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) 1997-2012 PhiloSoft Design"
__license__ = "GNU GPL"
__all__ = ['parse_html', 'universal_charset']
parsers.append(bkmk_ph_beautifulsoup.parse_html)
try:
- from . import bkmk_ph_lxml
+ from . import bkmk_ph_html5
except ImportError:
pass
else:
- parsers.append(bkmk_ph_lxml.parse_html)
+ parsers.append(bkmk_ph_html5.parse_html)
try:
- from . import bkmk_ph_htmlparser
+ from . import bkmk_ph_lxml
except ImportError:
pass
else:
- parsers.append(bkmk_ph_htmlparser.parse_html)
+ parsers.append(bkmk_ph_lxml.parse_html)
try:
- from . import bkmk_ph_html5
+ from . import bkmk_ph_htmlparser
except ImportError:
pass
else:
- parsers.append(bkmk_ph_html5.parse_html)
+ parsers.append(bkmk_ph_htmlparser.parse_html)
# ElementTidy often segfaults
#try:
return ''.join(output)
+import os
+BKMK_DEBUG_HTML_PARSERS = os.environ.get("BKMK_DEBUG_HTML_PARSERS")
+
def parse_html(filename, charset=None, log=None):
if not parsers:
return None
charsets.remove(charset)
charsets.insert(0, charset)
+ if BKMK_DEBUG_HTML_PARSERS:
+ _parsers = []
for p in parsers:
parser = None
for c in charsets:
try:
parser = p(filename, c, log)
- except UnicodeEncodeError:
+ except UnicodeError:
pass
else:
- break
- if parser:
- break
+ if parser:
+ if BKMK_DEBUG_HTML_PARSERS:
+ if log: log(" Parser %s: ok" % p.__module__)
+ _parsers.append((p, parser))
+ break
else:
- if log: log("Parser %s.%s failed, trying next one." % (p.__module__, p.__name__))
+ if log: log(" Parser %s: fail" % p.__module__)
+ if not BKMK_DEBUG_HTML_PARSERS and parser:
+ break
- if not parser:
- if log: log("All parser has failed.")
+ if BKMK_DEBUG_HTML_PARSERS:
+ if not _parsers:
+ if log: log(" All parsers have failed")
+ return None
+ elif not parser:
+ if log: log(" All parsers have failed")
return None
+ if BKMK_DEBUG_HTML_PARSERS:
+ p, parser = _parsers[0]
if log: log(" Using %s" % p.__module__)
converted_title = title = parser.title