v1.9.3

Changes in this release:

• Bumps yt-dlp to 2026.03.17 in the uv lock file

v1.9.2

Changes in this release:

Increase the wait time on try-login to allow for multiple redirects after the login process, especially on JS-heavy pages.

v1.9.1

This release:

• Prevents capture of AI-upscaled super-resolution videos
• Includes some site-specific behavior updates
• Bumps some dependency versions

v1.9.0

This release contains several new features along with enhancements for performance and reliability. It also contains several dependency updates, including yt-dlp 2025.12.08.

New behavior controls

There are several new features making behaviors more flexible, allowing navigating and gathering outlinks from more complex sites than in previous versions.

• A new repeatUntilSelector control makes it possible to instruct behaviors when to stop looping, providing a more robust means to control when looping should stop early. (#431 - @mistydemeo)
• It's now possible for custom behavior JavaScript to call the pre-defined outlink gathering function. (#429 - @mistydemeo)
• It's now possible to gather outlinks while behaviors are running, which ensures that it's possible to gather outlinks from sites whose content changes at runtime like sites using JavaScript pagination. (#433/#434 - @mistydemeo)

Performance enhancements

Two new minor performance enhancements limit the default size of page screenshots (#420 - @vbanos) and reduces crawl startup time by caching the Chrome version (#424 - @vbanos).

Reliability enhancements

• Brozzler now detects Chrome error pages and will retry the affected page. (#438 - @adam-miller)
• try-login will now perform form validation before submitting and trigger input events. (#432/#440 - @netoarmando)
• Improved error handling in retry loops. (#441 - @adam-miller)

v1.8.1

This is a bugfix release which fixes an issue which caused the max_sites_to_claim filter to be applied too early, resulting in claimable sites sometimes being skipped. (#422)

This release also drops support for Python 3.9; the new minimum version is Python 3.10.

v1.8.0

This release contains a number of new capture features. It also removes support for Python 3.8.

• The new --disable-video-capture commandline flag allows excluding video captures in the brozzler-new-site commandline tool. (#379)
• When video capture has been disabled, brozzler will no longer browse YouTube UMP packets. (#378)
• Audio content types are now also skipped when video capture is disabled. (#380)
• It's now possible to control whether Chrome is launched in headless mode using the --no-headless commandline flag and the headless keyword parameter in the Chrome.start method. The default hasn't changed. (#373)
• Improved performance when visiting page anchors. (#394)
• Improved compatibility when fetching robots.txt and page headers by enabling legacy renegotiation and disabling errors from unexpected SSL EOFs. (#411)
• Fixed a bug which could break captures under certain specific circumstances when a page interstitial is encountered. (#375)
• The header request timeout has been increased from 30 to 60 seconds. (#367)
• Updated the versions of several dependencies.

Thanks to @TheTechRobo for the features in pull requests #373 and #394!

v1.7.0

• Specifies additional extras when installing yt-dlp. The default group ensures that yt-dlp keeps its full feature set in future releases, and curl-cffi improves compatibility. (#366)
• Adds new controls for capture - an option to capture only PDFs, and an option to exclude capturing video. (#288)

v1.6.13

• Updates Chrome config to disable the HttpsFirstBalancedModeAutoEnable feature in order to disable auto HTTP-to-HTTPS upgrades. (#355)

v1.6.12

• Create new claim_sites() query (#353)

v1.6.11

• Updates yt-dlp config (#353)