Remove robots based on Twisted

[bookmarks_db.git] / bkmk_objects.py

diff --git a/bkmk_objects.py b/bkmk_objects.py
index 89647aa52eb57ecced2d4681834bba4b79f4ffaf..b09d81d89d3db6f8751613c1fd5074b753e53b26 100644 (file)
--- a/bkmk_objects.py
+++ b/bkmk_objects.py
@@ -5,8 +5,9 @@ This file is a part of Bookmarks database and Internet robot.
 """
 
 __author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 2000-2023 PhiloSoft Design"
+__copyright__ = "Copyright (C) 2000-2025 PhiloSoft Design"
 __license__ = "GNU GPL"
+__version__ = '6.4.0'
 
 __all__ = ['Folder', 'Bookmark', 'Ruler', 'Walker', 'Writer', 'Robot',
            'InverseLinker', 'Linear', 'make_linear', 'make_tree', 'break_tree',
@@ -14,8 +15,8 @@ __all__ = ['Folder', 'Bookmark', 'Ruler', 'Walker', 'Writer', 'Robot',
            ]
 
 
+from urllib.parse import unquote
 import os
-import urllib
 
 BKMK_FORMAT = os.environ.get("BKMK_FORMAT", "MOZILLA")
 
@@ -58,32 +59,6 @@ class Bookmark(object):
     def __init__(self, href, add_date, last_visit=None, last_modified=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)
-
-        if protocol == 'place':
-            href = protocol + ":"
-        else:
-            href = protocol + "://"
-        if user:
-            href += urllib.quote(user)
-            if password:
-                href += ':' + urllib.quote(password)
-            href += '@'
-        if host:
-            href += host.decode(parser_charset or 'utf-8').encode('idna')
-            if port:
-                href += ':%d' % port
-        if path:
-            href += path
-
         self.href = href
         self.add_date = add_date
         self.last_visit = last_visit
@@ -95,6 +70,30 @@ class Bookmark(object):
         self.charset = charset
 
 
+bkmk_attrs = {
+    "charset": "Charset",
+    "error": "Error",
+    "icon": "Icon",
+    "icon_href": "IconURI",
+    "md5": "Md5",
+    "moved": "Moved",
+    "no_error": "NoError",
+    "real_title": "RealTitle",
+    "size": "Size",
+    "test_time": "TestTime",
+}
+
+
+bkmk_attrs_sentinel = object()
+
+
+def copy_bkmk(old_bkmk, new_bkmk):
+    for attr in bkmk_attrs.keys():
+        value = getattr(old_bkmk, attr, bkmk_attrs_sentinel)
+        if value is not bkmk_attrs_sentinel:
+            setattr(new_bkmk, attr, value)
+
+
 class Ruler(object):
     isFolder = 0
     isBookmark = 0
@@ -221,10 +220,15 @@ def quote_title(title):
 
 def unquote_title(title):
     if BKMK_FORMAT == "MOZILLA":
-        from HTMLParser import HTMLParser
-        title = HTMLParser().unescape(
-            title.replace("&amp;", '&').decode('utf-8'))
-        title = title.encode('utf-8').replace("&#39;", "'")
+        try:
+            from HTMLParser import HTMLParser
+        except ImportError:
+            from html import unescape
+        else:
+            unescape = HTMLParser().unescape
+        title = unescape(
+            title.replace("&amp;", '&'))
+        title = title.replace("&#39;", "'")
     return title
 
 
@@ -234,7 +238,7 @@ def parse_params(param_str):
     param_list = {}
     for param in params:
         key, value = param.split('=', 1)
-        param_list[key] = value
+        param_list[key] = unquote(value)
     return main_param, param_list
 
 
@@ -243,3 +247,7 @@ def set_params(obj, params):
         params = params.items()
     for key, value in params:
         setattr(obj, key, value)
+
+
+# Global var to use in robots
+tty_pbar = None