List and download tender documents - worker auto translation to english job

pkg/ai_summarizer/entities.go

@@ -30,6 +30,15 @@ type DocumentSummary struct {
 	Error        string `json:"error"` // empty string when no error
 }
 
+// StoredDocument represents a document object stored for a tender in MinIO.
+type StoredDocument struct {
+	ObjectName   string `json:"-"`
+	DocumentName string `json:"document_name"`
+	Size         int64  `json:"size"`
+	LastModified int64  `json:"last_modified"`
+	DocumentType string `json:"document_type,omitempty"`
+}
+
 // TranslateRequest represents the request payload sent to the AI service
 // POST /ai/translate endpoint.
 type TranslateRequest struct {

pkg/ai_summarizer/storage.go

@@ -7,6 +7,7 @@ import (
 	"fmt"
 	"io"
 	"net/http"
+	"path/filepath"
 	"strings"
 
 	"github.com/minio/minio-go/v7"
@@ -153,6 +154,89 @@ func (s *StorageClient) IsSummaryReady(ctx context.Context, noticeID string) (bo
 	return tenderJSON.SummarizeStatus.Done, nil
 }
 
+// ListDocuments lists document files stored under <noticeID>/.
+func (s *StorageClient) ListDocuments(ctx context.Context, noticeID string) ([]StoredDocument, error) {
+	noticeID = strings.TrimSpace(noticeID)
+	if noticeID == "" {
+		return nil, fmt.Errorf("ai_summarizer: noticeID is required")
+	}
+
+	prefix := fmt.Sprintf("%s/", noticeID)
+	objectCh := s.client.ListObjects(ctx, s.config.MinioBucket, minio.ListObjectsOptions{
+		Prefix:    prefix,
+		Recursive: true,
+	})
+
+	documents := make([]StoredDocument, 0)
+	for object := range objectCh {
+		if object.Err != nil {
+			return nil, fmt.Errorf("ai_summarizer: failed to list documents for notice %s: %w", noticeID, object.Err)
+		}
+		if !isTenderDocumentObject(noticeID, object.Key) {
+			continue
+		}
+
+		documentName := filepath.Base(object.Key)
+		documents = append(documents, StoredDocument{
+			ObjectName:   object.Key,
+			DocumentName: documentName,
+			Size:         object.Size,
+			LastModified: object.LastModified.Unix(),
+			DocumentType: documentType(documentName),
+		})
+	}
+
+	s.logger.Info("Listed tender documents from storage", map[string]interface{}{
+		"notice_id":       noticeID,
+		"documents_count": len(documents),
+	})
+
+	return documents, nil
+}
+
+func isTenderDocumentObject(noticeID, objectName string) bool {
+	if objectName == "" || strings.HasSuffix(objectName, "/") {
+		return false
+	}
+
+	relativePath := strings.TrimPrefix(objectName, fmt.Sprintf("%s/", noticeID))
+	if relativePath == "" {
+		return false
+	}
+
+	filename := strings.ToLower(filepath.Base(relativePath))
+	switch filename {
+	case "tender.json", "ai_cache.db":
+		return false
+	}
+
+	switch strings.ToLower(filepath.Ext(filename)) {
+	case ".json", ".db", ".sqlite", ".sqlite3":
+		return false
+	default:
+		return true
+	}
+}
+
+// DownloadDocument opens a document object from the configured MinIO bucket.
+func (s *StorageClient) DownloadDocument(ctx context.Context, objectName string) (io.ReadCloser, error) {
+	objectName = strings.TrimSpace(objectName)
+	if objectName == "" {
+		return nil, fmt.Errorf("ai_summarizer: objectName is required")
+	}
+
+	object, err := s.client.GetObject(ctx, s.config.MinioBucket, objectName, minio.GetObjectOptions{})
+	if err != nil {
+		var minioErr minio.ErrorResponse
+		if errors.As(err, &minioErr) && minioErr.StatusCode == http.StatusNotFound {
+			return nil, ErrObjectNotFound
+		}
+		return nil, fmt.Errorf("ai_summarizer: failed to get document object: %w", err)
+	}
+
+	return object, nil
+}
+
 // GetAllSummaries lists all tender.json files in the MinIO bucket and returns
 // the overall_summary for each tender where summarize_status.done is true.
 // This is useful for batch-retrieving all summaries that the AI pipeline has
@@ -233,3 +317,11 @@ func countDone(items []TenderSummaryItem) int {
 	}
 	return count
 }
+
+func documentType(filename string) string {
+	extension := strings.TrimPrefix(strings.ToLower(filepath.Ext(filename)), ".")
+	if extension == "" {
+		return "unknown"
+	}
+	return extension
+}