feat(query): add histogram param parsing

source/libs/function/src/builtinsimpl.c

@@ -14,13 +14,14 @@
  */
 
 #include "builtinsimpl.h"
+#include "cJSON.h"
 #include "function.h"
 #include "querynodes.h"
 #include "taggfunction.h"
 #include "tdatablock.h"
 #include "tpercentile.h"
 
-#define HISTOGRAM_MAX_BINS_NUM 100
+#define HISTOGRAM_MAX_BINS_NUM   100
 
 typedef struct SSumRes {
   union {
@@ -106,6 +107,13 @@ typedef struct SHistoFuncInfo {
   SHistoFuncBin bins[];
 } SHistoFuncInfo;
 
+typedef enum {
+  UNKNOWN_BIN = 0,
+  USER_INPUT_BIN,
+  LINEAR_BIN,
+  LOG_BIN
+} EHistoBinType;
+
 
 #define SET_VAL(_info, numOfElem, res) \
   do {                                 \
@@ -1801,16 +1809,165 @@ bool getHistogramFuncEnv(SFunctionNode* UNUSED_PARAM(pFunc), SFuncExecEnv* pEnv)
   return true;
 }
 
-bool histogramFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo) {
+static int8_t getHistogramBinType(char *binTypeStr) {
+  int8_t binType;
+  if (strcasecmp(binTypeStr, "user_input") == 0) {
+    binType = USER_INPUT_BIN;
+  } else if (strcasecmp(binTypeStr, "linear_bin") == 0) {
+    binType = LINEAR_BIN;
+  } else if (strcasecmp(binTypeStr, "log_bin") == 0) {
+    binType = LOG_BIN;
+  } else {
+    binType = UNKNOWN_BIN;
+  }
+
+  return binType;
+}
+
+static bool getHistogramBinDesc(SHistoFuncInfo *pInfo, char *binDescStr, int8_t binType, bool normalized) {
+  cJSON*  binDesc = cJSON_Parse(binDescStr);
+  int32_t counter;
+  int32_t numOfBins;
+  double* intervals;
+  if (cJSON_IsObject(binDesc)) { /* linaer/log bins */
+    int32_t numOfParams = cJSON_GetArraySize(binDesc);
+    int32_t startIndex;
+    if (numOfParams != 4) {
+      return false;
+    }
+
+    cJSON* start    = cJSON_GetObjectItem(binDesc, "start");
+    cJSON* factor   = cJSON_GetObjectItem(binDesc, "factor");
+    cJSON* width    = cJSON_GetObjectItem(binDesc, "width");
+    cJSON* count    = cJSON_GetObjectItem(binDesc, "count");
+    cJSON* infinity = cJSON_GetObjectItem(binDesc, "infinity");
+
+    if (!cJSON_IsNumber(start) || !cJSON_IsNumber(count) || !cJSON_IsBool(infinity)) {
+      return false;
+    }
+
+    if (count->valueint <= 0 || count->valueint > 1000) { // limit count to 1000
+      return false;
+    }
+
+    if (isinf(start->valuedouble) || (width != NULL && isinf(width->valuedouble)) ||
+        (factor != NULL && isinf(factor->valuedouble)) || (count != NULL && isinf(count->valuedouble))) {
+      return false;
+    }
+
+    counter = (int32_t)count->valueint;
+    if (infinity->valueint == false) {
+      startIndex = 0;
+      numOfBins = counter + 1;
+    } else {
+      startIndex = 1;
+      numOfBins = counter + 3;
+    }
+
+    intervals = taosMemoryCalloc(numOfBins, sizeof(double));
+    if (cJSON_IsNumber(width) && factor == NULL && binType == LINEAR_BIN) {
+      // linear bin process
+      if (width->valuedouble == 0) {
+        taosMemoryFree(intervals);
+        return false;
+      }
+      for (int i = 0; i < counter + 1; ++i) {
+        intervals[startIndex] = start->valuedouble + i * width->valuedouble;
+        if (isinf(intervals[startIndex])) {
+          taosMemoryFree(intervals);
+          return false;
+        }
+        startIndex++;
+      }
+    } else if (cJSON_IsNumber(factor) && width == NULL && binType == LOG_BIN) {
+      // log bin process
+      if (start->valuedouble == 0) {
+        taosMemoryFree(intervals);
+        return false;
+      }
+      if (factor->valuedouble < 0 || factor->valuedouble == 0 || factor->valuedouble == 1) {
+        taosMemoryFree(intervals);
+        return false;
+      }
+      for (int i = 0; i < counter + 1; ++i) {
+        intervals[startIndex] = start->valuedouble * pow(factor->valuedouble, i * 1.0);
+        if (isinf(intervals[startIndex])) {
+          taosMemoryFree(intervals);
+          return false;
+        }
+        startIndex++;
+      }
+    } else {
+      taosMemoryFree(intervals);
+      return false;
+    }
+
+    if (infinity->valueint == true) {
+      intervals[0] = -INFINITY;
+      intervals[numOfBins - 1] = INFINITY;
+      // in case of desc bin orders, -inf/inf should be swapped
+      ASSERT(numOfBins >= 4);
+      if (intervals[1] > intervals[numOfBins - 2]) {
+        TSWAP(intervals[0], intervals[numOfBins - 1]);
+      }
+    }
+  } else if (cJSON_IsArray(binDesc)) { /* user input bins */
+    if (binType != USER_INPUT_BIN) {
+      return false;
+    }
+    counter = numOfBins = cJSON_GetArraySize(binDesc);
+    intervals = taosMemoryCalloc(numOfBins, sizeof(double));
+    cJSON* bin = binDesc->child;
+    if (bin == NULL) {
+      taosMemoryFree(intervals);
+      return false;
+    }
+    int i = 0;
+    while (bin) {
+      intervals[i] = bin->valuedouble;
+      if (!cJSON_IsNumber(bin)) {
+        taosMemoryFree(intervals);
+        return false;
+      }
+      if (i != 0 && intervals[i] <= intervals[i - 1]) {
+        taosMemoryFree(intervals);
+        return false;
+      }
+      bin = bin->next;
+      i++;
+    }
+  } else {
+    return false;
+  }
+
+  pInfo->numOfBins  = numOfBins;
+  pInfo->normalized = normalized;
+  for (int32_t i = 0; i < numOfBins; ++i) {
+    pInfo->bins[i].lower = intervals[i] < intervals[i + 1] ? intervals[i] : intervals[i + 1];
+    pInfo->bins[i].upper = intervals[i + 1] > intervals[i] ? intervals[i + 1] : intervals[i];
+    pInfo->bins[i].count = 0;
+  }
+
+  taosMemoryFree(intervals);
+  return true;
+}
+
+bool histogramFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo *pResultInfo) {
   if (!functionSetup(pCtx, pResultInfo)) {
     return false;
   }
 
-  SHistoFuncInfo* pInfo = GET_ROWCELL_INTERBUF(pResultInfo);
-  char* binType = pCtx->param[1].param.pz;
-  char* binDesc = pCtx->param[2].param.pz;
-  int64_t nornalized = pCtx->param[3].param.i;
+  SHistoFuncInfo *pInfo = GET_ROWCELL_INTERBUF(pResultInfo);
 
+  int8_t binType = getHistogramBinType(varDataVal(pCtx->param[1].param.pz));
+  if (binType == UNKNOWN_BIN) {
+    return false;
+  }
+  char* binDesc = varDataVal(pCtx->param[2].param.pz);
+  int64_t normalized = pCtx->param[3].param.i;
+  if (!getHistogramBinDesc(pInfo, binDesc, binType, (bool)normalized)) {
+    return false;
+  }
 
   return true;
 }