I've been tweaking my website statistics scripts a bit, and one of the things I've been working on is user agent parsing.
(I'm not sure whether to classify this as client-side or server-side. The information I am interested in comes from the client, but I'm logging and analyzing it server side. Let me know if there's a different place where I should post it.)
I don't do sniffing, because I refuse to maintain a bunch of different versions of the site; but I do very much like to be able to identify the agents that are being used on the site for other reasons, such as planning. For example, I got serious about handling preposterously small screens in a reasonable way a couple of years ago when I saw the number of phones being used to access the site go from negligible to significant in the space of a few months -- I guess one of the cellphone companies around here started pushing smartphones or stopped giving away non-web-enabled phones with the purchase of a contract or something.
I also like to be able to identify bots as bots when possible, mainly so I can exclude them from certain statistics.
Anyway, there are some UA string tokens that I just can't seem to identify what they mean or what generates them, and I'm wondering if anyone else knows anything about them. (If anyone is interested, I'd be happy to post some of the non-trivial ones I have identified.)
Here are some of the biggies I'd like to identify:
I would really like to know this one, as it occurs in 10-15% of the UAs we see, clearly enough that I ought to be paying attention to it, if only I know what I should pay attention to. These UAs generally look like IE. It may be some kind of BHO, but what?
Also occurs in UAs that look like IE.
Also in UAs that look like IE, but almost always right before the last closing parenthesis.
Looks like IE. NP08 disappeared 2013 June, but NP06 is still active.
I suspect that these are OEM IDs, but I don't know which OEMs or whether they have other implications.
Looks like IE. Probably related to HPDTDF, HPNTDF, CMNTDF (see below) but this one is still active.
Could be related to BOIE (Bing-branded IE) and yIE (Yahoo-branded IE), both of which also use all-caps and no hyphen for the l10n info and follow it up with other letters (but not these other letters, WOL). Is there a version of IE branded for a search engine or portal site that starts with E?
Here are some that technically don't matter anymore, but I'm still curious:
These used to occur very frequently (also in UAs that otherwise looked like IE) but suddenly disappeared in May of 2012. Some kind of bot?
Disappeared 2013 March.
Looks like IE. Disappeared 2012 July.
Looks like IE. Disappeared 2013 February.
Some ones I did manage to identify:
These are models in the Acer Iconia series of tablets.
These are various models in the Kindle Fire line.
These are Coby Kyros tablets.
These are various models of Optimus-series smartphones.
BRI Mobile, an add-on that has something to do with online banking.
Realtime Communications library, used e.g. by LiveMeeting. Not the same as WebRTC.
nnnnnn appears to be some kind of build identifier (a hexadecimal number, if I'm looking at it right). The brackets contain a semicolon-delimited list of key/value pairs. The keys (and some of the values) are four-character codes starting with FB, which is the clue. This is Facebook's mobile app. I haven't deciphered the meanings of all the keys yet, but some of them are clear enough (e.g., FBDV indicates the device the thing is running on, and FBCR is the "carrier" (cell service provider). SN and SV appear to be the host OS and a version number thereto pertaining, respectively. So far I've only seen this thing running on iOS. It appears to be Webkit-based, but not having an iOS device to test on I can't prove that.
There are more where these came from, but I'll start with that many for now.
Does anybody have a tangible lead on any of the ones I couldn't pin down? Is this an interesting topic to anyone besides me? Am I even in the right forum?