Merge pull request #9112 from taosdata/feature/fst

fix search bug

source/libs/index/inc/index_fst.h

@@ -30,6 +30,7 @@ extern "C" {
 
 typedef struct Fst     Fst;
 typedef struct FstNode FstNode;
+typedef struct StreamWithState StreamWithState;
 
 typedef enum { Included, Excluded, Unbounded} FstBound; 
 
@@ -283,6 +284,9 @@ Output       fstEmptyFinalOutput(Fst *fst, bool *null);
 FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx);
 FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx);
 
+// into stream to expand later 
+StreamWithState* streamBuilderIntoStream(FstStreamBuilder *sb);
+
 bool         fstVerify(Fst *fst);
 
 

source/libs/index/src/index_fst.c

@@ -1094,6 +1094,10 @@ bool fstGet(Fst *fst, FstSlice *b, Output *out) {
 FstStreamBuilder *fstSearch(Fst *fst, AutomationCtx *ctx) {
   return fstStreamBuilderCreate(fst, ctx);
 }
+StreamWithState* streamBuilderIntoStream(FstStreamBuilder *sb) {
+  if (sb == NULL) { return NULL; } 
+  return streamWithStateCreate(sb->fst, sb->aut, sb->min, sb->max);
+}
 FstStreamWithStateBuilder *fstSearchWithState(Fst *fst, AutomationCtx *ctx) {
   return fstStreamBuilderCreate(fst, ctx);
 }

source/libs/index/test/indexTests.cpp

@@ -59,9 +59,22 @@ class FstReadMemory {
      return ok; 
    }
    // add later
-   bool Search(const std::string &key, std::vector<uint64_t> &result) {
+   bool Search(AutomationCtx *ctx, std::vector<uint64_t> &result) { 
+      FstStreamBuilder *sb = fstSearch(_fst, ctx);
+      StreamWithState *st = streamBuilderIntoStream(sb);  
+      StreamWithStateResult *rt = NULL;  
+      
+      while ((rt = streamWithStateNextWith(st, NULL)) != NULL) {
+        result.push_back((uint64_t)(rt->out.out));
+      }
       return true;
    }
+   bool SearchWithTimeCostUs(AutomationCtx *ctx, std::vector<uint64_t> &result) {
+     int64_t s = taosGetTimestampUs();
+     bool ok = this->Search(ctx, result); 
+     int64_t e = taosGetTimestampUs();
+     return ok;
+   }
     
    ~FstReadMemory() {
     fstCountingWriterDestroy(_w);
@@ -186,11 +199,43 @@ void checkFstPerf() {
     printf("success to init fst read");  
   }  
   Performance_fstReadRecords(m); 
-   
   delete m;
 } 
 
+void checkFstPrefixSearch() {
+  FstWriter *fw = new FstWriter;
+  int64_t s = taosGetTimestampUs();
+  int count = 2;
+  std::string key("ab");
+  
+  for (int i = 0; i < count; i++) {
+    key[1] = key[1] + i;   
+    fw->Put(key, i); 
+  }
+  int64_t e = taosGetTimestampUs();
+  
+  std::cout << "insert data count :  " << count << "elapas time: " << e - s << std::endl;
+  delete fw;
 
+  FstReadMemory *m = new FstReadMemory(1024 * 64);
+  if (m->init() == false) {
+    std::cout << "init readMemory failed" << std::endl; 
+    delete m;
+    return;
+  }
+  
+  // prefix search 
+  std::vector<uint64_t> result;
+  AutomationCtx *ctx = automCtxCreate((void *)"ab", AUTOMATION_PREFIX); 
+  m->Search(ctx, result);   
+  assert(result.size() == count); 
+  for (int i = 0; i < result.size(); i++) {
+    assert(result[i] == i); // check result
+  }
+
+  free(ctx);
+  delete m;
+}  
 void validateFst() {
   int val = 100;
   int count = 100;
@@ -209,6 +254,8 @@ void validateFst() {
   FstReadMemory *m = new FstReadMemory(1024 * 64);
   if (m->init() == false) { 
     std::cout << "init readMemory failed" << std::endl; 
+    delete m;
+    return;
   }
 
   {
@@ -230,10 +277,12 @@ void validateFst() {
    }
   } 
   delete m;
-
 } 
+
+
 int main(int argc, char** argv) {
   checkFstPerf(); 
+  //checkFstPrefixSearch();
   return 1;
 }