Minor refactoring

[bookmarks_db.git] / bkmk_parser.py

diff --git a/bkmk_parser.py b/bkmk_parser.py
index e3b3db02532567aef8060c4546f1f3a6af6039e2..02116ccf025c9edacd119362331b64174ce89767 100644 (file)
--- a/bkmk_parser.py
+++ b/bkmk_parser.py
@@ -3,17 +3,15 @@
 This file is a part of Bookmarks database and Internet robot.
 """
 
-__version__ = "$Revision$"[11:-2]
-__revision__ = "$Id$"[5:-2]
-__date__ = "$Date$"[7:-2]
 __author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 1997-2011 PhiloSoft Design"
+__copyright__ = "Copyright (C) 1997-2012 PhiloSoft Design"
 __license__ = "GNU GPL"
 
 __all__ = ['BkmkParser']
 
 
-import sys, os
+import os
+from m_lib.defenc import default_encoding
 from m_lib.net.www.html import HTMLParser
 from bkmk_objects import Folder, Bookmark, Ruler
 
@@ -37,8 +35,6 @@ else:
    dump_names = debug
 
 
-DEFAULT_CHARSET = None
-
 class BkmkParser(HTMLParser):
    def __init__(self):
       HTMLParser.__init__(self)
@@ -46,13 +42,13 @@ class BkmkParser(HTMLParser):
       self.urls = 0
       self.objects = 0
 
-      self.charset = ""
+      self.charset = None
       self.recode = None
 
    def handle_data(self, data):
       if data:
-         if DEFAULT_CHARSET:
-            data = unicode(data, self.charset, "replace").encode(DEFAULT_CHARSET, "xmlcharrefreplace")
+         if self.charset and default_encoding:
+            data = unicode(data, self.charset, "replace").encode(default_encoding, "xmlcharrefreplace")
          self.accumulator += data
 
    # Mozilla - get charset
@@ -73,20 +69,10 @@ 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 += '<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=%s">\n' % DEFAULT_CHARSET
+      if default_encoding:
+         self.accumulator += '<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=%s">\n' % default_encoding
       self.accumulator += "<TITLE>"
 
    def end_title(self):
@@ -140,7 +126,7 @@ class BkmkParser(HTMLParser):
       add_date = None
       last_visit = None
       last_modified = None
-      keyword = None
+      keyword = ''
       icon = None
       charset = None
 
@@ -163,7 +149,8 @@ class BkmkParser(HTMLParser):
 
       debug("Bookmark points to: `%s'" % href)
       bookmark = Bookmark(href, add_date, last_visit, last_modified,
-         keyword or '', '', icon, charset)
+         keyword=keyword, icon=icon,
+         charset=charset, parser_charset=self.charset or default_encoding)
       self.current_object = bookmark
       self.current_folder.append(bookmark)
       self.urls += 1