Skip to content
Efe0729Public

About

A desktop video downloader that learns any site you add — browse, queue and resume HLS/m3u8 & MP4 downloads, then auto-upload to Google Drive, Mega or S3 via rclone.

Topics

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

Siphon

Download from any site you add, upload to the cloud.
A desktop downloader that learns a site from its own pages, so adding one takes an address instead of a code change.


What it does

Siphon ships with no sites in it. You add the sites you use, and it works out the rest by reading them:

  • Browse — your sites appear with their own categories, search and pages, with thumbnails, titles and running times.
  • Download — paste a link, import a list, or pick videos while browsing. A queue with positions, pause/resume, retries, speed limit, scheduling and an estimate of the time left worked out from the speed actually measured, and downloads that pick up where they stopped after a pause or restart.
  • Upload — send finished files to Google Drive: per-upload destination, parallel transfers, pause/resume, skipping files already there, and a queue that survives a restart.

Adding a site

Settings → My own sites → type an address (example.com is enough, a video page works too) → Test and add. Several addresses at once are fine.

Siphon then reads the site itself: where its videos are, how its listings page, how its search box is built, and which engine it was built on (Kernel Video Sharing and friends are recognised and come with their layout already known). What it learns is remembered, and a layout learned from one site is reused on the next site built the same way.

If a site cannot be opened from here at all — members only, region blocked — save one of its video pages from your browser (Ctrl+S) and use Teach from a saved page: the same information is read out of the file.

Sites that fight back

A site behind a "verify you are human" check, or one that builds its player or its listings with JavaScript, is opened in the browser already installed on the computer (Edge or Chrome), off-screen and in Siphon's own profile. The page is taken from there, the pass is kept, and ordinary requests work again afterwards. Nothing to configure.

Command line and Docker

Siphon also runs without a window (siphon-cli, or the Docker image) for a NAS or a server; see docs/docker-cli.md.

Installing (Windows, from source)

  1. Install Python — tick Add python.exe to PATH while installing.
  2. Put this folder wherever you like, for example %USERPROFILE%\Siphon.
  3. Double-click Siphon.bat. The first run builds Siphon its own Python environment and installs what it needs, which takes a few minutes; after that it starts straight away.

Derle.bat turns the same folder into a single dist\Siphon.exe that needs no Python. It is only worth running once you are done changing things.

Where your things are kept

Settings, the download queue, the history and the upload queue live in %APPDATA%\Siphon. They are stored scrambled: on Windows they are tied to your Windows account through DPAPI, so another account on the same computer cannot read which sites you added or what you downloaded.

A log is kept in the app's own folder, under logs\, to make problems easier to report. Video names, file names and the addresses of individual videos are left out of it — it records that a download failed and which site answered, not what you were downloading.

Known limitations

  • Screen readers and keyboard navigation. The interface draws its own controls, so a screen reader sees very little of it and not everything can be reached from the keyboard. This is a real gap, not an oversight; it would take a different toolkit to fix properly.
  • Sites that show no videos on their front page (some are built for each visitor, or are search-only) open on their first section when they have one; otherwise Browse says so and points at the search box.
  • Update checking looks at this project's Releases page (GITHUB_REPOSITORY in src/siphon/metadata.py) and only installs a build published there.

When a site is read wrongly

Siphon works out each site by itself, so sooner or later it will get one wrong: names that all read the same, a search box that shows the same results whatever you type, pages that will not turn.

Press Report — on the result screen after adding a site, or beside the site in Settings - My Sites. It checks the site again and writes down what it found: the engine, the listing, the rule the names came from, three example names, and the search and paging addresses it tried with the reason each was refused. That last part is usually the whole answer.

You then choose how to send it:

  • Open an issue on GitHub - fastest to get fixed, and public: everything in the report, the site's address included, can be read by anyone.
  • Copy, or Send with Gmail to contact.efe0729@gmail.com - for anyone who would rather not post a site they visit on a public page. The Gmail button writes the message in your browser; if you use something else, Copy and send it from there.

The report holds one site and nothing else: no download history, no queue, no other sites you have added, no file paths from your computer. Ticking Include the log adds the log lines about that one site, and the whole of it is shown in the window before anything is sent.

Your responsibility

Siphon ships with no sites in it and takes no view on what you add. What you download, and whether it is yours to download, is between you and the site you take it from: respect their terms and the law where you live.

License

Licensed under the Apache License 2.0 (Apache-2.0) — see LICENSE and THIRD_PARTY_NOTICES.md.

About

A desktop video downloader that learns any site you add — browse, queue and resume HLS/m3u8 & MP4 downloads, then auto-upload to Google Drive, Mega or S3 via rclone.

Topics

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Contributors

Languages