3.0.2

Fixed

• Fix extraction of header on non-stackoverflow and non-stackexchange domains (#385)

3.0.1

• Fix scraper is looping over failing URLs forever (#379)
• Fix extraction of header content for stackoverflow domains (#378)
• Rewrite CSS files (#383)

3.0.0

Changed

• Upgrade to Python 3.14 + Debian bookworm (#351)
• BREAKING: Replace usage of multiple --tag with single CSV --tags for Zimfarm integration (#351)
• Enhance logic to download images to better respect upstream servers (#367 and #369)

### Fixed

• Docker image does not build anymore on main branch (#346)
• Scraper fails while processing posts tags (#338)
• Posts tags can be split by | or by >< characters (#356)
• Fix numbers passed to Users progresser (#353)
• Fix display of comments randomly not present (#361)
• Add SVG icons for all left menu entries (#365)
• Fetch, rewrite and include real online page header (including potential SVG image) (#241)
• Simplify mobile topbar to better reuse styles already fetched (#364)
• Fix display of user infos in posts (#362)
• Fix display of post vote counts (#355)
• Add support for SVG images (#364)
• Fix computation of article excerpt to produce valid safe HTML (#320)
• Adjust computation of linked questions and remove related ones (#263)
• Scraper keeps re-uploading same images (#373)

2.2.1

Fixed

• Reduce number of workers and add backoff on 429 and non-HTTP (network, ...) errors

Changed

• Changed default --redis-url behavior (#333)
  • It now uses the REDIS_URL environment variable if set, falling back to redis://localhost:6379.
  • Docker image sets REDIS_URL to unix:///var/run/redis.sock by default.

2.2.0

Changed

• Breaking changes: adapt to new StackExchange dumps and missing Sites.xml (#322)
  • Only working on recent dumps (June 2024 and later)
  • --title and --description CLI parameters are now mandatory to specify ZIM metadata
  • Dropped -l/--list-all CLI action to list all SE sites (not working anymore)

Fixed

• Fix duplicate english Language metadata (#321)
• Change image processing order to save memory (#325)
• Fix confusion between selection and flavour in ZIM name (#327)
• New XML dump files have changed (#329)

2.1.3

Fixed

• Fix Mathjax equations not displayed properly (#283)

2.1.2

Fixed

• User icons don't load properly (#301)
• Revert adaptations to upstream XML format changes (#313)

2.1.1

Fixed

• Adapt to upstream XML format changes (#305)
• Add continuous delivery to Pypi (#303)

2.1.0

Added

• Redirection from /questions/{questionId} to the question page (#277)

Changed

• ZIM Tags now include _videos:no;_details:no and conditionaly include _pictures:no (#278)
• Default filename now uses nopic instead of all if using --without-images (#278)
• Multi-language domains now handled as such:
  • Language metadata to be set to eng,xxx (xxx being the second language)
  • Name metadata to be like "{domain}mul{variant}"
  • Filename metadata to match Name
• Using zimscraperlib 3.3
• Changed default publisher metadata from 'Kiwix' to 'openZIM'
• description metadata is now limited to 80 chars, full description goes to the long_description (#290)

Fixed

• Multilanguage ZIM are not perfectly handled (#259)
• Incorrect image displayed (#284)
• Markdown text formatting is not rendered (#286)
• Harmonize default publisher to openZIM (#291)
• Docker image: align redis binaries with Python distribution (#294)
• Issue with xml.sax.saxutils (#298)

2.0.2

Changed

• Fixed language-code-looking project codes setting incorrect Language (ell, or, vi)
• Fixed --name parameter not being used to set Name nor filename (#267)
• Sax parser now explicitly closed after use
• Fixed same-protocol links being considered relative paths during rewriting (#265)
• More reliable database commits
• Updated to zimscraperlib 1.8.0 and lxml 4.9.1
• Removed inline JS to comply with some CSP
• renamed redis module to avoid confusion
• External link icon now included