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 __author__ = "Oleg Broytman <phd@phdru.name>"
   7 __copyright__ = "Copyright (C) 2000-2012 PhiloSoft Design"
   8 __license__ = "GNU GPL"
   9
  10 __all__ = ['Folder', 'Bookmark', 'Ruler', 'Walker', 'Writer', 'Robot',
  11     'InverseLinker', 'Linear', 'make_linear', 'make_tree', 'break_tree',
  12     'quote_title', 'unquote_title',
  13 ]
  14
  15
  16 import os, urllib
  17
  18 BKMK_FORMAT = os.environ.get("BKMK_FORMAT", "MOZILLA")
  19
  20 class Folder(list):
  21    isFolder = 1
  22    isBookmark = 0
  23
  24    def __init__(self, add_date=None, comment='', last_modified=None):
  25       super(Folder, self).__init__()
  26       self.comment = comment
  27       self.add_date = add_date
  28       self.last_modified = last_modified
  29
  30    def walk_depth(self, walker, level=0):
  31       if hasattr(self, "header"): # root folder
  32          prune = 0
  33          walker.root_folder(self)
  34       else:
  35          prune = walker.prune_folder(self)
  36          if not prune:
  37             walker.start_folder(self, level)
  38
  39       if not prune:
  40          for object in self:
  41             if object.isFolder:
  42                object.walk_depth(walker, level+1)
  43             elif object.isBookmark:
  44                walker.bookmark(object, level)
  45             else:
  46                walker.ruler(object, level)
  47
  48          walker.end_folder(self, level)
  49
  50
  51 class Bookmark:
  52    isFolder = 0
  53    isBookmark = 1
  54
  55    def __init__(self, href, add_date, last_visit=None, last_modified=None,
  56          keyword=None, comment='', icon_href=None, icon=None,
  57          charset=None, parser_charset=None):
  58       protocol, request = urllib.splittype(href)
  59       user, password, port = None, None, None
  60       host, path = urllib.splithost(request)
  61       if host:
  62          user, host = urllib.splituser(host)
  63          if user:
  64             user, password = urllib.splitpasswd(user)
  65          host, port = urllib.splitport(host)
  66          if port: port = int(port)
  67
  68       href = protocol + "://"
  69       if user:
  70          href += urllib.quote(user)
  71          if password:
  72             href += ':' + urllib.quote(password)
  73          href += '@'
  74       if host:
  75          href += host.decode(parser_charset or 'utf-8').encode('idna')
  76          if port:
  77             href += ':%d' % port
  78       if path:
  79          href += path
  80
  81       self.href = href
  82       self.add_date = add_date
  83       self.last_visit = last_visit
  84       self.last_modified = last_modified
  85       self.keyword = keyword
  86       self.comment = comment
  87       self.icon_href = icon_href
  88       self.icon = icon
  89       self.charset = charset
  90
  91
  92 class Ruler:
  93    isFolder = 0
  94    isBookmark = 0
  95
  96
  97 class Walker:
  98    """
  99       Interface class. Any instance that will be passed to Folder.walk_depth
 100       may be derived from this class. It is not mandatory - unlike Java
 101       Python does not require interface classes; but it is convenient to have
 102       some methods predefined to no-op, in case you do not want to
 103       provide end_folder etc.
 104    """
 105
 106    def root_folder(self, r):
 107       pass
 108
 109    def start_folder(self, f, level):
 110       pass
 111
 112    def end_folder(self, f, level):
 113       pass
 114
 115    def bookmark(self, b, level):
 116       pass
 117
 118    def ruler(self, r, level):
 119       pass
 120
 121    def prune_folder(self, folder):
 122       return 0
 123
 124
 125 class Writer(Walker):
 126    def __init__(self, outfile, prune=None):
 127       self.outfile = outfile
 128       self.prune = prune
 129
 130    def prune_folder(self, folder):
 131       return self.prune == folder.name
 132
 133
 134 class Robot:
 135    def __init__(self, log):
 136       self.log = log
 137
 138    def stop(self):
 139       pass # Nothing to do on cleanup
 140
 141
 142 # Helper class to make inverese links (nodes linked to their parent)
 143 class InverseLinker(Walker):
 144    def root_folder(self, r):
 145       self.parent_stack = [r]
 146
 147    def start_folder(self, f, level):
 148       f.parent = self.parent_stack[-1]
 149       self.parent_stack.append(f) # Push the folder onto the stack of parents
 150
 151    def end_folder(self, f, level):
 152       del self.parent_stack[-1]   # Pop off the stack
 153
 154    def bookmark(self, b, level):
 155       b.parent = self.parent_stack[-1]
 156
 157    def ruler(self, r, level):
 158       r.parent = self.parent_stack[-1]
 159
 160
 161 # Helper class to make linear represenatation of the tree
 162 class Linear(Walker):
 163    def root_folder(self, r):
 164       r.linear = [r]
 165       self.linear = r.linear
 166
 167    def add_object(self, object):
 168       self.linear.append(object)
 169
 170    def start_folder(self, f, level):
 171       self.add_object(f)
 172
 173    def bookmark(self, b, level):
 174       self.add_object(b)
 175
 176    def ruler(self, r, level):
 177       self.add_object(r)
 178
 179
 180 # Helper - make linked linear represenatation of the tree, suitable to be stored in sequential storage
 181 def make_linear(root_folder):
 182    linker = InverseLinker()
 183    root_folder.walk_depth(linker)
 184
 185    linear = Linear()
 186    root_folder.walk_depth(linear)
 187
 188
 189 # Helper, opposite of make_linear - make a tree from the linked linear representation
 190 def make_tree(linear):
 191    root_folder = linear[0]
 192    del linear[0]
 193
 194    for object in linear:
 195       object.parent.append(object)
 196
 197    return root_folder
 198
 199 def break_tree(linear):
 200    del linear[0]
 201
 202    for object in linear:
 203       del object.parent
 204
 205
 206 def quote_title(title):
 207    if BKMK_FORMAT == "MOZILLA":
 208       title = title.replace("'", "&#39;")
 209    return title
 210
 211 def unquote_title(title):
 212    if BKMK_FORMAT == "MOZILLA":
 213       from HTMLParser import HTMLParser
 214       title = HTMLParser().unescape(title.replace("&amp;", '&').decode('utf-8'))
 215       title = title.encode('utf-8').replace("&#39;", "'")
 216    return title