AgentZero v1.2.1

This is a maintenance release:

• Added OpenAI and Claude bots
• Added tests
• Updated dependencies

AgentZero v1.2.0

This update add support for lots of new crawlers, and improves token handling internally:

• Reworked how tokens are ignored in agentzero::getTokens() to make them case-insensitive
• Added loads of new crawlers to crawlers::get()
• Added .NET CLR framework to frameworks::get()
• Added special class to handle difficult captures
• Updated Yahoo! Slurp parser
• Fix issue #6 typo in README.md
• "arm" is now "Arm" in architectures::get()
• Improved how apps are normalised in crawlers::getApp()
• Applebot is now categorised as AI
• Added bytespider crawler
• Fixed issue in agentzero::getTokens() where the single values array was encoded with preg_quote(), but the delimiter was not supplied
• AppleWebKit/537.36 is now marked as single as it is sometime preceded by other tokens that get captured together
• Added captures for Baidu and MicroMessenger apps
• Fixed PHPStan issues to level 9
• Added and updated test
• Updated dependencies

AgentZero v1.1.0

This release enables capture of more crawlers, and adds a new ai category:

• Apps are now not automatically type human
• Moved Powershell and AppleExchangeWebServices to crawlers, as there is no indication that these are human UA's
• Added Alamofire as a framework
• Fixed issue in other::getApp() where the app was expected to exist
• Added capture for iOS with a space after it
• Added Blazer browser (PalmOS)
• Added PalmOS and updated Windows captures to ignore PalmOS
• Added a new ai category, to mark crawlers that are specifically for AI
• Added specific mispellings that indicate a web scraper instead of the human it is trying to spoof
• Added and updated tests.

AgentZero v1.0.1

This release adds detection for more strings and improves how some strings are handled internally:

• Fixed issues parsing Google-Read-Aloud
• Cleaned up how specific category names and app names are specified for generic bot captures
• Added Cloudflare crawlers to capture list
• Fixes issue #5, missing Huawei Browser
• Added tests

AgentZero v1.0.0

This release improves detection, adds some new properties, and improves performance. Also is now marked as stable:

• Added PS Vita and PSP support
• Added Moblie as an alternative to Mobile as some Samsung devices misspell it
• Added new properties framework and frameworkversion to capture the runtime that the app was built in
• Added support for Weibo and WeChat apps
• Added support for QQ Browser
• Improved how IE and Trident is captured to account for the fact that the QQ Browser can be either Trident or WebKit
• Updated Chromium detection to set the rendering engine to WebKit when the version is less than 28
• Languages can now be captured from language/ or ByteLocale tokens
• Updated config in apps::get() to unify how app names are processed, now going through apps::getApp()
• Updated other::get() to normalise app names through apps::getApp()
• Moved capture for ByteFullLocale/ to languages::get().
• Updated crawlers::get() to improve normalisation of the Qwant crawler
• Improved how iPhone is captured for certain app strings
• Added tests/profile.php to process a load of UA strings so the code can be profiled
• Updated agentzero::parse() to generate a lowercase copy of the tokens, which is then passed to props::match(), this is to prevent the tokens being lowercased for every match that is made
• Added declare(strict_types = 1) to all test files
• Added and updated tests
• Updated README.md to remove beta flag

AgentZero v0.6.0

This release fixes some bugs and adds a new metric::

• Fixed #2 where the normalisation of the app name was not specific enough, and lost detail. There is now a new metric appname that extracts the original app name as specified in the UA string
• Fixed #4 where the version of Safari is specified in a different token, it is now extracted where available
• Updated test
• Updated docs and readme
• Updated dependencies

AgentZero v0.5.1

This is a service release that adds some improvements:

• The original UA string is now a property of the returned object
• Added snap preview crawler
• Updated tests
• Updated dependencies

AgentZero v0.5.0

This release improves the output quality, and adds some new properties:

• Tightened up what device and model represent, and upgraded parsers to produce the correct value
• Added RAM, CPU model number, and CPU clock metrics to output, as this is revealed by some app UA strings
• Added lowercase versions of some ignore strings in config::get()
• Added Google Quick Search Box to the apps list, fixed duplication with OcIdWebView, added imo, Amazon Kids Browser, and Viber
• Added support for detecting edge on Android
• Added Phone mapping in categories::get() to map to mobile
• Added a load of scrapers to crawlers::get()
• Fixed issues capturing some Amazon Fire Tablets in devices::get()
• Added support for capturing Samsung and Blackberry devices
• Redmi devices are now not mapped as Xiaomi devices
• Fixed issue in other::get() where Chrome was mapped as an app
• Added and updated tests
• Updated README.md to show the correct example now that the return value of agentzero::parse() is an agentzero object

AgentZero v0.4.0

This release normalises the output names better for platforms and browsers, improves information capture and parsing, and supports more browsers, vendors, and crawlers:

• Improved mappings for handling apps, to account for Oracle Webview, and information contained in the ByteDance (TikTok) app UA string
• Improved how Chrome browsers are handled and added detection from Cronet Android Webview
• Fixed issue in devices::getDevice() where the build and device was not separated properly if the build marker was a different case
• Added darkmode output from other::get(), where the information is contained within a JSON string
• Updated platforms::get() to handle short and full locale markers in the build string for Android devices
• Updated tests to only filter out values that are null for when tests are run, also in the indexphp test script
• Swapped other ahead of apps in the config as it decodes app JSON, which should override standard app info
• Updated apps::get() to normalise the app names for some common apps
• Updated browsers::get() to improve how the browser version is detected, to compensate for some odd strings
• Added detection for Vivo devices
• Fixed platform issues to allow text-based platform versions
• Some app names and device vendor names are now normalised
• Added more vendors to devices::getDevice()
• Platform names are now normalised to make sure the case is always consistent
• Improved parsing of the Instagram app when it is in proprietary format
• Added libcurl to scrapers
• Added in as a language code
• Added missing end command to agentzero::parse() to match strings at the end of a token
• Added some App captures, OculusBrowser and YaBrowser are now considered apps instead of browsers
• Improved architecture captures
• Added a load of new browsers such as Links, Elinks, BonEcho (Firefox), minimo, and some new Firefox captures to take into account unmatched patterns
• Improved and added crawler captures
• Improved and added device captures, added vendors, and added a generic capture for screen dimensions
• Engines now define captures as human, improved capturing
• Added more Linux flavours, and improved generic Linux version capturing
• Added and updated tests
• Updated support documentation

AgentZero v0.3.0

This release tightens up how human devices are detected, improves string parsing, and adds more devices/crawlers:

• Updated agentzero::getTokens() to split tokens by commas
• Updated configurations to handle the new token layout
• Updated browsers::get() to set all browsers as type=human
• Removed setting of type=human from platforms::get()
• Mappings in architectures::get() now specify category=desktop where possible
• Moved engines and browsers up the pecking order over apps, platforms and architectures, as browsers define whether the UA is human
• Updated browsers::get() to normalise the browser name
• Updated Facebook app language parser to normalise it, and account for odd strings
• Fixed issue with Silk browser where it shouldn't define that the category is tablet
• Normalised the name of Samsung Browser in browsers::get()
• Added AndroidDownloadManager to apps list
• Added support for more Kindle Fire tablets.
• Added OS/2 to platform mappings.
• Added some monitoring service crawlers
• Added exact match for Android as well as with a space after it
• Updated workflow to fix run issue
• Added and updated tests.