X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fparse_html.py;h=b30c66458eaa32960b5badce9bc849a2082213d7;hb=0274654a95196ee2b70720b4708a4dd76107a2ba;hp=6ec63369065035ca61308d3ad5cd65738dabd554;hpb=beae1a1df44dc95377bbb0a28e1eb2eac827c8b5;p=bookmarks_db.git diff --git a/Robots/parse_html.py b/Robots/parse_html.py index 6ec6336..b30c664 100755 --- a/Robots/parse_html.py +++ b/Robots/parse_html.py @@ -5,20 +5,12 @@ Written by BroytMann. Copyright (C) 1997-2007 PhiloSoft Design """ +import codecs -import sys -current_charset = sys.getdefaultencoding() -if current_charset == "ascii": - try: - import locale - except ImportError: - pass - else: - current_charset = locale.getpreferredencoding() -current_charset = current_charset.replace("windows-", "cp").lower() +from m_lib.defenc import default_encoding +current_charset = default_encoding.replace("windows-", "cp") DEFAULT_CHARSET = "cp1251" # Stupid default for Russian Cyrillic - from HTMLParser import HTMLParseError from m_lib.net.www.html import HTMLParser as _HTMLParser @@ -79,15 +71,13 @@ class HTMLParser(_HTMLParser): for attrname, value in attrs: if value: value = value.strip().lower() - if (attrname == 'rel') and (value in ('icon', 'shortcut icon')): + if (attrname == 'rel') and (value.lower() in ('icon', 'shortcut icon')): has_icon = True elif attrname == 'href': href = value if has_icon: self.icon = href - else: - self.icon = None import re