quentin389 / useragentinfo Goto Github PK

This external dependency is missing

Update performance tests to have 4 versions:

bulk + no opcache
non bulk + no opcache
bukl + opcache
no bulk + opcache

Hey thanks for the email about this library.

Currently I need a portable solution. That means not needing to use php.ini to point to browsecap. Would it be possible to change browsecap to something other than an ini file or parse it independent from php.ini. This could allow other languages to get involved. Furthermore are you going to add this composer?

User agent Mozilla/4.0 (compatible; MSIE 5.0; Windows NT; DigExt) is only partially overriden from the local array and shows isBanned and Crawler as true, but still identifies as Internet Explorer instead of E-mail address crawler. That's because the override array is treated as part of browscap.
However, I should create an array in the main class which overrides EVERYTHING. That won't be always useful, but will allow for correction of specific cases, like the above one.

Add more classes that implement UaiCacheInterface. I should at least add:

• memcache
• memcached
• APC

Several times now I needed to check one user agent using UserAgentInfo, and each time I had to modify my application to do that. Of course it's extremely simple, but still - it would be nicer if I had a website like user-agent-string.info that shows the UA results.

I've been needing a nice icon pack with browsers and devices in addition to the parser.
So far, haven't found anything good apart from: https://github.com/paulirish/browser-logos

Maybe we could include that into the master or a separate branch?

Try to optimize parsing and including speed based on xhprof profiler results.

Test if uaparser leverages opcache when loading the source JSON file. I'm not sure how opcaches work - will they cache such file? If not, then moving the source file to PHP will be a huuuuge performance boost for non bulk + opcache situation. Which is the most common scenarion on high traffic production servers.

If such switch should be made - create a commit to uaparser project.

There are notes in the readme mentioning weekly updates. It looks like this has not been updated in some time. Is there still activity on this?

Thank you.

I have been using this library at v1.3 for some time and I see it has not been updated in a long time. Is this project more or less dead? If so are there any forks still being updated or other projects being maintained that offer similar functionality?

One major sticking point for me is this library does not support composer which it badly needs to do moving forward (see #4).

In theory this information should not be available, because that defies the purpose of such compatibility mode. However, I can't predict all situations in which this might or might not be useful, so why not include it?
Of course a property disclaimer should be added.

http://stackoverflow.com/a/17942323

Some user agents I override from local class arrays are too specific or too vague for my liking.
For example, I identify Mozilla/5.0 (Windows; U; Windows NT 6.0; en-GB; rv:1.0; trendictionbot0.5.0; trendiction search; http://www.trendiction.de/bot; please let us know of any problems; web at trendiction.com) Gecko/20071127 Firefox/3.0.0.11 as trendiction bot... which is ridiculously over-specific.

Or not...

I have to see if browscap results for that make any sense. If they do then I could add it.

The get_browser() function is cited but there are no information why to use a class instead a build-in function. Is interesting to inform user about performance, and about strategies — eg. a fast detection and log with get_browser(), and later using UserAgentInfo for report analyzing log or building a detailed database.

The caching mechanism has a built-in version check for all the source parser files.
Browscap version is checked using filesize(ini_get('browscap')), as it is a good enough approximation of the file contents.

The problem is that when you're running a multi server configuration and each server uses its own browscap file, not the one supplied with this class (available in imports/browcap.ini), the files may differ on each server. If that happens and you have a cache that is common for all the servers then the cached UserAgentInfo objects will constantly be invalidated and regenerated, slowing down the servers, as parsing ua strings is a costly operation.

This is not a bug, the class actually behaves correctly, because the parsing results using different browscap version will differ. However, it is an issue, because it's not obvious that this can happen.
The proper way to fix it, without changing anything in UserAgentInfo project is to force your admins to keep the browscap files updated on all the servers, or to point the servers to UserAgentInfo internal browscap.ini file.

However, since that requires changes on the servers, it may not always be an option, it may take long time to change, and so on and so forth.

This will stop being an issue when I stop relying on browscap php.ini configuration and the PHP module for browscap. Using this module is not a good idea for other reasons (such as: required server configuration and parsing speed), so this issue is another important reason to start using browscap fully in PHP, as it is shown in https://github.com/GaretJax/phpbrowscap/

This will accept a file from modified https://github.com/GaretJax/phpbrowscap and use its own parser to parse data in PHP (parser based on Garet Jax solution).

Browscap removed ->is_banned property which I don't agree with, and I'm actually using this property in my work. It's active for few browser I've added but all the data researched by Gary when he had this property in his browscap project is gone from the files.
I've obtained the latest browscap file version where this information is present and I'm going to extract information from there and put it into nice array to be added to the rest of browscap data when parsing.
I've added raised an issue with browscap team asking them to restore the property.

The list has to be verified before using.