From: Brett Parker <iDunno@sommitrealweird.co.uk>
Date: Sun, 13 Jan 2008 21:02:24 +0000 (+0000)
Subject: * Begin fixes to list handling code - there's 2 unittests that are failing due
X-Git-Url: https://git.sommitrealweird.co.uk/rss2maildir.git/commitdiff_plain/764a363a914d6799f6d97c2402898fd85210b5e1?ds=inline;hp=5d3c474c51fe355462197e515cbc1d23be4a6cfa

* Begin fixes to list handling code - there's 2 unittests that are failing due
  to this. (Previous revision had 7 unittests fail - bother)
---

diff --git a/rss2maildir.py b/rss2maildir.py
index 113d931..353787c 100755
--- a/rss2maildir.py
+++ b/rss2maildir.py
@@ -70,6 +70,9 @@ class HTML2Text(HTMLParser):
         u'ul',
         u'ol',
         u'dl',
+        u'li',
+        u'dt',
+        u'dd',
         u'div',
         #u'blockquote',
         ]
@@ -219,10 +222,14 @@ class HTML2Text(HTMLParser):
             if self.ignorenodata:
                 newlinerequired = False
             self.ignorenodata = False
-            if newlinerequired \
-                and len(self.text) > 2 \
-                and self.text[-1] != u'\n' \
-                and self.text[-2] != u'\n':
+            if newlinerequired:
+                if tag_thats_done in [u'dt', u'dd', u'li'] \
+                    and len(self.text) > 1 \
+                    and self.text[-1] != u'\n':
+                        self.text = self.text + u'\n'
+                elif len(self.text) > 2 \
+                    and self.text[-1] != u'\n' \
+                    and self.text[-2] != u'\n':
                     self.text = self.text + u'\n\n'
 
         if tag_thats_done in ["h1", "h2", "h3", "h4", "h5", "h6"]: