The user-agent header

As mentioned in our overview of HTTP requeset headers, the user-agent header specifies information about the browser or robot that made the request. Here are some examples of user-agent strings:


1. Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.1; .NET CLR 1.1.4322; .NET CLR 1.0.3705)
2. Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en) AppleWebKit/418.9 (KHTML, like Gecko) Safari/419.3
3. Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9) Gecko/2008052906 Firefox/3.0
4. Opera/9.27 (Windows NT 5.1; U; en)
5. Opera/9.50 (J2ME/MIDP; Opera Mini/4.1.11320/534; U; en)
6. BlackBerry8100/4.2.1 Profile/MIDP-2.0 Configuration/CLDC-1.1 VendorID/125
7. T-Mobile Dash Mozilla/4.0 (compatible; MSIE 4.01; Windows CE; Smartphone; 320x240)
8. msnbot/1.1 (+http://search.msn.com/msnbot.htm)
9. libwww-perl/5.800

As you can see, there is some variety in the format of user-agent headers. However, the following generally apply:

Browser, client and sometimes module names can be suffixed with a slash followed by the version number, e.g. Opera/9.27, Safari/419.3;
In general, we can take the first substring outside brackets of the form xxxx/yyyy as being the client and version except in the case of Mozilla/yyyy, which many agents insert anyway;
Examples like [2] are a problem, however: we need to know in advance that "Safari" is the name of the browser and not "AppleWebKit";
Robots may insert + and a URL that is intended to be the "home page" of the robot, giving information about its purpose etc; genuine robots generally provide this;
Elements are generally semicolon separated;
There's no strict interpretation we can make of the brackets (compare [6] and [7] above, or [2];
There's no easy way to tell a plugin/module name from an operating system name– we just have to know some common cases to help us make a guess (see the list below).

Common client strings in the user-agent header

For reference, the following commonly occur in the non-bracketed section of the user-agent string. I've indicated the approximate frequency of given agents in requests to my server (taken over a sample of requests from approximately 12,000 unique clients). Obviously, these figures could differ according to the target audience of your site.

Agent name	% occurrence	Type	Explanation
Mozilla	99.5 %	Compatibility	Essentially meaningless nowadays. Originally, this agent name denoted either a version of Netscape or the version of Netscape with which another agent wanted to claim compatibility. Nowadays, even the most trivial home-made bot tends to include `Mozilla/4.0` or `Mozilla/5.0` at the start of the user-agent string.
Gecko	24.8 %	Compatibility	Gecko is the rendering engine used by Firefox and a handful of other Mozilla spin-offs such as SeaMonkey. In principle, you could take notice of this field to alter page contents to work around some rendering bug, or offer enhanced content if you were aware of some feature that Gecko rendered better than other engines. In practice, I suspect practically all servers ignore this field. The Safari web browser amusingly includes the string `(KHTML, like Gecko)` in its user-agent string.
Firefox	24.4 %	Browser	Identifies the Firefox web browser.
Safari	7.0 %	Browser	Apple's Mac OS browser. More information on the user-agent string sent by Safari is given in Apple's Safari FAQ.
AppleWebKit	7.0 %	Compatibility	What Apple describe as their "web technology framework", used by the Safari browser.
Version	6.1 %	Compatibility	Used by version 3.0 or greater of the Apple's Safari browser to indicate the "Safari family version number" (see entry for Safari above).
msnbot	< 1 %	Robot	Microsoft's search bot for their MSN Live Search engine.
Profile	< 1 %	Compatibility	Sent by some Java-copmpatible mobile devices to indicate which Java Mobile Information Device Profile (MIDP) they support. A typical specification would be `Profile/MIDP-2.0`. BlackBerries and Nokia and Sony Ericcson mobile phones (possibly among others) send this tag.
Configuration	< 1 %	Compatibility	Sent by some Java-copmpatible mobile devices to indicate which version of the Java Connected Limited Device Configuration they support.
BlackBerryXXXX	< 1 %	OS / Browser	BlackBerry devices appear to send as their browser/OS identifier a string that includes the model number (e.g. a BlackBerry 8310 with OS version 4.3 would send `BlackBerry8310/4.3.0`).

If you enjoy this Java programming article, please share with friends and colleagues. Follow the author on Twitter for the latest news and rants. Follow @BitterCoffey