Merge pull request #4442 from taosdata/feature/query

[TD-2236]<fix>: fix bugs in groupby normal colunms + last query

src/client/src/tscFunctionImpl.c

@@ -54,8 +54,8 @@
 
 #define DO_UPDATE_TAG_COLUMNS(ctx, ts)                                           \
   do {                                                                           \
-    for (int32_t i = 0; i < (ctx)->tagInfo.numOfTagCols; ++i) {                  \
-      SQLFunctionCtx *__ctx = (ctx)->tagInfo.pTagCtxList[i];                     \
+    for (int32_t _i = 0; _i < (ctx)->tagInfo.numOfTagCols; ++_i) {                  \
+      SQLFunctionCtx *__ctx = (ctx)->tagInfo.pTagCtxList[_i];                     \
       if (__ctx->functionId == TSDB_FUNC_TS_DUMMY) {                             \
         __ctx->tag.i64Key = (ts); \
         __ctx->tag.nType = TSDB_DATA_TYPE_BIGINT; \
@@ -66,8 +66,8 @@
 
 #define DO_UPDATE_TAG_COLUMNS_WITHOUT_TS(ctx)                   \
   do {                                                          \
-    for (int32_t i = 0; i < (ctx)->tagInfo.numOfTagCols; ++i) { \
-      SQLFunctionCtx *__ctx = (ctx)->tagInfo.pTagCtxList[i];    \
+    for (int32_t _i = 0; _i < (ctx)->tagInfo.numOfTagCols; ++_i) { \
+      SQLFunctionCtx *__ctx = (ctx)->tagInfo.pTagCtxList[_i];    \
       aAggs[TSDB_FUNC_TAG].xFunction(__ctx);                    \
     }                                                           \
   } while (0);
@@ -305,7 +305,7 @@ int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionI
   } else if (functionId == TSDB_FUNC_FIRST || functionId == TSDB_FUNC_LAST) {
     *type = (int16_t)dataType;
     *bytes = (int16_t)dataBytes;
-    *interBytes = dataBytes;
+    *interBytes = (int16_t)(dataBytes + sizeof(SFirstLastInfo));
   } else if (functionId == TSDB_FUNC_SPREAD) {
     *type = (int16_t)TSDB_DATA_TYPE_DOUBLE;
     *bytes = sizeof(double);
@@ -1169,8 +1169,8 @@ static int32_t minmax_merge_impl(SQLFunctionCtx *pCtx, int32_t bytes, char *outp
         if ((*(int32_t *)output < v) ^ isMin) {
           *(int32_t *)output = v;
           
-          for (int32_t i = 0; i < pCtx->tagInfo.numOfTagCols; ++i) {
-            SQLFunctionCtx *__ctx = pCtx->tagInfo.pTagCtxList[i];
+          for (int32_t j = 0; j < pCtx->tagInfo.numOfTagCols; ++j) {
+            SQLFunctionCtx *__ctx = pCtx->tagInfo.pTagCtxList[j];
             aAggs[TSDB_FUNC_TAG].xFunction(__ctx);
           }
           
@@ -1680,6 +1680,12 @@ static void last_function_f(SQLFunctionCtx *pCtx, int32_t index) {
     return;
   }
 
+  // the scan order is not the required order, ignore it
+  if (pCtx->order != pCtx->param[0].i64Key) {
+    return;
+  }
+
+  if (pCtx->order == TSDB_ORDER_DESC) {
     SET_VAL(pCtx, 1, 1);
     memcpy(pCtx->aOutputBuf, pData, pCtx->inputBytes);
 
@@ -1689,6 +1695,19 @@ static void last_function_f(SQLFunctionCtx *pCtx, int32_t index) {
     SResultRowCellInfo *pResInfo = GET_RES_INFO(pCtx);
     pResInfo->hasResult = DATA_SET_FLAG;
     pResInfo->complete = true;  // set query completed
+  } else { // in case of ascending order check, all data needs to be checked
+    SResultRowCellInfo* pResInfo = GET_RES_INFO(pCtx);
+    TSKEY ts = pCtx->ptsList[index];
+
+    char* buf = GET_ROWCELL_INTERBUF(pResInfo);
+    if (pResInfo->hasResult != DATA_SET_FLAG || (*(TSKEY*)buf) < ts) {
+      pResInfo->hasResult = DATA_SET_FLAG;
+      memcpy(pCtx->aOutputBuf, pData, pCtx->inputBytes);
+
+      *(TSKEY*)buf = ts;
+      DO_UPDATE_TAG_COLUMNS(pCtx, ts);
+    }
+  }
 }
 
 static void last_data_assign_impl(SQLFunctionCtx *pCtx, char *pData, int32_t index) {
@@ -1711,7 +1730,7 @@ static void last_data_assign_impl(SQLFunctionCtx *pCtx, char *pData, int32_t ind
 
 static void last_dist_function(SQLFunctionCtx *pCtx) {
   /*
-   * 1. for scan data in asc order, no need to check data
+   * 1. for scan data is not the required order
    * 2. for data blocks that are not loaded, no need to check data
    */
   if (pCtx->order != pCtx->param[0].i64Key) {
@@ -2447,7 +2466,7 @@ static bool percentile_function_setup(SQLFunctionCtx *pCtx) {
 static void percentile_function(SQLFunctionCtx *pCtx) {
   int32_t notNullElems = 0;
   
-  SResultRowCellInfo *    pResInfo = GET_RES_INFO(pCtx);
+  SResultRowCellInfo *pResInfo = GET_RES_INFO(pCtx);
   SPercentileInfo *pInfo = GET_ROWCELL_INTERBUF(pResInfo);
 
   // the first stage, only acquire the min/max value
@@ -2548,12 +2567,14 @@ static void percentile_finalizer(SQLFunctionCtx *pCtx) {
   double v = pCtx->param[0].nType == TSDB_DATA_TYPE_INT ? pCtx->param[0].i64Key : pCtx->param[0].dKey;
   
   SResultRowCellInfo *pResInfo = GET_RES_INFO(pCtx);
-  tMemBucket * pMemBucket = ((SPercentileInfo *)GET_ROWCELL_INTERBUF(pResInfo))->pMemBucket;
+  SPercentileInfo* ppInfo = (SPercentileInfo *) GET_ROWCELL_INTERBUF(pResInfo);
 
-  if (pMemBucket->total > 0) {  // check for null
-    *(double *)pCtx->aOutputBuf = getPercentile(pMemBucket, v);
-  } else {
+  tMemBucket * pMemBucket = ppInfo->pMemBucket;
+  if (pMemBucket == NULL || pMemBucket->total == 0) {  // check for null
+    assert(ppInfo->numOfElems == 0);
     setNull(pCtx->aOutputBuf, pCtx->outputType, pCtx->outputBytes);
+  } else {
+    *(double *)pCtx->aOutputBuf = getPercentile(pMemBucket, v);
   }
   
   tMemBucketDestroy(pMemBucket);

tests/script/general/parser/groupby.sim

@@ -606,6 +606,44 @@ sql insert into t1 values ('2020-03-27 04:21:16.000', 1)('2020-03-27 04:31:17.00
 sql insert into t2 values ('2020-03-27 04:11:16.000', 1)('2020-03-27 04:11:17.000', 2) ('2020-03-27 04:11:18.000', 3) ('2020-03-27 04:11:19.000', 4) ; 
 sql insert into t2 values ('2020-03-27 04:21:16.000', 1)('2020-03-27 04:31:17.000', 2) ('2020-03-27 04:51:18.000', 3) ('2020-03-27 05:10:19.000', 4) ; 
 
+print =================>td-2236
+sql select first(ts),last(ts) from t1 group by c;
+if $rows != 4 then
+  return -1
+endi
+
+if $data00 != @20-03-27 04:11:16.000@ then
+  return -1
+endi
+
+if $data01 != @20-03-27 04:21:16.000@ then
+  return -1
+endi
+
+if $data10 != @20-03-27 04:11:17.000@ then
+  return -1
+endi
+
+if $data11 != @20-03-27 04:31:17.000@ then
+  return -1
+endi
+
+if $data20 != @20-03-27 04:11:18.000@ then
+  return -1
+endi
+
+if $data21 != @20-03-27 04:51:18.000@ then
+  return -1
+endi
+
+if $data30 != @20-03-27 04:11:19.000@ then
+  return -1
+endi
+
+if $data31 != @20-03-27 05:10:19.000@ then
+  return -1
+endi
+
 #sql select irate(c) from st where t1="1" and ts >= '2020-03-27 04:11:17.732' and ts < '2020-03-27 05:11:17.732'  interval(1m) sliding(15s) group by tbname,t1,t2;
 #if $rows != 40 then
 #  return -1