apt_dht/MirrorManager.py

   1
   2 from urlparse import urlparse
   3 import os
   4
   5 from twisted.python import log, filepath
   6 from twisted.internet import defer
   7 from twisted.trial import unittest
   8 from twisted.web2 import stream
   9 from twisted.web2.http import splitHostPort
  10
  11 from AptPackages import AptPackages
  12
  13 aptpkg_dir='.apt-dht'
  14
  15 class MirrorError(Exception):
  16     """Exception raised when there's a problem with the mirror."""
  17
  18 class ProxyFileStream(stream.SimpleStream):
  19     """Saves a stream to a file while providing a new stream."""
  20
  21     def __init__(self, stream, outFile):
  22         """Initializes the proxy.
  23
  24         @type stream: C{twisted.web2.stream.IByteStream}
  25         @param stream: the input stream to read from
  26         @type outFile: C{twisted.python.filepath.FilePath}
  27         @param outFile: the file to write to
  28         """
  29         self.stream = stream
  30         self.outFile = outFile.open('w')
  31         self.length = self.stream.length
  32         self.start = 0
  33         self.doneDefer = defer.Deferred()
  34
  35     def _done(self):
  36         """Close the output file."""
  37         if not self.outFile.closed:
  38             self.outFile.close()
  39             self.doneDefer.callback(1)
  40
  41     def read(self):
  42         """Read some data from the stream."""
  43         if self.outFile.closed:
  44             return None
  45
  46         data = self.stream.read()
  47         if isinstance(data, defer.Deferred):
  48             data.addCallbacks(self._write, self._done)
  49             return data
  50
  51         self._write(data)
  52         return data
  53
  54     def _write(self, data):
  55         """Write the stream data to the file and return it for others to use."""
  56         if data is None:
  57             self._done()
  58             return data
  59
  60         self.outFile.write(data)
  61         return data
  62
  63     def close(self):
  64         """Clean everything up and return None to future reads."""
  65         self.length = 0
  66         self._done()
  67         self.stream.close()
  68
  69 class MirrorManager:
  70     """Manages all requests for mirror objects."""
  71
  72     def __init__(self, cache_dir):
  73         self.cache_dir = cache_dir
  74         self.cache = filepath.FilePath(self.cache_dir)
  75         self.apt_caches = {}
  76
  77     def extractPath(self, url):
  78         parsed = urlparse(url)
  79         host, port = splitHostPort(parsed[0], parsed[1])
  80         site = host + ":" + str(port)
  81         path = parsed[2]
  82
  83         i = max(path.rfind('/dists/'), path.rfind('/pool/'))
  84         if i >= 0:
  85             baseDir = path[:i]
  86             path = path[i:]
  87         else:
  88             # Uh oh, this is not good
  89             log.msg("Couldn't find a good base directory for path: %s" % (site + path))
  90             baseDir = ''
  91             if site in self.apt_caches:
  92                 longest_match = 0
  93                 for base in self.apt_caches[site]:
  94                     base_match = ''
  95                     for dirs in path.split('/'):
  96                         if base.startswith(base_match + '/' + dirs):
  97                             base_match += '/' + dirs
  98                         else:
  99                             break
 100                     if len(base_match) > longest_match:
 101                         longest_match = len(base_match)
 102                         baseDir = base_match
 103             log.msg("Settled on baseDir: %s" % baseDir)
 104
 105         return site, baseDir, path
 106
 107     def init(self, site, baseDir):
 108         if site not in self.apt_caches:
 109             self.apt_caches[site] = {}
 110
 111         if baseDir not in self.apt_caches[site]:
 112             site_cache = os.path.join(self.cache_dir, aptpkg_dir, 'mirrors', site + baseDir.replace('/', '_'))
 113             self.apt_caches[site][baseDir] = AptPackages(site_cache)
 114
 115     def updatedFile(self, url, file_path):
 116         site, baseDir, path = self.extractPath(url)
 117         self.init(site, baseDir)
 118         self.apt_caches[site][baseDir].file_updated(path, file_path)
 119
 120     def findHash(self, url):
 121         site, baseDir, path = self.extractPath(url)
 122         if site in self.apt_caches and baseDir in self.apt_caches[site]:
 123             return self.apt_caches[site][baseDir].findHash(path)
 124         d = defer.Deferred()
 125         d.errback(MirrorError("Site Not Found"))
 126         return d
 127
 128     def save_file(self, response, hash, size, url):
 129         """Save a downloaded file to the cache and stream it."""
 130         log.msg('Returning file: %s' % url)
 131
 132         parsed = urlparse(url)
 133         destFile = self.cache.preauthChild(parsed[1] + parsed[2])
 134         log.msg('Saving returned %r byte file to cache: %s' % (response.stream.length, destFile.path))
 135
 136         if destFile.exists():
 137             log.msg('File already exists, removing: %s' % destFile.path)
 138             destFile.remove()
 139         else:
 140             destFile.parent().makedirs()
 141
 142         orig_stream = response.stream
 143         response.stream = ProxyFileStream(orig_stream, destFile)
 144         response.stream.doneDefer.addCallback(self.save_complete, url, destFile,
 145                                               response.headers.getHeader('Last-Modified'))
 146         response.stream.doneDefer.addErrback(self.save_error, url)
 147         return response
 148
 149     def save_complete(self, result, url, destFile, modtime = None):
 150         """Update the modification time and AptPackages."""
 151         if modtime:
 152             os.utime(destFile.path, (modtime, modtime))
 153
 154         site, baseDir, path = self.extractPath(url)
 155         self.init(site, baseDir)
 156         self.apt_caches[site][baseDir].file_updated(path, destFile.path)
 157
 158     def save_error(self, failure, url):
 159         """An error has occurred in downloadign or saving the file."""
 160         log.msg('Error occurred downloading %s' % url)
 161         log.err(failure)
 162         return failure
 163
 164 class TestMirrorManager(unittest.TestCase):
 165     """Unit tests for the mirror manager."""
 166
 167     timeout = 20
 168     pending_calls = []
 169     client = None
 170
 171     def setUp(self):
 172         self.client = MirrorManager('/tmp')
 173
 174     def test_extractPath(self):
 175         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org/debian/dists/unstable/Release')
 176         self.failUnless(site == "ftp.us.debian.org:80", "no match: %s" % site)
 177         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
 178         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
 179
 180         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org:16999/debian/pool/d/dpkg/dpkg_1.2.1-1.tar.gz')
 181         self.failUnless(site == "ftp.us.debian.org:16999", "no match: %s" % site)
 182         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
 183         self.failUnless(path == "/pool/d/dpkg/dpkg_1.2.1-1.tar.gz", "no match: %s" % path)
 184
 185         site, baseDir, path = self.client.extractPath('http://debian.camrdale.org/dists/unstable/Release')
 186         self.failUnless(site == "debian.camrdale.org:80", "no match: %s" % site)
 187         self.failUnless(baseDir == "", "no match: %s" % baseDir)
 188         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
 189
 190     def verifyHash(self, found_hash, path, true_hash):
 191         self.failUnless(found_hash[0] == true_hash,
 192                     "%s hashes don't match: %s != %s" % (path, found_hash[0], true_hash))
 193
 194     def test_findHash(self):
 195         self.packagesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Packages$" | tail -n 1').read().rstrip('\n')
 196         self.sourcesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Sources$" | tail -n 1').read().rstrip('\n')
 197         for f in os.walk('/var/lib/apt/lists').next()[2]:
 198             if f[-7:] == "Release" and self.packagesFile.startswith(f[:-7]):
 199                 self.releaseFile = f
 200                 break
 201
 202         self.client.updatedFile('http://' + self.releaseFile.replace('_','/'),
 203                                 '/var/lib/apt/lists/' + self.releaseFile)
 204         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 205                                 self.packagesFile[self.packagesFile.find('_dists_')+1:].replace('_','/'),
 206                                 '/var/lib/apt/lists/' + self.packagesFile)
 207         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 208                                 self.sourcesFile[self.sourcesFile.find('_dists_')+1:].replace('_','/'),
 209                                 '/var/lib/apt/lists/' + self.sourcesFile)
 210
 211         lastDefer = defer.Deferred()
 212
 213         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 214                             '/var/lib/apt/lists/' + self.releaseFile +
 215                             ' | grep -E " main/binary-i386/Packages.bz2$"'
 216                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 217         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/binary-i386/Packages.bz2'
 218
 219         d = self.client.findHash(idx_path)
 220         d.addCallback(self.verifyHash, idx_path, idx_hash)
 221
 222         pkg_hash = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 223                             '/var/lib/apt/lists/' + self.packagesFile +
 224                             ' | grep -E "^SHA1:" | head -n 1' +
 225                             ' | cut -d\  -f 2').read().rstrip('\n')
 226         pkg_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + \
 227                    os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 228                             '/var/lib/apt/lists/' + self.packagesFile +
 229                             ' | grep -E "^Filename:" | head -n 1' +
 230                             ' | cut -d\  -f 2').read().rstrip('\n')
 231
 232         d = self.client.findHash(pkg_path)
 233         d.addCallback(self.verifyHash, pkg_path, pkg_hash)
 234
 235         src_dir = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 236                             '/var/lib/apt/lists/' + self.sourcesFile +
 237                             ' | grep -E "^Directory:" | head -n 1' +
 238                             ' | cut -d\  -f 2').read().rstrip('\n')
 239         src_hashes = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 240                             '/var/lib/apt/lists/' + self.sourcesFile +
 241                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 242                             ' | cut -d\  -f 2').read().split('\n')[:-1]
 243         src_paths = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 244                             '/var/lib/apt/lists/' + self.sourcesFile +
 245                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 246                             ' | cut -d\  -f 4').read().split('\n')[:-1]
 247
 248         for i in range(len(src_hashes)):
 249             src_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + src_dir + '/' + src_paths[i]
 250             d = self.client.findHash(src_path)
 251             d.addCallback(self.verifyHash, src_path, src_hashes[i])
 252
 253         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 254                             '/var/lib/apt/lists/' + self.releaseFile +
 255                             ' | grep -E " main/source/Sources.bz2$"'
 256                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 257         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/source/Sources.bz2'
 258
 259         d = self.client.findHash(idx_path)
 260         d.addCallback(self.verifyHash, idx_path, idx_hash)
 261
 262         d.addBoth(lastDefer.callback)
 263         return lastDefer
 264
 265     def tearDown(self):
 266         for p in self.pending_calls:
 267             if p.active():
 268                 p.cancel()
 269         self.client = None
 270