apt_dht/MirrorManager.py

   1
   2 from urlparse import urlparse
   3 import os
   4
   5 from twisted.python import log
   6 from twisted.python.filepath import FilePath
   7 from twisted.internet import defer
   8 from twisted.trial import unittest
   9 from twisted.web2.http import splitHostPort
  10
  11 from AptPackages import AptPackages
  12
  13 aptpkg_dir='apt-packages'
  14
  15 class MirrorError(Exception):
  16     """Exception raised when there's a problem with the mirror."""
  17
  18 class MirrorManager:
  19     """Manages all requests for mirror objects."""
  20
  21     def __init__(self, cache_dir):
  22         self.cache_dir = cache_dir
  23         self.apt_caches = {}
  24
  25     def extractPath(self, url):
  26         parsed = urlparse(url)
  27         host, port = splitHostPort(parsed[0], parsed[1])
  28         site = host + ":" + str(port)
  29         path = parsed[2]
  30
  31         i = max(path.rfind('/dists/'), path.rfind('/pool/'))
  32         if i >= 0:
  33             baseDir = path[:i]
  34             path = path[i:]
  35         else:
  36             # Uh oh, this is not good
  37             log.msg("Couldn't find a good base directory for path: %s" % (site + path))
  38             baseDir = ''
  39             if site in self.apt_caches:
  40                 longest_match = 0
  41                 for base in self.apt_caches[site]:
  42                     base_match = ''
  43                     for dirs in path.split('/'):
  44                         if base.startswith(base_match + '/' + dirs):
  45                             base_match += '/' + dirs
  46                         else:
  47                             break
  48                     if len(base_match) > longest_match:
  49                         longest_match = len(base_match)
  50                         baseDir = base_match
  51             log.msg("Settled on baseDir: %s" % baseDir)
  52
  53         return site, baseDir, path
  54
  55     def init(self, site, baseDir):
  56         if site not in self.apt_caches:
  57             self.apt_caches[site] = {}
  58
  59         if baseDir not in self.apt_caches[site]:
  60             site_cache = self.cache_dir.child(aptpkg_dir).child('mirrors').child(site + baseDir.replace('/', '_'))
  61             site_cache.makedirs
  62             self.apt_caches[site][baseDir] = AptPackages(site_cache)
  63
  64     def updatedFile(self, url, file_path):
  65         site, baseDir, path = self.extractPath(url)
  66         self.init(site, baseDir)
  67         self.apt_caches[site][baseDir].file_updated(path, file_path)
  68
  69     def findHash(self, url):
  70         site, baseDir, path = self.extractPath(url)
  71         if site in self.apt_caches and baseDir in self.apt_caches[site]:
  72             return self.apt_caches[site][baseDir].findHash(path)
  73         d = defer.Deferred()
  74         d.errback(MirrorError("Site Not Found"))
  75         return d
  76
  77 class TestMirrorManager(unittest.TestCase):
  78     """Unit tests for the mirror manager."""
  79
  80     timeout = 20
  81     pending_calls = []
  82     client = None
  83
  84     def setUp(self):
  85         self.client = MirrorManager(FilePath('/tmp/.apt-dht'))
  86
  87     def test_extractPath(self):
  88         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org/debian/dists/unstable/Release')
  89         self.failUnless(site == "ftp.us.debian.org:80", "no match: %s" % site)
  90         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
  91         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
  92
  93         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org:16999/debian/pool/d/dpkg/dpkg_1.2.1-1.tar.gz')
  94         self.failUnless(site == "ftp.us.debian.org:16999", "no match: %s" % site)
  95         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
  96         self.failUnless(path == "/pool/d/dpkg/dpkg_1.2.1-1.tar.gz", "no match: %s" % path)
  97
  98         site, baseDir, path = self.client.extractPath('http://debian.camrdale.org/dists/unstable/Release')
  99         self.failUnless(site == "debian.camrdale.org:80", "no match: %s" % site)
 100         self.failUnless(baseDir == "", "no match: %s" % baseDir)
 101         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
 102
 103     def verifyHash(self, found_hash, path, true_hash):
 104         self.failUnless(found_hash.hexexpected() == true_hash,
 105                     "%s hashes don't match: %s != %s" % (path, found_hash.hexexpected(), true_hash))
 106
 107     def test_findHash(self):
 108         self.packagesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Packages$" | tail -n 1').read().rstrip('\n')
 109         self.sourcesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Sources$" | tail -n 1').read().rstrip('\n')
 110         for f in os.walk('/var/lib/apt/lists').next()[2]:
 111             if f[-7:] == "Release" and self.packagesFile.startswith(f[:-7]):
 112                 self.releaseFile = f
 113                 break
 114
 115         self.client.updatedFile('http://' + self.releaseFile.replace('_','/'),
 116                                 FilePath('/var/lib/apt/lists/' + self.releaseFile))
 117         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 118                                 self.packagesFile[self.packagesFile.find('_dists_')+1:].replace('_','/'),
 119                                 FilePath('/var/lib/apt/lists/' + self.packagesFile))
 120         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 121                                 self.sourcesFile[self.sourcesFile.find('_dists_')+1:].replace('_','/'),
 122                                 FilePath('/var/lib/apt/lists/' + self.sourcesFile))
 123
 124         lastDefer = defer.Deferred()
 125
 126         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 127                             '/var/lib/apt/lists/' + self.releaseFile +
 128                             ' | grep -E " main/binary-i386/Packages.bz2$"'
 129                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 130         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/binary-i386/Packages.bz2'
 131
 132         d = self.client.findHash(idx_path)
 133         d.addCallback(self.verifyHash, idx_path, idx_hash)
 134
 135         pkg_hash = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 136                             '/var/lib/apt/lists/' + self.packagesFile +
 137                             ' | grep -E "^SHA1:" | head -n 1' +
 138                             ' | cut -d\  -f 2').read().rstrip('\n')
 139         pkg_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + \
 140                    os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 141                             '/var/lib/apt/lists/' + self.packagesFile +
 142                             ' | grep -E "^Filename:" | head -n 1' +
 143                             ' | cut -d\  -f 2').read().rstrip('\n')
 144
 145         d = self.client.findHash(pkg_path)
 146         d.addCallback(self.verifyHash, pkg_path, pkg_hash)
 147
 148         src_dir = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 149                             '/var/lib/apt/lists/' + self.sourcesFile +
 150                             ' | grep -E "^Directory:" | head -n 1' +
 151                             ' | cut -d\  -f 2').read().rstrip('\n')
 152         src_hashes = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 153                             '/var/lib/apt/lists/' + self.sourcesFile +
 154                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 155                             ' | cut -d\  -f 2').read().split('\n')[:-1]
 156         src_paths = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 157                             '/var/lib/apt/lists/' + self.sourcesFile +
 158                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 159                             ' | cut -d\  -f 4').read().split('\n')[:-1]
 160
 161         for i in range(len(src_hashes)):
 162             src_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + src_dir + '/' + src_paths[i]
 163             d = self.client.findHash(src_path)
 164             d.addCallback(self.verifyHash, src_path, src_hashes[i])
 165
 166         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 167                             '/var/lib/apt/lists/' + self.releaseFile +
 168                             ' | grep -E " main/source/Sources.bz2$"'
 169                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 170         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/source/Sources.bz2'
 171
 172         d = self.client.findHash(idx_path)
 173         d.addCallback(self.verifyHash, idx_path, idx_hash)
 174
 175         d.addBoth(lastDefer.callback)
 176         return lastDefer
 177
 178     def tearDown(self):
 179         for p in self.pending_calls:
 180             if p.active():
 181                 p.cancel()
 182         self.client = None
 183