Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-07 10:14:46 -04:00
Code Issues Packages Projects Releases Wiki Activity

MOBI Input: Fix regression that broke reading of some documents

Browse Source 
Apparently lxml.html is super fragile on Windows with python 3, so
fallback to html5-parser when it barfs.
This commit is contained in:
Kovid Goyal 2020-10-05 16:00:54 +05:30
parent 02095fcf81
commit 56a9b9529a
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C
1 changed files with 9 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/calibre/ebooks/mobi/reader/mobi6.py
View File

@ -192,7 +192,15 @@ class MobiReader(object):
except Exception:
self.log.warning('MOBI markup appears to contain random bytes. Stripping.')
self.processed_html = self.remove_random_bytes(self.processed_html)
try:
root = html.fromstring(self.processed_html)
except Exception:
self.log.warning('MOBI markup could not be parsed by lxml using html5-parser')
# Happens on windows with python 3 where lxml causes libxml to die with an
# error about using UCS-4 little endian encoding if certain
# characters are present in the input
from html5_parser import parse
root = parse(self.processed_html, keep_doctype=False, namespace_elements=False, maybe_xhtml=False, sanitize_names=True)
if root.xpath('descendant::p/descendant::p'):
from html5_parser import parse
self.log.warning('Malformed markup, parsing using html5-parser')