xinyin025/LangBot
1
0
Fork 0
mirror of https://github.com/langbot-app/LangBot.git synced 2026-08-09 12:40:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(provider): preserve think tags in streamed reasoning

Browse Source
This commit is contained in:
fdc310
2026-08-06 01:50:20 +08:00
parent d6f01adb8a
commit a1c2c975cb
2 changed files with 74 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py
+32 -5
View File
@@ -1135,6 +1135,8 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
role = 'assistant'
tool_call_state: dict[int, dict[str, typing.Any]] = {}
think_state = _ThinkStripState() if remove_think else None
reasoning_started = False
reasoning_closed = False
try:
response = await acompletion(**args)
@@ -1165,18 +1167,36 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
if 'role' in delta and delta['role']:
role = delta['role']
delta_content = delta.get('content', '')
reasoning_content = delta.get('reasoning_content', '')
delta_content = delta.get('content') or ''
reasoning_content = delta.get('reasoning_content') or ''
provider_fields = dict(delta.get('provider_specific_fields') or {})
# Handle reasoning_content based on remove_think flag
if reasoning_content:
provider_fields['reasoning_content'] = reasoning_content
if remove_think:
delta_content = None
delta_content = delta_content or None
else:
# Use reasoning_content as the displayed content
delta_content = reasoning_content
# Stream explicit markers so downstream adapters and
# the debug page see the same format as non-streaming
# responses.
if not reasoning_started:
delta_content = '<think>\n'
reasoning_started = True
else:
delta_content = ''
delta_content += reasoning_content
if delta.get('content'):
delta_content += f'\n</think>\n{delta.get("content")}'
reasoning_closed = True
elif delta_content and not remove_think and reasoning_started and not reasoning_closed:
delta_content = f'\n</think>\n{delta_content}'
reasoning_closed = True
if finish_reason and not remove_think and reasoning_started and not reasoning_closed:
delta_content = f'{delta_content}\n</think>\n'
reasoning_closed = True
if think_state is not None and delta_content:
delta_content = think_state.feed(delta_content)
@@ -1205,6 +1225,13 @@ class LiteLLMRequester(requester.ProviderAPIRequester):
yield provider_message.MessageChunk(**chunk_data)
chunk_idx += 1
if reasoning_started and not reasoning_closed:
yield provider_message.MessageChunk(
role=role,
content='\n</think>\n',
is_final=True,
)
if think_state is not None:
pending_content = think_state.flush()
if pending_content: