fix(inference_store): on duplicate chat completion IDs, replace

This commit is contained in:
Ashwin Bharambe 2025-09-10 12:25:04 -07:00
parent c04f1c1e8c
commit 7eedb88edc
2 changed files with 52 additions and 9 deletions

View file

@ -5,6 +5,7 @@
# the root directory of this source tree.
import asyncio
from typing import Any
from sqlalchemy.exc import IntegrityError
from llama_stack.apis.inference import (
ListOpenAIChatCompletionResponse,
@ -129,16 +130,44 @@ class InferenceStore:
raise ValueError("Inference store is not initialized")
data = chat_completion.model_dump()
record_data = {
"id": data["id"],
"created": data["created"],
"model": data["model"],
"choices": data["choices"],
"input_messages": [message.model_dump() for message in input_messages],
}
await self.sql_store.insert(
table="chat_completions",
data={
"id": data["id"],
"created": data["created"],
"model": data["model"],
"choices": data["choices"],
"input_messages": [message.model_dump() for message in input_messages],
},
try:
await self.sql_store.insert(
table="chat_completions",
data=record_data,
)
except IntegrityError as e:
# Check if it's a unique constraint violation
error_message = str(e.orig) if e.orig else str(e)
if self._is_unique_constraint_error(error_message):
# Update the existing record instead
await self.sql_store.update(
table="chat_completions",
data=record_data,
where={"id": data["id"]}
)
else:
# Re-raise if it's not a unique constraint error
raise
def _is_unique_constraint_error(self, error_message: str) -> bool:
"""Check if the error is specifically a unique constraint violation."""
error_lower = error_message.lower()
return any(
indicator in error_lower
for indicator in [
"unique constraint failed", # SQLite
"duplicate key", # PostgreSQL
"unique violation", # PostgreSQL alternative
"duplicate entry", # MySQL
]
)
async def list_chat_completions(