improved url matching, and texpdf tolerance and indentation levels set

1 files changed, 3 insertions, 14 deletions

diff --git a/lib/sisu/v0/shared_html_lite.rb b/lib/sisu/v0/shared_html_lite.rb
index ed622fb8..5285b6db 100644
--- a/lib/sisu/v0/shared_html_lite.rb
+++ b/lib/sisu/v0/shared_html_lite.rb
@@ -120,20 +120,9 @@ module SiSU_Format_Shared
         words=word_mode.join(' ')
         para.gsub!(/.+/,words)
       end
-      if (para !~/(\"\w+:\/\/\S+?\"|>\s*\w+:\/\/\w+?\S*<)/) #url markup http etc.
-        if para=~/\w+:\/\/\S+?\.\S+?[.,] /
-          para.gsub!(/(\w+:\/\/\S+?\.\S+?)([.,] )/,%{#{@url_brace.xml_open}<a href="\\1" target="_top">\\1</a>#{@url_brace.xml_close}\\2}) #full stops ! have been a bother
-        else
-          para.gsub!(/(\w+:\/\/\S+?\.\S+)/,%{#{@url_brace.xml_open}<a href="\\1" target="_top">\\1</a>#{@url_brace.xml_close}})
-        end
-      end
-      if (para =~/[ ^](?:https?|ftp):\/\/\S+/) #url markup http leftovers watch carefully may accept too much 2004w46
-        if para=~/([ ^])((?:https?|ftp):\/\/\S+?)([\.,] )/
-        para.gsub!(/([ ^])((?:https?|ftp):\/\/\S+?)([\.,] )/,%{\\1#{@url_brace.xml_open}<a href="\\2" target="_top">\\2</a>#{@url_brace.xml_close}\\3}) #full stops ! have been a bother
-        else
-          para.gsub!(/([ ^])((?:https?|ftp):\/\/\S+)/,%{\\1#{@url_brace.xml_open}<a href="\\2" target="_top">\\2</a>#{@url_brace.xml_close}})
-        end
-      end
+      para.gsub!(/(^|\s)[_\\]((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1<a href="\2" target="_top">\2</a>\3') #http ftp matches escaped, no decoration
+      para.gsub!(/((?:^|\s)[}])((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,'\1<a href="\2" target="_top">\2</a>\3') #special case \{ e.g. \}http://url
+      para.gsub!(/(^|\s)((?:https?|ftp):\/\/\S+?\.[^'"><\s]+?)([.,]?(?:\s|$))/,%{\\1#{@url_brace.xml_open}<a href="\\2" target="_top">\\2</a>#{@url_brace.xml_close}\\3}) #http ftp matches with decoration
       para
     end
     def paragraph