Files
rag-local/kb/service/parse_task_service.go
T
2026-08-05 10:28:44 +08:00

131 lines
4.1 KiB
Go

package service
import (
"context"
"path/filepath"
"time"
"rag-local/common"
"rag-local/kb/consts"
"rag-local/kb/dao"
"rag-local/kb/model/entity"
eembedding "github.com/cloudwego/eino/components/embedding"
"github.com/gogf/gf/v2/errors/gerror"
"github.com/gogf/gf/v2/frame/g"
)
var ParseTaskService = &parseTaskService{}
type parseTaskService struct{}
// StartParsePoller 启动任务轮询:单 goroutine 串行消费待处理任务(与 video-factory StartVideoPoller 同模式)
func (s *parseTaskService) StartParsePoller(ctx context.Context) {
go func() {
g.Log().Info(ctx, "parse task poller started")
for {
select {
case <-ctx.Done():
return
case <-time.After(consts.ParsePollIntervalSeconds * time.Second):
s.processOne(ctx)
}
}
}()
}
// processOne 处理一个待处理任务:解析 → 分块 → 落库(向量化在 M3 接入)
func (s *parseTaskService) processOne(ctx context.Context) {
task, err := dao.ParseTask.NextPending(ctx)
if err != nil {
g.Log().Errorf(ctx, "next parse task failed: %v", err)
return
}
if task == nil {
return
}
if err := dao.ParseTask.UpdateStatus(ctx, task.Id, consts.TaskStatusRunning, ""); err != nil {
g.Log().Errorf(ctx, "mark task running failed: %v", err)
return
}
doc, err := dao.Document.GetOne(ctx, task.DocumentId)
if err != nil {
s.fail(ctx, task, "读取文档失败: "+err.Error())
return
}
if doc == nil {
s.fail(ctx, task, "文档不存在")
return
}
if err := dao.Document.UpdateFields(ctx, doc.Id, g.Map{"status": consts.DocumentStatusParsing}); err != nil {
s.fail(ctx, task, "更新文档状态失败: "+err.Error())
return
}
text, err := common.ParseFile(filepath.Join("workspace", doc.FilePath))
if err != nil {
s.fail(ctx, task, "解析文件失败: "+err.Error())
return
}
chunks := ChunkService.SplitText(text)
// 数据集绑定 embedding 配置时构建向量模型,无配置降级为仅全文索引
var embedder eembedding.Embedder
if cfgId, err := dao.Dataset.GetEmbeddingCfgId(ctx, task.DatasetId); err == nil && cfgId > 0 {
if em, err := BuildEmbedder(ctx, cfgId); err == nil {
embedder = em
if dim := em.Dim(); dim != g.Cfg().MustGet(ctx, "vector.dim", consts.DefaultEmbeddingDim).Int() {
g.Log().Warningf(ctx, "embedding 维度 %d 与 vec0 表维度不一致,请确认 vector.dim 配置", dim)
}
} else {
g.Log().Warningf(ctx, "build embedder failed, fallback to fts only: %v", err)
}
}
if err := ChunkService.InsertAll(ctx, task.DatasetId, doc.Id, chunks, embedder); err != nil {
s.fail(ctx, task, "写入分块失败: "+err.Error())
return
}
// 第 3.5 步:知识图谱抽取(失败不阻断流水线)
if err := KgEntityService.ExtractDocument(ctx, task.DatasetId, doc.Id); err != nil {
g.Log().Warningf(ctx, "kg extract skipped for doc %d: %v", doc.Id, err)
}
if err := dao.ParseTask.UpdateStatus(ctx, task.Id, consts.TaskStatusDone, ""); err != nil {
g.Log().Errorf(ctx, "mark task done failed: %v", err)
}
}
func (s *parseTaskService) fail(ctx context.Context, task *entity.ParseTask, msg string) {
_ = dao.ParseTask.UpdateStatus(ctx, task.Id, consts.TaskStatusFailed, msg)
_ = dao.Document.UpdateFields(ctx, task.DocumentId, g.Map{
"status": consts.DocumentStatusFailed,
"error_msg": msg,
})
g.Log().Errorf(ctx, "parse task %d failed: %s", task.Id, msg)
}
// List 任务列表
func (s *parseTaskService) List(ctx context.Context, page, pageSize int) ([]*entity.ParseTask, int, error) {
return dao.ParseTask.List(ctx, page, pageSize)
}
// Retry 失败任务重置为待处理(文档状态同步重置)
func (s *parseTaskService) Retry(ctx context.Context, id int64) error {
task, err := dao.ParseTask.GetOne(ctx, id)
if err != nil {
return err
}
if task == nil {
return gerror.New("任务不存在")
}
if task.Status != consts.TaskStatusFailed {
return gerror.New("仅失败任务可重试")
}
if err := dao.ParseTask.UpdateStatus(ctx, id, consts.TaskStatusPending, ""); err != nil {
return err
}
return dao.Document.UpdateFields(ctx, task.DocumentId, g.Map{
"status": consts.DocumentStatusPending,
"error_msg": "",
})
}