Merge pull request #728 from i-dot-ai/feature/put-endpoint-for-reinge…

…stion

New core-api PUT endpoint for file reingestion

core_api/src/routes/file.py

@@ -200,6 +200,46 @@ def delete_file(file_uuid: UUID, user_uuid: Annotated[UUID, Depends(get_user_uui
     return file
 
 
+@file_app.put(
+    "/{file_uuid}",
+    response_model=File,
+    tags=["file"],
+    responses={404: {"model": APIError404, "description": "The file was not found"}},
+)
+async def reingest_file(file_uuid: UUID, user_uuid: Annotated[UUID, Depends(get_user_uuid)]) -> File:
+    """Deletes exisiting file chunks and regenerates embeddings
+
+    Args:
+        file_uuid (UUID): The UUID of the file to delete
+        user_uuid (UUID): The UUID of the user
+
+    Returns:
+        File: The file that was deleted
+
+    Raises:
+        404: If the file isn't found, or the creator and requester don't match
+    """
+    try:
+        file = storage_handler.read_item(file_uuid, model_type="File")
+    except NotFoundError:
+        return file_not_found_response(file_uuid=file_uuid)
+
+    if file.creator_user_uuid != user_uuid:
+        return file_not_found_response(file_uuid=file_uuid)
+
+    log.info("reingesting %s", file.uuid)
+
+    # Remove old chunks
+    chunks = storage_handler.get_file_chunks(file.uuid, user_uuid)
+    storage_handler.delete_items(chunks)
+
+    # Add new chunks
+    log.info("publishing %s", file.uuid)
+    await file_publisher.publish(file)
+
+    return file
+
+
 @file_app.get(
     "/{file_uuid}/chunks",
     tags=["file"],

core_api/tests/routes/test_file.py

@@ -108,6 +108,24 @@ def test_delete_missing_file(app_client, headers):
     assert response.status_code == HTTPStatus.NOT_FOUND
 
 
+def test_reingest_file(app_client, chunked_file, elasticsearch_storage_handler, headers):
+    """
+    Given a previously chunked file
+    When I PUT it to /file/uuid/
+    I Expect the old chunks to be removed
+    """
+    previous_chunks = elasticsearch_storage_handler.get_file_chunks(chunked_file.uuid, chunked_file.creator_user_uuid)
+
+    response = app_client.put(f"/file/{chunked_file.uuid}", headers=headers)
+    assert response.status_code == HTTPStatus.OK
+
+    elasticsearch_storage_handler.refresh()
+    assert (
+        elasticsearch_storage_handler.get_file_chunks(chunked_file.uuid, chunked_file.creator_user_uuid)
+        != previous_chunks
+    )
+
+
 def test_get_file_chunks(app_client, chunked_file, headers):
     """
     Given a previously chunked file

django_app/redbox_app/redbox_core/admin.py

@@ -41,28 +41,17 @@ class FileAdmin(admin.ModelAdmin):
     def reupload(self, request, queryset):  # noqa:ARG002
         for file in queryset:
             try:
-                logger.info("Deleting existing file from core-api: %s", file)
-                core_api.delete_file(file.core_file_uuid, file.user)
+                logger.info("Re-uploading file to core-api: %s", file)
+                core_api.reingest_file(file.core_file_uuid, file.user)
             except RequestException as e:
-                logger.exception("Error deleting File model object %s.", file, exc_info=e)
-
+                logger.exception("Error re-uploading File model object %s.", file, exc_info=e)
+                file.status = models.StatusEnum.errored
+                file.save()
             else:
-                file.status = models.StatusEnum.deleted
+                file.status = models.StatusEnum.uploaded
                 file.save()
 
-                try:
-                    logger.info("Re-uploading file to core-api: %s", file)
-                    upload_file_response = core_api.upload_file(file.unique_name, file.user)
-                except RequestException as e:
-                    logger.exception("Error re-uploading File model object %s.", file, exc_info=e)
-                    file.status = models.StatusEnum.errored
-                    file.save()
-                else:
-                    file.core_file_uuid = upload_file_response.uuid
-                    file.status = models.StatusEnum.uploaded
-                    file.save()
-
-        logger.info("Successfully reuploaded file %s.", file)
+                logger.info("Successfully reuploaded file %s.", file)
 
     list_display = ["original_file_name", "user", "status", "created_at", "last_referenced"]
     list_filter = ["user", "status"]

django_app/redbox_app/redbox_core/client.py

@@ -114,3 +114,9 @@ def delete_file(self, file_id: UUID, user: User) -> FileOperation:
         response = requests.delete(url, headers={"Authorization": user.get_bearer_token()}, timeout=60)
         response.raise_for_status()
         return FileOperation.schema().loads(response.content)
+
+    def reingest_file(self, file_id: UUID, user: User) -> FileOperation:
+        url = self.url / "file" / str(file_id)
+        response = requests.put(url, headers={"Authorization": user.get_bearer_token()}, timeout=60)
+        response.raise_for_status()
+        return FileOperation.schema().loads(response.content)