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