Split hrefs into domain and path components; recode only domain.

[bookmarks_db.git] / bkmk_objects.py

diff --git a/bkmk_objects.py b/bkmk_objects.py
index d15c38e52d440c2a7bc29a827b75e86953917d3a..be139edb3e205379733c8e7d38f8db42f2ce4c77 100644 (file)
--- a/bkmk_objects.py
+++ b/bkmk_objects.py
@@ -10,9 +10,15 @@ __author__ = "Oleg Broytman <phd@phdru.name>"
 __copyright__ = "Copyright (C) 2000-2011 PhiloSoft Design"
 __license__ = "GNU GPL"
 
-import os, cgi
-BKMK_FORMAT = os.environ.get("BKMK_FORMAT", "MOZILLA")
+__all__ = ['Folder', 'Bookmark', 'Ruler', 'Walker', 'Writer', 'Robot',
+    'InverseLinker', 'Linear', 'make_linear', 'make_tree', 'break_tree',
+    'quote_title', 'unquote_title',
+]
+
 
+import os, urllib
+
+BKMK_FORMAT = os.environ.get("BKMK_FORMAT", "MOZILLA")
 
 class Folder(list):
    isFolder = 1
@@ -50,7 +56,45 @@ class Bookmark:
    isBookmark = 1
 
    def __init__(self, href, add_date, last_visit=None, last_modified=None,
-         keyword=None, comment='', icon_href=None, icon=None, charset=None):
+         keyword=None, comment='', icon_href=None, icon=None,
+         charset=None, parser_charset=None):
+      protocol, request = urllib.splittype(href)
+      user, password, port = None, None, None
+      host, path = urllib.splithost(request)
+      if host:
+         user, host = urllib.splituser(host)
+         if user:
+            user, password = urllib.splitpasswd(user)
+         host, port = urllib.splitport(host)
+         if port: port = int(port)
+      path, tag = urllib.splittag(path)
+      path, query = urllib.splitquery(path)
+      path = urllib.unquote(path)
+      if tag: tag = urllib.unquote_plus(tag)
+
+      if host: # host can be None for Mozilla's place: URLs
+          host = host.decode(parser_charset or 'utf-8').encode('idna')
+
+      href = protocol + "://"
+      if user:
+         href += urllib.quote(user)
+         if password:
+            href += ':' + urllib.quote(password)
+         href += '@'
+      if host:
+         href += host
+         if port:
+            href += ':%d' % port
+      if path:
+         if protocol == "file":
+            href += urllib.quote(path)
+         else:
+            href += urllib.quote(path)
+      if query:
+         href += '?' + query
+      if tag:
+         href += '#' + urllib.quote_plus(tag)
+
       self.href = href
       self.add_date = add_date
       self.last_visit = last_visit