From 9c1f340ee6f0c4b76985ab080f994cd5e7cb97d0 Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Wed, 10 Sep 2008 13:45:02 -0700 Subject: [PATCH] IGN:Workaround invalid tag nesting in mobipocket HTML --- src/calibre/ebooks/mobi/reader.py | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/src/calibre/ebooks/mobi/reader.py b/src/calibre/ebooks/mobi/reader.py index fccb4f68b4..334b34e8d3 100644 --- a/src/calibre/ebooks/mobi/reader.py +++ b/src/calibre/ebooks/mobi/reader.py @@ -181,10 +181,12 @@ class MobiReader(object): '\n', self.processed_html) - soup = BeautifulSoup(self.processed_html.replace('> <', '>\n<')) + soup = BeautifulSoup(self.processed_html) self.cleanup_soup(soup) guide = soup.find('guide') for elem in soup.findAll(['metadata', 'guide']): @@ -210,6 +212,11 @@ class MobiReader(object): self.processed_html = re.sub(r'
', '', self.processed_html) if self.book_header.ancient and '')+'' + self.processed_html = self.processed_html.replace('> <', '>\n<') + self.processed_html = self.processed_html.replace('', '') + self.processed_html = self.processed_html.replace('', '') + self.processed_html = self.processed_html.replace('', '') + self.processed_html = self.processed_html.replace('', '') def cleanup_soup(self, soup): for tag in soup.recursiveChildGenerator():