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