From fdd1489c82a274615e46e3c67fc5707e3fb0465f Mon Sep 17 00:00:00 2001
From: Ralph Amissah <ralph@amissah.com>
Date: Mon, 9 Jul 2007 11:19:48 +0100
Subject: improved url matching, and texpdf tolerance and indentation levels
 set

---
 lib/sisu/v0/shared_xml.rb | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

(limited to 'lib/sisu/v0/shared_xml.rb')

diff --git a/lib/sisu/v0/shared_xml.rb b/lib/sisu/v0/shared_xml.rb
index bcdf44f5..995044db 100644
--- a/lib/sisu/v0/shared_xml.rb
+++ b/lib/sisu/v0/shared_xml.rb
@@ -347,8 +347,7 @@ module SiSU_XML_munge
     def markup(para='')
       wordlist=para.scan(/\S+|\n/) #\n needed for tables, check though added 2005w17
       para=tidywords(wordlist).join(' ').strip
-      para.gsub!(/(^|\s+)<\s+/,'\1&lt; ')
-      para.gsub!(/\s+>(\s+|$)/,' &gt;\1')
+      para.gsub!(/(^|\s+)<\s+/,'\1&lt; '); para.gsub!(/\s+>(\s+|$)/,' &gt;\1')
       para.gsub!(/<:pb>\s*/,'')
       para.gsub!(/<+[-~]#>+/,'')
       para.gsub!(/<0;\w\d+;[um]\d+><#@dp:#@dp>/,'')
@@ -357,8 +356,12 @@ module SiSU_XML_munge
       #para.gsub!(/^_\*\s+/,'<image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="/usr/share/sisu/image/bullet_red.png" width="12" height="12" alt="*" /> ')
       para.gsub!(/(^|\s)\{\s*(\S+?\.(?:jpg|png|gif))\s+(\d+)x(\d+)(\s+[^}]+)?\}(https?:\/\/\S+)/,%{\\1<image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="#{@dir.url.images_local}/\\2" width="\\3" height="\\4" />[\\2] \\5})
       para.gsub!(/(^|\s)\{\s*(\S+?\.(?:jpg|png|gif))(\s+[^}]+)?\}(https?:\/\/\S+)/,%{\\1<image xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:actuate="onLoad" xlink:show="embed" xlink:href="#{@dir.url.images_local}/\\2"/>\\2})
-      para.gsub!(/(^|\s)\{([^}]+)\}(https?:\/\/[^"><]+?)([,.:;"><]?(?:\s|$))/,'\1<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\3">\2</link>\4')
-      para.gsub!(/(^|\s)(https?:\/\/[^"><]+?)([,.:;"><]?(?:\s|$))/,%{\\1#{@url_brace.xml_open}<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\\2">\\2</link>#{@url_brace.xml_close}\\3})
+      para.gsub!(/(^|\s)\{([^}]+)\}(https?:\/\/[^"><]+?)([,.:;"><]?(?:\s|$))/,
+        '\1<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\3">\2</link>\4') #watch, compare html_tune
+      para.gsub!(/(^|\s)((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,
+        %{\\1#{@url_brace.xml_open}<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\\2">\\2</link>#{@url_brace.xml_close}\\3})
+      para.gsub!(/(^|\s)[_\\]((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1\2\3') #escaped urls not linked, deal with later
+      #para.gsub!(/(^|\s)[_\\]((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1<link xmlns:xlink="http://www.w3.org/1999/xlink" xlink:type="simple" xlink:href="\2">\2</link>\3') #escaped urls not linked, deal with later
       para.gsub!(/&nbsp;/,' ') #clean
       para
     end
-- 
cgit v1.2.3