[BugFix]: properly catch templating error when preprocess input (#13976)

Signed-off-by: Guillaume Calmettes <gcalmettes@scaleway.com>

vllm/entrypoints/openai/serving_chat.py

@@ -7,6 +7,7 @@ from collections.abc import AsyncGenerator, AsyncIterator
 from collections.abc import Sequence as GenericSequence
 from typing import Callable, Final, Optional, Union
 
+import jinja2
 from fastapi import Request
 
 from vllm.config import ModelConfig
@@ -199,6 +200,15 @@ class OpenAIServingChat(OpenAIServing):
         except ValueError as e:
             logger.exception("Error in preprocessing prompt inputs")
             return self.create_error_response(str(e))
+        except TypeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
+        except RuntimeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
+        except jinja2.TemplateError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
 
         request_id = "chatcmpl-" \
                      f"{self._base_request_id(raw_request, request.request_id)}"

vllm/entrypoints/openai/serving_completion.py

@@ -6,6 +6,7 @@ from collections.abc import AsyncGenerator, AsyncIterator
 from collections.abc import Sequence as GenericSequence
 from typing import Optional, Union, cast
 
+import jinja2
 from fastapi import Request
 
 from vllm.config import ModelConfig
@@ -114,6 +115,15 @@ class OpenAIServingCompletion(OpenAIServing):
         except ValueError as e:
             logger.exception("Error in preprocessing prompt inputs")
             return self.create_error_response(str(e))
+        except TypeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
+        except RuntimeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
+        except jinja2.TemplateError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
 
         # Schedule the request and get the result generator.
         generators: list[AsyncGenerator[RequestOutput, None]] = []

vllm/entrypoints/openai/serving_embedding.py

@@ -142,6 +142,9 @@ class OpenAIServingEmbedding(OpenAIServing):
         except ValueError as e:
             logger.exception("Error in preprocessing prompt inputs")
             return self.create_error_response(str(e))
+        except TypeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
 
         # Schedule the request and get the result generator.
         generators: list[AsyncGenerator[PoolingRequestOutput, None]] = []

vllm/entrypoints/openai/serving_pooling.py

@@ -6,6 +6,7 @@ import time
 from collections.abc import AsyncGenerator
 from typing import Final, Literal, Optional, Union, cast
 
+import jinja2
 import numpy as np
 from fastapi import Request
 from typing_extensions import assert_never
@@ -138,6 +139,12 @@ class OpenAIServingPooling(OpenAIServing):
         except ValueError as e:
             logger.exception("Error in preprocessing prompt inputs")
             return self.create_error_response(str(e))
+        except TypeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
+        except jinja2.TemplateError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
 
         # Schedule the request and get the result generator.
         generators: list[AsyncGenerator[PoolingRequestOutput, None]] = []

vllm/entrypoints/openai/serving_tokenization.py

@@ -2,6 +2,7 @@
 
 from typing import Final, Optional, Union
 
+import jinja2
 from fastapi import Request
 
 from vllm.config import ModelConfig
@@ -91,6 +92,12 @@ class OpenAIServingTokenization(OpenAIServing):
         except ValueError as e:
             logger.exception("Error in preprocessing prompt inputs")
             return self.create_error_response(str(e))
+        except TypeError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
+        except jinja2.TemplateError as e:
+            logger.exception("Error in preprocessing prompt inputs")
+            return self.create_error_response(str(e))
 
         input_ids: list[int] = []
         for i, engine_prompt in enumerate(engine_prompts):