[med-svn] r2406 - trunk/community/talks/200808_debconf8

[tille at alioth.debian.org]

Author: tille
Date: 2008-08-11 10:29:10 +0000 (Mon, 11 Aug 2008)
New Revision: 2406

Modified:
   trunk/community/talks/200808_debconf8/get-archive-pages
Log:
Get subject and author of alioth index in one line, not finished yet


Modified: trunk/community/talks/200808_debconf8/get-archive-pages
===================================================================
--- trunk/community/talks/200808_debconf8/get-archive-pages	2008-08-11 04:29:19 UTC (rev 2405)
+++ trunk/community/talks/200808_debconf8/get-archive-pages	2008-08-11 10:29:10 UTC (rev 2406)
@@ -143,16 +143,22 @@
 		    @data = ();
 		    foreach $content (@tmpdata) {
 			@lines = split(/(\n)/, $content);
-			foreach $line (@lines) {
-			    if ( $line =~ /^\s*$/   || $line =~ /^<!--\d+ /        ||
-				 $line =~ /^<\/I>$/ || $line =~ /^\s*<\/?UL>\s*$/i ||
-				 $line =~ /^<\/A><A NAME="\d+">&nbsp;<\/A>$/ ) { next ; }
-			    if ( ($subject) = $line =~ /^\s*<LI><A HREF="\d+.html">\[[-\w]+\]\s*(.+)$/ ) {
+			foreach (@lines) {
+			    s/\s+/ /g;
+			    if ( $_ =~ /^\s*$/   || $_ =~ /^<!--\d+ /      ||
+				 $_ =~ /^<\/I>$/ || $_ =~ /^\s*<\/?p>\s*$/ ||
+                                 $_ =~ /^\s*<\/?UL>\s*$/i ||
+				 $_ =~ /^<\/A><A NAME="\d+">&nbsp;<\/A>$/ ) { next ; }
+			    if ( ($subject) = $_ =~ /^\s*<LI><A HREF="\d+.html">\[[-\w]+\]\s*(.+)$/ ) {
 				$_ = $subject ;
-				$_ =~ s/^Re:\s*//i ;       # Remove Re:
+				$_ =~ s/^\s*Re:\s*//i ;       # Remove Re:
 				@data = (@data, $subject) ;
 			    } else {
-				@data = (@data, "$line\n" ) ;
+				if ( $_ =~ /<I>/ || $_ =~ /<b>Messages:<\/b>/ ) {
+				    @data = (@data, "$_\n" ) ;
+				} else {
+				    @data = (@data, "$_" ) ;
+				}
 			    }
 			}
 		    }