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 // ops
@ -255,7 +247,6 @@ typedef struct FstIndexedValue {
FstLastTransition *fstLastTransitionCreate(uint8_t inp, Output out); FstLastTransition *fstLastTransitionCreate(uint8_t inp, Output out);
void fstLastTransitionDestroy(FstLastTransition *trn); void fstLastTransitionDestroy(FstLastTransition *trn);
typedef struct FstMeta { typedef struct FstMeta {
uint64_t version; uint64_t version;
CompiledAddr rootAddr; CompiledAddr rootAddr;
@ -265,67 +256,69 @@ typedef struct FstMeta {
} FstMeta; } FstMeta;
typedef struct Fst { typedef struct Fst {
FstMeta *meta; FstMeta * meta;
FstSlice *data; // FstSlice *data; //
FstNode *root; // FstNode * root; //
} Fst; } Fst;
// refactor simple function // refactor simple function
Fst* fstCreate(FstSlice *data); Fst *fstCreate(FstSlice *data);
void fstDestroy(Fst *fst); void fstDestroy(Fst *fst);
bool fstGet(Fst *fst, FstSlice *b, Output *out); bool fstGet(Fst *fst, FstSlice *b, Output *out);
FstNode* fstGetNode(Fst *fst, CompiledAddr); FstNode * fstGetNode(Fst *fst, CompiledAddr);
FstNode* fstGetRoot(Fst *fst); FstNode * fstGetRoot(Fst *fst);
FstType fstGetType(Fst *fst); FstType fstGetType(Fst *fst);
CompiledAddr fstGetRootAddr(Fst *fst); CompiledAddr fstGetRootAddr(Fst *fst);
Output fstEmptyFinalOutput(Fst *fst, bool *null); Output fstEmptyFinalOutput(Fst *fst, bool *null);
FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx); FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx);
FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx);
FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx);
// into stream to expand later // into stream to expand later
StreamWithState* streamBuilderIntoStream(FstStreamBuilder *sb); StreamWithState *streamBuilderIntoStream(FstStreamBuilder *sb);
bool fstVerify(Fst *fst); bool fstVerify(Fst *fst);
// refactor this function
//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 { typedef struct StreamState {
FstNode *node; FstNode * node;
uint64_t trans; uint64_t trans;
FstOutput out; FstOutput out;
void *autState; void * autState;
} StreamState; } StreamState;
void streamStateDestroy(void *s); void streamStateDestroy(void *s);
typedef struct StreamWithState { typedef struct StreamWithState {
Fst *fst; Fst * fst;
AutomationCtx *aut; AutomationCtx * aut;
SArray *inp; SArray * inp;
FstOutput emptyOutput; FstOutput emptyOutput;
SArray *stack; // <StreamState> SArray * stack; // <StreamState>
FstBoundWithData *endAt; FstBoundWithData *endAt;
} StreamWithState; } StreamWithState;
typedef struct StreamWithStateResult { typedef struct StreamWithStateResult {
FstSlice data; FstSlice data;
FstOutput out; FstOutput out;
void *state; void * state;
} StreamWithStateResult; } StreamWithStateResult;
StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *state); StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *state);
void swsResultDestroy(StreamWithStateResult *result); void swsResultDestroy(StreamWithStateResult *result);
typedef void* (*StreamCallback)(void *); typedef void *(*StreamCallback)(void *);
StreamWithState *streamWithStateCreate(Fst *fst, AutomationCtx *automation, FstBoundWithData *min, FstBoundWithData *max) ; 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 // set up bound range
@ -333,7 +326,6 @@ FstStreamBuilder *fstStreamBuilderCreate(Fst *fst, AutomationCtx *aut);
FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type); FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type);
#ifdef __cplusplus #ifdef __cplusplus
} }
#endif #endif

24
source/libs/index/inc/index_fst_automation.h
View File

@ -21,13 +21,9 @@ extern "C" {
#include "index_fst_util.h" #include "index_fst_util.h"
typedef struct AutomationCtx AutomationCtx; typedef struct AutomationCtx AutomationCtx;
typedef enum AutomationType { typedef enum AutomationType { AUTOMATION_PREFIX, AUTMMATION_MATCH } AutomationType;
AUTOMATION_PREFIX,
AUTMMATION_MATCH
} AutomationType;
typedef struct StartWith { typedef struct StartWith {
AutomationCtx *autoSelf; AutomationCtx *autoSelf;
@ -40,12 +36,11 @@ typedef struct Complement {
// automation // automation
typedef struct AutomationCtx { typedef struct AutomationCtx {
AutomationType type; AutomationType type;
void *stdata; void * stdata;
char *data; char * data;
} AutomationCtx; } AutomationCtx;
typedef enum ValueType { FST_INT, FST_CHAR, FST_ARRAY } ValueType;
typedef enum ValueType { FST_INT, FST_CHAR, FST_ARRAY} ValueType;
typedef enum StartWithStateKind { Done, Running } StartWithStateKind; typedef enum StartWithStateKind { Done, Running } StartWithStateKind;
typedef struct StartWithStateValue { typedef struct StartWithStateValue {
@ -53,24 +48,23 @@ typedef struct StartWithStateValue {
ValueType type; ValueType type;
union { union {
int val; int val;
char *ptr; char * ptr;
SArray *arr; SArray *arr;
// add more type // add more type
} ; };
} StartWithStateValue; } StartWithStateValue;
StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void *val); StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void *val);
StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv); StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv);
void startWithStateValueDestroy(void *sv); void startWithStateValueDestroy(void *sv);
typedef struct AutomationFunc { typedef struct AutomationFunc {
void* (*start)(AutomationCtx *ctx) ; void *(*start)(AutomationCtx *ctx);
bool (*isMatch)(AutomationCtx *ctx, void *); bool (*isMatch)(AutomationCtx *ctx, void *);
bool (*canMatch)(AutomationCtx *ctx, void *data); bool (*canMatch)(AutomationCtx *ctx, void *data);
bool (*willAlwaysMatch)(AutomationCtx *ctx, void *state); bool (*willAlwaysMatch)(AutomationCtx *ctx, void *state);
void* (*accept)(AutomationCtx *ctx, void *state, uint8_t byte); void *(*accept)(AutomationCtx *ctx, void *state, uint8_t byte);
void* (*acceptEof)(AutomationCtx *ct, void *state); void *(*acceptEof)(AutomationCtx *ct, void *state);
} AutomationFunc; } AutomationFunc;
AutomationCtx *automCtxCreate(void *data, AutomationType atype); AutomationCtx *automCtxCreate(void *data, AutomationType atype);

4
source/libs/index/inc/index_fst_common.h
View File

@ -1,9 +1,9 @@
#ifndef __INDEX_FST_COMM_H__ #ifndef __INDEX_FST_COMM_H__
#define __INDEX_FST_COMM_H__ #define __INDEX_FST_COMM_H__
#include "tutil.h"
extern const uint8_t COMMON_INPUTS[]; extern const uint8_t COMMON_INPUTS[];
extern char const COMMON_INPUTS_INV[]; extern const char COMMON_INPUTS_INV[];
#ifdef __cplusplus #ifdef __cplusplus
extern "C" { extern "C" {

17
source/libs/index/inc/index_fst_counting_writer.h
View File

@ -22,11 +22,10 @@ extern "C" {
#include "tfile.h" #include "tfile.h"
#define DefaultMem 1024 * 1024
#define DefaultMem 1024*1024
static char tmpFile[] = "./index"; static char tmpFile[] = "./index";
typedef enum WriterType {TMemory, TFile} WriterType; typedef enum WriterType { TMemory, TFile } WriterType;
typedef struct WriterCtx { typedef struct WriterCtx {
int (*write)(struct WriterCtx *ctx, uint8_t *buf, int len); int (*write)(struct WriterCtx *ctx, uint8_t *buf, int len);
@ -40,7 +39,7 @@ typedef struct WriterCtx {
} file; } file;
struct { struct {
int32_t capa; int32_t capa;
char *buf; char * buf;
} mem; } mem;
}; };
int32_t offset; int32_t offset;
@ -51,14 +50,13 @@ static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len);
static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len); static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len);
static int writeCtxDoFlush(WriterCtx *ctx); static int writeCtxDoFlush(WriterCtx *ctx);
WriterCtx* writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity); WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity);
void writerCtxDestroy(WriterCtx *w); void writerCtxDestroy(WriterCtx *w);
typedef uint32_t CheckSummer; typedef uint32_t CheckSummer;
typedef struct FstCountingWriter { 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; uint64_t count;
CheckSummer summer; CheckSummer summer;
} FstCountingWriter; } FstCountingWriter;
@ -69,17 +67,14 @@ 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); FstCountingWriter *fstCountingWriterCreate(void *wtr);
void fstCountingWriterDestroy(FstCountingWriter *w); void fstCountingWriterDestroy(FstCountingWriter *w);
void fstCountingWriterPackUintIn(FstCountingWriter *writer, uint64_t n, uint8_t nBytes); void fstCountingWriterPackUintIn(FstCountingWriter *writer, uint64_t n, uint8_t nBytes);
uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n); uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n);
#define FST_WRITER_COUNT(writer) (writer->count) #define FST_WRITER_COUNT(writer) (writer->count)
#define FST_WRITER_INTER_WRITER(writer) (writer->wtr) #define FST_WRITER_INTER_WRITER(writer) (writer->wtr)
#define FST_WRITE_CHECK_SUMMER(writer) (writer->summer) #define FST_WRITE_CHECK_SUMMER(writer) (writer->summer)
@ -89,5 +84,3 @@ uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n);
#endif #endif
#endif #endif

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

@ -20,17 +20,17 @@
extern "C" { extern "C" {
#endif #endif
#include "index_fst_util.h"
#include "index_fst_counting_writer.h" #include "index_fst_counting_writer.h"
#include "index_fst_util.h"
#define FST_BUILDER_NODE_IS_FINAL(bn) (bn->isFinal) #define FST_BUILDER_NODE_IS_FINAL(bn) (bn->isFinal)
#define FST_BUILDER_NODE_TRANS_ISEMPTY(bn) (taosArrayGetSize(bn->trans) == 0) #define FST_BUILDER_NODE_TRANS_ISEMPTY(bn) (taosArrayGetSize(bn->trans) == 0)
#define FST_BUILDER_NODE_FINALOUTPUT_ISZERO(bn) (bn->finalOutput == 0) #define FST_BUILDER_NODE_FINALOUTPUT_ISZERO(bn) (bn->finalOutput == 0)
typedef struct FstTransition { typedef struct FstTransition {
uint8_t inp; //The byte input associated with this transition. uint8_t inp; // The byte input associated with this transition.
Output out; //The output associated with this transition Output out; // The output associated with this transition
CompiledAddr addr; //The address of the node that this transition points to CompiledAddr addr; // The address of the node that this transition points to
} FstTransition; } FstTransition;
typedef struct FstBuilderNode { typedef struct FstBuilderNode {
@ -45,7 +45,8 @@ 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 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);

23
source/libs/index/inc/index_fst_registry.h
View File

@ -19,9 +19,9 @@
extern "C" { extern "C" {
#endif #endif
#include "index_fst_node.h"
#include "index_fst_util.h" #include "index_fst_util.h"
#include "tarray.h" #include "tarray.h"
#include "index_fst_node.h"
typedef struct FstRegistryCell { typedef struct FstRegistryCell {
CompiledAddr addr; CompiledAddr addr;
@ -29,38 +29,37 @@ typedef struct FstRegistryCell {
} FstRegistryCell; } FstRegistryCell;
#define FST_REGISTRY_CELL_IS_EMPTY(cell) (cell->addr == NONE_ADDRESS) #define FST_REGISTRY_CELL_IS_EMPTY(cell) (cell->addr == NONE_ADDRESS)
#define FST_REGISTRY_CELL_INSERT(cell, tAddr) do {cell->addr = tAddr;} while(0) #define FST_REGISTRY_CELL_INSERT(cell, tAddr) \
do { \
cell->addr = tAddr; \
} while (0)
// typedef struct FstRegistryCache {
//typedef struct FstRegistryCache {
// SArray *cells; // SArray *cells;
// uint32_t start; // uint32_t start;
// uint32_t end; // uint32_t end;
//} FstRegistryCache; //} FstRegistryCache;
typedef enum {FOUND, NOTFOUND, REJECTED} FstRegistryEntryState; typedef enum { FOUND, NOTFOUND, REJECTED } FstRegistryEntryState;
typedef struct FstRegistryEntry { typedef struct FstRegistryEntry {
FstRegistryEntryState state; FstRegistryEntryState state;
CompiledAddr addr; CompiledAddr addr;
FstRegistryCell *cell; FstRegistryCell * cell;
} FstRegistryEntry; } FstRegistryEntry;
// Registry relation function // Registry relation function
typedef struct FstRegistry { typedef struct FstRegistry {
SArray *table; //<FstRegistryCell> SArray * table; //<FstRegistryCell>
uint64_t tableSize; // num of rows uint64_t tableSize; // num of rows
uint64_t mruSize; // num of columns uint64_t mruSize; // num of columns
} FstRegistry; } FstRegistry;
// //
FstRegistry* fstRegistryCreate(uint64_t tableSize, uint64_t mruSize); FstRegistry *fstRegistryCreate(uint64_t tableSize, uint64_t mruSize);
void fstRegistryDestroy(FstRegistry *registry); void fstRegistryDestroy(FstRegistry *registry);
FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNode);
FstRegistryEntry* fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNode);
void fstRegistryEntryDestroy(FstRegistryEntry *entry); void fstRegistryEntryDestroy(FstRegistryEntry *entry);
#ifdef __cplusplus #ifdef __cplusplus

54
source/libs/index/inc/index_fst_util.h
View File

@ -13,7 +13,6 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>. * along with this program. If not, see <http://www.gnu.org/licenses/>.
*/ */
#ifndef __INDEX_FST_UTIL_H__ #ifndef __INDEX_FST_UTIL_H__
#define __INDEX_FST_UTIL_H__ #define __INDEX_FST_UTIL_H__
@ -21,16 +20,15 @@
extern "C" { extern "C" {
#endif #endif
#include "tarray.h"
#include "index_fst_common.h" #include "index_fst_common.h"
#include "tarray.h"
typedef uint64_t FstType; typedef uint64_t FstType;
typedef uint64_t CompiledAddr; typedef uint64_t CompiledAddr;
typedef uint64_t Output; typedef uint64_t Output;
typedef uint8_t PackSizes; typedef uint8_t PackSizes;
// A sentinel value used to indicate an empty final state
//A sentinel value used to indicate an empty final state
extern const CompiledAddr EMPTY_ADDRESS; extern const CompiledAddr EMPTY_ADDRESS;
/// A sentinel value used to indicate an invalid state. /// A sentinel value used to indicate an invalid state.
extern const CompiledAddr NONE_ADDRESS; extern const CompiledAddr NONE_ADDRESS;
@ -48,29 +46,33 @@ extern const uint64_t TRANS_INDEX_THRESHOLD;
// //
// `0` is a legal value which means there are no transitions/outputs // `0` is a legal value which means there are no transitions/outputs
#define FST_SET_TRANSITION_PACK_SIZE(v, sz) \
do { \
v = (v & 0b00001111) | (sz << 4); \
} while (0)
#define FST_GET_TRANSITION_PACK_SIZE(v) (((v)&0b11110000) >> 4)
#define FST_SET_OUTPUT_PACK_SIZE(v, sz) \
do { \
v = (v & 0b11110000) | sz; \
} while (0)
#define FST_GET_OUTPUT_PACK_SIZE(v) ((v)&0b00001111)
#define FST_SET_TRANSITION_PACK_SIZE(v, sz) do {v = (v & 0b00001111) | (sz << 4); } while(0) #define COMMON_INPUT(idx) COMMON_INPUTS_INV[(idx)-1]
#define FST_GET_TRANSITION_PACK_SIZE(v) (((v) & 0b11110000) >> 4)
#define FST_SET_OUTPUT_PACK_SIZE(v, sz) do { v = (v & 0b11110000) | sz; } while(0)
#define FST_GET_OUTPUT_PACK_SIZE(v) ((v) & 0b00001111)
#define COMMON_INPUT(idx) COMMON_INPUTS_INV[(idx) - 1] #define COMMON_INDEX(v, max, val) \
do { \
val = ((uint16_t)COMMON_INPUTS[v] + 1) % 256; \
val = val > max ? 0 : val; \
} while (0)
#define COMMON_INDEX(v, max, val) do { \ // uint8_t commonInput(uint8_t idx);
val = ((uint16_t)COMMON_INPUTS[v] + 1)%256; \ // uint8_t commonIdx(uint8_t v, uint8_t max);
val = val > max ? 0: val; \
} while(0)
//uint8_t commonInput(uint8_t idx);
//uint8_t commonIdx(uint8_t v, uint8_t max);
uint8_t packSize(uint64_t n); 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); 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 { typedef struct FstString {
uint8_t *data; uint8_t *data;
uint32_t len; uint32_t len;
@ -95,9 +97,9 @@ uint8_t *fstSliceData(FstSlice *s, int32_t *sz);
//// stack //// stack
// //
//typedef (*StackFreeElemFn)(void *elem); // typedef (*StackFreeElemFn)(void *elem);
// //
//typedef struct FstStack { // typedef struct FstStack {
// void *first; // void *first;
// void *end; // void *end;
// size_t elemSize; // size_t elemSize;
@ -106,15 +108,13 @@ uint8_t *fstSliceData(FstSlice *s, int32_t *sz);
//} FstStack; //} FstStack;
// //
// //
//FstStack* fstStackCreate(size_t elemSize, stackFreeElem); // FstStack* fstStackCreate(size_t elemSize, stackFreeElem);
//void *fstStackPush(FstStack *s, void *elem); // void *fstStackPush(FstStack *s, void *elem);
//void *fstStackTop(FstStack *s); // void *fstStackTop(FstStack *s);
//size_t fstStackLen(FstStack *s); // size_t fstStackLen(FstStack *s);
//void fstStackDestory(FstStack *); // void fstStackDestory(FstStack *);
// //
#ifdef __cplusplus #ifdef __cplusplus
} }
#endif #endif

35
source/libs/index/inc/index_tfile.h
View File

@ -17,10 +17,10 @@
#include "index.h" #include "index.h"
#include "indexInt.h" #include "indexInt.h"
#include "tlockfree.h"
#include "index_tfile.h"
#include "index_fst_counting_writer.h"
#include "index_fst.h" #include "index_fst.h"
#include "index_fst_counting_writer.h"
#include "index_tfile.h"
#include "tlockfree.h"
#ifdef __cplusplus #ifdef __cplusplus
extern "C" { extern "C" {
@ -48,7 +48,6 @@ typedef struct TFileCacheKey {
int32_t nColName; int32_t nColName;
} TFileCacheKey; } TFileCacheKey;
// table cache // table cache
// refactor to LRU cache later // refactor to LRU cache later
typedef struct TFileCache { typedef struct TFileCache {
@ -57,49 +56,47 @@ typedef struct TFileCache {
// add more param // add more param
} TFileCache; } TFileCache;
typedef struct TFileWriter { typedef struct TFileWriter {
FstBuilder *fb; FstBuilder *fb;
WriterCtx *ctx; WriterCtx * ctx;
} TFileWriter; } TFileWriter;
typedef struct TFileReader { typedef struct TFileReader {
T_REF_DECLARE() T_REF_DECLARE()
Fst *fst; Fst * fst;
WriterCtx *ctx; WriterCtx * ctx;
TFileReadHeader header;
} TFileReader; } TFileReader;
typedef struct IndexTFile { typedef struct IndexTFile {
char *path; char * path;
TFileCache *cache; TFileCache * cache;
TFileWriter *tw; TFileWriter *tw;
} IndexTFile; } IndexTFile;
typedef struct TFileWriterOpt { typedef struct TFileWriterOpt {
uint64_t suid; uint64_t suid;
int8_t colType; int8_t colType;
char *colName; char * colName;
int32_t nColName; int32_t nColName;
int32_t version; int32_t version;
} TFileWriterOpt; } TFileWriterOpt;
typedef struct TFileReaderOpt { typedef struct TFileReaderOpt {
uint64_t suid; uint64_t suid;
char *colName; char * colName;
int32_t nColName; int32_t nColName;
} TFileReaderOpt; } TFileReaderOpt;
// tfile cache, manage tindex reader // tfile cache, manage tindex reader
TFileCache *tfileCacheCreate(const char *path); TFileCache * tfileCacheCreate(const char *path);
void tfileCacheDestroy(TFileCache *tcache); void tfileCacheDestroy(TFileCache *tcache);
TFileReader* tfileCacheGet(TFileCache *tcache, TFileCacheKey *key); TFileReader *tfileCacheGet(TFileCache *tcache, TFileCacheKey *key);
void tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader); void tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader);
TFileReader* tfileReaderCreate(); TFileReader *tfileReaderCreate();
void TFileReaderDestroy(TFileReader *reader); void TFileReaderDestroy(TFileReader *reader);
TFileWriter *tfileWriterCreate(const char *suid, const char *colName); TFileWriter *tfileWriterCreate(const char *suid, const char *colName);
void tfileWriterDestroy(TFileWriter *tw); void tfileWriterDestroy(TFileWriter *tw);
@ -108,13 +105,9 @@ IndexTFile *indexTFileCreate(const char *path);
int indexTFilePut(void *tfile, SIndexTerm *term, uint64_t uid); int indexTFilePut(void *tfile, SIndexTerm *term, uint64_t uid);
int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result); int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result);
#ifdef __cplusplus #ifdef __cplusplus
} }
#endif #endif
#endif #endif

5
source/libs/index/inc/index_util.h
View File

@ -34,7 +34,7 @@ extern "C" {
#define SERIALIZE_VAR_TO_BUF(buf, var, type) \ #define SERIALIZE_VAR_TO_BUF(buf, var, type) \
do { \ do { \
type c = var; \ type c = var; \
assert(sizeof(var) == sizeof(type));\ assert(sizeof(var) == sizeof(type)); \
memcpy((void *)buf, (void *)&c, sizeof(c)); \ memcpy((void *)buf, (void *)&c, sizeof(c)); \
buf += sizeof(c); \ buf += sizeof(c); \
} while (0) } while (0)
@ -42,10 +42,9 @@ extern "C" {
#define SERIALIZE_STR_VAR_TO_BUF(buf, var, len) \ #define SERIALIZE_STR_VAR_TO_BUF(buf, var, len) \
do { \ do { \
memcpy((void *)buf, (void *)var, len); \ memcpy((void *)buf, (void *)var, len); \
buf += len;\ buf += len; \
} while (0) } while (0)
#ifdef __cplusplus #ifdef __cplusplus
} }
#endif #endif

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

@ -26,8 +26,11 @@
static int uidCompare(const void *a, const void *b) { static int uidCompare(const void *a, const void *b) {
uint64_t u1 = *(uint64_t *)a; uint64_t u1 = *(uint64_t *)a;
uint64_t u2 = *(uint64_t *)b; uint64_t u2 = *(uint64_t *)b;
if (u1 == u2) { return 0; } if (u1 == u2) {
else { return u1 < u2 ? -1 : 1; } return 0;
} else {
return u1 < u2 ? -1 : 1;
}
} }
typedef struct SIdxColInfo { typedef struct SIdxColInfo {
int colId; // generated by index internal int colId; // generated by index internal
@ -37,7 +40,6 @@ typedef struct SIdxColInfo {
static pthread_once_t isInit = PTHREAD_ONCE_INIT; static pthread_once_t isInit = PTHREAD_ONCE_INIT;
static void indexInit(); static void indexInit();
static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *term, SArray **result); static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *term, SArray **result);
static int indexMergeCacheIntoTindex(SIndex *sIdx); static int indexMergeCacheIntoTindex(SIndex *sIdx);
@ -47,14 +49,16 @@ static int indexMergeFinalResults(SArray *interResults, EIndexOperatorType oType
int indexOpen(SIndexOpts *opts, const char *path, SIndex **index) { int indexOpen(SIndexOpts *opts, const char *path, SIndex **index) {
pthread_once(&isInit, indexInit); pthread_once(&isInit, indexInit);
SIndex *sIdx = calloc(1, sizeof(SIndex)); SIndex *sIdx = calloc(1, sizeof(SIndex));
if (sIdx == NULL) { return -1; } if (sIdx == NULL) {
return -1;
}
#ifdef USE_LUCENE #ifdef USE_LUCENE
index_t *index = index_open(path); index_t *index = index_open(path);
sIdx->index = index; sIdx->index = index;
#endif #endif
sIdx->cache = (void*)indexCacheCreate(); sIdx->cache = (void *)indexCacheCreate();
sIdx->tindex = NULL; sIdx->tindex = NULL;
sIdx->colObj = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK); sIdx->colObj = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
sIdx->colId = 1; sIdx->colId = 1;
@ -80,8 +84,7 @@ void indexClose(SIndex *sIdx) {
return; return;
} }
int indexPut(SIndex *index, SIndexMultiTerm * fVals, uint64_t uid) { int indexPut(SIndex *index, SIndexMultiTerm *fVals, uint64_t uid) {
#ifdef USE_LUCENE #ifdef USE_LUCENE
index_document_t *doc = index_document_create(); index_document_t *doc = index_document_create();
@ -100,10 +103,10 @@ int indexPut(SIndex *index, SIndexMultiTerm * fVals, uint64_t uid) {
#ifdef USE_INVERTED_INDEX #ifdef USE_INVERTED_INDEX
//TODO(yihao): reduce the lock range // TODO(yihao): reduce the lock range
pthread_mutex_lock(&index->mtx); pthread_mutex_lock(&index->mtx);
for (int i = 0; i < taosArrayGetSize(fVals); i++) { for (int i = 0; i < taosArrayGetSize(fVals); i++) {
SIndexTerm *p = taosArrayGetP(fVals, i); SIndexTerm * p = taosArrayGetP(fVals, i);
SIdxColInfo *fi = taosHashGet(index->colObj, p->colName, p->nColName); SIdxColInfo *fi = taosHashGet(index->colObj, p->colName, p->nColName);
if (fi == NULL) { if (fi == NULL) {
SIdxColInfo tfi = {.colId = index->colId}; SIdxColInfo tfi = {.colId = index->colId};
@ -111,13 +114,13 @@ int indexPut(SIndex *index, SIndexMultiTerm * fVals, uint64_t uid) {
index->colId++; index->colId++;
taosHashPut(index->colObj, p->colName, p->nColName, &tfi, sizeof(tfi)); taosHashPut(index->colObj, p->colName, p->nColName, &tfi, sizeof(tfi));
} else { } else {
//TODO, del // TODO, del
} }
} }
pthread_mutex_unlock(&index->mtx); pthread_mutex_unlock(&index->mtx);
for (int i = 0; i < taosArrayGetSize(fVals); i++) { for (int i = 0; i < taosArrayGetSize(fVals); i++) {
SIndexTerm *p = taosArrayGetP(fVals, i); SIndexTerm * p = taosArrayGetP(fVals, i);
SIdxColInfo *fi = taosHashGet(index->colObj, p->colName, p->nColName); SIdxColInfo *fi = taosHashGet(index->colObj, p->colName, p->nColName);
assert(fi != NULL); assert(fi != NULL);
int32_t colId = fi->colId; int32_t colId = fi->colId;
@ -138,11 +141,11 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
int nQuery = taosArrayGetSize(multiQuerys->query); int nQuery = taosArrayGetSize(multiQuerys->query);
char **fields = malloc(sizeof(char *) * nQuery); char **fields = malloc(sizeof(char *) * nQuery);
char **keys = malloc(sizeof(char *) * nQuery); char **keys = malloc(sizeof(char *) * nQuery);
int *types = malloc(sizeof(int) * nQuery); int * types = malloc(sizeof(int) * nQuery);
for (int i = 0; i < nQuery; i++) { for (int i = 0; i < nQuery; i++) {
SIndexTermQuery *p = taosArrayGet(multiQuerys->query, i); SIndexTermQuery *p = taosArrayGet(multiQuerys->query, i);
SIndexTerm *term = p->field_value; SIndexTerm * term = p->field_value;
fields[i] = calloc(1, term->nKey + 1); fields[i] = calloc(1, term->nKey + 1);
keys[i] = calloc(1, term->nVal + 1); keys[i] = calloc(1, term->nVal + 1);
@ -152,7 +155,7 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
types[i] = (int)(p->type); types[i] = (int)(p->type);
} }
int *tResult = NULL; int *tResult = NULL;
int tsz= 0; 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++) { for (int i = 0; i < tsz; i++) {
@ -175,7 +178,7 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
int nQuery = taosArrayGetSize(multiQuerys->query); int nQuery = taosArrayGetSize(multiQuerys->query);
for (size_t i = 0; i < nQuery; i++) { for (size_t i = 0; i < nQuery; i++) {
SIndexTermQuery *qTerm = taosArrayGet(multiQuerys->query, i); SIndexTermQuery *qTerm = taosArrayGet(multiQuerys->query, i);
SArray *tResult = NULL; SArray * tResult = NULL;
indexTermSearch(index, qTerm, &tResult); indexTermSearch(index, qTerm, &tResult);
taosArrayPush(interResults, (void *)&tResult); taosArrayPush(interResults, (void *)&tResult);
} }
@ -186,38 +189,36 @@ int indexSearch(SIndex *index, SIndexMultiTermQuery *multiQuerys, SArray *result
return 1; return 1;
} }
int indexDelete(SIndex *index, SIndexMultiTermQuery *query) { int indexDelete(SIndex *index, SIndexMultiTermQuery *query) {
#ifdef USE_INVERTED_INDEX #ifdef USE_INVERTED_INDEX
#endif #endif
return 1; return 1;
} }
int indexRebuild(SIndex *index, SIndexOpts *opts) { int indexRebuild(SIndex *index, SIndexOpts *opts){
#ifdef USE_INVERTED_INDEX #ifdef USE_INVERTED_INDEX
#endif #endif
} }
SIndexOpts *indexOptsCreate() { SIndexOpts *indexOptsCreate() {
#ifdef USE_LUCENE #ifdef USE_LUCENE
#endif #endif
return NULL; return NULL;
} }
void indexOptsDestroy(SIndexOpts *opts) { void indexOptsDestroy(SIndexOpts *opts){
#ifdef USE_LUCENE #ifdef USE_LUCENE
#endif #endif
} } /*
/*
* @param: oper * @param: oper
* *
*/ */
SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType opera) { SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType opera) {
SIndexMultiTermQuery *p = (SIndexMultiTermQuery *)malloc(sizeof(SIndexMultiTermQuery)); SIndexMultiTermQuery *p = (SIndexMultiTermQuery *)malloc(sizeof(SIndexMultiTermQuery));
if (p == NULL) { return NULL; } if (p == NULL) {
return NULL;
}
p->opera = opera; p->opera = opera;
p->query = taosArrayInit(4, sizeof(SIndexTermQuery)); p->query = taosArrayInit(4, sizeof(SIndexTermQuery));
return p; return p;
@ -230,19 +231,21 @@ void indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery) {
taosArrayDestroy(pQuery->query); taosArrayDestroy(pQuery->query);
free(pQuery); free(pQuery);
}; };
int indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType qType){ int indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType qType) {
SIndexTermQuery q = {.qType = qType, .term = term}; SIndexTermQuery q = {.qType = qType, .term = term};
taosArrayPush(pQuery->query, &q); taosArrayPush(pQuery->query, &q);
return 0; return 0;
} }
SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn oper, uint8_t colType, const char *colName,
SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn oper, uint8_t colType, const char *colName, int32_t nColName, const char *colVal, int32_t nColVal) { int32_t nColName, const char *colVal, int32_t nColVal) {
SIndexTerm *t = (SIndexTerm *)calloc(1, (sizeof(SIndexTerm))); SIndexTerm *t = (SIndexTerm *)calloc(1, (sizeof(SIndexTerm)));
if (t == NULL) { return NULL; } if (t == NULL) {
return NULL;
}
t->suid = suid; t->suid = suid;
t->operType= oper; t->operType = oper;
t->colType = colType; t->colType = colType;
t->colName = (char *)calloc(1, nColName + 1); t->colName = (char *)calloc(1, nColName + 1);
@ -260,9 +263,7 @@ void indexTermDestroy(SIndexTerm *p) {
free(p); free(p);
} }
SIndexMultiTerm *indexMultiTermCreate() { SIndexMultiTerm *indexMultiTermCreate() { return taosArrayInit(4, sizeof(SIndexTerm *)); }
return taosArrayInit(4, sizeof(SIndexTerm *));
}
int indexMultiTermAdd(SIndexMultiTerm *terms, SIndexTerm *term) { int indexMultiTermAdd(SIndexMultiTerm *terms, SIndexTerm *term) {
taosArrayPush(terms, &term); taosArrayPush(terms, &term);
@ -277,7 +278,7 @@ void indexMultiTermDestroy(SIndexMultiTerm *terms) {
} }
void indexInit() { void indexInit() {
//do nothing // do nothing
} }
static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *query, SArray **result) { static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *query, SArray **result) {
int32_t version = -1; int32_t version = -1;
@ -299,7 +300,7 @@ static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *query, SArray **result
pthread_mutex_unlock(&sIdx->mtx); pthread_mutex_unlock(&sIdx->mtx);
*result = taosArrayInit(4, sizeof(uint64_t)); *result = taosArrayInit(4, sizeof(uint64_t));
//TODO: iterator mem and tidex // TODO: iterator mem and tidex
STermValueType s; STermValueType s;
if (0 == indexCacheSearch(sIdx->cache, query, colId, version, *result, &s)) { if (0 == indexCacheSearch(sIdx->cache, query, colId, version, *result, &s)) {
if (s == kTypeDeletion) { if (s == kTypeDeletion) {
@ -319,7 +320,9 @@ static int indexTermSearch(SIndex *sIdx, SIndexTermQuery *query, SArray **result
return 0; return 0;
} }
static void indexInterResultsDestroy(SArray *results) { static void indexInterResultsDestroy(SArray *results) {
if (results == NULL) { return; } if (results == NULL) {
return;
}
size_t sz = taosArrayGetSize(results); size_t sz = taosArrayGetSize(results);
for (size_t i = 0; i < sz; i++) { for (size_t i = 0; i < sz; i++) {
@ -327,10 +330,9 @@ static void indexInterResultsDestroy(SArray *results) {
taosArrayDestroy(p); taosArrayDestroy(p);
} }
taosArrayDestroy(results); 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 // refactor, merge interResults into fResults by oType
SArray *first = taosArrayGetP(interResults, 0); SArray *first = taosArrayGetP(interResults, 0);
taosArraySort(first, uidCompare); taosArraySort(first, uidCompare);
taosArrayRemoveDuplicate(first, uidCompare, NULL); taosArrayRemoveDuplicate(first, uidCompare, NULL);

64
source/libs/index/src/index_cache.c
View File

@ -14,17 +14,17 @@
*/ */
#include "index_cache.h" #include "index_cache.h"
#include "tcompare.h"
#include "index_util.h" #include "index_util.h"
#include "tcompare.h"
#define MAX_INDEX_KEY_LEN 256// test only, change later #define MAX_INDEX_KEY_LEN 256 // test only, change later
// ref index_cache.h:22 // 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)) #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))
static char* getIndexKey(const void *pData) { static char * getIndexKey(const void *pData) { return NULL; }
return NULL;
}
static int32_t compareKey(const void *l, const void *r) { static int32_t compareKey(const void *l, const void *r) {
char *lp = (char *)l; char *lp = (char *)l;
char *rp = (char *)r; char *rp = (char *)r;
@ -41,7 +41,7 @@ static int32_t compareKey(const void *l, const void *r) {
memcpy(&lf, lp, sizeof(lf)); memcpy(&lf, lp, sizeof(lf));
memcpy(&rf, rp, sizeof(rf)); memcpy(&rf, rp, sizeof(rf));
if (lf != rf) { if (lf != rf) {
return lf < rf ? -1: 1; return lf < rf ? -1 : 1;
} }
lp += sizeof(lf); lp += sizeof(lf);
rp += sizeof(rf); rp += sizeof(rf);
@ -64,11 +64,17 @@ static int32_t compareKey(const void *l, const void *r) {
// compare value // compare value
int32_t i, j; int32_t i, j;
for (i = 0, j = 0; i < lfl && j < rfl; i++, j++) { for (i = 0, j = 0; i < lfl && j < rfl; i++, j++) {
if (lp[i] == rp[j]) { continue; } if (lp[i] == rp[j]) {
else { return lp[i] < rp[j] ? -1 : 1;} continue;
} else {
return lp[i] < rp[j] ? -1 : 1;
}
}
if (i < lfl) {
return 1;
} else if (j < rfl) {
return -1;
} }
if (i < lfl) { return 1;}
else if (j < rfl) { return -1; }
lp += lfl; lp += lfl;
rp += rfl; rp += rfl;
@ -91,32 +97,35 @@ static int32_t compareKey(const void *l, const void *r) {
// not care item type // not care item type
return 0; return 0;
} }
IndexCache *indexCacheCreate() { IndexCache *indexCacheCreate() {
IndexCache *cache = calloc(1, sizeof(IndexCache)); 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); cache->skiplist = tSkipListCreate(
MAX_SKIP_LIST_LEVEL, TSDB_DATA_TYPE_BINARY, MAX_INDEX_KEY_LEN, compareKey, SL_ALLOW_DUP_KEY, getIndexKey);
return cache; return cache;
} }
void indexCacheDestroy(void *cache) { void indexCacheDestroy(void *cache) {
IndexCache *pCache = cache; IndexCache *pCache = cache;
if (pCache == NULL) { return; } if (pCache == NULL) {
return;
}
tSkipListDestroy(pCache->skiplist); tSkipListDestroy(pCache->skiplist);
free(pCache); free(pCache);
} }
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) {
if (cache == NULL) { return -1;} if (cache == NULL) {
return -1;
}
IndexCache *pCache = cache; IndexCache *pCache = cache;
// encode data // encode data
int32_t total = CACHE_KEY_LEN(term); int32_t total = CACHE_KEY_LEN(term);
char *buf = calloc(1, total); char * buf = calloc(1, total);
char *p = buf; char * p = buf;
SERIALIZE_VAR_TO_BUF(p, total,int32_t); SERIALIZE_VAR_TO_BUF(p, total, int32_t);
SERIALIZE_VAR_TO_BUF(p, colId, int16_t); SERIALIZE_VAR_TO_BUF(p, colId, int16_t);
SERIALIZE_MEM_TO_BUF(p, term, colType); SERIALIZE_MEM_TO_BUF(p, term, colType);
@ -131,31 +140,28 @@ int indexCachePut(void *cache, SIndexTerm *term, int16_t colId, int32_t version,
tSkipListPut(pCache->skiplist, (void *)buf); tSkipListPut(pCache->skiplist, (void *)buf);
return 0; return 0;
// encode end // encode end
} }
int indexCacheDel(void *cache, int32_t fieldId, const char *fieldValue, int32_t fvlen, uint64_t uid, int8_t operType) { int indexCacheDel(void *cache, int32_t fieldId, const char *fieldValue, int32_t fvlen, uint64_t uid, int8_t operType) {
IndexCache *pCache = cache; IndexCache *pCache = cache;
return 0; return 0;
} }
int indexCacheSearch(void *cache, SIndexTermQuery *query, int16_t colId, int32_t version, SArray *result, STermValueType *s) { int indexCacheSearch(
if (cache == NULL) { return -1; } void *cache, SIndexTermQuery *query, int16_t colId, int32_t version, SArray *result, STermValueType *s) {
IndexCache *pCache = cache; if (cache == NULL) {
SIndexTerm *term = query->term; return -1;
}
IndexCache * pCache = cache;
SIndexTerm * term = query->term;
EIndexQueryType qtype = query->qType; EIndexQueryType qtype = query->qType;
int32_t keyLen = CACHE_KEY_LEN(term); int32_t keyLen = CACHE_KEY_LEN(term);
char *buf = calloc(1, keyLen); char *buf = calloc(1, keyLen);
if (qtype == QUERY_TERM) { if (qtype == QUERY_TERM) {
} else if (qtype == QUERY_PREFIX) { } else if (qtype == QUERY_PREFIX) {
} else if (qtype == QUERY_SUFFIX) { } else if (qtype == QUERY_SUFFIX) {
} else if (qtype == QUERY_REGEX) { } else if (qtype == QUERY_REGEX) {
} }
return 0; return 0;
} }

422
source/libs/index/src/index_fst.c
View File

@ -14,11 +14,10 @@
*/ */
#include "index_fst.h" #include "index_fst.h"
#include "tcoding.h"
#include "tchecksum.h"
#include "indexInt.h" #include "indexInt.h"
#include "index_fst_automation.h" #include "index_fst_automation.h"
#include "tchecksum.h"
#include "tcoding.h"
static void fstPackDeltaIn(FstCountingWriter *wrt, CompiledAddr nodeAddr, CompiledAddr transAddr, uint8_t nBytes) { static void fstPackDeltaIn(FstCountingWriter *wrt, CompiledAddr nodeAddr, CompiledAddr transAddr, uint8_t nBytes) {
CompiledAddr deltaAddr = (transAddr == EMPTY_ADDRESS) ? EMPTY_ADDRESS : nodeAddr - transAddr; CompiledAddr deltaAddr = (transAddr == EMPTY_ADDRESS) ? EMPTY_ADDRESS : nodeAddr - transAddr;
@ -32,20 +31,24 @@ static uint8_t fstPackDetla(FstCountingWriter *wrt, CompiledAddr nodeAddr, Compi
FstUnFinishedNodes *fstUnFinishedNodesCreate() { FstUnFinishedNodes *fstUnFinishedNodesCreate() {
FstUnFinishedNodes *nodes = malloc(sizeof(FstUnFinishedNodes)); FstUnFinishedNodes *nodes = malloc(sizeof(FstUnFinishedNodes));
if (nodes == NULL) { return NULL; } if (nodes == NULL) {
return NULL;
}
nodes->stack = (SArray *)taosArrayInit(64, sizeof(FstBuilderNodeUnfinished)); nodes->stack = (SArray *)taosArrayInit(64, sizeof(FstBuilderNodeUnfinished));
fstUnFinishedNodesPushEmpty(nodes, false); fstUnFinishedNodesPushEmpty(nodes, false);
return nodes; return nodes;
} }
void unFinishedNodeDestroyElem(void* elem) { void unFinishedNodeDestroyElem(void *elem) {
FstBuilderNodeUnfinished *b = (FstBuilderNodeUnfinished*)elem; FstBuilderNodeUnfinished *b = (FstBuilderNodeUnfinished *)elem;
fstBuilderNodeDestroy(b->node); fstBuilderNodeDestroy(b->node);
free(b->last); free(b->last);
b->last = NULL; b->last = NULL;
} }
void fstUnFinishedNodesDestroy(FstUnFinishedNodes *nodes) { void fstUnFinishedNodesDestroy(FstUnFinishedNodes *nodes) {
if (nodes == NULL) { return; } if (nodes == NULL) {
return;
}
taosArrayDestroyEx(nodes->stack, unFinishedNodeDestroyElem); taosArrayDestroyEx(nodes->stack, unFinishedNodeDestroyElem);
free(nodes); free(nodes);
@ -59,7 +62,6 @@ void fstUnFinishedNodesPushEmpty(FstUnFinishedNodes *nodes, bool isFinal) {
FstBuilderNodeUnfinished un = {.node = node, .last = NULL}; FstBuilderNodeUnfinished un = {.node = node, .last = NULL};
taosArrayPush(nodes->stack, &un); taosArrayPush(nodes->stack, &un);
} }
FstBuilderNode *fstUnFinishedNodesPopRoot(FstUnFinishedNodes *nodes) { FstBuilderNode *fstUnFinishedNodesPopRoot(FstUnFinishedNodes *nodes) {
assert(taosArrayGetSize(nodes->stack) == 1); assert(taosArrayGetSize(nodes->stack) == 1);
@ -72,8 +74,8 @@ FstBuilderNode *fstUnFinishedNodesPopRoot(FstUnFinishedNodes *nodes) {
FstBuilderNode *fstUnFinishedNodesPopFreeze(FstUnFinishedNodes *nodes, CompiledAddr addr) { FstBuilderNode *fstUnFinishedNodesPopFreeze(FstUnFinishedNodes *nodes, CompiledAddr addr) {
FstBuilderNodeUnfinished *un = taosArrayPop(nodes->stack); FstBuilderNodeUnfinished *un = taosArrayPop(nodes->stack);
fstBuilderNodeUnfinishedLastCompiled(un, addr); fstBuilderNodeUnfinishedLastCompiled(un, addr);
//free(un->last); // TODO add func FstLastTransitionFree() // free(un->last); // TODO add func FstLastTransitionFree()
//un->last = NULL; // un->last = NULL;
return un->node; return un->node;
} }
@ -81,13 +83,12 @@ FstBuilderNode *fstUnFinishedNodesPopEmpty(FstUnFinishedNodes *nodes) {
FstBuilderNodeUnfinished *un = taosArrayPop(nodes->stack); FstBuilderNodeUnfinished *un = taosArrayPop(nodes->stack);
assert(un->last == NULL); assert(un->last == NULL);
return un->node; return un->node;
} }
void fstUnFinishedNodesSetRootOutput(FstUnFinishedNodes *nodes, Output out) { void fstUnFinishedNodesSetRootOutput(FstUnFinishedNodes *nodes, Output out) {
FstBuilderNodeUnfinished *un = taosArrayGet(nodes->stack, 0); FstBuilderNodeUnfinished *un = taosArrayGet(nodes->stack, 0);
un->node->isFinal = true; un->node->isFinal = true;
un->node->finalOutput = out; un->node->finalOutput = out;
//un->node->trans = NULL; // un->node->trans = NULL;
} }
void fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes *nodes, CompiledAddr addr) { void fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes *nodes, CompiledAddr addr) {
size_t sz = taosArrayGetSize(nodes->stack) - 1; size_t sz = taosArrayGetSize(nodes->stack) - 1;
@ -103,9 +104,9 @@ void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *nodes, FstSlice bs, Output
FstBuilderNodeUnfinished *un = taosArrayGet(nodes->stack, sz); FstBuilderNodeUnfinished *un = taosArrayGet(nodes->stack, sz);
assert(un->last == NULL); assert(un->last == NULL);
//FstLastTransition *trn = malloc(sizeof(FstLastTransition)); // FstLastTransition *trn = malloc(sizeof(FstLastTransition));
//trn->inp = s->data[s->start]; // trn->inp = s->data[s->start];
//trn->out = out; // trn->out = out;
int32_t len = 0; int32_t len = 0;
uint8_t *data = fstSliceData(s, &len); uint8_t *data = fstSliceData(s, &len);
un->last = fstLastTransitionCreate(data[0], out); un->last = fstLastTransitionCreate(data[0], out);
@ -116,9 +117,9 @@ void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *nodes, FstSlice bs, Output
n->finalOutput = 0; n->finalOutput = 0;
n->trans = taosArrayInit(16, sizeof(FstTransition)); n->trans = taosArrayInit(16, sizeof(FstTransition));
//FstLastTransition *trn = malloc(sizeof(FstLastTransition)); // FstLastTransition *trn = malloc(sizeof(FstLastTransition));
//trn->inp = s->data[i]; // trn->inp = s->data[i];
//trn->out = out; // trn->out = out;
FstLastTransition *trn = fstLastTransitionCreate(data[i], 0); FstLastTransition *trn = fstLastTransitionCreate(data[i], 0);
FstBuilderNodeUnfinished un = {.node = n, .last = trn}; FstBuilderNodeUnfinished un = {.node = n, .last = trn};
@ -127,7 +128,6 @@ void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *nodes, FstSlice bs, Output
fstUnFinishedNodesPushEmpty(nodes, true); fstUnFinishedNodesPushEmpty(nodes, true);
} }
uint64_t fstUnFinishedNodesFindCommPrefix(FstUnFinishedNodes *node, FstSlice bs) { uint64_t fstUnFinishedNodesFindCommPrefix(FstUnFinishedNodes *node, FstSlice bs) {
FstSlice *s = &bs; FstSlice *s = &bs;
@ -157,7 +157,7 @@ uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes *node,
FstLastTransition *t = un->last; FstLastTransition *t = un->last;
uint64_t addPrefix = 0; uint64_t addPrefix = 0;
uint8_t *data = fstSliceData(s, NULL); uint8_t * data = fstSliceData(s, NULL);
if (t && t->inp == data[i]) { if (t && t->inp == data[i]) {
uint64_t commPrefix = MIN(t->out, *out); uint64_t commPrefix = MIN(t->out, *out);
uint64_t tAddPrefix = t->out - commPrefix; uint64_t tAddPrefix = t->out - commPrefix;
@ -177,8 +177,7 @@ uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes *node,
return i; return i;
} }
FstState fstStateCreateFrom(FstSlice *slice, CompiledAddr addr) {
FstState fstStateCreateFrom(FstSlice* slice, CompiledAddr addr) {
FstState fs = {.state = EmptyFinal, .val = 0}; FstState fs = {.state = EmptyFinal, .val = 0};
if (addr == EMPTY_ADDRESS) { if (addr == EMPTY_ADDRESS) {
return fs; return fs;
@ -198,20 +197,16 @@ FstState fstStateCreateFrom(FstSlice* slice, CompiledAddr addr) {
return fs; return fs;
} }
static FstState fstStateDict[] = { static FstState fstStateDict[] = {{.state = OneTransNext, .val = 0b11000000}, {.state = OneTrans, .val = 0b10000000},
{.state = OneTransNext, .val = 0b11000000}, {.state = AnyTrans, .val = 0b00000000}, {.state = EmptyFinal, .val = 0b00000000}};
{.state = OneTrans, .val = 0b10000000},
{.state = AnyTrans, .val = 0b00000000},
{.state = EmptyFinal, .val = 0b00000000}
};
// debug // debug
static const char *fstStateStr[] = {"ONE_TRANS_NEXT", "ONE_TRANS", "ANY_TRANS", "EMPTY_FINAL"}; static const char *fstStateStr[] = {"ONE_TRANS_NEXT", "ONE_TRANS", "ANY_TRANS", "EMPTY_FINAL"};
FstState fstStateCreate(State state){ FstState fstStateCreate(State state) {
uint8_t idx = (uint8_t)state; uint8_t idx = (uint8_t)state;
return fstStateDict[idx]; return fstStateDict[idx];
} }
//compile // compile
void fstStateCompileForOneTransNext(FstCountingWriter *w, CompiledAddr addr, uint8_t inp) { void fstStateCompileForOneTransNext(FstCountingWriter *w, CompiledAddr addr, uint8_t inp) {
FstState s = fstStateCreate(OneTransNext); FstState s = fstStateCreate(OneTransNext);
fstStateSetCommInput(&s, inp); fstStateSetCommInput(&s, inp);
@ -226,7 +221,7 @@ void fstStateCompileForOneTransNext(FstCountingWriter *w, CompiledAddr addr, uin
// w->write_all(&[s.val]) // w->write_all(&[s.val])
return; return;
} }
void fstStateCompileForOneTrans(FstCountingWriter *w, CompiledAddr addr, FstTransition* trn) { void fstStateCompileForOneTrans(FstCountingWriter *w, CompiledAddr addr, FstTransition *trn) {
Output out = trn->out; Output out = trn->out;
uint8_t outPackSize = (out == 0 ? 0 : fstCountingWriterPackUint(w, out)); uint8_t outPackSize = (out == 0 ? 0 : fstCountingWriterPackUint(w, out));
uint8_t transPackSize = fstPackDetla(w, addr, trn->addr); uint8_t transPackSize = fstPackDetla(w, addr, trn->addr);
@ -245,18 +240,17 @@ void fstStateCompileForOneTrans(FstCountingWriter *w, CompiledAddr addr, FstTran
fstCountingWriterWrite(w, (char *)&trn->inp, sizeof(trn->inp)); fstCountingWriterWrite(w, (char *)&trn->inp, sizeof(trn->inp));
} }
fstCountingWriterWrite(w, (char *)(&(st.val)), sizeof(st.val)); fstCountingWriterWrite(w, (char *)(&(st.val)), sizeof(st.val));
return ; return;
} }
void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuilderNode *node) { void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuilderNode *node) {
size_t sz = taosArrayGetSize(node->trans); size_t sz = taosArrayGetSize(node->trans);
assert(sz <= 256); assert(sz <= 256);
uint8_t tSize = 0; uint8_t tSize = 0;
uint8_t oSize = packSize(node->finalOutput) ; uint8_t oSize = packSize(node->finalOutput);
// finalOutput.is_zero() // finalOutput.is_zero()
bool anyOuts = (node->finalOutput != 0) ; bool anyOuts = (node->finalOutput != 0);
for (size_t i = 0; i < sz; i++) { for (size_t i = 0; i < sz; i++) {
FstTransition *t = taosArrayGet(node->trans, i); FstTransition *t = taosArrayGet(node->trans, i);
tSize = MAX(tSize, packDeltaSize(addr, t->addr)); tSize = MAX(tSize, packDeltaSize(addr, t->addr));
@ -265,8 +259,11 @@ void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuil
} }
PackSizes packSizes = 0; PackSizes packSizes = 0;
if (anyOuts) { FST_SET_OUTPUT_PACK_SIZE(packSizes, oSize); } if (anyOuts) {
else { FST_SET_OUTPUT_PACK_SIZE(packSizes, 0); } FST_SET_OUTPUT_PACK_SIZE(packSizes, oSize);
} else {
FST_SET_OUTPUT_PACK_SIZE(packSizes, 0);
}
FST_SET_TRANSITION_PACK_SIZE(packSizes, tSize); FST_SET_TRANSITION_PACK_SIZE(packSizes, tSize);
@ -290,7 +287,7 @@ void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuil
for (int32_t i = sz - 1; i >= 0; i--) { for (int32_t i = sz - 1; i >= 0; i--) {
FstTransition *t = taosArrayGet(node->trans, i); FstTransition *t = taosArrayGet(node->trans, i);
fstCountingWriterWrite(w, (char *)&t->inp, 1); fstCountingWriterWrite(w, (char *)&t->inp, 1);
//fstPackDeltaIn(w, addr, t->addr, tSize); // fstPackDeltaIn(w, addr, t->addr, tSize);
} }
if (sz > TRANS_INDEX_THRESHOLD) { if (sz > TRANS_INDEX_THRESHOLD) {
// A value of 255 indicates that no transition exists for the byte // A value of 255 indicates that no transition exists for the byte
@ -299,13 +296,13 @@ void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuil
// this node indicates an absent transition. // this node indicates an absent transition.
uint8_t *index = (uint8_t *)malloc(sizeof(uint8_t) * 256); uint8_t *index = (uint8_t *)malloc(sizeof(uint8_t) * 256);
memset(index, 255, sizeof(uint8_t) * 256); memset(index, 255, sizeof(uint8_t) * 256);
///for (uint8_t i = 0; i < 256; i++) { /// for (uint8_t i = 0; i < 256; i++) {
// index[i] = 255; // index[i] = 255;
///} ///}
for (size_t i = 0; i < sz; i++) { for (size_t i = 0; i < sz; i++) {
FstTransition *t = taosArrayGet(node->trans, i); FstTransition *t = taosArrayGet(node->trans, i);
index[t->inp] = i; index[t->inp] = i;
//fstPackDeltaIn(w, addr, t->addr, tSize); // fstPackDeltaIn(w, addr, t->addr, tSize);
} }
fstCountingWriterWrite(w, (char *)index, 256); fstCountingWriterWrite(w, (char *)index, 256);
free(index); free(index);
@ -318,15 +315,18 @@ void fstStateCompileForAnyTrans(FstCountingWriter *w, CompiledAddr addr, FstBuil
// the # of transitions can never be 1 here, since 1 is always // the # of transitions can never be 1 here, since 1 is always
// encoded in the state byte. // encoded in the state byte.
uint8_t v = 1; uint8_t v = 1;
if (sz == 256) { fstCountingWriterWrite(w, (char *)&v, 1); } if (sz == 256) {
else { fstCountingWriterWrite(w, (char *)&sz, 1); } fstCountingWriterWrite(w, (char *)&v, 1);
} else {
fstCountingWriterWrite(w, (char *)&sz, 1);
}
} }
fstCountingWriterWrite(w, (char *)(&(st.val)), 1); fstCountingWriterWrite(w, (char *)(&(st.val)), 1);
return; return;
} }
// set_comm_input // set_comm_input
void fstStateSetCommInput(FstState* s, uint8_t inp) { void fstStateSetCommInput(FstState *s, uint8_t inp) {
assert(s->state == OneTransNext || s->state == OneTrans); assert(s->state == OneTransNext || s->state == OneTrans);
uint8_t val; uint8_t val;
@ -335,49 +335,42 @@ void fstStateSetCommInput(FstState* s, uint8_t inp) {
} }
// comm_input // comm_input
uint8_t fstStateCommInput(FstState* s, bool *null) { uint8_t fstStateCommInput(FstState *s, bool *null) {
assert(s->state == OneTransNext || s->state == OneTrans); assert(s->state == OneTransNext || s->state == OneTrans);
uint8_t v = s->val & 0b00111111; uint8_t v = s->val & 0b00111111;
if (v == 0) { if (v == 0) {
*null = true; *null = true;
return v; return v;
} }
//v = 0 indicate that common_input is None // v = 0 indicate that common_input is None
return v == 0 ? 0 : COMMON_INPUT(v); return v == 0 ? 0 : COMMON_INPUT(v);
} }
// input_len // input_len
uint64_t fstStateInputLen(FstState* s) { uint64_t fstStateInputLen(FstState *s) {
assert(s->state == OneTransNext || s->state == OneTrans); assert(s->state == OneTransNext || s->state == OneTrans);
bool null = false; bool null = false;
fstStateCommInput(s, &null); fstStateCommInput(s, &null);
return null ? 1 : 0 ; return null ? 1 : 0;
} }
// end_addr // end_addr
uint64_t fstStateEndAddrForOneTransNext(FstState* s, FstSlice *data) { uint64_t fstStateEndAddrForOneTransNext(FstState *s, FstSlice *data) {
assert(s->state == OneTransNext); assert(s->state == OneTransNext);
return FST_SLICE_LEN(data) - 1 - fstStateInputLen(s); return FST_SLICE_LEN(data) - 1 - fstStateInputLen(s);
} }
uint64_t fstStateEndAddrForOneTrans(FstState *s, FstSlice *data, PackSizes sizes) { uint64_t fstStateEndAddrForOneTrans(FstState *s, FstSlice *data, PackSizes sizes) {
assert(s->state == OneTrans); assert(s->state == OneTrans);
return FST_SLICE_LEN(data) return FST_SLICE_LEN(data) - 1 - fstStateInputLen(s) - 1 // pack size
- 1 - FST_GET_TRANSITION_PACK_SIZE(sizes) - FST_GET_OUTPUT_PACK_SIZE(sizes);
- fstStateInputLen(s)
- 1 // pack size
- FST_GET_TRANSITION_PACK_SIZE(sizes)
- FST_GET_OUTPUT_PACK_SIZE(sizes);
} }
uint64_t fstStateEndAddrForAnyTrans(FstState *state, uint64_t version, FstSlice *date, PackSizes sizes, uint64_t nTrans) { uint64_t fstStateEndAddrForAnyTrans(
FstState *state, uint64_t version, FstSlice *date, PackSizes sizes, uint64_t nTrans) {
uint8_t oSizes = FST_GET_OUTPUT_PACK_SIZE(sizes); uint8_t oSizes = FST_GET_OUTPUT_PACK_SIZE(sizes);
uint8_t finalOsize = !fstStateIsFinalState(state) ? 0 : oSizes; uint8_t finalOsize = !fstStateIsFinalState(state) ? 0 : oSizes;
return FST_SLICE_LEN(date) return FST_SLICE_LEN(date) - 1 - fstStateNtransLen(state) - 1 // pack size
- 1 - fstStateTotalTransSize(state, version, sizes, nTrans) - nTrans * oSizes // output values
- fstStateNtransLen(state)
- 1 //pack size
- fstStateTotalTransSize(state, version, sizes, nTrans)
- nTrans * oSizes // output values
- finalOsize; // final output - finalOsize; // final output
} }
// input // input
@ -386,19 +379,15 @@ uint8_t fstStateInput(FstState *s, FstNode *node) {
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
bool null = false; bool null = false;
uint8_t inp = fstStateCommInput(s, &null); uint8_t inp = fstStateCommInput(s, &null);
uint8_t *data = fstSliceData(slice, NULL); uint8_t * data = fstSliceData(slice, NULL);
return null == false ? inp : data[-1]; return null == false ? inp : data[-1];
} }
uint8_t fstStateInputForAnyTrans(FstState *s, FstNode *node, uint64_t i) { uint8_t fstStateInputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
assert(s->state == AnyTrans); assert(s->state == AnyTrans);
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
uint64_t at = node->start uint64_t at = node->start - fstStateNtransLen(s) - 1 // pack size
- fstStateNtransLen(s) - fstStateTransIndexSize(s, node->version, node->nTrans) - i - 1; // the output size
- 1 // pack size
- fstStateTransIndexSize(s, node->version, node->nTrans)
- i
- 1; // the output size
uint8_t *data = fstSliceData(slice, NULL); uint8_t *data = fstSliceData(slice, NULL);
return data[at]; return data[at];
@ -413,14 +402,12 @@ CompiledAddr fstStateTransAddr(FstState *s, FstNode *node) {
} else { } else {
PackSizes sizes = node->sizes; PackSizes sizes = node->sizes;
uint8_t tSizes = FST_GET_TRANSITION_PACK_SIZE(sizes); uint8_t tSizes = FST_GET_TRANSITION_PACK_SIZE(sizes);
uint64_t i = node->start uint64_t i = node->start - fstStateInputLen(s) - 1 // PackSizes
- fstStateInputLen(s)
- 1 // PackSizes
- tSizes; - tSizes;
// refactor error logic // refactor error logic
uint8_t *data = fstSliceData(slice, NULL); uint8_t *data = fstSliceData(slice, NULL);
return unpackDelta(data +i, tSizes, node->end); return unpackDelta(data + i, tSizes, node->end);
} }
} }
CompiledAddr fstStateTransAddrForAnyTrans(FstState *s, FstNode *node, uint64_t i) { CompiledAddr fstStateTransAddrForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
@ -428,20 +415,15 @@ CompiledAddr fstStateTransAddrForAnyTrans(FstState *s, FstNode *node, uint64_t i
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
uint8_t tSizes = FST_GET_TRANSITION_PACK_SIZE(node->sizes); uint8_t tSizes = FST_GET_TRANSITION_PACK_SIZE(node->sizes);
uint64_t at = node->start uint64_t at = node->start - fstStateNtransLen(s) - 1 - fstStateTransIndexSize(s, node->version, node->nTrans) -
- fstStateNtransLen(s) node->nTrans - (i * tSizes) - tSizes;
- 1
- fstStateTransIndexSize(s, node->version, node->nTrans)
- node->nTrans
- (i * tSizes)
- tSizes;
uint8_t *data = fstSliceData(slice, NULL); uint8_t *data = fstSliceData(slice, NULL);
return unpackDelta(data + at, tSizes, node->end); return unpackDelta(data + at, tSizes, node->end);
} }
// sizes // sizes
PackSizes fstStateSizes(FstState *s, FstSlice *slice) { PackSizes fstStateSizes(FstState *s, FstSlice *slice) {
assert(s->state == OneTrans || s->state == AnyTrans) ; assert(s->state == OneTrans || s->state == AnyTrans);
uint64_t i; uint64_t i;
if (s->state == OneTrans) { if (s->state == OneTrans) {
i = FST_SLICE_LEN(slice) - 1 - fstStateInputLen(s) - 1; i = FST_SLICE_LEN(slice) - 1 - fstStateInputLen(s) - 1;
@ -450,7 +432,7 @@ PackSizes fstStateSizes(FstState *s, FstSlice *slice) {
} }
uint8_t *data = fstSliceData(slice, NULL); uint8_t *data = fstSliceData(slice, NULL);
return (PackSizes)(*(data +i)); return (PackSizes)(*(data + i));
} }
// Output // Output
Output fstStateOutput(FstState *s, FstNode *node) { Output fstStateOutput(FstState *s, FstNode *node) {
@ -463,14 +445,9 @@ Output fstStateOutput(FstState *s, FstNode *node) {
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
uint8_t tSizes = FST_GET_TRANSITION_PACK_SIZE(node->sizes); uint8_t tSizes = FST_GET_TRANSITION_PACK_SIZE(node->sizes);
uint64_t i = node->start uint64_t i = node->start - fstStateInputLen(s) - 1 - tSizes - oSizes;
- fstStateInputLen(s)
- 1
- tSizes
- oSizes;
uint8_t *data = fstSliceData(slice, NULL); uint8_t *data = fstSliceData(slice, NULL);
return unpackUint64(data + i, oSizes); return unpackUint64(data + i, oSizes);
} }
Output fstStateOutputForAnyTrans(FstState *s, FstNode *node, uint64_t i) { Output fstStateOutputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
assert(s->state == AnyTrans); assert(s->state == AnyTrans);
@ -480,13 +457,9 @@ Output fstStateOutputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
return 0; return 0;
} }
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
uint8_t *data = fstSliceData(slice, NULL); uint8_t * data = fstSliceData(slice, NULL);
uint64_t at = node->start uint64_t at = node->start - fstStateNtransLen(s) - 1 // pack size
- fstStateNtransLen(s) - fstStateTotalTransSize(s, node->version, node->sizes, node->nTrans) - (i * oSizes) - oSizes;
- 1 // pack size
- fstStateTotalTransSize(s, node->version, node->sizes, node->nTrans)
- (i * oSizes)
- oSizes;
return unpackUint64(data + at, oSizes); return unpackUint64(data + at, oSizes);
} }
@ -495,7 +468,9 @@ Output fstStateOutputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
void fstStateSetFinalState(FstState *s, bool yes) { void fstStateSetFinalState(FstState *s, bool yes) {
assert(s->state == AnyTrans); assert(s->state == AnyTrans);
if (yes) { s->val |= 0b01000000; } if (yes) {
s->val |= 0b01000000;
}
return; return;
} }
bool fstStateIsFinalState(FstState *s) { bool fstStateIsFinalState(FstState *s) {
@ -528,7 +503,7 @@ uint64_t fstStateTotalTransSize(FstState *s, uint64_t version, PackSizes sizes,
} }
uint64_t fstStateTransIndexSize(FstState *s, uint64_t version, uint64_t nTrans) { uint64_t fstStateTransIndexSize(FstState *s, uint64_t version, uint64_t nTrans) {
assert(s->state == AnyTrans); assert(s->state == AnyTrans);
return (version >= 2 &&nTrans > TRANS_INDEX_THRESHOLD) ? 256 : 0; return (version >= 2 && nTrans > TRANS_INDEX_THRESHOLD) ? 256 : 0;
} }
uint64_t fstStateNtransLen(FstState *s) { uint64_t fstStateNtransLen(FstState *s) {
assert(s->state == AnyTrans); assert(s->state == AnyTrans);
@ -545,8 +520,9 @@ uint64_t fstStateNtrans(FstState *s, FstSlice *slice) {
int32_t len; int32_t len;
uint8_t *data = fstSliceData(slice, &len); uint8_t *data = fstSliceData(slice, &len);
n = data[len - 2]; n = data[len - 2];
//n = data[slice->end - 1]; // data[data.len() - 2] // n = data[slice->end - 1]; // data[data.len() - 2]
return n == 1 ? 256: n; // // "1" is never a normal legal value here, because if there, // is only 1 transition, then it is encoded in the state byte return n == 1 ? 256 : n; // // "1" is never a normal legal value here, because if there, // is only 1 transition,
// then it is encoded in the state byte
} }
Output fstStateFinalOutput(FstState *s, uint64_t version, FstSlice *slice, PackSizes sizes, uint64_t nTrans) { Output fstStateFinalOutput(FstState *s, uint64_t version, FstSlice *slice, PackSizes sizes, uint64_t nTrans) {
uint8_t oSizes = FST_GET_OUTPUT_PACK_SIZE(sizes); uint8_t oSizes = FST_GET_OUTPUT_PACK_SIZE(sizes);
@ -554,61 +530,54 @@ Output fstStateFinalOutput(FstState *s, uint64_t version, FstSlice *slice, Pack
return 0; return 0;
} }
uint64_t at = FST_SLICE_LEN(slice) uint64_t at = FST_SLICE_LEN(slice) - 1 - fstStateNtransLen(s) - 1 // pack size
- 1 - fstStateTotalTransSize(s, version, sizes, nTrans) - (nTrans * oSizes) - oSizes;
- fstStateNtransLen(s)
- 1 // pack size
- fstStateTotalTransSize(s, version, sizes, nTrans)
- (nTrans * oSizes)
- oSizes;
uint8_t *data = fstSliceData(slice, NULL); uint8_t *data = fstSliceData(slice, NULL);
return unpackUint64(data + at, (uint8_t)oSizes); return unpackUint64(data + at, (uint8_t)oSizes);
} }
uint64_t fstStateFindInput(FstState *s, FstNode *node, uint8_t b, bool *null) { uint64_t fstStateFindInput(FstState *s, FstNode *node, uint8_t b, bool *null) {
assert(s->state == AnyTrans); assert(s->state == AnyTrans);
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
if (node->version >= 2 && node->nTrans > TRANS_INDEX_THRESHOLD) { if (node->version >= 2 && node->nTrans > TRANS_INDEX_THRESHOLD) {
uint64_t at = node->start uint64_t at = node->start - fstStateNtransLen(s) - 1 // pack size
- fstStateNtransLen(s)
- 1 // pack size
- fstStateTransIndexSize(s, node->version, node->nTrans); - fstStateTransIndexSize(s, node->version, node->nTrans);
int32_t dlen = 0; int32_t dlen = 0;
uint8_t *data = fstSliceData(slice, &dlen); uint8_t *data = fstSliceData(slice, &dlen);
uint64_t i = data[at + b]; uint64_t i = data[at + b];
//uint64_t i = slice->data[slice->start + at + b]; // uint64_t i = slice->data[slice->start + at + b];
if (i >= node->nTrans) { if (i >= node->nTrans) {
*null = true; *null = true;
} }
return i; return i;
} else { } else {
uint64_t start = node->start uint64_t start = node->start - fstStateNtransLen(s) - 1 // pack size
- fstStateNtransLen(s)
- 1 // pack size
- node->nTrans; - node->nTrans;
uint64_t end = start + node->nTrans; uint64_t end = start + node->nTrans;
FstSlice t = fstSliceCopy(slice, start, end - 1); FstSlice t = fstSliceCopy(slice, start, end - 1);
int32_t len = 0; int32_t len = 0;
uint8_t *data = fstSliceData(&t, &len); uint8_t *data = fstSliceData(&t, &len);
int i = 0; int i = 0;
for(; i < len; i++) { for (; i < len; i++) {
uint8_t v = data[i]; uint8_t v = data[i];
if (v == b) { if (v == b) {
fstSliceDestroy(&t); fstSliceDestroy(&t);
return node->nTrans - i - 1; // bug return node->nTrans - i - 1; // bug
} }
} }
if (i == len) { *null = true; } if (i == len) {
*null = true;
}
fstSliceDestroy(&t); fstSliceDestroy(&t);
} }
} }
// fst node function // fst node function
FstNode *fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice *slice) { FstNode *fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice *slice) {
FstNode *n = (FstNode *)malloc(sizeof(FstNode)); FstNode *n = (FstNode *)malloc(sizeof(FstNode));
if (n == NULL) { return NULL; } if (n == NULL) {
return NULL;
}
FstState st = fstStateCreateFrom(slice, addr); FstState st = fstStateCreateFrom(slice, addr);
@ -656,7 +625,8 @@ FstNode *fstNodeCreate(int64_t version, CompiledAddr addr, FstSlice *slice) {
n->isFinal = fstStateIsFinalState(&st); // s.is_final_state(); n->isFinal = fstStateIsFinalState(&st); // s.is_final_state();
n->nTrans = nTrans; n->nTrans = nTrans;
n->sizes = sz; n->sizes = sz;
n->finalOutput = fstStateFinalOutput(&st, version, &data, sz, nTrans); // s.final_output(version, data, sz, ntrans); n->finalOutput =
fstStateFinalOutput(&st, version, &data, sz, nTrans); // s.final_output(version, data, sz, ntrans);
} }
return n; return n;
} }
@ -667,12 +637,11 @@ static const char *fstNodeState(FstNode *node) {
return fstStateStr[st->state]; return fstStateStr[st->state];
} }
void fstNodeDestroy(FstNode *node) { void fstNodeDestroy(FstNode *node) {
fstSliceDestroy(&node->data); fstSliceDestroy(&node->data);
free(node); free(node);
} }
FstTransitions* fstNodeTransitions(FstNode *node) { FstTransitions *fstNodeTransitions(FstNode *node) {
FstTransitions *t = malloc(sizeof(FstTransitions)); FstTransitions *t = malloc(sizeof(FstTransitions));
if (NULL == t) { if (NULL == t) {
return NULL; return NULL;
@ -717,7 +686,7 @@ bool fstNodeGetTransitionAddrAt(FstNode *node, uint64_t i, CompiledAddr *res) {
fstStateTransAddr(st, node); fstStateTransAddr(st, node);
} else if (st->state == AnyTrans) { } else if (st->state == AnyTrans) {
fstStateTransAddrForAnyTrans(st, node, i); fstStateTransAddrForAnyTrans(st, node, i);
} else if (FST_STATE_EMPTY_FINAL(node)){ } else if (FST_STATE_EMPTY_FINAL(node)) {
s = false; s = false;
} else { } else {
assert(0); assert(0);
@ -731,16 +700,25 @@ bool fstNodeFindInput(FstNode *node, uint8_t b, uint64_t *res) {
bool s = true; bool s = true;
FstState *st = &node->state; FstState *st = &node->state;
if (st->state == OneTransNext) { if (st->state == OneTransNext) {
if (fstStateInput(st,node) == b) { *res = 0; } if (fstStateInput(st, node) == b) {
else { s = false; } } *res = 0;
else if (st->state == OneTrans) { } else {
if (fstStateInput(st, node) == b) { *res = 0 ;} s = false;
else { s = false; } }
} else if (st->state == OneTrans) {
if (fstStateInput(st, node) == b) {
*res = 0;
} else {
s = false;
}
} else if (st->state == AnyTrans) { } else if (st->state == AnyTrans) {
bool null = false; bool null = false;
uint64_t out = fstStateFindInput(st, node, b, &null); uint64_t out = fstStateFindInput(st, node, b, &null);
if (null == false) { *res = out; } if (null == false) {
else { s = false;} *res = out;
} else {
s = false;
}
} }
return s; return s;
} }
@ -757,11 +735,11 @@ bool fstNodeCompile(FstNode *node, void *w, CompiledAddr lastAddr, CompiledAddr
FstTransition *tran = taosArrayGet(builderNode->trans, 0); FstTransition *tran = taosArrayGet(builderNode->trans, 0);
if (tran->addr == lastAddr && tran->out == 0) { if (tran->addr == lastAddr && tran->out == 0) {
fstStateCompileForOneTransNext(w, addr, tran->inp); fstStateCompileForOneTransNext(w, addr, tran->inp);
//OneTransNext::compile(w, lastAddr, tran->inp); // OneTransNext::compile(w, lastAddr, tran->inp);
return true; return true;
} else { } else {
fstStateCompileForOneTrans(w, addr, tran); fstStateCompileForOneTrans(w, addr, tran);
//OneTrans::Compile(w, lastAddr, *tran); // OneTrans::Compile(w, lastAddr, *tran);
return true; return true;
} }
} }
@ -772,16 +750,15 @@ bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, Compiled
return fstNodeCompile(NULL, wrt, lastAddr, startAddr, b); return fstNodeCompile(NULL, wrt, lastAddr, startAddr, b);
} }
FstBuilder *fstBuilderCreate(void *w, FstType ty) { FstBuilder *fstBuilderCreate(void *w, FstType ty) {
FstBuilder *b = malloc(sizeof(FstBuilder)); FstBuilder *b = malloc(sizeof(FstBuilder));
if (NULL == b) { return b; } if (NULL == b) {
return b;
}
b->wrt = fstCountingWriterCreate(w); b->wrt = fstCountingWriterCreate(w);
b->unfinished = fstUnFinishedNodesCreate(); b->unfinished = fstUnFinishedNodesCreate();
b->registry = fstRegistryCreate(10000, 2) ; b->registry = fstRegistryCreate(10000, 2);
b->last = fstSliceCreate(NULL, 0); b->last = fstSliceCreate(NULL, 0);
b->lastAddr = NONE_ADDRESS; b->lastAddr = NONE_ADDRESS;
b->len = 0; b->len = 0;
@ -799,7 +776,9 @@ FstBuilder *fstBuilderCreate(void *w, FstType ty) {
return b; return b;
} }
void fstBuilderDestroy(FstBuilder *b) { void fstBuilderDestroy(FstBuilder *b) {
if (b == NULL) { return; } if (b == NULL) {
return;
}
fstCountingWriterDestroy(b->wrt); fstCountingWriterDestroy(b->wrt);
fstUnFinishedNodesDestroy(b->unfinished); fstUnFinishedNodesDestroy(b->unfinished);
@ -808,7 +787,6 @@ void fstBuilderDestroy(FstBuilder *b) {
free(b); free(b);
} }
bool fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in) { bool fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in) {
OrderType t = fstBuilderCheckLastKey(b, bs, true); OrderType t = fstBuilderCheckLastKey(b, bs, true);
if (t == Ordered) { if (t == Ordered) {
@ -827,7 +805,7 @@ void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in) {
fstUnFinishedNodesSetRootOutput(b->unfinished, in); fstUnFinishedNodesSetRootOutput(b->unfinished, in);
return; return;
} }
//if (in != 0) { //if let Some(in) = in // if (in != 0) { //if let Some(in) = in
// prefixLen = fstUnFinishedNodesFindCommPrefixAndSetOutput(b->unfinished, bs, in, &out); // prefixLen = fstUnFinishedNodesFindCommPrefixAndSetOutput(b->unfinished, bs, in, &out);
//} else { //} else {
// prefixLen = fstUnFinishedNodesFindCommPrefix(b->unfinished, bs); // prefixLen = fstUnFinishedNodesFindCommPrefix(b->unfinished, bs);
@ -848,7 +826,7 @@ void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in) {
fstUnFinishedNodesAddSuffix(b->unfinished, sub, out); fstUnFinishedNodesAddSuffix(b->unfinished, sub, out);
fstSliceDestroy(&sub); fstSliceDestroy(&sub);
return; return;
} }
OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup) { OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup) {
FstSlice *input = &bs; FstSlice *input = &bs;
@ -882,15 +860,13 @@ void fstBuilderCompileFrom(FstBuilder *b, uint64_t istate) {
fstBuilderNodeDestroy(bn); fstBuilderNodeDestroy(bn);
assert(addr != NONE_ADDRESS); assert(addr != NONE_ADDRESS);
//fstBuilderNodeDestroy(n); // fstBuilderNodeDestroy(n);
} }
fstUnFinishedNodesTopLastFreeze(b->unfinished, addr); fstUnFinishedNodesTopLastFreeze(b->unfinished, addr);
return; return;
} }
CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn) { CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn) {
if (FST_BUILDER_NODE_IS_FINAL(bn) if (FST_BUILDER_NODE_IS_FINAL(bn) && FST_BUILDER_NODE_TRANS_ISEMPTY(bn) && FST_BUILDER_NODE_FINALOUTPUT_ISZERO(bn)) {
&& FST_BUILDER_NODE_TRANS_ISEMPTY(bn)
&& FST_BUILDER_NODE_FINALOUTPUT_ISZERO(bn)) {
return EMPTY_ADDRESS; return EMPTY_ADDRESS;
} }
FstRegistryEntry *entry = fstRegistryGetEntry(b->registry, bn); FstRegistryEntry *entry = fstRegistryGetEntry(b->registry, bn);
@ -911,7 +887,7 @@ CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn) {
return b->lastAddr; return b->lastAddr;
} }
void* fstBuilderInsertInner(FstBuilder *b) { void *fstBuilderInsertInner(FstBuilder *b) {
fstBuilderCompileFrom(b, 0); fstBuilderCompileFrom(b, 0);
FstBuilderNode *rootNode = fstUnFinishedNodesPopRoot(b->unfinished); FstBuilderNode *rootNode = fstUnFinishedNodesPopRoot(b->unfinished);
CompiledAddr rootAddr = fstBuilderCompile(b, rootNode); CompiledAddr rootAddr = fstBuilderCompile(b, rootNode);
@ -928,21 +904,17 @@ void* fstBuilderInsertInner(FstBuilder *b) {
fstCountingWriterWrite(b->wrt, buf64, sizeof(buf64)); fstCountingWriterWrite(b->wrt, buf64, sizeof(buf64));
char buf32[4] = {0}; char buf32[4] = {0};
void *pBuf32 = buf32; void * pBuf32 = buf32;
uint32_t sum = fstCountingWriterMaskedCheckSum(b->wrt); uint32_t sum = fstCountingWriterMaskedCheckSum(b->wrt);
taosEncodeFixedU32(&pBuf32, sum); taosEncodeFixedU32(&pBuf32, sum);
fstCountingWriterWrite(b->wrt, buf32, sizeof(buf32)); fstCountingWriterWrite(b->wrt, buf32, sizeof(buf32));
fstCountingWriterFlush(b->wrt); fstCountingWriterFlush(b->wrt);
//fstCountingWriterDestroy(b->wrt); // fstCountingWriterDestroy(b->wrt);
//b->wrt = NULL; // b->wrt = NULL;
return b->wrt; return b->wrt;
} }
void fstBuilderFinish(FstBuilder *b) { void fstBuilderFinish(FstBuilder *b) { fstBuilderInsertInner(b); }
fstBuilderInsertInner(b);
}
FstSlice fstNodeAsSlice(FstNode *node) { FstSlice fstNodeAsSlice(FstNode *node) {
FstSlice *slice = &node->data; FstSlice *slice = &node->data;
@ -952,19 +924,21 @@ FstSlice fstNodeAsSlice(FstNode *node) {
FstLastTransition *fstLastTransitionCreate(uint8_t inp, Output out) { FstLastTransition *fstLastTransitionCreate(uint8_t inp, Output out) {
FstLastTransition *trn = malloc(sizeof(FstLastTransition)); FstLastTransition *trn = malloc(sizeof(FstLastTransition));
if (trn == NULL) { return NULL; } if (trn == NULL) {
return NULL;
}
trn->inp = inp; trn->inp = inp;
trn->out = out; trn->out = out;
return trn; return trn;
} }
void fstLastTransitionDestroy(FstLastTransition *trn) { void fstLastTransitionDestroy(FstLastTransition *trn) { free(trn); }
free(trn);
}
void fstBuilderNodeUnfinishedLastCompiled(FstBuilderNodeUnfinished *unNode, CompiledAddr addr) { void fstBuilderNodeUnfinishedLastCompiled(FstBuilderNodeUnfinished *unNode, CompiledAddr addr) {
FstLastTransition *trn = unNode->last; FstLastTransition *trn = unNode->last;
if (trn == NULL) { return; } if (trn == NULL) {
return;
}
FstTransition t = {.inp = trn->inp, .out = trn->out, .addr = addr}; FstTransition t = {.inp = trn->inp, .out = trn->out, .addr = addr};
taosArrayPush(unNode->node->trans, &t); taosArrayPush(unNode->node->trans, &t);
fstLastTransitionDestroy(trn); fstLastTransitionDestroy(trn);
@ -987,9 +961,9 @@ void fstBuilderNodeUnfinishedAddOutputPrefix(FstBuilderNodeUnfinished *unNode, O
return; return;
} }
Fst* fstCreate(FstSlice *slice) { Fst *fstCreate(FstSlice *slice) {
int32_t slen; int32_t slen;
char *buf = fstSliceData(slice, &slen); char * buf = fstSliceData(slice, &slen);
if (slen < 36) { if (slen < 36) {
return NULL; return NULL;
} }
@ -1018,9 +992,11 @@ Fst* fstCreate(FstSlice *slice) {
uint64_t fstLen; uint64_t fstLen;
len -= sizeof(fstLen); len -= sizeof(fstLen);
taosDecodeFixedU64(buf + len, &fstLen); taosDecodeFixedU64(buf + len, &fstLen);
//TODO(validate root addr) // TODO(validate root addr)
Fst *fst= (Fst *)calloc(1, sizeof(Fst)); Fst *fst = (Fst *)calloc(1, sizeof(Fst));
if (fst == NULL) { return NULL; } if (fst == NULL) {
return NULL;
}
fst->meta = (FstMeta *)malloc(sizeof(FstMeta)); fst->meta = (FstMeta *)malloc(sizeof(FstMeta));
if (NULL == fst->meta) { if (NULL == fst->meta) {
@ -1042,7 +1018,6 @@ Fst* fstCreate(FstSlice *slice) {
FST_CREAT_FAILED: FST_CREAT_FAILED:
free(fst->meta); free(fst->meta);
free(fst); free(fst);
} }
void fstDestroy(Fst *fst) { void fstDestroy(Fst *fst) {
if (fst) { if (fst) {
@ -1091,16 +1066,14 @@ bool fstGet(Fst *fst, FstSlice *b, Output *out) {
return true; return true;
} }
FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx) { FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx) { return fstStreamBuilderCreate(fst, ctx); }
return fstStreamBuilderCreate(fst, ctx); StreamWithState * streamBuilderIntoStream(FstStreamBuilder *sb) {
} if (sb == NULL) {
StreamWithState* streamBuilderIntoStream(FstStreamBuilder *sb) { return NULL;
if (sb == NULL) { return NULL; } }
return streamWithStateCreate(sb->fst, sb->aut, sb->min, sb->max); return streamWithStateCreate(sb->fst, sb->aut, sb->min, sb->max);
} }
FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx) { FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx) { return fstStreamBuilderCreate(fst, ctx); }
return fstStreamBuilderCreate(fst, ctx);
}
FstNode *fstGetRoot(Fst *fst) { FstNode *fstGetRoot(Fst *fst) {
if (fst->root != NULL) { if (fst->root != NULL) {
@ -1110,16 +1083,9 @@ FstNode *fstGetRoot(Fst *fst) {
fst->root = fstGetNode(fst, rAddr); fst->root = fstGetNode(fst, rAddr);
return fst->root; return fst->root;
} }
FstNode* fstGetNode(Fst *fst, CompiledAddr addr) { FstNode * fstGetNode(Fst *fst, CompiledAddr addr) { return fstNodeCreate(fst->meta->version, addr, fst->data); }
return fstNodeCreate(fst->meta->version, addr, fst->data); FstType fstGetType(Fst *fst) { return fst->meta->ty; }
CompiledAddr fstGetRootAddr(Fst *fst) { return fst->meta->rootAddr; }
}
FstType fstGetType(Fst *fst) {
return fst->meta->ty;
}
CompiledAddr fstGetRootAddr(Fst *fst) {
return fst->meta->rootAddr;
}
Output fstEmptyFinalOutput(Fst *fst, bool *null) { Output fstEmptyFinalOutput(Fst *fst, bool *null) {
Output res = 0; Output res = 0;
@ -1145,9 +1111,11 @@ bool fstVerify(Fst *fst) {
} }
// data bound function // data bound function
FstBoundWithData* fstBoundStateCreate(FstBound type, FstSlice *data) { FstBoundWithData *fstBoundStateCreate(FstBound type, FstSlice *data) {
FstBoundWithData *b = calloc(1, sizeof(FstBoundWithData)); FstBoundWithData *b = calloc(1, sizeof(FstBoundWithData));
if (b == NULL) { return NULL; } if (b == NULL) {
return NULL;
}
if (data != NULL) { if (data != NULL) {
b->data = fstSliceCopy(data, data->start, data->end); b->data = fstSliceCopy(data, data->start, data->end);
@ -1159,7 +1127,6 @@ FstBoundWithData* fstBoundStateCreate(FstBound type, FstSlice *data) {
return b; return b;
} }
bool fstBoundWithDataExceededBy(FstBoundWithData *bound, FstSlice *slice) { bool fstBoundWithDataExceededBy(FstBoundWithData *bound, FstSlice *slice) {
int comp = fstSliceCompare(slice, &bound->data); int comp = fstSliceCompare(slice, &bound->data);
if (bound->type == Included) { if (bound->type == Included) {
@ -1178,18 +1145,16 @@ bool fstBoundWithDataIsEmpty(FstBoundWithData *bound) {
} }
} }
bool fstBoundWithDataIsIncluded(FstBoundWithData *bound) { return bound->type == Excluded ? false : true; }
bool fstBoundWithDataIsIncluded(FstBoundWithData *bound) { void fstBoundDestroy(FstBoundWithData *bound) { free(bound); }
return bound->type == Excluded? false : true;
}
void fstBoundDestroy(FstBoundWithData *bound) { StreamWithState *streamWithStateCreate(
free(bound); Fst *fst, AutomationCtx *automation, FstBoundWithData *min, FstBoundWithData *max) {
}
StreamWithState *streamWithStateCreate(Fst *fst, AutomationCtx *automation, FstBoundWithData *min, FstBoundWithData *max) {
StreamWithState *sws = calloc(1, sizeof(StreamWithState)); StreamWithState *sws = calloc(1, sizeof(StreamWithState));
if (sws == NULL) { return NULL; } if (sws == NULL) {
return NULL;
}
sws->fst = fst; sws->fst = fst;
sws->aut = automation; sws->aut = automation;
@ -1205,7 +1170,9 @@ StreamWithState *streamWithStateCreate(Fst *fst, AutomationCtx *automation, FstB
return sws; return sws;
} }
void streamWithStateDestroy(StreamWithState *sws) { void streamWithStateDestroy(StreamWithState *sws) {
if (sws == NULL) { return; } if (sws == NULL) {
return;
}
taosArrayDestroy(sws->inp); taosArrayDestroy(sws->inp);
taosArrayDestroyEx(sws->stack, streamStateDestroy); taosArrayDestroyEx(sws->stack, streamStateDestroy);
@ -1214,7 +1181,6 @@ void streamWithStateDestroy(StreamWithState *sws) {
} }
bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) { bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
AutomationCtx *aut = sws->aut; AutomationCtx *aut = sws->aut;
if (fstBoundWithDataIsEmpty(min)) { if (fstBoundWithDataIsEmpty(min)) {
if (fstBoundWithDataIsIncluded(min)) { if (fstBoundWithDataIsIncluded(min)) {
@ -1228,7 +1194,8 @@ bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
return true; return true;
} }
FstSlice *key = NULL; FstSlice *key = NULL;
bool inclusize = false;; bool inclusize = false;
;
if (min->type == Included) { if (min->type == Included) {
key = &min->data; key = &min->data;
@ -1241,8 +1208,8 @@ bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
FstNode *node = fstGetRoot(sws->fst); FstNode *node = fstGetRoot(sws->fst);
Output out = 0; Output out = 0;
//void* autState = sws->aut->start(); // void* autState = sws->aut->start();
void* autState = automFuncs[aut->type].start(aut); void *autState = automFuncs[aut->type].start(aut);
int32_t len; int32_t len;
uint8_t *data = fstSliceData(key, &len); uint8_t *data = fstSliceData(key, &len);
@ -1257,16 +1224,12 @@ bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
// autState = sws->aut->accept(preState, b); // autState = sws->aut->accept(preState, b);
autState = automFuncs[aut->type].accept(aut, preState, b); autState = automFuncs[aut->type].accept(aut, preState, b);
taosArrayPush(sws->inp, &b); taosArrayPush(sws->inp, &b);
StreamState s = {.node = node, StreamState s = {.node = node, .trans = res + 1, .out = {.null = false, .out = out}, .autState = preState};
.trans = res + 1,
.out = {.null = false, .out = out},
.autState = preState};
taosArrayPush(sws->stack, &s); taosArrayPush(sws->stack, &s);
out += trn.out; out += trn.out;
node = fstGetNode(sws->fst, trn.addr); node = fstGetNode(sws->fst, trn.addr);
fstNodeDestroy(node); fstNodeDestroy(node);
} else { } else {
// This is a little tricky. We're in this case if the // This is a little tricky. We're in this case if the
// given bound is not a prefix of any key in the FST. // given bound is not a prefix of any key in the FST.
// Since this is a minimum bound, we need to find the // Since this is a minimum bound, we need to find the
@ -1281,10 +1244,7 @@ bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
} }
} }
StreamState s = {.node = node, StreamState s = {.node = node, .trans = i, .out = {.null = false, .out = out}, .autState = autState};
.trans = i,
.out = {.null = false, .out = out},
.autState = autState};
taosArrayPush(sws->stack, &s); taosArrayPush(sws->stack, &s);
return true; return true;
} }
@ -1296,14 +1256,12 @@ bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
s->trans -= 1; s->trans -= 1;
taosArrayPop(sws->inp); taosArrayPop(sws->inp);
} else { } else {
FstNode *n = s->node; FstNode * n = s->node;
uint64_t trans = s->trans; uint64_t trans = s->trans;
FstTransition trn; FstTransition trn;
fstNodeGetTransitionAt(n, trans - 1, &trn); fstNodeGetTransitionAt(n, trans - 1, &trn);
StreamState s = {.node = fstGetNode(sws->fst, trn.addr), StreamState s = {
.trans= 0, .node = fstGetNode(sws->fst, trn.addr), .trans = 0, .out = {.null = false, .out = out}, .autState = autState};
.out = {.null = false, .out = out},
.autState = autState};
taosArrayPush(sws->stack, &s); taosArrayPush(sws->stack, &s);
return true; return true;
} }
@ -1340,27 +1298,26 @@ StreamWithStateResult *streamWithStateNextWith(StreamWithState *sws, StreamCallb
FstTransition trn; FstTransition trn;
fstNodeGetTransitionAt(p->node, p->trans, &trn); fstNodeGetTransitionAt(p->node, p->trans, &trn);
Output out = p->out.out + trn.out; Output out = p->out.out + trn.out;
void* nextState = automFuncs[aut->type].accept(aut, p->autState, trn.inp); void * nextState = automFuncs[aut->type].accept(aut, p->autState, trn.inp);
void* tState = callback(nextState); void * tState = callback(nextState);
bool isMatch = automFuncs[aut->type].isMatch(aut, nextState); bool isMatch = automFuncs[aut->type].isMatch(aut, nextState);
FstNode *nextNode = fstGetNode(sws->fst, trn.addr); FstNode *nextNode = fstGetNode(sws->fst, trn.addr);
taosArrayPush(nodes, &nextNode); taosArrayPush(nodes, &nextNode);
taosArrayPush(sws->inp, &(trn.inp)); taosArrayPush(sws->inp, &(trn.inp));
if (FST_NODE_IS_FINAL(nextNode)) { if (FST_NODE_IS_FINAL(nextNode)) {
//void *eofState = sws->aut->acceptEof(nextState); // void *eofState = sws->aut->acceptEof(nextState);
void *eofState = automFuncs[aut->type].acceptEof(aut, nextState); void *eofState = automFuncs[aut->type].acceptEof(aut, nextState);
if (eofState != NULL) { if (eofState != NULL) {
isMatch = automFuncs[aut->type].isMatch(aut, eofState); isMatch = automFuncs[aut->type].isMatch(aut, eofState);
} }
} }
StreamState s1 = { .node = p->node, .trans = p->trans + 1, .out = p->out, .autState = p->autState}; StreamState s1 = {.node = p->node, .trans = p->trans + 1, .out = p->out, .autState = p->autState};
taosArrayPush(sws->stack, &s1); taosArrayPush(sws->stack, &s1);
StreamState s2 = {.node = nextNode, .trans = 0, .out = {.null = false, .out = out}, .autState = nextState}; StreamState s2 = {.node = nextNode, .trans = 0, .out = {.null = false, .out = out}, .autState = nextState};
taosArrayPush(sws->stack, &s2); taosArrayPush(sws->stack, &s2);
size_t isz = taosArrayGetSize(sws->inp); size_t isz = taosArrayGetSize(sws->inp);
uint8_t *buf = (uint8_t *)malloc(isz * sizeof(uint8_t)); uint8_t *buf = (uint8_t *)malloc(isz * sizeof(uint8_t));
for (uint32_t i = 0; i < isz; i++) { for (uint32_t i = 0; i < isz; i++) {
@ -1385,17 +1342,18 @@ StreamWithStateResult *streamWithStateNextWith(StreamWithState *sws, StreamCallb
fstSliceDestroy(&slice); fstSliceDestroy(&slice);
} }
for (size_t i = 0; i < taosArrayGetSize(nodes); i++) { for (size_t i = 0; i < taosArrayGetSize(nodes); i++) {
FstNode** node = (FstNode **)taosArrayGet(nodes, i); FstNode **node = (FstNode **)taosArrayGet(nodes, i);
fstNodeDestroy(*node); fstNodeDestroy(*node);
} }
taosArrayDestroy(nodes); taosArrayDestroy(nodes);
return NULL; return NULL;
} }
StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *state) { StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *state) {
StreamWithStateResult *result = calloc(1, sizeof(StreamWithStateResult)); StreamWithStateResult *result = calloc(1, sizeof(StreamWithStateResult));
if (result == NULL) { return NULL; } if (result == NULL) {
return NULL;
}
result->data = fstSliceCopy(data, 0, FST_SLICE_LEN(data) - 1); result->data = fstSliceCopy(data, 0, FST_SLICE_LEN(data) - 1);
result->out = fOut; result->out = fOut;
@ -1404,7 +1362,9 @@ StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *sta
return result; return result;
} }
void swsResultDestroy(StreamWithStateResult *result) { void swsResultDestroy(StreamWithStateResult *result) {
if (NULL == result) { return; } if (NULL == result) {
return;
}
fstSliceDestroy(&result->data); fstSliceDestroy(&result->data);
startWithStateValueDestroy(result->state); startWithStateValueDestroy(result->state);
@ -1412,16 +1372,20 @@ void swsResultDestroy(StreamWithStateResult *result) {
} }
void streamStateDestroy(void *s) { void streamStateDestroy(void *s) {
if (NULL == s) { return; } if (NULL == s) {
return;
}
StreamState *ss = (StreamState *)s; StreamState *ss = (StreamState *)s;
fstNodeDestroy(ss->node); fstNodeDestroy(ss->node);
//free(s->autoState); // free(s->autoState);
} }
FstStreamBuilder *fstStreamBuilderCreate(Fst *fst, AutomationCtx *aut) { FstStreamBuilder *fstStreamBuilderCreate(Fst *fst, AutomationCtx *aut) {
FstStreamBuilder *b = calloc(1, sizeof(FstStreamBuilder)); FstStreamBuilder *b = calloc(1, sizeof(FstStreamBuilder));
if (NULL == b) { return NULL; } if (NULL == b) {
return NULL;
}
b->fst = fst; b->fst = fst;
b->aut = aut; b->aut = aut;
@ -1437,7 +1401,9 @@ void fstStreamBuilderDestroy(FstStreamBuilder *b) {
free(b); free(b);
} }
FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type) { FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type) {
if (b == NULL) { return NULL; } if (b == NULL) {
return NULL;
}
if (type == GE) { if (type == GE) {
b->min->type = Included; b->min->type = Included;
@ -1458,9 +1424,3 @@ FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, Rang
} }
return b; return b;
} }

96
source/libs/index/src/index_fst_automation.c
View File

@ -15,10 +15,11 @@
#include "index_fst_automation.h" #include "index_fst_automation.h"
StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void *val) { StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void *val) {
StartWithStateValue *nsv = calloc(1, sizeof(StartWithStateValue)); StartWithStateValue *nsv = calloc(1, sizeof(StartWithStateValue));
if (nsv == NULL) { return NULL; } if (nsv == NULL) {
return NULL;
}
nsv->kind = kind; nsv->kind = kind;
nsv->type = ty; nsv->type = ty;
@ -29,14 +30,16 @@ StartWithStateValue *startWithStateValueCreate(StartWithStateKind kind, ValueTyp
nsv->ptr = (char *)calloc(1, len + 1); nsv->ptr = (char *)calloc(1, len + 1);
memcpy(nsv->ptr, val, len); memcpy(nsv->ptr, val, len);
} else if (ty == FST_ARRAY) { } else if (ty == FST_ARRAY) {
//TODO, // TODO,
//nsv->arr = taosArrayFromList() // nsv->arr = taosArrayFromList()
} }
return nsv; return nsv;
} }
void startWithStateValueDestroy(void *val) { void startWithStateValueDestroy(void *val) {
StartWithStateValue *sv = (StartWithStateValue *)val; StartWithStateValue *sv = (StartWithStateValue *)val;
if (sv == NULL) { return; } if (sv == NULL) {
return;
}
if (sv->type == FST_INT) { if (sv->type == FST_INT) {
// //
@ -49,10 +52,12 @@ void startWithStateValueDestroy(void *val) {
} }
StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv) { StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv) {
StartWithStateValue *nsv = calloc(1, sizeof(StartWithStateValue)); StartWithStateValue *nsv = calloc(1, sizeof(StartWithStateValue));
if (nsv == NULL) { return NULL; } if (nsv == NULL) {
return NULL;
}
nsv->kind = sv->kind; nsv->kind = sv->kind;
nsv->type= sv->type; nsv->type = sv->type;
if (nsv->type == FST_INT) { if (nsv->type == FST_INT) {
nsv->val = sv->val; nsv->val = sv->val;
} else if (nsv->type == FST_CHAR) { } else if (nsv->type == FST_CHAR) {
@ -64,27 +69,26 @@ StartWithStateValue *startWithStateValueDump(StartWithStateValue *sv) {
return nsv; return nsv;
} }
// prefix query, impl later // prefix query, impl later
static void* prefixStart(AutomationCtx *ctx) { static void *prefixStart(AutomationCtx *ctx) {
StartWithStateValue *data = (StartWithStateValue *)(ctx->stdata); StartWithStateValue *data = (StartWithStateValue *)(ctx->stdata);
return startWithStateValueDump(data); return startWithStateValueDump(data);
}; };
static bool prefixIsMatch(AutomationCtx *ctx, void *sv) { static bool prefixIsMatch(AutomationCtx *ctx, void *sv) {
StartWithStateValue* ssv = (StartWithStateValue *)sv; StartWithStateValue *ssv = (StartWithStateValue *)sv;
return ssv->val == strlen(ctx->data); return ssv->val == strlen(ctx->data);
} }
static bool prefixCanMatch(AutomationCtx *ctx, void *sv) { static bool prefixCanMatch(AutomationCtx *ctx, void *sv) {
StartWithStateValue* ssv = (StartWithStateValue *)sv; StartWithStateValue *ssv = (StartWithStateValue *)sv;
return ssv->val >= 0; return ssv->val >= 0;
} }
static bool prefixWillAlwaysMatch(AutomationCtx *ctx, void *state) { static bool prefixWillAlwaysMatch(AutomationCtx *ctx, void *state) { return true; }
return true; static void *prefixAccept(AutomationCtx *ctx, void *state, uint8_t byte) {
} StartWithStateValue *ssv = (StartWithStateValue *)state;
static void* prefixAccept(AutomationCtx *ctx, void *state, uint8_t byte) { if (ssv == NULL || ctx == NULL) {
StartWithStateValue* ssv = (StartWithStateValue *)state; return NULL;
if (ssv == NULL || ctx == NULL) {return NULL;} }
char *data = ctx->data; char *data = ctx->data;
if (ssv->kind == Done) { if (ssv->kind == Done) {
@ -102,55 +106,30 @@ static void* prefixAccept(AutomationCtx *ctx, void *state, uint8_t byte) {
} }
return NULL; return NULL;
} }
static void* prefixAcceptEof(AutomationCtx *ctx, void *state) { static void *prefixAcceptEof(AutomationCtx *ctx, void *state) { return NULL; }
return NULL;
}
// pattern query, impl later // pattern query, impl later
static void* patternStart(AutomationCtx *ctx) { static void *patternStart(AutomationCtx *ctx) { return NULL; }
return NULL; static bool patternIsMatch(AutomationCtx *ctx, void *data) { return true; }
} static bool patternCanMatch(AutomationCtx *ctx, void *data) { return true; }
static bool patternIsMatch(AutomationCtx *ctx, void *data) { static bool patternWillAlwaysMatch(AutomationCtx *ctx, void *state) { return true; }
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) { static void *patternAccept(AutomationCtx *ctx, void *state, uint8_t byte) { return NULL; }
return NULL;
}
static void* patternAcceptEof(AutomationCtx *ctx, void *state) { static void *patternAcceptEof(AutomationCtx *ctx, void *state) { return NULL; }
return NULL;
}
AutomationFunc automFuncs[] = {{ AutomationFunc automFuncs[] = {
prefixStart, {prefixStart, prefixIsMatch, prefixCanMatch, prefixWillAlwaysMatch, prefixAccept, prefixAcceptEof},
prefixIsMatch, {patternStart, patternIsMatch, patternCanMatch, patternWillAlwaysMatch, patternAccept, patternAcceptEof}
prefixCanMatch,
prefixWillAlwaysMatch,
prefixAccept,
prefixAcceptEof
},
{
patternStart,
patternIsMatch,
patternCanMatch,
patternWillAlwaysMatch,
patternAccept,
patternAcceptEof
}
// add more search type // add more search type
}; };
AutomationCtx* automCtxCreate(void *data,AutomationType atype) { AutomationCtx *automCtxCreate(void *data, AutomationType atype) {
AutomationCtx *ctx = calloc(1, sizeof(AutomationCtx)); AutomationCtx *ctx = calloc(1, sizeof(AutomationCtx));
if (ctx == NULL) { return NULL; } if (ctx == NULL) {
return NULL;
}
StartWithStateValue *sv = NULL; StartWithStateValue *sv = NULL;
if (atype == AUTOMATION_PREFIX) { if (atype == AUTOMATION_PREFIX) {
@ -158,14 +137,13 @@ AutomationCtx* automCtxCreate(void *data,AutomationType atype) {
sv = startWithStateValueCreate(Running, FST_INT, &val); sv = startWithStateValueCreate(Running, FST_INT, &val);
ctx->stdata = (void *)sv; ctx->stdata = (void *)sv;
} else if (atype == AUTMMATION_MATCH) { } else if (atype == AUTMMATION_MATCH) {
} else { } else {
// add more search type // add more search type
} }
char* src = (char *)data; char * src = (char *)data;
size_t len = strlen(src); 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); memcpy(dst, src, len);
dst[len] = 0; dst[len] = 0;

288
source/libs/index/src/index_fst_common.c
View File

@ -12,8 +12,7 @@
* You should have received a copy of the GNU Affero General Public License * 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/>. * along with this program. If not, see <http://www.gnu.org/licenses/>.
*/ */
#include "index_fst_common.h"
#include "tutil.h"
const uint8_t COMMON_INPUTS[] = { const uint8_t COMMON_INPUTS[] = {
84, // '\x00' 84, // '\x00'
@ -275,33 +274,260 @@ const uint8_t COMMON_INPUTS[] = {
}; };
const char COMMON_INPUTS_INV[] = { const char COMMON_INPUTS_INV[] = {
't', 'e', '/', 'o', 'a', 's', 'r', 'i', 'p', 'c', 'n', 'w', 't',
'.', 'h', 'l', 'm', '-', 'd', 'u', '0', '1', '2', 'g', '=', 'e',
':', 'b', 'f', '3', 'y', '5', '&', '_', '4', 'v', '9', '6', '/',
'7', '8', 'k', '%', '?', 'x', 'C', 'D', 'A', 'S', 'F', 'I', 'o',
'B', 'E', 'j', 'P', 'T', 'z', 'R', 'N', 'M', '+', 'L', 'O', 'a',
'q', 'H', 'G', 'W', 'U', 'V', ',', 'Y', 'K', 'J', 'Z', 'X', 's',
'Q', ';', ')', '(', '~', '[', ']', '$', '!', '\'', '*', '@', 'r',
'\x00', '\x01', '\x02', '\x03', '\x04', '\x05', '\x06', '\x07', 'i',
'\x08', '\t', '\n', '\x0b', '\x0c', '\r', '\x0e', '\x0f', '\x10', 'p',
'\x11', '\x12', '\x13', '\x14', '\x15', '\x16', '\x17', '\x18', 'c',
'\x19', '\x1a', '\x1b', '\x1c', '\x1d', '\x1e', '\x1f', ' ', '"', 'n',
'#', '<', '>', '\\', '^', '`', '{', '|', '}','\x7f','\x80', 'w',
'\x81', '\x82', '\x83', '\x84', '\x85', '\x86', '\x87', '\x88', '.',
'\x89', '\x8a', '\x8b', '\x8c', '\x8d', '\x8e', '\x8f', '\x90', 'h',
'\x91', '\x92', '\x93', '\x94', '\x95', '\x96', '\x97', '\x98', 'l',
'\x99', '\x9a', '\x9b', '\x9c', '\x9d', '\x9e', '\x9f', '\xa0', 'm',
'\xa1', '\xa2', '\xa3', '\xa4', '\xa5', '\xa6', '\xa7', '\xa8', '-',
'\xa9', '\xaa', '\xab', '\xac', '\xad', '\xae', '\xaf', '\xb0', 'd',
'\xb1', '\xb2', '\xb3', '\xb4', '\xb5', '\xb6', '\xb7', '\xb8', 'u',
'\xb9', '\xba', '\xbb', '\xbc', '\xbd', '\xbe', '\xbf', '\xc0', '0',
'\xc1', '\xc2', '\xc3', '\xc4', '\xc5', '\xc6', '\xc7', '\xc8', '1',
'\xc9', '\xca', '\xcb', '\xcc', '\xcd', '\xce', '\xcf', '\xd0', '2',
'\xd1', '\xd2', '\xd3', '\xd4', '\xd5', '\xd6', '\xd7', '\xd8', 'g',
'\xd9', '\xda', '\xdb', '\xdc', '\xdd', '\xde', '\xdf', '\xe0', '=',
'\xe1', '\xe2', '\xe3', '\xe4', '\xe5', '\xe6', '\xe7', '\xe8', ':',
'\xe9', '\xea', '\xeb', '\xec', '\xed', '\xee', '\xef', '\xf0', 'b',
'\xf1', '\xf2', '\xf3', '\xf4', '\xf5', '\xf6', '\xf7', '\xf8', 'f',
'\xf9', '\xfa', '\xfb', '\xfc', '\xfd', '\xfe', '\xff', '3',
'y',
'5',
'&',
'_',
'4',
'v',
'9',
'6',
'7',
'8',
'k',
'%',
'?',
'x',
'C',
'D',
'A',
'S',
'F',
'I',
'B',
'E',
'j',
'P',
'T',
'z',
'R',
'N',
'M',
'+',
'L',
'O',
'q',
'H',
'G',
'W',
'U',
'V',
',',
'Y',
'K',
'J',
'Z',
'X',
'Q',
';',
')',
'(',
'~',
'[',
']',
'$',
'!',
'\'',
'*',
'@',
'\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',
}; };

46
source/libs/index/src/index_fst_counting_writer.c
View File

@ -12,10 +12,10 @@
* You should have received a copy of the GNU Affero General Public License * 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/>. * along with this program. If not, see <http://www.gnu.org/licenses/>.
*/ */
#include "tutil.h" #include "index_fst_counting_writer.h"
#include "indexInt.h" #include "indexInt.h"
#include "index_fst_util.h" #include "index_fst_util.h"
#include "index_fst_counting_writer.h" #include "tutil.h"
static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) { static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
if (ctx->offset + len > ctx->limit) { if (ctx->offset + len > ctx->limit) {
@ -25,7 +25,7 @@ static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
if (ctx->type == TFile) { if (ctx->type == TFile) {
assert(len == tfWrite(ctx->file.fd, buf, len)); assert(len == tfWrite(ctx->file.fd, buf, len));
} else { } else {
memcpy(ctx->mem.buf+ ctx->offset, buf, len); memcpy(ctx->mem.buf + ctx->offset, buf, len);
} }
ctx->offset += len; ctx->offset += len;
return len; return len;
@ -43,17 +43,19 @@ static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
} }
static int writeCtxDoFlush(WriterCtx *ctx) { static int writeCtxDoFlush(WriterCtx *ctx) {
if (ctx->type == TFile) { if (ctx->type == TFile) {
//tfFsync(ctx->fd); // tfFsync(ctx->fd);
//tfFlush(ctx->file.fd); // tfFlush(ctx->file.fd);
} else { } else {
// do nothing // do nothing
} }
return 1; return 1;
} }
WriterCtx* writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity) { WriterCtx *writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity) {
WriterCtx *ctx = calloc(1, sizeof(WriterCtx)); WriterCtx *ctx = calloc(1, sizeof(WriterCtx));
if (ctx == NULL) { return NULL; } if (ctx == NULL) {
return NULL;
}
ctx->type = type; ctx->type = type;
if (ctx->type == TFile) { if (ctx->type == TFile) {
@ -81,7 +83,9 @@ WriterCtx* writerCtxCreate(WriterType type, const char *path, bool readOnly, int
return ctx; return ctx;
END: END:
if (ctx->type == TMemory) { free(ctx->mem.buf); } if (ctx->type == TMemory) {
free(ctx->mem.buf);
}
free(ctx); free(ctx);
} }
void writerCtxDestroy(WriterCtx *ctx) { void writerCtxDestroy(WriterCtx *ctx) {
@ -93,10 +97,11 @@ void writerCtxDestroy(WriterCtx *ctx) {
free(ctx); free(ctx);
} }
FstCountingWriter *fstCountingWriterCreate(void *wrt) { FstCountingWriter *fstCountingWriterCreate(void *wrt) {
FstCountingWriter *cw = calloc(1, sizeof(FstCountingWriter)); FstCountingWriter *cw = calloc(1, sizeof(FstCountingWriter));
if (cw == NULL) { return NULL; } if (cw == NULL) {
return NULL;
}
cw->wrt = wrt; cw->wrt = wrt;
//(void *)(writerCtxCreate(TFile, readOnly)); //(void *)(writerCtxCreate(TFile, readOnly));
@ -105,12 +110,14 @@ FstCountingWriter *fstCountingWriterCreate(void *wrt) {
void fstCountingWriterDestroy(FstCountingWriter *cw) { void fstCountingWriterDestroy(FstCountingWriter *cw) {
// free wrt object: close fd or free mem // free wrt object: close fd or free mem
fstCountingWriterFlush(cw); fstCountingWriterFlush(cw);
//writerCtxDestroy((WriterCtx *)(cw->wrt)); // writerCtxDestroy((WriterCtx *)(cw->wrt));
free(cw); free(cw);
} }
int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t len) { int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t len) {
if (write == NULL) { return 0; } if (write == NULL) {
return 0;
}
// update checksum // update checksum
// write data to file/socket or mem // write data to file/socket or mem
WriterCtx *ctx = write->wrt; WriterCtx *ctx = write->wrt;
@ -121,21 +128,20 @@ int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t len)
return len; return len;
} }
int fstCountingWriterRead(FstCountingWriter *write, uint8_t *buf, uint32_t len) { int fstCountingWriterRead(FstCountingWriter *write, uint8_t *buf, uint32_t len) {
if (write == NULL) { return 0; } if (write == NULL) {
return 0;
}
WriterCtx *ctx = write->wrt; WriterCtx *ctx = write->wrt;
int nRead = ctx->read(ctx, buf, len); int nRead = ctx->read(ctx, buf, len);
//assert(nRead == len); // assert(nRead == len);
return nRead; return nRead;
} }
uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write) { uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write) { return 0; }
return 0;
}
int fstCountingWriterFlush(FstCountingWriter *write) { int fstCountingWriterFlush(FstCountingWriter *write) {
WriterCtx *ctx = write->wrt; WriterCtx *ctx = write->wrt;
ctx->flush(ctx); ctx->flush(ctx);
//write->wtr->flush // write->wtr->flush
return 1; return 1;
} }
@ -156,5 +162,3 @@ uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n) {
fstCountingWriterPackUintIn(writer, n, nBytes); fstCountingWriterPackUintIn(writer, n, nBytes);
return nBytes; return nBytes;
} }

69
source/libs/index/src/index_fst_node.c
View File

@ -22,23 +22,27 @@ FstBuilderNode *fstBuilderNodeDefault() {
return bn; return bn;
} }
void fstBuilderNodeDestroy(FstBuilderNode *node) { void fstBuilderNodeDestroy(FstBuilderNode *node) {
if (node == NULL) { return; } if (node == NULL) {
return;
}
taosArrayDestroy(node->trans); taosArrayDestroy(node->trans);
free(node); free(node);
} }
bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2) { bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2) {
if (n1 == n2) { return true; } if (n1 == n2) {
if (n1 == NULL || n2 == NULL ) { return true;
}
if (n1 == NULL || n2 == NULL) {
return false; return false;
} }
if (n1->isFinal != n2->isFinal || n1->finalOutput != n2->finalOutput) { if (n1->isFinal != n2->isFinal || n1->finalOutput != n2->finalOutput) {
return false; return false;
} }
size_t s1 = n1->trans? taosArrayGetSize(n1->trans): 0; size_t s1 = n1->trans ? taosArrayGetSize(n1->trans) : 0;
size_t s2 = n2->trans? taosArrayGetSize(n2->trans): 0; size_t s2 = n2->trans ? taosArrayGetSize(n2->trans) : 0;
if (s1 != s2) { if (s1 != s2) {
return false; return false;
} }
@ -54,7 +58,9 @@ bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2) {
} }
FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src) { FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src) {
FstBuilderNode *node = malloc(sizeof(FstBuilderNode)); FstBuilderNode *node = malloc(sizeof(FstBuilderNode));
if (node == NULL) { return NULL; } if (node == NULL) {
return NULL;
}
// //
size_t sz = taosArrayGetSize(src->trans); size_t sz = taosArrayGetSize(src->trans);
@ -69,16 +75,17 @@ FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src) {
node->isFinal = src->isFinal; node->isFinal = src->isFinal;
node->finalOutput = src->finalOutput; node->finalOutput = src->finalOutput;
return node; return node;
} }
// not destroy src, User's bussiness // not destroy src, User's bussiness
void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src) { void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src) {
if (dst == NULL || src == NULL) { return; } if (dst == NULL || src == NULL) {
return;
}
dst->isFinal = src->isFinal; dst->isFinal = src->isFinal;
dst->finalOutput = src->finalOutput; dst->finalOutput = src->finalOutput;
//release free avoid mem leak // release free avoid mem leak
taosArrayDestroy(dst->trans); taosArrayDestroy(dst->trans);
size_t sz = taosArrayGetSize(src->trans); size_t sz = taosArrayGetSize(src->trans);
dst->trans = taosArrayInit(sz, sizeof(FstTransition)); dst->trans = taosArrayInit(sz, sizeof(FstTransition));
@ -88,28 +95,26 @@ void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src) {
} }
} }
// bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr
// startAddr) {
//bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr startAddr) { // size_t sz = taosArrayGetSize(b->trans);
// assert(sz < 256);
//size_t sz = taosArrayGetSize(b->trans); // if (FST_BUILDER_NODE_IS_FINAL(b)
//assert(sz < 256); // && FST_BUILDER_NODE_TRANS_ISEMPTY(b)
//if (FST_BUILDER_NODE_IS_FINAL(b) // && FST_BUILDER_NODE_FINALOUTPUT_ISZERO(b)) {
// && FST_BUILDER_NODE_TRANS_ISEMPTY(b) // return true;
// && FST_BUILDER_NODE_FINALOUTPUT_ISZERO(b)) { //} else if (sz != 1 || b->isFinal) {
// return true; // // AnyTrans->Compile(w, addr, node);
//} else if (sz != 1 || b->isFinal) { //} else {
// // AnyTrans->Compile(w, addr, node); // FstTransition *tran = taosArrayGet(b->trans, 0);
//} else { // if (tran->addr == lastAddr && tran->out == 0) {
// FstTransition *tran = taosArrayGet(b->trans, 0); // //OneTransNext::compile(w, lastAddr, tran->inp);
// if (tran->addr == lastAddr && tran->out == 0) { // return true;
// //OneTransNext::compile(w, lastAddr, tran->inp); // } else {
// return true; // //OneTrans::Compile(w, lastAddr, *tran);
// } else { // return true;
// //OneTrans::Compile(w, lastAddr, *tran); // }
// return true; //}
// } // return true;
//}
//return true;
//} //}

44
source/libs/index/src/index_fst_registry.c
View File

@ -15,9 +15,8 @@
#include "index_fst_registry.h" #include "index_fst_registry.h"
uint64_t fstRegistryHash(FstRegistry *registry, FstBuilderNode *bNode) { uint64_t fstRegistryHash(FstRegistry *registry, FstBuilderNode *bNode) {
//TODO(yihaoDeng): refactor later // TODO(yihaoDeng): refactor later
const uint64_t FNV_PRIME = 1099511628211; const uint64_t FNV_PRIME = 1099511628211;
uint64_t h = 14695981039346656037u; uint64_t h = 14695981039346656037u;
@ -29,14 +28,15 @@ uint64_t fstRegistryHash(FstRegistry *registry, FstBuilderNode *bNode) {
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->inp)) * FNV_PRIME;
h = (h ^ (uint64_t)(trn->out)) * FNV_PRIME; h = (h ^ (uint64_t)(trn->out)) * FNV_PRIME;
h = (h ^ (uint64_t)(trn->addr))* FNV_PRIME; h = (h ^ (uint64_t)(trn->addr)) * FNV_PRIME;
} }
return h %(registry->tableSize); 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); size_t sz = taosArrayGetSize(arr);
if (a >= sz || b >= sz) { return; } if (a >= sz || b >= sz) {
return;
}
FstRegistryCell *cell1 = (FstRegistryCell *)taosArrayGet(arr, a); FstRegistryCell *cell1 = (FstRegistryCell *)taosArrayGet(arr, a);
FstRegistryCell *cell2 = (FstRegistryCell *)taosArrayGet(arr, b); FstRegistryCell *cell2 = (FstRegistryCell *)taosArrayGet(arr, b);
@ -53,24 +53,28 @@ static void fstRegistryCellSwap(SArray *arr, uint32_t a, uint32_t b) {
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); size_t sz = taosArrayGetSize(arr);
if (start >= sz && end >= sz) {return; } if (start >= sz && end >= sz) {
return;
}
assert(start >= end); assert(start >= end);
int32_t s = (int32_t)start; int32_t s = (int32_t)start;
int32_t e = (int32_t)end; int32_t e = (int32_t)end;
while(s > e) { while (s > e) {
fstRegistryCellSwap(arr, s - 1, s); fstRegistryCellSwap(arr, s - 1, s);
s -= 1; s -= 1;
} }
} }
FstRegistry* fstRegistryCreate(uint64_t tableSize, uint64_t mruSize) { FstRegistry *fstRegistryCreate(uint64_t tableSize, uint64_t mruSize) {
FstRegistry *registry = malloc(sizeof(FstRegistry)); FstRegistry *registry = malloc(sizeof(FstRegistry));
if (registry == NULL) { return NULL ;} if (registry == NULL) {
return NULL;
}
uint64_t nCells = tableSize * mruSize; uint64_t nCells = tableSize * mruSize;
SArray* tb = (SArray *)taosArrayInit(nCells, sizeof(FstRegistryCell)); SArray * tb = (SArray *)taosArrayInit(nCells, sizeof(FstRegistryCell));
if (NULL == tb) { if (NULL == tb) {
free(registry); free(registry);
return NULL; return NULL;
@ -88,7 +92,9 @@ FstRegistry* fstRegistryCreate(uint64_t tableSize, uint64_t mruSize) {
} }
void fstRegistryDestroy(FstRegistry *registry) { void fstRegistryDestroy(FstRegistry *registry) {
if (registry == NULL) { return; } if (registry == NULL) {
return;
}
SArray *tb = registry->table; SArray *tb = registry->table;
size_t sz = taosArrayGetSize(tb); size_t sz = taosArrayGetSize(tb);
@ -111,10 +117,10 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
FstRegistryEntry *entry = malloc(sizeof(FstRegistryEntry)); FstRegistryEntry *entry = malloc(sizeof(FstRegistryEntry));
if (end - start == 1) { if (end - start == 1) {
FstRegistryCell *cell = taosArrayGet(registry->table, start); FstRegistryCell *cell = taosArrayGet(registry->table, start);
//cell->isNode && // cell->isNode &&
if (cell->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell->node, bNode)) { if (cell->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell->node, bNode)) {
entry->state = FOUND; entry->state = FOUND;
entry->addr = cell->addr ; entry->addr = cell->addr;
return entry; return entry;
} else { } else {
fstBuilderNodeCloneFrom(cell->node, bNode); fstBuilderNodeCloneFrom(cell->node, bNode);
@ -136,7 +142,7 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
fstRegistryCellSwap(registry->table, start, start + 1); fstRegistryCellSwap(registry->table, start, start + 1);
return entry; return entry;
} }
//clone from bNode, refactor later // clone from bNode, refactor later
fstBuilderNodeCloneFrom(cell2->node, bNode); fstBuilderNodeCloneFrom(cell2->node, bNode);
fstRegistryCellSwap(registry->table, start, start + 1); fstRegistryCellSwap(registry->table, start, start + 1);
@ -157,7 +163,7 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
if (i >= end) { if (i >= end) {
uint64_t last = end - 1; uint64_t last = end - 1;
FstRegistryCell *cell = (FstRegistryCell *)taosArrayGet(registry->table, last); FstRegistryCell *cell = (FstRegistryCell *)taosArrayGet(registry->table, last);
//clone from bNode, refactor later // clone from bNode, refactor later
fstBuilderNodeCloneFrom(cell->node, bNode); fstBuilderNodeCloneFrom(cell->node, bNode);
fstRegistryCellPromote(registry->table, last, start); fstRegistryCellPromote(registry->table, last, start);
@ -168,8 +174,4 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
} }
return entry; return entry;
} }
void fstRegistryEntryDestroy(FstRegistryEntry *entry) { void fstRegistryEntryDestroy(FstRegistryEntry *entry) { free(entry); }
free(entry);
}

57
source/libs/index/src/index_fst_util.c
View File

@ -15,9 +15,7 @@
#include "index_fst_util.h" #include "index_fst_util.h"
#include "index_fst_common.h" #include "index_fst_common.h"
// A sentinel value used to indicate an empty final state
//A sentinel value used to indicate an empty final state
const CompiledAddr EMPTY_ADDRESS = 0; const CompiledAddr EMPTY_ADDRESS = 0;
/// A sentinel value used to indicate an invalid state. /// A sentinel value used to indicate an invalid state.
const CompiledAddr NONE_ADDRESS = 1; const CompiledAddr NONE_ADDRESS = 1;
@ -31,21 +29,18 @@ const uint64_t VERSION = 3;
const uint64_t TRANS_INDEX_THRESHOLD = 32; const uint64_t TRANS_INDEX_THRESHOLD = 32;
// uint8_t commonInput(uint8_t idx) {
//uint8_t commonInput(uint8_t idx) {
// if (idx == 0) { return -1; } // if (idx == 0) { return -1; }
// else { // else {
// return COMMON_INPUTS_INV[idx - 1]; // return COMMON_INPUTS_INV[idx - 1];
// } // }
//} //}
// //
//uint8_t commonIdx(uint8_t v, uint8_t max) { // uint8_t commonIdx(uint8_t v, uint8_t max) {
// uint8_t v = ((uint16_t)tCOMMON_INPUTS[v] + 1)%256; // uint8_t v = ((uint16_t)tCOMMON_INPUTS[v] + 1)%256;
// return v > max ? 0: v; // return v > max ? 0: v;
//} //}
uint8_t packSize(uint64_t n) { uint8_t packSize(uint64_t n) {
if (n < (1u << 8)) { if (n < (1u << 8)) {
return 1; return 1;
@ -107,15 +102,14 @@ FstSlice fstSliceCreate(uint8_t *data, uint64_t len) {
FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end) { FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end) {
FstString *str = s->str; FstString *str = s->str;
str->ref++; str->ref++;
//uint8_t *buf = fstSliceData(s, &alen); // uint8_t *buf = fstSliceData(s, &alen);
//start = buf + start - (buf - s->start); // start = buf + start - (buf - s->start);
//end = buf + end - (buf - s->start); // end = buf + end - (buf - s->start);
FstSlice t = {.str = str, .start = start + s->start, .end = end + s->start}; FstSlice t = {.str = str, .start = start + s->start, .end = end + s->start};
return t; 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 tlen = end - start + 1;
int32_t slen; int32_t slen;
uint8_t *data = fstSliceData(s, &slen); uint8_t *data = fstSliceData(s, &slen);
@ -135,9 +129,7 @@ FstSlice fstSliceDeepCopy(FstSlice *s, int32_t start, int32_t end) {
ans.end = tlen - 1; ans.end = tlen - 1;
return ans; return ans;
} }
bool fstSliceIsEmpty(FstSlice *s) { bool fstSliceIsEmpty(FstSlice *s) { return s->str == NULL || s->str->len == 0 || s->start < 0 || s->end < 0; }
return s->str == NULL || s->str->len == 0 || s->start < 0 || s->end < 0;
}
uint8_t *fstSliceData(FstSlice *s, int32_t *size) { uint8_t *fstSliceData(FstSlice *s, int32_t *size) {
FstString *str = s->str; FstString *str = s->str;
@ -165,27 +157,32 @@ int fstSliceCompare(FstSlice *a, FstSlice *b) {
for (i = 0, j = 0; i < alen && j < blen; i++, j++) { for (i = 0, j = 0; i < alen && j < blen; i++, j++) {
uint8_t x = aBuf[i]; uint8_t x = aBuf[i];
uint8_t y = bBuf[j]; uint8_t y = bBuf[j];
if (x == y) { continue;} if (x == y) {
else if (x < y) { return -1; } continue;
else { return 1; }; } else if (x < y) {
return -1;
} else {
return 1;
};
}
if (i < alen) {
return 1;
} else if (j < blen) {
return -1;
} else {
return 0;
} }
if (i < alen) { return 1; }
else if (j < blen) { return -1; }
else { return 0; }
} }
//FstStack* fstStackCreate(size_t elemSize, StackFreeElem freeFn) { // FstStack* fstStackCreate(size_t elemSize, StackFreeElem freeFn) {
// FstStack *s = calloc(1, sizeof(FstStack)); // FstStack *s = calloc(1, sizeof(FstStack));
// if (s == NULL) { return NULL; } // if (s == NULL) { return NULL; }
// s-> // s->
// s->freeFn // s->freeFn
// //
//} //}
//void *fstStackPush(FstStack *s, void *elem); // void *fstStackPush(FstStack *s, void *elem);
//void *fstStackTop(FstStack *s); // void *fstStackTop(FstStack *s);
//size_t fstStackLen(FstStack *s); // size_t fstStackLen(FstStack *s);
//void *fstStackGetAt(FstStack *s, size_t i); // void *fstStackGetAt(FstStack *s, size_t i);
//void fstStackDestory(FstStack *); // void fstStackDestory(FstStack *);

101
source/libs/index/src/index_tfile.c
View File

@ -16,18 +16,18 @@
//#include <sys/types.h> //#include <sys/types.h>
//#include <dirent.h> //#include <dirent.h>
#include "index_tfile.h" #include "index_tfile.h"
#include "index.h"
#include "index_fst.h" #include "index_fst.h"
#include "index_fst_counting_writer.h"
#include "index_util.h" #include "index_util.h"
#include "taosdef.h" #include "taosdef.h"
#include "index.h"
#include "index_fst_counting_writer.h"
static FORCE_INLINE int tfileReadLoadHeader(TFileReader *reader) {
static FORCE_INLINE int tfileLoadHeader(WriterCtx *ctx, TFileReadHeader *header) { // TODO simple tfile header later
//TODO simple tfile header later
char buf[TFILE_HADER_PRE_SIZE]; char buf[TFILE_HADER_PRE_SIZE];
char *p = buf; char * p = buf;
int64_t nread = ctx->read(ctx, buf, TFILE_HADER_PRE_SIZE); TFileReadHeader *header = &reader->header;
int64_t nread = reader->ctx->read(reader->ctx, buf, TFILE_HADER_PRE_SIZE);
assert(nread == TFILE_HADER_PRE_SIZE); assert(nread == TFILE_HADER_PRE_SIZE);
memcpy(&header->suid, p, sizeof(header->suid)); memcpy(&header->suid, p, sizeof(header->suid));
@ -39,20 +39,22 @@ static FORCE_INLINE int tfileLoadHeader(WriterCtx *ctx, TFileReadHeader *header)
int32_t colLen = 0; int32_t colLen = 0;
memcpy(&colLen, p, sizeof(colLen)); memcpy(&colLen, p, sizeof(colLen));
assert(colLen < sizeof(header->colName)); assert(colLen < sizeof(header->colName));
nread = ctx->read(ctx, header->colName, colLen); nread = reader->ctx->read(reader->ctx, header->colName, colLen);
assert(nread == colLen); assert(nread == colLen);
nread = ctx->read(ctx, &header->colType, sizeof(header->colType)); nread = reader->ctx->read(reader->ctx, &header->colType, sizeof(header->colType));
return 0; return 0;
}; };
static int tfileGetFileList(const char *path, SArray *result) { static int tfileGetFileList(const char *path, SArray *result) {
DIR *dir = opendir(path); DIR *dir = opendir(path);
if (NULL == dir) { return -1; } if (NULL == dir) {
return -1;
}
struct dirent *entry; struct dirent *entry;
while ((entry = readdir(dir)) != NULL) { while ((entry = readdir(dir)) != NULL) {
size_t len = strlen(entry->d_name); 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); memcpy(buf, entry->d_name, len);
taosArrayPush(result, &buf); taosArrayPush(result, &buf);
} }
@ -90,7 +92,9 @@ static void tfileSerialCacheKey(TFileCacheKey *key, char *buf) {
TFileCache *tfileCacheCreate(const char *path) { TFileCache *tfileCacheCreate(const char *path) {
TFileCache *tcache = calloc(1, sizeof(TFileCache)); TFileCache *tcache = calloc(1, sizeof(TFileCache));
if (tcache == NULL) { return NULL; } if (tcache == NULL) {
return NULL;
}
tcache->tableCache = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK); tcache->tableCache = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
tcache->capacity = 64; tcache->capacity = 64;
@ -99,7 +103,7 @@ TFileCache *tfileCacheCreate(const char *path) {
tfileGetFileList(path, files); tfileGetFileList(path, files);
taosArraySort(files, tfileCompare); taosArraySort(files, tfileCompare);
for (size_t i = 0; i < taosArrayGetSize(files); i++) { for (size_t i = 0; i < taosArrayGetSize(files); i++) {
char *file = taosArrayGetP(files, i); char * file = taosArrayGetP(files, i);
uint64_t suid; uint64_t suid;
int colId, version; int colId, version;
if (0 != tfileParseFileName(file, &suid, &colId, &version)) { if (0 != tfileParseFileName(file, &suid, &colId, &version)) {
@ -107,29 +111,41 @@ TFileCache *tfileCacheCreate(const char *path) {
continue; continue;
} }
TFileReader *reader = calloc(1, sizeof(TFileReader)); WriterCtx *wc = writerCtxCreate(TFile, file, true, 1024 * 64);
reader->ctx = writerCtxCreate(TFile, file, true, 1024 * 64); if (wc == NULL) {
if (reader->ctx == NULL) {
TFileReaderDestroy(reader);
indexError("failed to open index: %s", file); indexError("failed to open index: %s", file);
goto End; goto End;
} }
TFileReadHeader header = {0}; TFileReader *reader = tfileReaderCreate(wc);
if (0 != tfileLoadHeader(reader->ctx, &header)) { if (0 != tfileReadLoadHeader(reader)) {
TFileReaderDestroy(reader); TFileReaderDestroy(reader);
indexError("failed to load index header, index Id: %s", file); indexError("failed to load index header, index Id: %s", file);
goto End;
} }
} }
taosArrayDestroyEx(files, tfileDestroyFileName); taosArrayDestroyEx(files, tfileDestroyFileName);
return tcache; return tcache;
End: End:
tfileCacheDestroy(tcache);
taosArrayDestroyEx(files, tfileDestroyFileName); taosArrayDestroyEx(files, tfileDestroyFileName);
return NULL; return NULL;
} }
void tfileCacheDestroy(TFileCache *tcache) { void tfileCacheDestroy(TFileCache *tcache) {
if (tcache == NULL) {
return;
}
// free table cache
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);
reader = taosHashIterate(tcache->tableCache, reader);
}
taosHashCleanup(tcache->tableCache);
free(tcache); free(tcache);
} }
TFileReader *tfileCacheGet(TFileCache *tcache, TFileCacheKey *key) { TFileReader *tfileCacheGet(TFileCache *tcache, TFileCacheKey *key) {
@ -145,56 +161,47 @@ void tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader)
return; return;
} }
TFileReader *tfileReaderCreate(WriterCtx *ctx) {
TFileReader* tfileReaderCreate() { TFileReader *reader = calloc(1, sizeof(TFileReader));
if (reader == NULL) {
return NULL;
}
reader->ctx = ctx;
// T_REF_INC(reader);
return reader;
} }
void TFileReaderDestroy(TFileReader *reader) { void TFileReaderDestroy(TFileReader *reader) {
if (reader == NULL) { return; } if (reader == NULL) {
return;
}
// T_REF_INC(reader);
writerCtxDestroy(reader->ctx); writerCtxDestroy(reader->ctx);
free(reader); free(reader);
} }
TFileWriter *tfileWriterCreate(const char *suid, const char *colName); TFileWriter *tfileWriterCreate(const char *suid, const char *colName);
void tfileWriterDestroy(TFileWriter *tw); void tfileWriterDestroy(TFileWriter *tw);
IndexTFile *indexTFileCreate(const char *path) { IndexTFile *indexTFileCreate(const char *path) {
IndexTFile *tfile = calloc(1, sizeof(IndexTFile)); IndexTFile *tfile = calloc(1, sizeof(IndexTFile));
tfile->cache = tfileCacheCreate(path); tfile->cache = tfileCacheCreate(path);
return tfile; return tfile;
} }
void IndexTFileDestroy(IndexTFile *tfile) { void IndexTFileDestroy(IndexTFile *tfile) { free(tfile); }
free(tfile);
}
int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result) { int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result) {
IndexTFile *pTfile = (IndexTFile *)tfile; IndexTFile *pTfile = (IndexTFile *)tfile;
SIndexTerm *term = query->term; SIndexTerm * term = query->term;
TFileCacheKey key = {.suid = term->suid, TFileCacheKey key = {
.colType = term->colType, .suid = term->suid, .colType = term->colType, .version = 0, .colName = term->colName, .nColName = term->nColName};
.version = 0,
.colName = term->colName,
.nColName= term->nColName};
TFileReader *reader = tfileCacheGet(pTfile->cache, &key); TFileReader *reader = tfileCacheGet(pTfile->cache, &key);
return 0; return 0;
} }
int indexTFilePut(void *tfile, SIndexTerm *term, uint64_t uid) { int indexTFilePut(void *tfile, SIndexTerm *term, uint64_t uid) {
TFileWriterOpt wOpt = {.suid = term->suid, TFileWriterOpt wOpt = {
.colType = term->colType, .suid = term->suid, .colType = term->colType, .colName = term->colName, .nColName = term->nColName, .version = 1};
.colName = term->colName,
.nColName= term->nColName,
.version = 1};
return 0; return 0;
} }