X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=bkmk_parser.py;h=3f2182cf4c15e3f73631874d54aada5ff4ac55e0;hb=c05c3463a8d992c798443d7a6fe8c01de1b776fa;hp=be39a322bb803032daa46e2d9c740985508b1d71;hpb=5a3a639d28368406ca2e02584ee7924cb0ae8089;p=bookmarks_db.git diff --git a/bkmk_parser.py b/bkmk_parser.py old mode 100755 new mode 100644 index be39a32..3f2182c --- a/bkmk_parser.py +++ b/bkmk_parser.py @@ -1,7 +1,7 @@ """ Parser for Netscape Navigator's and Mozilla's bookmarks.html - Written by BroytMann. Copyright (C) 1997-2004 PhiloSoft Design + Written by Broytman. Copyright (C) 1997-2008 PhiloSoft Design """ @@ -45,7 +45,7 @@ class BkmkParser(HTMLParser): def handle_data(self, data): if data: if DEFAULT_CHARSET: - data = unicode(data, self.charset, "replace").encode(DEFAULT_CHARSET, "replace") + data = unicode(data, self.charset, "replace").encode(DEFAULT_CHARSET, "xmlcharrefreplace") self.accumulator += data @@ -70,9 +70,18 @@ class BkmkParser(HTMLParser): 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 += "" def end_title(self): @@ -100,13 +109,16 @@ class BkmkParser(HTMLParser): # Start a folder def start_h3(self, attrs): + last_modified = None for attrname, value in attrs: value = value.strip() if attrname == 'add_date': add_date = value + elif attrname == 'last_modified': + last_modified = value debug("New folder...") - folder = Folder(add_date) + folder = Folder(add_date, last_modified=last_modified) self.current_object = folder self.current_folder.append(folder) self.folder_stack.append(folder) # push new folder @@ -123,9 +135,12 @@ class BkmkParser(HTMLParser): # Start a bookmark def start_a(self, attrs): + add_date = None last_visit = None last_modified = None keyword = None + icon = None + charset = None for attrname, value in attrs: value = value.strip() @@ -139,9 +154,14 @@ class BkmkParser(HTMLParser): last_modified = value elif attrname == "shortcuturl": keyword = value + elif attrname == "icon": + icon = value + elif attrname == "last_charset": + charset = value debug("Bookmark points to: `%s'" % href) - bookmark = Bookmark(href, add_date, last_visit, last_modified, keyword) + bookmark = Bookmark(href, add_date, last_visit, last_modified, + keyword or '', '', icon, charset) self.current_object = bookmark self.current_folder.append(bookmark) self.urls += 1