[med-svn] r2404 - trunk/community/talks/200808_debconf8

[tille at alioth.debian.org]

Author: tille
Date: 2008-08-11 00:06:25 +0000 (Mon, 11 Aug 2008)
New Revision: 2404

Modified:
   trunk/community/talks/200808_debconf8/get-archive-pages
Log:
Further work on Alioth lists, not working yet


Modified: trunk/community/talks/200808_debconf8/get-archive-pages
===================================================================
--- trunk/community/talks/200808_debconf8/get-archive-pages	2008-08-10 23:36:05 UTC (rev 2403)
+++ trunk/community/talks/200808_debconf8/get-archive-pages	2008-08-11 00:06:25 UTC (rev 2404)
@@ -37,7 +37,8 @@
 my @MONTHES  = ('01', '02', '03', '04', '05', '06', '07', '08', '09', '10', '11', '12');
 my @ROBOTS   = ('Debian Installer', 'bugzilla-skolelinux', 'Archive Administrator', 'hostmaster',
                 'Debian-med-request', 'Debian testing watch', 'Debian Bug Tracking System',
-                'Skolelinux archive Installer', 'Debian Wiki', 'gentoo-\w+\+help');
+                'Skolelinux archive Installer', 'Debian Wiki', 'gentoo-\w+\+help',
+                'Debichem-commits');
 
 ## TODO: just consider mails containing these strings as SPAM
 ##       This has to be implemented in the code below
@@ -143,11 +144,13 @@
 		    foreach $content (@tmpdata) {
 			@lines = split(/(\n)/, $content);
 			foreach $line (@lines) {
-			    if ( $line =~ /^\s*$/   || $line =~ /^<!--\d+ / ||
-				 $line =~ /^<\/I>$/ || $line =~ /^<UL>$/    ||
+			    if ( $line =~ /^\s*$/   || $line =~ /^<!--\d+ /        ||
+				 $line =~ /^<\/I>$/ || $line =~ /^\s*<\/?UL>\s*$/i ||
 				 $line =~ /^<\/A><A NAME="\d+">&nbsp;<\/A>$/ ) { next ; }
-			    if ( $line =~ /^<LI><A HREF="\d+.html">\[[-\w]+\]/ ) {
-				@data = (@data, $line) ;
+			    if ( ($subject) = $line =~ /^\s*<LI><A HREF="\d+.html">\[[-\w]+\]\s*(.+)$/ ) {
+				$_ = $subject ;
+				$_ =~ s/^Re:\s*//i ;       # Remove Re:
+				@data = (@data, $subject) ;
 			    } else {
 				@data = (@data, "$line\n" ) ;
 			    }