• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

taosdata / TDengine / #3621

22 Feb 2025 11:44AM UTC coverage: 2.037% (-61.5%) from 63.573%
#3621

push

travis-ci

web-flow
Merge pull request #29874 from taosdata/merge/mainto3.0

merge: from main to 3.0 branch

4357 of 287032 branches covered (1.52%)

Branch coverage included in aggregate %.

0 of 174 new or added lines in 18 files covered. (0.0%)

213359 existing lines in 469 files now uncovered.

7260 of 283369 relevant lines covered (2.56%)

23737.72 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

0.0
/source/libs/executor/src/hashjoinoperator.c
1
/*
2
 * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
3
 *
4
 * This program is free software: you can use, redistribute, and/or modify
5
 * it under the terms of the GNU Affero General Public License, version 3
6
 * or later ("AGPL"), as published by the Free Software Foundation.
7
 *
8
 * This program is distributed in the hope that it will be useful, but WITHOUT
9
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10
 * FITNESS FOR A PARTICULAR PURPOSE.
11
 *
12
 * You should have received a copy of the GNU Affero General Public License
13
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
14
 */
15

16
#include "executorInt.h"
17
#include "filter.h"
18
#include "function.h"
19
#include "operator.h"
20
#include "os.h"
21
#include "querynodes.h"
22
#include "querytask.h"
23
#include "tcompare.h"
24
#include "tdatablock.h"
25
#include "thash.h"
26
#include "tmsg.h"
27
#include "ttypes.h"
28
#include "hashjoin.h"
29
#include "functionMgt.h"
30

31

UNCOV
32
bool hJoinBlkReachThreshold(SHJoinOperatorInfo* pInfo, int64_t blkRows) {
×
UNCOV
33
  if (INT64_MAX == pInfo->ctx.limit || pInfo->pFinFilter != NULL) {
×
UNCOV
34
    return blkRows >= pInfo->blkThreshold;
×
35
  }
36
  
37
  return (pInfo->execInfo.resRows + blkRows) >= pInfo->ctx.limit;
×
38
}
39

40
int32_t hJoinHandleMidRemains(SHJoinOperatorInfo* pJoin, SHJoinCtx* pCtx) {
×
41
  TSWAP(pJoin->midBlk, pJoin->finBlk);
×
42

43
  pCtx->midRemains = false;
×
44

45
  return TSDB_CODE_SUCCESS;
×
46
}
47

48
int32_t hJoinCopyMergeMidBlk(SHJoinCtx* pCtx, SSDataBlock** ppMid, SSDataBlock** ppFin) {
×
49
  SSDataBlock* pLess = *ppMid;
×
50
  SSDataBlock* pMore = *ppFin;
×
51

52
/*
53
  if ((*ppMid)->info.rows < (*ppFin)->info.rows) {
54
    pLess = (*ppMid);
55
    pMore = (*ppFin);
56
  } else {
57
    pLess = (*ppFin);
58
    pMore = (*ppMid);
59
  }
60
*/
61

62
  int32_t totalRows = pMore->info.rows + pLess->info.rows;
×
63
  if (totalRows <= pMore->info.capacity) {
×
64
    HJ_ERR_RET(blockDataMerge(pMore, pLess));
×
65
    blockDataCleanup(pLess);
×
66
    pCtx->midRemains = false;
×
67
  } else {
68
    int32_t copyRows = pMore->info.capacity - pMore->info.rows;
×
69
    HJ_ERR_RET(blockDataMergeNRows(pMore, pLess, pLess->info.rows - copyRows, copyRows));
×
70
    blockDataShrinkNRows(pLess, copyRows);
×
71
    pCtx->midRemains = true;
×
72
  }
73

74
/*
75
  if (pMore != (*ppFin)) {
76
    TSWAP(*ppMid, *ppFin);
77
  }
78
*/
79

80
  return TSDB_CODE_SUCCESS;
×
81
}
82

UNCOV
83
int32_t hJoinSetImplFp(SHJoinOperatorInfo* pJoin) {
×
UNCOV
84
  switch (pJoin->joinType) {
×
UNCOV
85
    case JOIN_TYPE_INNER:
×
UNCOV
86
      pJoin->joinFp = hInnerJoinDo;
×
UNCOV
87
      break;
×
88
    case JOIN_TYPE_LEFT:
×
89
    case JOIN_TYPE_RIGHT: {
90
      switch (pJoin->subType) {
×
91
        case JOIN_STYPE_OUTER:          
×
92
          //pJoin->joinFp = hLeftJoinDo; TOOPEN
93
          break;
×
94
        default:
×
95
          break;
×
96
      }
97
      break;
×
98
    }      
99
    default:
×
100
      break;
×
101
  }
102

UNCOV
103
  return TSDB_CODE_SUCCESS;
×
104
}
105

106

UNCOV
107
int32_t hJoinLaunchPrimExpr(SSDataBlock* pBlock, SHJoinTableCtx* pTable, int32_t startIdx, int32_t endIdx) {
×
UNCOV
108
  if (NULL == pTable->primExpr) {
×
UNCOV
109
    return TSDB_CODE_SUCCESS;
×
110
  }
111

112
  SHJoinPrimExprCtx* pCtx = &pTable->primCtx;
×
113
  SColumnInfoData* pPrimIn = taosArrayGet(pBlock->pDataBlock, pTable->primCol->srcSlot);
×
114
  SColumnInfoData* pPrimOut = taosArrayGet(pBlock->pDataBlock, pTable->primCtx.targetSlotId);
×
115
  if (0 != pCtx->timezoneUnit) {
×
116
    for (int32_t i = startIdx; i <= endIdx; ++i) {
×
117
      ((int64_t*)pPrimOut->pData)[i] = ((int64_t*)pPrimIn->pData)[i] - (((int64_t*)pPrimIn->pData)[i] + pCtx->timezoneUnit) % pCtx->truncateUnit;
×
118
    }
119
  } else {
120
    for (int32_t i = startIdx; i <= endIdx; ++i) {
×
121
      ((int64_t*)pPrimOut->pData)[i] = ((int64_t*)pPrimIn->pData)[i] / pCtx->truncateUnit * pCtx->truncateUnit;
×
122
    }
123
  }
124

125
  return TSDB_CODE_SUCCESS;
×
126
}
127

128

129
static int64_t hJoinGetSingleKeyRowsNum(SBufRowInfo* pRow) {
×
130
  int64_t rows = 0;
×
131
  while (pRow) {
×
132
    rows++;
×
133
    pRow = pRow->next;
×
134
  }
135
  return rows;
×
136
}
137

138
static int64_t hJoinGetRowsNumOfKeyHash(SSHashObj* pHash) {
×
139
  SGroupData* pGroup = NULL;
×
140
  int32_t iter = 0;
×
141
  int64_t rowsNum = 0;
×
142
  
143
  while (NULL != (pGroup = tSimpleHashIterate(pHash, pGroup, &iter))) {
×
144
    int32_t* pKey = tSimpleHashGetKey(pGroup, NULL);
×
145
    int64_t rows = hJoinGetSingleKeyRowsNum(pGroup->rows);
×
146
    //qTrace("build_key:%d, rows:%" PRId64, *pKey, rows);
147
    rowsNum += rows;
×
148
  }
149

150
  return rowsNum;
×
151
}
152

UNCOV
153
static int32_t hJoinInitKeyColsInfo(SHJoinTableCtx* pTable, SNodeList* pList) {
×
UNCOV
154
  pTable->keyNum = LIST_LENGTH(pList);
×
155
  
UNCOV
156
  pTable->keyCols = taosMemoryMalloc(pTable->keyNum * sizeof(SHJoinColInfo));
×
UNCOV
157
  if (NULL == pTable->keyCols) {
×
158
    return terrno;
×
159
  }
160

UNCOV
161
  int64_t bufSize = 0;
×
UNCOV
162
  int32_t i = 0;
×
UNCOV
163
  SNode* pNode = NULL;
×
UNCOV
164
  FOREACH(pNode, pList) {
×
UNCOV
165
    SColumnNode* pColNode = (SColumnNode*)pNode;
×
UNCOV
166
    pTable->keyCols[i].srcSlot = pColNode->slotId;
×
UNCOV
167
    pTable->keyCols[i].vardata = IS_VAR_DATA_TYPE(pColNode->node.resType.type);
×
UNCOV
168
    pTable->keyCols[i].bytes = pColNode->node.resType.bytes;
×
UNCOV
169
    bufSize += pColNode->node.resType.bytes;
×
UNCOV
170
    ++i;
×
171
  }  
172

UNCOV
173
  if (pTable->keyNum > 1) {
×
UNCOV
174
    pTable->keyBuf = taosMemoryMalloc(bufSize);
×
UNCOV
175
    if (NULL == pTable->keyBuf) {
×
176
      return terrno;
×
177
    }
178
  }
179

UNCOV
180
  return TSDB_CODE_SUCCESS;
×
181
}
182

UNCOV
183
static void hJoinGetValColsNum(SNodeList* pList, int32_t blkId, int32_t* colNum) {
×
UNCOV
184
  *colNum = 0;
×
185
  
UNCOV
186
  SNode* pNode = NULL;
×
UNCOV
187
  FOREACH(pNode, pList) {
×
UNCOV
188
    STargetNode* pTarget = (STargetNode*)pNode;
×
UNCOV
189
    SColumnNode* pCol = (SColumnNode*)pTarget->pExpr;
×
UNCOV
190
    if (pCol->dataBlockId == blkId) {
×
UNCOV
191
      (*colNum)++;
×
192
    }
193
  }
UNCOV
194
}
×
195

UNCOV
196
static bool hJoinIsValColInKeyCols(int16_t slotId, int32_t keyNum, SHJoinColInfo* pKeys, int32_t* pKeyIdx) {
×
UNCOV
197
  for (int32_t i = 0; i < keyNum; ++i) {
×
UNCOV
198
    if (pKeys[i].srcSlot == slotId) {
×
199
      *pKeyIdx = i;
×
200
      return true;
×
201
    }
202
  }
203

UNCOV
204
  return false;
×
205
}
206

UNCOV
207
static int32_t hJoinInitValColsInfo(SHJoinTableCtx* pTable, SNodeList* pList) {
×
UNCOV
208
  hJoinGetValColsNum(pList, pTable->blkId, &pTable->valNum);
×
UNCOV
209
  if (pTable->valNum == 0) {
×
210
    return TSDB_CODE_SUCCESS;
×
211
  }
212
  
UNCOV
213
  pTable->valCols = taosMemoryMalloc(pTable->valNum * sizeof(SHJoinColInfo));
×
UNCOV
214
  if (NULL == pTable->valCols) {
×
215
    return terrno;
×
216
  }
217

UNCOV
218
  int32_t i = 0;
×
UNCOV
219
  int32_t colNum = 0;
×
UNCOV
220
  SNode* pNode = NULL;
×
UNCOV
221
  FOREACH(pNode, pList) {
×
UNCOV
222
    STargetNode* pTarget = (STargetNode*)pNode;
×
UNCOV
223
    SColumnNode* pColNode = (SColumnNode*)pTarget->pExpr;
×
UNCOV
224
    if (pColNode->dataBlockId == pTable->blkId) {
×
UNCOV
225
      if (hJoinIsValColInKeyCols(pColNode->slotId, pTable->keyNum, pTable->keyCols, &pTable->valCols[i].srcSlot)) {
×
226
        pTable->valCols[i].keyCol = true;
×
227
      } else {
UNCOV
228
        pTable->valCols[i].keyCol = false;
×
UNCOV
229
        pTable->valCols[i].srcSlot = pColNode->slotId;
×
UNCOV
230
        pTable->valColExist = true;
×
UNCOV
231
        colNum++;
×
232
      }
UNCOV
233
      pTable->valCols[i].dstSlot = pTarget->slotId;
×
UNCOV
234
      pTable->valCols[i].vardata = IS_VAR_DATA_TYPE(pColNode->node.resType.type);
×
UNCOV
235
      if (pTable->valCols[i].vardata && !pTable->valCols[i].keyCol) {
×
236
        if (NULL == pTable->valVarCols) {
×
237
          pTable->valVarCols = taosArrayInit(pTable->valNum, sizeof(int32_t));
×
238
          if (NULL == pTable->valVarCols) {
×
239
            return terrno;
×
240
          }
241
        }
242
        if (NULL == taosArrayPush(pTable->valVarCols, &i)) {
×
243
          return terrno;
×
244
        }
245
      }
UNCOV
246
      pTable->valCols[i].bytes = pColNode->node.resType.bytes;
×
UNCOV
247
      if (!pTable->valCols[i].keyCol && !pTable->valCols[i].vardata) {
×
UNCOV
248
        pTable->valBufSize += pColNode->node.resType.bytes;
×
249
      }
UNCOV
250
      i++;
×
251
    }
252
  }
253

UNCOV
254
  pTable->valBitMapSize = BitmapLen(colNum);
×
UNCOV
255
  pTable->valBufSize += pTable->valBitMapSize;
×
256

UNCOV
257
  return TSDB_CODE_SUCCESS;
×
258
}
259

UNCOV
260
static int32_t hJoinInitPrimKeyInfo(SHJoinTableCtx* pTable, int32_t slotId) {
×
UNCOV
261
  pTable->primCol = taosMemoryMalloc(sizeof(SHJoinColMap));
×
UNCOV
262
  if (NULL == pTable->primCol) {
×
263
    return terrno;
×
264
  }
265

UNCOV
266
  pTable->primCol->srcSlot = slotId;
×
267

UNCOV
268
  return TSDB_CODE_SUCCESS;
×
269
}
270

271

UNCOV
272
static int32_t hJoinInitPrimExprCtx(SNode* pNode, SHJoinPrimExprCtx* pCtx, SHJoinTableCtx* pTable) {
×
UNCOV
273
  if (NULL == pNode) {
×
UNCOV
274
    pCtx->targetSlotId = pTable->primCol->srcSlot;
×
UNCOV
275
    return TSDB_CODE_SUCCESS;
×
276
  }
277
  
278
  if (QUERY_NODE_TARGET != nodeType(pNode)) {
×
279
    return TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR;
×
280
  }  
281

282
  STargetNode* pTarget = (STargetNode*)pNode;
×
283
  if (QUERY_NODE_FUNCTION != nodeType(pTarget->pExpr)) {
×
284
    return TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR;
×
285
  }
286

287
  SFunctionNode* pFunc = (SFunctionNode*)pTarget->pExpr;
×
288
  if (FUNCTION_TYPE_TIMETRUNCATE != pFunc->funcType) {
×
289
    return TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR;
×
290
  }
291

292
  if (4 != pFunc->pParameterList->length && 5 != pFunc->pParameterList->length) {
×
293
    return TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR;
×
294
  }
295

296
  SValueNode* pUnit = (SValueNode*)nodesListGetNode(pFunc->pParameterList, 1);
×
297
  SValueNode* pCurrTz = (5 == pFunc->pParameterList->length) ? (SValueNode*)nodesListGetNode(pFunc->pParameterList, 2) : NULL;
×
298
  SValueNode* pTimeZone = (5 == pFunc->pParameterList->length) ? (SValueNode*)nodesListGetNode(pFunc->pParameterList, 4) : (SValueNode*)nodesListGetNode(pFunc->pParameterList, 3);
×
299

300
  pCtx->truncateUnit = pUnit->typeData;
×
301
  if ((NULL == pCurrTz || 1 == pCurrTz->typeData) && pCtx->truncateUnit >= (86400 * TSDB_TICK_PER_SECOND(pFunc->node.resType.precision))) {
×
302
    pCtx->timezoneUnit = offsetFromTz(varDataVal(pTimeZone->datum.p), TSDB_TICK_PER_SECOND(pFunc->node.resType.precision));
×
303
  }
304

305
  pCtx->targetSlotId = pTarget->slotId;
×
306

307
  return TSDB_CODE_SUCCESS;
×
308
}
309

310

UNCOV
311
static int32_t hJoinInitTableInfo(SHJoinOperatorInfo* pJoin, SHashJoinPhysiNode* pJoinNode, SOperatorInfo** pDownstream, int32_t idx, SQueryStat* pStat) {
×
UNCOV
312
  SNodeList* pKeyList = NULL;
×
UNCOV
313
  SHJoinTableCtx* pTable = &pJoin->tbs[idx];
×
UNCOV
314
  pTable->downStream = pDownstream[idx];
×
UNCOV
315
  pTable->blkId = pDownstream[idx]->resultDataBlockId;
×
UNCOV
316
  if (0 == idx) {
×
UNCOV
317
    pKeyList = pJoinNode->pOnLeft;
×
UNCOV
318
    pTable->hasTimeRange = pJoinNode->timeRangeTarget & 0x1;
×
319
  } else {
UNCOV
320
    pKeyList = pJoinNode->pOnRight;
×
UNCOV
321
    pTable->hasTimeRange = pJoinNode->timeRangeTarget & 0x2;
×
322
  }
323

UNCOV
324
  HJ_ERR_RET(hJoinInitPrimKeyInfo(pTable, (0 == idx) ? pJoinNode->leftPrimSlotId : pJoinNode->rightPrimSlotId));
×
325
  
UNCOV
326
  int32_t code = hJoinInitKeyColsInfo(pTable, pKeyList);
×
UNCOV
327
  if (code) {
×
328
    return code;
×
329
  }
UNCOV
330
  code = hJoinInitValColsInfo(pTable, pJoinNode->pTargets);
×
UNCOV
331
  if (code) {
×
332
    return code;
×
333
  }
334

UNCOV
335
  TAOS_MEMCPY(&pTable->inputStat, pStat, sizeof(*pStat));
×
336

UNCOV
337
  HJ_ERR_RET(hJoinInitPrimExprCtx(pTable->primExpr, &pTable->primCtx, pTable));
×
338

UNCOV
339
  return TSDB_CODE_SUCCESS;
×
340
}
341

UNCOV
342
static void hJoinSetBuildAndProbeTable(SHJoinOperatorInfo* pInfo, SHashJoinPhysiNode* pJoinNode) {
×
UNCOV
343
  int32_t buildIdx = 0;
×
UNCOV
344
  int32_t probeIdx = 1;
×
345

UNCOV
346
  pInfo->joinType = pJoinNode->joinType;
×
UNCOV
347
  pInfo->subType = pJoinNode->subType;
×
348
  
UNCOV
349
  switch (pInfo->joinType) {
×
UNCOV
350
    case JOIN_TYPE_INNER:
×
351
    case JOIN_TYPE_FULL:
UNCOV
352
      if (pInfo->tbs[0].inputStat.inputRowNum <= pInfo->tbs[1].inputStat.inputRowNum) {
×
UNCOV
353
        buildIdx = 0;
×
UNCOV
354
        probeIdx = 1;
×
355
      } else {
356
        buildIdx = 1;
×
357
        probeIdx = 0;
×
358
      }
UNCOV
359
      break;
×
360
    case JOIN_TYPE_LEFT:
×
361
      buildIdx = 1;
×
362
      probeIdx = 0;
×
363
      break;
×
364
    case JOIN_TYPE_RIGHT:
×
365
      buildIdx = 0;
×
366
      probeIdx = 1;
×
367
      break;
×
368
    default:
×
369
      break;
×
370
  } 
371
  
UNCOV
372
  pInfo->pBuild = &pInfo->tbs[buildIdx];
×
UNCOV
373
  pInfo->pProbe = &pInfo->tbs[probeIdx];
×
374
  
UNCOV
375
  pInfo->pBuild->downStreamIdx = buildIdx;
×
UNCOV
376
  pInfo->pProbe->downStreamIdx = probeIdx;
×
377

UNCOV
378
  if (0 == buildIdx) {
×
UNCOV
379
    pInfo->pBuild->primExpr = pJoinNode->leftPrimExpr;
×
UNCOV
380
    pInfo->pProbe->primExpr = pJoinNode->rightPrimExpr;
×
381
  } else {
382
    pInfo->pBuild->primExpr = pJoinNode->rightPrimExpr;
×
383
    pInfo->pProbe->primExpr = pJoinNode->leftPrimExpr;
×
384
  }
UNCOV
385
}
×
386

UNCOV
387
static int32_t hJoinBuildResColsMap(SHJoinOperatorInfo* pInfo, SHashJoinPhysiNode* pJoinNode) {
×
UNCOV
388
  pInfo->pResColNum = pJoinNode->pTargets->length;
×
UNCOV
389
  pInfo->pResColMap = taosMemoryCalloc(pJoinNode->pTargets->length, sizeof(int8_t));
×
UNCOV
390
  if (NULL == pInfo->pResColMap) {
×
391
    return terrno;
×
392
  }
393
  
UNCOV
394
  SNode* pNode = NULL;
×
UNCOV
395
  int32_t i = 0;
×
UNCOV
396
  FOREACH(pNode, pJoinNode->pTargets) {
×
UNCOV
397
    STargetNode* pTarget = (STargetNode*)pNode;
×
UNCOV
398
    SColumnNode* pCol = (SColumnNode*)pTarget->pExpr;
×
UNCOV
399
    if (pCol->dataBlockId == pInfo->pBuild->blkId) {
×
UNCOV
400
      pInfo->pResColMap[i] = 1;
×
401
    }
402
    
UNCOV
403
    i++;
×
404
  }
405

UNCOV
406
  return TSDB_CODE_SUCCESS;
×
407
}
408

409

410
static FORCE_INLINE int32_t hJoinAddPageToBufs(SArray* pRowBufs) {
411
  SBufPageInfo page;
UNCOV
412
  page.pageSize = HASH_JOIN_DEFAULT_PAGE_SIZE;
×
UNCOV
413
  page.offset = 0;
×
UNCOV
414
  page.data = taosMemoryMalloc(page.pageSize);
×
UNCOV
415
  if (NULL == page.data) {
×
416
    return terrno;
×
417
  }
418

UNCOV
419
  if (NULL == taosArrayPush(pRowBufs, &page)) {
×
420
    return terrno;
×
421
  }
UNCOV
422
  return TSDB_CODE_SUCCESS;
×
423
}
424

UNCOV
425
static int32_t hJoinInitBufPages(SHJoinOperatorInfo* pInfo) {
×
UNCOV
426
  pInfo->pRowBufs = taosArrayInit(32, sizeof(SBufPageInfo));
×
UNCOV
427
  if (NULL == pInfo->pRowBufs) {
×
428
    return terrno;
×
429
  }
430

UNCOV
431
  return hJoinAddPageToBufs(pInfo->pRowBufs);
×
432
}
433

UNCOV
434
static void hJoinFreeTableInfo(SHJoinTableCtx* pTable) {
×
UNCOV
435
  taosMemoryFreeClear(pTable->keyCols);
×
UNCOV
436
  taosMemoryFreeClear(pTable->keyBuf);
×
UNCOV
437
  taosMemoryFreeClear(pTable->valCols);
×
UNCOV
438
  taosArrayDestroy(pTable->valVarCols);
×
UNCOV
439
  taosMemoryFree(pTable->primCol);
×
UNCOV
440
}
×
441

UNCOV
442
static void hJoinFreeBufPage(void* param) {
×
UNCOV
443
  SBufPageInfo* pInfo = (SBufPageInfo*)param;
×
UNCOV
444
  taosMemoryFree(pInfo->data);
×
UNCOV
445
}
×
446

UNCOV
447
static void hJoinDestroyKeyHash(SSHashObj** ppHash) {
×
UNCOV
448
  if (NULL == ppHash || NULL == (*ppHash)) {
×
UNCOV
449
    return;
×
450
  }
451

UNCOV
452
  void*   pIte = NULL;
×
UNCOV
453
  int32_t iter = 0;
×
UNCOV
454
  while ((pIte = tSimpleHashIterate(*ppHash, pIte, &iter)) != NULL) {
×
UNCOV
455
    SGroupData* pGroup = pIte;
×
UNCOV
456
    SBufRowInfo* pRow = pGroup->rows;
×
UNCOV
457
    SBufRowInfo* pNext = NULL;
×
UNCOV
458
    while (pRow) {
×
UNCOV
459
      pNext = pRow->next;
×
UNCOV
460
      taosMemoryFree(pRow);
×
UNCOV
461
      pRow = pNext;
×
462
    }
463
  }
464

UNCOV
465
  tSimpleHashCleanup(*ppHash);
×
UNCOV
466
  *ppHash = NULL;
×
467
}
468

469
static FORCE_INLINE int32_t hJoinRetrieveColDataFromRowBufs(SArray* pRowBufs, SBufRowInfo* pRow, char** ppData) {
UNCOV
470
  *ppData = NULL;
×
471
  
UNCOV
472
  if ((uint16_t)-1 == pRow->pageId) {
×
473
    return TSDB_CODE_SUCCESS;
×
474
  }
UNCOV
475
  SBufPageInfo *pPage = taosArrayGet(pRowBufs, pRow->pageId);
×
UNCOV
476
  if (NULL == pPage) {
×
477
    qError("fail to get %d page, total:%d", pRow->pageId, (int32_t)taosArrayGetSize(pRowBufs));
×
478
    QRY_ERR_RET(TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR);
×
479
  }
480
  
UNCOV
481
  *ppData = pPage->data + pRow->offset;
×
482

UNCOV
483
  return TSDB_CODE_SUCCESS;
×
484
}
485

UNCOV
486
static int32_t hJoinCopyResRowsToBlock(SHJoinOperatorInfo* pJoin, int32_t rowNum, SBufRowInfo* pStart, SSDataBlock* pRes) {
×
UNCOV
487
  SHJoinTableCtx* pBuild = pJoin->pBuild;
×
UNCOV
488
  SHJoinTableCtx* pProbe = pJoin->pProbe;
×
UNCOV
489
  int32_t buildIdx = 0, buildValIdx = 0;
×
UNCOV
490
  int32_t probeIdx = 0;
×
UNCOV
491
  SBufRowInfo* pRow = pStart;
×
UNCOV
492
  int32_t code = 0;
×
UNCOV
493
  char* pData = NULL;
×
494

UNCOV
495
  for (int32_t r = 0; r < rowNum; ++r) {
×
UNCOV
496
    HJ_ERR_RET(hJoinRetrieveColDataFromRowBufs(pJoin->pRowBufs, pRow, &pData));
×
497
    
UNCOV
498
    char* pValData = pData + pBuild->valBitMapSize;
×
UNCOV
499
    char* pKeyData = pProbe->keyData;
×
UNCOV
500
    buildIdx = buildValIdx = probeIdx = 0;
×
UNCOV
501
    for (int32_t i = 0; i < pJoin->pResColNum; ++i) {
×
UNCOV
502
      if (pJoin->pResColMap[i]) {
×
UNCOV
503
        SColumnInfoData* pDst = taosArrayGet(pRes->pDataBlock, pBuild->valCols[buildIdx].dstSlot);
×
UNCOV
504
        if (pBuild->valCols[buildIdx].keyCol) {
×
505
          code = colDataSetVal(pDst, pRes->info.rows + r, pKeyData, false);
×
506
          if (code) {
×
507
            return code;
×
508
          }
509
          pKeyData += pBuild->valCols[buildIdx].vardata ? varDataTLen(pKeyData) : pBuild->valCols[buildIdx].bytes;
×
510
        } else {
UNCOV
511
          if (colDataIsNull_f(pData, buildValIdx)) {
×
512
            code = colDataSetVal(pDst, pRes->info.rows + r, NULL, true);
×
513
            if (code) {
×
514
              return code;
×
515
            }
516
          } else {
UNCOV
517
            code = colDataSetVal(pDst, pRes->info.rows + r, pValData, false);
×
UNCOV
518
            if (code) {
×
519
              return code;
×
520
            }
UNCOV
521
            pValData += pBuild->valCols[buildIdx].vardata ? varDataTLen(pValData) : pBuild->valCols[buildIdx].bytes;
×
522
          }
UNCOV
523
          buildValIdx++;
×
524
        }
UNCOV
525
        buildIdx++;
×
UNCOV
526
      } else if (0 == r) {
×
UNCOV
527
        SColumnInfoData* pSrc = taosArrayGet(pJoin->ctx.pProbeData->pDataBlock, pProbe->valCols[probeIdx].srcSlot);
×
UNCOV
528
        SColumnInfoData* pDst = taosArrayGet(pRes->pDataBlock, pProbe->valCols[probeIdx].dstSlot);
×
529
    
UNCOV
530
        code = colDataCopyNItems(pDst, pRes->info.rows, colDataGetData(pSrc, pJoin->ctx.probeStartIdx), rowNum, colDataIsNull_s(pSrc, pJoin->ctx.probeStartIdx));
×
UNCOV
531
        if (code) {
×
532
          return code;
×
533
        }
UNCOV
534
        probeIdx++;
×
535
      }
536
    }
UNCOV
537
    pRow = pRow->next;
×
538
  }
539

UNCOV
540
  return TSDB_CODE_SUCCESS;
×
541
}
542

543
int32_t hJoinCopyNMatchRowsToBlock(SHJoinOperatorInfo* pJoin, SSDataBlock* pRes, int32_t startIdx, int32_t rows) {
×
544
  SHJoinTableCtx* pBuild = pJoin->pBuild;
×
545
  SHJoinTableCtx* pProbe = pJoin->pProbe;
×
546
  int32_t buildIdx = 0;
×
547
  int32_t probeIdx = 0;
×
548
  int32_t code = 0;
×
549

550
  for (int32_t i = 0; i < pJoin->pResColNum; ++i) {
×
551
    if (pJoin->pResColMap[i]) {
×
552
      SColumnInfoData* pDst = taosArrayGet(pRes->pDataBlock, pBuild->valCols[buildIdx].dstSlot);
×
553
      colDataSetNItemsNull(pDst, pRes->info.rows, rows);
×
554

555
      buildIdx++;
×
556
    } else {
557
      SColumnInfoData* pSrc = taosArrayGet(pJoin->ctx.pProbeData->pDataBlock, pProbe->valCols[probeIdx].srcSlot);
×
558
      SColumnInfoData* pDst = taosArrayGet(pRes->pDataBlock, pProbe->valCols[probeIdx].dstSlot);
×
559

560
      QRY_ERR_RET(colDataAssignNRows(pDst, pRes->info.rows, pSrc, startIdx, rows));
×
561

562
      probeIdx++;
×
563
    }
564
  }
565

566
  pRes->info.rows += rows;
×
567

568
  return TSDB_CODE_SUCCESS;
×
569
}
570

571

572

UNCOV
573
void hJoinAppendResToBlock(struct SOperatorInfo* pOperator, SSDataBlock* pRes, bool* allFetched) {
×
UNCOV
574
  SHJoinOperatorInfo* pJoin = pOperator->info;
×
UNCOV
575
  SHJoinCtx* pCtx = &pJoin->ctx;
×
UNCOV
576
  SBufRowInfo* pStart = pCtx->pBuildRow;
×
UNCOV
577
  int32_t rowNum = 0;
×
UNCOV
578
  int32_t resNum = pRes->info.rows;
×
579
  
UNCOV
580
  while (pCtx->pBuildRow && (resNum < pRes->info.capacity)) {
×
UNCOV
581
    rowNum++;
×
UNCOV
582
    resNum++;
×
UNCOV
583
    pCtx->pBuildRow = pCtx->pBuildRow->next;
×
584
  }
585

UNCOV
586
  pJoin->execInfo.resRows += rowNum;
×
587

UNCOV
588
  int32_t code = hJoinCopyResRowsToBlock(pJoin, rowNum, pStart, pRes);
×
UNCOV
589
  if (code) {
×
590
    pOperator->pTaskInfo->code = code;
×
591
    T_LONG_JMP(pOperator->pTaskInfo->env, code);
×
592
  }
593

UNCOV
594
  pRes->info.rows = resNum;
×
UNCOV
595
  *allFetched = pCtx->pBuildRow ? false : true;
×
UNCOV
596
}
×
597

598

UNCOV
599
bool hJoinCopyKeyColsDataToBuf(SHJoinTableCtx* pTable, int32_t rowIdx, size_t *pBufLen) {
×
UNCOV
600
  char *pData = NULL;
×
UNCOV
601
  size_t bufLen = 0;
×
602
  
UNCOV
603
  if (1 == pTable->keyNum) {
×
UNCOV
604
    if (colDataIsNull_s(pTable->keyCols[0].colData, rowIdx)) {
×
UNCOV
605
      return true;
×
606
    }
UNCOV
607
    if (pTable->keyCols[0].vardata) {
×
UNCOV
608
      pData = pTable->keyCols[0].data + pTable->keyCols[0].offset[rowIdx];
×
UNCOV
609
      bufLen = varDataTLen(pData);
×
610
    } else {
UNCOV
611
      pData = pTable->keyCols[0].data + pTable->keyCols[0].bytes * rowIdx;
×
UNCOV
612
      bufLen = pTable->keyCols[0].bytes;
×
613
    }
UNCOV
614
    pTable->keyData = pData;
×
615
  } else {
UNCOV
616
    for (int32_t i = 0; i < pTable->keyNum; ++i) {
×
UNCOV
617
      if (colDataIsNull_s(pTable->keyCols[i].colData, rowIdx)) {
×
618
        return true;
×
619
      }
UNCOV
620
      if (pTable->keyCols[i].vardata) {
×
UNCOV
621
        pData = pTable->keyCols[i].data + pTable->keyCols[i].offset[rowIdx];
×
UNCOV
622
        TAOS_MEMCPY(pTable->keyBuf + bufLen, pData, varDataTLen(pData));
×
UNCOV
623
        bufLen += varDataTLen(pData);
×
624
      } else {
UNCOV
625
        pData = pTable->keyCols[i].data + pTable->keyCols[i].bytes * rowIdx;
×
UNCOV
626
        TAOS_MEMCPY(pTable->keyBuf + bufLen, pData, pTable->keyCols[i].bytes);
×
UNCOV
627
        bufLen += pTable->keyCols[i].bytes;
×
628
      }
629
    }
UNCOV
630
    pTable->keyData = pTable->keyBuf;
×
631
  }
632

UNCOV
633
  if (pBufLen) {
×
UNCOV
634
    *pBufLen = bufLen;
×
635
  }
636

UNCOV
637
  return false;
×
638
}
639

UNCOV
640
static int32_t hJoinSetKeyColsData(SSDataBlock* pBlock, SHJoinTableCtx* pTable) {
×
UNCOV
641
  for (int32_t i = 0; i < pTable->keyNum; ++i) {
×
UNCOV
642
    SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, pTable->keyCols[i].srcSlot);
×
UNCOV
643
    if (NULL == pCol) {
×
644
      qError("fail to get %d col, total:%d", pTable->keyCols[i].srcSlot, (int32_t)taosArrayGetSize(pBlock->pDataBlock));
×
645
      QRY_ERR_RET(TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR);
×
646
    }
UNCOV
647
    if (pTable->keyCols[i].vardata != IS_VAR_DATA_TYPE(pCol->info.type))  {
×
648
      qError("column type mismatch, idx:%d, slotId:%d, type:%d, vardata:%d", i, pTable->keyCols[i].srcSlot, pCol->info.type, pTable->keyCols[i].vardata);
×
649
      return TSDB_CODE_INVALID_PARA;
×
650
    }
UNCOV
651
    if (pTable->keyCols[i].bytes != pCol->info.bytes)  {
×
652
      qError("column bytes mismatch, idx:%d, slotId:%d, bytes:%d, %d", i, pTable->keyCols[i].srcSlot, pCol->info.bytes, pTable->keyCols[i].bytes);
×
653
      return TSDB_CODE_INVALID_PARA;
×
654
    }
UNCOV
655
    pTable->keyCols[i].data = pCol->pData;
×
UNCOV
656
    if (pTable->keyCols[i].vardata) {
×
UNCOV
657
      pTable->keyCols[i].offset = pCol->varmeta.offset;
×
658
    }
UNCOV
659
    pTable->keyCols[i].colData = pCol;
×
660
  }
661

UNCOV
662
  return TSDB_CODE_SUCCESS;
×
663
}
664

UNCOV
665
static int32_t hJoinSetValColsData(SSDataBlock* pBlock, SHJoinTableCtx* pTable) {
×
UNCOV
666
  if (!pTable->valColExist) {
×
667
    return TSDB_CODE_SUCCESS;
×
668
  }
UNCOV
669
  for (int32_t i = 0; i < pTable->valNum; ++i) {
×
UNCOV
670
    if (pTable->valCols[i].keyCol) {
×
671
      continue;
×
672
    }
UNCOV
673
    SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, pTable->valCols[i].srcSlot);
×
UNCOV
674
    if (NULL == pCol) {
×
675
      qError("fail to get %d col, total:%d", pTable->valCols[i].srcSlot, (int32_t)taosArrayGetSize(pBlock->pDataBlock));
×
676
      QRY_ERR_RET(TSDB_CODE_QRY_EXECUTOR_INTERNAL_ERROR);
×
677
    }
UNCOV
678
    if (pTable->valCols[i].vardata != IS_VAR_DATA_TYPE(pCol->info.type))  {
×
679
      qError("column type mismatch, idx:%d, slotId:%d, type:%d, vardata:%d", i, pTable->valCols[i].srcSlot, pCol->info.type, pTable->valCols[i].vardata);
×
680
      return TSDB_CODE_INVALID_PARA;
×
681
    }
UNCOV
682
    if (pTable->valCols[i].bytes != pCol->info.bytes)  {
×
683
      qError("column bytes mismatch, idx:%d, slotId:%d, bytes:%d, %d", i, pTable->valCols[i].srcSlot, pCol->info.bytes, pTable->valCols[i].bytes);
×
684
      return TSDB_CODE_INVALID_PARA;
×
685
    }
UNCOV
686
    if (!pTable->valCols[i].vardata) {
×
UNCOV
687
      pTable->valCols[i].bitMap = pCol->nullbitmap;
×
688
    }
UNCOV
689
    pTable->valCols[i].data = pCol->pData;
×
UNCOV
690
    if (pTable->valCols[i].vardata) {
×
691
      pTable->valCols[i].offset = pCol->varmeta.offset;
×
692
    }
693
  }
694

UNCOV
695
  return TSDB_CODE_SUCCESS;
×
696
}
697

698

699

700
static FORCE_INLINE void hJoinCopyValColsDataToBuf(SHJoinTableCtx* pTable, int32_t rowIdx) {
UNCOV
701
  if (!pTable->valColExist) {
×
702
    return;
×
703
  }
704

UNCOV
705
  char *pData = NULL;
×
UNCOV
706
  size_t bufLen = pTable->valBitMapSize;
×
UNCOV
707
  TAOS_MEMSET(pTable->valData, 0, pTable->valBitMapSize);
×
UNCOV
708
  for (int32_t i = 0, m = 0; i < pTable->valNum; ++i) {
×
UNCOV
709
    if (pTable->valCols[i].keyCol) {
×
710
      continue;
×
711
    }
UNCOV
712
    if (pTable->valCols[i].vardata) {
×
713
      if (-1 == pTable->valCols[i].offset[rowIdx]) {
×
714
        colDataSetNull_f(pTable->valData, m);
×
715
      } else {
716
        pData = pTable->valCols[i].data + pTable->valCols[i].offset[rowIdx];
×
717
        TAOS_MEMCPY(pTable->valData + bufLen, pData, varDataTLen(pData));
×
718
        bufLen += varDataTLen(pData);
×
719
      }
720
    } else {
UNCOV
721
      if (colDataIsNull_f(pTable->valCols[i].bitMap, rowIdx)) {
×
722
        colDataSetNull_f(pTable->valData, m);
×
723
      } else {
UNCOV
724
        pData = pTable->valCols[i].data + pTable->valCols[i].bytes * rowIdx;
×
UNCOV
725
        TAOS_MEMCPY(pTable->valData + bufLen, pData, pTable->valCols[i].bytes);
×
UNCOV
726
        bufLen += pTable->valCols[i].bytes;
×
727
      }
728
    }
UNCOV
729
    m++;
×
730
  }
731
}
732

733

734
static FORCE_INLINE int32_t hJoinGetValBufFromPages(SArray* pPages, int32_t bufSize, char** pBuf, SBufRowInfo* pRow) {
UNCOV
735
  if (0 == bufSize) {
×
736
    pRow->pageId = -1;
×
737
    return TSDB_CODE_SUCCESS;
×
738
  }
739

UNCOV
740
  if (bufSize > HASH_JOIN_DEFAULT_PAGE_SIZE) {
×
741
    qError("invalid join value buf size:%d", bufSize);
×
742
    return TSDB_CODE_INVALID_PARA;
×
743
  }
744
  
745
  do {
×
UNCOV
746
    SBufPageInfo* page = taosArrayGetLast(pPages);
×
UNCOV
747
    if ((page->pageSize - page->offset) >= bufSize) {
×
UNCOV
748
      *pBuf = page->data + page->offset;
×
UNCOV
749
      pRow->pageId = taosArrayGetSize(pPages) - 1;
×
UNCOV
750
      pRow->offset = page->offset;
×
UNCOV
751
      page->offset += bufSize;
×
UNCOV
752
      return TSDB_CODE_SUCCESS;
×
753
    }
754

755
    int32_t code = hJoinAddPageToBufs(pPages);
×
756
    if (code) {
×
757
      return code;
×
758
    }
759
  } while (true);
760
}
761

762
static FORCE_INLINE int32_t hJoinGetValBufSize(SHJoinTableCtx* pTable, int32_t rowIdx) {
UNCOV
763
  if (NULL == pTable->valVarCols) {
×
UNCOV
764
    return pTable->valBufSize;
×
765
  }
766

767
  int32_t* varColIdx = NULL;
×
768
  int32_t bufLen = pTable->valBufSize;
×
769
  int32_t varColNum = taosArrayGetSize(pTable->valVarCols);
×
770
  for (int32_t i = 0; i < varColNum; ++i) {
×
771
    varColIdx = taosArrayGet(pTable->valVarCols, i);
×
772
    char* pData = pTable->valCols[*varColIdx].data + pTable->valCols[*varColIdx].offset[rowIdx];
×
773
    bufLen += varDataTLen(pData);
×
774
  }
775

776
  return bufLen;
×
777
}
778

779

UNCOV
780
static int32_t hJoinAddRowToHashImpl(SHJoinOperatorInfo* pJoin, SGroupData* pGroup, SHJoinTableCtx* pTable, size_t keyLen, int32_t rowIdx) {
×
UNCOV
781
  SGroupData group = {0};
×
UNCOV
782
  SBufRowInfo* pRow = NULL;
×
783

UNCOV
784
  if (NULL == pGroup) {
×
UNCOV
785
    group.rows = taosMemoryMalloc(sizeof(SBufRowInfo));
×
UNCOV
786
    if (NULL == group.rows) {
×
787
      return terrno;
×
788
    }
UNCOV
789
    pRow = group.rows;
×
790
  } else {
UNCOV
791
    pRow = taosMemoryMalloc(sizeof(SBufRowInfo));
×
UNCOV
792
    if (NULL == pRow) {
×
793
      return terrno;
×
794
    }
795
  }
796

UNCOV
797
  int32_t code = hJoinGetValBufFromPages(pJoin->pRowBufs, hJoinGetValBufSize(pTable, rowIdx), &pTable->valData, pRow);
×
UNCOV
798
  if (code) {
×
799
    taosMemoryFree(pRow);
×
800
    return code;
×
801
  }
802

UNCOV
803
  if (NULL == pGroup) {
×
UNCOV
804
    pRow->next = NULL;
×
UNCOV
805
    if (tSimpleHashPut(pJoin->pKeyHash, pTable->keyData, keyLen, &group, sizeof(group))) {
×
806
      taosMemoryFree(pRow);
×
807
      return terrno;
×
808
    }
809
  } else {
UNCOV
810
    pRow->next = pGroup->rows;
×
UNCOV
811
    pGroup->rows = pRow;
×
812
  }
813

UNCOV
814
  return TSDB_CODE_SUCCESS;
×
815
}
816

UNCOV
817
static int32_t hJoinAddRowToHash(SHJoinOperatorInfo* pJoin, SSDataBlock* pBlock, size_t keyLen, int32_t rowIdx) {
×
UNCOV
818
  SHJoinTableCtx* pBuild = pJoin->pBuild;
×
UNCOV
819
  int32_t code = hJoinSetValColsData(pBlock, pBuild);
×
UNCOV
820
  if (code) {
×
821
    return code;
×
822
  }
823

UNCOV
824
  SGroupData* pGroup = tSimpleHashGet(pJoin->pKeyHash, pBuild->keyData, keyLen);
×
UNCOV
825
  code = hJoinAddRowToHashImpl(pJoin, pGroup, pBuild, keyLen, rowIdx);
×
UNCOV
826
  if (code) {
×
827
    return code;
×
828
  }
829
  
830
  hJoinCopyValColsDataToBuf(pBuild, rowIdx);
831

UNCOV
832
  return TSDB_CODE_SUCCESS;
×
833
}
834

835
static bool hJoinFilterTimeRange(SSDataBlock* pBlock, STimeWindow* pRange, int32_t primSlot, int32_t* startIdx, int32_t* endIdx) {
×
836
  SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, primSlot);
×
837
  if (NULL == pCol) {
×
838
    qError("hash join can't get prim col, slot:%d, slotNum:%d", primSlot, (int32_t)taosArrayGetSize(pBlock->pDataBlock));
×
839
    return false;
×
840
  }
841

842
  TSKEY skey = *(TSKEY*)colDataGetData(pCol, 0);
×
843
  TSKEY ekey = *(TSKEY*)colDataGetData(pCol, (pBlock->info.rows - 1));
×
844

845
  if (ekey < pRange->skey || skey > pRange->ekey) {
×
846
    return false;
×
847
  }
848

849
  if (skey >= pRange->skey && ekey <= pRange->ekey) {
×
850
    *startIdx = 0;
×
851
    *endIdx = pBlock->info.rows - 1;
×
852
    return true;
×
853
  }
854

855
  if (skey < pRange->skey && ekey > pRange->ekey) {
×
856
    TSKEY *pStart = (TSKEY*)taosbsearch(&pRange->skey, pCol->pData, pBlock->info.rows, sizeof(TSKEY), compareInt64Val, TD_GE);
×
857
    TSKEY *pEnd = (TSKEY*)taosbsearch(&pRange->ekey, pCol->pData, pBlock->info.rows, sizeof(TSKEY), compareInt64Val, TD_LE);
×
858
    *startIdx = ((uint64_t)pStart - (uint64_t)pCol->pData) / sizeof(int64_t);
×
859
    *endIdx = ((uint64_t)pEnd - (uint64_t)pCol->pData) / sizeof(int64_t);
×
860
    return true;
×
861
  }
862

863
  if (skey >= pRange->skey) {
×
864
    TSKEY *pEnd = (TSKEY*)taosbsearch(&pRange->ekey, pCol->pData, pBlock->info.rows, sizeof(TSKEY), compareInt64Val, TD_LE);
×
865
    *startIdx = 0;
×
866
    *endIdx = ((uint64_t)pEnd - (uint64_t)pCol->pData) / sizeof(int64_t);
×
867
    return true;
×
868
  }
869

870
  TSKEY *pStart = (TSKEY*)taosbsearch(&pRange->skey, pCol->pData, pBlock->info.rows, sizeof(TSKEY), compareInt64Val, TD_GE);
×
871
  *startIdx = ((uint64_t)pStart - (uint64_t)pCol->pData) / sizeof(int64_t);
×
872
  *endIdx = pBlock->info.rows - 1;
×
873
  
874
  return true;
×
875
}
876

UNCOV
877
static int32_t hJoinAddBlockRowsToHash(SSDataBlock* pBlock, SHJoinOperatorInfo* pJoin) {
×
UNCOV
878
  SHJoinTableCtx* pBuild = pJoin->pBuild;
×
UNCOV
879
  int32_t startIdx = 0, endIdx = pBlock->info.rows - 1;
×
UNCOV
880
  if (pBuild->hasTimeRange && !hJoinFilterTimeRange(pBlock, &pJoin->tblTimeRange, pBuild->primCol->srcSlot, &startIdx, &endIdx)) {
×
881
    return TSDB_CODE_SUCCESS;
×
882
  }
883

UNCOV
884
  HJ_ERR_RET(hJoinLaunchPrimExpr(pBlock, pBuild, startIdx, endIdx));
×
885

UNCOV
886
  int32_t code = hJoinSetKeyColsData(pBlock, pBuild);
×
UNCOV
887
  if (code) {
×
888
    return code;
×
889
  }
890

UNCOV
891
  size_t bufLen = 0;
×
UNCOV
892
  for (int32_t i = startIdx; i <= endIdx; ++i) {
×
UNCOV
893
    if (hJoinCopyKeyColsDataToBuf(pBuild, i, &bufLen)) {
×
UNCOV
894
      continue;
×
895
    }
UNCOV
896
    code = hJoinAddRowToHash(pJoin, pBlock, bufLen, i);
×
UNCOV
897
    if (code) {
×
898
      return code;
×
899
    }
900
  }
901

UNCOV
902
  return code;
×
903
}
904

UNCOV
905
static int32_t hJoinBuildHash(struct SOperatorInfo* pOperator, bool* queryDone) {
×
UNCOV
906
  SHJoinOperatorInfo* pJoin = pOperator->info;
×
UNCOV
907
  SSDataBlock*        pBlock = NULL;
×
UNCOV
908
  int32_t             code = TSDB_CODE_SUCCESS;
×
909

910
  while (true) {
UNCOV
911
    pBlock = getNextBlockFromDownstream(pOperator, pJoin->pBuild->downStreamIdx);
×
UNCOV
912
    if (NULL == pBlock) {
×
UNCOV
913
      break;
×
914
    }
915

UNCOV
916
    pJoin->execInfo.buildBlkNum++;
×
UNCOV
917
    pJoin->execInfo.buildBlkRows += pBlock->info.rows;
×
918

UNCOV
919
    code = hJoinAddBlockRowsToHash(pBlock, pJoin);
×
UNCOV
920
    if (code) {
×
921
      return code;
×
922
    }
923
  }
924

UNCOV
925
  if (IS_INNER_NONE_JOIN(pJoin->joinType, pJoin->subType) && tSimpleHashGetSize(pJoin->pKeyHash) <= 0) {
×
UNCOV
926
    hJoinSetDone(pOperator);
×
UNCOV
927
    *queryDone = true;
×
928
  }
929
  
930
  //qTrace("build table rows:%" PRId64, hJoinGetRowsNumOfKeyHash(pJoin->pKeyHash));
931

UNCOV
932
  return TSDB_CODE_SUCCESS;
×
933
}
934

UNCOV
935
static int32_t hJoinPrepareStart(struct SOperatorInfo* pOperator, SSDataBlock* pBlock) {
×
UNCOV
936
  SHJoinOperatorInfo* pJoin = pOperator->info;
×
UNCOV
937
  SHJoinTableCtx* pProbe = pJoin->pProbe;
×
UNCOV
938
  int32_t startIdx = 0, endIdx = pBlock->info.rows - 1;
×
UNCOV
939
  if (pProbe->hasTimeRange && !hJoinFilterTimeRange(pBlock, &pJoin->tblTimeRange, pProbe->primCol->srcSlot, &startIdx, &endIdx)) {
×
940
    if (!IS_INNER_NONE_JOIN(pJoin->joinType, pJoin->subType)) {
×
941
      pJoin->ctx.probeEndIdx = -1;
×
942
      pJoin->ctx.probePostIdx = 0;
×
943
      pJoin->ctx.pProbeData = pBlock;
×
944
      pJoin->ctx.rowRemains = true;
×
945
      pJoin->ctx.probePhase = E_JOIN_PHASE_POST;
×
946
      
947
      HJ_ERR_RET((*pJoin->joinFp)(pOperator));
×
948
    }
949
    
950
    return TSDB_CODE_SUCCESS;
×
951
  }
952

UNCOV
953
  HJ_ERR_RET(hJoinLaunchPrimExpr(pBlock, pProbe, startIdx, endIdx));
×
954

UNCOV
955
  int32_t code = hJoinSetKeyColsData(pBlock, pProbe);
×
UNCOV
956
  if (code) {
×
957
    return code;
×
958
  }
UNCOV
959
  code = hJoinSetValColsData(pBlock, pProbe);
×
UNCOV
960
  if (code) {
×
961
    return code;
×
962
  }
963

UNCOV
964
  pJoin->ctx.probeStartIdx = startIdx;
×
UNCOV
965
  pJoin->ctx.probeEndIdx = endIdx;
×
UNCOV
966
  pJoin->ctx.pBuildRow = NULL;
×
UNCOV
967
  pJoin->ctx.pProbeData = pBlock;
×
UNCOV
968
  pJoin->ctx.rowRemains = true;
×
UNCOV
969
  pJoin->ctx.probePreIdx = 0;
×
UNCOV
970
  pJoin->ctx.probePostIdx = endIdx + 1;
×
971

UNCOV
972
  if (!IS_INNER_NONE_JOIN(pJoin->joinType, pJoin->subType) && startIdx > 0) {
×
973
    pJoin->ctx.probePhase = E_JOIN_PHASE_PRE;
×
974
  } else {
UNCOV
975
    pJoin->ctx.probePhase = E_JOIN_PHASE_CUR;
×
976
  }
977

UNCOV
978
  HJ_ERR_RET((*pJoin->joinFp)(pOperator));
×
979

UNCOV
980
  return TSDB_CODE_SUCCESS;
×
981
}
982

UNCOV
983
void hJoinSetDone(struct SOperatorInfo* pOperator) {
×
UNCOV
984
  setOperatorCompleted(pOperator);
×
985

UNCOV
986
  SHJoinOperatorInfo* pInfo = pOperator->info;
×
UNCOV
987
  hJoinDestroyKeyHash(&pInfo->pKeyHash);
×
988

UNCOV
989
  qDebug("hash Join done");  
×
UNCOV
990
}
×
991

UNCOV
992
static int32_t hJoinMainProcess(struct SOperatorInfo* pOperator, SSDataBlock** pResBlock) {
×
UNCOV
993
  SHJoinOperatorInfo* pJoin = pOperator->info;
×
UNCOV
994
  SExecTaskInfo*      pTaskInfo = pOperator->pTaskInfo;
×
UNCOV
995
  int32_t             code = TSDB_CODE_SUCCESS;
×
UNCOV
996
  int32_t             lino = 0;
×
UNCOV
997
  SSDataBlock*        pRes = pJoin->finBlk;
×
UNCOV
998
  int64_t             st = 0;
×
999

UNCOV
1000
  QRY_PARAM_CHECK(pResBlock);
×
UNCOV
1001
  if (pOperator->cost.openCost == 0) {
×
UNCOV
1002
    st = taosGetTimestampUs();
×
1003
  }
1004

UNCOV
1005
  if (pOperator->status == OP_EXEC_DONE) {
×
UNCOV
1006
    pRes->info.rows = 0;
×
UNCOV
1007
    goto _end;
×
1008
  }
1009

UNCOV
1010
  if (!pJoin->keyHashBuilt) {
×
UNCOV
1011
    pJoin->keyHashBuilt = true;
×
1012

UNCOV
1013
    bool queryDone = false;
×
UNCOV
1014
    code = hJoinBuildHash(pOperator, &queryDone);
×
UNCOV
1015
    QUERY_CHECK_CODE(code, lino, _end);
×
1016

UNCOV
1017
    if (queryDone) {
×
UNCOV
1018
      goto _end;
×
1019
    }
1020
  }
1021

UNCOV
1022
  blockDataCleanup(pRes);
×
1023

UNCOV
1024
  if (pJoin->ctx.rowRemains) {
×
1025
    code = (*pJoin->joinFp)(pOperator);
×
1026
    QUERY_CHECK_CODE(code, lino, _end);
×
1027

1028
    if (pRes->info.rows > 0 && pJoin->pFinFilter != NULL) {
×
1029
      code = doFilter(pRes, pJoin->pFinFilter, NULL);
×
1030
      QUERY_CHECK_CODE(code, lino, _end);
×
1031
    }
1032

1033
    if (pRes->info.rows > 0) {
×
1034
      *pResBlock = pRes;
×
1035
      return code;
×
1036
    }
1037
  }
1038

UNCOV
1039
  while (true) {
×
UNCOV
1040
    SSDataBlock* pBlock = getNextBlockFromDownstream(pOperator, pJoin->pProbe->downStreamIdx);
×
UNCOV
1041
    if (NULL == pBlock) {
×
UNCOV
1042
      hJoinSetDone(pOperator);
×
UNCOV
1043
      break;
×
1044
    }
1045

UNCOV
1046
    pJoin->execInfo.probeBlkNum++;
×
UNCOV
1047
    pJoin->execInfo.probeBlkRows += pBlock->info.rows;
×
1048

UNCOV
1049
    code = hJoinPrepareStart(pOperator, pBlock);
×
UNCOV
1050
    QUERY_CHECK_CODE(code, lino, _end);
×
1051

UNCOV
1052
    if (!hJoinBlkReachThreshold(pJoin, pRes->info.rows)) {
×
UNCOV
1053
      continue;
×
1054
    }
1055

1056
    if (pRes->info.rows > 0 && pJoin->pFinFilter != NULL) {
×
1057
      code = doFilter(pRes, pJoin->pFinFilter, NULL);
×
1058
      QUERY_CHECK_CODE(code, lino, _end);
×
1059
    }
1060

1061
    if (pRes->info.rows > 0) {
×
1062
      break;
×
1063
    }
1064
  }
1065

UNCOV
1066
_end:
×
UNCOV
1067
  if (pOperator->cost.openCost == 0) {
×
UNCOV
1068
    pOperator->cost.openCost = (taosGetTimestampUs() - st) / 1000.0;
×
1069
  }
UNCOV
1070
  if (code != TSDB_CODE_SUCCESS) {
×
1071
    qError("%s failed at line %d since %s", __func__, lino, tstrerror(code));
×
1072
    pTaskInfo->code = code;
×
1073
    T_LONG_JMP(pTaskInfo->env, code);
×
1074
  }
UNCOV
1075
  if (pRes->info.rows > 0) {
×
UNCOV
1076
    *pResBlock = pRes;
×
1077
  }
1078

UNCOV
1079
  return code;
×
1080
}
1081

UNCOV
1082
static void destroyHashJoinOperator(void* param) {
×
UNCOV
1083
  SHJoinOperatorInfo* pJoinOperator = (SHJoinOperatorInfo*)param;
×
UNCOV
1084
  qDebug("hashJoin exec info, buildBlk:%" PRId64 ", buildRows:%" PRId64 ", probeBlk:%" PRId64 ", probeRows:%" PRId64 ", resRows:%" PRId64, 
×
1085
         pJoinOperator->execInfo.buildBlkNum, pJoinOperator->execInfo.buildBlkRows, pJoinOperator->execInfo.probeBlkNum, 
1086
         pJoinOperator->execInfo.probeBlkRows, pJoinOperator->execInfo.resRows);
1087

UNCOV
1088
  hJoinDestroyKeyHash(&pJoinOperator->pKeyHash);
×
1089

UNCOV
1090
  hJoinFreeTableInfo(&pJoinOperator->tbs[0]);
×
UNCOV
1091
  hJoinFreeTableInfo(&pJoinOperator->tbs[1]);
×
UNCOV
1092
  blockDataDestroy(pJoinOperator->finBlk);
×
UNCOV
1093
  pJoinOperator->finBlk = NULL;
×
UNCOV
1094
  taosMemoryFreeClear(pJoinOperator->pResColMap);
×
UNCOV
1095
  taosArrayDestroyEx(pJoinOperator->pRowBufs, hJoinFreeBufPage);
×
1096

UNCOV
1097
  taosMemoryFreeClear(param);
×
UNCOV
1098
}
×
1099

UNCOV
1100
int32_t hJoinHandleConds(SHJoinOperatorInfo* pJoin, SHashJoinPhysiNode* pJoinNode) {
×
UNCOV
1101
  switch (pJoin->joinType) {
×
UNCOV
1102
    case JOIN_TYPE_INNER: {
×
UNCOV
1103
      SNode* pCond = NULL;
×
UNCOV
1104
      if (pJoinNode->pFullOnCond != NULL) {
×
1105
        if (pJoinNode->node.pConditions != NULL) {
×
1106
          HJ_ERR_RET(mergeJoinConds(&pJoinNode->pFullOnCond, &pJoinNode->node.pConditions));
×
1107
        }
1108
        pCond = pJoinNode->pFullOnCond;
×
UNCOV
1109
      } else if (pJoinNode->node.pConditions != NULL) {
×
1110
        pCond = pJoinNode->node.pConditions;
×
1111
      }
1112
      
UNCOV
1113
      HJ_ERR_RET(filterInitFromNode(pCond, &pJoin->pFinFilter, 0));
×
UNCOV
1114
      break;
×
1115
    }
1116
    case JOIN_TYPE_LEFT:
×
1117
    case JOIN_TYPE_RIGHT:
1118
    case JOIN_TYPE_FULL:
1119
      if (pJoinNode->pFullOnCond != NULL) {
×
1120
        HJ_ERR_RET(filterInitFromNode(pJoinNode->pFullOnCond, &pJoin->pPreFilter, 0));
×
1121
      }
1122
      if (pJoinNode->node.pConditions != NULL) {
×
1123
        HJ_ERR_RET(filterInitFromNode(pJoinNode->node.pConditions, &pJoin->pFinFilter, 0));
×
1124
      }
1125
      break;
×
1126
    default:
×
1127
      break;
×
1128
  }
1129

UNCOV
1130
  return TSDB_CODE_SUCCESS;
×
1131
}
1132

UNCOV
1133
static uint32_t hJoinGetFinBlkCapacity(SHJoinOperatorInfo* pJoin, SHashJoinPhysiNode* pJoinNode) {
×
UNCOV
1134
  uint32_t maxRows = TMAX(HJOIN_DEFAULT_BLK_ROWS_NUM, HJOIN_BLK_SIZE_LIMIT/pJoinNode->node.pOutputDataBlockDesc->totalRowSize);
×
UNCOV
1135
  if (INT64_MAX != pJoin->ctx.limit && NULL == pJoin->pFinFilter) {
×
1136
    uint32_t limitMaxRows = pJoin->ctx.limit / HJOIN_BLK_THRESHOLD_RATIO + 1;
×
1137
    return (maxRows > limitMaxRows) ? limitMaxRows : maxRows;
×
1138
  }
1139

UNCOV
1140
  return maxRows;
×
1141
}
1142

1143

UNCOV
1144
int32_t hJoinInitResBlocks(SHJoinOperatorInfo* pJoin, SHashJoinPhysiNode* pJoinNode) {
×
UNCOV
1145
  pJoin->finBlk = createDataBlockFromDescNode(pJoinNode->node.pOutputDataBlockDesc);
×
UNCOV
1146
  if (NULL == pJoin->finBlk) {
×
1147
    QRY_ERR_RET(terrno);
×
1148
  }
1149

UNCOV
1150
  int32_t code = blockDataEnsureCapacity(pJoin->finBlk, hJoinGetFinBlkCapacity(pJoin, pJoinNode));
×
UNCOV
1151
  if (TSDB_CODE_SUCCESS != code) {
×
1152
    QRY_ERR_RET(code);
×
1153
  }
1154
  
UNCOV
1155
  if (NULL != pJoin->pPreFilter) {
×
1156
    pJoin->midBlk = NULL;
×
1157
    code = createOneDataBlock(pJoin->finBlk, false, &pJoin->midBlk);
×
1158
    if (code) {
×
1159
      QRY_ERR_RET(code);
×
1160
    }
1161

1162
    code = blockDataEnsureCapacity(pJoin->midBlk, pJoin->finBlk->info.capacity);
×
1163
    if (TSDB_CODE_SUCCESS != code) {
×
1164
      QRY_ERR_RET(code);
×
1165
    }
1166
  }
1167

UNCOV
1168
  pJoin->blkThreshold = pJoin->finBlk->info.capacity * HJOIN_BLK_THRESHOLD_RATIO;
×
UNCOV
1169
  return TSDB_CODE_SUCCESS;
×
1170
}
1171

1172

UNCOV
1173
int32_t createHashJoinOperatorInfo(SOperatorInfo** pDownstream, int32_t numOfDownstream,
×
1174
                                           SHashJoinPhysiNode* pJoinNode, SExecTaskInfo* pTaskInfo, SOperatorInfo** pOptrInfo) {
UNCOV
1175
  QRY_PARAM_CHECK(pOptrInfo);
×
1176

UNCOV
1177
  int32_t             code = TSDB_CODE_SUCCESS;
×
UNCOV
1178
  SHJoinOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SHJoinOperatorInfo));
×
UNCOV
1179
  SOperatorInfo*      pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
×
UNCOV
1180
  if (pOperator == NULL || pInfo == NULL) {
×
1181
    code = terrno;
×
1182
    goto _return;
×
1183
  }
1184

UNCOV
1185
  pInfo->tblTimeRange.skey = pJoinNode->timeRange.skey;
×
UNCOV
1186
  pInfo->tblTimeRange.ekey = pJoinNode->timeRange.ekey;
×
1187
  
UNCOV
1188
  pInfo->ctx.limit = (pJoinNode->node.pLimit && ((SLimitNode*)pJoinNode->node.pLimit)->limit) ? ((SLimitNode*)pJoinNode->node.pLimit)->limit->datum.i : INT64_MAX;
×
1189

UNCOV
1190
  setOperatorInfo(pOperator, "HashJoinOperator", QUERY_NODE_PHYSICAL_PLAN_HASH_JOIN, false, OP_NOT_OPENED, pInfo, pTaskInfo);
×
1191

UNCOV
1192
  HJ_ERR_JRET(hJoinInitTableInfo(pInfo, pJoinNode, pDownstream, 0, &pJoinNode->inputStat[0]));
×
UNCOV
1193
  HJ_ERR_JRET(hJoinInitTableInfo(pInfo, pJoinNode, pDownstream, 1, &pJoinNode->inputStat[1]));
×
1194

UNCOV
1195
  hJoinSetBuildAndProbeTable(pInfo, pJoinNode);
×
1196
  
UNCOV
1197
  HJ_ERR_JRET(hJoinBuildResColsMap(pInfo, pJoinNode));
×
1198

UNCOV
1199
  HJ_ERR_JRET(hJoinInitBufPages(pInfo));
×
1200

UNCOV
1201
  size_t hashCap = pInfo->pBuild->inputStat.inputRowNum > 0 ? (pInfo->pBuild->inputStat.inputRowNum * 1.5) : 1024;
×
UNCOV
1202
  pInfo->pKeyHash = tSimpleHashInit(hashCap, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY));
×
UNCOV
1203
  if (pInfo->pKeyHash == NULL) {
×
1204
    code = terrno;
×
1205
    goto _return;
×
1206
  }
1207

UNCOV
1208
  HJ_ERR_JRET(hJoinHandleConds(pInfo, pJoinNode));
×
1209

UNCOV
1210
  HJ_ERR_JRET(hJoinInitResBlocks(pInfo, pJoinNode));
×
1211

UNCOV
1212
  HJ_ERR_JRET(hJoinSetImplFp(pInfo));
×
1213

UNCOV
1214
  HJ_ERR_JRET(appendDownstream(pOperator, pDownstream, numOfDownstream));
×
1215

UNCOV
1216
  pOperator->fpSet = createOperatorFpSet(optrDummyOpenFn, hJoinMainProcess, NULL, destroyHashJoinOperator, optrDefaultBufFn, NULL, optrDefaultGetNextExtFn, NULL);
×
1217

UNCOV
1218
  qDebug("create hash Join operator done");
×
1219

UNCOV
1220
  *pOptrInfo = pOperator;
×
UNCOV
1221
  return TSDB_CODE_SUCCESS;
×
1222

1223
_return:
×
1224

1225
  if (pInfo != NULL) {
×
1226
    destroyHashJoinOperator(pInfo);
×
1227
  }
1228
  destroyOperatorAndDownstreams(pOperator, pDownstream, numOfDownstream);
×
1229
  pTaskInfo->code = code;
×
1230
  return code;
×
1231
}
1232

1233

STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc