zeroinstall/injector/fetch.py

   1 """
   2 Downloads feeds, keys, packages and icons.
   3 """
   4
   5 # Copyright (C) 2009, Thomas Leonard
   6 # See the README file for details, or visit http://0install.net.
   7
   8 from zeroinstall import _
   9 import os
  10 from logging import info, debug, warn
  11
  12 from zeroinstall.support import tasks, basedir
  13 from zeroinstall.injector.namespaces import XMLNS_IFACE, config_site
  14 from zeroinstall.injector.model import DownloadSource, Recipe, SafeException, escape
  15 from zeroinstall.injector.iface_cache import PendingFeed, ReplayAttack
  16 from zeroinstall.injector.handler import NoTrustedKeys
  17 from zeroinstall.injector import download
  18
  19 DEFAULT_KEY_LOOKUP_SERVER = 'https://keylookup.appspot.com'
  20
  21 def _escape_slashes(path):
  22         return path.replace('/', '%23')
  23
  24 def _get_feed_dir(feed):
  25         """The algorithm from 0mirror."""
  26         if '#' in feed:
  27                 raise SafeException(_("Invalid URL '%s'") % feed)
  28         scheme, rest = feed.split('://', 1)
  29         domain, rest = rest.split('/', 1)
  30         for x in [scheme, domain, rest]:
  31                 if not x or x.startswith(','):
  32                         raise SafeException(_("Invalid URL '%s'") % feed)
  33         return os.path.join('feeds', scheme, domain, _escape_slashes(rest))
  34
  35 class KeyInfoFetcher:
  36         """Fetches information about a GPG key from a key-info server.
  37         See L{Fetcher.fetch_key_info} for details.
  38         @since: 0.42
  39
  40         Example:
  41
  42         >>> kf = KeyInfoFetcher('https://server', fingerprint)
  43         >>> while True:
  44                 print kf.info
  45                 if kf.blocker is None: break
  46                 print kf.status
  47                 yield kf.blocker
  48         """
  49         def __init__(self, server, fingerprint):
  50                 self.fingerprint = fingerprint
  51                 self.info = []
  52                 self.blocker = None
  53
  54                 if server is None: return
  55
  56                 self.status = _('Fetching key information from %s...') % server
  57
  58                 dl = download.Download(server + '/key/' + fingerprint)
  59                 dl.start()
  60
  61                 from xml.dom import minidom
  62
  63                 @tasks.async
  64                 def fetch_key_info():
  65                         try:
  66                                 tempfile = dl.tempfile
  67                                 yield dl.downloaded
  68                                 self.blocker = None
  69                                 tasks.check(dl.downloaded)
  70                                 tempfile.seek(0)
  71                                 doc = minidom.parse(tempfile)
  72                                 if doc.documentElement.localName != 'key-lookup':
  73                                         raise SafeException(_('Expected <key-lookup>, not <%s>') % doc.documentElement.localName)
  74                                 self.info += doc.documentElement.childNodes
  75                         except Exception, ex:
  76                                 doc = minidom.parseString('<item vote="bad"/>')
  77                                 root = doc.documentElement
  78                                 root.appendChild(doc.createTextNode(_('Error getting key information: %s') % ex))
  79                                 self.info.append(root)
  80
  81                 self.blocker = fetch_key_info()
  82
  83 class Fetcher(object):
  84         """Downloads and stores various things.
  85         @ivar handler: handler to use for user-interaction
  86         @type handler: L{handler.Handler}
  87         @ivar key_info: caches information about GPG keys
  88         @type key_info: {str: L{KeyInfoFetcher}}
  89         @ivar key_info_server: the base URL of a key information server
  90         @type key_info_server: str
  91         @ivar feed_mirror: the base URL of a mirror site for keys and feeds
  92         @type feed_mirror: str | None
  93         """
  94         __slots__ = ['handler', 'feed_mirror', 'key_info_server', 'key_info']
  95
  96         def __init__(self, handler):
  97                 self.handler = handler
  98                 self.feed_mirror = "http://roscidus.com/0mirror"
  99                 self.key_info_server = DEFAULT_KEY_LOOKUP_SERVER
 100                 self.key_info = {}
 101
 102         @tasks.async
 103         def cook(self, required_digest, recipe, stores, force = False, impl_hint = None):
 104                 """Follow a Recipe.
 105                 @param impl_hint: the Implementation this is for (if any) as a hint for the GUI
 106                 @see: L{download_impl} uses this method when appropriate"""
 107                 # Maybe we're taking this metaphor too far?
 108
 109                 # Start downloading all the ingredients.
 110                 downloads = {}  # Downloads that are not yet successful
 111                 streams = {}    # Streams collected from successful downloads
 112
 113                 # Start a download for each ingredient
 114                 blockers = []
 115                 for step in recipe.steps:
 116                         blocker, stream = self.download_archive(step, force = force, impl_hint = impl_hint)
 117                         assert stream
 118                         blockers.append(blocker)
 119                         streams[step] = stream
 120
 121                 while blockers:
 122                         yield blockers
 123                         tasks.check(blockers)
 124                         blockers = [b for b in blockers if not b.happened]
 125
 126                 from zeroinstall.zerostore import unpack
 127
 128                 # Create an empty directory for the new implementation
 129                 store = stores.stores[0]
 130                 tmpdir = store.get_tmp_dir_for(required_digest)
 131                 try:
 132                         # Unpack each of the downloaded archives into it in turn
 133                         for step in recipe.steps:
 134                                 stream = streams[step]
 135                                 stream.seek(0)
 136                                 unpack.unpack_archive_over(step.url, stream, tmpdir, step.extract)
 137                         # Check that the result is correct and store it in the cache
 138                         store.check_manifest_and_rename(required_digest, tmpdir)
 139                         tmpdir = None
 140                 finally:
 141                         # If unpacking fails, remove the temporary directory
 142                         if tmpdir is not None:
 143                                 from zeroinstall import support
 144                                 support.ro_rmtree(tmpdir)
 145
 146         def get_feed_mirror(self, url):
 147                 """Return the URL of a mirror for this feed."""
 148                 if self.feed_mirror is None:
 149                         return None
 150                 import urlparse
 151                 if urlparse.urlparse(url).hostname == 'localhost':
 152                         return None
 153                 return '%s/%s/latest.xml' % (self.feed_mirror, _get_feed_dir(url))
 154
 155         def download_and_import_feed(self, feed_url, iface_cache, force = False):
 156                 """Download the feed, download any required keys, confirm trust if needed and import.
 157                 @param feed_url: the feed to be downloaded
 158                 @type feed_url: str
 159                 @param iface_cache: cache in which to store the feed
 160                 @type iface_cache: L{iface_cache.IfaceCache}
 161                 @param force: whether to abort and restart an existing download"""
 162                 from download import DownloadAborted
 163
 164                 debug(_("download_and_import_feed %(url)s (force = %(force)d)"), {'url': feed_url, 'force': force})
 165                 assert not feed_url.startswith('/')
 166
 167                 primary = self._download_and_import_feed(feed_url, iface_cache, force, use_mirror = False)
 168
 169                 @tasks.named_async("monitor feed downloads for " + feed_url)
 170                 def wait_for_downloads(primary):
 171                         # Download just the upstream feed, unless it takes too long...
 172                         timeout = tasks.TimeoutBlocker(5, 'Mirror timeout')             # 5 seconds
 173
 174                         yield primary, timeout
 175                         tasks.check(timeout)
 176
 177                         try:
 178                                 tasks.check(primary)
 179                                 if primary.happened:
 180                                         return          # OK, primary succeeded!
 181                                 # OK, maybe it's just being slow...
 182                                 info("Feed download from %s is taking a long time. Trying mirror too...", feed_url)
 183                                 primary_ex = None
 184                         except NoTrustedKeys, ex:
 185                                 raise                   # Don't bother trying the mirror if we have a trust problem
 186                         except ReplayAttack, ex:
 187                                 raise                   # Don't bother trying the mirror if we have a replay attack
 188                         except DownloadAborted, ex:
 189                                 raise                   # Don't bother trying the mirror if the user cancelled
 190                         except SafeException, ex:
 191                                 # Primary failed
 192                                 primary = None
 193                                 primary_ex = ex
 194                                 warn(_("Trying mirror, as feed download from %(url)s failed: %(exception)s"), {'url': feed_url, 'exception': ex})
 195
 196                         # Start downloading from mirror...
 197                         mirror = self._download_and_import_feed(feed_url, iface_cache, force, use_mirror = True)
 198
 199                         # Wait until both mirror and primary tasks are complete...
 200                         while True:
 201                                 blockers = filter(None, [primary, mirror])
 202                                 if not blockers:
 203                                         break
 204                                 yield blockers
 205
 206                                 if primary:
 207                                         try:
 208                                                 tasks.check(primary)
 209                                                 if primary.happened:
 210                                                         primary = None
 211                                                         # No point carrying on with the mirror once the primary has succeeded
 212                                                         if mirror:
 213                                                                 info(_("Primary feed download succeeded; aborting mirror download for %s") % feed_url)
 214                                                                 mirror.dl.abort()
 215                                         except SafeException, ex:
 216                                                 primary = None
 217                                                 primary_ex = ex
 218                                                 info(_("Feed download from %(url)s failed; still trying mirror: %(exception)s"), {'url': feed_url, 'exception': ex})
 219
 220                                 if mirror:
 221                                         try:
 222                                                 tasks.check(mirror)
 223                                                 if mirror.happened:
 224                                                         mirror = None
 225                                                         if primary_ex:
 226                                                                 # We already warned; no need to raise an exception too,
 227                                                                 # as the mirror download succeeded.
 228                                                                 primary_ex = None
 229                                         except ReplayAttack, ex:
 230                                                 info(_("Version from mirror is older than cached version; ignoring it: %s"), ex)
 231                                                 mirror = None
 232                                                 primary_ex = None
 233                                         except SafeException, ex:
 234                                                 info(_("Mirror download failed: %s"), ex)
 235                                                 mirror = None
 236
 237                         if primary_ex:
 238                                 raise primary_ex
 239
 240                 return wait_for_downloads(primary)
 241
 242         def _download_and_import_feed(self, feed_url, iface_cache, force, use_mirror):
 243                 """Download and import a feed.
 244                 @param use_mirror: False to use primary location; True to use mirror."""
 245                 if use_mirror:
 246                         url = self.get_feed_mirror(feed_url)
 247                         if url is None: return None
 248                 else:
 249                         url = feed_url
 250
 251                 dl = self.handler.get_download(url, force = force, hint = feed_url)
 252                 stream = dl.tempfile
 253
 254                 @tasks.named_async("fetch_feed " + url)
 255                 def fetch_feed():
 256                         yield dl.downloaded
 257                         tasks.check(dl.downloaded)
 258
 259                         pending = PendingFeed(feed_url, stream)
 260
 261                         if use_mirror:
 262                                 # If we got the feed from a mirror, get the key from there too
 263                                 key_mirror = self.feed_mirror + '/keys/'
 264                         else:
 265                                 key_mirror = None
 266
 267                         keys_downloaded = tasks.Task(pending.download_keys(self.handler, feed_hint = feed_url, key_mirror = key_mirror), _("download keys for %s") % feed_url)
 268                         yield keys_downloaded.finished
 269                         tasks.check(keys_downloaded.finished)
 270
 271                         iface = iface_cache.get_interface(pending.url)
 272                         if not iface_cache.update_interface_if_trusted(iface, pending.sigs, pending.new_xml):
 273                                 blocker = self.handler.confirm_keys(pending, self.fetch_key_info)
 274                                 if blocker:
 275                                         yield blocker
 276                                         tasks.check(blocker)
 277                                 if not iface_cache.update_interface_if_trusted(iface, pending.sigs, pending.new_xml):
 278                                         raise NoTrustedKeys(_("No signing keys trusted; not importing"))
 279
 280                 task = fetch_feed()
 281                 task.dl = dl
 282                 return task
 283
 284         def fetch_key_info(self, fingerprint):
 285                 try:
 286                         return self.key_info[fingerprint]
 287                 except KeyError:
 288                         self.key_info[fingerprint] = info = KeyInfoFetcher(self.key_info_server, fingerprint)
 289                         return info
 290
 291         def download_impl(self, impl, retrieval_method, stores, force = False):
 292                 """Download an implementation.
 293                 @param impl: the selected implementation
 294                 @type impl: L{model.ZeroInstallImplementation}
 295                 @param retrieval_method: a way of getting the implementation (e.g. an Archive or a Recipe)
 296                 @type retrieval_method: L{model.RetrievalMethod}
 297                 @param stores: where to store the downloaded implementation
 298                 @type stores: L{zerostore.Stores}
 299                 @param force: whether to abort and restart an existing download
 300                 @rtype: L{tasks.Blocker}"""
 301                 assert impl
 302                 assert retrieval_method
 303
 304                 from zeroinstall.zerostore import manifest
 305                 best = None
 306                 for digest in impl.digests:
 307                         alg_name = digest.split('=', 1)[0]
 308                         alg = manifest.algorithms.get(alg_name, None)
 309                         if alg and (best is None or best.rating < alg.rating):
 310                                 best = alg
 311                                 required_digest = digest
 312
 313                 if best is None:
 314                         if not impl.digests:
 315                                 raise SafeException(_("No <manifest-digest> given for '%(implementation)s' version %(version)s") %
 316                                                 {'implementation': impl.feed.get_name(), 'version': impl.get_version()})
 317                         raise SafeException(_("Unknown digest algorithms '%(algorithms)s' for '%(implementation)s' version %(version)s") %
 318                                         {'algorithms': impl.digests, 'implementation': impl.feed.get_name(), 'version': impl.get_version()})
 319
 320                 @tasks.async
 321                 def download_impl():
 322                         if isinstance(retrieval_method, DownloadSource):
 323                                 blocker, stream = self.download_archive(retrieval_method, force = force, impl_hint = impl)
 324                                 yield blocker
 325                                 tasks.check(blocker)
 326
 327                                 stream.seek(0)
 328                                 self._add_to_cache(required_digest, stores, retrieval_method, stream)
 329                         elif isinstance(retrieval_method, Recipe):
 330                                 blocker = self.cook(required_digest, retrieval_method, stores, force, impl_hint = impl)
 331                                 yield blocker
 332                                 tasks.check(blocker)
 333                         else:
 334                                 raise Exception(_("Unknown download type for '%s'") % retrieval_method)
 335
 336                         self.handler.impl_added_to_store(impl)
 337                 return download_impl()
 338
 339         def _add_to_cache(self, required_digest, stores, retrieval_method, stream):
 340                 assert isinstance(retrieval_method, DownloadSource)
 341                 url = retrieval_method.url
 342                 stores.add_archive_to_cache(required_digest, stream, retrieval_method.url, retrieval_method.extract,
 343                                                  type = retrieval_method.type, start_offset = retrieval_method.start_offset or 0)
 344
 345         def download_archive(self, download_source, force = False, impl_hint = None):
 346                 """Fetch an archive. You should normally call L{download_impl}
 347                 instead, since it handles other kinds of retrieval method too."""
 348                 from zeroinstall.zerostore import unpack
 349
 350                 url = download_source.url
 351                 if not (url.startswith('http:') or url.startswith('https:') or url.startswith('ftp:')):
 352                         raise SafeException(_("Unknown scheme in download URL '%s'") % url)
 353
 354                 mime_type = download_source.type
 355                 if not mime_type:
 356                         mime_type = unpack.type_from_url(download_source.url)
 357                 if not mime_type:
 358                         raise SafeException(_("No 'type' attribute on archive, and I can't guess from the name (%s)") % download_source.url)
 359                 unpack.check_type_ok(mime_type)
 360                 dl = self.handler.get_download(download_source.url, force = force, hint = impl_hint)
 361                 dl.expected_size = download_source.size + (download_source.start_offset or 0)
 362                 return (dl.downloaded, dl.tempfile)
 363
 364         def download_icon(self, interface, force = False, modification_time = None):
 365                 """Download an icon for this interface and add it to the
 366                 icon cache. If the interface has no icon or we are offline, do nothing.
 367                 @return: the task doing the import, or None
 368                 @rtype: L{tasks.Task}"""
 369                 debug(_("download_icon %(interface)s (force = %(force)d)"), {'interface': interface, 'force': force})
 370
 371                 # Find a suitable icon to download
 372                 for icon in interface.get_metadata(XMLNS_IFACE, 'icon'):
 373                         type = icon.getAttribute('type')
 374                         if type != 'image/png':
 375                                 debug(_('Skipping non-PNG icon'))
 376                                 continue
 377                         source = icon.getAttribute('href')
 378                         if source:
 379                                 break
 380                         warn(_('Missing "href" attribute on <icon> in %s'), interface)
 381                 else:
 382                         info(_('No PNG icons found in %s'), interface)
 383                         return
 384
 385                 try:
 386                         dl = self.handler.monitored_downloads[source]
 387                         if dl and force:
 388                                 dl.abort()
 389                                 raise KeyError
 390                 except KeyError:
 391                         dl = download.Download(source, hint = interface, modification_time = modification_time)
 392                         self.handler.monitor_download(dl)
 393
 394                 @tasks.async
 395                 def download_and_add_icon():
 396                         stream = dl.tempfile
 397                         yield dl.downloaded
 398                         try:
 399                                 tasks.check(dl.downloaded)
 400                                 if dl.unmodified: return
 401                                 stream.seek(0)
 402
 403                                 import shutil
 404                                 icons_cache = basedir.save_cache_path(config_site, 'interface_icons')
 405                                 icon_file = file(os.path.join(icons_cache, escape(interface.uri)), 'w')
 406                                 shutil.copyfileobj(stream, icon_file)
 407                         except Exception, ex:
 408                                 self.handler.report_error(ex)
 409
 410                 return download_and_add_icon()
 411
 412         def download_impls(self, implementations, stores):
 413                 """Download the given implementations, choosing a suitable retrieval method for each."""
 414                 blockers = []
 415
 416                 to_download = []
 417                 for impl in implementations:
 418                         debug(_("start_downloading_impls: for %(feed)s get %(implementation)s"), {'feed': impl.feed, 'implementation': impl})
 419                         source = self.get_best_source(impl)
 420                         if not source:
 421                                 raise SafeException(_("Implementation %(implementation_id)s of interface %(interface)s"
 422                                         " cannot be downloaded (no download locations given in "
 423                                         "interface!)") % {'implementation_id': impl.id, 'interface': impl.feed.get_name()})
 424                         to_download.append((impl, source))
 425
 426                 for impl, source in to_download:
 427                         blockers.append(self.download_impl(impl, source, stores))
 428
 429                 if not blockers:
 430                         return None
 431
 432                 @tasks.async
 433                 def download_impls(blockers):
 434                         # Record the first error log the rest
 435                         error = []
 436                         def dl_error(ex, tb = None):
 437                                 if error:
 438                                         self.handler.report_error(ex)
 439                                 else:
 440                                         error.append(ex)
 441                         while blockers:
 442                                 yield blockers
 443                                 tasks.check(blockers, dl_error)
 444
 445                                 blockers = [b for b in blockers if not b.happened]
 446                         if error:
 447                                 raise error[0]
 448
 449                 return download_impls(blockers)
 450
 451         def get_best_source(self, impl):
 452                 """Return the best download source for this implementation.
 453                 @rtype: L{model.RetrievalMethod}"""
 454                 if impl.download_sources:
 455                         return impl.download_sources[0]
 456                 return None
 457