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