apt_dht/CacheManager.py

   1
   2 from bz2 import BZ2Decompressor
   3 from zlib import decompressobj, MAX_WBITS
   4 from gzip import FCOMMENT, FEXTRA, FHCRC, FNAME, FTEXT
   5 from urlparse import urlparse
   6 import os
   7
   8 from twisted.python import log
   9 from twisted.python.filepath import FilePath
  10 from twisted.internet import defer, reactor
  11 from twisted.trial import unittest
  12 from twisted.web2 import stream
  13 from twisted.web2.http import splitHostPort
  14
  15 from Hash import HashObject
  16
  17 aptpkg_dir='apt-packages'
  18
  19 DECOMPRESS_EXTS = ['.gz', '.bz2']
  20 DECOMPRESS_FILES = ['release', 'sources', 'packages']
  21
  22 class ProxyFileStream(stream.SimpleStream):
  23     """Saves a stream to a file while providing a new stream."""
  24
  25     def __init__(self, stream, outFile, hash, decompress = None, decFile = None):
  26         """Initializes the proxy.
  27
  28         @type stream: C{twisted.web2.stream.IByteStream}
  29         @param stream: the input stream to read from
  30         @type outFile: C{twisted.python.FilePath}
  31         @param outFile: the file to write to
  32         @type hash: L{Hash.HashObject}
  33         @param hash: the hash object to use for the file
  34         @type decompress: C{string}
  35         @param decompress: also decompress the file as this type
  36             (currently only '.gz' and '.bz2' are supported)
  37         @type decFile: C{twisted.python.FilePath}
  38         @param decFile: the file to write the decompressed data to
  39         """
  40         self.stream = stream
  41         self.outFile = outFile.open('w')
  42         self.hash = hash
  43         self.hash.new()
  44         self.gzfile = None
  45         self.bz2file = None
  46         if decompress == ".gz":
  47             self.gzheader = True
  48             self.gzfile = decFile.open('w')
  49             self.gzdec = decompressobj(-MAX_WBITS)
  50         elif decompress == ".bz2":
  51             self.bz2file = decFile.open('w')
  52             self.bz2dec = BZ2Decompressor()
  53         self.length = self.stream.length
  54         self.start = 0
  55         self.doneDefer = defer.Deferred()
  56
  57     def _done(self):
  58         """Close the output file."""
  59         if not self.outFile.closed:
  60             self.outFile.close()
  61             self.hash.digest()
  62             if self.gzfile:
  63                 data_dec = self.gzdec.flush()
  64                 self.gzfile.write(data_dec)
  65                 self.gzfile.close()
  66                 self.gzfile = None
  67             if self.bz2file:
  68                 self.bz2file.close()
  69                 self.bz2file = None
  70
  71             self.doneDefer.callback(self.hash)
  72
  73     def read(self):
  74         """Read some data from the stream."""
  75         if self.outFile.closed:
  76             return None
  77
  78         data = self.stream.read()
  79         if isinstance(data, defer.Deferred):
  80             data.addCallbacks(self._write, self._done)
  81             return data
  82
  83         self._write(data)
  84         return data
  85
  86     def _write(self, data):
  87         """Write the stream data to the file and return it for others to use."""
  88         if data is None:
  89             self._done()
  90             return data
  91
  92         self.outFile.write(data)
  93         self.hash.update(data)
  94         if self.gzfile:
  95             if self.gzheader:
  96                 self.gzheader = False
  97                 new_data = self._remove_gzip_header(data)
  98                 dec_data = self.gzdec.decompress(new_data)
  99             else:
 100                 dec_data = self.gzdec.decompress(data)
 101             self.gzfile.write(dec_data)
 102         if self.bz2file:
 103             dec_data = self.bz2dec.decompress(data)
 104             self.bz2file.write(dec_data)
 105         return data
 106
 107     def _remove_gzip_header(self, data):
 108         if data[:2] != '\037\213':
 109             raise IOError, 'Not a gzipped file'
 110         if ord(data[2]) != 8:
 111             raise IOError, 'Unknown compression method'
 112         flag = ord(data[3])
 113         # modtime = self.fileobj.read(4)
 114         # extraflag = self.fileobj.read(1)
 115         # os = self.fileobj.read(1)
 116
 117         skip = 10
 118         if flag & FEXTRA:
 119             # Read & discard the extra field, if present
 120             xlen = ord(data[10])
 121             xlen = xlen + 256*ord(data[11])
 122             skip = skip + 2 + xlen
 123         if flag & FNAME:
 124             # Read and discard a null-terminated string containing the filename
 125             while True:
 126                 if not data[skip] or data[skip] == '\000':
 127                     break
 128                 skip += 1
 129             skip += 1
 130         if flag & FCOMMENT:
 131             # Read and discard a null-terminated string containing a comment
 132             while True:
 133                 if not data[skip] or data[skip] == '\000':
 134                     break
 135                 skip += 1
 136             skip += 1
 137         if flag & FHCRC:
 138             skip += 2     # Read & discard the 16-bit header CRC
 139         return data[skip:]
 140
 141     def close(self):
 142         """Clean everything up and return None to future reads."""
 143         self.length = 0
 144         self._done()
 145         self.stream.close()
 146
 147 class CacheManager:
 148     """Manages all requests for cached objects."""
 149
 150     def __init__(self, cache_dir, db, other_dirs = [], manager = None):
 151         self.cache_dir = cache_dir
 152         self.other_dirs = other_dirs
 153         self.all_dirs = self.other_dirs[:]
 154         self.all_dirs.insert(0, self.cache_dir)
 155         self.db = db
 156         self.manager = manager
 157         self.scanning = []
 158
 159         # Init the database, remove old files, init the HTTP dirs
 160         self.db.removeUntrackedFiles(self.all_dirs)
 161         self.db.reconcileDirectories()
 162         self.manager.setDirectories(self.db.getAllDirectories())
 163
 164
 165     def scanDirectories(self):
 166         """Scan the cache directories, hashing new and rehashing changed files."""
 167         assert not self.scanning, "a directory scan is already under way"
 168         self.scanning = self.all_dirs[:]
 169         self._scanDirectories()
 170
 171     def _scanDirectories(self, walker = None):
 172         # Need to start waling a new directory
 173         if walker is None:
 174             # If there are any left, get them
 175             if self.scanning:
 176                 log.msg('started scanning directory: %s' % self.scanning[0].path)
 177                 walker = self.scanning[0].walk()
 178             else:
 179                 # Done, just check if the HTTP directories need updating
 180                 log.msg('cache directory scan complete')
 181                 if self.db.reconcileDirectories():
 182                     self.manager.setDirectories(self.db.getAllDirectories())
 183                 return
 184
 185         try:
 186             # Get the next file in the directory
 187             file = walker.next()
 188         except StopIteration:
 189             # No files left, go to the next directory
 190             log.msg('done scanning directory: %s' % self.scanning[0].path)
 191             self.scanning.pop(0)
 192             reactor.callLater(0, self._scanDirectories)
 193             return
 194
 195         # If it's not a file, or it's already properly in the DB, ignore it
 196         if not file.isfile() or self.db.isUnchanged(file):
 197             if not file.isfile():
 198                 log.msg('entering directory: %s' % file.path)
 199             else:
 200                 log.msg('file is unchanged: %s' % file.path)
 201             reactor.callLater(0, self._scanDirectories, walker)
 202             return
 203
 204         # Otherwise hash it
 205         log.msg('start hash checking file: %s' % file.path)
 206         hash = HashObject()
 207         df = hash.hashInThread(file)
 208         df.addBoth(self._doneHashing, file, walker)
 209         df.addErrback(log.err)
 210
 211     def _doneHashing(self, result, file, walker):
 212         reactor.callLater(0, self._scanDirectories, walker)
 213
 214         if isinstance(result, HashObject):
 215             log.msg('hash check of %s completed with hash: %s' % (file.path, result.hexdigest()))
 216             if self.scanning[0] == self.cache_dir:
 217                 mirror_dir = self.cache_dir.child(file.path[len(self.cache_dir.path)+1:].split('/', 1)[0])
 218                 urlpath, newdir = self.db.storeFile(file, result.digest(), mirror_dir)
 219                 url = 'http:/' + file.path[len(self.cache_dir.path):]
 220             else:
 221                 urlpath, newdir = self.db.storeFile(file, result.digest(), self.scanning[0])
 222                 url = None
 223             if newdir:
 224                 self.manager.setDirectories(self.db.getAllDirectories())
 225             self.manager.new_cached_file(file, result, urlpath, url)
 226         else:
 227             log.msg('hash check of %s failed' % file.path)
 228             log.err(result)
 229
 230     def save_file(self, response, hash, url):
 231         """Save a downloaded file to the cache and stream it."""
 232         if response.code != 200:
 233             log.msg('File was not found (%r): %s' % (response, url))
 234             return response
 235
 236         log.msg('Returning file: %s' % url)
 237
 238         parsed = urlparse(url)
 239         destFile = self.cache_dir.preauthChild(parsed[1] + parsed[2])
 240         log.msg('Saving returned %r byte file to cache: %s' % (response.stream.length, destFile.path))
 241
 242         if destFile.exists():
 243             log.msg('File already exists, removing: %s' % destFile.path)
 244             destFile.remove()
 245         elif not destFile.parent().exists():
 246             destFile.parent().makedirs()
 247
 248         root, ext = os.path.splitext(destFile.basename())
 249         if root.lower() in DECOMPRESS_FILES and ext.lower() in DECOMPRESS_EXTS:
 250             ext = ext.lower()
 251             decFile = destFile.sibling(root)
 252             log.msg('Decompressing to: %s' % decFile.path)
 253             if decFile.exists():
 254                 log.msg('File already exists, removing: %s' % decFile.path)
 255                 decFile.remove()
 256         else:
 257             ext = None
 258             decFile = None
 259
 260         orig_stream = response.stream
 261         response.stream = ProxyFileStream(orig_stream, destFile, hash, ext, decFile)
 262         response.stream.doneDefer.addCallback(self._save_complete, url, destFile,
 263                                               response.headers.getHeader('Last-Modified'),
 264                                               ext, decFile)
 265         response.stream.doneDefer.addErrback(self.save_error, url)
 266         return response
 267
 268     def _save_complete(self, hash, url, destFile, modtime = None, ext = None, decFile = None):
 269         """Update the modification time and AptPackages."""
 270         if modtime:
 271             os.utime(destFile.path, (modtime, modtime))
 272             if ext:
 273                 os.utime(decFile.path, (modtime, modtime))
 274
 275         result = hash.verify()
 276         if result or result is None:
 277             if result:
 278                 log.msg('Hashes match: %s' % url)
 279             else:
 280                 log.msg('Hashed file to %s: %s' % (hash.hexdigest(), url))
 281
 282             mirror_dir = self.cache_dir.child(destFile.path[len(self.cache_dir.path)+1:].split('/', 1)[0])
 283             urlpath, newdir = self.db.storeFile(destFile, hash.digest(), mirror_dir)
 284             log.msg('now avaliable at %s: %s' % (urlpath, url))
 285
 286             if self.manager:
 287                 if newdir:
 288                     log.msg('A new web directory was created, so enable it')
 289                     self.manager.setDirectories(self.db.getAllDirectories())
 290
 291                 self.manager.new_cached_file(destFile, hash, urlpath, url)
 292                 if ext:
 293                     self.manager.new_cached_file(decFile, None, urlpath, url[:-len(ext)])
 294         else:
 295             log.msg("Hashes don't match %s != %s: %s" % (hash.hexexpected(), hash.hexdigest(), url))
 296             destFile.remove()
 297             if ext:
 298                 decFile.remove()
 299
 300     def save_error(self, failure, url):
 301         """An error has occurred in downloadign or saving the file."""
 302         log.msg('Error occurred downloading %s' % url)
 303         log.err(failure)
 304         return failure
 305
 306 class TestMirrorManager(unittest.TestCase):
 307     """Unit tests for the mirror manager."""
 308
 309     timeout = 20
 310     pending_calls = []
 311     client = None
 312
 313     def setUp(self):
 314         self.client = CacheManager(FilePath('/tmp/.apt-dht'))
 315
 316     def tearDown(self):
 317         for p in self.pending_calls:
 318             if p.active():
 319                 p.cancel()
 320         self.client = None
 321