Merge pull request #6338 from taosdata/feature/TD-2574

[TD-2574]<enhance>: refactored algorithms for top and bottom functions.

src/query/src/qAggMain.c

@@ -1758,6 +1758,49 @@ static void valuePairAssign(tValuePair *dst, int16_t type, const char *val, int6
     memcpy((dst)->pTags, (src)->pTags, (size_t)(__l)); \
   } while (0)
 
+static int32_t topBotComparFn(const void *p1, const void *p2, const void *param)
+{
+  uint16_t     type = *(uint16_t *) param;
+  tValuePair  *val1 = *(tValuePair **) p1;
+  tValuePair  *val2 = *(tValuePair **) p2;
+
+  if (IS_SIGNED_NUMERIC_TYPE(type)) {
+    if (val1->v.i64 == val2->v.i64) {
+      return 0;
+    }
+
+    return (val1->v.i64 > val2->v.i64) ? 1 : -1;
+   } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
+     if (val1->v.u64 == val2->v.u64) {
+       return 0;
+     }
+
+     return (val1->v.u64 > val2->v.u64) ? 1 : -1;
+  }
+
+  if (val1->v.dKey == val2->v.dKey) {
+    return 0;
+  }
+
+  return (val1->v.dKey > val2->v.dKey) ? 1 : -1;
+}
+
+static void topBotSwapFn(void *dst, void *src, const void *param)
+{
+  char         tag[32768];
+  tValuePair   temp;
+  uint16_t     tagLen = *(uint16_t *) param;
+  tValuePair  *vdst = *(tValuePair **) dst;
+  tValuePair  *vsrc = *(tValuePair **) src;
+
+  memset(tag, 0, sizeof(tag));
+  temp.pTags = tag;
+
+  VALUEPAIRASSIGN(&temp, vdst, tagLen);
+  VALUEPAIRASSIGN(vdst, vsrc, tagLen);
+  VALUEPAIRASSIGN(vsrc, &temp, tagLen);
+}
+
 static void do_top_function_add(STopBotInfo *pInfo, int32_t maxLen, void *pData, int64_t ts, uint16_t type,
                                 SExtTagsInfo *pTagInfo, char *pTags, int16_t stage) {
   tVariant val = {0};
@@ -1765,61 +1808,19 @@ static void do_top_function_add(STopBotInfo *pInfo, int32_t maxLen, void *pData,
   
   tValuePair **pList = pInfo->res;
   assert(pList != NULL);
-  
+
   if (pInfo->num < maxLen) {
-    if (pInfo->num == 0 ||
-        (IS_SIGNED_NUMERIC_TYPE(type) && val.i64 >= pList[pInfo->num - 1]->v.i64) ||
-        (IS_UNSIGNED_NUMERIC_TYPE(type) && val.u64 >= pList[pInfo->num - 1]->v.u64) ||
-        (IS_FLOAT_TYPE(type) && val.dKey >= pList[pInfo->num - 1]->v.dKey)) {
-      valuePairAssign(pList[pInfo->num], type, (const char*)&val.i64, ts, pTags, pTagInfo, stage);
-    } else {
-      int32_t i = pInfo->num - 1;
-      if (IS_SIGNED_NUMERIC_TYPE(type)) {
-        while (i >= 0 && pList[i]->v.i64 > val.i64) {
-          VALUEPAIRASSIGN(pList[i + 1], pList[i], pTagInfo->tagsLen);
-          i -= 1;
-        }
-      } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
-        while (i >= 0 && pList[i]->v.u64 > val.u64) {
-          VALUEPAIRASSIGN(pList[i + 1], pList[i], pTagInfo->tagsLen);
-          i -= 1;
-        }
-      } else {
-        while (i >= 0 && pList[i]->v.dKey > val.dKey) {
-          VALUEPAIRASSIGN(pList[i + 1], pList[i], pTagInfo->tagsLen);
-          i -= 1;
-        }
-      }
-      
-      valuePairAssign(pList[i + 1], type, (const char*) &val.i64, ts, pTags, pTagInfo, stage);
-    }
-    
+    valuePairAssign(pList[pInfo->num], type, (const char *)&val.i64, ts, pTags, pTagInfo, stage);
+
+    taosheapsort((void *) pList, sizeof(tValuePair **), pInfo->num + 1, (const void *) &type, topBotComparFn, (const void *) &pTagInfo->tagsLen, topBotSwapFn, 0);
+ 
     pInfo->num++;
   } else {
-    int32_t i = 0;
-
     if ((IS_SIGNED_NUMERIC_TYPE(type) && val.i64 > pList[0]->v.i64) ||
         (IS_UNSIGNED_NUMERIC_TYPE(type) && val.u64 > pList[0]->v.u64) ||
         (IS_FLOAT_TYPE(type) && val.dKey > pList[0]->v.dKey)) {
-      // find the appropriate the slot position
-      if (IS_SIGNED_NUMERIC_TYPE(type)) {
-        while (i + 1 < maxLen && pList[i + 1]->v.i64 < val.i64) {
-          VALUEPAIRASSIGN(pList[i], pList[i + 1], pTagInfo->tagsLen);
-          i += 1;
-        }
-      } if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
-        while (i + 1 < maxLen && pList[i + 1]->v.u64 < val.u64) {
-          VALUEPAIRASSIGN(pList[i], pList[i + 1], pTagInfo->tagsLen);
-          i += 1;
-        }
-      } else {
-        while (i + 1 < maxLen && pList[i + 1]->v.dKey < val.dKey) {
-          VALUEPAIRASSIGN(pList[i], pList[i + 1], pTagInfo->tagsLen);
-          i += 1;
-        }
-      }
-
-      valuePairAssign(pList[i], type, (const char *)&val.i64, ts, pTags, pTagInfo, stage);
+      valuePairAssign(pList[0], type, (const char *)&val.i64, ts, pTags, pTagInfo, stage);
+      taosheapadjust((void *) pList, sizeof(tValuePair **), 0, maxLen - 1, (const void *) &type, topBotComparFn, (const void *) &pTagInfo->tagsLen, topBotSwapFn, 0);
     }
   }
 }
@@ -1833,57 +1834,17 @@ static void do_bottom_function_add(STopBotInfo *pInfo, int32_t maxLen, void *pDa
   assert(pList != NULL);
 
   if (pInfo->num < maxLen) {
-    if (pInfo->num == 0) {
-      valuePairAssign(pList[pInfo->num], type, (const char*) &val.i64, ts, pTags, pTagInfo, stage);
-    } else {
-      int32_t i = pInfo->num - 1;
-      
-      if (IS_SIGNED_NUMERIC_TYPE(type)) {
-        while (i >= 0 && pList[i]->v.i64 < val.i64) {
-          VALUEPAIRASSIGN(pList[i + 1], pList[i], pTagInfo->tagsLen);
-          i -= 1;
-        }
-      } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
-          while (i >= 0 && pList[i]->v.u64 < val.u64) {
-            VALUEPAIRASSIGN(pList[i + 1], pList[i], pTagInfo->tagsLen);
-            i -= 1;
-          }
-      } else {
-        while (i >= 0 && pList[i]->v.dKey < val.dKey) {
-          VALUEPAIRASSIGN(pList[i + 1], pList[i], pTagInfo->tagsLen);
-          i -= 1;
-        }
-      }
-      
-      valuePairAssign(pList[i + 1], type, (const char*)&val.i64, ts, pTags, pTagInfo, stage);
-    }
-    
+    valuePairAssign(pList[pInfo->num], type, (const char *)&val.i64, ts, pTags, pTagInfo, stage);
+
+    taosheapsort((void *) pList, sizeof(tValuePair **), pInfo->num + 1, (const void *) &type, topBotComparFn, (const void *) &pTagInfo->tagsLen, topBotSwapFn, 1);
+
     pInfo->num++;
   } else {
-    int32_t i = 0;
-    
     if ((IS_SIGNED_NUMERIC_TYPE(type) && val.i64 < pList[0]->v.i64) ||
         (IS_UNSIGNED_NUMERIC_TYPE(type) && val.u64 < pList[0]->v.u64) ||
         (IS_FLOAT_TYPE(type) && val.dKey < pList[0]->v.dKey)) {
-      // find the appropriate the slot position
-      if (IS_SIGNED_NUMERIC_TYPE(type)) {
-        while (i + 1 < maxLen && pList[i + 1]->v.i64 > val.i64) {
-          VALUEPAIRASSIGN(pList[i], pList[i + 1], pTagInfo->tagsLen);
-          i += 1;
-        }
-      } if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
-        while (i + 1 < maxLen && pList[i + 1]->v.u64 > val.u64) {
-          VALUEPAIRASSIGN(pList[i], pList[i + 1], pTagInfo->tagsLen);
-          i += 1;
-        }
-      } else {
-        while (i + 1 < maxLen && pList[i + 1]->v.dKey > val.dKey) {
-          VALUEPAIRASSIGN(pList[i], pList[i + 1], pTagInfo->tagsLen);
-          i += 1;
-        }
-      }
-      
-      valuePairAssign(pList[i], type, (const char*)&val.i64, ts, pTags, pTagInfo, stage);
+      valuePairAssign(pList[0], type, (const char *)&val.i64, ts, pTags, pTagInfo, stage);
+      taosheapadjust((void *) pList, sizeof(tValuePair **), 0, maxLen - 1, (const void *) &type, topBotComparFn, (const void *) &pTagInfo->tagsLen, topBotSwapFn, 1);
     }
   }
 }

src/util/inc/talgo.h

@@ -34,6 +34,7 @@ typedef int (*__compar_fn_t) (const void *, const void *);
 #define elePtrAt(base, size, idx) (void *)((char *)(base) + (size) * (idx))
 
 typedef int32_t (*__ext_compar_fn_t)(const void *p1, const void *p2, const void *param);
+typedef void (*__ext_swap_fn_t)(void *p1, void *p2, const void *param);
 
 /**
  * quick sort, with the compare function requiring additional parameters support
@@ -59,6 +60,38 @@ void taosqsort(void *src, size_t numOfElem, size_t size, const void* param, __ex
  */
 void *taosbsearch(const void *key, const void *base, size_t nmemb, size_t size, __compar_fn_t fn, int flags);
 
+/**
+ * adjust heap
+ *
+ * @param base: the start address of array
+ * @param size: size of every item in array
+ * @param start: the first index
+ * @param end: the last index
+ * @param parcompar: parameters for compare function
+ * @param compar: user defined compare function
+ * @param parswap: parameters for swap function
+ * @param swap: user defined swap function, the default swap function doswap will be used if swap is NULL
+ * @param maxroot: if heap is max root heap
+ * @return
+ */
+void taosheapadjust(void *base, int32_t size, int32_t start, int32_t end, const void *parcompar, __ext_compar_fn_t compar, const void *parswap, __ext_swap_fn_t swap, bool maxroot);
+
+/**
+ * sort heap to make sure it is a max/min root heap
+ *
+ * @param base: the start address of array
+ * @param size: size of every item in array
+ * @param len: the length of array
+ * @param parcompar: parameters for compare function
+ * @param compar: user defined compare function
+ * @param parswap: parameters for swap function
+ * @param swap: user defined swap function, the default swap function doswap will be used if swap is NULL
+ * @param maxroot: if heap is max root heap
+ * @return
+ */
+void taosheapsort(void *base, int32_t size, int32_t len, const void *parcompar, __ext_compar_fn_t compar, const void *parswap, __ext_swap_fn_t swap, bool maxroot);
+
+
 #ifdef __cplusplus
 }
 #endif

src/util/src/talgo.c

@@ -225,3 +225,89 @@ void * taosbsearch(const void *key, const void *base, size_t nmemb, size_t size,
   
   return NULL;
 }
+
+void taosheapadjust(void *base, int32_t size, int32_t start, int32_t end, const void *parcompar, __ext_compar_fn_t compar, const void *parswap, __ext_swap_fn_t swap, bool maxroot)
+{
+  int32_t  parent;
+  int32_t  child;
+  char    *buf;
+
+  if (base && size > 0 && compar) {
+    parent = start;
+    child = 2 * parent + 1;
+
+    if (swap == NULL) {
+      buf = calloc(1, size);
+      if (buf == NULL) {
+        return;
+      }
+    }
+
+    if (maxroot) {
+      while (child <= end) {
+        if (child + 1 <= end && (*compar)(elePtrAt(base, size, child), elePtrAt(base, size, child + 1), parcompar) < 0) {
+          child++;
+        }
+
+        if ((*compar)(elePtrAt(base, size, parent), elePtrAt(base, size, child), parcompar) > 0) {
+          break;
+        }
+
+        if (swap == NULL) {
+          doswap(elePtrAt(base, size, parent), elePtrAt(base, size, child), size, buf);
+        } else {
+          (*swap)(elePtrAt(base, size, parent), elePtrAt(base, size, child), parswap);
+        }
+
+        parent = child;
+        child = 2 * parent + 1;
+      }
+    } else {
+      while (child <= end) {
+        if (child + 1 <= end && (*compar)(elePtrAt(base, size, child), elePtrAt(base, size, child + 1), parcompar) > 0) {
+          child++;
+        }
+
+        if ((*compar)(elePtrAt(base, size, parent), elePtrAt(base, size, child), parcompar) < 0) {
+          break;
+        }
+
+        if (swap == NULL) {
+          doswap(elePtrAt(base, size, parent), elePtrAt(base, size, child), size, buf);
+        } else {
+          (*swap)(elePtrAt(base, size, parent), elePtrAt(base, size, child), parswap);
+        }
+
+        parent = child;
+        child = 2 * parent + 1;
+      }
+    }
+
+    if (swap == NULL) {
+      tfree(buf);
+    }
+  }
+}
+
+void taosheapsort(void *base, int32_t size, int32_t len, const void *parcompar, __ext_compar_fn_t compar, const void *parswap, __ext_swap_fn_t swap, bool maxroot)
+{
+  int32_t  i;
+
+  if (base && size > 0) {
+    for (i = len / 2 - 1; i >= 0; i--) {
+      taosheapadjust(base, size, i, len - 1, parcompar, compar, parswap, swap, maxroot);
+    }
+  }
+
+/*
+  char *buf = calloc(1, size);
+
+  for (i = len - 1; i > 0; i--) {
+    doswap(elePtrAt(base, size, 0), elePtrAt(base, size, i));
+    taosheapadjust(base, size, 0, i - 1, parcompar, compar, parswap, swap, maxroot);
+  }
+
+  tfree(buf);
+*/
+}
+

tests/script/general/parser/select_with_tags.sim

@@ -190,32 +190,32 @@ if $rows != 12800 then
   return -1
 endi
 
-sql select top(c1, 100), tbname, t1, t2 from select_tags_mt0;
-if $rows != 100 then
+sql select top(c1, 80), tbname, t1, t2 from select_tags_mt0;
+if $rows != 80 then
    return -1
 endi
 
-if $data00 != @70-01-01 08:03:30.100@ then
+if $data00 != @70-01-01 08:03:40.100@ then
    return -1
 endi
 
-if $data10 != @70-01-01 08:03:30.200@ then
+if $data10 != @70-01-01 08:03:40.200@ then
    return -1
 endi
 
-if $data01 != 110 then
+if $data01 != 111 then
    return -1
 endi
 
-if $data02 != @select_tags_tb11@ then
+if $data02 != @select_tags_tb12@ then
    return -1
 endi
 
-if $data03 != 11 then
+if $data03 != 12 then
    return -1
 endi
 
-if $data04 != @abc11@ then
+if $data04 != @abc12@ then
    return -1
 endi
 
@@ -248,8 +248,8 @@ if $data04 != @abc12@ then
    return -1
 endi
 
-sql select bottom(c1, 100), tbname, t1, t2 from select_tags_mt0;
-if $rows != 100 then
+sql select bottom(c1, 72), tbname, t1, t2 from select_tags_mt0;
+if $rows != 72 then
    return -1
 endi