TDengine/source/dnode/vnode/src/tsdb/dev/tsdbSttFReaderWriter.c

759 lines
24 KiB
C
Raw Normal View History

2023-03-23 08:28:13 +00:00
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
2023-05-26 15:42:47 +00:00
#include "inc/tsdbSttFReaderWriter.h"
2023-03-23 10:40:44 +00:00
2023-04-23 08:18:18 +00:00
typedef struct {
int64_t prevFooter;
SFDataPtr dict[4]; // 0:bloom filter, 1:SSttBlk, 2:STbStatisBlk, 3:SDelBlk
uint8_t reserved[24];
} SFSttFooter;
2023-03-24 06:24:34 +00:00
2023-04-23 08:31:11 +00:00
// SSttFReader ============================================================
2023-04-23 09:37:12 +00:00
struct SSttFileReader {
2023-05-26 15:42:47 +00:00
SSttFileReaderConfig config;
TSttSegReaderArray segReaderArray;
STsdbFD *fd;
};
struct SSttSegReader {
SSttFileReader *reader;
struct {
bool bloomFilterLoaded;
bool sttBlkLoaded;
bool delBlkLoaded;
bool statisBlkLoaded;
} ctx;
SFSttFooter footer;
void *bloomFilter;
TSttBlkArray sttBlkArray;
TDelBlkArray delBlkArray;
TStatisBlkArray statisBlkArray;
2023-04-23 08:31:11 +00:00
};
2023-04-26 03:16:49 +00:00
// SSttFileReader
2023-05-26 15:42:47 +00:00
static int32_t tsdbSttSegReaderOpen(SSttFileReader *reader, int64_t offset, SSttSegReader **segReader) {
2023-04-23 09:37:12 +00:00
int32_t code = 0;
2023-05-26 15:42:47 +00:00
int32_t lino = 0;
int32_t vid = TD_VID(reader->config.tsdb->pVnode);
ASSERT(offset >= TSDB_FHDR_SIZE);
segReader[0] = taosMemoryCalloc(1, sizeof(*segReader[0]));
if (segReader[0] == NULL) return TSDB_CODE_OUT_OF_MEMORY;
segReader[0]->reader = reader;
code = tsdbReadFile(reader->fd, offset, (uint8_t *)(&segReader[0]->footer), sizeof(segReader[0]->footer));
TSDB_CHECK_CODE(code, lino, _exit);
_exit:
if (code) {
tsdbError("vgId:%d %s failed at line %d since %s", vid, __func__, lino, tstrerror(code));
taosMemoryFree(segReader[0]);
segReader[0] = NULL;
}
2023-04-23 09:37:12 +00:00
return code;
}
2023-05-26 15:42:47 +00:00
static int32_t tsdbSttSegReaderClose(SSttSegReader **segReader) {
if (!segReader[0]) return 0;
if (segReader[0]->ctx.bloomFilterLoaded) {
// TODO
}
if (segReader[0]->ctx.sttBlkLoaded) {
TARRAY2_FREE(&segReader[0]->sttBlkArray);
}
if (segReader[0]->ctx.delBlkLoaded) {
TARRAY2_FREE(&segReader[0]->delBlkArray);
}
if (segReader[0]->ctx.statisBlkLoaded) {
TARRAY2_FREE(&segReader[0]->statisBlkArray);
}
taosMemoryFree(segReader[0]);
segReader[0] = NULL;
return 0;
2023-04-23 09:37:12 +00:00
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReaderOpen(const SSttFileReaderConfig *config, SSttFileReader **reader) {
2023-04-26 03:16:49 +00:00
int32_t code = 0;
2023-05-26 15:42:47 +00:00
int32_t lino = 0;
int32_t vid = TD_VID(config->tsdb->pVnode);
reader[0] = taosMemoryCalloc(1, sizeof(*reader[0]));
if (reader[0] == NULL) return TSDB_CODE_OUT_OF_MEMORY;
reader[0]->config = config[0];
TARRAY2_INIT(&reader[0]->segReaderArray);
// open file
char fname[TSDB_FILENAME_LEN];
tsdbTFileName(config->tsdb, &config->file, fname);
code = tsdbOpenFile(fname, config->szPage, TD_FILE_READ, &reader[0]->fd);
TSDB_CHECK_CODE(code, lino, _exit);
// open each segment reader
int64_t size = config->file.size;
while (size > 0) {
SSttSegReader *segReader;
code = tsdbSttSegReaderOpen(reader[0], size - sizeof(SFSttFooter), &segReader);
TSDB_CHECK_CODE(code, lino, _exit);
code = TARRAY2_APPEND(&reader[0]->segReaderArray, segReader);
TSDB_CHECK_CODE(code, lino, _exit);
size = segReader->footer.prevFooter;
}
ASSERT(TARRAY2_SIZE(&reader[0]->segReaderArray) == config->file.stt.nseg);
_exit:
if (code) {
tsdbError("vgId:%d %s failed at line %d since %s", vid, __func__, lino, tstrerror(code));
tsdbSttFReaderClose(reader);
}
2023-04-26 03:16:49 +00:00
return code;
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReaderClose(SSttFileReader **reader) {
tsdbCloseFile(&reader[0]->fd);
TARRAY2_CLEAR_FREE(&reader[0]->segReaderArray, tsdbSttSegReaderClose);
taosMemoryFree(reader[0]);
reader[0] = NULL;
return 0;
2023-04-26 03:16:49 +00:00
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReaderGetSegReader(SSttFileReader *reader, const TSttSegReaderArray **segReaderArray) {
segReaderArray[0] = &reader->segReaderArray;
return 0;
2023-04-26 03:16:49 +00:00
}
2023-05-26 15:42:47 +00:00
// SSttFSegReader
int32_t tsdbSttFReadBloomFilter(SSttSegReader *reader, const void *pFilter) {
2023-04-26 03:16:49 +00:00
int32_t code = 0;
// TODO
return code;
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReadStatisBlk(SSttSegReader *reader, const TStatisBlkArray **statisBlkArray) {
if (!reader->ctx.statisBlkLoaded) {
SFDataPtr fptr = reader->footer.dict[2];
if (fptr.size > 0) {
ASSERT(fptr.size % sizeof(STbStatisBlk) == 0);
int32_t size = fptr.size / sizeof(STbStatisBlk);
void *data = taosMemoryMalloc(fptr.size);
if (!data) return TSDB_CODE_OUT_OF_MEMORY;
int32_t code = tsdbReadFile(reader->reader->fd, fptr.offset, data, fptr.size);
if (code) return code;
TARRAY2_INIT_EX(&reader->statisBlkArray, size, size, data);
} else {
TARRAY2_INIT(&reader->statisBlkArray);
}
reader->ctx.statisBlkLoaded = true;
}
statisBlkArray[0] = &reader->statisBlkArray;
return 0;
2023-04-26 03:16:49 +00:00
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReadDelBlk(SSttSegReader *reader, const TDelBlkArray **delBlkArray) {
if (!reader->ctx.delBlkLoaded) {
SFDataPtr fptr = reader->footer.dict[3];
if (fptr.size > 0) {
ASSERT(fptr.size % sizeof(SDelBlk) == 0);
int32_t size = fptr.size / sizeof(SDelBlk);
void *data = taosMemoryMalloc(fptr.size);
if (!data) return TSDB_CODE_OUT_OF_MEMORY;
int32_t code = tsdbReadFile(reader->reader->fd, fptr.offset, data, fptr.size);
if (code) return code;
TARRAY2_INIT_EX(&reader->delBlkArray, size, size, data);
} else {
TARRAY2_INIT(&reader->delBlkArray);
}
reader->ctx.delBlkLoaded = true;
}
delBlkArray[0] = &reader->delBlkArray;
return 0;
}
int32_t tsdbSttFReadSttBlk(SSttSegReader *reader, const TSttBlkArray **sttBlkArray) {
if (!reader->ctx.sttBlkLoaded) {
SFDataPtr fptr = reader->footer.dict[1];
if (fptr.size > 0) {
ASSERT(fptr.size % sizeof(SSttBlk) == 0);
int32_t size = fptr.size / sizeof(SSttBlk);
void *data = taosMemoryMalloc(fptr.size);
if (!data) return TSDB_CODE_OUT_OF_MEMORY;
int32_t code = tsdbReadFile(reader->reader->fd, fptr.offset, data, fptr.size);
if (code) return code;
TARRAY2_INIT_EX(&reader->sttBlkArray, size, size, data);
} else {
TARRAY2_INIT(&reader->sttBlkArray);
}
reader->ctx.sttBlkLoaded = true;
}
sttBlkArray[0] = &reader->sttBlkArray;
return 0;
2023-04-26 03:16:49 +00:00
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReadSttBlock(SSttSegReader *reader, const SSttBlk *sttBlk, SBlockData *bData) {
2023-04-26 03:16:49 +00:00
int32_t code = 0;
// TODO
return code;
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReadDelBlock(SSttSegReader *reader, const SDelBlk *delBlk, SDelBlock *dData) {
2023-04-26 03:16:49 +00:00
int32_t code = 0;
// TODO
return code;
}
2023-05-26 15:42:47 +00:00
int32_t tsdbSttFReadStatisBlock(SSttSegReader *reader, const STbStatisBlk *statisBlk, STbStatisBlock *sData) {
2023-04-26 03:16:49 +00:00
int32_t code = 0;
// TODO
return code;
}
2023-04-23 08:31:11 +00:00
// SSttFWriter ============================================================
2023-04-23 09:37:12 +00:00
struct SSttFileWriter {
SSttFileWriterConfig config;
2023-05-26 11:32:02 +00:00
struct {
bool opened;
} ctx;
2023-04-11 06:22:36 +00:00
// file
2023-05-26 11:32:02 +00:00
STFile file;
2023-03-29 09:34:36 +00:00
// data
2023-05-26 15:51:18 +00:00
TSttBlkArray sttBlkArray;
TDelBlkArray delBlkArray;
TStatisBlkArray statisBlkArray;
void *bloomFilter; // TODO
SFSttFooter footer;
SBlockData bData[1];
SDelBlock dData[1];
STbStatisBlock sData[1];
2023-03-24 06:24:34 +00:00
// helper data
SSkmInfo skmTb;
SSkmInfo skmRow;
2023-03-27 08:30:22 +00:00
int32_t aBufSize[5];
uint8_t *aBuf[5];
2023-05-26 11:32:02 +00:00
STsdbFD *fd;
2023-03-23 10:40:44 +00:00
};
2023-03-23 08:28:13 +00:00
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteTSDataBlock(SSttFileWriter *writer) {
if (writer->bData->nRow == 0) return 0;
2023-03-23 08:28:13 +00:00
int32_t code = 0;
2023-05-26 11:32:02 +00:00
int32_t lino = 0;
SSttBlk sttBlk[1];
sttBlk->suid = writer->bData->suid;
sttBlk->minUid = writer->bData->uid ? writer->bData->uid : writer->bData->aUid[0];
sttBlk->maxUid = writer->bData->uid ? writer->bData->uid : writer->bData->aUid[writer->bData->nRow - 1];
sttBlk->minKey = sttBlk->maxKey = writer->bData->aTSKEY[0];
sttBlk->minVer = sttBlk->maxVer = writer->bData->aVersion[0];
sttBlk->nRow = writer->bData->nRow;
for (int32_t iRow = 1; iRow < writer->bData->nRow; iRow++) {
if (sttBlk->minKey > writer->bData->aTSKEY[iRow]) sttBlk->minKey = writer->bData->aTSKEY[iRow];
if (sttBlk->maxKey < writer->bData->aTSKEY[iRow]) sttBlk->maxKey = writer->bData->aTSKEY[iRow];
if (sttBlk->minVer > writer->bData->aVersion[iRow]) sttBlk->minVer = writer->bData->aVersion[iRow];
if (sttBlk->maxVer < writer->bData->aVersion[iRow]) sttBlk->maxVer = writer->bData->aVersion[iRow];
}
code = tCmprBlockData(writer->bData, writer->config.cmprAlg, NULL, NULL, writer->config.aBuf, writer->aBufSize);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-23 10:40:44 +00:00
2023-05-26 11:32:02 +00:00
sttBlk->bInfo.offset = writer->file.size;
sttBlk->bInfo.szKey = writer->aBufSize[2] + writer->aBufSize[3];
sttBlk->bInfo.szBlock = writer->aBufSize[0] + writer->aBufSize[1] + sttBlk->bInfo.szKey;
2023-03-24 09:57:37 +00:00
2023-05-26 11:32:02 +00:00
for (int32_t i = 3; i >= 0; i--) {
if (writer->aBufSize[i]) {
code = tsdbWriteFile(writer->fd, writer->file.size, writer->config.aBuf[i], writer->aBufSize[i]);
TSDB_CHECK_CODE(code, lino, _exit);
writer->file.size += writer->aBufSize[i];
2023-03-27 08:30:22 +00:00
}
}
2023-05-26 11:32:02 +00:00
tBlockDataClear(writer->bData);
2023-03-27 08:30:22 +00:00
2023-05-26 11:32:02 +00:00
code = TARRAY2_APPEND_P(&writer->sttBlkArray, sttBlk);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-23 10:40:44 +00:00
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
tstrerror(code));
2023-03-23 10:40:44 +00:00
}
2023-03-23 08:28:13 +00:00
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteStatisBlock(SSttFileWriter *writer) {
if (writer->sData->nRow == 0) return 0;
2023-03-30 09:05:58 +00:00
int32_t code = 0;
2023-05-26 11:32:02 +00:00
int32_t lino = 0;
2023-04-03 05:59:37 +00:00
2023-05-26 11:32:02 +00:00
STbStatisBlk statisBlk[1];
2023-04-03 05:59:37 +00:00
2023-05-26 11:32:02 +00:00
statisBlk->nRow = writer->sData->nRow;
statisBlk->minTid.suid = writer->sData->aData[0][0];
statisBlk->minTid.uid = writer->sData->aData[1][0];
statisBlk->maxTid.suid = writer->sData->aData[0][writer->sData->nRow - 1];
statisBlk->maxTid.uid = writer->sData->aData[1][writer->sData->nRow - 1];
statisBlk->minVer = statisBlk->maxVer = statisBlk->maxVer = writer->sData->aData[2][0];
for (int32_t iRow = 1; iRow < writer->sData->nRow; iRow++) {
if (statisBlk->minVer > writer->sData->aData[2][iRow]) statisBlk->minVer = writer->sData->aData[2][iRow];
if (statisBlk->maxVer < writer->sData->aData[2][iRow]) statisBlk->maxVer = writer->sData->aData[2][iRow];
2023-04-03 05:59:37 +00:00
}
2023-05-26 11:32:02 +00:00
statisBlk->dp.offset = writer->file.size;
statisBlk->dp.size = 0;
2023-04-03 05:59:37 +00:00
2023-04-11 06:22:36 +00:00
// TODO: add compression here
2023-05-26 11:32:02 +00:00
int64_t tsize = sizeof(int64_t) * writer->sData->nRow;
for (int32_t i = 0; i < ARRAY_SIZE(writer->sData->aData); i++) {
code = tsdbWriteFile(writer->fd, writer->file.size, (const uint8_t *)writer->sData->aData[i], tsize);
TSDB_CHECK_CODE(code, lino, _exit);
statisBlk->dp.size += tsize;
writer->file.size += tsize;
2023-04-03 05:59:37 +00:00
}
2023-05-26 11:32:02 +00:00
tTbStatisBlockClear(writer->sData);
2023-03-30 09:05:58 +00:00
2023-05-26 11:32:02 +00:00
code = TARRAY2_APPEND_P(&writer->statisBlkArray, statisBlk);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-30 09:05:58 +00:00
2023-04-03 05:59:37 +00:00
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
tstrerror(code));
2023-04-03 05:59:37 +00:00
}
2023-03-30 09:05:58 +00:00
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteDelBlock(SSttFileWriter *writer) {
if (writer->dData->nRow == 0) return 0;
2023-03-30 09:05:58 +00:00
int32_t code = 0;
2023-04-03 05:59:37 +00:00
int32_t lino;
2023-05-26 11:32:02 +00:00
SDelBlk delBlk[1];
2023-04-03 05:59:37 +00:00
2023-05-26 11:32:02 +00:00
delBlk->nRow = writer->sData->nRow;
delBlk->minTid.suid = writer->sData->aData[0][0];
delBlk->minTid.uid = writer->sData->aData[1][0];
delBlk->maxTid.suid = writer->sData->aData[0][writer->sData->nRow - 1];
delBlk->maxTid.uid = writer->sData->aData[1][writer->sData->nRow - 1];
delBlk->minVer = delBlk->maxVer = delBlk->maxVer = writer->sData->aData[2][0];
for (int32_t iRow = 1; iRow < writer->sData->nRow; iRow++) {
if (delBlk->minVer > writer->sData->aData[2][iRow]) delBlk->minVer = writer->sData->aData[2][iRow];
if (delBlk->maxVer < writer->sData->aData[2][iRow]) delBlk->maxVer = writer->sData->aData[2][iRow];
2023-04-03 05:59:37 +00:00
}
2023-05-26 11:32:02 +00:00
delBlk->dp.offset = writer->file.size;
delBlk->dp.size = 0; // TODO
2023-04-03 05:59:37 +00:00
2023-05-26 11:32:02 +00:00
int64_t tsize = sizeof(int64_t) * writer->dData->nRow;
for (int32_t i = 0; i < ARRAY_SIZE(writer->dData->aData); i++) {
code = tsdbWriteFile(writer->fd, writer->file.size, (const uint8_t *)writer->dData->aData[i], tsize);
2023-04-03 05:59:37 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-05-26 11:32:02 +00:00
delBlk->dp.size += tsize;
writer->file.size += tsize;
2023-04-03 05:59:37 +00:00
}
2023-05-26 11:32:02 +00:00
tDelBlockDestroy(writer->dData);
2023-04-03 05:59:37 +00:00
2023-05-26 11:32:02 +00:00
code = TARRAY2_APPEND_P(&writer->delBlkArray, delBlk);
TSDB_CHECK_CODE(code, lino, _exit);
2023-04-03 05:59:37 +00:00
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
2023-04-03 05:59:37 +00:00
tstrerror(code));
} else {
// tsdbTrace();
}
2023-03-30 09:05:58 +00:00
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteSttBlk(SSttFileWriter *writer) {
2023-03-24 09:57:37 +00:00
int32_t code = 0;
2023-03-27 08:30:22 +00:00
int32_t lino;
2023-05-26 11:32:02 +00:00
writer->footer.dict[1].offset = writer->file.size;
writer->footer.dict[1].size = sizeof(SSttBlk) * TARRAY2_SIZE(&writer->sttBlkArray);
2023-03-30 09:43:54 +00:00
2023-05-26 11:32:02 +00:00
if (writer->footer.dict[1].size) {
code = tsdbWriteFile(writer->fd, writer->file.size, (const uint8_t *)TARRAY2_DATA(&writer->sttBlkArray),
writer->footer.dict[1].size);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-30 09:43:54 +00:00
2023-05-26 11:32:02 +00:00
writer->file.size += writer->footer.dict[1].size;
2023-03-27 08:30:22 +00:00
}
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
tstrerror(code));
2023-03-27 08:30:22 +00:00
}
2023-03-24 09:57:37 +00:00
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteStatisBlk(SSttFileWriter *writer) {
2023-03-24 09:57:37 +00:00
int32_t code = 0;
2023-03-30 09:43:54 +00:00
int32_t lino;
2023-03-27 08:30:22 +00:00
2023-05-26 11:32:02 +00:00
writer->footer.dict[2].offset = writer->file.size;
writer->footer.dict[2].size = sizeof(STbStatisBlock) * TARRAY2_SIZE(&writer->statisBlkArray);
2023-03-30 09:43:54 +00:00
2023-05-26 11:32:02 +00:00
if (writer->footer.dict[2].size) {
code = tsdbWriteFile(writer->fd, writer->file.size, (const uint8_t *)TARRAY2_DATA(&writer->statisBlkArray),
writer->footer.dict[2].size);
TSDB_CHECK_CODE(code, lino, _exit);
writer->file.size += writer->footer.dict[2].size;
2023-03-30 09:43:54 +00:00
}
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
tstrerror(code));
2023-03-30 09:43:54 +00:00
}
2023-03-24 09:57:37 +00:00
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteDelBlk(SSttFileWriter *writer) {
2023-03-24 09:57:37 +00:00
int32_t code = 0;
2023-03-30 09:43:54 +00:00
int32_t lino;
2023-05-26 11:32:02 +00:00
writer->footer.dict[3].offset = writer->file.size;
writer->footer.dict[3].size = sizeof(SDelBlk) * TARRAY2_SIZE(&writer->delBlkArray);
2023-03-30 09:43:54 +00:00
2023-05-26 11:32:02 +00:00
if (writer->footer.dict[3].size) {
code = tsdbWriteFile(writer->fd, writer->file.size, (const uint8_t *)TARRAY2_DATA(&writer->delBlkArray),
writer->footer.dict[3].size);
TSDB_CHECK_CODE(code, lino, _exit);
writer->file.size += writer->footer.dict[3].size;
2023-03-30 09:43:54 +00:00
}
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
tstrerror(code));
2023-03-30 09:43:54 +00:00
}
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteFooter(SSttFileWriter *writer) {
int32_t code = tsdbWriteFile(writer->fd, writer->file.size, (const uint8_t *)&writer->footer, sizeof(writer->footer));
writer->file.size += sizeof(writer->footer);
2023-03-24 09:57:37 +00:00
return code;
}
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFWriterDoOpen(SSttFileWriter *writer) {
2023-03-23 08:10:08 +00:00
int32_t code = 0;
2023-05-26 11:32:02 +00:00
int32_t lino = 0;
int32_t vid = TD_VID(writer->config.tsdb->pVnode);
2023-03-22 11:24:40 +00:00
2023-05-26 11:32:02 +00:00
// set
writer->file = writer->config.file;
writer->file.stt.nseg++;
if (!writer->config.skmTb) writer->config.skmTb = &writer->skmTb;
if (!writer->config.skmRow) writer->config.skmRow = &writer->skmRow;
if (!writer->config.aBuf) writer->config.aBuf = writer->aBuf;
2023-03-22 11:24:40 +00:00
2023-05-26 11:32:02 +00:00
// open file
int32_t flag;
char fname[TSDB_FILENAME_LEN];
2023-04-11 06:22:36 +00:00
2023-05-26 11:32:02 +00:00
if (writer->file.size) {
flag = TD_FILE_READ | TD_FILE_WRITE;
} else {
flag = TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE | TD_FILE_TRUNC;
2023-03-24 09:57:37 +00:00
}
2023-05-26 11:32:02 +00:00
tsdbTFileName(writer->config.tsdb, &writer->file, fname);
code = tsdbOpenFile(fname, writer->config.szPage, flag, &writer->fd);
TSDB_CHECK_CODE(code, lino, _exit);
if (!writer->file.size) {
uint8_t hdr[TSDB_FHDR_SIZE] = {0};
code = tsdbWriteFile(writer->fd, 0, hdr, sizeof(hdr));
TSDB_CHECK_CODE(code, lino, _exit);
writer->file.size += sizeof(hdr);
2023-03-24 09:57:37 +00:00
}
2023-03-22 11:24:40 +00:00
2023-03-24 09:57:37 +00:00
_exit:
2023-05-26 11:32:02 +00:00
if (code) {
tsdbError("vgId:%d %s failed at line %d since %s", vid, __func__, lino, tstrerror(code));
} else {
writer->ctx.opened = true;
2023-03-24 09:57:37 +00:00
}
2023-05-26 11:32:02 +00:00
return 0;
2023-03-24 09:57:37 +00:00
}
2023-05-26 11:32:02 +00:00
static void tsdbSttFWriterDoClose(SSttFileWriter *pWriter) {
// TODO: do clear the struct
}
2023-04-11 07:01:42 +00:00
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoWriteBloomFilter(SSttFileWriter *writer) {
// TODO
return 0;
}
2023-05-26 11:41:27 +00:00
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFileDoUpdateHeader(SSttFileWriter *writer) {
// TODO
return 0;
}
2023-05-26 11:41:27 +00:00
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFWriterCloseCommit(SSttFileWriter *writer, STFileOp *op) {
int32_t lino;
int32_t code;
int32_t vid = TD_VID(writer->config.tsdb->pVnode);
2023-03-27 08:30:22 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteTSDataBlock(writer);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-27 08:30:22 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteStatisBlock(writer);
2023-03-27 08:30:22 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteDelBlock(writer);
TSDB_CHECK_CODE(code, lino, _exit);
2023-04-11 06:22:36 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteSttBlk(writer);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-30 09:05:58 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteStatisBlk(writer);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-24 09:57:37 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteDelBlk(writer);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-24 09:57:37 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoWriteBloomFilter(writer);
TSDB_CHECK_CODE(code, lino, _exit);
code = tsdbSttFileDoWriteFooter(writer);
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-24 09:57:37 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFileDoUpdateHeader(writer);
2023-03-24 09:57:37 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-05-26 11:32:02 +00:00
code = tsdbFsyncFile(writer->fd);
2023-03-22 11:24:40 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-05-26 11:41:27 +00:00
tsdbCloseFile(&writer->fd);
2023-05-26 11:32:02 +00:00
ASSERT(writer->config.file.size > writer->file.size);
op->optype = writer->config.file.size ? TSDB_FOP_MODIFY : TSDB_FOP_CREATE;
op->fid = writer->config.file.fid;
op->of = writer->config.file;
op->nf = writer->file;
2023-03-22 11:24:40 +00:00
_exit:
if (code) {
2023-05-11 10:16:55 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", vid, __func__, lino, tstrerror(code));
2023-03-22 11:24:40 +00:00
}
2023-03-24 06:24:34 +00:00
return code;
2023-03-22 11:24:40 +00:00
}
2023-05-26 11:41:27 +00:00
2023-05-26 11:32:02 +00:00
static int32_t tsdbSttFWriterCloseAbort(SSttFileWriter *writer) {
2023-05-26 11:41:27 +00:00
char fname[TSDB_FILENAME_LEN];
tsdbTFileName(writer->config.tsdb, &writer->config.file, fname);
if (writer->config.file.size) { // truncate the file to the original size
ASSERT(writer->config.file.size <= writer->file.size);
if (writer->config.file.size < writer->file.size) {
taosFtruncateFile(writer->fd->pFD, writer->config.file.size);
tsdbCloseFile(&writer->fd);
}
} else { // remove the file
tsdbCloseFile(&writer->fd);
taosRemoveFile(fname);
}
2023-05-26 11:32:02 +00:00
return 0;
}
2023-05-26 11:41:27 +00:00
int32_t tsdbSttFWriterOpen(const SSttFileWriterConfig *config, SSttFileWriter **writer) {
writer[0] = taosMemoryMalloc(sizeof(*writer[0]));
if (writer[0] == NULL) return TSDB_CODE_OUT_OF_MEMORY;
writer[0]->config = config[0];
writer[0]->ctx.opened = false;
return 0;
}
2023-05-26 11:32:02 +00:00
int32_t tsdbSttFWriterClose(SSttFileWriter **writer, int8_t abort, STFileOp *op) {
2023-03-24 06:24:34 +00:00
int32_t code = 0;
2023-05-11 10:16:55 +00:00
int32_t lino = 0;
2023-05-26 11:32:02 +00:00
int32_t vid = TD_VID(writer[0]->config.tsdb->pVnode);
2023-03-24 09:57:37 +00:00
2023-05-26 11:32:02 +00:00
if (!writer[0]->ctx.opened) {
op->optype = TSDB_FOP_NONE;
} else {
2023-05-26 11:41:27 +00:00
if (abort) {
code = tsdbSttFWriterCloseAbort(writer[0]);
2023-05-11 10:16:55 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-05-26 11:32:02 +00:00
} else {
2023-05-26 11:41:27 +00:00
code = tsdbSttFWriterCloseCommit(writer[0], op);
2023-05-11 10:16:55 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-04-11 07:01:42 +00:00
}
2023-05-26 11:32:02 +00:00
tsdbSttFWriterDoClose(writer[0]);
2023-04-11 07:01:42 +00:00
}
2023-05-26 11:32:02 +00:00
taosMemoryFree(writer[0]);
writer[0] = NULL;
2023-03-24 09:57:37 +00:00
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", vid, __func__, lino, tstrerror(code));
2023-03-24 09:57:37 +00:00
}
2023-03-24 06:24:34 +00:00
return code;
2023-03-22 11:24:40 +00:00
}
2023-05-26 11:32:02 +00:00
int32_t tsdbSttFWriteTSData(SSttFileWriter *writer, SRowInfo *row) {
2023-03-23 08:10:08 +00:00
int32_t code = 0;
2023-05-26 11:32:02 +00:00
int32_t lino = 0;
2023-03-23 08:10:08 +00:00
2023-05-26 11:32:02 +00:00
if (!writer->ctx.opened) {
code = tsdbSttFWriterDoOpen(writer);
TSDB_CHECK_CODE(code, lino, _exit);
}
2023-03-30 09:05:58 +00:00
2023-05-26 11:32:02 +00:00
TABLEID *tbid = (TABLEID *)row;
TSDBROW *pRow = &row->row;
TSDBKEY key = TSDBROW_KEY(pRow);
if (!TABLE_SAME_SCHEMA(writer->bData[0].suid, writer->bData[0].uid, tbid->suid, tbid->uid)) {
if (writer->bData[0].nRow > 0) {
code = tsdbSttFileDoWriteTSDataBlock(writer);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-23 08:10:08 +00:00
}
2023-05-26 11:32:02 +00:00
if (writer->sData[0].nRow >= writer->config.maxRow) {
code = tsdbSttFileDoWriteStatisBlock(writer);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-30 09:05:58 +00:00
}
2023-05-26 11:32:02 +00:00
writer->sData[0].aData[0][writer->sData[0].nRow] = tbid->suid; // suid
writer->sData[0].aData[1][writer->sData[0].nRow] = tbid->uid; // uid
writer->sData[0].aData[2][writer->sData[0].nRow] = key.ts; // skey
writer->sData[0].aData[3][writer->sData[0].nRow] = key.version; // sver
writer->sData[0].aData[4][writer->sData[0].nRow] = key.ts; // ekey
writer->sData[0].aData[5][writer->sData[0].nRow] = key.version; // ever
writer->sData[0].aData[6][writer->sData[0].nRow] = 1; // count
writer->sData[0].nRow++;
2023-03-30 09:05:58 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbUpdateSkmTb(writer->config.tsdb, tbid, writer->config.skmTb);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-04-11 06:22:36 +00:00
2023-05-26 06:49:50 +00:00
TABLEID id = {
.suid = tbid->suid,
.uid = tbid->uid ? 0 : tbid->uid,
};
2023-05-26 11:32:02 +00:00
code = tBlockDataInit(&writer->bData[0], &id, writer->config.skmTb->pTSchema, NULL, 0);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-24 06:24:34 +00:00
}
2023-05-26 11:32:02 +00:00
if (row->row.type == TSDBROW_ROW_FMT) {
code = tsdbUpdateSkmRow(writer->config.tsdb, tbid, TSDBROW_SVERSION(pRow), writer->config.skmRow);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-04-11 06:22:36 +00:00
}
2023-05-26 11:32:02 +00:00
code = tBlockDataAppendRow(&writer->bData[0], pRow, writer->config.skmRow->pTSchema, tbid->uid);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-23 08:10:08 +00:00
2023-05-26 11:32:02 +00:00
if (writer->bData[0].nRow >= writer->config.maxRow) {
code = tsdbSttFileDoWriteTSDataBlock(writer);
2023-05-26 06:49:50 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
2023-03-23 08:10:08 +00:00
}
2023-05-26 11:32:02 +00:00
if (key.ts > writer->sData[0].aData[4][writer->sData[0].nRow - 1]) {
writer->sData[0].aData[4][writer->sData[0].nRow - 1] = key.ts; // ekey
writer->sData[0].aData[5][writer->sData[0].nRow - 1] = key.version; // ever
writer->sData[0].aData[6][writer->sData[0].nRow - 1]++; // count
} else if (key.ts == writer->sData[0].aData[4][writer->sData[0].nRow - 1]) {
writer->sData[0].aData[4][writer->sData[0].nRow - 1] = key.ts; // ekey
writer->sData[0].aData[5][writer->sData[0].nRow - 1] = key.version; // ever
2023-03-30 09:05:58 +00:00
} else {
ASSERTS(0, "timestamp should be in ascending order");
}
2023-03-23 08:10:08 +00:00
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
2023-05-26 06:49:50 +00:00
tstrerror(code));
2023-03-23 08:10:08 +00:00
}
2023-03-24 06:24:34 +00:00
return code;
2023-03-23 08:10:08 +00:00
}
2023-05-26 11:32:02 +00:00
int32_t tsdbSttFWriteTSDataBlock(SSttFileWriter *writer, SBlockData *bdata) {
2023-05-26 07:25:29 +00:00
int32_t code = 0;
int32_t lino = 0;
SRowInfo rowInfo;
2023-05-26 11:32:02 +00:00
rowInfo.suid = bdata->suid;
for (int32_t i = 0; i < bdata->nRow; i++) {
rowInfo.uid = bdata->uid ? bdata->uid : bdata->aUid[i];
rowInfo.row = tsdbRowFromBlockData(bdata, i);
2023-05-26 07:25:29 +00:00
2023-05-26 11:32:02 +00:00
code = tsdbSttFWriteTSData(writer, &rowInfo);
2023-05-26 07:25:29 +00:00
TSDB_CHECK_CODE(code, lino, _exit);
}
_exit:
if (code) {
2023-05-26 11:32:02 +00:00
tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(writer->config.tsdb->pVnode), __func__, lino,
2023-05-26 07:25:29 +00:00
tstrerror(code));
}
2023-05-26 06:49:50 +00:00
return 0;
}
2023-05-26 11:32:02 +00:00
int32_t tsdbSttFWriteDLData(SSttFileWriter *writer, TABLEID *tbid, SDelData *pDelData) {
2023-04-11 06:22:36 +00:00
ASSERTS(0, "TODO: Not implemented yet");
2023-05-26 11:32:02 +00:00
int32_t code;
if (!writer->ctx.opened) {
code = tsdbSttFWriterDoOpen(writer);
return code;
}
writer->dData[0].aData[0][writer->dData[0].nRow] = tbid->suid; // suid
writer->dData[0].aData[1][writer->dData[0].nRow] = tbid->uid; // uid
writer->dData[0].aData[2][writer->dData[0].nRow] = pDelData->version; // version
writer->dData[0].aData[3][writer->dData[0].nRow] = pDelData->sKey; // skey
writer->dData[0].aData[4][writer->dData[0].nRow] = pDelData->eKey; // ekey
writer->dData[0].nRow++;
2023-03-30 09:05:58 +00:00
2023-05-26 11:32:02 +00:00
if (writer->dData[0].nRow >= writer->config.maxRow) {
return tsdbSttFileDoWriteDelBlock(writer);
2023-03-30 09:05:58 +00:00
} else {
return 0;
}
2023-03-24 06:24:34 +00:00
}