Merge pull request #8918 from taosdata/origin/3.0/fst

Origin/3.0/fst

source/libs/index/inc/index_fst.h

@@ -16,6 +16,9 @@
 #ifndef __INDEX_FST_H__
 #define __INDEX_FST_H__
 
+#ifdef __cplusplus
+extern "C" {
+#endif
 
 #include "tarray.h"
 #include "index_fst_util.h"
@@ -34,6 +37,7 @@ typedef struct FstRange {
 } FstRange;
 
 
+typedef enum {GE, GT, LE, LT} RangeType;
 typedef enum { OneTransNext, OneTrans, AnyTrans, EmptyFinal} State;
 
 typedef enum {Ordered, OutOfOrdered, DuplicateKey} OrderType;
@@ -93,8 +97,11 @@ typedef struct FstBuilder {
 
 
 FstBuilder *fstBuilderCreate(void *w, FstType ty);
+
+
 void fstBuilderDestroy(FstBuilder *b);
 void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in);
+bool fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in);
 OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup);
 void fstBuilderCompileFrom(FstBuilder *b, uint64_t istate);
 CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn);
@@ -169,11 +176,6 @@ uint64_t fstStateFindInput(FstState *state, FstNode *node, uint8_t b, bool *null
 
 
 
-
-
-
-  
-
 #define FST_STATE_ONE_TRNAS_NEXT(node) (node->state.state == OneTransNext) 
 #define FST_STATE_ONE_TRNAS(node) (node->state.state == OneTrans)
 #define FST_STATE_ANY_TRANS(node) (node->state.state == AnyTrans)
@@ -272,7 +274,6 @@ FstNode*     fstGetNode(Fst *fst, CompiledAddr);
 FstNode*     fstGetRoot(Fst *fst);
 FstType      fstGetType(Fst *fst); 
 CompiledAddr fstGetRootAddr(Fst *fst);
-
 Output       fstEmptyFinalOutput(Fst *fst, bool *null);
 bool         fstVerify(Fst *fst);
 
@@ -280,10 +281,6 @@ bool fstVerify(Fst *fst);
 //refactor this function 
 bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr startAddr); 
 
-
-
-
-
 typedef struct StreamState {
   FstNode   *node; 
   uint64_t  trans;
@@ -310,10 +307,30 @@ typedef struct StreamWithStateResult {
 } StreamWithStateResult;
 
 StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *state);
+void swsResultDestroy(StreamWithStateResult *result);
 
 typedef void* (*StreamCallback)(void *);
 StreamWithState *streamWithStateCreate(Fst *fst, Automation *automation, FstBoundWithData *min, FstBoundWithData *max) ;
 void streamWithStateDestroy(StreamWithState *sws);
 bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min);           
 StreamWithStateResult* streamWithStateNextWith(StreamWithState *sws, StreamCallback callback);
+
+typedef struct FstStreamBuilder {
+  Fst *fst; 
+  Automation *aut;
+  FstBoundWithData *min;  
+  FstBoundWithData *max;
+} FstStreamBuilder;
+
+FstStreamBuilder *fstStreamBuilderCreate(Fst *fst, Automation *aut); 
+// set up bound range
+// refator, simple code by marco 
+
+FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type);
+
+
+#ifdef __cplusplus
+}
+#endif
+
 #endif

source/libs/index/inc/index_fst_automation.h

@@ -15,6 +15,10 @@
 #ifndef __INDEX_FST_AUTAOMATION_H__
 #define __INDEX_FST_AUTAOMATION_H__
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 typedef struct AutomationCtx AutomationCtx;
 
 typedef struct StartWith {
@@ -42,6 +46,8 @@ typedef struct Automation {
   void *data;
 } Automation; 
 
-
+#ifdef __cplusplus
+}
+#endif
 
 #endif

source/libs/index/inc/index_fst_common.h

@@ -1,7 +1,16 @@
 #ifndef __INDEX_FST_COMM_H__
 #define __INDEX_FST_COMM_H__
 
+
 extern const uint8_t COMMON_INPUTS[];
 extern char const COMMON_INPUTS_INV[];
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifdef __cplusplus
+}
+#endif
+
 #endif

source/libs/index/inc/index_fst_counting_writer.h

@@ -16,6 +16,38 @@
 #ifndef __INDEX_FST_COUNTING_WRITER_H__
 #define __INDEX_FST_COUNTING_WRITER_H__
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "tfile.h"
+
+
+#define DefaultMem  1024*1024
+
+static char tmpFile[] = "/tmp/index";
+typedef enum WriterType {TMemory, TFile} WriterType;   
+
+typedef struct WriterCtx {
+  int (*write)(struct WriterCtx *ctx, uint8_t *buf, int len);
+  int (*read)(struct WriterCtx *ctx, uint8_t *buf, int len);
+  int (*flush)(struct WriterCtx *ctx);
+  WriterType type; 
+  union {
+    int fd;
+    void *mem;
+  };
+  int32_t offset;
+  int32_t limit;
+} WriterCtx;
+
+static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len);
+static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len);
+static int writeCtxDoFlush(WriterCtx *ctx);
+
+WriterCtx* writerCtxCreate(WriterType type);
+void writerCtxDestroy(WriterCtx *w);
+
 typedef uint32_t CheckSummer;
 
 
@@ -25,7 +57,7 @@ typedef struct FstCountingWriter {
   CheckSummer summer;  
 } FstCountingWriter;
 
-uint64_t fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t bufLen); 
+int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t bufLen); 
 
 int fstCountingWriterFlush(FstCountingWriter *write);
 
@@ -44,6 +76,10 @@ uint8_t fstCountingWriterPackUint(FstCountingWriter *writer, uint64_t n);
 #define FST_WRITER_INTER_WRITER(writer) (writer->wtr)
 #define FST_WRITE_CHECK_SUMMER(writer) (writer->summer)
 
+#ifdef __cplusplus
+}
+#endif
+
 #endif
 
 

source/libs/index/inc/index_fst_node.h

@@ -16,6 +16,10 @@
 #ifndef __INDEX_FST_NODE_H__
 #define __INDEX_FST_NODE_H__
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 #include "index_fst_util.h"
 #include "index_fst_counting_writer.h"
 
@@ -42,7 +46,12 @@ FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src);
 void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src);
 
 //bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr startAddr); 
+bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2); 
 
 void fstBuilderNodeDestroy(FstBuilderNode *node);
 
+#ifdef __cplusplus
+}
+#endif
+
 #endif

source/libs/index/inc/index_fst_registry.h

@@ -15,6 +15,10 @@
 #ifndef __FST_REGISTRY_H__
 #define __FST_REGISTRY_H__
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 #include "index_fst_util.h"
 #include "tarray.h"
 #include "index_fst_node.h"
@@ -59,4 +63,8 @@ void fstRegistryDestroy(FstRegistry *registry);
 FstRegistryEntry* fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNode);
 void fstRegistryEntryDestroy(FstRegistryEntry *entry);
 
+#ifdef __cplusplus
+}
+#endif
+
 #endif

source/libs/index/inc/index_fst_util.h

@@ -17,6 +17,10 @@
 #ifndef __INDEX_FST_UTIL_H__
 #define __INDEX_FST_UTIL_H__
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 #include "tarray.h"
 #include "index_fst_common.h"
 
@@ -67,20 +71,30 @@ uint8_t      packDeltaSize(CompiledAddr nodeAddr, CompiledAddr transAddr);
 CompiledAddr unpackDelta(char *data, uint64_t len, uint64_t nodeAddr);
 
 
+typedef struct FstString {
+  uint8_t *data; 
+  uint32_t len;
+  int32_t ref;
+} FstString;
 
 typedef struct FstSlice {
-  uint8_t *data; 
+  FstString   *str;     
-  uint64_t dLen;
   int32_t start; 
   int32_t end; 
 } FstSlice;
 
-FstSlice fstSliceCopy(FstSlice *slice, int32_t start, int32_t end);
+FstSlice fstSliceCreate(uint8_t *data, uint64_t len);
-FstSlice fstSliceCreate(uint8_t *data, uint64_t dLen);
+FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end);
-bool fstSliceEmpty(FstSlice *slice);
+FstSlice fstSliceDeepCopy(FstSlice *s, int32_t start, int32_t end);
-int fstSliceCompare(FstSlice *a, FstSlice *b);
+bool fstSliceIsEmpty(FstSlice *s);
+int  fstSliceCompare(FstSlice *s1, FstSlice *s2);
+void fstSliceDestroy(FstSlice *s); 
+uint8_t *fstSliceData(FstSlice *s, int32_t *sz); 
 
-#define FST_SLICE_LEN(s) ((s)->end - (s)->start + 1)
+#define FST_SLICE_LEN(s) (s->end - s->start + 1)
 
+#ifdef __cplusplus
+}
+#endif
 
 #endif

source/libs/index/src/index_fst.c

@@ -52,7 +52,7 @@ void fstUnFinishedNodesPushEmpty(FstUnFinishedNodes *nodes, bool isFinal) {
   FstBuilderNode *node = malloc(sizeof(FstBuilderNode));
   node->isFinal     = isFinal;
   node->finalOutput = 0;
-  node->trans       = NULL;
+  node->trans       = taosArrayInit(16, sizeof(FstTransition));
 
   FstBuilderNodeUnfinished un = {.node = node, .last = NULL}; 
   taosArrayPush(nodes->stack, &un);
@@ -92,7 +92,7 @@ void fstUnFinishedNodesTopLastFreeze(FstUnFinishedNodes *nodes, CompiledAddr add
 }
 void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *nodes, FstSlice bs, Output out) {
   FstSlice *s = &bs;
-  if (s->data == NULL || s->dLen == 0 || s->start > s->end) {
+  if (fstSliceIsEmpty(s)) {
     return;
   }
   size_t sz = taosArrayGetSize(nodes->stack) - 1; 
@@ -104,18 +104,20 @@ void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *nodes, FstSlice bs, Output
   //FstLastTransition *trn = malloc(sizeof(FstLastTransition)); 
   //trn->inp = s->data[s->start]; 
   //trn->out = out;
-  un->last = fstLastTransitionCreate(s->data[s->start], out); 
+  int32_t len = 0;
+  uint8_t *data = fstSliceData(s, &len);
+  un->last = fstLastTransitionCreate(data[0], out); 
 
-  for (uint64_t i = s->start; i <= s->end; i++) {
+  for (uint64_t i = 0; i < len; i++) {
     FstBuilderNode *n = malloc(sizeof(FstBuilderNode));
     n->isFinal     = false;
     n->finalOutput = 0;
-    n->trans       = NULL;
+    n->trans       = taosArrayInit(16, sizeof(FstTransition));
     
     //FstLastTransition *trn = malloc(sizeof(FstLastTransition)); 
     //trn->inp = s->data[i];
     //trn->out = out; 
-    FstLastTransition *trn = fstLastTransitionCreate(s->data[i], out);
+    FstLastTransition *trn = fstLastTransitionCreate(data[i], out);
 
     FstBuilderNodeUnfinished un = {.node = n, .last = trn}; 
     taosArrayPush(nodes->stack, &un); 
@@ -127,13 +129,13 @@ void fstUnFinishedNodesAddSuffix(FstUnFinishedNodes *nodes, FstSlice bs, Output
 uint64_t fstUnFinishedNodesFindCommPrefix(FstUnFinishedNodes *node, FstSlice bs) {
   FstSlice *s = &bs;
 
-  size_t lsz = (size_t)(s->end - s->start + 1);          // data len 
   size_t ssz = taosArrayGetSize(node->stack);  // stack size
-    
   uint64_t count = 0;
+  int32_t lsz;          // data len 
+  uint8_t *data = fstSliceData(s, &lsz);
   for (size_t i = 0; i < ssz && i < lsz; i++) {
     FstBuilderNodeUnfinished *un = taosArrayGet(node->stack, i); 
-    if (un->last->inp == s->data[s->start + i]) {
+    if (un->last->inp == data[i]) {
       count++;
     } else {
       break;
@@ -153,7 +155,8 @@ uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes *node,
 
     FstLastTransition *t = un->last; 
     uint64_t addPrefix = 0;  
-    if (t && t->inp == s->data[s->start + i]) {
+    uint8_t *data = fstSliceData(s, NULL);
+    if (t && t->inp == data[i]) {
       uint64_t commPrefix = MIN(t->out, *out);      
       uint64_t tAddPrefix  = t->out - commPrefix; 
       (*out) = (*out) - commPrefix; 
@@ -164,7 +167,6 @@ uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes *node,
     }
     if (addPrefix != 0) {
       fstBuilderNodeUnfinishedAddOutputPrefix(un, addPrefix);  
-
     }
   }   
   return i;
@@ -176,7 +178,9 @@ FstState fstStateCreateFrom(FstSlice* slice, CompiledAddr addr) {
   if (addr == EMPTY_ADDRESS) {
     return fs; 
   }
-  uint8_t v = slice->data[addr]; 
+  
+  uint8_t *data = fstSliceData(slice, NULL);
+  uint8_t v = data[addr]; 
   uint8_t t = (v & 0b11000000) >> 6;
   if (t == 0b11) {
     fs.state = OneTransNext;
@@ -376,7 +380,8 @@ uint8_t  fstStateInput(FstState *s, FstNode *node) {
   FstSlice *slice = &node->data;
   bool null = false;
   uint8_t inp = fstStateCommInput(s, &null);
-  return null == false ? inp : slice->data[slice->start - 1];
+  uint8_t *data = fstSliceData(slice, NULL); 
+  return null == false ? inp : data[-1];
 }
 uint8_t  fstStateInputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
   assert(s->state == AnyTrans);
@@ -388,7 +393,9 @@ uint8_t  fstStateInputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
             - fstStateTransIndexSize(s, node->version, node->nTrans) 
             - i 
             - 1; // the output size 
-  return slice->data[at];
+
+  uint8_t *data = fstSliceData(slice, NULL);
+  return data[at];
 }
 
 // trans_addr
@@ -406,7 +413,8 @@ CompiledAddr fstStateTransAddr(FstState *s, FstNode *node) {
                 - tSizes;
 
     // refactor error logic 
-    return unpackDelta(slice->data + slice->start + i, tSizes, node->end);      
+    uint8_t *data = fstSliceData(slice, NULL);
+    return unpackDelta(data +i, tSizes, node->end);      
   }  
 }
 CompiledAddr fstStateTransAddrForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
@@ -421,7 +429,8 @@ CompiledAddr fstStateTransAddrForAnyTrans(FstState *s, FstNode *node, uint64_t i
                - node->nTrans
                - (i * tSizes)
                - tSizes;
-  return unpackDelta(slice->data + slice->start + at, tSizes, node->end); 
+  uint8_t *data = fstSliceData(slice, NULL);
+  return unpackDelta(data + at, tSizes, node->end); 
 }
 
 // sizes 
@@ -434,7 +443,8 @@ PackSizes fstStateSizes(FstState *s, FstSlice *slice) {
     i = FST_SLICE_LEN(slice) - 1 - fstStateNtransLen(s) - 1;
   }
 
-  return (PackSizes)(slice->data[slice->start + i]);
+  uint8_t *data = fstSliceData(slice, NULL);
+  return (PackSizes)(*(data +i));
 }
 // Output 
 Output fstStateOutput(FstState *s, FstNode *node) {
@@ -452,7 +462,8 @@ Output fstStateOutput(FstState *s, FstNode *node) {
                 - 1
                 - tSizes 
                 - oSizes;
-  return unpackUint64(slice->data + slice->start + i, oSizes);
+  uint8_t *data = fstSliceData(slice, NULL);              
+  return unpackUint64(data + i, oSizes);
   
 }
 Output fstStateOutputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
@@ -469,7 +480,9 @@ Output fstStateOutputForAnyTrans(FstState *s, FstNode *node, uint64_t i) {
                 - fstStateTotalTransSize(s, node->version, node->sizes, node->nTrans)
                 - (i * oSizes)
                 - oSizes;
-  return unpackUint64(slice->data + slice->start + at, oSizes);
+
+  uint8_t *data = fstSliceData(slice, NULL);              
+  return unpackUint64(data + at, oSizes);
 }
 
 // anyTrans specify function
@@ -523,7 +536,10 @@ uint64_t fstStateNtrans(FstState *s, FstSlice *slice) {
   if (null != true) {
     return n;   
   }  
-  n = slice->data[slice->end - 1]; // data[data.len() - 2]
+  int32_t len;
+  uint8_t *data = fstSliceData(slice, &len);
+  n = 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  
 }
 Output  fstStateFinalOutput(FstState *s, uint64_t version, FstSlice *slice, PackSizes sizes, uint64_t nTrans) {
@@ -538,7 +554,8 @@ Output  fstStateFinalOutput(FstState *s, uint64_t version, FstSlice *slice, Pack
                  - fstStateTotalTransSize(s, version, sizes, nTrans)
                  - (nTrans * oSizes)
                  - oSizes;
-  return unpackUint64(slice->data + slice->start + at, (uint8_t)oSizes);    
+  uint8_t *data = fstSliceData(slice, NULL);
+  return unpackUint64(data + at, (uint8_t)oSizes);    
 
 }
 uint64_t fstStateFindInput(FstState *s, FstNode *node, uint8_t b, bool *null) {
@@ -549,7 +566,10 @@ uint64_t fstStateFindInput(FstState *s, FstNode *node, uint8_t b, bool *null) {
                   - fstStateNtransLen(s)
                   - 1 // pack size 
                   - fstStateTransIndexSize(s, node->version, node->nTrans);
-    uint64_t i = slice->data[slice->start + at + b];  
+    int32_t dlen = 0; 
+    uint8_t *data = fstSliceData(slice, &dlen);
+    uint64_t i = data[at + b];
+    //uint64_t i = slice->data[slice->start + at + b];  
     if (i >= node->nTrans) {
       *null = true;
     } 
@@ -561,8 +581,13 @@ uint64_t fstStateFindInput(FstState *s, FstNode *node, uint8_t b, bool *null) {
                     - node->nTrans;
     uint64_t end =  start + node->nTrans;
     uint64_t len = end - start; 
+    int32_t dlen = 0; 
+    uint8_t *data = fstSliceData(slice, &dlen);
     for(int i = 0; i < len; i++) {
-      uint8_t v = slice->data[slice->start + i];
+      //uint8_t v = slice->data[slice->start + i];
+      ////slice->data[slice->start + i];
+      uint8_t v = data[i]; 
+      
       if (v == b) {
         return node->nTrans - i - 1; // bug  
       }
@@ -635,6 +660,7 @@ static const char *fstNodeState(FstNode *node) {
 
 
 void fstNodeDestroy(FstNode *node) {
+  fstSliceDestroy(&node->data); 
   free(node);
 }
 FstTransitions* fstNodeTransitions(FstNode *node) {
@@ -774,18 +800,18 @@ bool fstBuilderInsert(FstBuilder *b, FstSlice bs, Output in) {
 
 void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in) {
    FstSlice *s = &bs;
-   if (fstSliceEmpty(s)) {
+   if (fstSliceIsEmpty(s)) {
      b->len = 1; 
      fstUnFinishedNodesSetRootOutput(b->unfinished, in);
      return;
    }
-   Output out; 
    //if (in != 0) { //if let Some(in) = in 
    //   prefixLen = fstUnFinishedNodesFindCommPrefixAndSetOutput(b->unfinished, bs, in, &out);  
    //} else {
    //   prefixLen = fstUnFinishedNodesFindCommPrefix(b->unfinished, bs);
    //   out = 0;
    //}
+   Output out; 
    uint64_t prefixLen = fstUnFinishedNodesFindCommPrefixAndSetOutput(b->unfinished, bs, in, &out);
   
    if (prefixLen == FST_SLICE_LEN(s)) {
@@ -798,12 +824,13 @@ void fstBuilderInsertOutput(FstBuilder *b, FstSlice bs, Output in) {
    
    FstSlice sub = fstSliceCopy(s, prefixLen, s->end);
    fstUnFinishedNodesAddSuffix(b->unfinished, sub, out);
+   fstSliceDestroy(&sub);
    return;
  }
 
 OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup) {
   FstSlice *input = &bs;
-  if (fstSliceEmpty(&b->last)) {
+  if (fstSliceIsEmpty(&b->last)) {
     // deep copy or not
     b->last = fstSliceCopy(&bs, input->start, input->end);
   } else {
@@ -829,7 +856,7 @@ void fstBuilderCompileFrom(FstBuilder *b, uint64_t istate) {
     }
     addr =  fstBuilderCompile(b, n);
     assert(addr != NONE_ADDRESS);      
-    fstBuilderNodeDestroy(n);
+    //fstBuilderNodeDestroy(n);
   }
   fstUnFinishedNodesTopLastFreeze(b->unfinished, addr);
   return; 
@@ -888,7 +915,7 @@ void fstBuilderFinish(FstBuilder *b) {
 
 FstSlice fstNodeAsSlice(FstNode *node) {
   FstSlice *slice = &node->data; 
-  FstSlice s = fstSliceCopy(slice, slice->end, slice->dLen - 1);   
+  FstSlice s = fstSliceCopy(slice, slice->end, FST_SLICE_LEN(slice) - 1);   
   return s; 
 }
 
@@ -929,12 +956,13 @@ void fstBuilderNodeUnfinishedAddOutputPrefix(FstBuilderNodeUnfinished *unNode, O
 }
 
 Fst* fstCreate(FstSlice *slice) {
-  char *buf = slice->data;
+  int32_t slen;
-  uint64_t skip = 0;  
+  char *buf = fstSliceData(slice, &slen);
-  uint64_t len = slice->dLen;
+  if (slen < 36) { 
-  if (len < 36) { 
     return NULL; 
   }
+  uint64_t len = slen;
+  uint64_t skip = 0;  
 
   uint64_t version; 
   taosDecodeFixedU64(buf, &version); 
@@ -992,8 +1020,10 @@ void fstDestroy(Fst *fst) {
 bool fstGet(Fst *fst, FstSlice *b, Output *out) {
   FstNode *root = fstGetRoot(fst);    
   Output tOut = 0; 
-  for (uint32_t i = 0; i < b->dLen; i++) {
+  int32_t len;
-    uint8_t inp = b->data[i];
+  uint8_t *data = fstSliceData(b, &len);
+  for (uint32_t i = 0; i < len; i++) {
+    uint8_t inp = data[i];
     Output  res = 0;
     bool null = fstNodeFindInput(root, inp, &res);    
     if (null) { return false; }    
@@ -1046,9 +1076,10 @@ Output fstEmptyFinalOutput(Fst *fst, bool *null) {
 
 bool fstVerify(Fst *fst) {
   uint32_t checkSum = fst->meta->checkSum;
-  FstSlice *data    = fst->data;          
+  int32_t len;
+  uint8_t *data = fstSliceData(fst->data, &len);
   TSCKSUM initSum  = 0;  
-  if (!taosCheckChecksumWhole(data->data, data->dLen)) {
+  if (!taosCheckChecksumWhole(data, len)) {
     return false;
   }
   return true;
@@ -1059,8 +1090,13 @@ FstBoundWithData* fstBoundStateCreate(FstBound type, FstSlice *data) {
   FstBoundWithData *b = calloc(1, sizeof(FstBoundWithData));
   if (b == NULL) { return NULL; }
 
-  b->type = type;
+  if (data != NULL) {
     b->data = fstSliceCopy(data, data->start, data->end);
+  } else {
+    b->data = fstSliceCreate(NULL, 0);
+  }
+  b->type = type;
+
   return b; 
 }
 
@@ -1078,7 +1114,7 @@ bool fstBoundWithDataIsEmpty(FstBoundWithData *bound) {
   if (bound->type == Unbounded) {
     return true;
   } else { 
-    return fstSliceEmpty(&bound->data);  
+    return fstSliceIsEmpty(&bound->data);  
   }     
 }
 
@@ -1145,8 +1181,10 @@ bool streamWithStateSeekMin(StreamWithState *sws, FstBoundWithData *min) {
   Output  out = 0;
   void*   autState = sws->aut->start();  
 
-  for (uint32_t i = 0; i < key->dLen; i++) {
+  int32_t len; 
-    uint8_t b = key->data[i];
+  uint8_t *data = fstSliceData(key, &len);  
+  for (uint32_t i = 0; i < len; i++) {
+    uint8_t b = data[i];
     uint64_t res = 0;
     bool null = fstNodeFindInput(node, b, &res); 
     if (null == false) {
@@ -1262,12 +1300,16 @@ StreamWithStateResult *streamWithStateNextWith(StreamWithState *sws, StreamCallb
     if (fstBoundWithDataExceededBy(sws->endAt, &slice)) {
       taosArrayDestroyEx(sws->stack, streamStateDestroy);
       sws->stack = (SArray *)taosArrayInit(256, sizeof(StreamState)); 
+      fstSliceDestroy(&slice);
       return NULL;
     }
     if (FST_NODE_IS_FINAL(nextNode) && isMatch) {
-      FstOutput fOutput = {.null = false, out = out + FST_NODE_FINAL_OUTPUT(nextNode)};
+      FstOutput fOutput = {.null = false, .out = out + FST_NODE_FINAL_OUTPUT(nextNode)};
-      return swsResultCreate(&slice, fOutput , tState);
+      StreamWithStateResult *result =  swsResultCreate(&slice, fOutput , tState);     
+      fstSliceDestroy(&slice);
+      return result; 
     }
+    fstSliceDestroy(&slice);
   }
   return NULL; 
   
@@ -1277,14 +1319,19 @@ StreamWithStateResult *swsResultCreate(FstSlice *data, FstOutput fOut, void *sta
   StreamWithStateResult *result = calloc(1, sizeof(StreamWithStateResult));  
   if (result == NULL) { return NULL; }
   
-  FstSlice slice = fstSliceCopy(data, 0, data->dLen - 1);
+  result->data   = fstSliceCopy(data, 0, FST_SLICE_LEN(data) - 1);
-  result->data  = slice;
   result->out   = fOut;
   result->state = state; 
 
   return result;
-   
 }
+void swsResultDestroy(StreamWithStateResult *result) {
+  if (NULL == result) { return; }
+  
+  fstSliceDestroy(&result->data);
+  free(result);
+}
+
 void streamStateDestroy(void *s) {
   if (NULL == s) { return; }
   StreamState *ss = (StreamState *)s;
@@ -1293,5 +1340,44 @@ void streamStateDestroy(void *s) {
   //free(s->autoState);
 }
 
+FstStreamBuilder *fstStreamBuilderCreate(Fst *fst, Automation *aut) {
+  FstStreamBuilder *b = calloc(1, sizeof(FstStreamBuilder));
+  if (NULL == b) { return NULL; }
+
+  b->fst = fst;
+  b->aut = aut;
+  b->min = fstBoundStateCreate(Unbounded, NULL);
+  b->max = fstBoundStateCreate(Unbounded, NULL);  
+  return b;
+}
+void fstStreamBuilderDestroy(FstStreamBuilder *b) {
+  fstSliceDestroy(&b->min->data);
+  fstSliceDestroy(&b->max->data);
+  free(b);
+}
+FstStreamBuilder *fstStreamBuilderRange(FstStreamBuilder *b, FstSlice *val, RangeType type) {
+  if (b == NULL) { return NULL; }
+
+  if (type == GE) {
+    b->min->type = Included;
+    fstSliceDestroy(&(b->min->data));
+    b->min->data = fstSliceDeepCopy(val, 0, FST_SLICE_LEN(val) - 1);
+  } else if (type == GT) {
+    b->min->type = Excluded;
+    fstSliceDestroy(&(b->min->data));
+    b->min->data = fstSliceDeepCopy(val, 0, FST_SLICE_LEN(val) - 1);
+  } else if (type == LE) {
+    b->max->type = Included;
+    fstSliceDestroy(&(b->max->data));
+    b->max->data = fstSliceDeepCopy(val, 0, FST_SLICE_LEN(val) - 1);
+  } else if (type == LT) {
+    b->max->type = Excluded;
+    fstSliceDestroy(&(b->max->data));
+    b->max->data = fstSliceDeepCopy(val, 0, FST_SLICE_LEN(val) - 1);
+  }
+  return b;
+}
+
+
 
 

source/libs/index/src/index_fst_automation.c

@@ -12,3 +12,4 @@
  * 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/>.
  */
+

source/libs/index/src/index_fst_counting_writer.c

@@ -16,24 +16,88 @@
 #include "index_fst_util.h"
 #include "index_fst_counting_writer.h"
 
+static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
+  if (ctx->offset + len > ctx->limit) {
+    return -1;
+  }
+
+  if (ctx->type == TFile) {
+    assert(len != tfWrite(ctx->fd, buf, len));
+  } else {
+    memcpy(ctx->mem + ctx->offset, buf, len);
+  } 
+  ctx->offset += len;
+  return len;
+}
+static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
+  if (ctx->type == TFile) {
+    tfRead(ctx->fd, buf, len);
+  } else {
+    memcpy(buf, ctx->mem + ctx->offset, len);
+  }
+  ctx->offset += len;
+
+  return 1;
+} 
+static int writeCtxDoFlush(WriterCtx *ctx) {
+  if (ctx->type == TFile) {
+    //tfFlush(ctx->fd);
+  } else {
+    // do nothing
+  }
+  return 1;
+}
+
+WriterCtx* writerCtxCreate(WriterType type) {     
+  WriterCtx *ctx = calloc(1, sizeof(WriterCtx));
+  if (ctx == NULL) { return NULL; }
+
+  ctx->type == type;
+  if (ctx->type == TFile) {
+    ctx->fd = tfOpenCreateWriteAppend(tmpFile);  
+  } else if (ctx->type == TMemory) {
+    ctx->mem = calloc(1, DefaultMem * sizeof(uint8_t));
+  } 
+  ctx->write = writeCtxDoWrite;
+  ctx->read  = writeCtxDoRead;
+  ctx->flush = writeCtxDoFlush;
+
+  ctx->offset = 0;
+  ctx->limit  = DefaultMem;
+
+  return ctx;
+}
+void writerCtxDestroy(WriterCtx *ctx) {
+  if (ctx->type == TMemory) {
+    free(ctx->mem);
+  } else {
+    tfClose(ctx->fd);    
+  }
+  free(ctx);
+}
+
+
 FstCountingWriter *fstCountingWriterCreate(void *wrt) {
   FstCountingWriter *cw = calloc(1, sizeof(FstCountingWriter)); 
   if (cw == NULL) { return NULL; }
   
-  cw->wrt = wrt; 
+  cw->wrt = (void *)(writerCtxCreate(TFile)); 
   return cw; 
 }
 void fstCountingWriterDestroy(FstCountingWriter *cw) {
   // free wrt object: close fd or free mem 
+  writerCtxDestroy((WriterCtx *)(cw->wrt));
   free(cw);
 }
 
-uint64_t fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t bufLen) {
+int fstCountingWriterWrite(FstCountingWriter *write, uint8_t *buf, uint32_t bufLen) {
   if (write == NULL) { return 0; } 
   // update checksum 
   // write data to file/socket or mem
+  WriterCtx *ctx = write->wrt;
 
-  write->count += bufLen;
+  int nWrite = ctx->write(ctx, buf, bufLen); 
+  write->count += nWrite;
   return bufLen; 
 } 
 
@@ -41,6 +105,8 @@ uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write) {
   return 0;
 }
 int fstCountingWriterFlush(FstCountingWriter *write) {
+  WriterCtx *ctx = write->wrt;
+  ctx->flush(ctx);
   //write->wtr->flush
   return 1;
 }

source/libs/index/src/index_fst_node.c

@@ -18,7 +18,7 @@ FstBuilderNode *fstBuilderNodeDefault() {
   FstBuilderNode *bn = malloc(sizeof(FstBuilderNode));
   bn->isFinal     = false; 
   bn->finalOutput = 0; 
-  bn->trans       = NULL; 
+  bn->trans       = taosArrayInit(16, sizeof(FstTransition)); 
   return bn;
 }
 void fstBuilderNodeDestroy(FstBuilderNode *node) {
@@ -27,6 +27,25 @@ void fstBuilderNodeDestroy(FstBuilderNode *node) {
   taosArrayDestroy(node->trans);
   free(node);
 } 
+
+bool fstBuilderNodeEqual(FstBuilderNode *n1, FstBuilderNode *n2) {
+  if (n1 == n2) { return true; }
+
+  if (n1->isFinal != n2->isFinal || 
+      n1->finalOutput != n2->finalOutput ||
+      taosArrayGetSize(n1->trans) != taosArrayGetSize(n2->trans)) {
+    return false;  
+  } 
+  size_t sz =  taosArrayGetSize(n1->trans); 
+  for (size_t i = 0; i < sz; i++) {
+    FstTransition *t1 = taosArrayGet(n1->trans, i);
+    FstTransition *t2 = taosArrayGet(n2->trans, i);
+    if (t1->inp != t2->inp || t1->out != t2->out || t1->addr != t2->addr) {
+      return false;
+    }  
+  }
+  return true;
+}
 FstBuilderNode *fstBuilderNodeClone(FstBuilderNode *src) {
   FstBuilderNode *node = malloc(sizeof(FstBuilderNode));  
   if (node == NULL) { return NULL; }
@@ -53,12 +72,17 @@ void fstBuilderNodeCloneFrom(FstBuilderNode *dst, FstBuilderNode *src) {
   dst->isFinal     = src->isFinal;
   dst->finalOutput = src->finalOutput;
 
-  // avoid mem leak
+  //release free  avoid mem leak
   taosArrayDestroy(dst->trans); 
-  dst->trans       = src->trans;    
+  size_t sz = taosArrayGetSize(src->trans);
-  src->trans = NULL;
+  dst->trans = taosArrayInit(sz, sizeof(FstTransition)); 
+  for (size_t i = 0; i < sz; i++) {
+    FstTransition *trn = taosArrayGet(src->trans, i); 
+    taosArrayPush(dst->trans, trn);
+  } 
 }
 
+
 //bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr startAddr) {
   
   //size_t sz = taosArrayGetSize(b->trans);  

source/libs/index/src/index_fst_registry.c

@@ -112,7 +112,7 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
   if (end - start == 1) {
     FstRegistryCell *cell = taosArrayGet(registry->table, start); 
     //cell->isNode && 
-    if (cell->addr != NONE_ADDRESS && cell->node == bNode) {
+    if (cell->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell->node, bNode)) {
        entry->state = FOUND;
        entry->addr  = cell->addr ;
        return entry; 
@@ -123,13 +123,13 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
     }
   } else if (end - start == 2) {
     FstRegistryCell *cell1 = taosArrayGet(registry->table, start); 
-    if (cell1->addr != NONE_ADDRESS && cell1->node == bNode) {
+    if (cell1->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell1->node, bNode)) {
       entry->state = FOUND;   
       entry->addr  = cell1->addr;
       return entry;
     }      
     FstRegistryCell *cell2 = taosArrayGet(registry->table, start + 1); 
-    if (cell2->addr != NONE_ADDRESS && cell2->node == bNode) {
+    if (cell2->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell2->node, bNode)) {
       entry->state = FOUND; 
       entry->addr  = cell2->addr;
       // must swap here
@@ -147,7 +147,7 @@ FstRegistryEntry *fstRegistryGetEntry(FstRegistry *registry, FstBuilderNode *bNo
     uint32_t i = start; 
     for (; i < end; i++) {
       FstRegistryCell *cell = (FstRegistryCell *)taosArrayGet(registry->table, i);
-      if (cell->addr != NONE_ADDRESS && cell->node == bNode) {
+      if (cell->addr != NONE_ADDRESS && fstBuilderNodeEqual(cell->node, bNode)) {
         entry->state = FOUND;  
         entry->addr  = cell->addr; 
         fstRegistryCellPromote(registry->table, i, start);

source/libs/index/src/index_fst_util.c

@@ -91,42 +91,87 @@ CompiledAddr unpackDelta(char *data, uint64_t len, uint64_t nodeAddr) {
 }
 
 // fst slice func
-FstSlice fstSliceCreate(uint8_t *data, uint64_t dLen) {
+//
-  FstSlice slice = {.data = data, .dLen = dLen, .start = 0, .end = dLen - 1};
+
-  return slice;
+FstSlice fstSliceCreate(uint8_t *data, uint64_t len) {
+  FstString *str = (FstString *)malloc(sizeof(FstString));
+  str->ref  = 1; 
+  str->len  = len;
+  str->data = malloc(len * sizeof(uint8_t));
+  memcpy(str->data, data, len);
+  
+  FstSlice s = {.str = str, .start = 0, .end = len - 1};
+  return s;
 } 
 // just shallow copy
-FstSlice fstSliceCopy(FstSlice *slice, int32_t start, int32_t end) {
+FstSlice fstSliceCopy(FstSlice *s, int32_t start, int32_t end) {
-  FstSlice t;
+  FstString *str = s->str;
-  if (start >= slice->dLen || end >= slice->dLen || start > end) {
+  str->ref++;
-    t.data = NULL;
+  //uint8_t *buf = fstSliceData(s, &alen); 
-    return t;
+  //start = buf + start - (buf - s->start);
-  };
+  //end   = buf + end - (buf - s->start);  
 
-  t.data  = slice->data;
+  FstSlice t = {.str = str, .start = start + s->start, .end = end + s->start};
-  t.dLen  = slice->dLen;
-  t.start = start; 
-  t.end   = end; 
   return t;
 }
-bool fstSliceEmpty(FstSlice *slice) {
+FstSlice fstSliceDeepCopy(FstSlice *s, int32_t start, int32_t end) {
-  return slice->data == NULL || slice->dLen <= 0;
+
+  int32_t tlen = end - start + 1;
+  int32_t slen;
+  uint8_t *data = fstSliceData(s, &slen);      
+  assert(tlen <= slen);
+
+  uint8_t *buf  = malloc(sizeof(uint8_t) * tlen);  
+  memcpy(buf, data + start, tlen);
+   
+  FstString *str = malloc(sizeof(FstString));  
+  str->data = buf;
+  str->len  = tlen; 
+  str->ref  = 1;
+
+  FstSlice ans;
+  ans.str   = str;
+  ans.start = 0; 
+  ans.end   = tlen - 1;
+  return ans; 
+}
+bool fstSliceIsEmpty(FstSlice *s) {
+  return s->str == NULL || s->str->len == 0 || s->start < 0 || s->end < 0;
+}
+
+uint8_t *fstSliceData(FstSlice *s, int32_t *size) {
+   FstString *str = s->str;
+   if (size != NULL) {
+    *size = s->end - s->start + 1;
+   }
+   return str->data + s->start; 
+}
+void fstSliceDestroy(FstSlice *s) {
+  FstString *str = s->str;
+  str->ref--;
+  if (str->ref <= 0) {
+    free(str->data);
+    free(str);
+    s->str = NULL; 
+  }
 }
 
 int fstSliceCompare(FstSlice *a, FstSlice *b) {
-  int32_t aLen = (a->end - a->start + 1);  
+  int32_t alen, blen;
-  int32_t bLen = (b->end - b->start + 1);
+  uint8_t *aBuf = fstSliceData(a, &alen); 
-  int32_t mLen = (aLen < bLen ? aLen : bLen);
+  uint8_t *bBuf = fstSliceData(b, &blen); 
-  for (int i = 0; i < mLen; i++) {
+
-    uint8_t x = a->data[i + a->start];
+  uint32_t i, j;
-    uint8_t y = b->data[i + b->start];
+  for (i = 0, j = 0; i < alen && j < blen; i++, j++) {
-    if (x == y) { continue; }
+    uint8_t x = aBuf[i];
+    uint8_t y = bBuf[j];
+    if (x == y) { continue;}
     else if (x < y) { return -1; }
-    else { return 1; } 
+    else { return 1; };
   }
-  if (aLen == bLen) { return 0; }
+  if (i < alen) { return 1; } 
-  else if (aLen < bLen) { return -1; }
+  else if (j < blen) { return -1; }
-  else { return 1; } 
+  else { return 0; } 
 } 
 
 

source/libs/index/test/CMakeLists.txt

@@ -1,7 +1,6 @@
 add_executable(indexTest "")
 target_sources(indexTest
   PRIVATE 
-  "../src/index.c"  
   "indexTests.cpp" 
 )
 target_include_directories ( indexTest

source/libs/index/test/indexTests.cpp

@@ -3,58 +3,84 @@
 #include <iostream>
 #include "index.h"
 #include "indexInt.h"
+#include "index_fst.h"
+#include "index_fst_util.h"
+#include "index_fst_counting_writer.h"
 
 
 
 
-TEST(IndexTest, index_create_test) {
+//TEST(IndexTest, index_create_test) {
-  SIndexOpts *opts = indexOptsCreate();
+//  SIndexOpts *opts = indexOptsCreate();
-  SIndex *index = indexOpen(opts, "./test");
+//  SIndex *index = indexOpen(opts, "./test");
-  if (index == NULL) {
+//  if (index == NULL) {
-    std::cout << "index open failed" << std::endl; 
+//    std::cout << "index open failed" << std::endl; 
-  }
+//  }
+//
+//  
+//  // write   
+//  for (int i = 0; i < 100000; i++) {
+//    SIndexMultiTerm* terms = indexMultiTermCreate();
+//    std::string val = "field";    
+//
+//    indexMultiTermAdd(terms, "tag1", strlen("tag1"), val.c_str(), val.size());
+//
+//    val.append(std::to_string(i)); 
+//    indexMultiTermAdd(terms, "tag2", strlen("tag2"), val.c_str(), val.size());
+//
+//    val.insert(0, std::to_string(i));
+//    indexMultiTermAdd(terms, "tag3", strlen("tag3"), val.c_str(), val.size());
+//
+//    val.append("const");    
+//    indexMultiTermAdd(terms, "tag4", strlen("tag4"), val.c_str(), val.size());
+//
+//     
+//    indexPut(index, terms, i);
+//    indexMultiTermDestroy(terms);
+//  } 
+// 
+//
+//  // query
+//  SIndexMultiTermQuery *multiQuery = indexMultiTermQueryCreate(MUST); 
+//  
+//  indexMultiTermQueryAdd(multiQuery, "tag1", strlen("tag1"), "field", strlen("field"), QUERY_PREFIX);
+//  indexMultiTermQueryAdd(multiQuery, "tag3", strlen("tag3"), "0field0", strlen("0field0"), QUERY_TERM);
+//
+//  SArray *result = (SArray *)taosArrayInit(10, sizeof(int));   
+//  indexSearch(index, multiQuery, result);
+//
+//  std::cout << "taos'size : " << taosArrayGetSize(result) << std::endl;
+//  for (int i = 0;  i < taosArrayGetSize(result); i++) {
+//    int *v = (int *)taosArrayGet(result, i);
+//    std::cout << "value --->" << *v  << std::endl;
+//  }
+//  // add more test case 
+//  indexMultiTermQueryDestroy(multiQuery);
+//
+//  indexOptsDestroy(opts); 
+//  indexClose(index); 
+//  //
+//}
 
+int main(int argc, char** argv) {
+  std::string str("abc");
+  FstSlice key = fstSliceCreate((uint8_t *)str.c_str(), str.size());
+  Output   val = 10;
 
-  // write   
+  std::string str1("bcd");
-  for (int i = 0; i < 100000; i++) {
+  FstSlice key1 = fstSliceCreate((uint8_t *)str1.c_str(), str1.size());
-    SIndexMultiTerm* terms = indexMultiTermCreate();
+  Output   val2 = 10;
-    std::string val = "field";    
+  FstBuilder *b = fstBuilderCreate(NULL, 1);
-
+  fstBuilderInsert(b, key, val); 
-    indexMultiTermAdd(terms, "tag1", strlen("tag1"), val.c_str(), val.size());
+  fstBuilderInsert(b, key1, val2); 
-
+  fstBuilderFinish(b);
-    val.append(std::to_string(i)); 
+  fstBuilderDestroy(b);
-    indexMultiTermAdd(terms, "tag2", strlen("tag2"), val.c_str(), val.size());
+  fstSliceDestroy(&key);
-
+  return 1;
-    val.insert(0, std::to_string(i));
-    indexMultiTermAdd(terms, "tag3", strlen("tag3"), val.c_str(), val.size());
-
-    val.append("const");    
-    indexMultiTermAdd(terms, "tag4", strlen("tag4"), val.c_str(), val.size());
-
-     
-    indexPut(index, terms, i);
-    indexMultiTermDestroy(terms);
-  } 
- 
-
-  // query
-  SIndexMultiTermQuery *multiQuery = indexMultiTermQueryCreate(MUST); 
-  
-  indexMultiTermQueryAdd(multiQuery, "tag1", strlen("tag1"), "field", strlen("field"), QUERY_PREFIX);
-  indexMultiTermQueryAdd(multiQuery, "tag3", strlen("tag3"), "0field0", strlen("0field0"), QUERY_TERM);
-
-  SArray *result = (SArray *)taosArrayInit(10, sizeof(int));   
-  indexSearch(index, multiQuery, result);
-
-  std::cout << "taos'size : " << taosArrayGetSize(result) << std::endl;
-  for (int i = 0;  i < taosArrayGetSize(result); i++) {
-    int *v = (int *)taosArrayGet(result, i);
-    std::cout << "value --->" << *v  << std::endl;
-  }
-  // add more test case 
-  indexMultiTermQueryDestroy(multiQuery);
-
-  indexOptsDestroy(opts); 
-  indexClose(index); 
-  //
 }
+
+//TEST(IndexFstBuilder, IndexFstInput) {
+//
+//}
+
+