Рабочий пайплайн: YouTube -> транскрипция -> выжимка -> Postgres

- новый api/: /pipeline/process (yt-dlp, субтитры или Vosk, LLM/экстрактивная
  суммаризация), /videos, /llm; старый код перенесён в api_legacy/
- web/: Flet UI (flet 0.28.3 + flet-web, порт 8550)
- Dockerfile.api: ffmpeg слоем из mwader/static-ffmpeg, pip с кэш-маунтом
- requirements/pyproject: убраны moviepy, imageio-ffmpeg, битый asyncio
- README с инструкцией запуска и загрузкой Vosk-модели

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
jze9
2026-07-14 11:12:46 +05:00
parent 564c9c2d2f
commit 2697e01714
51 changed files with 2313 additions and 284 deletions

View File

@@ -0,0 +1,97 @@
from fastapi import APIRouter, HTTPException
from pydantic import BaseModel
from fastapi.responses import FileResponse
from pathlib import Path
import asyncio
router = APIRouter(prefix="/moviepy", tags=["moviepy"])
class VideoToMp3Request(BaseModel):
filename: str
class VideoToWavRequest(BaseModel):
filename: str
def _run_moviepy(src: Path, dest: Path):
from moviepy import VideoFileClip
clip = VideoFileClip(str(src))
try:
if clip.audio is None:
raise RuntimeError("Source file has no audio stream")
clip.audio.write_audiofile(str(dest), codec="libmp3lame", bitrate="192k")
finally:
clip.close()
def _run_moviepy_to_wav(src: Path, dest: Path):
from moviepy import VideoFileClip
clip = VideoFileClip(str(src))
try:
if clip.audio is None:
raise RuntimeError("Source file has no audio stream")
clip.audio.write_audiofile(str(dest), fps=16000, nbytes=2)
finally:
clip.close()
@router.post("/video-to-mp3")
async def video_to_mp3(req: VideoToMp3Request):
try:
import moviepy # noqa: F401
except Exception:
raise HTTPException(status_code=503, detail="moviepy is not installed in the runtime. Install moviepy/imageio-ffmpeg")
project_root = Path(__file__).resolve().parent.parent.parent
data_dir = project_root / "data"
src = (data_dir / req.filename).resolve()
try:
src.relative_to(data_dir)
except Exception:
raise HTTPException(status_code=400, detail="Invalid filename")
if not src.exists() or not src.is_file():
raise HTTPException(status_code=404, detail="Source file not found")
out_dir = data_dir / "audio"
out_dir.mkdir(parents=True, exist_ok=True)
dest = out_dir / f"{src.stem}.mp3"
try:
await asyncio.get_running_loop().run_in_executor(None, _run_moviepy, src, dest)
except Exception as e:
raise HTTPException(status_code=500, detail=f"conversion failed: {e}")
return FileResponse(str(dest), filename=dest.name, media_type="audio/mpeg")
@router.post("/video-to-wav")
async def video_to_wav(req: VideoToWavRequest):
try:
import moviepy # noqa: F401
except Exception:
raise HTTPException(status_code=503, detail="moviepy is not installed in the runtime. Install moviepy/imageio-ffmpeg")
project_root = Path(__file__).resolve().parent.parent.parent
data_dir = project_root / "data"
src = (data_dir / req.filename).resolve()
try:
src.relative_to(data_dir)
except Exception:
raise HTTPException(status_code=400, detail="Invalid filename")
if not src.exists() or not src.is_file():
raise HTTPException(status_code=404, detail="Source file not found")
out_dir = data_dir / "audio"
out_dir.mkdir(parents=True, exist_ok=True)
dest = out_dir / f"{src.stem}.wav"
try:
await asyncio.get_running_loop().run_in_executor(None, _run_moviepy_to_wav, src, dest)
except Exception as e:
raise HTTPException(status_code=500, detail=f"conversion failed: {e}")
return FileResponse(str(dest), filename=dest.name, media_type="audio/wav")