src/allmydata/download.py

   1
   2 import os, random, weakref, itertools, time
   3 from zope.interface import implements
   4 from twisted.internet import defer
   5 from twisted.internet.interfaces import IPushProducer, IConsumer
   6 from twisted.application import service
   7 from foolscap.eventual import eventually
   8
   9 from allmydata.util import base32, mathutil, hashutil, log
  10 from allmydata.util.assertutil import _assert
  11 from allmydata import codec, hashtree, storage, uri
  12 from allmydata.interfaces import IDownloadTarget, IDownloader, IFileURI, \
  13      IDownloadStatus, IDownloadResults
  14 from allmydata.encode import NotEnoughPeersError
  15 from pycryptopp.cipher.aes import AES
  16
  17 class HaveAllPeersError(Exception):
  18     # we use this to jump out of the loop
  19     pass
  20
  21 class BadURIExtensionHashValue(Exception):
  22     pass
  23 class BadPlaintextHashValue(Exception):
  24     pass
  25 class BadCrypttextHashValue(Exception):
  26     pass
  27
  28 class DownloadStopped(Exception):
  29     pass
  30
  31 class DownloadResults:
  32     implements(IDownloadResults)
  33
  34     def __init__(self):
  35         self.servers_used = set()
  36         self.server_problems = {}
  37         self.servermap = {}
  38         self.timings = {}
  39         self.file_size = None
  40
  41 class Output:
  42     def __init__(self, downloadable, key, total_length, log_parent,
  43                  download_status):
  44         self.downloadable = downloadable
  45         self._decryptor = AES(key)
  46         self._crypttext_hasher = hashutil.crypttext_hasher()
  47         self._plaintext_hasher = hashutil.plaintext_hasher()
  48         self.length = 0
  49         self.total_length = total_length
  50         self._segment_number = 0
  51         self._plaintext_hash_tree = None
  52         self._crypttext_hash_tree = None
  53         self._opened = False
  54         self._log_parent = log_parent
  55         self._status = download_status
  56         self._status.set_progress(0.0)
  57
  58     def log(self, *args, **kwargs):
  59         if "parent" not in kwargs:
  60             kwargs["parent"] = self._log_parent
  61         if "facility" not in kwargs:
  62             kwargs["facility"] = "download.output"
  63         return log.msg(*args, **kwargs)
  64
  65     def setup_hashtrees(self, plaintext_hashtree, crypttext_hashtree):
  66         self._plaintext_hash_tree = plaintext_hashtree
  67         self._crypttext_hash_tree = crypttext_hashtree
  68
  69     def write_segment(self, crypttext):
  70         self.length += len(crypttext)
  71         self._status.set_progress( float(self.length) / self.total_length )
  72
  73         # memory footprint: 'crypttext' is the only segment_size usage
  74         # outstanding. While we decrypt it into 'plaintext', we hit
  75         # 2*segment_size.
  76         self._crypttext_hasher.update(crypttext)
  77         if self._crypttext_hash_tree:
  78             ch = hashutil.crypttext_segment_hasher()
  79             ch.update(crypttext)
  80             crypttext_leaves = {self._segment_number: ch.digest()}
  81             self.log(format="crypttext leaf hash (%(bytes)sB) [%(segnum)d] is %(hash)s",
  82                      bytes=len(crypttext),
  83                      segnum=self._segment_number, hash=base32.b2a(ch.digest()),
  84                      level=log.NOISY)
  85             self._crypttext_hash_tree.set_hashes(leaves=crypttext_leaves)
  86
  87         plaintext = self._decryptor.process(crypttext)
  88         del crypttext
  89
  90         # now we're back down to 1*segment_size.
  91
  92         self._plaintext_hasher.update(plaintext)
  93         if self._plaintext_hash_tree:
  94             ph = hashutil.plaintext_segment_hasher()
  95             ph.update(plaintext)
  96             plaintext_leaves = {self._segment_number: ph.digest()}
  97             self.log(format="plaintext leaf hash (%(bytes)sB) [%(segnum)d] is %(hash)s",
  98                      bytes=len(plaintext),
  99                      segnum=self._segment_number, hash=base32.b2a(ph.digest()),
 100                      level=log.NOISY)
 101             self._plaintext_hash_tree.set_hashes(leaves=plaintext_leaves)
 102
 103         self._segment_number += 1
 104         # We're still at 1*segment_size. The Downloadable is responsible for
 105         # any memory usage beyond this.
 106         if not self._opened:
 107             self._opened = True
 108             self.downloadable.open(self.total_length)
 109         self.downloadable.write(plaintext)
 110
 111     def fail(self, why):
 112         # this is really unusual, and deserves maximum forensics
 113         self.log("download failed!", failure=why, level=log.SCARY)
 114         self.downloadable.fail(why)
 115
 116     def close(self):
 117         self.crypttext_hash = self._crypttext_hasher.digest()
 118         self.plaintext_hash = self._plaintext_hasher.digest()
 119         self.log("download finished, closing IDownloadable", level=log.NOISY)
 120         self.downloadable.close()
 121
 122     def finish(self):
 123         return self.downloadable.finish()
 124
 125 class ValidatedBucket:
 126     """I am a front-end for a remote storage bucket, responsible for
 127     retrieving and validating data from that bucket.
 128
 129     My get_block() method is used by BlockDownloaders.
 130     """
 131
 132     def __init__(self, sharenum, bucket,
 133                  share_hash_tree, roothash,
 134                  num_blocks):
 135         self.sharenum = sharenum
 136         self.bucket = bucket
 137         self._share_hash = None # None means not validated yet
 138         self.share_hash_tree = share_hash_tree
 139         self._roothash = roothash
 140         self.block_hash_tree = hashtree.IncompleteHashTree(num_blocks)
 141         self.started = False
 142
 143     def get_block(self, blocknum):
 144         if not self.started:
 145             d = self.bucket.start()
 146             def _started(res):
 147                 self.started = True
 148                 return self.get_block(blocknum)
 149             d.addCallback(_started)
 150             return d
 151
 152         # the first time we use this bucket, we need to fetch enough elements
 153         # of the share hash tree to validate it from our share hash up to the
 154         # hashroot.
 155         if not self._share_hash:
 156             d1 = self.bucket.get_share_hashes()
 157         else:
 158             d1 = defer.succeed([])
 159
 160         # we might need to grab some elements of our block hash tree, to
 161         # validate the requested block up to the share hash
 162         needed = self.block_hash_tree.needed_hashes(blocknum)
 163         if needed:
 164             # TODO: get fewer hashes, use get_block_hashes(needed)
 165             d2 = self.bucket.get_block_hashes()
 166         else:
 167             d2 = defer.succeed([])
 168
 169         d3 = self.bucket.get_block(blocknum)
 170
 171         d = defer.gatherResults([d1, d2, d3])
 172         d.addCallback(self._got_data, blocknum)
 173         return d
 174
 175     def _got_data(self, res, blocknum):
 176         sharehashes, blockhashes, blockdata = res
 177         blockhash = None # to make logging it safe
 178
 179         try:
 180             if not self._share_hash:
 181                 sh = dict(sharehashes)
 182                 sh[0] = self._roothash # always use our own root, from the URI
 183                 sht = self.share_hash_tree
 184                 if sht.get_leaf_index(self.sharenum) not in sh:
 185                     raise hashtree.NotEnoughHashesError
 186                 sht.set_hashes(sh)
 187                 self._share_hash = sht.get_leaf(self.sharenum)
 188
 189             blockhash = hashutil.block_hash(blockdata)
 190             #log.msg("checking block_hash(shareid=%d, blocknum=%d) len=%d "
 191             #        "%r .. %r: %s" %
 192             #        (self.sharenum, blocknum, len(blockdata),
 193             #         blockdata[:50], blockdata[-50:], base32.b2a(blockhash)))
 194
 195             # we always validate the blockhash
 196             bh = dict(enumerate(blockhashes))
 197             # replace blockhash root with validated value
 198             bh[0] = self._share_hash
 199             self.block_hash_tree.set_hashes(bh, {blocknum: blockhash})
 200
 201         except (hashtree.BadHashError, hashtree.NotEnoughHashesError):
 202             # log.WEIRD: indicates undetected disk/network error, or more
 203             # likely a programming error
 204             log.msg("hash failure in block=%d, shnum=%d on %s" %
 205                     (blocknum, self.sharenum, self.bucket))
 206             if self._share_hash:
 207                 log.msg(""" failure occurred when checking the block_hash_tree.
 208                 This suggests that either the block data was bad, or that the
 209                 block hashes we received along with it were bad.""")
 210             else:
 211                 log.msg(""" the failure probably occurred when checking the
 212                 share_hash_tree, which suggests that the share hashes we
 213                 received from the remote peer were bad.""")
 214             log.msg(" have self._share_hash: %s" % bool(self._share_hash))
 215             log.msg(" block length: %d" % len(blockdata))
 216             log.msg(" block hash: %s" % base32.b2a_or_none(blockhash))
 217             if len(blockdata) < 100:
 218                 log.msg(" block data: %r" % (blockdata,))
 219             else:
 220                 log.msg(" block data start/end: %r .. %r" %
 221                         (blockdata[:50], blockdata[-50:]))
 222             log.msg(" root hash: %s" % base32.b2a(self._roothash))
 223             log.msg(" share hash tree:\n" + self.share_hash_tree.dump())
 224             log.msg(" block hash tree:\n" + self.block_hash_tree.dump())
 225             lines = []
 226             for i,h in sorted(sharehashes):
 227                 lines.append("%3d: %s" % (i, base32.b2a_or_none(h)))
 228             log.msg(" sharehashes:\n" + "\n".join(lines) + "\n")
 229             lines = []
 230             for i,h in enumerate(blockhashes):
 231                 lines.append("%3d: %s" % (i, base32.b2a_or_none(h)))
 232             log.msg(" blockhashes:\n" + "\n".join(lines) + "\n")
 233             raise
 234
 235         # If we made it here, the block is good. If the hash trees didn't
 236         # like what they saw, they would have raised a BadHashError, causing
 237         # our caller to see a Failure and thus ignore this block (as well as
 238         # dropping this bucket).
 239         return blockdata
 240
 241
 242
 243 class BlockDownloader:
 244     """I am responsible for downloading a single block (from a single bucket)
 245     for a single segment.
 246
 247     I am a child of the SegmentDownloader.
 248     """
 249
 250     def __init__(self, vbucket, blocknum, parent, results):
 251         self.vbucket = vbucket
 252         self.blocknum = blocknum
 253         self.parent = parent
 254         self.results = results
 255         self._log_number = self.parent.log("starting block %d" % blocknum)
 256
 257     def log(self, msg, parent=None):
 258         if parent is None:
 259             parent = self._log_number
 260         return self.parent.log(msg, parent=parent)
 261
 262     def start(self, segnum):
 263         lognum = self.log("get_block(segnum=%d)" % segnum)
 264         started = time.time()
 265         d = self.vbucket.get_block(segnum)
 266         d.addCallbacks(self._hold_block, self._got_block_error,
 267                        callbackArgs=(started, lognum,), errbackArgs=(lognum,))
 268         return d
 269
 270     def _hold_block(self, data, started, lognum):
 271         if self.results:
 272             elapsed = time.time() - started
 273             peerid = self.vbucket.bucket.get_peerid()
 274             if peerid not in self.results.timings["fetch_per_server"]:
 275                 self.results.timings["fetch_per_server"][peerid] = []
 276             self.results.timings["fetch_per_server"][peerid].append(elapsed)
 277         self.log("got block", parent=lognum)
 278         self.parent.hold_block(self.blocknum, data)
 279
 280     def _got_block_error(self, f, lognum):
 281         self.log("BlockDownloader[%d] got error: %s" % (self.blocknum, f),
 282                  parent=lognum)
 283         if self.results:
 284             peerid = self.vbucket.bucket.get_peerid()
 285             self.results.server_problems[peerid] = str(f)
 286         self.parent.bucket_failed(self.vbucket)
 287
 288 class SegmentDownloader:
 289     """I am responsible for downloading all the blocks for a single segment
 290     of data.
 291
 292     I am a child of the FileDownloader.
 293     """
 294
 295     def __init__(self, parent, segmentnumber, needed_shares, results):
 296         self.parent = parent
 297         self.segmentnumber = segmentnumber
 298         self.needed_blocks = needed_shares
 299         self.blocks = {} # k: blocknum, v: data
 300         self.results = results
 301         self._log_number = self.parent.log("starting segment %d" %
 302                                            segmentnumber)
 303
 304     def log(self, msg, parent=None):
 305         if parent is None:
 306             parent = self._log_number
 307         return self.parent.log(msg, parent=parent)
 308
 309     def start(self):
 310         return self._download()
 311
 312     def _download(self):
 313         d = self._try()
 314         def _done(res):
 315             if len(self.blocks) >= self.needed_blocks:
 316                 # we only need self.needed_blocks blocks
 317                 # we want to get the smallest blockids, because they are
 318                 # more likely to be fast "primary blocks"
 319                 blockids = sorted(self.blocks.keys())[:self.needed_blocks]
 320                 blocks = []
 321                 for blocknum in blockids:
 322                     blocks.append(self.blocks[blocknum])
 323                 return (blocks, blockids)
 324             else:
 325                 return self._download()
 326         d.addCallback(_done)
 327         return d
 328
 329     def _try(self):
 330         # fill our set of active buckets, maybe raising NotEnoughPeersError
 331         active_buckets = self.parent._activate_enough_buckets()
 332         # Now we have enough buckets, in self.parent.active_buckets.
 333
 334         # in test cases, bd.start might mutate active_buckets right away, so
 335         # we need to put off calling start() until we've iterated all the way
 336         # through it.
 337         downloaders = []
 338         for blocknum, vbucket in active_buckets.iteritems():
 339             bd = BlockDownloader(vbucket, blocknum, self, self.results)
 340             downloaders.append(bd)
 341             if self.results:
 342                 self.results.servers_used.add(vbucket.bucket.get_peerid())
 343         l = [bd.start(self.segmentnumber) for bd in downloaders]
 344         return defer.DeferredList(l, fireOnOneErrback=True)
 345
 346     def hold_block(self, blocknum, data):
 347         self.blocks[blocknum] = data
 348
 349     def bucket_failed(self, vbucket):
 350         self.parent.bucket_failed(vbucket)
 351
 352 class DownloadStatus:
 353     implements(IDownloadStatus)
 354     statusid_counter = itertools.count(0)
 355
 356     def __init__(self):
 357         self.storage_index = None
 358         self.size = None
 359         self.helper = False
 360         self.status = "Not started"
 361         self.progress = 0.0
 362         self.paused = False
 363         self.stopped = False
 364         self.active = True
 365         self.results = None
 366         self.counter = self.statusid_counter.next()
 367         self.started = time.time()
 368
 369     def get_started(self):
 370         return self.started
 371     def get_storage_index(self):
 372         return self.storage_index
 373     def get_size(self):
 374         return self.size
 375     def using_helper(self):
 376         return self.helper
 377     def get_status(self):
 378         status = self.status
 379         if self.paused:
 380             status += " (output paused)"
 381         if self.stopped:
 382             status += " (output stopped)"
 383         return status
 384     def get_progress(self):
 385         return self.progress
 386     def get_active(self):
 387         return self.active
 388     def get_results(self):
 389         return self.results
 390     def get_counter(self):
 391         return self.counter
 392
 393     def set_storage_index(self, si):
 394         self.storage_index = si
 395     def set_size(self, size):
 396         self.size = size
 397     def set_helper(self, helper):
 398         self.helper = helper
 399     def set_status(self, status):
 400         self.status = status
 401     def set_paused(self, paused):
 402         self.paused = paused
 403     def set_stopped(self, stopped):
 404         self.stopped = stopped
 405     def set_progress(self, value):
 406         self.progress = value
 407     def set_active(self, value):
 408         self.active = value
 409     def set_results(self, value):
 410         self.results = value
 411
 412 class FileDownloader:
 413     implements(IPushProducer)
 414     check_crypttext_hash = True
 415     check_plaintext_hash = True
 416     _status = None
 417
 418     def __init__(self, client, u, downloadable):
 419         self._client = client
 420
 421         u = IFileURI(u)
 422         self._storage_index = u.storage_index
 423         self._uri_extension_hash = u.uri_extension_hash
 424         self._total_shares = u.total_shares
 425         self._size = u.size
 426         self._num_needed_shares = u.needed_shares
 427
 428         self._si_s = storage.si_b2a(self._storage_index)
 429         self.init_logging()
 430
 431         self._started = time.time()
 432         self._status = s = DownloadStatus()
 433         s.set_status("Starting")
 434         s.set_storage_index(self._storage_index)
 435         s.set_size(self._size)
 436         s.set_helper(False)
 437         s.set_active(True)
 438
 439         self._results = DownloadResults()
 440         s.set_results(self._results)
 441         self._results.file_size = self._size
 442         self._results.timings["servers_peer_selection"] = {}
 443         self._results.timings["fetch_per_server"] = {}
 444         self._results.timings["cumulative_fetch"] = 0.0
 445         self._results.timings["cumulative_decode"] = 0.0
 446         self._results.timings["cumulative_decrypt"] = 0.0
 447
 448         if IConsumer.providedBy(downloadable):
 449             downloadable.registerProducer(self, True)
 450         self._downloadable = downloadable
 451         self._output = Output(downloadable, u.key, self._size, self._log_number,
 452                               self._status)
 453         self._paused = False
 454         self._stopped = False
 455
 456         self.active_buckets = {} # k: shnum, v: bucket
 457         self._share_buckets = [] # list of (sharenum, bucket) tuples
 458         self._share_vbuckets = {} # k: shnum, v: set of ValidatedBuckets
 459         self._uri_extension_sources = []
 460
 461         self._uri_extension_data = None
 462
 463         self._fetch_failures = {"uri_extension": 0,
 464                                 "plaintext_hashroot": 0,
 465                                 "plaintext_hashtree": 0,
 466                                 "crypttext_hashroot": 0,
 467                                 "crypttext_hashtree": 0,
 468                                 }
 469
 470     def init_logging(self):
 471         self._log_prefix = prefix = storage.si_b2a(self._storage_index)[:5]
 472         num = self._client.log(format="FileDownloader(%(si)s): starting",
 473                                si=storage.si_b2a(self._storage_index))
 474         self._log_number = num
 475
 476     def log(self, *args, **kwargs):
 477         if "parent" not in kwargs:
 478             kwargs["parent"] = self._log_number
 479         if "facility" not in kwargs:
 480             kwargs["facility"] = "tahoe.download"
 481         return log.msg(*args, **kwargs)
 482
 483     def pauseProducing(self):
 484         if self._paused:
 485             return
 486         self._paused = defer.Deferred()
 487         if self._status:
 488             self._status.set_paused(True)
 489
 490     def resumeProducing(self):
 491         if self._paused:
 492             p = self._paused
 493             self._paused = None
 494             eventually(p.callback, None)
 495             if self._status:
 496                 self._status.set_paused(False)
 497
 498     def stopProducing(self):
 499         self.log("Download.stopProducing")
 500         self._stopped = True
 501         if self._status:
 502             self._status.set_stopped(True)
 503             self._status.set_active(False)
 504
 505     def start(self):
 506         self.log("starting download")
 507
 508         # first step: who should we download from?
 509         d = defer.maybeDeferred(self._get_all_shareholders)
 510         d.addCallback(self._got_all_shareholders)
 511         # now get the uri_extension block from somebody and validate it
 512         d.addCallback(self._obtain_uri_extension)
 513         d.addCallback(self._got_uri_extension)
 514         d.addCallback(self._get_hashtrees)
 515         d.addCallback(self._create_validated_buckets)
 516         # once we know that, we can download blocks from everybody
 517         d.addCallback(self._download_all_segments)
 518         def _finished(res):
 519             if self._status:
 520                 self._status.set_status("Finished")
 521                 self._status.set_active(False)
 522                 self._status.set_paused(False)
 523             if IConsumer.providedBy(self._downloadable):
 524                 self._downloadable.unregisterProducer()
 525             return res
 526         d.addBoth(_finished)
 527         def _failed(why):
 528             if self._status:
 529                 self._status.set_status("Failed")
 530                 self._status.set_active(False)
 531             self._output.fail(why)
 532             return why
 533         d.addErrback(_failed)
 534         d.addCallback(self._done)
 535         return d
 536
 537     def _get_all_shareholders(self):
 538         dl = []
 539         for (peerid,ss) in self._client.get_permuted_peers("storage",
 540                                                            self._storage_index):
 541             d = ss.callRemote("get_buckets", self._storage_index)
 542             d.addCallbacks(self._got_response, self._got_error,
 543                            callbackArgs=(peerid,))
 544             dl.append(d)
 545         self._responses_received = 0
 546         self._queries_sent = len(dl)
 547         if self._status:
 548             self._status.set_status("Locating Shares (%d/%d)" %
 549                                     (self._responses_received,
 550                                      self._queries_sent))
 551         return defer.DeferredList(dl)
 552
 553     def _got_response(self, buckets, peerid):
 554         self._responses_received += 1
 555         if self._results:
 556             elapsed = time.time() - self._started
 557             self._results.timings["servers_peer_selection"][peerid] = elapsed
 558         if self._status:
 559             self._status.set_status("Locating Shares (%d/%d)" %
 560                                     (self._responses_received,
 561                                      self._queries_sent))
 562         for sharenum, bucket in buckets.iteritems():
 563             b = storage.ReadBucketProxy(bucket, peerid, self._si_s)
 564             self.add_share_bucket(sharenum, b)
 565             self._uri_extension_sources.append(b)
 566             if self._results:
 567                 if peerid not in self._results.servermap:
 568                     self._results.servermap[peerid] = set()
 569                 self._results.servermap[peerid].add(sharenum)
 570
 571     def add_share_bucket(self, sharenum, bucket):
 572         # this is split out for the benefit of test_encode.py
 573         self._share_buckets.append( (sharenum, bucket) )
 574
 575     def _got_error(self, f):
 576         self._client.log("Somebody failed. -- %s" % (f,))
 577
 578     def bucket_failed(self, vbucket):
 579         shnum = vbucket.sharenum
 580         del self.active_buckets[shnum]
 581         s = self._share_vbuckets[shnum]
 582         # s is a set of ValidatedBucket instances
 583         s.remove(vbucket)
 584         # ... which might now be empty
 585         if not s:
 586             # there are no more buckets which can provide this share, so
 587             # remove the key. This may prompt us to use a different share.
 588             del self._share_vbuckets[shnum]
 589
 590     def _got_all_shareholders(self, res):
 591         if self._results:
 592             now = time.time()
 593             self._results.timings["peer_selection"] = now - self._started
 594
 595         if len(self._share_buckets) < self._num_needed_shares:
 596             raise NotEnoughPeersError
 597
 598         #for s in self._share_vbuckets.values():
 599         #    for vb in s:
 600         #        assert isinstance(vb, ValidatedBucket), \
 601         #               "vb is %s but should be a ValidatedBucket" % (vb,)
 602
 603     def _unpack_uri_extension_data(self, data):
 604         return uri.unpack_extension(data)
 605
 606     def _obtain_uri_extension(self, ignored):
 607         # all shareholders are supposed to have a copy of uri_extension, and
 608         # all are supposed to be identical. We compute the hash of the data
 609         # that comes back, and compare it against the version in our URI. If
 610         # they don't match, ignore their data and try someone else.
 611         if self._status:
 612             self._status.set_status("Obtaining URI Extension")
 613
 614         self._uri_extension_fetch_started = time.time()
 615         def _validate(proposal, bucket):
 616             h = hashutil.uri_extension_hash(proposal)
 617             if h != self._uri_extension_hash:
 618                 self._fetch_failures["uri_extension"] += 1
 619                 msg = ("The copy of uri_extension we received from "
 620                        "%s was bad: wanted %s, got %s" %
 621                        (bucket,
 622                         base32.b2a(self._uri_extension_hash),
 623                         base32.b2a(h)))
 624                 self.log(msg, level=log.SCARY)
 625                 raise BadURIExtensionHashValue(msg)
 626             return self._unpack_uri_extension_data(proposal)
 627         return self._obtain_validated_thing(None,
 628                                             self._uri_extension_sources,
 629                                             "uri_extension",
 630                                             "get_uri_extension", (), _validate)
 631
 632     def _obtain_validated_thing(self, ignored, sources, name, methname, args,
 633                                 validatorfunc):
 634         if not sources:
 635             raise NotEnoughPeersError("started with zero peers while fetching "
 636                                       "%s" % name)
 637         bucket = sources[0]
 638         sources = sources[1:]
 639         #d = bucket.callRemote(methname, *args)
 640         d = bucket.startIfNecessary()
 641         d.addCallback(lambda res: getattr(bucket, methname)(*args))
 642         d.addCallback(validatorfunc, bucket)
 643         def _bad(f):
 644             self.log("%s from vbucket %s failed:" % (name, bucket),
 645                      failure=f, level=log.WEIRD)
 646             if not sources:
 647                 raise NotEnoughPeersError("ran out of peers, last error was %s"
 648                                           % (f,))
 649             # try again with a different one
 650             return self._obtain_validated_thing(None, sources, name,
 651                                                 methname, args, validatorfunc)
 652         d.addErrback(_bad)
 653         return d
 654
 655     def _got_uri_extension(self, uri_extension_data):
 656         if self._results:
 657             elapsed = time.time() - self._uri_extension_fetch_started
 658             self._results.timings["uri_extension"] = elapsed
 659
 660         d = self._uri_extension_data = uri_extension_data
 661
 662         self._codec = codec.get_decoder_by_name(d['codec_name'])
 663         self._codec.set_serialized_params(d['codec_params'])
 664         self._tail_codec = codec.get_decoder_by_name(d['codec_name'])
 665         self._tail_codec.set_serialized_params(d['tail_codec_params'])
 666
 667         crypttext_hash = d['crypttext_hash']
 668         assert isinstance(crypttext_hash, str)
 669         assert len(crypttext_hash) == 32
 670         self._crypttext_hash = crypttext_hash
 671         self._plaintext_hash = d['plaintext_hash']
 672         self._roothash = d['share_root_hash']
 673
 674         self._segment_size = segment_size = d['segment_size']
 675         self._total_segments = mathutil.div_ceil(self._size, segment_size)
 676         self._current_segnum = 0
 677
 678         self._share_hashtree = hashtree.IncompleteHashTree(d['total_shares'])
 679         self._share_hashtree.set_hashes({0: self._roothash})
 680
 681     def _get_hashtrees(self, res):
 682         self._get_hashtrees_started = time.time()
 683         if self._status:
 684             self._status.set_status("Retrieving Hash Trees")
 685         d = self._get_plaintext_hashtrees()
 686         d.addCallback(self._get_crypttext_hashtrees)
 687         d.addCallback(self._setup_hashtrees)
 688         return d
 689
 690     def _get_plaintext_hashtrees(self):
 691         def _validate_plaintext_hashtree(proposal, bucket):
 692             if proposal[0] != self._uri_extension_data['plaintext_root_hash']:
 693                 self._fetch_failures["plaintext_hashroot"] += 1
 694                 msg = ("The copy of the plaintext_root_hash we received from"
 695                        " %s was bad" % bucket)
 696                 raise BadPlaintextHashValue(msg)
 697             pt_hashtree = hashtree.IncompleteHashTree(self._total_segments)
 698             pt_hashes = dict(list(enumerate(proposal)))
 699             try:
 700                 pt_hashtree.set_hashes(pt_hashes)
 701             except hashtree.BadHashError:
 702                 # the hashes they gave us were not self-consistent, even
 703                 # though the root matched what we saw in the uri_extension
 704                 # block
 705                 self._fetch_failures["plaintext_hashtree"] += 1
 706                 raise
 707             self._plaintext_hashtree = pt_hashtree
 708         d = self._obtain_validated_thing(None,
 709                                          self._uri_extension_sources,
 710                                          "plaintext_hashes",
 711                                          "get_plaintext_hashes", (),
 712                                          _validate_plaintext_hashtree)
 713         return d
 714
 715     def _get_crypttext_hashtrees(self, res):
 716         def _validate_crypttext_hashtree(proposal, bucket):
 717             if proposal[0] != self._uri_extension_data['crypttext_root_hash']:
 718                 self._fetch_failures["crypttext_hashroot"] += 1
 719                 msg = ("The copy of the crypttext_root_hash we received from"
 720                        " %s was bad" % bucket)
 721                 raise BadCrypttextHashValue(msg)
 722             ct_hashtree = hashtree.IncompleteHashTree(self._total_segments)
 723             ct_hashes = dict(list(enumerate(proposal)))
 724             try:
 725                 ct_hashtree.set_hashes(ct_hashes)
 726             except hashtree.BadHashError:
 727                 self._fetch_failures["crypttext_hashtree"] += 1
 728                 raise
 729             ct_hashtree.set_hashes(ct_hashes)
 730             self._crypttext_hashtree = ct_hashtree
 731         d = self._obtain_validated_thing(None,
 732                                          self._uri_extension_sources,
 733                                          "crypttext_hashes",
 734                                          "get_crypttext_hashes", (),
 735                                          _validate_crypttext_hashtree)
 736         return d
 737
 738     def _setup_hashtrees(self, res):
 739         self._output.setup_hashtrees(self._plaintext_hashtree,
 740                                      self._crypttext_hashtree)
 741         if self._results:
 742             elapsed = time.time() - self._get_hashtrees_started
 743             self._results.timings["hashtrees"] = elapsed
 744
 745     def _create_validated_buckets(self, ignored=None):
 746         self._share_vbuckets = {}
 747         for sharenum, bucket in self._share_buckets:
 748             vbucket = ValidatedBucket(sharenum, bucket,
 749                                       self._share_hashtree,
 750                                       self._roothash,
 751                                       self._total_segments)
 752             s = self._share_vbuckets.setdefault(sharenum, set())
 753             s.add(vbucket)
 754
 755     def _activate_enough_buckets(self):
 756         """either return a mapping from shnum to a ValidatedBucket that can
 757         provide data for that share, or raise NotEnoughPeersError"""
 758
 759         while len(self.active_buckets) < self._num_needed_shares:
 760             # need some more
 761             handled_shnums = set(self.active_buckets.keys())
 762             available_shnums = set(self._share_vbuckets.keys())
 763             potential_shnums = list(available_shnums - handled_shnums)
 764             if not potential_shnums:
 765                 raise NotEnoughPeersError
 766             # choose a random share
 767             shnum = random.choice(potential_shnums)
 768             # and a random bucket that will provide it
 769             validated_bucket = random.choice(list(self._share_vbuckets[shnum]))
 770             self.active_buckets[shnum] = validated_bucket
 771         return self.active_buckets
 772
 773
 774     def _download_all_segments(self, res):
 775         # the promise: upon entry to this function, self._share_vbuckets
 776         # contains enough buckets to complete the download, and some extra
 777         # ones to tolerate some buckets dropping out or having errors.
 778         # self._share_vbuckets is a dictionary that maps from shnum to a set
 779         # of ValidatedBuckets, which themselves are wrappers around
 780         # RIBucketReader references.
 781         self.active_buckets = {} # k: shnum, v: ValidatedBucket instance
 782
 783         self._started_fetching = time.time()
 784
 785         d = defer.succeed(None)
 786         for segnum in range(self._total_segments-1):
 787             d.addCallback(self._download_segment, segnum)
 788             # this pause, at the end of write, prevents pre-fetch from
 789             # happening until the consumer is ready for more data.
 790             d.addCallback(self._check_for_pause)
 791         d.addCallback(self._download_tail_segment, self._total_segments-1)
 792         return d
 793
 794     def _check_for_pause(self, res):
 795         if self._paused:
 796             d = defer.Deferred()
 797             self._paused.addCallback(lambda ignored: d.callback(res))
 798             return d
 799         if self._stopped:
 800             raise DownloadStopped("our Consumer called stopProducing()")
 801         return res
 802
 803     def _download_segment(self, res, segnum):
 804         if self._status:
 805             self._status.set_status("Downloading segment %d of %d" %
 806                                     (segnum+1, self._total_segments))
 807         self.log("downloading seg#%d of %d (%d%%)"
 808                  % (segnum, self._total_segments,
 809                     100.0 * segnum / self._total_segments))
 810         # memory footprint: when the SegmentDownloader finishes pulling down
 811         # all shares, we have 1*segment_size of usage.
 812         segmentdler = SegmentDownloader(self, segnum, self._num_needed_shares,
 813                                         self._results)
 814         started = time.time()
 815         d = segmentdler.start()
 816         def _finished_fetching(res):
 817             elapsed = time.time() - started
 818             self._results.timings["cumulative_fetch"] += elapsed
 819             return res
 820         if self._results:
 821             d.addCallback(_finished_fetching)
 822         # pause before using more memory
 823         d.addCallback(self._check_for_pause)
 824         # while the codec does its job, we hit 2*segment_size
 825         def _started_decode(res):
 826             self._started_decode = time.time()
 827             return res
 828         if self._results:
 829             d.addCallback(_started_decode)
 830         d.addCallback(lambda (shares, shareids):
 831                       self._codec.decode(shares, shareids))
 832         # once the codec is done, we drop back to 1*segment_size, because
 833         # 'shares' goes out of scope. The memory usage is all in the
 834         # plaintext now, spread out into a bunch of tiny buffers.
 835         def _finished_decode(res):
 836             elapsed = time.time() - self._started_decode
 837             self._results.timings["cumulative_decode"] += elapsed
 838             return res
 839         if self._results:
 840             d.addCallback(_finished_decode)
 841
 842         # pause/check-for-stop just before writing, to honor stopProducing
 843         d.addCallback(self._check_for_pause)
 844         def _done(buffers):
 845             # we start by joining all these buffers together into a single
 846             # string. This makes Output.write easier, since it wants to hash
 847             # data one segment at a time anyways, and doesn't impact our
 848             # memory footprint since we're already peaking at 2*segment_size
 849             # inside the codec a moment ago.
 850             segment = "".join(buffers)
 851             del buffers
 852             # we're down to 1*segment_size right now, but write_segment()
 853             # will decrypt a copy of the segment internally, which will push
 854             # us up to 2*segment_size while it runs.
 855             started_decrypt = time.time()
 856             self._output.write_segment(segment)
 857             if self._results:
 858                 elapsed = time.time() - started_decrypt
 859                 self._results.timings["cumulative_decrypt"] += elapsed
 860         d.addCallback(_done)
 861         return d
 862
 863     def _download_tail_segment(self, res, segnum):
 864         self.log("downloading seg#%d of %d (%d%%)"
 865                  % (segnum, self._total_segments,
 866                     100.0 * segnum / self._total_segments))
 867         segmentdler = SegmentDownloader(self, segnum, self._num_needed_shares,
 868                                         self._results)
 869         started = time.time()
 870         d = segmentdler.start()
 871         def _finished_fetching(res):
 872             elapsed = time.time() - started
 873             self._results.timings["cumulative_fetch"] += elapsed
 874             return res
 875         if self._results:
 876             d.addCallback(_finished_fetching)
 877         # pause before using more memory
 878         d.addCallback(self._check_for_pause)
 879         def _started_decode(res):
 880             self._started_decode = time.time()
 881             return res
 882         if self._results:
 883             d.addCallback(_started_decode)
 884         d.addCallback(lambda (shares, shareids):
 885                       self._tail_codec.decode(shares, shareids))
 886         def _finished_decode(res):
 887             elapsed = time.time() - self._started_decode
 888             self._results.timings["cumulative_decode"] += elapsed
 889             return res
 890         if self._results:
 891             d.addCallback(_finished_decode)
 892         # pause/check-for-stop just before writing, to honor stopProducing
 893         d.addCallback(self._check_for_pause)
 894         def _done(buffers):
 895             # trim off any padding added by the upload side
 896             segment = "".join(buffers)
 897             del buffers
 898             # we never send empty segments. If the data was an exact multiple
 899             # of the segment size, the last segment will be full.
 900             pad_size = mathutil.pad_size(self._size, self._segment_size)
 901             tail_size = self._segment_size - pad_size
 902             segment = segment[:tail_size]
 903             started_decrypt = time.time()
 904             self._output.write_segment(segment)
 905             if self._results:
 906                 elapsed = time.time() - started_decrypt
 907                 self._results.timings["cumulative_decrypt"] += elapsed
 908         d.addCallback(_done)
 909         return d
 910
 911     def _done(self, res):
 912         self.log("download done")
 913         if self._results:
 914             now = time.time()
 915             self._results.timings["total"] = now - self._started
 916             self._results.timings["segments"] = now - self._started_fetching
 917         self._output.close()
 918         if self.check_crypttext_hash:
 919             _assert(self._crypttext_hash == self._output.crypttext_hash,
 920                     "bad crypttext_hash: computed=%s, expected=%s" %
 921                     (base32.b2a(self._output.crypttext_hash),
 922                      base32.b2a(self._crypttext_hash)))
 923         if self.check_plaintext_hash:
 924             _assert(self._plaintext_hash == self._output.plaintext_hash,
 925                     "bad plaintext_hash: computed=%s, expected=%s" %
 926                     (base32.b2a(self._output.plaintext_hash),
 927                      base32.b2a(self._plaintext_hash)))
 928         _assert(self._output.length == self._size,
 929                 got=self._output.length, expected=self._size)
 930         return self._output.finish()
 931
 932     def get_download_status(self):
 933         return self._status
 934
 935
 936 class LiteralDownloader:
 937     def __init__(self, client, u, downloadable):
 938         self._uri = IFileURI(u)
 939         assert isinstance(self._uri, uri.LiteralFileURI)
 940         self._downloadable = downloadable
 941         self._status = s = DownloadStatus()
 942         s.set_storage_index(None)
 943         s.set_helper(False)
 944         s.set_status("Done")
 945         s.set_active(False)
 946         s.set_progress(1.0)
 947
 948     def start(self):
 949         data = self._uri.data
 950         self._status.set_size(len(data))
 951         self._downloadable.open(len(data))
 952         self._downloadable.write(data)
 953         self._downloadable.close()
 954         return defer.maybeDeferred(self._downloadable.finish)
 955
 956     def get_download_status(self):
 957         return self._status
 958
 959 class FileName:
 960     implements(IDownloadTarget)
 961     def __init__(self, filename):
 962         self._filename = filename
 963         self.f = None
 964     def open(self, size):
 965         self.f = open(self._filename, "wb")
 966         return self.f
 967     def write(self, data):
 968         self.f.write(data)
 969     def close(self):
 970         if self.f:
 971             self.f.close()
 972     def fail(self, why):
 973         if self.f:
 974             self.f.close()
 975             os.unlink(self._filename)
 976     def register_canceller(self, cb):
 977         pass # we won't use it
 978     def finish(self):
 979         pass
 980
 981 class Data:
 982     implements(IDownloadTarget)
 983     def __init__(self):
 984         self._data = []
 985     def open(self, size):
 986         pass
 987     def write(self, data):
 988         self._data.append(data)
 989     def close(self):
 990         self.data = "".join(self._data)
 991         del self._data
 992     def fail(self, why):
 993         del self._data
 994     def register_canceller(self, cb):
 995         pass # we won't use it
 996     def finish(self):
 997         return self.data
 998
 999 class FileHandle:
1000     """Use me to download data to a pre-defined filehandle-like object. I
1001     will use the target's write() method. I will *not* close the filehandle:
1002     I leave that up to the originator of the filehandle. The download process
1003     will return the filehandle when it completes.
1004     """
1005     implements(IDownloadTarget)
1006     def __init__(self, filehandle):
1007         self._filehandle = filehandle
1008     def open(self, size):
1009         pass
1010     def write(self, data):
1011         self._filehandle.write(data)
1012     def close(self):
1013         # the originator of the filehandle reserves the right to close it
1014         pass
1015     def fail(self, why):
1016         pass
1017     def register_canceller(self, cb):
1018         pass
1019     def finish(self):
1020         return self._filehandle
1021
1022 class Downloader(service.MultiService):
1023     """I am a service that allows file downloading.
1024     """
1025     implements(IDownloader)
1026     name = "downloader"
1027     MAX_DOWNLOAD_STATUSES = 10
1028
1029     def __init__(self):
1030         service.MultiService.__init__(self)
1031         self._all_downloads = weakref.WeakKeyDictionary()
1032         self._recent_download_status = []
1033
1034     def download(self, u, t):
1035         assert self.parent
1036         assert self.running
1037         u = IFileURI(u)
1038         t = IDownloadTarget(t)
1039         assert t.write
1040         assert t.close
1041         if isinstance(u, uri.LiteralFileURI):
1042             dl = LiteralDownloader(self.parent, u, t)
1043         elif isinstance(u, uri.CHKFileURI):
1044             dl = FileDownloader(self.parent, u, t)
1045         else:
1046             raise RuntimeError("I don't know how to download a %s" % u)
1047         self._all_downloads[dl] = None
1048         self._recent_download_status.append(dl.get_download_status())
1049         while len(self._recent_download_status) > self.MAX_DOWNLOAD_STATUSES:
1050             self._recent_download_status.pop(0)
1051         d = dl.start()
1052         return d
1053
1054     # utility functions
1055     def download_to_data(self, uri):
1056         return self.download(uri, Data())
1057     def download_to_filename(self, uri, filename):
1058         return self.download(uri, FileName(filename))
1059     def download_to_filehandle(self, uri, filehandle):
1060         return self.download(uri, FileHandle(filehandle))
1061
1062
1063     def list_all_downloads(self):
1064         return self._all_downloads.keys()
1065     def list_active_downloads(self):
1066         return [d.get_download_status() for d in self._all_downloads.keys()
1067                 if d.get_download_status().get_active()]
1068     def list_recent_downloads(self):
1069         return self._recent_download_status