[Distutils] [issue29] SVN detection for package indices breaks on Sourceforge.net

[whit morriss]

Nathan R. Yergler wrote:
> New submission from Nathan R. Yergler <nathan at yergler.net>:
> 
> PackageIndex._download_html (setuptools/package_index.py) attempts to detect if
> an HTML page is a Subversion repository using the regular expression
> r'<title>Revision \d+:'.  This breaks for Sourceforge subversion repositories,
> which include the project name before the "Revision" text (see
> http://cctools.svn.sourceforge.net/svnroot/cctools/vendorlibs/utidylib/ for an
> arbitrary example).
> 
> ----------
> messages: 64
> nosy: nyergler
> priority: bug
> status: unread
> title: SVN detection for package indices breaks on Sourceforge.net
> 
> _______________________________________________
> Setuptools tracker <setuptools at bugs.python.org>
> <http://bugs.python.org/setuptools/issue29>
> _______________________________________________
> _______________________________________________
> Distutils-SIG maillist  -  Distutils-SIG at python.org
> http://mail.python.org/mailman/listinfo/distutils-sig
> 
looks like sf upgraded to svn 1.5?

this issue actually breaks setuptools for anyone using the default xslt 
template for svn 1.5 (though that template is inadequate for fixing it)


r'<title>\s*[a-zA-Z]+\s*\-?\s*Revision \d+:'

will match both old and new:

<title>almanac - Revision 1179: /
<title>Revision 19507: /

seems like there ought to be a better way to do this than sniffing title 
tags though... especially since svn offers easy ways to break this via 
the svnindex.xsl template.

-w