原逻辑 if True: 导致数据库瞬时错误也终止整个语音事件循环, AI 回复被中途截断。改为仅当 reason == insufficient_tokens 时 才终止,其他扣费错误(如 DB 超时)不中断会话。 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| api | ||
| config | ||
| migrations | ||
| models | ||
| scripts | ||
| services | ||
| utils | ||
| __init__.py | ||
| main.py | ||
| nohup.out | ||
| pet_data.db | ||
| requirements.txt | ||
| run.py | ||
| start.sh | ||
| test_token_charge.py | ||
| test_voice_conversation.py | ||