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