Commit: ed712d441a3abeb61f516c2d82775e85a118b734 - newsfour (git)

news4 - RSS aggrigation system

Commit MetaInfo

• Révision: ed712d441a3abeb61f516c2d82775e85a118b734 (tree)
• l'heure: 2012-10-04 04:21:36
• Auteur: hylom <hylom@hylo...>
• Commiter: hylom

Message de Log

add auto breaking with p tag in slashdotjp filter

Change Summary

• modified: filters/slashdotjp.py

Modification

--- a/filters/slashdotjp.py
+++ b/filters/slashdotjp.py
@@ -7,6 +7,8 @@ re_read_all = re.compile(ur'''<p>\s*<a href=['"][^'"]+['"]>\s*すべて読む\s*
 re_related = re.compile(ur'''<p>\s*関連ストーリー：.*?</p>''')
 re_topics = re.compile(ur'''<a href="http://slashdot.jp/stories/\w+">(.*?)</a>''')
 
+re_break = re.compile(r'''\n\n(.*?)\n''')
+
 def entry_filter(entry):
     # すべて読む、関連ストーリーを削除
     body = entry['body']
@@ -17,6 +19,9 @@ def entry_filter(entry):
         itr = re_topics.findall(s)
         for items in itr:
             topics.append(items)
+
+    while re_break.search(body):
+        body = re_break.sub(r'</p><p>\1</p><p>', body)
         
     body = re_read_all.sub('', body)
     body = re_related.sub('', body)