Store charset in the DB and the generated HTML.

[bookmarks_db.git] / bkmk_parser.py

diff --git a/bkmk_parser.py b/bkmk_parser.py
index be39a322bb803032daa46e2d9c740985508b1d71..7ce99e8412447cbba334291783149f80cc2c3e46 100755 (executable)
--- 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 BroytMann. Copyright (C) 1997-2005 PhiloSoft Design
 """
 
 
@@ -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 += '<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=%s">\n' % DEFAULT_CHARSET
       self.accumulator += "<TITLE>"
 
    def end_title(self):
@@ -141,7 +150,7 @@ class BkmkParser(HTMLParser):
             keyword = 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 '')
       self.current_object = bookmark
       self.current_folder.append(bookmark)
       self.urls += 1