v2.1.0

Hyperflux 2.1.0 (2026-06-19)

--extract-scan now reaches players that sit a few clicks away. When the landing
page has no direct media, flux follows internal watch/play/embed links a bounded
number of hops, re-scans each page, and when it finds the media it writes a
multi-step config that walks the same path at download time. Series index pages
are recognised too: the generated config lists the episodes and resolves each one
through the per-episode pipeline. The crawl is content-driven: it follows
promising links even when their URLs do not match a known watch/play pattern,
ranks them by relevance, stays inside the series you are scanning, and skips ad
and onclick/popunder traps. Control how far it looks with --extract-scan-depth=N
(default 2, max 3).

Scanning no longer writes straight into your active config dir. flux prints the
config it generates, stashes a pending copy, and tells you the id to run
"flux --save-config " with when you want it active. The id is stable per
source URL, so re-scanning the same site gives you the same one, and saving
refuses to overwrite an existing active config.

v2.0.0

Hyperflux 2.0.0 (2026-06-18)

Download a whole series, not just one file. Point flux at any episode page of a
supported site and it opens an interactive picker listing every episode: the ones
already on disk are ticked, the rest start selected, and one key selects them all.
Confirm and flux downloads only what's missing, keeping each file under the same
name it has on the site, so re-running later grabs just the new episodes.

• The episode picker is a clean terminal UI: arrow keys or j/k to move, space to
  toggle, "a" to select all, enter to download, q to quit. It scrolls through long
  series with no cap and redraws cleanly when the terminal is resized.

• Extractor configs now work out of the box: the bundled config installs active in a
  system directory, so a fresh install resolves matching pages with no setup, and
  "flux --extract-scan " writes the config it generates straight into your
  active config dir. A download that turns out to be a web page now hints you to run
  --extract-scan instead of silently saving the HTML.

• Config regexes can interpolate captured variables, which is what lets a config
  enumerate a series' episodes.

Everything from 1.x still applies: a page that matches no config is a normal direct
download, HLS streams are fetched in parallel and assembled, and packages are on the
apt/dnf/apk Cloudsmith repositories.

v1.1.0

Hyperflux 1.1.0 (2026-06-18)

Download media from a web page, not just a direct file URL. Point flux at a
page and, when a user-written extractor config matches the domain, it resolves
the real media URL by running a small pipeline of HTTP requests and regex
captures, then downloads it with the usual multiple connections. A page that
matches no config falls through to the normal direct download, unchanged.

• HLS (.m3u8) streams: segments are fetched in parallel, decrypted (AES-128),
  assembled, and remuxed to MP4 when ffmpeg is present (otherwise kept as a
  .ts file). Choose quality with --quality and the container with --mux.

• Config generator: "flux --extract-scan " inspects a page, scores the
  media candidates against ads, and writes a commented starter config. An
  interactive picker steps in when several candidates compete; --yes takes the
  best one without asking.

• Series: a config can list the episodes of a show. flux opens a multi-select
  screen, or takes --all or --episodes 1,3-5, and downloads the selection,
  carrying on past any episode that fails.

Extractor configs live in ~/.config/hyperflux/extractors/ as plain text and run
arbitrary HTTP requests, so review one from someone else before using it.
JavaScript-signed and DRM-protected sites are out of scope.

You can also install and update Hyperflux from signed apt, dnf and apk
repositories on Cloudsmith now, instead of downloading a package each time; the
README has the one-line setup per distribution.

v1.0.0

Hyperflux 1.0.0