"""HTML Parser using lxml.html
This file is a part of Bookmarks database and Internet robot.
+
"""
__author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 2010-2012 PhiloSoft Design"
+__copyright__ = "Copyright (C) 2010-2014 PhiloSoft Design"
__license__ = "GNU GPL"
__all__ = ['parse_html']
-from lxml.html import parse
+from lxml.html import fromtring
from .bkmk_ph_util import HTMLParser
-def parse_html(filename, charset=None, log=None):
- html_tree = parse(filename)
+def parse_html(html_text, charset=None, log=None):
+ html_tree = fromtring(html_text)
if html_tree.getroot() is None:
return None
title = html_tree.findtext('head/title')
if title is None:
title = html_tree.findtext('title')
- if title is None:
- return None
meta = html_tree.findall('head/meta')
for m in meta:
break
except IndexError:
meta_charset = False
+ elif m.get('charset', ''):
+ meta_charset = m.get('charset').lower()
+ break
else:
meta_charset = False
- if charset or meta_charset:
+ if title and (charset or meta_charset):
title = title.encode(charset or meta_charset)
for m in meta:
else:
icon = None
+ if (title is None) and (refresh is None) and (icon is None):
+ return None
return HTMLParser(charset, meta_charset, title, refresh, icon)