API changes: use Feed rather than Interface in many places
[zeroinstall/zeroinstall-afb.git] / zeroinstall / injector / iface_cache.py
blob6a0853b2aef1abe4572540e5b549230e94830038
1 """
2 Manages the feed cache.
4 @var iface_cache: A singleton cache object. You should normally use this rather than
5 creating new cache objects.
7 """
8 # Copyright (C) 2009, Thomas Leonard
9 # See the README file for details, or visit http://0install.net.
11 # Note:
13 # We need to know the modification time of each interface, because we refuse
14 # to update to an older version (this prevents an attack where the attacker
15 # sends back an old version which is correctly signed but has a known bug).
17 # The way we store this is a bit complicated due to backward compatibility:
19 # - GPG-signed interfaces have their signatures removed and a last-modified
20 # attribute is stored containing the date from the signature.
22 # - XML-signed interfaces are stored unmodified with their signatures. The
23 # date is extracted from the signature when needed.
25 # - Older versions used to add the last-modified attribute even to files
26 # with XML signatures - these files therefore have invalid signatures and
27 # we extract from the attribute for these.
29 # Eventually, support for the first and third cases will be removed.
31 import os, sys, time
32 from logging import debug, info, warn
33 from cStringIO import StringIO
35 from zeroinstall import _
36 from zeroinstall.support import basedir
37 from zeroinstall.injector import reader, model
38 from zeroinstall.injector.namespaces import config_site, config_prog
39 from zeroinstall.injector.model import Interface, escape, unescape
40 from zeroinstall import zerostore, SafeException
42 def _pretty_time(t):
43 assert isinstance(t, (int, long)), t
44 return time.strftime('%Y-%m-%d %H:%M:%S UTC', time.localtime(t))
46 class ReplayAttack(SafeException):
47 """Attempt to import a feed that's older than the one in the cache."""
48 pass
50 class PendingFeed(object):
51 """A feed that has been downloaded but not yet added to the interface cache.
52 Feeds remain in this state until the user confirms that they trust at least
53 one of the signatures.
54 @ivar url: URL for the feed
55 @type url: str
56 @ivar signed_data: the untrusted data
57 @type signed_data: stream
58 @ivar sigs: signatures extracted from signed_data
59 @type sigs: [L{gpg.Signature}]
60 @ivar new_xml: the payload of the signed_data, or the whole thing if XML
61 @type new_xml: str
62 @since: 0.25"""
63 __slots__ = ['url', 'signed_data', 'sigs', 'new_xml']
65 def __init__(self, url, signed_data):
66 """Downloaded data is a GPG-signed message.
67 @param url: the URL of the downloaded feed
68 @type url: str
69 @param signed_data: the downloaded data (not yet trusted)
70 @type signed_data: stream
71 @raise SafeException: if the data is not signed, and logs the actual data"""
72 self.url = url
73 self.signed_data = signed_data
74 self.recheck()
76 def download_keys(self, handler, feed_hint = None, key_mirror = None):
77 """Download any required GPG keys not already on our keyring.
78 When all downloads are done (successful or otherwise), add any new keys
79 to the keyring, L{recheck}.
80 @param handler: handler to manage the download
81 @type handler: L{handler.Handler}
82 @param key_mirror: URL of directory containing keys, or None to use feed's directory
83 @type key_mirror: str
84 """
85 downloads = {}
86 blockers = []
87 for x in self.sigs:
88 key_id = x.need_key()
89 if key_id:
90 import urlparse
91 key_url = urlparse.urljoin(key_mirror or self.url, '%s.gpg' % key_id)
92 info(_("Fetching key from %s"), key_url)
93 dl = handler.get_download(key_url, hint = feed_hint)
94 downloads[dl.downloaded] = (dl, dl.tempfile)
95 blockers.append(dl.downloaded)
97 exception = None
98 any_success = False
100 from zeroinstall.support import tasks
102 while blockers:
103 yield blockers
105 old_blockers = blockers
106 blockers = []
108 for b in old_blockers:
109 try:
110 tasks.check(b)
111 if b.happened:
112 dl, stream = downloads[b]
113 stream.seek(0)
114 self._downloaded_key(stream)
115 any_success = True
116 else:
117 blockers.append(b)
118 except Exception:
119 _type, exception, tb = sys.exc_info()
120 warn(_("Failed to import key for '%(url)s': %(exception)s"), {'url': self.url, 'exception': str(exception)})
122 if exception and not any_success:
123 raise exception, None, tb
125 self.recheck()
127 def _downloaded_key(self, stream):
128 import shutil, tempfile
129 from zeroinstall.injector import gpg
131 info(_("Importing key for feed '%s'"), self.url)
133 # Python2.4: can't call fileno() on stream, so save to tmp file instead
134 tmpfile = tempfile.TemporaryFile(prefix = 'injector-dl-data-')
135 try:
136 shutil.copyfileobj(stream, tmpfile)
137 tmpfile.flush()
139 tmpfile.seek(0)
140 gpg.import_key(tmpfile)
141 finally:
142 tmpfile.close()
144 def recheck(self):
145 """Set new_xml and sigs by reading signed_data.
146 You need to call this when previously-missing keys are added to the GPG keyring."""
147 import gpg
148 try:
149 self.signed_data.seek(0)
150 stream, sigs = gpg.check_stream(self.signed_data)
151 assert sigs
153 data = stream.read()
154 if stream is not self.signed_data:
155 stream.close()
157 self.new_xml = data
158 self.sigs = sigs
159 except:
160 self.signed_data.seek(0)
161 info(_("Failed to check GPG signature. Data received was:\n") + repr(self.signed_data.read()))
162 raise
164 class IfaceCache(object):
166 The interface cache stores downloaded and verified interfaces in
167 ~/.cache/0install.net/interfaces (by default).
169 There are methods to query the cache, add to it, check signatures, etc.
171 The cache is updated by L{fetch.Fetcher}.
173 Confusingly, this class is really two caches combined: the in-memory
174 cache of L{model.Interface} objects, and an on-disk cache of L{model.ZeroInstallFeed}s.
175 It will probably be split into two in future.
177 @see: L{iface_cache} - the singleton IfaceCache instance.
180 __slots__ = ['_interfaces', 'stores', '_feeds']
182 def __init__(self):
183 self._interfaces = {}
184 self._feeds = {}
186 self.stores = zerostore.Stores()
188 def update_interface_if_trusted(self, interface, sigs, xml):
189 import warnings
190 warnings.warn(_("Use update_feed_if_trusted instead"), DeprecationWarning, stacklevel = 2)
191 return self.update_feed_if_trusted(interface.uri, sigs, xml)
193 def update_feed_if_trusted(self, feed_url, sigs, xml):
194 """Update a cached feed (using L{update_feed_from_network})
195 if we trust the signatures.
196 If we don't trust any of the signatures, do nothing.
197 @param feed_url: the feed being updated
198 @type feed_url: str
199 @param sigs: signatures from L{gpg.check_stream}
200 @type sigs: [L{gpg.Signature}]
201 @param xml: the downloaded replacement feed document
202 @type xml: str
203 @return: True if the feed was updated
204 @rtype: bool
206 import trust
207 updated = self._oldest_trusted(sigs, trust.domain_from_url(feed_url))
208 if updated is None: return False # None are trusted
210 self.update_feed_from_network(feed_url, xml, updated)
211 return True
213 def update_interface_from_network(self, interface, new_xml, modified_time):
214 import warnings
215 warnings.warn(_("Use update_feed_from_network instead"), DeprecationWarning, stacklevel = 2)
216 self.update_feed_from_network(interface.uri, new_xml, modified_time)
218 def update_feed_from_network(self, feed_url, new_xml, modified_time):
219 """Update a cached feed.
220 Called by L{update_feed_if_trusted} if we trust this data.
221 After a successful update, L{writer} is used to update the feed's
222 last_checked time.
223 @param feed_url: the feed being updated
224 @type feed_url: L{model.Interface}
225 @param new_xml: the downloaded replacement feed document
226 @type new_xml: str
227 @param modified_time: the timestamp of the oldest trusted signature
228 (used as an approximation to the feed's modification time)
229 @type modified_time: long
230 @raises ReplayAttack: if modified_time is older than the currently cached time
232 debug(_("Updating '%(interface)s' from network; modified at %(time)s") %
233 {'interface': feed_url, 'time': _pretty_time(modified_time)})
235 if '\n<!-- Base64 Signature' not in new_xml:
236 # Only do this for old-style feeds without
237 # signatures Otherwise, we can get the time from the
238 # signature, and adding this attribute just makes the
239 # signature invalid.
240 from xml.dom import minidom
241 doc = minidom.parseString(new_xml)
242 doc.documentElement.setAttribute('last-modified', str(modified_time))
243 new_xml = StringIO()
244 doc.writexml(new_xml)
245 new_xml = new_xml.getvalue()
247 self._import_new_feed(feed_url, new_xml, modified_time)
249 feed = self.get_feed(feed_url)
251 import writer
252 feed.last_checked = long(time.time())
253 writer.save_feed(feed)
255 info(_("Updated feed cache entry for %(interface)s (modified %(time)s)"),
256 {'interface': feed.get_name(), 'time': _pretty_time(modified_time)})
258 def _import_new_feed(self, feed_url, new_xml, modified_time):
259 """Write new_xml into the cache.
260 @param feed_url: the URL for the feed being updated
261 @param new_xml: the data to write
262 @param modified_time: when new_xml was modified
263 @raises ReplayAttack: if the new mtime is older than the current one
265 assert modified_time
267 upstream_dir = basedir.save_cache_path(config_site, 'interfaces')
268 cached = os.path.join(upstream_dir, escape(feed_url))
270 old_modified = None
271 if os.path.exists(cached):
272 old_xml = file(cached).read()
273 if old_xml == new_xml:
274 debug(_("No change"))
275 # Update in-memory copy, in case someone else updated the disk copy
276 self.get_feed(feed_url, force = True)
277 return
278 old_modified = int(os.stat(cached).st_mtime)
280 # Do we need to write this temporary file now?
281 stream = file(cached + '.new', 'w')
282 stream.write(new_xml)
283 stream.close()
284 os.utime(cached + '.new', (modified_time, modified_time))
285 new_mtime = reader.check_readable(feed_url, cached + '.new')
286 assert new_mtime == modified_time
288 old_modified = self._get_signature_date(feed_url) or old_modified
290 if old_modified:
291 if new_mtime < old_modified:
292 os.unlink(cached + '.new')
293 raise ReplayAttack(_("New feed's modification time is "
294 "before old version!\nInterface: %(iface)s\nOld time: %(old_time)s\nNew time: %(new_time)s\n"
295 "Refusing update.")
296 % {'iface': feed_url, 'old_time': _pretty_time(old_modified), 'new_time': _pretty_time(new_mtime)})
297 if new_mtime == old_modified:
298 # You used to have to update the modification time manually.
299 # Now it comes from the signature, this check isn't useful
300 # and often causes problems when the stored format changes
301 # (e.g., when we stopped writing last-modified attributes)
302 pass
303 #raise SafeException("Interface has changed, but modification time "
304 # "hasn't! Refusing update.")
305 os.rename(cached + '.new', cached)
306 debug(_("Saved as %s") % cached)
308 self.get_feed(feed_url, force = True)
310 def get_feed(self, url, force = False):
311 """Get a feed from the cache.
312 @param url: the URL of the feed
313 @param force: load the file from disk again
314 @return: the feed, or None if it isn't cached
315 @rtype: L{model.ZeroInstallFeed}"""
316 if not force:
317 feed = self._feeds.get(url, False)
318 if feed != False:
319 return feed
321 feed = self._feeds[url] = reader.load_feed_from_cache(url)
322 return feed
324 def get_interface(self, uri):
325 """Get the interface for uri, creating a new one if required.
326 New interfaces are initialised from the disk cache, but not from
327 the network.
328 @param uri: the URI of the interface to find
329 @rtype: L{model.Interface}
331 if type(uri) == str:
332 uri = unicode(uri)
333 assert isinstance(uri, unicode)
335 if uri in self._interfaces:
336 return self._interfaces[uri]
338 debug(_("Initialising new interface object for %s"), uri)
339 self._interfaces[uri] = Interface(uri)
340 reader.update_from_cache(self._interfaces[uri])
341 return self._interfaces[uri]
343 def list_all_interfaces(self):
344 """List all interfaces in the cache.
345 @rtype: [str]
347 all = set()
348 for d in basedir.load_cache_paths(config_site, 'interfaces'):
349 for leaf in os.listdir(d):
350 if not leaf.startswith('.'):
351 all.add(unescape(leaf))
352 for d in basedir.load_config_paths(config_site, config_prog, 'user_overrides'):
353 for leaf in os.listdir(d):
354 if not leaf.startswith('.'):
355 all.add(unescape(leaf))
356 return list(all) # Why not just return the set?
358 def get_icon_path(self, iface):
359 """Get the path of a cached icon for an interface.
360 @param iface: interface whose icon we want
361 @return: the path of the cached icon, or None if not cached.
362 @rtype: str"""
363 return basedir.load_first_cache(config_site, 'interface_icons',
364 escape(iface.uri))
366 def get_cached_signatures(self, uri):
367 """Verify the cached interface using GPG.
368 Only new-style XML-signed interfaces retain their signatures in the cache.
369 @param uri: the feed to check
370 @type uri: str
371 @return: a list of signatures, or None
372 @rtype: [L{gpg.Signature}] or None
373 @since: 0.25"""
374 import gpg
375 if uri.startswith('/'):
376 old_iface = uri
377 else:
378 old_iface = basedir.load_first_cache(config_site, 'interfaces', escape(uri))
379 if old_iface is None:
380 return None
381 try:
382 return gpg.check_stream(file(old_iface))[1]
383 except SafeException, ex:
384 debug(_("No signatures (old-style interface): %s") % ex)
385 return None
387 def _get_signature_date(self, uri):
388 """Read the date-stamp from the signature of the cached interface.
389 If the date-stamp is unavailable, returns None."""
390 import trust
391 sigs = self.get_cached_signatures(uri)
392 if sigs:
393 return self._oldest_trusted(sigs, trust.domain_from_url(uri))
395 def _oldest_trusted(self, sigs, domain):
396 """Return the date of the oldest trusted signature in the list, or None if there
397 are no trusted sigs in the list."""
398 trusted = [s.get_timestamp() for s in sigs if s.is_trusted(domain)]
399 if trusted:
400 return min(trusted)
401 return None
403 def mark_as_checking(self, url):
404 """Touch a 'last_check_attempt_timestamp' file for this feed.
405 If url is a local path, nothing happens.
406 This prevents us from repeatedly trying to download a failing feed many
407 times in a short period."""
408 if url.startswith('/'):
409 return
410 feeds_dir = basedir.save_cache_path(config_site, config_prog, 'last-check-attempt')
411 timestamp_path = os.path.join(feeds_dir, model._pretty_escape(url))
412 fd = os.open(timestamp_path, os.O_WRONLY | os.O_CREAT, 0644)
413 os.close(fd)
414 os.utime(timestamp_path, None) # In case file already exists
416 def get_last_check_attempt(self, url):
417 """Return the time of the most recent update attempt for a feed.
418 @see: L{mark_as_checking}
419 @return: The time, or None if none is recorded
420 @rtype: float | None"""
421 timestamp_path = basedir.load_first_cache(config_site, config_prog, 'last-check-attempt', model._pretty_escape(url))
422 if timestamp_path:
423 return os.stat(timestamp_path).st_mtime
424 return None
426 def get_feed_imports(self, iface):
427 """Get all feeds that add to this interface.
428 This is the feeds explicitly added by the user, feeds added by the distribution,
429 and feeds imported by a <feed> in the main feed (but not recursively, at present).
430 @rtype: L{Feed}"""
431 main_feed = self.get_feed(iface.uri)
432 if main_feed:
433 return iface.extra_feeds + main_feed.feeds
434 else:
435 return iface.extra_feeds
437 def get_feeds(self, iface):
438 """Get all feeds for this interface. This is a mapping from feed URLs
439 to ZeroInstallFeeds. It includes the interface's main feed, plus the
440 resolution of every feed returne by L{get_feed_imports}. Uncached
441 feeds are indicated by a value of None.
442 @rtype: {str: L{ZeroInstallFeed} | None}"""
443 main_feed = self.get_feed(iface.uri)
444 results = {iface.uri: main_feed}
445 for imp in iface.extra_feeds:
446 results[imp.uri] = self.get_feed(imp.uri)
447 if main_feed:
448 for imp in main_feed.feeds:
449 results[imp.uri] = self.get_feed(imp.uri)
450 return results
452 def get_implementations(self, iface):
453 """Return all implementations from all of iface's feeds.
454 @rtype: [L{Implementation}]"""
455 impls = []
456 for feed in self.get_feeds(iface).itervalues():
457 if feed:
458 impls += feed.implementations.values()
459 return impls
461 iface_cache = IfaceCache()