导出系统
All checks were successful
构建并部署到测试环境(无 SSH) / build-and-deploy (push) Successful in 7m50s

This commit is contained in:
Break
2026-06-15 15:28:29 +08:00
parent 2f0b24ce88
commit 7ec84fbc0f
17 changed files with 939 additions and 260 deletions

View File

@@ -5,6 +5,7 @@ import (
"encoding/csv"
"errors"
"fmt"
"io"
"os"
"path/filepath"
"time"
@@ -82,18 +83,18 @@ func isFinalRetry(ctx context.Context) bool {
return retryCount >= maxRetry
}
func writeExportFile(format string, headers []string, rows [][]string) (string, int64, error) {
func writeExportFile(format string, headers []string, rows [][]string, withoutHeader bool) (string, int64, error) {
switch format {
case constants.ExportTaskFormatCSV:
return writeCSVFile(headers, rows)
return writeCSVFile(headers, rows, withoutHeader)
case constants.ExportTaskFormatXLSX:
return writeXLSXFile(headers, rows)
return writeXLSXFile(headers, rows, withoutHeader)
default:
return "", 0, fmt.Errorf("不支持的导出格式: %s", format)
}
}
func writeCSVFile(headers []string, rows [][]string) (string, int64, error) {
func writeCSVFile(headers []string, rows [][]string, withoutHeader bool) (string, int64, error) {
file, err := os.CreateTemp("", "export-*.csv")
if err != nil {
return "", 0, err
@@ -101,8 +102,10 @@ func writeCSVFile(headers []string, rows [][]string) (string, int64, error) {
defer file.Close()
writer := csv.NewWriter(file)
if err := writer.Write(headers); err != nil {
return "", 0, err
if !withoutHeader {
if err := writer.Write(headers); err != nil {
return "", 0, err
}
}
for _, row := range rows {
if err := writer.Write(row); err != nil {
@@ -125,7 +128,7 @@ func writeCSVFile(headers []string, rows [][]string) (string, int64, error) {
return file.Name(), stat.Size(), nil
}
func writeXLSXFile(headers []string, rows [][]string) (string, int64, error) {
func writeXLSXFile(headers []string, rows [][]string, withoutHeader bool) (string, int64, error) {
file := excelize.NewFile()
sheet := file.GetSheetName(file.GetActiveSheetIndex())
@@ -134,16 +137,20 @@ func writeXLSXFile(headers []string, rows [][]string) (string, int64, error) {
return "", 0, err
}
headerValues := make([]interface{}, 0, len(headers))
for _, item := range headers {
headerValues = append(headerValues, item)
}
if err := streamWriter.SetRow("A1", headerValues); err != nil {
return "", 0, err
rowIndex := 1
if !withoutHeader {
headerValues := make([]interface{}, 0, len(headers))
for _, item := range headers {
headerValues = append(headerValues, item)
}
if err := streamWriter.SetRow("A1", headerValues); err != nil {
return "", 0, err
}
rowIndex++
}
for i, row := range rows {
axis, err := excelize.CoordinatesToCellName(1, i+2)
for _, row := range rows {
axis, err := excelize.CoordinatesToCellName(1, rowIndex)
if err != nil {
return "", 0, err
}
@@ -154,6 +161,7 @@ func writeXLSXFile(headers []string, rows [][]string) (string, int64, error) {
if err := streamWriter.SetRow(axis, values); err != nil {
return "", 0, err
}
rowIndex++
}
if err := streamWriter.Flush(); err != nil {
@@ -175,6 +183,90 @@ func writeXLSXFile(headers []string, rows [][]string) (string, int64, error) {
return path, stat.Size(), nil
}
func writeXLSXFromCSV(csvPath string) (string, int64, error) {
source, err := os.Open(csvPath)
if err != nil {
return "", 0, err
}
defer source.Close()
file := excelize.NewFile()
sheet := file.GetSheetName(file.GetActiveSheetIndex())
streamWriter, err := file.NewStreamWriter(sheet)
if err != nil {
return "", 0, err
}
reader := csv.NewReader(source)
rowIndex := 1
for {
row, err := reader.Read()
if errors.Is(err, io.EOF) {
break
}
if err != nil {
return "", 0, err
}
values := make([]interface{}, 0, len(row))
for _, item := range row {
values = append(values, item)
}
axis, err := excelize.CoordinatesToCellName(1, rowIndex)
if err != nil {
return "", 0, err
}
if err := streamWriter.SetRow(axis, values); err != nil {
return "", 0, err
}
rowIndex++
}
if err := streamWriter.Flush(); err != nil {
return "", 0, err
}
path := filepath.Join(os.TempDir(), fmt.Sprintf("export-%d.xlsx", time.Now().UnixNano()))
if err := file.SaveAs(path); err != nil {
return "", 0, err
}
if err := file.Close(); err != nil {
return "", 0, err
}
stat, err := os.Stat(path)
if err != nil {
return "", 0, err
}
return path, stat.Size(), nil
}
func alignRowsToHeaders(rows [][]string, headers []string) [][]string {
if len(headers) == 0 {
return rows
}
aligned := make([][]string, 0, len(rows))
for _, row := range rows {
current := make([]string, len(headers))
copy(current, row)
aligned = append(aligned, current)
}
return aligned
}
func hasMisalignedRows(rows [][]string, headers []string) bool {
if len(headers) == 0 {
return false
}
for _, row := range rows {
if len(row) != len(headers) {
return true
}
}
return false
}
func uploadExportFile(ctx context.Context, storageSvc *storage.Service, localPath, fileName string) (string, error) {
if storageSvc == nil || storageSvc.Provider() == nil {
return "", errors.New("对象存储服务未配置")

View File

@@ -119,14 +119,22 @@ func (h *ExportDispatchHandler) HandleExportDispatch(ctx context.Context, task *
return asynq.SkipRetry
}
shards, totalRows, err := h.buildShards(ctx, exportTask, strategy, updater)
params := exporter.ParseExportParams(exportTask)
headers, err := strategy.Headers(ctx, params)
if err != nil {
_ = h.taskStore.MarkFailed(ctx, exportTask.ID, updater, "解析导出表头失败")
h.logger.Error("解析导出表头失败", zap.Uint("task_id", exportTask.ID), zap.Error(err))
return err
}
shards, totalRows, err := h.buildShards(ctx, strategy, params, exportTask.CancelRequested, exportTask.ID, updater)
if err != nil {
_ = h.taskStore.MarkFailed(ctx, exportTask.ID, updater, "构建导出分片失败")
h.logger.Error("构建导出分片失败", zap.Uint("task_id", exportTask.ID), zap.Error(err))
return err
}
if err := h.persistShardsAndPlan(ctx, exportTask.ID, updater, shards, totalRows); err != nil {
if err := h.persistShardsAndPlan(ctx, exportTask.ID, updater, shards, totalRows, headers); err != nil {
_ = h.taskStore.MarkFailed(ctx, exportTask.ID, updater, "保存导出分片失败")
return err
}
@@ -141,58 +149,53 @@ func (h *ExportDispatchHandler) HandleExportDispatch(ctx context.Context, task *
h.logger.Info("导出 dispatch 完成",
zap.Uint("task_id", exportTask.ID),
zap.Uint("total_rows", totalRows),
zap.Int("total_rows", totalRows),
zap.Int("total_shards", len(shards)),
)
return nil
}
func (h *ExportDispatchHandler) buildShards(ctx context.Context, task *model.ExportTask, strategy exporter.SceneStrategy, updater uint) ([]*model.ExportShardTask, uint, error) {
func (h *ExportDispatchHandler) buildShards(ctx context.Context, source exporter.DataSource, params exporter.ExportParams, cancelRequested bool, taskID uint, updater uint) ([]*model.ExportShardTask, int, error) {
shards := make([]*model.ExportShardTask, 0)
totalRows := uint(0)
afterID := uint64(0)
if cancelRequested {
return shards, 0, nil
}
totalRows, err := source.Count(ctx, params)
if err != nil {
return nil, 0, err
}
if totalRows <= 0 {
return shards, 0, nil
}
shardNo := 1
for {
if task.CancelRequested {
break
for offset := 0; offset < totalRows; offset += constants.ExportDefaultShardSize {
limit := constants.ExportDefaultShardSize
if remaining := totalRows - offset; remaining < limit {
limit = remaining
}
boundary, err := strategy.NextShardBoundary(ctx, task, afterID, constants.ExportDefaultShardSize)
if err != nil {
return nil, 0, err
}
if boundary == nil {
break
}
rowCount := boundary.RowCount
if rowCount < 0 {
rowCount = 0
}
totalRows += uint(rowCount)
shard := &model.ExportShardTask{
TaskID: task.ID,
TaskID: taskID,
ShardNo: shardNo,
Status: constants.ExportShardStatusPending,
CursorStart: boundary.StartID,
CursorEnd: boundary.EndID,
RowCount: rowCount,
ShardOffset: offset,
ShardLimit: limit,
RowCount: limit,
}
shard.Creator = updater
shard.Updater = updater
shards = append(shards, shard)
afterID = boundary.EndID
shardNo++
}
return shards, totalRows, nil
}
func (h *ExportDispatchHandler) persistShardsAndPlan(ctx context.Context, taskID uint, updater uint, shards []*model.ExportShardTask, totalRows uint) error {
func (h *ExportDispatchHandler) persistShardsAndPlan(ctx context.Context, taskID uint, updater uint, shards []*model.ExportShardTask, totalRows int, headers []string) error {
return h.db.WithContext(ctx).Transaction(func(tx *gorm.DB) error {
txShardStore := postgres.NewExportShardTaskStore(tx, h.redis)
txTaskStore := postgres.NewExportTaskStore(tx, h.redis)
@@ -200,7 +203,10 @@ func (h *ExportDispatchHandler) persistShardsAndPlan(ctx context.Context, taskID
if err := txShardStore.CreateBatch(ctx, shards); err != nil {
return err
}
if err := txTaskStore.UpdateDispatchPlan(ctx, taskID, totalRows, uint(len(shards)), updater); err != nil {
if err := txTaskStore.UpdateResolvedHeaders(ctx, taskID, headers, updater); err != nil {
return err
}
if err := txTaskStore.UpdateDispatchPlan(ctx, taskID, totalRows, len(shards), updater); err != nil {
return err
}
return nil

View File

@@ -2,7 +2,10 @@ package task
import (
"context"
"encoding/csv"
"errors"
"fmt"
"io"
"os"
"github.com/hibiken/asynq"
@@ -10,6 +13,7 @@ import (
"go.uber.org/zap"
"github.com/break/junhong_cmp_fiber/internal/exporter"
"github.com/break/junhong_cmp_fiber/internal/model"
"github.com/break/junhong_cmp_fiber/internal/store/postgres"
"github.com/break/junhong_cmp_fiber/pkg/constants"
"github.com/break/junhong_cmp_fiber/pkg/storage"
@@ -102,8 +106,7 @@ func (h *ExportFinalizeHandler) HandleExportFinalize(ctx context.Context, task *
return nil
}
strategy, ok := h.sceneRegistry.Get(exportTask.Scene)
if !ok {
if !h.sceneRegistry.IsSupported(exportTask.Scene) {
if err := h.taskStore.MarkFailed(ctx, exportTask.ID, updater, "导出场景未注册"); err != nil {
return err
}
@@ -115,23 +118,22 @@ func (h *ExportFinalizeHandler) HandleExportFinalize(ctx context.Context, task *
return err
}
rows := make([][]string, 0)
for _, shard := range shards {
if shard.Status != constants.ExportShardStatusSuccess {
continue
}
currentRows, err := strategy.QueryRows(ctx, exportTask, shard.CursorStart, shard.CursorEnd)
if err != nil {
return h.handleFinalizeError(ctx, exportTask.ID, updater, "读取分片数据失败", err)
}
rows = append(rows, currentRows...)
headers := exporter.ResolvedHeadersFromTask(exportTask)
if len(headers) == 0 {
return h.handleFinalizeError(ctx, exportTask.ID, updater, "读取导出表头失败", errors.New("resolved_headers 为空"))
}
localPath, fileSize, err := writeExportFile(exportTask.Format, strategy.Headers(), rows)
mergedCSVPath, cleanupCSV, err := h.mergeShardCSVFiles(ctx, headers, shards)
if err != nil {
return h.handleFinalizeError(ctx, exportTask.ID, updater, "生成汇总文件失败", err)
}
defer func() { _ = os.Remove(localPath) }()
defer cleanupCSV()
localPath, fileSize, cleanupFinal, err := prepareFinalExportFile(exportTask.Format, mergedCSVPath)
if err != nil {
return h.handleFinalizeError(ctx, exportTask.ID, updater, "生成汇总文件失败", err)
}
defer cleanupFinal()
fileName := fmt.Sprintf("%s.%s", exportTask.TaskNo, exportTask.Format)
fileKey, err := uploadExportFile(ctx, h.storageSvc, localPath, fileName)
@@ -146,6 +148,104 @@ func (h *ExportFinalizeHandler) HandleExportFinalize(ctx context.Context, task *
return nil
}
func (h *ExportFinalizeHandler) mergeShardCSVFiles(ctx context.Context, headers []string, shards []*model.ExportShardTask) (string, func(), error) {
if h.storageSvc == nil || h.storageSvc.Provider() == nil {
return "", nil, errors.New("对象存储服务未配置")
}
file, err := os.CreateTemp("", "export-merged-*.csv")
if err != nil {
return "", nil, err
}
cleanup := func() {
_ = os.Remove(file.Name())
}
writer := csv.NewWriter(file)
if err := writer.Write(headers); err != nil {
file.Close()
cleanup()
return "", nil, err
}
for _, shard := range shards {
if shard.Status != constants.ExportShardStatusSuccess {
continue
}
if shard.FileKey == "" {
file.Close()
cleanup()
return "", nil, fmt.Errorf("分片 %d 缺少文件Key", shard.ShardNo)
}
if err := h.appendShardFile(ctx, writer, shard.FileKey); err != nil {
file.Close()
cleanup()
return "", nil, err
}
}
writer.Flush()
if err := writer.Error(); err != nil {
file.Close()
cleanup()
return "", nil, err
}
if err := file.Sync(); err != nil {
file.Close()
cleanup()
return "", nil, err
}
if err := file.Close(); err != nil {
cleanup()
return "", nil, err
}
return file.Name(), cleanup, nil
}
func (h *ExportFinalizeHandler) appendShardFile(ctx context.Context, writer *csv.Writer, key string) error {
reader, err := h.storageSvc.Provider().Download(ctx, key)
if err != nil {
return err
}
defer reader.Close()
csvReader := csv.NewReader(reader)
for {
row, err := csvReader.Read()
if errors.Is(err, io.EOF) {
break
}
if err != nil {
return err
}
if err := writer.Write(row); err != nil {
return err
}
}
return nil
}
func prepareFinalExportFile(format, mergedCSVPath string) (string, int64, func(), error) {
switch format {
case constants.ExportTaskFormatCSV:
stat, err := os.Stat(mergedCSVPath)
if err != nil {
return "", 0, nil, err
}
return mergedCSVPath, stat.Size(), func() {}, nil
case constants.ExportTaskFormatXLSX:
xlsxPath, fileSize, err := writeXLSXFromCSV(mergedCSVPath)
if err != nil {
return "", 0, nil, err
}
return xlsxPath, fileSize, func() { _ = os.Remove(xlsxPath) }, nil
default:
return "", 0, nil, fmt.Errorf("不支持的导出格式: %s", format)
}
}
func (h *ExportFinalizeHandler) handleFinalizeError(ctx context.Context, taskID uint, updater uint, prefix string, err error) error {
if isFinalRetry(ctx) {
markErr := h.taskStore.MarkFailed(ctx, taskID, updater, prefix+": "+err.Error())

View File

@@ -107,24 +107,38 @@ func (h *ExportShardHandler) HandleExportShard(ctx context.Context, task *asynq.
}
}
strategy, ok := h.sceneRegistry.Get(exportTask.Scene)
source, ok := h.sceneRegistry.Get(exportTask.Scene)
if !ok {
_ = h.markShardFinalFailed(ctx, exportTask.ID, shardTask.ID, updater, "导出场景未注册")
return asynq.SkipRetry
}
rows, err := strategy.QueryRows(ctx, exportTask, shardTask.CursorStart, shardTask.CursorEnd)
headers := exporter.ResolvedHeadersFromTask(exportTask)
if len(headers) == 0 {
return h.handleShardError(ctx, exportTask.ID, shardTask.ID, updater, "读取导出表头失败", fmt.Errorf("resolved_headers 为空"))
}
params := exporter.ParseExportParams(exportTask)
rows, err := source.Fetch(ctx, params, shardTask.ShardOffset, shardTask.ShardLimit)
if err != nil {
return h.handleShardError(ctx, exportTask.ID, shardTask.ID, updater, "查询分片数据失败", err)
}
if hasMisalignedRows(rows, headers) {
h.logger.Warn("导出分片数据列数与表头不一致,已按固定表头对齐",
zap.Uint("task_id", exportTask.ID),
zap.Uint("shard_id", shardTask.ID),
zap.Int("header_count", len(headers)),
)
}
rows = alignRowsToHeaders(rows, headers)
localPath, fileSize, err := writeExportFile(exportTask.Format, strategy.Headers(), rows)
localPath, fileSize, err := writeExportFile(constants.ExportTaskFormatCSV, headers, rows, true)
if err != nil {
return h.handleShardError(ctx, exportTask.ID, shardTask.ID, updater, "生成分片文件失败", err)
}
defer func() { _ = os.Remove(localPath) }()
fileName := fmt.Sprintf("%s-shard-%04d.%s", exportTask.TaskNo, shardTask.ShardNo, exportTask.Format)
fileName := fmt.Sprintf("%s-shard-%04d.csv.shard", exportTask.TaskNo, shardTask.ShardNo)
fileKey, err := uploadExportFile(ctx, h.storageSvc, localPath, fileName)
if err != nil {
return h.handleShardError(ctx, exportTask.ID, shardTask.ID, updater, "上传分片文件失败", err)