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, unload_delay):
  22         self.cache_dir = cache_dir
  23         self.unload_delay = unload_delay
  24         self.apt_caches = {}
  25
  26     def extractPath(self, url):
  27         parsed = urlparse(url)
  28         host, port = splitHostPort(parsed[0], parsed[1])
  29         site = host + ":" + str(port)
  30         path = parsed[2]
  31
  32         i = max(path.rfind('/dists/'), path.rfind('/pool/'))
  33         if i >= 0:
  34             baseDir = path[:i]
  35             path = path[i:]
  36         else:
  37             # Uh oh, this is not good
  38             log.msg("Couldn't find a good base directory for path: %s" % (site + path))
  39             baseDir = ''
  40             if site in self.apt_caches:
  41                 longest_match = 0
  42                 for base in self.apt_caches[site]:
  43                     base_match = ''
  44                     for dirs in path.split('/'):
  45                         if base.startswith(base_match + '/' + dirs):
  46                             base_match += '/' + dirs
  47                         else:
  48                             break
  49                     if len(base_match) > longest_match:
  50                         longest_match = len(base_match)
  51                         baseDir = base_match
  52             log.msg("Settled on baseDir: %s" % baseDir)
  53
  54         return site, baseDir, path
  55
  56     def init(self, site, baseDir):
  57         if site not in self.apt_caches:
  58             self.apt_caches[site] = {}
  59
  60         if baseDir not in self.apt_caches[site]:
  61             site_cache = self.cache_dir.child(aptpkg_dir).child('mirrors').child(site + baseDir.replace('/', '_'))
  62             site_cache.makedirs
  63             self.apt_caches[site][baseDir] = AptPackages(site_cache, self.unload_delay)
  64
  65     def updatedFile(self, url, file_path):
  66         site, baseDir, path = self.extractPath(url)
  67         self.init(site, baseDir)
  68         self.apt_caches[site][baseDir].file_updated(path, file_path)
  69
  70     def findHash(self, url):
  71         site, baseDir, path = self.extractPath(url)
  72         if site in self.apt_caches and baseDir in self.apt_caches[site]:
  73             return self.apt_caches[site][baseDir].findHash(path)
  74         d = defer.Deferred()
  75         d.errback(MirrorError("Site Not Found"))
  76         return d
  77
  78 class TestMirrorManager(unittest.TestCase):
  79     """Unit tests for the mirror manager."""
  80
  81     timeout = 20
  82     pending_calls = []
  83     client = None
  84
  85     def setUp(self):
  86         self.client = MirrorManager(FilePath('/tmp/.apt-dht'), 300)
  87
  88     def test_extractPath(self):
  89         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org/debian/dists/unstable/Release')
  90         self.failUnless(site == "ftp.us.debian.org:80", "no match: %s" % site)
  91         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
  92         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
  93
  94         site, baseDir, path = self.client.extractPath('http://ftp.us.debian.org:16999/debian/pool/d/dpkg/dpkg_1.2.1-1.tar.gz')
  95         self.failUnless(site == "ftp.us.debian.org:16999", "no match: %s" % site)
  96         self.failUnless(baseDir == "/debian", "no match: %s" % baseDir)
  97         self.failUnless(path == "/pool/d/dpkg/dpkg_1.2.1-1.tar.gz", "no match: %s" % path)
  98
  99         site, baseDir, path = self.client.extractPath('http://debian.camrdale.org/dists/unstable/Release')
 100         self.failUnless(site == "debian.camrdale.org:80", "no match: %s" % site)
 101         self.failUnless(baseDir == "", "no match: %s" % baseDir)
 102         self.failUnless(path == "/dists/unstable/Release", "no match: %s" % path)
 103
 104     def verifyHash(self, found_hash, path, true_hash):
 105         self.failUnless(found_hash.hexexpected() == true_hash,
 106                     "%s hashes don't match: %s != %s" % (path, found_hash.hexexpected(), true_hash))
 107
 108     def test_findHash(self):
 109         self.packagesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Packages$" | tail -n 1').read().rstrip('\n')
 110         self.sourcesFile = os.popen('ls -Sr /var/lib/apt/lists/ | grep -E "_main_.*Sources$" | tail -n 1').read().rstrip('\n')
 111         for f in os.walk('/var/lib/apt/lists').next()[2]:
 112             if f[-7:] == "Release" and self.packagesFile.startswith(f[:-7]):
 113                 self.releaseFile = f
 114                 break
 115
 116         self.client.updatedFile('http://' + self.releaseFile.replace('_','/'),
 117                                 FilePath('/var/lib/apt/lists/' + self.releaseFile))
 118         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 119                                 self.packagesFile[self.packagesFile.find('_dists_')+1:].replace('_','/'),
 120                                 FilePath('/var/lib/apt/lists/' + self.packagesFile))
 121         self.client.updatedFile('http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') +
 122                                 self.sourcesFile[self.sourcesFile.find('_dists_')+1:].replace('_','/'),
 123                                 FilePath('/var/lib/apt/lists/' + self.sourcesFile))
 124
 125         lastDefer = defer.Deferred()
 126
 127         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 128                             '/var/lib/apt/lists/' + self.releaseFile +
 129                             ' | grep -E " main/binary-i386/Packages.bz2$"'
 130                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 131         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/binary-i386/Packages.bz2'
 132
 133         d = self.client.findHash(idx_path)
 134         d.addCallback(self.verifyHash, idx_path, idx_hash)
 135
 136         pkg_hash = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 137                             '/var/lib/apt/lists/' + self.packagesFile +
 138                             ' | grep -E "^SHA1:" | head -n 1' +
 139                             ' | cut -d\  -f 2').read().rstrip('\n')
 140         pkg_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + \
 141                    os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 142                             '/var/lib/apt/lists/' + self.packagesFile +
 143                             ' | grep -E "^Filename:" | head -n 1' +
 144                             ' | cut -d\  -f 2').read().rstrip('\n')
 145
 146         d = self.client.findHash(pkg_path)
 147         d.addCallback(self.verifyHash, pkg_path, pkg_hash)
 148
 149         src_dir = os.popen('grep -A 30 -E "^Package: dpkg$" ' +
 150                             '/var/lib/apt/lists/' + self.sourcesFile +
 151                             ' | grep -E "^Directory:" | head -n 1' +
 152                             ' | cut -d\  -f 2').read().rstrip('\n')
 153         src_hashes = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 154                             '/var/lib/apt/lists/' + self.sourcesFile +
 155                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 156                             ' | cut -d\  -f 2').read().split('\n')[:-1]
 157         src_paths = os.popen('grep -A 20 -E "^Package: dpkg$" ' +
 158                             '/var/lib/apt/lists/' + self.sourcesFile +
 159                             ' | grep -A 4 -E "^Files:" | grep -E "^ " ' +
 160                             ' | cut -d\  -f 4').read().split('\n')[:-1]
 161
 162         for i in range(len(src_hashes)):
 163             src_path = 'http://' + self.releaseFile[:self.releaseFile.find('_dists_')+1].replace('_','/') + src_dir + '/' + src_paths[i]
 164             d = self.client.findHash(src_path)
 165             d.addCallback(self.verifyHash, src_path, src_hashes[i])
 166
 167         idx_hash = os.popen('grep -A 3000 -E "^SHA1:" ' +
 168                             '/var/lib/apt/lists/' + self.releaseFile +
 169                             ' | grep -E " main/source/Sources.bz2$"'
 170                             ' | head -n 1 | cut -d\  -f 2').read().rstrip('\n')
 171         idx_path = 'http://' + self.releaseFile.replace('_','/')[:-7] + 'main/source/Sources.bz2'
 172
 173         d = self.client.findHash(idx_path)
 174         d.addCallback(self.verifyHash, idx_path, idx_hash)
 175
 176         d.addBoth(lastDefer.callback)
 177         return lastDefer
 178
 179     def tearDown(self):
 180         for p in self.pending_calls:
 181             if p.active():
 182                 p.cancel()
 183         self.client = None
 184