[CLOSED] [Bug] Open-API Crashes on Truncated JSON from Android Clients #50
Description
Sunday Aug 10, 2025 at 02:22 GMT
Originally opened as aws-samples#55
Describe the bug
The open-api seems to have a flaw: when it receives a truncated JSON (which frequently occurs when an Android client connects to an LLM inference engine on a high-speed local network), it immediately throws an error and terminates the chat.
I suspect this is a flaw in the file:
react-native/src/api/open-api.ts
Steps To Reproduce
Steps to reproduce the behavior:
Use mlc-llm to deploy a openai compatible llm engine.
Use the Android client in the same LAN to connect to it and chat.
This may be a more edge case, and it is not easy to reproduce the environment. I can help test it.Expected behavior
No more errors.
App Version
2.4.0
React Native version
.
The platform of the bug
Android
Relevant log output
<details>
<summary>Log Messages</summary>
During streaming reception, the following appears at the end:
data: {"id":"chatcmpl-c435dcca209f4943b474e0bfbca0413f","choices":[{"finish_reason":null,"index":0,"delta":{"content":"hello","role":"assistant","name":null,"tool_calls":null,"tool_call_id":null},"logprobs":null}],"created":1754622244,"model":"q
-------------------- or -------------------
data: {"id":"chatcmpl-45cf0a3f0ee74983a4b3c2cf3bc2f368","choices":[{"finish_reason":null,"index":0,"delta":{"content":"”","role":"assistant","name":null,"tool_calls":null,"tool_call_id":null},"logprobs":null}],"created":1754622149,"model":"qwen","
</details>Is this a regression?
no
Regression additional context
No response
Additional context
No response