www/pages/torrent/client/bencode.py

   1 # Written by Petru Paler, Uoti Urpala, Ross Cohen and John Hoffman
   2 # see LICENSE.txt for license information
   3
   4 from types import IntType, LongType, StringType, ListType, TupleType, DictType
   5 try:
   6     from types import BooleanType
   7 except ImportError:
   8     BooleanType = None
   9 try:
  10     from types import UnicodeType
  11 except ImportError:
  12     UnicodeType = None
  13 from cStringIO import StringIO
  14
  15 def decode_int(x, f):
  16     f += 1
  17     newf = x.index('e', f)
  18     try:
  19         n = int(x[f:newf])
  20     except:
  21         n = long(x[f:newf])
  22     if x[f] == '-':
  23         if x[f + 1] == '0':
  24             raise ValueError
  25     elif x[f] == '0' and newf != f+1:
  26         raise ValueError
  27     return (n, newf+1)
  28
  29 def decode_string(x, f):
  30     colon = x.index(':', f)
  31     try:
  32         n = int(x[f:colon])
  33     except (OverflowError, ValueError):
  34         n = long(x[f:colon])
  35     if x[f] == '0' and colon != f+1:
  36         raise ValueError
  37     colon += 1
  38     return (x[colon:colon+n], colon+n)
  39
  40 def decode_unicode(x, f):
  41     s, f = decode_string(x, f+1)
  42     return (s.decode('UTF-8'),f)
  43
  44 def decode_list(x, f):
  45     r, f = [], f+1
  46     while x[f] != 'e':
  47         v, f = decode_func[x[f]](x, f)
  48         r.append(v)
  49     return (r, f + 1)
  50
  51 def decode_dict(x, f):
  52     r, f = {}, f+1
  53     lastkey = None
  54     while x[f] != 'e':
  55         k, f = decode_string(x, f)
  56         if lastkey >= k:
  57             raise ValueError
  58         lastkey = k
  59         r[k], f = decode_func[x[f]](x, f)
  60     return (r, f + 1)
  61
  62 decode_func = {}
  63 decode_func['l'] = decode_list
  64 decode_func['d'] = decode_dict
  65 decode_func['i'] = decode_int
  66 decode_func['0'] = decode_string
  67 decode_func['1'] = decode_string
  68 decode_func['2'] = decode_string
  69 decode_func['3'] = decode_string
  70 decode_func['4'] = decode_string
  71 decode_func['5'] = decode_string
  72 decode_func['6'] = decode_string
  73 decode_func['7'] = decode_string
  74 decode_func['8'] = decode_string
  75 decode_func['9'] = decode_string
  76 #decode_func['u'] = decode_unicode
  77
  78 def bdecode(x, sloppy = 0):
  79     try:
  80         r, l = decode_func[x[0]](x, 0)
  81 #    except (IndexError, KeyError):
  82     except (IndexError, KeyError, ValueError):
  83         raise ValueError, "bad bencoded data"
  84     if not sloppy and l != len(x):
  85         raise ValueError, "bad bencoded data"
  86     return r
  87
  88 def test_bdecode():
  89     try:
  90         bdecode('0:0:')
  91         assert 0
  92     except ValueError:
  93         pass
  94     try:
  95         bdecode('ie')
  96         assert 0
  97     except ValueError:
  98         pass
  99     try:
 100         bdecode('i341foo382e')
 101         assert 0
 102     except ValueError:
 103         pass
 104     assert bdecode('i4e') == 4L
 105     assert bdecode('i0e') == 0L
 106     assert bdecode('i123456789e') == 123456789L
 107     assert bdecode('i-10e') == -10L
 108     try:
 109         bdecode('i-0e')
 110         assert 0
 111     except ValueError:
 112         pass
 113     try:
 114         bdecode('i123')
 115         assert 0
 116     except ValueError:
 117         pass
 118     try:
 119         bdecode('')
 120         assert 0
 121     except ValueError:
 122         pass
 123     try:
 124         bdecode('i6easd')
 125         assert 0
 126     except ValueError:
 127         pass
 128     try:
 129         bdecode('35208734823ljdahflajhdf')
 130         assert 0
 131     except ValueError:
 132         pass
 133     try:
 134         bdecode('2:abfdjslhfld')
 135         assert 0
 136     except ValueError:
 137         pass
 138     assert bdecode('0:') == ''
 139     assert bdecode('3:abc') == 'abc'
 140     assert bdecode('10:1234567890') == '1234567890'
 141     try:
 142         bdecode('02:xy')
 143         assert 0
 144     except ValueError:
 145         pass
 146     try:
 147         bdecode('l')
 148         assert 0
 149     except ValueError:
 150         pass
 151     assert bdecode('le') == []
 152     try:
 153         bdecode('leanfdldjfh')
 154         assert 0
 155     except ValueError:
 156         pass
 157     assert bdecode('l0:0:0:e') == ['', '', '']
 158     try:
 159         bdecode('relwjhrlewjh')
 160         assert 0
 161     except ValueError:
 162         pass
 163     assert bdecode('li1ei2ei3ee') == [1, 2, 3]
 164     assert bdecode('l3:asd2:xye') == ['asd', 'xy']
 165     assert bdecode('ll5:Alice3:Bobeli2ei3eee') == [['Alice', 'Bob'], [2, 3]]
 166     try:
 167         bdecode('d')
 168         assert 0
 169     except ValueError:
 170         pass
 171     try:
 172         bdecode('defoobar')
 173         assert 0
 174     except ValueError:
 175         pass
 176     assert bdecode('de') == {}
 177     assert bdecode('d3:agei25e4:eyes4:bluee') == {'age': 25, 'eyes': 'blue'}
 178     assert bdecode('d8:spam.mp3d6:author5:Alice6:lengthi100000eee') == {'spam.mp3': {'author': 'Alice', 'length': 100000}}
 179     try:
 180         bdecode('d3:fooe')
 181         assert 0
 182     except ValueError:
 183         pass
 184     try:
 185         bdecode('di1e0:e')
 186         assert 0
 187     except ValueError:
 188         pass
 189     try:
 190         bdecode('d1:b0:1:a0:e')
 191         assert 0
 192     except ValueError:
 193         pass
 194     try:
 195         bdecode('d1:a0:1:a0:e')
 196         assert 0
 197     except ValueError:
 198         pass
 199     try:
 200         bdecode('i03e')
 201         assert 0
 202     except ValueError:
 203         pass
 204     try:
 205         bdecode('l01:ae')
 206         assert 0
 207     except ValueError:
 208         pass
 209     try:
 210         bdecode('9999:x')
 211         assert 0
 212     except ValueError:
 213         pass
 214     try:
 215         bdecode('l0:')
 216         assert 0
 217     except ValueError:
 218         pass
 219     try:
 220         bdecode('d0:0:')
 221         assert 0
 222     except ValueError:
 223         pass
 224     try:
 225         bdecode('d0:')
 226         assert 0
 227     except ValueError:
 228         pass
 229
 230 bencached_marker = []
 231
 232 class Bencached:
 233     def __init__(self, s):
 234         self.marker = bencached_marker
 235         self.bencoded = s
 236
 237 BencachedType = type(Bencached('')) # insufficient, but good as a filter
 238
 239 def encode_bencached(x,r):
 240     assert x.marker == bencached_marker
 241     r.append(x.bencoded)
 242
 243 def encode_int(x,r):
 244     r.extend(('i',str(x),'e'))
 245
 246 def encode_bool(x,r):
 247     encode_int(int(x),r)
 248
 249 def encode_string(x,r):
 250     r.extend((str(len(x)),':',x))
 251
 252 def encode_unicode(x,r):
 253     #r.append('u')
 254     encode_string(x.encode('UTF-8'),r)
 255
 256 def encode_list(x,r):
 257         r.append('l')
 258         for e in x:
 259             encode_func[type(e)](e, r)
 260         r.append('e')
 261
 262 def encode_dict(x,r):
 263     r.append('d')
 264     ilist = x.items()
 265     ilist.sort()
 266     for k,v in ilist:
 267         r.extend((str(len(k)),':',k))
 268         encode_func[type(v)](v, r)
 269     r.append('e')
 270
 271 encode_func = {}
 272 encode_func[BencachedType] = encode_bencached
 273 encode_func[IntType] = encode_int
 274 encode_func[LongType] = encode_int
 275 encode_func[StringType] = encode_string
 276 encode_func[ListType] = encode_list
 277 encode_func[TupleType] = encode_list
 278 encode_func[DictType] = encode_dict
 279 if BooleanType:
 280     encode_func[BooleanType] = encode_bool
 281 if UnicodeType:
 282     encode_func[UnicodeType] = encode_unicode
 283
 284 def bencode(x):
 285     r = []
 286     try:
 287         encode_func[type(x)](x, r)
 288     except:
 289         print "*** error *** could not encode type %s (value: %s)" % (type(x), x)
 290         assert 0
 291     return ''.join(r)
 292
 293 def test_bencode():
 294     assert bencode(4) == 'i4e'
 295     assert bencode(0) == 'i0e'
 296     assert bencode(-10) == 'i-10e'
 297     assert bencode(12345678901234567890L) == 'i12345678901234567890e'
 298     assert bencode('') == '0:'
 299     assert bencode('abc') == '3:abc'
 300     assert bencode('1234567890') == '10:1234567890'
 301     assert bencode([]) == 'le'
 302     assert bencode([1, 2, 3]) == 'li1ei2ei3ee'
 303     assert bencode([['Alice', 'Bob'], [2, 3]]) == 'll5:Alice3:Bobeli2ei3eee'
 304     assert bencode({}) == 'de'
 305     assert bencode({'age': 25, 'eyes': 'blue'}) == 'd3:agei25e4:eyes4:bluee'
 306     assert bencode({'spam.mp3': {'author': 'Alice', 'length': 100000}}) == 'd8:spam.mp3d6:author5:Alice6:lengthi100000eee'
 307     try:
 308         bencode({1: 'foo'})
 309         assert 0
 310     except AssertionError:
 311         pass
 312
 313
 314 try:
 315     import psyco
 316     psyco.bind(bdecode)
 317     psyco.bind(bencode)
 318 except ImportError:
 319     pass