TutorialMaker / pipeline /__init__.py
vivekchakraverty's picture
Re-architect: API-based acquisition, no download; ffmpeg -ss screenshots
f07812e
Raw
History Blame Contribute Delete
358 Bytes
"""Pipeline package: YouTube topic -> captioned .docx tutorial (API-based, no download).
Stages (see app.run_pipeline for orchestration):
search -> sentiment (YouTube Data API comments) -> transcribe (youtube-transcript-api)
-> tutorial (LLM) -> frames (yt-dlp stream URL + ffmpeg -ss, weighted timestamps)
-> captions (VLM) -> docx_builder
"""