f09ae248f9
* add staging ios app build script * feat(api): add OpenTelemetry LGTM stack for local observability Wire OTel traces, metrics, and logs through a collector to Tempo, Prometheus, and Loki, with custom LLM instrumentation, dev compose overlay, Grafana provisioning, env templates, and development.sh auto-start. * feat: expand observability, harden dev tooling, and fix expo staging UX Add business and LLM Prometheus metrics with Grafana dashboards, alerting, and a metrics verification script. Wire telemetry through adapters and core LLM paths, and document the local LGTM workflow. Fix development.sh for macOS bash 3.2, open Grafana and eval-web in Chrome, and repair eval-web auto-open (unbound EVAL_WEB_BROWSER_SCHEDULED). Merge internal-eval into the main dev script with improved compose handling. Require EXPO_PUBLIC_* at build time, improve iOS HTTP ATS for staging IPs, show memoir empty state instead of load errors when no chapters exist, and add jest env setup plus chapter list response normalization. * chore: enable Grafana Assistant Cursor plugin * fix: memoir empty state and repair withdrawn 0020_chapters_book_id stamp Show empty memoir UI when the chapter list succeeds with no items; treat auth/404 as non-fatal. Extend alembic revision repair so local dev DBs stamped with the removed 0020_chapters_book_id migration can roll back and upgrade to 0019. --------- Co-authored-by: Kevin <kevin@brighteng.org> Co-authored-by: Cursor <cursoragent@cursor.com>
125 lines
3.8 KiB
Python
125 lines
3.8 KiB
Python
"""Use-case oriented LLM gateway.
|
||
|
||
This is a small compatibility layer over the existing provider and JSON helper
|
||
functions. It gives new code a stable place to request model capabilities while
|
||
older agents continue to use LangChain directly during the transition.
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
from dataclasses import dataclass
|
||
from typing import Any, Callable, TypeVar
|
||
|
||
from pydantic import BaseModel
|
||
|
||
from app.core.dependencies import get_llm_provider, get_llm_provider_fast
|
||
from app.core.llm_call import allm_json_call, llm_json_call
|
||
from app.core.llm_telemetry import langchain_invoke_span
|
||
|
||
T = TypeVar("T", bound=BaseModel)
|
||
|
||
|
||
@dataclass(frozen=True)
|
||
class LlmUseCase:
|
||
name: str
|
||
fast: bool = False
|
||
max_tokens: int | None = None
|
||
temperature: float | None = None
|
||
model: str | None = None
|
||
|
||
|
||
class LlmGateway:
|
||
"""Facade for text and JSON LLM calls."""
|
||
|
||
def provider_for(self, use_case: LlmUseCase | None = None):
|
||
if use_case and use_case.fast:
|
||
return get_llm_provider_fast()
|
||
return get_llm_provider()
|
||
|
||
def langchain_llm_for(self, use_case: LlmUseCase | None = None) -> Any | None:
|
||
provider = self.provider_for(use_case)
|
||
return getattr(provider, "langchain_llm", None)
|
||
|
||
async def chat_text(
|
||
self,
|
||
messages: list[dict],
|
||
*,
|
||
use_case: LlmUseCase | None = None,
|
||
temperature: float | None = None,
|
||
model: str | None = None,
|
||
max_tokens: int | None = None,
|
||
) -> str:
|
||
provider = self.provider_for(use_case)
|
||
resolved_temperature = (
|
||
temperature
|
||
if temperature is not None
|
||
else (
|
||
use_case.temperature
|
||
if use_case and use_case.temperature is not None
|
||
else 0.7
|
||
)
|
||
)
|
||
resolved_model = model if model is not None else (use_case.model if use_case else None)
|
||
agent_name = use_case.name if use_case else "llm_gateway.chat"
|
||
kwargs = dict(
|
||
messages=messages,
|
||
temperature=resolved_temperature,
|
||
model=resolved_model,
|
||
max_tokens=(
|
||
max_tokens
|
||
if max_tokens is not None
|
||
else (use_case.max_tokens if use_case else None)
|
||
),
|
||
)
|
||
# DeepSeekProvider.complete 已包 langchain_invoke_span，避免双层 span
|
||
from app.adapters.llm.deepseek import DeepSeekLLMProvider
|
||
|
||
if isinstance(provider, DeepSeekLLMProvider):
|
||
return await provider.complete(**kwargs)
|
||
|
||
provider_label = type(provider).__name__.replace("Provider", "").lower() or "unknown"
|
||
with langchain_invoke_span(
|
||
agent=agent_name,
|
||
provider=provider_label,
|
||
model=resolved_model or "unknown",
|
||
call_type="chat",
|
||
):
|
||
return await provider.complete(**kwargs)
|
||
|
||
async def json_object(
|
||
self,
|
||
prompt: str,
|
||
schema: type[T],
|
||
*,
|
||
use_case: LlmUseCase,
|
||
fallback_factory: Callable[[], T] | None = None,
|
||
) -> T:
|
||
return await allm_json_call(
|
||
self.langchain_llm_for(use_case),
|
||
prompt,
|
||
schema,
|
||
max_tokens=use_case.max_tokens or 1024,
|
||
agent=use_case.name,
|
||
fallback_factory=fallback_factory,
|
||
)
|
||
|
||
def sync_json_object(
|
||
self,
|
||
prompt: str,
|
||
schema: type[T],
|
||
*,
|
||
use_case: LlmUseCase,
|
||
fallback_factory: Callable[[], T] | None = None,
|
||
) -> T:
|
||
return llm_json_call(
|
||
self.langchain_llm_for(use_case),
|
||
prompt,
|
||
schema,
|
||
max_tokens=use_case.max_tokens or 1024,
|
||
agent=use_case.name,
|
||
fallback_factory=fallback_factory,
|
||
)
|
||
|
||
|
||
__all__ = ["LlmGateway", "LlmUseCase"]
|