X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;ds=sidebyside;f=bkmk_parser.py;h=4cc8897e23f7b5737f9203e5664ef988b459acb3;hb=ca8624faf5b9a31e76b7bf3d441f446c8b62c8f6;hp=37f5f7743d33e0fc48527832559677f8775ba66a;hpb=364bdf891fa30093391583c449ebc7cbab4ec2a5;p=bookmarks_db.git diff --git a/bkmk_parser.py b/bkmk_parser.py old mode 100755 new mode 100644 index 37f5f77..4cc8897 --- a/bkmk_parser.py +++ b/bkmk_parser.py @@ -1,11 +1,11 @@ """ Parser for Netscape Navigator's and Mozilla's bookmarks.html - Written by BroytMann. Copyright (C) 1997-2003 PhiloSoft Design + Written by BroytMann. Copyright (C) 1997-2007 PhiloSoft Design """ -import os +import sys, os from m_lib.net.www.html import HTMLParser from bkmk_objects import Folder, Bookmark, Ruler @@ -29,6 +29,8 @@ else: dump_names = debug +DEFAULT_CHARSET = None + class BkmkParser(HTMLParser): def __init__(self): HTMLParser.__init__(self) @@ -42,8 +44,8 @@ class BkmkParser(HTMLParser): def handle_data(self, data): if data: - if self.charset: - data = unicode(data, self.charset).encode() + if DEFAULT_CHARSET: + data = unicode(data, self.charset, "replace").encode(DEFAULT_CHARSET, "xmlcharrefreplace") self.accumulator += data @@ -65,9 +67,21 @@ class BkmkParser(HTMLParser): self.charset = content.split('=')[1] except IndexError: pass + else: + global DEFAULT_CHARSET + DEFAULT_CHARSET = sys.getdefaultencoding() + if DEFAULT_CHARSET == "ascii": + try: + import locale + except ImportError: + pass + else: + DEFAULT_CHARSET = locale.getpreferredencoding() def start_title(self, attrs): + if DEFAULT_CHARSET: + self.accumulator += '\n' % DEFAULT_CHARSET self.accumulator += "