feat: add 422 response for unprocessable entity in API endpoints

feat: add 422 response for unprocessable entity in API endpoints

Author: a-klos

admin-api-lib/openapi.yaml

@@ -64,6 +64,8 @@ paths:
           description: ok
         "400":
           description: Bad request
+        "422":
+          description: If no text has been extracted from the file.
         "500":
           description: Internal server error
       tags:

admin-api-lib/src/admin_api_lib/apis/admin_api.py

@@ -11,6 +11,7 @@
 from admin_api_lib.apis.admin_api_base import BaseAdminApi
 from admin_api_lib.models.document_status import DocumentStatus
 
+
 router = APIRouter()
 
 ns_pkg = admin_api_lib.impl
@@ -129,6 +130,7 @@ async def load_confluence_post() -> None:
     responses={
         200: {"description": "ok"},
         400: {"description": "Bad request"},
+        422: {"description": "If no text has been extracted from the file."},
         500: {"description": "Internal server error"},
     },
     tags=["admin"],

admin-api-lib/src/admin_api_lib/impl/api_endpoints/default_document_uploader.py

@@ -161,6 +161,10 @@ async def _aparse_document(
         self._key_value_store.upsert(filename, Status.PROCESSING)
 
         information_pieces = self._document_extractor.extract_from_file_post(ExtractionRequest(path_on_s3=filename))
+        if not information_pieces:
+            self._key_value_store.upsert(filename, Status.ERROR)
+            logger.error("No information pieces found in the document: %s", filename)
+            raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail="No information pieces found")
         documents = [self._information_mapper.extractor_information_piece2document(x) for x in information_pieces]
         host_base_url = str(request.base_url)
         document_url = f"{host_base_url.rstrip('/')}/document_reference/{urllib.parse.quote_plus(filename)}"