From dc0bbee020a72a0ad1f868527c7af6a4cb06184e Mon Sep 17 00:00:00 2001 From: Yorik van Havre Date: Tue, 8 Apr 2014 16:09:37 -0300 Subject: [PATCH] More cleaning in offline doc scripts --- src/Tools/offlinedoc/downloadwiki.py | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/src/Tools/offlinedoc/downloadwiki.py b/src/Tools/offlinedoc/downloadwiki.py index c8e5f60fa..0c97387c7 100755 --- a/src/Tools/offlinedoc/downloadwiki.py +++ b/src/Tools/offlinedoc/downloadwiki.py @@ -183,15 +183,15 @@ def getimagelinks(html): def cleanhtml(html): "cleans given html code from dirty script stuff" html = html.replace('\n','Wlinebreak') # removing linebreaks for regex processing - html = re.compile('(.*)]+column-content+[^>]+>').sub('',html) # stripping before content - html = re.compile(']+column-one+[^>]+>.*').sub('',html) # stripping after content + html = re.compile('(.*)
]+>').sub('',html) # stripping before content + html = re.compile('
]*>.*?
').sub('',html) # removing nav div html = re.compile('

]*>.*?

').sub('',html) # removing print subtitle html = re.compile('Retrieved from').sub('Online version:',html) # changing online title - html = re.compile('
').sub('',html) # removing catlinks html = re.compile('
').sub('',html) # removing nav stuff html = re.compile('
" header += "" header += "" + header += "

" + title + "

" footer = "" html = header+html+footer filename = local(page.replace("/","-"))