Skip to content

Releases

v0.3.0

Choose a tag to compare

@github-actions github-actions released this 16 Nov 14:22
205b5a9

Added

  • A Markdown transcript in every gallery folder, named after the folder itself: a folder such as 123456789_Sample-Gallery/ holds 123456789_Sample-Gallery.md. It renders the page in page order with every image, the caption that belongs to it, every text block, and embed links, so the words that explain a slide are no longer lost. Images the page shows side by side stay on one line.
  • An offline HTML page in every gallery folder, named after the folder too. It mirrors the layout of the project page without Behance's own interface: the canvas colour and module spacing the author chose, full width slides that fill the column, narrower images at the size the author gave them, modules that hold several images as a row sized by each image's proportions, and text modules with their formatting. It carries its own styles and needs no network.
  • GalleryPlan carries the page's text modules and ImageAsset carries the caption the author wrote for it, so the renderings keep the author's words beside the images.
  • A py.typed marker, so type checkers apply the package's annotations.
  • mypy in strict mode, run over src and .github/scripts by CI.

Changed

  • GalleryPlan now holds slides instead of images, so a module that shows several images stays one slide. The flat image list is an images property derived from the slides, which keeps the download order and the reported count unchanged.
  • ImageAsset records display_width and display_height, the size the page lays the image out at.
  • Ruff now checks annotation presence, security rules, and pytest style, with documented waivers for tests.
  • Development and CI now use uv, with dependencies pinned in uv.lock. The dev dependencies moved from the dev extra to a dev dependency group, so pip install -e ".[dev]" no longer installs the test and lint tools.

Removed

  • Dropped Python 3.9, 3.10, and 3.11 support, so the package requires Python 3.12 or newer.

v0.2.0

Choose a tag to compare

@kevinkwee kevinkwee released this 15 Nov 03:56
a579b1a

Added

  • --videos to download video modules. Behance video is resolved through Adobe's Creative Cloud player to a direct MP4 rendition, and the best rendition is chosen unless a cap is set.
  • --video-height PX to cap the rendition height, falling back to the smallest available.
  • The video extra, pip install "behance-dl[video]", which adds yt-dlp so YouTube, Vimeo, and Dailymotion embeds download too. Without it those modules stay links and the run reports them as failed.
  • A redirect policy that refuses any redirect leaving https or changing host, so both URL allowlists still describe where a request can end up.

Changed

  • BehanceClient.download takes a media kind. An image download now rejects video bytes, and a video download rejects image bytes.
  • --force re-fetches videos as well as images.
  • A video is named from its validated asset id and height, so a re-run can match it on disk without trusting an expiring signed URL.

Fixed

  • When the derived rendition answers 404, the fallback file is named after the rendition that actually answered instead of the one that did not.
  • A playlist embed reports success once its single item is saved.

v0.1.0

Choose a tag to compare

@kevinkwee kevinkwee released this 15 Nov 03:56
fc34ba7

Added

  • Download every image of a public Behance gallery from the payload embedded in the page.
  • Variant selection by width with -W. The ranking prefers source, then a plain bucket, then an optimised copy, and drops single-frame stills when an animated version exists.
  • manifest.json resume, so a re-run only fetches what is missing and a width change re-fetches the affected images.
  • embeds.json for prototypes and video modules that are links rather than files.
  • Parallel downloads, retries with backoff, atomic writes, and content type plus magic byte validation.
  • The behance-dl console command and python -m behance_dl.