mirror/nanobot
1
0
Fork 0
mirror of https://github.com/HKUDS/nanobot.git synced 2026-05-10 03:36:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat(provider): support reasoning_content in OpenAI compat provider

Browse Source 
Extract reasoning_content from both non-streaming and streaming responses
in OpenAICompatProvider. Accumulate chunks during streaming and merge into
LLMResponse, enabling reasoning chain display for models like MiMo and DeepSeek-R1.

Signed-off-by: Lingao Meng <menglingao@xiaomi.com>
This commit is contained in:
Lingao Meng 2026-04-03 14:40:40 +08:00 committed by Xubin Ren
parent 0a31e84044
commit 210643ed68
1 changed files with 16 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
nanobot/providers/openai_compat_provider.py
View File

@ -385,9 +385,13 @@ class OpenAICompatProvider(LLMProvider):
content = self._extract_text_content( content = self._extract_text_content(
response_map.get("content") or response_map.get("output_text") response_map.get("content") or response_map.get("output_text")
) )
reasoning_content = self._extract_text_content(
response_map.get("reasoning_content")
)
if content is not None: if content is not None:
return LLMResponse( return LLMResponse(
content=content, content=content,
reasoning_content=reasoning_content,
finish_reason=str(response_map.get("finish_reason") or "stop"), finish_reason=str(response_map.get("finish_reason") or "stop"),
usage=self._extract_usage(response_map), usage=self._extract_usage(response_map),
) )
@ -482,6 +486,7 @@ class OpenAICompatProvider(LLMProvider):
@classmethod @classmethod
def _parse_chunks(cls, chunks: list[Any]) -> LLMResponse: def _parse_chunks(cls, chunks: list[Any]) -> LLMResponse:
content_parts: list[str] = [] content_parts: list[str] = []
reasoning_parts: list[str] = []
tc_bufs: dict[int, dict[str, Any]] = {} tc_bufs: dict[int, dict[str, Any]] = {}
finish_reason = "stop" finish_reason = "stop"
usage: dict[str, int] = {} usage: dict[str, int] = {}
@ -535,6 +540,9 @@ class OpenAICompatProvider(LLMProvider):
text = cls._extract_text_content(delta.get("content")) text = cls._extract_text_content(delta.get("content"))
if text: if text:
content_parts.append(text) content_parts.append(text)
text = cls._extract_text_content(delta.get("reasoning_content"))
if text:
reasoning_parts.append(text)
for idx, tc in enumerate(delta.get("tool_calls") or []): for idx, tc in enumerate(delta.get("tool_calls") or []):
_accum_tc(tc, idx) _accum_tc(tc, idx)
usage = cls._extract_usage(chunk_map) or usage usage = cls._extract_usage(chunk_map) or usage
@ -549,6 +557,10 @@ class OpenAICompatProvider(LLMProvider):
delta = choice.delta delta = choice.delta
if delta and delta.content: if delta and delta.content:
content_parts.append(delta.content) content_parts.append(delta.content)
if delta:
reasoning = getattr(delta, "reasoning_content", None)
if reasoning:
reasoning_parts.append(reasoning)
for tc in (delta.tool_calls or []) if delta else []: for tc in (delta.tool_calls or []) if delta else []:
_accum_tc(tc, getattr(tc, "index", 0)) _accum_tc(tc, getattr(tc, "index", 0))
@ -567,6 +579,7 @@ class OpenAICompatProvider(LLMProvider):
], ],
finish_reason=finish_reason, finish_reason=finish_reason,
usage=usage, usage=usage,
reasoning_content="".join(reasoning_parts) or None,
) )
@staticmethod @staticmethod
@ -630,6 +643,9 @@ class OpenAICompatProvider(LLMProvider):
break break
chunks.append(chunk) chunks.append(chunk)
if on_content_delta and chunk.choices: if on_content_delta and chunk.choices:
text = getattr(chunk.choices[0].delta, "reasoning_content", None)
if text:
await on_content_delta(text)
text = getattr(chunk.choices[0].delta, "content", None) text = getattr(chunk.choices[0].delta, "content", None)
if text: if text:
await on_content_delta(text) await on_content_delta(text)