Merge pull request #9548 from taosdata/fix/TS-802-D1

[TS-802]<fix>(query): first last API support order by desc in sub query

src/client/src/tscUtil.c

@@ -1479,6 +1479,18 @@ void handleDownstreamOperator(SSqlObj** pSqlObjList, int32_t numOfUpstream, SQue
           break;
         }
       }
+
+      // set input data order to param[1]
+      if(pex->base.functionId == TSDB_FUNC_FIRST || pex->base.functionId  ==  TSDB_FUNC_FIRST_DST ||
+         pex->base.functionId == TSDB_FUNC_LAST  || pex->base.functionId  ==  TSDB_FUNC_LAST_DST) {
+        // set input order
+        SQueryInfo* pInputQI = pSqlObjList[0]->cmd.pQueryInfo;
+        if(pInputQI) {
+          pex->base.numOfParams = 3;
+          pex->base.param[2].nType = TSDB_DATA_TYPE_INT;
+          pex->base.param[2].i64 = pInputQI->order.order;
+        }
+      }      
     }
 
     tscDebug("0x%"PRIx64" create QInfo 0x%"PRIx64" to execute the main query while all nest queries are ready", pSql->self, pSql->self);

src/query/src/qAggMain.c

@@ -1620,33 +1620,65 @@ static bool first_last_function_setup(SQLFunctionCtx *pCtx, SResultRowCellInfo*
 
 // todo opt for null block
 static void first_function(SQLFunctionCtx *pCtx) {
-  if (pCtx->order == TSDB_ORDER_DESC) {
-    return;
-  }
-  
+  SResultRowCellInfo* pResInfo = GET_RES_INFO(pCtx);
   int32_t notNullElems = 0;
-  
-  // handle the null value
-  for (int32_t i = 0; i < pCtx->size; ++i) {
-    char *data = GET_INPUT_DATA(pCtx, i);
-    if (pCtx->hasNull && isNull(data, pCtx->inputType)) {
-      continue;
-    }
-    
-    memcpy(pCtx->pOutput, data, pCtx->inputBytes);
-    if (pCtx->ptsList != NULL) {
-      TSKEY k = GET_TS_DATA(pCtx, i);
-      DO_UPDATE_TAG_COLUMNS(pCtx, k);
+  int32_t step = 1;
+  int32_t i = 0;
+  bool inputAsc = true;
+
+  // input data come from sub query, input data order equal to sub query order
+  if(pCtx->numOfParams == 3) {
+    if(pCtx->param[2].nType == TSDB_DATA_TYPE_INT && pCtx->param[2].i64 == TSDB_ORDER_DESC) {
+      step = -1;
+      i = pCtx->size - 1;
+      inputAsc = false;
+    }
+  } else if (pCtx->order == TSDB_ORDER_DESC) {
+    return ;
+  }
+
+  if(pCtx->order == TSDB_ORDER_ASC && inputAsc) {
+    for (int32_t m = 0; m < pCtx->size; ++m, i+=step) {
+      char *data = GET_INPUT_DATA(pCtx, i);
+      if (pCtx->hasNull && isNull(data, pCtx->inputType)) {
+        continue;
+      }
+      
+      memcpy(pCtx->pOutput, data, pCtx->inputBytes);
+      if (pCtx->ptsList != NULL) {
+        TSKEY k = GET_TS_DATA(pCtx, i);
+        DO_UPDATE_TAG_COLUMNS(pCtx, k);
+      }
+
+      SResultRowCellInfo *pInfo = GET_RES_INFO(pCtx);
+      pInfo->hasResult = DATA_SET_FLAG;
+      pInfo->complete = true;
+      
+      notNullElems++;
+      break;
     }
+  } else {  // desc order
+    for (int32_t m = 0; m < pCtx->size; ++m, i+=step) {
+      char *data = GET_INPUT_DATA(pCtx, i);
+      if (pCtx->hasNull && isNull(data, pCtx->inputType) && (!pCtx->requireNull)) {
+        continue;
+      }
 
-    SResultRowCellInfo *pInfo = GET_RES_INFO(pCtx);
-    pInfo->hasResult = DATA_SET_FLAG;
-    pInfo->complete = true;
-    
-    notNullElems++;
-    break;
+      TSKEY ts = pCtx->ptsList ? GET_TS_DATA(pCtx, i) : 0;
+
+      char* buf = GET_ROWCELL_INTERBUF(pResInfo);
+      if (pResInfo->hasResult != DATA_SET_FLAG || (*(TSKEY*)buf) > ts) {
+        pResInfo->hasResult = DATA_SET_FLAG;
+        memcpy(pCtx->pOutput, data, pCtx->inputBytes);
+
+        *(TSKEY*)buf = ts;
+        DO_UPDATE_TAG_COLUMNS(pCtx, ts);
+      }
+
+      notNullElems++;
+      break;
+    }
   }
-  
   SET_VAL(pCtx, notNullElems, 1);
 }
 
@@ -1730,16 +1762,23 @@ static void first_dist_func_merge(SQLFunctionCtx *pCtx) {
  *    least one data in this block that is not null.(TODO opt for this case)
  */
 static void last_function(SQLFunctionCtx *pCtx) {
-  if (pCtx->order != pCtx->param[0].i64) {
+  SResultRowCellInfo* pResInfo = GET_RES_INFO(pCtx);
+  int32_t notNullElems = 0;
+  int32_t step = -1;
+  int32_t i = pCtx->size - 1;
+
+  // input data come from sub query, input data order equal to sub query order
+  if(pCtx->numOfParams == 3) {
+    if(pCtx->param[2].nType == TSDB_DATA_TYPE_INT && pCtx->param[2].i64 == TSDB_ORDER_DESC) {
+      step = 1;
+      i = 0;
+    }
+  } else if (pCtx->order != pCtx->param[0].i64) {
     return;
   }
 
-  SResultRowCellInfo* pResInfo = GET_RES_INFO(pCtx);
-
-  int32_t notNullElems = 0;
   if (pCtx->order == TSDB_ORDER_DESC) {
-
-    for (int32_t i = pCtx->size - 1; i >= 0; --i) {
+    for (int32_t m = pCtx->size - 1; m >= 0; --m, i += step) {
       char *data = GET_INPUT_DATA(pCtx, i);
       if (pCtx->hasNull && isNull(data, pCtx->inputType) && (!pCtx->requireNull)) {
         continue;
@@ -1756,7 +1795,7 @@ static void last_function(SQLFunctionCtx *pCtx) {
       break;
     }
   } else {  // ascending order
-    for (int32_t i = pCtx->size - 1; i >= 0; --i) {
+    for (int32_t m = pCtx->size - 1; m >= 0; --m, i += step) {
       char *data = GET_INPUT_DATA(pCtx, i);
       if (pCtx->hasNull && isNull(data, pCtx->inputType) && (!pCtx->requireNull)) {
         continue;

src/query/src/qExecutor.c

@@ -5612,6 +5612,18 @@ static int32_t getTableScanOrder(STableScanInfo* pTableScanInfo) {
   return pTableScanInfo->order;
 }
 
+// check all SQLFunctionCtx is completed
+static bool allCtxCompleted(SOperatorInfo* pOperator, SQLFunctionCtx* pCtx) {
+  // only one false, return false
+  for(int32_t i = 0; i < pOperator->numOfOutput; i++) {
+    if(pCtx[i].resultInfo == NULL)
+      return false;
+    if(!pCtx[i].resultInfo->complete)
+      return false;
+  }
+  return true;
+}
+
 // this is a blocking operator
 static SSDataBlock* doAggregate(void* param, bool* newgroup) {
   SOperatorInfo* pOperator = (SOperatorInfo*) param;
@@ -5650,6 +5662,9 @@ static SSDataBlock* doAggregate(void* param, bool* newgroup) {
     // the pDataBlock are always the same one, no need to call this again
     setInputDataBlock(pOperator, pInfo->pCtx, pBlock, order);
     doAggregateImpl(pOperator, pQueryAttr->window.skey, pInfo->pCtx, pBlock);
+    // if all pCtx is completed, then query should be over
+    if(allCtxCompleted(pOperator, pInfo->pCtx))
+      break;    
   }
 
   doSetOperatorCompleted(pOperator);

tests/pytest/fulltest.sh

@@ -229,7 +229,8 @@ python3 test.py -f tools/taosdemoAllTest/taosdemoTestInsertAllType.py
 python3 test.py -f tools/taosdemoAllTest/taosdemoTestInsertShell.py
 
 #query
-python3 test.py -f query/distinctOneColTb.py
+python3 ./test.py -f query/queryBase.py
+python3 ./test.py -f query/distinctOneColTb.py
 python3 ./test.py -f query/filter.py
 python3 ./test.py -f query/filterCombo.py
 python3 ./test.py -f query/queryNormal.py

tests/pytest/query/queryBase.py

+###################################################################
+#           Copyright (c) 2016 by TAOS Technologies, Inc.
+#                     All rights reserved.
+#
+#  This file is proprietary and confidential to TAOS Technologies.
+#  No part of this file may be reproduced, stored, transmitted,
+#  disclosed or used in any form or by any means other than as
+#  expressly provided by the written permission from Jianhui Tao
+#
+###################################################################
+
+# -*- coding: utf-8 -*-
+#
+#     query base function test case
+#
+
+import sys
+
+from numpy.lib.function_base import insert
+import taos
+from util.log import *
+from util.cases import *
+from util.sql import *
+import numpy as np
+
+# constant define
+WAITS = 5 # wait seconds
+
+class TDTestCase:
+    #
+    # --------------- main frame -------------------
+    #
+    
+    def caseDescription(self):
+        '''
+        Query moudle base api or keyword test case:
+        case1: api first() last() 
+        case2: none
+        '''
+        return 
+
+    # init
+    def init(self, conn, logSql):
+        tdLog.debug("start to execute %s" % __file__)
+        tdSql.init(conn.cursor())
+        tdSql.prepare()
+        self.create_tables();
+        self.ts = 1500000000000
+
+
+    # run case  
+    def run(self):
+        # insert data
+        self.insert_data("t1", self.ts, 1*10000, 30000, 0);
+        self.insert_data("t2", self.ts, 2*10000, 30000, 100000);
+        self.insert_data("t3", self.ts, 3*10000, 30000, 200000);
+        # test base case
+        self.case_first()
+        tdLog.debug(" QUERYBASE first() api ............ [OK]")
+        # test advance case
+        self.case_last()
+        tdLog.debug(" QUERYBASE last() api  ............ [OK]")
+
+    # stop 
+    def stop(self):
+        tdSql.close()
+        tdLog.success("%s successfully executed" % __file__)
+
+    #
+    # --------------- case  -------------------
+    #
+
+    # create table
+    def create_tables(self):
+        # super table
+        tdSql.execute("create table st(ts timestamp, i1 int) tags(area int)");
+        # child table
+        tdSql.execute("create table t1 using st tags(1)");
+        tdSql.execute("create table t2 using st tags(2)");
+        tdSql.execute("create table t3 using st tags(3)");
+        return 
+
+    # insert data1
+    def insert_data(self, tbname, ts_start, count, batch_num, base):
+        pre_insert = "insert into %s values"%tbname
+        sql = pre_insert
+        tdLog.debug("doing insert table %s rows=%d ..."%(tbname, count))
+        for i in range(count):
+            sql += " (%d,%d)"%(ts_start + i*1000, base + i)
+            if i >0 and i%batch_num == 0:
+                tdSql.execute(sql)
+                sql = pre_insert
+        # end sql        
+        if sql != pre_insert:
+            tdSql.execute(sql)
+
+        tdLog.debug("INSERT TABLE DATA ............ [OK]")
+        return
+
+    # first case base 
+    def case_first(self):
+        # 
+        # last base function
+        #
+
+        # base t1 table
+        sql = "select first(*) from t1 where ts>='2017-07-14 12:40:00' order by ts asc;"
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 7200)
+        sql = "select first(*) from t1 where ts>='2017-07-14 12:40:00' order by ts desc;" # desc
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 7200)
+        # super table st
+        sql = "select first(*)  from st where ts>='2017-07-14 11:40:00' and ts<='2017-07-14 12:40:00' and tbname in('t1') order by ts;"
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 3600)
+        sql = "select first(*)  from st where ts>='2017-07-14 11:40:00' and ts<='2017-07-14 12:40:00' and tbname in('t1') order by ts desc;" # desc
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 3600)
+        # sub query
+        sql = "select first(*) from ( select sum(i1)  from st where ts>='2017-07-14 11:40:00' and ts<'2017-07-14 12:40:00' interval(10m) order by ts asc );"
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 187019100)
+        sql = "select first(*) from ( select sum(i1)  from st where ts>='2017-07-14 11:40:00' and ts<'2017-07-14 12:40:00' interval(10m) order by ts desc );" # desc
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 187019100)
+        return 
+
+    # last case
+    def case_last(self):
+        # 
+        #  last base test
+        #
+        
+        # base t1 table
+        sql = "select last(*) from t1 where ts<='2017-07-14 12:40:00' order by ts asc;"
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 7200)
+        sql = "select last(*) from t1 where ts<='2017-07-14 12:40:00' order by ts desc;" # desc
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 7200)        
+        # super table st
+        sql = "select last(*)  from st where ts>='2017-07-14 11:40:00' and ts<='2017-07-14 12:40:00' and tbname in('t1') order by ts;"
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 7200)
+        sql = "select last(*)  from st where ts>='2017-07-14 11:40:00' and ts<='2017-07-14 12:40:00' and tbname in('t1') order by ts desc;" # desc
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 7200)
+        
+        # sub query
+        sql = "select last(*) from ( select sum(i1)  from st where ts>='2017-07-14 11:40:00' and ts<'2017-07-14 12:40:00'  interval(10m) order by ts asc );"
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 192419100)
+        sql = "select last(*) from ( select sum(i1)  from st where ts>='2017-07-14 11:40:00' and ts<'2017-07-14 12:40:00'  interval(10m) order by ts desc );" # desc
+        tdSql.waitedQuery(sql, 1, WAITS)
+        tdSql.checkData(0, 1, 192419100)
+ 
+
+#
+# add case with filename
+#
+tdCases.addWindows(__file__, TDTestCase())
+tdCases.addLinux(__file__, TDTestCase())
\ No newline at end of file