X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=bkmk_parser.py;h=d43674138e33f4f2eccc2b577a9ff2656c725cb9;hb=62977582003da9f434cd4fd377837935b6f0520c;hp=327e0c45f2dd2beb56711cd2e42eb0747e13bfd7;hpb=1f4d6d3d0e3a55e3390ef0f0e20f2358baa57fc1;p=bookmarks_db.git diff --git a/bkmk_parser.py b/bkmk_parser.py index 327e0c4..d436741 100755 --- 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) 1997-2003 PhiloSoft Design + Written by BroytMann. Copyright (C) 1997-2005 PhiloSoft Design """ @@ -70,6 +70,13 @@ 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): @@ -98,7 +105,7 @@ class BkmkParser(HTMLParser): self.root_folder.name = accumulator - # Start next folder + # Start a folder def start_h3(self, attrs): for attrname, value in attrs: value = value.strip() @@ -121,24 +128,27 @@ class BkmkParser(HTMLParser): self.current_folder.name = accumulator - # Start bookmark + # Start a bookmark def start_a(self, attrs): last_visit = None last_modified = None + keyword = None for attrname, value in attrs: value = value.strip() - if attrname == 'href': + if attrname == "href": href = value - if attrname == 'add_date': + elif attrname == "add_date": add_date = value - if attrname == 'last_visit': + elif attrname == "last_visit": last_visit = value - if attrname == 'last_modified': + elif attrname == "last_modified": last_modified = value + elif attrname == "shortcuturl": + keyword = value debug("Bookmark points to: `%s'" % href) - bookmark = Bookmark(href, add_date, last_visit, last_modified) + bookmark = Bookmark(href, add_date, last_visit, last_modified, keyword or '') self.current_object = bookmark self.current_folder.append(bookmark) self.urls += 1