[owl] Make column dtype validation more robust by using field_validator (#822)

Backend - owl (API server)

- Make column dtype validation more robust by using `field_validator`
- Improve Docling time out error and log message

Author: jiahuei

services/api/src/owl/docparse.py

@@ -510,11 +510,12 @@ async def retrieve_document_content(
                 time_slept += sleep_for
             else:  # Executed if the while loop completes without a 'break'
                 logger.error(
-                    f"{self.request_id} - Polling timed out for Docling-Serve task {task_id} after {time_slept} seconds."
-                )
-                raise UnexpectedError(
-                    f"Polling timed out for Docling-Serve task {task_id} after {time_slept} seconds."
+                    (
+                        f'Docling-Serve task "{task_id}" for document "{file_name}" '
+                        f"timed out after {time_slept} seconds. ({self.request_id})"
+                    )
                 )
+                raise UnexpectedError(f'Your document "{file_name}" took too long to parse.')
 
             # Step 3: Fetch result
             result_url = f"{self.docling_serve_url}/v1alpha/result/{task_id}"

services/api/src/owl/types/__init__.py

@@ -11,6 +11,7 @@
     BaseModel,
     BeforeValidator,
     Field,
+    field_validator,
     model_validator,
 )
 
@@ -422,12 +423,12 @@ class ColumnSchemaCreate(t.ColumnSchemaCreate):
         description=f"Column data type, one of {list(map(str, ColumnDtypeCreate))}.",
     )
 
-    @model_validator(mode="before")
+    @field_validator("dtype", mode="before")
     @classmethod
-    def map_file_dtype_to_image(cls, data: dict[str, Any]) -> dict[str, Any]:
-        if data.get("dtype", "") == "file":
-            data["dtype"] = ColumnDtype.IMAGE
-        return data
+    def map_file_dtype_to_image(cls, v: Any) -> Any:
+        if v == "file":
+            v = ColumnDtype.IMAGE
+        return v
 
 
 class TableSchemaCreate(t.TableSchemaCreate):