apt_dht/MirrorManager.py

   1
   2 from bz2 import BZ2Decompressor
   3 from zlib import decompressobj, MAX_WBITS
   4 from gzip import FCOMMENT, FEXTRA, FHCRC, FNAME, FTEXT
   5 from urlparse import urlparse
   6 import os
   7
   8 from twisted.python import log, filepath
   9 from twisted.internet import defer
  10 from twisted.trial import unittest
  11 from twisted.web2 import stream
  12 from twisted.web2.http import splitHostPort
  13
  14 from AptPackages import AptPackages
  15
  16 aptpkg_dir='.apt-dht'
  17
  18 DECOMPRESS_EXTS = ['.gz', '.bz2']
  19 DECOMPRESS_FILES = ['release', 'sources', 'packages']
  20
  21 class MirrorError(Exception):
  22     """Exception raised when there's a problem with the mirror."""
  23
  24 class ProxyFileStream(stream.SimpleStream):
  25     """Saves a stream to a file while providing a new stream."""
  26
  27     def __init__(self, stream, outFile, hash, decompress = None, decFile = None):
  28         """Initializes the proxy.
  29
  30         @type stream: C{twisted.web2.stream.IByteStream}
  31         @param stream: the input stream to read from
  32         @type outFile: C{twisted.python.filepath.FilePath}
  33         @param outFile: the file to write to
  34         @type hash: L{Hash.HashObject}
  35         @param hash: the hash object to use for the file
  36         @type decompress: C{string}
  37         @param decompress: also decompress the file as this type
  38             (currently only '.gz' and '.bz2' are supported)
  39         @type decFile: C{twisted.python.filepath.FilePath}
  40         @param decFile: the file to write the decompressed data to
  41         """
  42         self.stream = stream
  43         self.outFile = outFile.open('w')
  44         self.hash = hash
  45         self.hash.new()
  46         self.gzfile = None
  47         self.bz2file = None
  48         if decompress == ".gz":
  49             self.gzheader = True
  50             self.gzfile = decFile.open('w')
  51             self.gzdec = decompressobj(-MAX_WBITS)
  52         elif decompress == ".bz2":
  53             self.bz2file = decFile.open('w')
  54             self.bz2dec = BZ2Decompressor()
  55         self.length = self.stream.length
  56         self.start = 0
  57         self.doneDefer = defer.Deferred()
  58
  59     def _done(self):
  60         """Close the output file."""
  61         if not self.outFile.closed:
  62             self.outFile.close()
  63             self.hash.digest()
  64             if self.gzfile:
  65                 data_dec = self.gzdec.flush()
  66                 self.gzfile.write(data_dec)
  67                 self.gzfile.close()
  68                 self.gzfile = None
  69             if self.bz2file:
  70                 self.bz2file.close()
  71                 self.bz2file = None
  72
  73             self.doneDefer.callback(self.hash)
  74
  75     def read(self):
  76         """Read some data from the stream."""
  77         if self.outFile.closed:
  78             return None
  79
  80         data = self.stream.read()
  81         if isinstance(data, defer.Deferred):
  82             data.addCallbacks(self._write, self._done)
  83             return data
  84
  85         self._write(data)
  86         return data
  87
  88     def _write(self, data):
  89         """Write the stream data to the file and return it for others to use."""
  90         if data is None:
  91             self._done()
  92             return data
  93
  94         self.outFile.write(data)
  95         self.hash.update(data)
  96         if self.gzfile:
  97             if self.gzheader:
  98                 self.gzheader = False
  99                 new_data = self._remove_gzip_header(data)
 100                 dec_data = self.gzdec.decompress(new_data)
 101             else:
 102                 dec_data = self.gzdec.decompress(data)
 103             self.gzfile.write(dec_data)
 104         if self.bz2file:
 105             dec_data = self.bz2dec.decompress(data)
 106             self.bz2file.write(dec_data)
 107         return data
 108
 109     def _remove_gzip_header(self, data):
 110         if data[:2] != '\037\213':
 111             raise IOError, 'Not a gzipped file'
 112         if ord(data[2]) != 8:
 113             raise IOError, 'Unknown compression method'
 114         flag = ord(data[3])
 115         # modtime = self.fileobj.read(4)
 116         # extraflag = self.fileobj.read(1)
 117         # os = self.fileobj.read(1)
 118
 119         skip = 10
 120         if flag & FEXTRA:
 121             # Read & discard the extra field, if present
 122             xlen = ord(data[10])
 123             xlen = xlen + 256*ord(data[11])
 124             skip = skip + 2 + xlen
 125         if flag & FNAME:
 126             # Read and discard a null-terminated string containing the filename
 127             while True:
 128                 if not data[skip] or data[skip] == '\000':
 129                     break
 130                 skip += 1
 131             skip += 1
 132         if flag & FCOMMENT:
 133             # Read and discard a null-terminated string containing a comment
 134             while True:
 135                 if not data[skip] or data[skip] == '\000':
 136                     break
 137                 skip += 1
 138             skip += 1
 139         if flag & FHCRC:
 140             skip += 2     # Read & discard the 16-bit header CRC
 141         return data[skip:]
 142
 143     def close(self):
 144         """Clean everything up and return None to future reads."""
 145         self.length = 0
 146         self._done()
 147         self.stream.close()
 148
 149 class MirrorManager:
 150     """Manages all requests for mirror objects."""
 151
 152     def __init__(self, cache_dir, manager = None):
 153         self.manager = manager
 154         self.cache_dir = cache_dir
 155         self.cache = filepath.FilePath(self.cache_dir)
 156         self.apt_caches = {}
 157
 158     def extractPath(self, url):
 159         parsed = urlparse(url)
 160         host, port = splitHostPort(parsed[0], parsed[1])
 161         site = host + ":" + str(port)
 162         path = parsed[2]
 163
 164         i = max(path.rfind('/dists/'), path.rfind('/pool/'))
 165         if i >= 0:
 166             baseDir = path[:i]
 167             path = path[i:]
 168         else:
 169             # Uh oh, this is not good
 170             log.msg("Couldn't find a good base directory for path: %s" % (site + path))
 171             baseDir = ''
 172             if site in self.apt_caches:
 173                 longest_match = 0
 174                 for base in self.apt_caches[site]:
 175                     base_match = ''
 176                     for dirs in path.split('/'):
 177                         if base.startswith(base_match + '/' + dirs):
 178                             base_match += '/' + dirs
 179                         else:
 180                             break
 181                     if len(base_match) > longest_match:
 182                         longest_match = len(base_match)
 183                         baseDir = base_match
 184             log.msg("Settled on baseDir: %s" % baseDir)
 185
 186         return site, baseDir, path
 187
 188     def init(self, site, baseDir):
 189         if site not in self.apt_caches:
 190             self.apt_caches[site] = {}
 191
 192         if baseDir not in self.apt_caches[site]:
 193             site_cache = os.path.join(self.cache_dir, aptpkg_dir, 'mirrors', site + baseDir.replace('/', '_'))
 194             self.apt_caches[site][baseDir] = AptPackages(site_cache)
 195
 196     def updatedFile(self, url, file_path):
 197         site, baseDir, path = self.extractPath(url)
 198         self.init(site, baseDir)
 199         self.apt_caches[site][baseDir].file_updated(path, file_path)
 200
 201     def findHash(self, url):
 202         site, baseDir, path = self.extractPath(url)
 203         if site in self.apt_caches and baseDir in self.apt_caches[site]:
 204             return self.apt_caches[site][baseDir].findHash(path)
 205         d = defer.Deferred()
 206         d.errback(MirrorError("Site Not Found"))
 207         return d
 208
 209     def save_file(self, response, hash, url):
 210         """Save a downloaded file to the cache and stream it."""
 211         if response.code != 200:
 212             log.msg('File was not found (%r): %s' % (response, url))
 213             return response
 214
 215         log.msg('Returning file: %s' % url)
 216
 217         parsed = urlparse(url)
 218         destFile = self.cache.preauthChild(parsed[1] + parsed[2])
 219         log.msg('Saving returned %r byte file to cache: %s' % (response.stream.length, destFile.path))
 220
 221         if destFile.exists():
 222             log.msg('File already exists, removing: %s' % destFile.path)
 223             destFile.remove()
 224         else:
 225             destFile.parent().makedirs()
 226
 227         root, ext = os.path.splitext(destFile.basename())
 228         if root.lower() in DECOMPRESS_FILES and ext.lower() in DECOMPRESS_EXTS:
 229             ext = ext.lower()
 230             decFile = destFile.sibling(root)
 231             log.msg('Decompressing to: %s' % decFile.path)
 232             if decFile.exists():
 233                 log.msg('File already exists, removing: %s' % decFile.path)
 234                 decFile.remove()
 235         else:
 236             ext = None
 237             decFile = None
 238
 239         orig_stream = response.stream
 240         response.stream = ProxyFileStream(orig_stream, destFile, hash, ext, decFile)
 241         response.stream.doneDefer.addCallback(self.save_complete, url, destFile,
 242                                               response.headers.getHeader('Last-Modified'),
 243                                               ext, decFile)
 244         response.stream.doneDefer.addErrback(self.save_error, url)
 245         return response
 246
 247     def save_complete(self, hash, url, destFile, modtime = None, ext = None, decFile = None):
 248         """Update the modification time and AptPackages."""
 249         if modtime:
 250             os.utime(destFile.path, (modtime, modtime))
 251             if ext:
 252                 os.utime(decFile.path, (modtime, modtime))
 253
 254         result = hash.verify()
 255         if result or result is None:
 256             if result:
 257                 log.msg('Hashes match: %s' % url)
 258             else:
 259                 log.msg('Hashed file to %s: %s' % (hash.hexdigest(), url))
 260
 261             self.updatedFile(url, destFile.path)
 262             if ext:
 263                 self.updatedFile(url[:-len(ext)], decFile.path)
 264
 265             if self.manager:
 266                 self.manager.cached_file(hash, url, destFile.path)
 267         else:
 268             log.msg("Hashes don't match %s != %s: %s" % (hash.hexexpected(), hash.hexdigest(), url))
 269
 270     def save_error(self, failure, url):
 271         """An error has occurred in downloadign or saving the file."""
 272         log.msg('Error occurred downloading %s' % url)
 273         log.err(failure)
 274         return failure
 275
 276 class TestMirrorManager(unittest.TestCase):
 277     """Unit tests for the mirror manager."""
 278
 279     timeout = 20
 280     pending_calls = []
 281     client = None
 282
 283     def setUp(self):
 284         self.client = MirrorManager('/tmp')
 285
 286     def test_extractPath(self):
 287         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org/debian/dists/unstable/Release')
 288         self.failUnless(site == "ftp.us.debian.org:80", "no match: %s" % site)
 289         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
 290         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
 291
 292         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org:16999/debian/pool/d/dpkg/dpkg_1.2.1-1.tar.gz')
 293         self.failUnless(site == "ftp.us.debian.org:16999", "no match: %s" % site)
 294         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
 295         self.failUnless(path == "/pool/d/dpkg/dpkg_1.2.1-1.tar.gz", "no match: %s" % path)
 296
 297         site, baseDir, path = self.client.extractPath('http://debian.camrdale.org/dists/unstable/Release')
 298         self.failUnless(site == "debian.camrdale.org:80", "no match: %s" % site)
 299         self.failUnless(baseDir == "", "no match: %s" % baseDir)
 300         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
 301
 302     def verifyHash(self, found_hash, path, true_hash):
 303         self.failUnless(found_hash.hexexpected() == true_hash,
 304                     "%s hashes don't match: %s != %s" % (path, found_hash.hexexpected(), true_hash))
 305
 306     def test_findHash(self):
 307         self.packagesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Packages$" | tail -n 1').read().rstrip('\n')
 308         self.sourcesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Sources$" | tail -n 1').read().rstrip('\n')
 309         for f in os.walk('/var/lib/apt/lists').next()[2]:
 310             if f[-7:] == "Release" and self.packagesFile.startswith(f[:-7]):
 311                 self.releaseFile = f
 312                 break
 313
 314         self.client.updatedFile('http://' + self.releaseFile.replace('_','/'),
 315                                 '/var/lib/apt/lists/' + self.releaseFile)
 316         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 317                                 self.packagesFile[self.packagesFile.find('_dists_')+1:].replace('_','/'),
 318                                 '/var/lib/apt/lists/' + self.packagesFile)
 319         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 320                                 self.sourcesFile[self.sourcesFile.find('_dists_')+1:].replace('_','/'),
 321                                 '/var/lib/apt/lists/' + self.sourcesFile)
 322
 323         lastDefer = defer.Deferred()
 324
 325         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 326                             '/var/lib/apt/lists/' + self.releaseFile +
 327                             ' | grep -E " main/binary-i386/Packages.bz2$"'
 328                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 329         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/binary-i386/Packages.bz2'
 330
 331         d = self.client.findHash(idx_path)
 332         d.addCallback(self.verifyHash, idx_path, idx_hash)
 333
 334         pkg_hash = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 335                             '/var/lib/apt/lists/' + self.packagesFile +
 336                             ' | grep -E "^SHA1:" | head -n 1' +
 337                             ' | cut -d\  -f 2').read().rstrip('\n')
 338         pkg_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + \
 339                    os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 340                             '/var/lib/apt/lists/' + self.packagesFile +
 341                             ' | grep -E "^Filename:" | head -n 1' +
 342                             ' | cut -d\  -f 2').read().rstrip('\n')
 343
 344         d = self.client.findHash(pkg_path)
 345         d.addCallback(self.verifyHash, pkg_path, pkg_hash)
 346
 347         src_dir = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 348                             '/var/lib/apt/lists/' + self.sourcesFile +
 349                             ' | grep -E "^Directory:" | head -n 1' +
 350                             ' | cut -d\  -f 2').read().rstrip('\n')
 351         src_hashes = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 352                             '/var/lib/apt/lists/' + self.sourcesFile +
 353                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 354                             ' | cut -d\  -f 2').read().split('\n')[:-1]
 355         src_paths = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 356                             '/var/lib/apt/lists/' + self.sourcesFile +
 357                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 358                             ' | cut -d\  -f 4').read().split('\n')[:-1]
 359
 360         for i in range(len(src_hashes)):
 361             src_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + src_dir + '/' + src_paths[i]
 362             d = self.client.findHash(src_path)
 363             d.addCallback(self.verifyHash, src_path, src_hashes[i])
 364
 365         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 366                             '/var/lib/apt/lists/' + self.releaseFile +
 367                             ' | grep -E " main/source/Sources.bz2$"'
 368                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 369         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/source/Sources.bz2'
 370
 371         d = self.client.findHash(idx_path)
 372         d.addCallback(self.verifyHash, idx_path, idx_hash)
 373
 374         d.addBoth(lastDefer.callback)
 375         return lastDefer
 376
 377     def tearDown(self):
 378         for p in self.pending_calls:
 379             if p.active():
 380                 p.cancel()
 381         self.client = None
 382