Strip multiple newlines to one in the post before parsing with bbcode

boards/mdx_neboard.py

@@ -1,152 +1,164
 # coding=utf-8
 
 import re
 import bbcode
 
 import boards
 
 
 __author__ = 'neko259'
 
 
 REFLINK_PATTERN = re.compile(r'\d+')
+MULTI_NEWLINES_PATTERN = re.compile(r'(\r?\n){2,}')
+ONE_NEWLINE = '\n'
 
 
 class TextFormatter():
     """
     An interface for formatter that can be used in the text format panel
     """
 
     def __init__(self):
         pass
 
     name = ''
 
     # Left and right tags for the button preview
     preview_left = ''
     preview_right = ''
 
     # Left and right characters for the textarea input
     format_left = ''
     format_right = ''
 
 
 class AutolinkPattern():
     def handleMatch(self, m):
         link_element = etree.Element('a')
         href = m.group(2)
         link_element.set('href', href)
         link_element.text = href
 
         return link_element
 
 
 class QuotePattern(TextFormatter):
     name = 'q'
     preview_left = '<span class="multiquote">'
     preview_right = '</span>'
 
     format_left = '[quote]'
     format_right = '[/quote]'
 
 
 class SpoilerPattern(TextFormatter):
     name = 'spoiler'
     preview_left = '<span class="spoiler">'
     preview_right = '</span>'
 
     format_left = '[spoiler]'
     format_right = '[/spoiler]'
 
     def handleMatch(self, m):
         quote_element = etree.Element('span')
         quote_element.set('class', 'spoiler')
         quote_element.text = m.group(2)
 
         return quote_element
 
 
 class CommentPattern(TextFormatter):
     name = ''
     preview_left = '<span class="comment">// '
     preview_right = '</span>'
 
     format_left = '[comment]'
     format_right = '[/comment]'
 
 
 class StrikeThroughPattern(TextFormatter):
     name = 's'
     preview_left = '<span class="strikethrough">'
     preview_right = '</span>'
 
     format_left = '[s]'
     format_right = '[/s]'
 
 
 class ItalicPattern(TextFormatter):
     name = 'i'
     preview_left = '<i>'
     preview_right = '</i>'
 
     format_left = '[i]'
     format_right = '[/i]'
 
 
 class BoldPattern(TextFormatter):
     name = 'b'
     preview_left = '<b>'
     preview_right = '</b>'
 
     format_left = '[b]'
     format_right = '[/b]'
 
 
 class CodePattern(TextFormatter):
     name = 'code'
     preview_left = '<code>'
     preview_right = '</code>'
 
     format_left = '[code]'
     format_right = '[/code]'
 
 
 def render_reflink(tag_name, value, options, parent, context):
     if not REFLINK_PATTERN.match(value):
         return u'>>%s' % value
 
     post_id = int(value)
 
     posts = boards.models.Post.objects.filter(id=post_id)
     if posts.exists():
         post = posts[0]
 
         return u'<a href=%s>&gt;&gt;%s</a>' % (post.get_url(), post_id)
     else:
         return u'>>%s' % value
 
 
+def preparse_text(text):
+    """
+    Performs manual parsing before the bbcode parser is used.
+    """
+
+    return MULTI_NEWLINES_PATTERN.sub(ONE_NEWLINE, text)
+
+
 def bbcode_extended(markup):
     parser = bbcode.Parser()
     parser.add_formatter('post', render_reflink, strip=True)
     parser.add_simple_formatter('quote',
             u'<span class="multiquote">%(value)s</span>', strip=True)
     parser.add_simple_formatter('comment',
             u'<span class="comment">//%(value)s</span>')
     parser.add_simple_formatter('spoiler',
             u'<span class="spoiler">%(value)s</span>')
     parser.add_simple_formatter('s',
             u'<span class="strikethrough">%(value)s</span>')
     parser.add_simple_formatter('code',
             u'<pre><code>%(value)s</pre></code>')
-    return parser.format(markup)
+
+    text = preparse_text(markup)
+    return parser.format(text)
 
 formatters = [
     QuotePattern,
     SpoilerPattern,
     ItalicPattern,
     BoldPattern,
     CommentPattern,
     StrikeThroughPattern,
     CodePattern,
 ]