admin/DataClaw
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: model arguments fixed

Browse Source
This commit is contained in:
qixinbo
2026-03-28 14:46:50 +08:00
parent 00e5587e75
commit bd731660ac
12 changed files with 361 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/app/core/llm_provider.py
+3 -2
View File
@@ -2,9 +2,10 @@ from typing import Optional, Dict
from nanobot.providers.azure_openai_provider import AzureOpenAIProvider
from nanobot.providers.openai_codex_provider import OpenAICodexProvider
from nanobot.providers.openai_compat_provider import OpenAICompatProvider
from nanobot.providers.registry import find_by_name
from app.core.patched_openai_compat_provider import PatchedOpenAICompatProvider
def normalize_provider_name(provider: Optional[str]) -> Optional[str]:
if not provider:
@@ -51,7 +52,7 @@ def build_llm_provider(
extra_headers=extra_headers,
)
return OpenAICompatProvider(
return PatchedOpenAICompatProvider(
api_key=api_key,
api_base=api_base,
default_model=model,
backend/app/core/nanobot.py
+28 -2
View File
@@ -19,7 +19,6 @@ from nanobot.bus.events import OutboundMessage
from nanobot.bus.queue import MessageBus
from nanobot.config.loader import load_config
from nanobot.cron.service import CronService
from nanobot.providers.openai_compat_provider import OpenAICompatProvider
from nanobot.providers.openai_codex_provider import OpenAICodexProvider
from nanobot.providers.azure_openai_provider import AzureOpenAIProvider
from nanobot.providers.base import GenerationSettings
@@ -32,6 +31,7 @@ from nanobot.config.schema import Config
# or just import here if we are confident.
# Given the structure, importing here should be fine as long as skills.py doesn't import nanobot.py.
from app.api.skills import load_skills
from app.core.patched_openai_compat_provider import PatchedOpenAICompatProvider
from app.services.llm_cache import get_llm_configs, get_active_llm_config
from app.core.data_root import get_workspace_root
@@ -45,6 +45,7 @@ class NanobotIntegration:
self._started = False
self._model_agent_cache: Dict[tuple[str | None, int | None], AgentLoop] = {}
self._model_agent_lock = asyncio.Lock()
self._last_usage_by_session: Dict[str, Dict[str, Any]] = {}
@staticmethod
def _normalize_config_value(value: Any) -> Any:
@@ -80,6 +81,28 @@ class NanobotIntegration:
return content
return str(response)
@staticmethod
def _normalize_usage(usage: Any) -> Dict[str, int] | None:
if not isinstance(usage, dict):
return None
normalized: Dict[str, int] = {}
prompt = int(usage.get("prompt_tokens", 0) or 0)
completion = int(usage.get("completion_tokens", 0) or 0)
total = int(usage.get("total_tokens", 0) or 0)
# If total_tokens is missing or zero, calculate it
if total == 0:
total = prompt + completion
normalized["prompt_tokens"] = prompt
normalized["completion_tokens"] = completion
normalized["total_tokens"] = total
return normalized if (prompt > 0 or completion > 0) else None
def get_last_usage(self, session_id: str) -> Dict[str, int] | None:
usage = self._last_usage_by_session.get(session_id)
return dict(usage) if usage else None
def _need_custom_agent_for_target(self, target_config: Dict[str, Any]) -> bool:
if not self.agent:
return False
@@ -220,7 +243,7 @@ class NanobotIntegration:
extra_headers=extra_headers,
)
return OpenAICompatProvider(
return PatchedOpenAICompatProvider(
api_key=api_key,
api_base=api_base,
default_model=model,
@@ -407,6 +430,9 @@ class NanobotIntegration:
on_progress=on_progress,
on_stream=on_stream,
)
usage = self._normalize_usage(getattr(agent_to_use, "_last_usage", None))
if usage:
self._last_usage_by_session[session_id] = usage
return self._extract_response_text(response)
def _normalize_session_messages(self, messages: List[Any]) -> List[dict[str, Any]]:
backend/app/core/patched_openai_compat_provider.py
+43
View File
@@ -0,0 +1,43 @@
from __future__ import annotations
from typing import Any
from nanobot.providers.openai_compat_provider import OpenAICompatProvider
class PatchedOpenAICompatProvider(OpenAICompatProvider):
_MAX_COMPLETION_TOKEN_MODELS = ("gpt-5", "o1", "o3", "o4")
def _build_kwargs(
self,
messages: list[dict[str, Any]],
tools: list[dict[str, Any]] | None,
model: str | None,
max_tokens: int,
temperature: float,
reasoning_effort: str | None,
tool_choice: str | dict[str, Any] | None,
) -> dict[str, Any]:
kwargs = super()._build_kwargs(
messages=messages,
tools=tools,
model=model,
max_tokens=max_tokens,
temperature=temperature,
reasoning_effort=reasoning_effort,
tool_choice=tool_choice,
)
model_name = (model or self.default_model or "").lower()
spec = self._spec
supports_max_completion_tokens = bool(
spec and getattr(spec, "supports_max_completion_tokens", False)
)
should_use_max_completion_tokens = supports_max_completion_tokens or any(
token in model_name for token in self._MAX_COMPLETION_TOKEN_MODELS
)
if should_use_max_completion_tokens and "max_tokens" in kwargs:
kwargs["max_completion_tokens"] = kwargs.pop("max_tokens")
return kwargs
backend/main.py
+31 -2
View File
@@ -268,6 +268,7 @@ def _persist_assistant_enrichment(
session_id: str,
viz_payload: Optional[Dict[str, Any]] = None,
artifacts: Optional[List[Dict[str, Any]]] = None,
usage: Optional[Dict[str, Any]] = None,
) -> None:
if not nanobot_service.agent:
return
@@ -281,9 +282,25 @@ def _persist_assistant_enrichment(
if artifacts:
session.messages[-1]["artifacts"] = artifacts
changed = True
if usage:
session.messages[-1]["usage"] = usage
changed = True
if changed:
nanobot_service.agent.sessions.save(session)
def _extract_reasoning_content(session_messages: List[Dict[str, Any]]) -> str:
for message in reversed(session_messages):
if not isinstance(message, dict):
continue
if message.get("role") != "assistant":
continue
reasoning_content = message.get("reasoning_content")
if isinstance(reasoning_content, str) and reasoning_content.strip():
return reasoning_content
break
return ""
@app.post("/nanobot/chat")
async def nanobot_chat(request: ChatRequest):
try:
@@ -321,10 +338,12 @@ async def nanobot_chat(request: ChatRequest):
artifacts = extract_artifacts(text, session_messages)
viz_payload = current_viz_data.get()
usage = nanobot_service.get_last_usage(request.session_id)
_persist_assistant_enrichment(
session_id=request.session_id,
viz_payload=viz_payload if isinstance(viz_payload, dict) else None,
artifacts=artifacts,
usage=usage,
)
payload = {
@@ -334,6 +353,8 @@ async def nanobot_chat(request: ChatRequest):
}
if artifacts:
payload["artifacts"] = artifacts
if usage:
payload["usage"] = usage
return payload
except Exception as e:
raise HTTPException(status_code=500, detail=str(e))
@@ -356,7 +377,9 @@ async def nanobot_chat_stream(request: ChatRequest):
async def _on_progress(content: str, **kwargs: Any) -> None:
if content:
await progress_queue.put(content)
payload: Dict[str, Any] = {"type": "progress", "content": content}
payload.update(kwargs)
await progress_queue.put(payload)
async def _on_stream(delta: str) -> None:
if delta:
@@ -427,9 +450,10 @@ async def nanobot_chat_stream(request: ChatRequest):
session = nanobot_service.agent.sessions.get_or_create(request.session_id)
session_messages = session.messages
artifacts = extract_artifacts(text, session_messages)
reasoning_content = _extract_reasoning_content(session_messages)
# Check again for viz payload after task completes if not sent yet
viz_payload = current_viz_data.get()
usage = nanobot_service.get_last_usage(request.session_id)
if viz_payload:
try:
current_hash = hash((
@@ -447,11 +471,16 @@ async def nanobot_chat_stream(request: ChatRequest):
session_id=request.session_id,
viz_payload=viz_payload if isinstance(viz_payload, dict) else None,
artifacts=artifacts,
usage=usage,
)
final_payload = {"type": "final", "content": text}
if reasoning_content:
final_payload["reasoning_content"] = reasoning_content
if artifacts:
final_payload["artifacts"] = artifacts
if usage:
final_payload["usage"] = usage
yield f"data: {json.dumps(final_payload, ensure_ascii=False)}\n\n"
yield f"data: {json.dumps({'type': 'done'}, ensure_ascii=False)}\n\n"
except asyncio.CancelledError:
backend/tests/test_chat_project_id.py
+45
View File
@@ -98,3 +98,48 @@ def test_nanobot_chat_stream_syncs_project_id(monkeypatch) -> None:
assert "stream-complete" in content
assert calls == [{"session_key": "api:test-3", "project_id": 202}]
assert process_kwargs and process_kwargs[0]["project_id"] == 202
def test_nanobot_chat_stream_emits_reasoning_flags_and_final_reasoning(monkeypatch) -> None:
async def fake_process_message(*args, **kwargs):
on_progress = kwargs.get("on_progress")
on_stream = kwargs.get("on_stream")
if on_progress:
await on_progress("模型正在拆解问题", is_reasoning=True)
await on_progress("开始执行工具", tool_hint=True)
if on_stream:
await on_stream("answer-token")
return "final-answer"
class _DummySession:
def __init__(self):
self.metadata = {}
self.messages = [
{"role": "assistant", "content": "final-answer", "reasoning_content": "完整思考过程"}
]
class _DummySessions:
def get_or_create(self, _key):
return _DummySession()
class _DummyAgent:
def __init__(self):
self.sessions = _DummySessions()
async def collect_stream_chunks(response) -> list[str]:
chunks: list[str] = []
async for chunk in response.body_iterator:
chunks.append(chunk.decode("utf-8") if isinstance(chunk, bytes) else chunk)
return chunks
monkeypatch.setattr(main.nanobot_service, "process_message", fake_process_message)
monkeypatch.setattr(main.nanobot_service, "agent", _DummyAgent())
request = main.ChatRequest(message="hello", session_id="api:test-4", project_id=303)
response = asyncio.run(main.nanobot_chat_stream(request))
chunks = asyncio.run(collect_stream_chunks(response))
content = "".join(chunks)
assert '"type": "progress", "content": "模型正在拆解问题", "is_reasoning": true' in content
assert '"type": "progress", "content": "开始执行工具", "tool_hint": true' in content
assert '"type": "final", "content": "final-answer", "reasoning_content": "完整思考过程"' in content
backend/tests/test_patched_openai_compat_provider.py
+62
View File
@@ -0,0 +1,62 @@
import sys
from pathlib import Path
BACKEND_ROOT = Path(__file__).resolve().parents[1]
REPO_ROOT = BACKEND_ROOT.parent
NANOBOT_ROOT = REPO_ROOT / "nanobot"
if str(BACKEND_ROOT) not in sys.path:
sys.path.insert(0, str(BACKEND_ROOT))
if str(NANOBOT_ROOT) not in sys.path:
sys.path.insert(0, str(NANOBOT_ROOT))
from app.core.llm_provider import build_llm_provider
from app.core.nanobot import NanobotIntegration
from app.core.patched_openai_compat_provider import PatchedOpenAICompatProvider
def test_build_llm_provider_uses_max_completion_tokens_for_gpt5() -> None:
provider = build_llm_provider(
model="gpt-5.4-nano",
provider="openai",
api_key="test-key",
api_base="https://example.com/v1",
)
assert isinstance(provider, PatchedOpenAICompatProvider)
kwargs = provider._build_kwargs(
messages=[{"role": "user", "content": "hello"}],
tools=None,
model="gpt-5.4-nano",
max_tokens=5,
temperature=0,
reasoning_effort=None,
tool_choice=None,
)
assert kwargs["max_completion_tokens"] == 5
assert "max_tokens" not in kwargs
def test_nanobot_provider_keeps_max_tokens_for_legacy_models() -> None:
integration = NanobotIntegration()
provider = integration._build_provider(
model="gpt-4o-mini",
provider_name="openai",
api_key="test-key",
api_base="https://example.com/v1",
extra_headers=None,
)
assert isinstance(provider, PatchedOpenAICompatProvider)
kwargs = provider._build_kwargs(
messages=[{"role": "user", "content": "hello"}],
tools=None,
model="gpt-4o-mini",
max_tokens=5,
temperature=0,
reasoning_effort=None,
tool_choice=None,
)
assert kwargs["max_tokens"] == 5
assert "max_completion_tokens" not in kwargs