bkmk_objects.py

   1 """Objects to represent bookmarks.html structure
   2
   3 This file is a part of Bookmarks database and Internet robot.
   4
   5 """
   6
   7 __author__ = "Oleg Broytman <phd@phdru.name>"
   8 __copyright__ = "Copyright (C) 2000-2024 PhiloSoft Design"
   9 __license__ = "GNU GPL"
  10
  11 __all__ = ['Folder', 'Bookmark', 'Ruler', 'Walker', 'Writer', 'Robot',
  12            'InverseLinker', 'Linear', 'make_linear', 'make_tree', 'break_tree',
  13            'quote_title', 'unquote_title', 'parse_params', 'set_params',
  14            ]
  15
  16
  17 from urllib.parse import urlsplit, quote, unquote
  18 import os
  19
  20 BKMK_FORMAT = os.environ.get("BKMK_FORMAT", "MOZILLA")
  21
  22
  23 class Folder(list):
  24     isFolder = 1
  25     isBookmark = 0
  26
  27     def __init__(self, add_date=None, comment='', last_modified=None):
  28         super(Folder, self).__init__()
  29         self.comment = comment
  30         self.add_date = add_date
  31         self.last_modified = last_modified
  32
  33     def walk_depth(self, walker, level=0):
  34         if hasattr(self, "header"):  # root folder
  35             prune = 0
  36             walker.root_folder(self)
  37         else:
  38             prune = walker.prune_folder(self)
  39             if not prune:
  40                 walker.start_folder(self, level)
  41
  42         if not prune:
  43             for object in self:
  44                 if object.isFolder:
  45                     object.walk_depth(walker, level+1)
  46                 elif object.isBookmark:
  47                     walker.bookmark(object, level)
  48                 else:
  49                     walker.ruler(object, level)
  50
  51             walker.end_folder(self, level)
  52
  53
  54 class Bookmark(object):
  55     isFolder = 0
  56     isBookmark = 1
  57
  58     def __init__(self, href, add_date, last_visit=None, last_modified=None,
  59                  keyword=None, comment='', icon_href=None, icon=None,
  60                  charset=None, parser_charset=None):
  61         split_results = urlsplit(href)
  62         protocol, netloc, path, query, tag = split_results
  63         user = split_results.username
  64         password = split_results.password
  65         host = split_results.hostname
  66         port = split_results.port
  67
  68         if protocol == 'place':
  69             href = protocol + ":"
  70         else:
  71             href = protocol + "://"
  72         if user:
  73             href += quote(user)
  74             if password:
  75                 href += ':' + quote(password)
  76             href += '@'
  77         if host:
  78             href += host.encode('idna').decode('ascii')
  79             if port:
  80                 href += ':%d' % port
  81         if path:
  82             href += path
  83
  84         self.href = href
  85         self.add_date = add_date
  86         self.last_visit = last_visit
  87         self.last_modified = last_modified
  88         self.keyword = keyword
  89         self.comment = comment
  90         self.icon_href = icon_href
  91         self.icon = icon
  92         self.charset = charset
  93
  94
  95 class Ruler(object):
  96     isFolder = 0
  97     isBookmark = 0
  98
  99
 100 class Walker(object):
 101     """
 102        Interface class. Any instance that will be passed to Folder.walk_depth
 103        may be derived from this class. It is not mandatory - unlike Java
 104        Python does not require interface classes; but it is convenient to have
 105        some methods predefined to no-op, in case you do not want to
 106        provide end_folder etc.
 107     """
 108
 109     def root_folder(self, r):
 110         pass
 111
 112     def start_folder(self, f, level):
 113         pass
 114
 115     def end_folder(self, f, level):
 116         pass
 117
 118     def bookmark(self, b, level):
 119         pass
 120
 121     def ruler(self, r, level):
 122         pass
 123
 124     def prune_folder(self, folder):
 125         return 0
 126
 127
 128 class Writer(Walker):
 129     def __init__(self, outfile, prune=None):
 130         self.outfile = outfile
 131         self.prune = prune
 132
 133     def prune_folder(self, folder):
 134         return self.prune == folder.name
 135
 136
 137 class Robot(object):
 138     def __init__(self, log):
 139         self.log = log
 140
 141     def stop(self):
 142         pass  # Nothing to do on cleanup
 143
 144
 145 # Helper class to make inverese links (nodes linked to their parent)
 146 class InverseLinker(Walker):
 147     def root_folder(self, r):
 148         self.parent_stack = [r]
 149
 150     def start_folder(self, f, level):
 151         f.parent = self.parent_stack[-1]
 152         # Push the folder onto the stack of parents
 153         self.parent_stack.append(f)
 154
 155     def end_folder(self, f, level):
 156         del self.parent_stack[-1]   # Pop off the stack
 157
 158     def bookmark(self, b, level):
 159         b.parent = self.parent_stack[-1]
 160
 161     def ruler(self, r, level):
 162         r.parent = self.parent_stack[-1]
 163
 164
 165 # Helper class to make linear represenatation of the tree
 166 class Linear(Walker):
 167     def root_folder(self, r):
 168         r.linear = [r]
 169         self.linear = r.linear
 170
 171     def add_object(self, object):
 172         self.linear.append(object)
 173
 174     def start_folder(self, f, level):
 175         self.add_object(f)
 176
 177     def bookmark(self, b, level):
 178         self.add_object(b)
 179
 180     def ruler(self, r, level):
 181         self.add_object(r)
 182
 183
 184 # Helper - make linked linear represenatation of the tree,
 185 # suitable to be stored in sequential storage.
 186 def make_linear(root_folder):
 187     linker = InverseLinker()
 188     root_folder.walk_depth(linker)
 189
 190     linear = Linear()
 191     root_folder.walk_depth(linear)
 192
 193
 194 # Helper, opposite of make_linear -
 195 # make a tree from the linked linear representation.
 196 def make_tree(linear):
 197     root_folder = linear[0]
 198     del linear[0]
 199
 200     for object in linear:
 201         object.parent.append(object)
 202
 203     return root_folder
 204
 205
 206 def break_tree(linear):
 207     del linear[0]
 208
 209     for object in linear:
 210         del object.parent
 211
 212
 213 def quote_title(title):
 214     if BKMK_FORMAT == "MOZILLA":
 215         title = title.replace("'", "&#39;")
 216     return title
 217
 218
 219 def unquote_title(title):
 220     if BKMK_FORMAT == "MOZILLA":
 221         try:
 222             from HTMLParser import HTMLParser
 223         except ImportError:
 224             from html import unescape
 225         else:
 226             unescape = HTMLParser().unescape
 227         title = unescape(
 228             title.replace("&amp;", '&'))
 229         title = title.replace("&#39;", "'")
 230     return title
 231
 232
 233 def parse_params(param_str):
 234     params = param_str.split(':')
 235     main_param = params.pop(0)
 236     param_list = {}
 237     for param in params:
 238         key, value = param.split('=', 1)
 239         param_list[key] = unquote(value)
 240     return main_param, param_list
 241
 242
 243 def set_params(obj, params):
 244     if hasattr(params, "items"):
 245         params = params.items()
 246     for key, value in params:
 247         setattr(obj, key, value)