Support S3 custom model without explicit credentials (#948)

src/marqo/s2_inference/model_downloading/from_s3.py

@@ -19,13 +19,7 @@ def get_presigned_s3_url(location: S3Location, auth: Optional[S3Auth] = None):
 
     TODO: add link to proper usage in error messages
     """
-    if auth is None:
-        raise ModelDownloadError(
-            "Error retrieving private model. s3 authorisation information is required to "
-            "download a model from an s3 bucket. "
-            "If the model is publicly accessible, please use the model's publicly accessible URL."
-        )
-    s3_client = boto3.client('s3', **auth.dict())
+    s3_client = boto3.client('s3', **(auth.dict() if auth is not None else {}))
     try:
         return s3_client.generate_presigned_url('get_object', Params=location.dict(exclude_unset=True))
     except NoCredentialsError:

src/marqo/s2_inference/processing/custom_clip_utils.py

@@ -120,7 +120,9 @@ def download_pretrained_from_s3(
             raise ModelDownloadError(
                 "Received 403 error when trying to retrieve model from s3 storage. "
                 "Please check the request's s3 credentials and try again. "
-            )
+            ) from e
+        else:
+            raise e
 
 def download_pretrained_from_url(
         url: str,

src/marqo/s2_inference/s2_inference.py

@@ -225,7 +225,7 @@ def _get_max_vectorise_batch_size() -> int:
     except (ValueError, TypeError) as e:
         value_error_msg = f"`{validation_error_msg} Current value: `{max_batch_size_value}`. Reason: {e}"
         logger.error(value_error_msg)
-        raise ConfigurationError(value_error_msg)
+        raise ConfigurationError(value_error_msg) from e
     if batch_size < 1:
         batch_size_too_small_msg = f"`{validation_error_msg} Current value: `{max_batch_size_value}`."
         logger.error(batch_size_too_small_msg)
@@ -298,17 +298,17 @@ def _update_available_models(model_cache_key: str, model_name: str, validated_mo
                     f"Unable to load model={model_name} on device={device} with normalization={normalize_embeddings}. "
                     f"If you are trying to load a custom model, "
                     f"please check that model_properties={validated_model_properties} is correct "
-                    f"and Marqo has access to the weights file.")
+                    f"and Marqo has access to the weights file.") from e
 
     else:
         most_recently_used_time = datetime.datetime.now()
         logger.debug(f'renewed {model_name} on device {device} with new most recently time={most_recently_used_time}.')
         try:
             _available_models[model_cache_key][AvailableModelsKey.most_recently_used_time] = most_recently_used_time
-        except KeyError:
+        except KeyError as e:
             raise ModelNotInCacheError(f"Marqo cannot renew model {model_name} on device {device} with normalization={normalize_embeddings}. "
                                        f"Maybe another thread is updating the model cache at the same time."
-                                       f"Please wait for 10 seconds and send the request again.\n")
+                                       f"Please wait for 10 seconds and send the request again.\n") from e
 
 
 def validate_model_properties(model_name: str, model_properties: dict) -> dict:

src/marqo/tensor_search/tensor_search.py

@@ -1664,15 +1664,15 @@ def vectorise_jobs(jobs: List[VectorisedJobs]) -> Dict[JHash, Dict[str, List[flo
         except (s2_inference_errors.UnknownModelError,
                 s2_inference_errors.InvalidModelPropertiesError,
                 s2_inference_errors.ModelLoadError,
-                s2_inference.ModelDownloadError) as model_error:
+                s2_inference.ModelDownloadError) as e:
             raise api_exceptions.BadRequestError(
-                message=f'Problem vectorising query. Reason: {str(model_error)}',
+                message=f'Problem vectorising query. Reason: {str(e)}',
                 link=marqo_docs.list_of_models()
-            )
+            ) from e
 
         except s2_inference_errors.S2InferenceError as e:
             # TODO: differentiate image processing errors from other types of vectorise errors
-            raise api_exceptions.InvalidArgError(message=f'Error vectorising content: {v.content}. Message: {e}')
+            raise api_exceptions.InvalidArgError(message=f'Error vectorising content: {v.content}. Message: {e}') from e
     return result
 
 

src/marqo/version.py

@@ -1,4 +1,4 @@
-__version__ = "2.11.2"
+__version__ = "2.11.3"
 
 def get_version() -> str:
     return f"{__version__}"