Highlights
- 15 new bot signatures (#659), including the Marginalia search crawler, Blackboard Ally, Amazon Quick agents, Meta's HatchFeed link preview, xAI's grok-agent, the opencode and agent-browser CLIs, Bun's default
fetch, Jetpack, Silktide and the Camoufox scraping browser. - Chromebook false positive fixed (#658). Android apps on Chromebooks with the
octopusboard were being flagged as crawlers. - Faster on long User-Agents (#634). The catch-all pattern no longer rescans long tokens, which without the PCRE JIT could take several seconds on very long strings. Results are unchanged.
- Much better real-browser test coverage (#652, #657). Around 2,900 current browser and in-app webview User-Agents have been added to the test data, so new signatures are checked against modern browsers. Contributions are welcome via the new "Contribute real browser user agents" issue form (#653).
What's Changed
- Add img2dataset by @uwereiher in #631
- Add PHP 8.6 to test matrix by @JayBizzle in #632
- Remove deprecated --no-suggest Composer flag by @JayBizzle in #633
- Stop the catch-all pattern rescanning long tokens by @JayBizzle in #634
- Mark unmaintained ports in the README and add a security policy by @JayBizzle in #651
- Add a way to contribute real browser user agents by @JayBizzle in #653
- Add current browser user agents to the devices test data by @JayBizzle in #652
- Cover the crawler regex failure branch directly by @JayBizzle in #655
- Add signatures for 15 undetected bots by @JayBizzle in #659
- Add more current browser user agents to the devices test data by @JayBizzle in #657
- Stop flagging Chromebooks with the octopus board name by @JayBizzle in #658
Full Changelog: v1.4.2...v1.4.3