Files
YG_FT/backend/app/main.py
caoxiaozhu 03254f8196 feat(data_process): 模型缓存统一仓库根 .cache 并修复 PDF 页眉页脚清理
- 新增 app/core/cache_paths.py:HF_HOME / tiktoken 缓存统一指向 <repo>/.cache,
  本地与 Docker 路径一致,离线部署打包 .cache 即可
- Dockerfile.backend 的 tiktoken 词表改用官方 SHA 文件名,避免运行时回退重建
- 修复 layout_hybrid 路径不运行 detect_pdf_document_noise 的缺陷:
  needs_pdf_noise 不再与 needs_layout_raw 互斥,PDF 智能预处理在版面切分下也生效
- 新增 layout_noise.py:识别跨页重复的页眉表格标签组并按行剔除,
  解决 docling layout 模型把中文企业 PDF 页眉识别成普通 Table 导致清不掉的问题
- 回收 HybridChunker 丢弃的末尾孤立标题,找回章节标题内容
2026-08-21 10:16:23 +08:00

49 lines
1.5 KiB
Python

import asyncio
from contextlib import suppress
from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from app.core.cache_paths import setup_local_caches
# 在任何 docling / tiktoken 模块被实例化之前设置缓存路径,避免首调用走到 ~/.cache。
setup_local_caches()
from app.api.v1.router import api_router # noqa: E402
from app.core.config import docs_kwargs, get_settings # noqa: E402
from app.core.logging import configure_logging, setup_request_logging # noqa: E402
from app.workers.compute_poller import run_compute_poller # noqa: E402
def create_app() -> FastAPI:
settings = get_settings()
configure_logging(settings)
app = FastAPI(title=settings.app_name, **docs_kwargs(settings.enable_docs))
app.add_middleware(
CORSMiddleware,
allow_origins=settings.cors_allow_origins,
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
setup_request_logging(app)
app.include_router(api_router, prefix=settings.route_prefix)
@app.on_event("startup")
async def start_workers() -> None:
app.state.compute_poller_task = asyncio.create_task(run_compute_poller())
@app.on_event("shutdown")
async def stop_workers() -> None:
task = getattr(app.state, "compute_poller_task", None)
if task:
task.cancel()
with suppress(asyncio.CancelledError):
await task
return app
app = create_app()