This file is indexed.

/usr/share/pyshared/DNS/Lib.py is in python-dns 2.3.6-3.

This file is owned by root:root, with mode 0o644.

The actual contents of the file can be viewed below.

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
# -*- encoding: utf-8 -*-
"""
 $Id: Lib.py,v 1.11.2.11 2011/11/23 17:07:14 customdesigned Exp $

 This file is part of the pydns project.
 Homepage: http://pydns.sourceforge.net

 This code is covered by the standard Python License. See LICENSE for details.

 Library code. Largely this is packers and unpackers for various types.
"""

#
#
# See RFC 1035:
# ------------------------------------------------------------------------
# Network Working Group                                     P. Mockapetris
# Request for Comments: 1035                                           ISI
#                                                            November 1987
# Obsoletes: RFCs 882, 883, 973
#
#             DOMAIN NAMES - IMPLEMENTATION AND SPECIFICATION
# ------------------------------------------------------------------------


import types

import Type
import Class
import Opcode
import Status
import DNS

from Base import DNSError

LABEL_UTF8 = False
LABEL_ENCODING = 'idna'

class UnpackError(DNSError): pass
class PackError(DNSError): pass

# Low-level 16 and 32 bit integer packing and unpacking

from struct import pack as struct_pack
from struct import unpack as struct_unpack
from socket import inet_ntoa, inet_aton

def pack16bit(n):
    return struct_pack('!H', n)

def pack32bit(n):
    return struct_pack('!L', n)

def unpack16bit(s):
    return struct_unpack('!H', s)[0]

def unpack32bit(s):
    return struct_unpack('!L', s)[0]

def addr2bin(addr):
    return struct_unpack('!l', inet_aton(addr))[0]

def bin2addr(n):
    return inet_ntoa(struct_pack('!L', n))

# Packing class

class Packer:
    " packer base class. supports basic byte/16bit/32bit/addr/string/name "
    def __init__(self):
        self.buf = ''
        self.index = {}
    def getbuf(self):
        return self.buf
    def addbyte(self, c):
        if len(c) != 1: raise TypeError, 'one character expected'
        self.buf = self.buf + c
    def addbytes(self, bytes):
        self.buf = self.buf + bytes
    def add16bit(self, n):
        self.buf = self.buf + pack16bit(n)
    def add32bit(self, n):
        self.buf = self.buf + pack32bit(n)
    def addaddr(self, addr):
        n = addr2bin(addr)
        self.buf = self.buf + pack32bit(n)
    def addstring(self, s):
        if len(s) > 255:
            raise ValueError, "Can't encode string of length "+ \
                            "%s (> 255)"%(len(s))
        self.addbyte(chr(len(s)))
        self.addbytes(s)
    def addname(self, name):
        # Domain name packing (section 4.1.4)
        # Add a domain name to the buffer, possibly using pointers.
        # The case of the first occurrence of a name is preserved.
        # One trailing dot is ignored.
        if name.endswith('.'):
            name = name[:-1]
        if name:
            nlist = name.split('.')
            for label in nlist:
                if not label:
                    raise PackError, 'empty label'
        else:
            nlist = []
        keys = []
        for i in range(len(nlist)):
            key = '.'.join(nlist[i:]).upper()
            keys.append(key)
            if self.index.has_key(key):
                pointer = self.index[key]
                break
        else:
            i = len(nlist)
            pointer = None
        # Do it into temporaries first so exceptions don't
        # mess up self.index and self.buf
        buf = ''
        offset = len(self.buf)
        index = []
        if DNS.LABEL_UTF8:
          enc = 'utf8'
        else:
          enc = DNS.LABEL_ENCODING
        for j in range(i):
            label = nlist[j]
            try:
                label = label.encode(enc)
            except UnicodeEncodeError:
                if not DNS.LABEL_UTF8: raise
                if not label.startswith('\ufeff'):
                    label = '\ufeff'+label
                label = label.encode(enc)
            n = len(label)
            if n > 63:
                raise PackError, 'label too long'
            if offset + len(buf) < 0x3FFF:
                index.append((keys[j], offset + len(buf)))
            else:
                print 'DNS.Lib.Packer.addname:',
                print 'warning: pointer too big'
            buf = buf + (chr(n) + label)
        if pointer:
            buf = buf + pack16bit(pointer | 0xC000)
        else:
            buf = buf + '\0'
        self.buf = self.buf + buf
        for key, value in index:
            self.index[key] = value
    def dump(self):
        keys = self.index.keys()
        keys.sort()
        print '-'*40
        for key in keys:
            print '%20s %3d' % (key, self.index[key])
        print '-'*40
        space = 1
        for i in range(0, len(self.buf)+1, 2):
            if self.buf[i:i+2] == '**':
                if not space: print
                space = 1
                continue
            space = 0
            print '%4d' % i,
            for c in self.buf[i:i+2]:
                if ' ' < c < '\177':
                    print ' %c' % c,
                else:
                    print '%2d' % ord(c),
            print
        print '-'*40


# Unpacking class


class Unpacker:
    def __init__(self, buf):
        self.buf = buf
        self.offset = 0
    def getbyte(self):
        if self.offset >= len(self.buf):
            raise UnpackError, "Ran off end of data"
        c = self.buf[self.offset]
        self.offset = self.offset + 1
        return c
    def getbytes(self, n):
        s = self.buf[self.offset : self.offset + n]
        if len(s) != n: raise UnpackError, 'not enough data left'
        self.offset = self.offset + n
        return s
    def get16bit(self):
        return unpack16bit(self.getbytes(2))
    def get32bit(self):
        return unpack32bit(self.getbytes(4))
    def getaddr(self):
        return bin2addr(self.get32bit())
    def getstring(self):
        return self.getbytes(ord(self.getbyte()))
    def getname(self):
        # Domain name unpacking (section 4.1.4)
        c = self.getbyte()
        i = ord(c)
        if i & 0xC0 == 0xC0:
            d = self.getbyte()
            j = ord(d)
            pointer = ((i<<8) | j) & ~0xC000
            save_offset = self.offset
            try:
                self.offset = pointer
                domain = self.getname()
            finally:
                self.offset = save_offset
            return domain
        if i == 0:
            return ''
        domain = self.getbytes(i)
        remains = self.getname()
        if not remains:
            return domain
        else:
            return domain + '.' + remains


# Test program for packin/unpacking (section 4.1.4)

def testpacker():
    N = 2500
    R = range(N)
    import timing
    # See section 4.1.4 of RFC 1035
    timing.start()
    for i in R:
        p = Packer()
        p.addaddr('192.168.0.1')
        p.addbytes('*' * 20)
        p.addname('f.ISI.ARPA')
        p.addbytes('*' * 8)
        p.addname('Foo.F.isi.arpa')
        p.addbytes('*' * 18)
        p.addname('arpa')
        p.addbytes('*' * 26)
        p.addname('')
    timing.finish()
    print timing.milli(), "ms total for packing"
    print round(timing.milli()  / i, 4), 'ms per packing'
    #p.dump()
    u = Unpacker(p.buf)
    u.getaddr()
    u.getbytes(20)
    u.getname()
    u.getbytes(8)
    u.getname()
    u.getbytes(18)
    u.getname()
    u.getbytes(26)
    u.getname()
    timing.start()
    for i in R:
        u = Unpacker(p.buf)

        res = (u.getaddr(),
               u.getbytes(20),
               u.getname(),
               u.getbytes(8),
               u.getname(),
               u.getbytes(18),
               u.getname(),
               u.getbytes(26),
               u.getname())
    timing.finish()
    print timing.milli(), "ms total for unpacking"
    print round(timing.milli() / i, 4), 'ms per unpacking'
    #for item in res: print item


# Pack/unpack RR toplevel format (section 3.2.1)

class RRpacker(Packer):
    def __init__(self):
        Packer.__init__(self)
        self.rdstart = None
    def addRRheader(self, name, type, klass, ttl, *rest):
        self.addname(name)
        self.add16bit(type)
        self.add16bit(klass)
        self.add32bit(ttl)
        if rest:
            if rest[1:]: raise TypeError, 'too many args'
            rdlength = rest[0]
        else:
            rdlength = 0
        self.add16bit(rdlength)
        self.rdstart = len(self.buf)
    def patchrdlength(self):
        rdlength = unpack16bit(self.buf[self.rdstart-2:self.rdstart])
        if rdlength == len(self.buf) - self.rdstart:
            return
        rdata = self.buf[self.rdstart:]
        save_buf = self.buf
        ok = 0
        try:
            self.buf = self.buf[:self.rdstart-2]
            self.add16bit(len(rdata))
            self.buf = self.buf + rdata
            ok = 1
        finally:
            if not ok: self.buf = save_buf
    def endRR(self):
        if self.rdstart is not None:
            self.patchrdlength()
        self.rdstart = None
    def getbuf(self):
        if self.rdstart is not None: self.patchrdlength()
        return Packer.getbuf(self)
    # Standard RRs (section 3.3)
    def addCNAME(self, name, klass, ttl, cname):
        self.addRRheader(name, Type.CNAME, klass, ttl)
        self.addname(cname)
        self.endRR()
    def addHINFO(self, name, klass, ttl, cpu, os):
        self.addRRheader(name, Type.HINFO, klass, ttl)
        self.addstring(cpu)
        self.addstring(os)
        self.endRR()
    def addMX(self, name, klass, ttl, preference, exchange):
        self.addRRheader(name, Type.MX, klass, ttl)
        self.add16bit(preference)
        self.addname(exchange)
        self.endRR()
    def addNS(self, name, klass, ttl, nsdname):
        self.addRRheader(name, Type.NS, klass, ttl)
        self.addname(nsdname)
        self.endRR()
    def addPTR(self, name, klass, ttl, ptrdname):
        self.addRRheader(name, Type.PTR, klass, ttl)
        self.addname(ptrdname)
        self.endRR()
    def addSOA(self, name, klass, ttl,
              mname, rname, serial, refresh, retry, expire, minimum):
        self.addRRheader(name, Type.SOA, klass, ttl)
        self.addname(mname)
        self.addname(rname)
        self.add32bit(serial)
        self.add32bit(refresh)
        self.add32bit(retry)
        self.add32bit(expire)
        self.add32bit(minimum)
        self.endRR()
    def addTXT(self, name, klass, ttl, tlist):
        self.addRRheader(name, Type.TXT, klass, ttl)
        if type(tlist) is types.StringType:
            tlist = [tlist]
        for txtdata in tlist:
            self.addstring(txtdata)
        self.endRR()
    # Internet specific RRs (section 3.4) -- class = IN
    def addA(self, name, klass, ttl, address):
        self.addRRheader(name, Type.A, klass, ttl)
        self.addaddr(address)
        self.endRR()
    def addWKS(self, name, ttl, address, protocol, bitmap):
        self.addRRheader(name, Type.WKS, Class.IN, ttl)
        self.addaddr(address)
        self.addbyte(chr(protocol))
        self.addbytes(bitmap)
        self.endRR()
    def addSRV(self):
        raise NotImplementedError

def prettyTime(seconds):
    if seconds<60:
        return seconds,"%d seconds"%(seconds)
    if seconds<3600:
        return seconds,"%d minutes"%(seconds/60)
    if seconds<86400:
        return seconds,"%d hours"%(seconds/3600)
    if seconds<604800:
        return seconds,"%d days"%(seconds/86400)
    else:
        return seconds,"%d weeks"%(seconds/604800)


class RRunpacker(Unpacker):
    def __init__(self, buf):
        Unpacker.__init__(self, buf)
        self.rdend = None
    def getRRheader(self):
        name = self.getname()
        rrtype = self.get16bit()
        klass = self.get16bit()
        ttl = self.get32bit()
        rdlength = self.get16bit()
        self.rdend = self.offset + rdlength
        return (name, rrtype, klass, ttl, rdlength)
    def endRR(self):
        if self.offset != self.rdend:
            raise UnpackError, 'end of RR not reached'
    def getCNAMEdata(self):
        return self.getname()
    def getHINFOdata(self):
        return self.getstring(), self.getstring()
    def getMXdata(self):
        return self.get16bit(), self.getname()
    def getNSdata(self):
        return self.getname()
    def getPTRdata(self):
        return self.getname()
    def getSOAdata(self):
        return self.getname(), \
               self.getname(), \
               ('serial',)+(self.get32bit(),), \
               ('refresh ',)+prettyTime(self.get32bit()), \
               ('retry',)+prettyTime(self.get32bit()), \
               ('expire',)+prettyTime(self.get32bit()), \
               ('minimum',)+prettyTime(self.get32bit())
    def getTXTdata(self):
        txt = []
        while self.offset != self.rdend:
            txt.append(self.getstring())
        return txt
    getSPFdata = getTXTdata
    def getAdata(self):
        return self.getaddr()
    def getWKSdata(self):
        address = self.getaddr()
        protocol = ord(self.getbyte())
        bitmap = self.getbytes(self.rdend - self.offset)
        return address, protocol, bitmap
    def getSRVdata(self):
        """
        _Service._Proto.Name TTL Class SRV Priority Weight Port Target
        """
        priority = self.get16bit()
        weight = self.get16bit()
        port = self.get16bit()
        target = self.getname()
        #print '***priority, weight, port, target', priority, weight, port, target
        return priority, weight, port, target


# Pack/unpack Message Header (section 4.1)

class Hpacker(Packer):
    def addHeader(self, id, qr, opcode, aa, tc, rd, ra, z, rcode,
              qdcount, ancount, nscount, arcount):
        self.add16bit(id)
        self.add16bit((qr&1)<<15 | (opcode&0xF)<<11 | (aa&1)<<10
                  | (tc&1)<<9 | (rd&1)<<8 | (ra&1)<<7
                  | (z&7)<<4 | (rcode&0xF))
        self.add16bit(qdcount)
        self.add16bit(ancount)
        self.add16bit(nscount)
        self.add16bit(arcount)

class Hunpacker(Unpacker):
    def getHeader(self):
        id = self.get16bit()
        flags = self.get16bit()
        qr, opcode, aa, tc, rd, ra, z, rcode = (
                  (flags>>15)&1,
                  (flags>>11)&0xF,
                  (flags>>10)&1,
                  (flags>>9)&1,
                  (flags>>8)&1,
                  (flags>>7)&1,
                  (flags>>4)&7,
                  (flags>>0)&0xF)
        qdcount = self.get16bit()
        ancount = self.get16bit()
        nscount = self.get16bit()
        arcount = self.get16bit()
        return (id, qr, opcode, aa, tc, rd, ra, z, rcode,
                  qdcount, ancount, nscount, arcount)


# Pack/unpack Question (section 4.1.2)

class Qpacker(Packer):
    def addQuestion(self, qname, qtype, qclass):
        self.addname(qname)
        self.add16bit(qtype)
        self.add16bit(qclass)

class Qunpacker(Unpacker):
    def getQuestion(self):
        return self.getname(), self.get16bit(), self.get16bit()


# Pack/unpack Message(section 4)
# NB the order of the base classes is important for __init__()!

class Mpacker(RRpacker, Qpacker, Hpacker):
    pass

class Munpacker(RRunpacker, Qunpacker, Hunpacker):
    pass


# Routines to print an unpacker to stdout, for debugging.
# These affect the unpacker's current position!

def dumpM(u):
    print 'HEADER:',
    (id, qr, opcode, aa, tc, rd, ra, z, rcode,
              qdcount, ancount, nscount, arcount) = u.getHeader()
    print 'id=%d,' % id,
    print 'qr=%d, opcode=%d, aa=%d, tc=%d, rd=%d, ra=%d, z=%d, rcode=%d,' \
              % (qr, opcode, aa, tc, rd, ra, z, rcode)
    if tc: print '*** response truncated! ***'
    if rcode: print '*** nonzero error code! (%d) ***' % rcode
    print '  qdcount=%d, ancount=%d, nscount=%d, arcount=%d' \
              % (qdcount, ancount, nscount, arcount)
    for i in range(qdcount):
        print 'QUESTION %d:' % i,
        dumpQ(u)
    for i in range(ancount):
        print 'ANSWER %d:' % i,
        dumpRR(u)
    for i in range(nscount):
        print 'AUTHORITY RECORD %d:' % i,
        dumpRR(u)
    for i in range(arcount):
        print 'ADDITIONAL RECORD %d:' % i,
        dumpRR(u)

class DnsResult:

    def __init__(self,u,args):
        self.header={}
        self.questions=[]
        self.answers=[]
        self.authority=[]
        self.additional=[]
        self.args=args
        self.storeM(u)

    def show(self):
        import time
        print '; <<>> PDG.py 1.0 <<>> %s %s'%(self.args['name'],
            self.args['qtype'])
        opt=""
        if self.args['rd']:
            opt=opt+'recurs '
        h=self.header
        print ';; options: '+opt
        print ';; got answer:'
        print ';; ->>HEADER<<- opcode %s, status %s, id %d'%(
            h['opcode'],h['status'],h['id'])
        flags=filter(lambda x,h=h:h[x],('qr','aa','rd','ra','tc'))
        print ';; flags: %s; Ques: %d, Ans: %d, Auth: %d, Addit: %d'%(
            ''.join(flags),h['qdcount'],h['ancount'],h['nscount'],
            h['arcount'])
        print ';; QUESTIONS:'
        for q in self.questions:
            print ';;      %s, type = %s, class = %s'%(q['qname'],q['qtypestr'],
                q['qclassstr'])
        print
        print ';; ANSWERS:'
        for a in self.answers:
            print '%-20s    %-6s  %-6s  %s'%(a['name'],`a['ttl']`,a['typename'],
                a['data'])
        print
        print ';; AUTHORITY RECORDS:'
        for a in self.authority:
            print '%-20s    %-6s  %-6s  %s'%(a['name'],`a['ttl']`,a['typename'],
                a['data'])
        print
        print ';; ADDITIONAL RECORDS:'
        for a in self.additional:
            print '%-20s    %-6s  %-6s  %s'%(a['name'],`a['ttl']`,a['typename'],
                a['data'])
        print
        if self.args.has_key('elapsed'):
            print ';; Total query time: %d msec'%self.args['elapsed']
        print ';; To SERVER: %s'%(self.args['server'])
        print ';; WHEN: %s'%time.ctime(time.time())

    def storeM(self,u):
        (self.header['id'], self.header['qr'], self.header['opcode'],
          self.header['aa'], self.header['tc'], self.header['rd'],
          self.header['ra'], self.header['z'], self.header['rcode'],
          self.header['qdcount'], self.header['ancount'],
          self.header['nscount'], self.header['arcount']) = u.getHeader()
        self.header['opcodestr']=Opcode.opcodestr(self.header['opcode'])
        self.header['status']=Status.statusstr(self.header['rcode'])
        for i in range(self.header['qdcount']):
            #print 'QUESTION %d:' % i,
            self.questions.append(self.storeQ(u))
        for i in range(self.header['ancount']):
            #print 'ANSWER %d:' % i,
            self.answers.append(self.storeRR(u))
        for i in range(self.header['nscount']):
            #print 'AUTHORITY RECORD %d:' % i,
            self.authority.append(self.storeRR(u))
        for i in range(self.header['arcount']):
            #print 'ADDITIONAL RECORD %d:' % i,
            self.additional.append(self.storeRR(u))

    def storeQ(self,u):
        q={}
        q['qname'], q['qtype'], q['qclass'] = u.getQuestion()
        q['qtypestr']=Type.typestr(q['qtype'])
        q['qclassstr']=Class.classstr(q['qclass'])
        return q

    def storeRR(self,u):
        r={}
        r['name'],r['type'],r['class'],r['ttl'],r['rdlength'] = u.getRRheader()
        r['typename'] = Type.typestr(r['type'])
        r['classstr'] = Class.classstr(r['class'])
        #print 'name=%s, type=%d(%s), class=%d(%s), ttl=%d' \
        #      % (name,
        #        type, typename,
        #        klass, Class.classstr(class),
        #        ttl)
        mname = 'get%sdata' % r['typename']
        if hasattr(u, mname):
            r['data']=getattr(u, mname)()
        else:
            r['data']=u.getbytes(r['rdlength'])
        return r

def dumpQ(u):
    qname, qtype, qclass = u.getQuestion()
    print 'qname=%s, qtype=%d(%s), qclass=%d(%s)' \
              % (qname,
                 qtype, Type.typestr(qtype),
                 qclass, Class.classstr(qclass))

def dumpRR(u):
    name, type, klass, ttl, rdlength = u.getRRheader()
    typename = Type.typestr(type)
    print 'name=%s, type=%d(%s), class=%d(%s), ttl=%d' \
              % (name,
                 type, typename,
                 klass, Class.classstr(klass),
                 ttl)
    mname = 'get%sdata' % typename
    if hasattr(u, mname):
        print '  formatted rdata:', getattr(u, mname)()
    else:
        print '  binary rdata:', u.getbytes(rdlength)

if __name__ == "__main__":
    testpacker()
#
# $Log: Lib.py,v $
# Revision 1.11.2.11  2011/11/23 17:07:14  customdesigned
# Rename some args in DNS.Lib to match py3dns.
#
# Revision 1.11.2.10  2011/08/04 22:23:55  customdesigned
# Allow empty label at end (trailing dot), so now '.' can lookup the
# root zone again.
#
# Revision 1.11.2.9  2011/03/21 21:03:22  customdesigned
# Get rid of obsolete string module
#
# Revision 1.11.2.8  2011/03/16 20:06:39  customdesigned
# Refer to explicit LICENSE file.
#
# Revision 1.11.2.7  2009/06/09 18:39:06  customdesigned
# Built-in SPF support
#
# Revision 1.11.2.6  2008/10/15 22:34:06  customdesigned
# Default to idna encoding.
#
# Revision 1.11.2.5  2008/09/17 17:35:14  customdesigned
# Use 7-bit ascii encoding, because case folding needs to be disabled
# before utf8 is safe to use, even experimentally.
#
# Revision 1.11.2.4  2008/09/17 16:09:53  customdesigned
# Encode unicode labels as UTF-8
#
# Revision 1.11.2.3  2007/05/22 20:27:40  customdesigned
# Fix unpacker underflow.
#
# Revision 1.11.2.2  2007/05/22 20:25:53  customdesigned
# Use socket.inetntoa,inetaton.
#
# Revision 1.11.2.1  2007/05/22 20:20:39  customdesigned
# Mark utf-8 encoding
#
# Revision 1.11  2002/03/19 13:05:02  anthonybaxter
# converted to class based exceptions (there goes the python1.4 compatibility :)
#
# removed a quite gross use of 'eval()'.
#
# Revision 1.10  2002/03/19 12:41:33  anthonybaxter
# tabnannied and reindented everything. 4 space indent, no tabs.
# yay.
#
# Revision 1.9  2002/03/19 10:30:33  anthonybaxter
# first round of major bits and pieces. The major stuff here (summarised
# from my local, off-net CVS server :/ this will cause some oddities with
# the
#
# tests/testPackers.py:
#   a large slab of unit tests for the packer and unpacker code in DNS.Lib
#
# DNS/Lib.py:
#   placeholder for addSRV.
#   added 'klass' to addA, make it the same as the other A* records.
#   made addTXT check for being passed a string, turn it into a length 1 list.
#   explicitly check for adding a string of length > 255 (prohibited).
#   a bunch of cleanups from a first pass with pychecker
#   new code for pack/unpack. the bitwise stuff uses struct, for a smallish
#     (disappointly small, actually) improvement, while addr2bin is much
#     much faster now.
#
# DNS/Base.py:
#   added DiscoverNameServers. This automatically does the right thing
#     on unix/ win32. No idea how MacOS handles this.  *sigh*
#     Incompatible change: Don't use ParseResolvConf on non-unix, use this
#     function, instead!
#   a bunch of cleanups from a first pass with pychecker
#
# Revision 1.8  2001/08/09 09:08:55  anthonybaxter
# added identifying header to top of each file
#
# Revision 1.7  2001/07/19 07:50:44  anthony
# Added SRV (RFC 2782) support. Code from Michael Ströder.
#
# Revision 1.6  2001/07/19 07:39:18  anthony
# 'type' -> 'rrtype' in getRRheader(). Fix from Michael Ströder.
#
# Revision 1.5  2001/07/19 07:34:19  anthony
# oops. glitch in storeRR (fixed now).
# Reported by Bastian Kleineidam and by greg lin.
#
# Revision 1.4  2001/07/19 07:16:42  anthony
# Changed (opcode&0xF)<<11 to (opcode*0xF)<<11.
# Patch from Timothy J. Miller.
#
# Revision 1.3  2001/07/19 06:57:07  anthony
# cvs keywords added
#
#