Vomy

Pulling a channel's back catalogue — video metadata, playlists, the comments under each one — is the starting point for any analysis of it. The official route is the YouTube Data API: a key, an OAuth flow, and a daily quota that a single large channel can exhaust before you have finished collecting it.

The obvious alternative is to fetch the public pages directly, and the obvious naive version of that stops working within a few hundred requests. You get rate-limited, then blocked, and the run dies halfway through with a partial dataset and no way to resume from anything meaningful.

Vomy is a CLI over a scraper core, with the two kept deliberately separate: `cli.py` parses and orchestrates, `scraper.py` does the fetching, `models.py` defines what a channel, video and comment actually are, and configuration lives in `config/settings.py` rather than in flags scattered through the code. The rate-limit problem is handled on two fronts — rotation across proxies, fetched from public lists automatically or supplied as your own file, and a configurable `--sleep-interval` so a run can be deliberately slowed to something polite. Requests are shaped to look like a browser rather than a script. An interactive mode lets you pick which of the four data types you actually want instead of collecting everything by default.

One command against a channel URL produces a timestamped output directory of JSON — channel info, video metadata, comments and playlists — with no key, no OAuth and no quota. MIT-licensed and pure Python.