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 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         self.href = href
  62         self.add_date = add_date
  63         self.last_visit = last_visit
  64         self.last_modified = last_modified
  65         self.keyword = keyword
  66         self.comment = comment
  67         self.icon_href = icon_href
  68         self.icon = icon
  69         self.charset = charset
  70
  71
  72 class Ruler(object):
  73     isFolder = 0
  74     isBookmark = 0
  75
  76
  77 class Walker(object):
  78     """
  79        Interface class. Any instance that will be passed to Folder.walk_depth
  80        may be derived from this class. It is not mandatory - unlike Java
  81        Python does not require interface classes; but it is convenient to have
  82        some methods predefined to no-op, in case you do not want to
  83        provide end_folder etc.
  84     """
  85
  86     def root_folder(self, r):
  87         pass
  88
  89     def start_folder(self, f, level):
  90         pass
  91
  92     def end_folder(self, f, level):
  93         pass
  94
  95     def bookmark(self, b, level):
  96         pass
  97
  98     def ruler(self, r, level):
  99         pass
 100
 101     def prune_folder(self, folder):
 102         return 0
 103
 104
 105 class Writer(Walker):
 106     def __init__(self, outfile, prune=None):
 107         self.outfile = outfile
 108         self.prune = prune
 109
 110     def prune_folder(self, folder):
 111         return self.prune == folder.name
 112
 113
 114 class Robot(object):
 115     def __init__(self, log):
 116         self.log = log
 117
 118     def stop(self):
 119         pass  # Nothing to do on cleanup
 120
 121
 122 # Helper class to make inverese links (nodes linked to their parent)
 123 class InverseLinker(Walker):
 124     def root_folder(self, r):
 125         self.parent_stack = [r]
 126
 127     def start_folder(self, f, level):
 128         f.parent = self.parent_stack[-1]
 129         # Push the folder onto the stack of parents
 130         self.parent_stack.append(f)
 131
 132     def end_folder(self, f, level):
 133         del self.parent_stack[-1]   # Pop off the stack
 134
 135     def bookmark(self, b, level):
 136         b.parent = self.parent_stack[-1]
 137
 138     def ruler(self, r, level):
 139         r.parent = self.parent_stack[-1]
 140
 141
 142 # Helper class to make linear represenatation of the tree
 143 class Linear(Walker):
 144     def root_folder(self, r):
 145         r.linear = [r]
 146         self.linear = r.linear
 147
 148     def add_object(self, object):
 149         self.linear.append(object)
 150
 151     def start_folder(self, f, level):
 152         self.add_object(f)
 153
 154     def bookmark(self, b, level):
 155         self.add_object(b)
 156
 157     def ruler(self, r, level):
 158         self.add_object(r)
 159
 160
 161 # Helper - make linked linear represenatation of the tree,
 162 # suitable to be stored in sequential storage.
 163 def make_linear(root_folder):
 164     linker = InverseLinker()
 165     root_folder.walk_depth(linker)
 166
 167     linear = Linear()
 168     root_folder.walk_depth(linear)
 169
 170
 171 # Helper, opposite of make_linear -
 172 # make a tree from the linked linear representation.
 173 def make_tree(linear):
 174     root_folder = linear[0]
 175     del linear[0]
 176
 177     for object in linear:
 178         object.parent.append(object)
 179
 180     return root_folder
 181
 182
 183 def break_tree(linear):
 184     del linear[0]
 185
 186     for object in linear:
 187         del object.parent
 188
 189
 190 def quote_title(title):
 191     if BKMK_FORMAT == "MOZILLA":
 192         title = title.replace("'", "&#39;")
 193     return title
 194
 195
 196 def unquote_title(title):
 197     if BKMK_FORMAT == "MOZILLA":
 198         try:
 199             from HTMLParser import HTMLParser
 200         except ImportError:
 201             from html import unescape
 202         else:
 203             unescape = HTMLParser().unescape
 204         title = unescape(
 205             title.replace("&amp;", '&'))
 206         title = title.replace("&#39;", "'")
 207     return title
 208
 209
 210 def parse_params(param_str):
 211     params = param_str.split(':')
 212     main_param = params.pop(0)
 213     param_list = {}
 214     for param in params:
 215         key, value = param.split('=', 1)
 216         param_list[key] = unquote(value)
 217     return main_param, param_list
 218
 219
 220 def set_params(obj, params):
 221     if hasattr(params, "items"):
 222         params = params.items()
 223     for key, value in params:
 224         setattr(obj, key, value)