Files
query-orchestration/internal/document/text/extract.go
T
Michael McGuinness 81e7223560 Merged in feature/textextraction (pull request #110)
Text Extraction

* bases

* go

* splitting

* structure

* movetoasync

* movetoasync

* settinguptrigger

* reorder

* storevent

* standardisepollingvalidation

* unittests

* fixlint

* fixlint

* awscfg

* generatesample

* followthrough

* tests

* clena

* store

* externalidcleanup

* clientid

* local

* baseunittests

* putobjecttests

* tests
2025-04-02 18:50:03 +00:00

52 lines
1.1 KiB
Go

package documenttext
import (
"context"
"log/slog"
querysyncrunner "queryorchestration/api/querySyncRunner"
"queryorchestration/internal/serviceconfig/queue"
"github.com/google/uuid"
)
func (s *Service) Extract(ctx context.Context, documentId uuid.UUID) error {
slog.Debug("extracting document text", "id", documentId.String())
isextracted, err := s.cfg.GetDBQueries().IsDocumentTextExtracted(ctx, documentId)
if err != nil {
return err
} else if isextracted {
return s.informExtraction(ctx, documentId)
}
entry, err := s.cfg.GetDBQueries().GetCleanEntryByDocId(ctx, documentId)
if err != nil {
return err
} else if entry.Fail.Valid {
slog.Error("document has no clean document", "id", documentId)
return nil
}
err = s.triggerExtract(ctx, entry)
if err != nil {
return err
}
return nil
}
func (s *Service) informExtraction(ctx context.Context, id uuid.UUID) error {
err := s.cfg.SendToQueue(ctx, &queue.SendParams{
QueueURL: s.cfg.GetQuerySyncURL(),
Body: querysyncrunner.Body{
DocumentID: id,
},
})
if err != nil {
return err
}
return nil
}