patches for UserAgent/*

[Joost van Baal]

Hi Egon,

On Sun, Aug 15, 2004 at 01:53:03AM +0200, Egon Willighagen wrote:
> 
> attached are three patches for Robot.pm, Browser.pm and OS.pm.
> 
> My local www report had to many "Unknown" in the reports so I did some 
> research and hacked a bit in my local Lire 1.5 installation...
> 
> On Browser.pm:
> 
> It did not recognize recent browsers, like Firefox, Epiphany or Safari...
> Nor did it recognize Java VM's that can retrieve over HTTP... The test
> for 'gecko' is no longer valid marker for Mozilla browsers, as there are many 
> browsers based on the Gecko engine, *like* Mozilla... but that does not 
> make them Mozilla...

Did you sent this patch to the AWStats maintainers too?

> On OS.pm:
> 
> After some heavy debugging, I found that the keys method for hashes does not 
> return them in order in which entered in the hashtable... and it turned out 
> that x11 ("Unknown X11 System") was tested before things like "linux" :(
> Therefor, I separate these more generic types into a separate hash.

This one didn't apply cleanly to the version in CVS: $Id: OS.pm,v 1.1
2004/05/24 15:24:33 flacoste Exp $ .  After some tweaking it applied.

> On Robot.pm:
> 
> Robots which user agent string started with "Mozilla/x.0_(compatible" were 
> never recognized as assumed to always be browsers, instead of robots...
> I a local log I found two exceptions: "Yahoo! Slurp" and "Grub.org" which
> is a distributed crawler... Because of these two (and likely more) exceptions,
> I made a new hash for these kind of robots that mimick a true browser...
> Furthermore, I added 4 robots found in this local log, which were not 
> available in the perl module...

Did you sent this to the http://www.robotstxt.org/wc/active.html
maintainer, Martijn Koster?

> I have not checked wether I still have CVS write access, but otherwise, 
> consider these patches GPL (that's the Lire license, correct?) and I hereby 
> transfer any copyright on these patches to the foundation.

OK.

> PS. the new HTML output is not in 1.5, is it?

I believe not, it's planned for Lire 2.0.  See
http://logreport.org/lire/roadmap-2.0.php .  It is in CVS though, IIRC.

Thanks a _lot_ for the patches, I've just applied them.  They'll get
distributed with Lire 2.0 which is expected to get shipped within a
couple of weeks.

Thanks again, Bye,

Joost

-- 
.    .                                        http://logreport.com/
| '.|                        /^LogReport$/
| Lire                                        http://logreport.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.logreport.org/pipermail/development/attachments/20040818/6c12d572/attachment.bin