"""HTML Parsers
This file is a part of Bookmarks database and Internet robot.
+
"""
-__version__ = "$Revision$"[11:-2]
-__revision__ = "$Id$"[5:-2]
-__date__ = "$Date$"[7:-2]
__author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 1997-2011 PhiloSoft Design"
+__copyright__ = "Copyright (C) 1997-2014 PhiloSoft Design"
__license__ = "GNU GPL"
-__all__ = ['parse_html', 'universal_charset']
+__all__ = ['parse_html', 'parse_filename', 'universal_charset']
import codecs
import os
BKMK_DEBUG_HTML_PARSERS = os.environ.get("BKMK_DEBUG_HTML_PARSERS")
-def parse_html(filename, charset=None, log=None):
+def parse_html(html_text, charset=None, log=None):
if not parsers:
return None
parser = None
for c in charsets:
try:
- parser = p(filename, c, log)
+ parser = p(html_text, c, log)
except UnicodeError:
pass
else:
if parser.charset:
parser.icon = icon.encode(parser.charset)
return parser
+
+def parse_filename(filename, charset=None, log=None):
+ fp = open(filename, 'r')
+ try:
+ parser = parse_html(fp.read(), charset=charset, log=log)
+ finally:
+ fp.close()
+ return parser