def resolve_guid(guid, suffix=None):
"""Resolve GUID to corresponding URL and return result of appropriate
view function. This effectively yields a redirect without changing the
displayed URL of the page.
:param guid: GUID value (not the object)
:param suffix: String to append to GUID route
:return: Werkzeug response
"""
# Get prefix; handles API routes
prefix = request.path.split(guid)[0].rstrip('/')
# Look up GUID
guid_object = Guid.load(guid)
if guid_object:
# verify that the object is a GuidStoredObject descendant. If a model
# was once a descendant but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a redirect_mode attribute or otherwise don't behave as
# expected.
if not isinstance(guid_object.referent, GuidStoredObject):
sentry.log_message(
'Guid `{}` resolved to non-guid object'.format(guid)
)
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
mode = referent.redirect_mode
if mode is None:
raise HTTPError(http.NOT_FOUND)
url = referent.deep_url if mode == 'proxy' else referent.url
url = _build_guid_url(url, prefix, suffix)
# Always redirect API URLs; URL should identify endpoint being called
if prefix or mode == 'redirect':
if request.query_string:
url += '?' + request.query_string
return redirect(url)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(
_build_guid_url(
guid.lower(), prefix, suffix
)
)
# GUID not found
raise HTTPError(http.NOT_FOUND)
def resolve_guid(guid, suffix=None):
"""Load GUID by primary key, look up the corresponding view function in the
routing table, and return the return value of the view function without
changing the URL.
:param str guid: GUID primary key
:param str suffix: Remainder of URL after the GUID
:return: Return value of proxied view function
"""
try:
# Look up
guid_object = Guid.load(guid)
except KeyError as e:
if e.message == 'osfstorageguidfile': # Used when an old detached OsfStorageGuidFile object is accessed
raise HTTPError(http.NOT_FOUND)
else:
raise e
if guid_object:
# verify that the object implements a GuidStoredObject-like interface. If a model
# was once GuidStoredObject-like but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a deep_url attribute or otherwise don't behave as
# expected.
if not hasattr(guid_object.referent, 'deep_url'):
sentry.log_message(
'Guid `{}` resolved to an object with no deep_url'.format(guid)
)
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
if not referent.deep_url:
raise HTTPError(http.NOT_FOUND)
if isinstance(referent, PreprintService):
return send_from_directory(
os.path.abspath(os.path.join(os.getcwd(), EXTERNAL_EMBER_APPS['preprints']['path'])),
'index.html'
)
url = _build_guid_url(urllib.unquote(referent.deep_url), suffix)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(
_build_guid_url(guid.lower(), suffix)
)
# GUID not found
raise HTTPError(http.NOT_FOUND)
def resolve_guid(guid, suffix=None):
"""Load GUID by primary key, look up the corresponding view function in the
routing table, and return the return value of the view function without
changing the URL.
:param str guid: GUID primary key
:param str suffix: Remainder of URL after the GUID
:return: Return value of proxied view function
"""
try:
# Look up
guid_object = Guid.load(guid)
except KeyError as e:
if e.message == 'osfstorageguidfile': # Used when an old detached OsfStorageGuidFile object is accessed
raise HTTPError(http.NOT_FOUND)
else:
raise e
if guid_object:
# verify that the object implements a GuidStoredObject-like interface. If a model
# was once GuidStoredObject-like but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a deep_url attribute or otherwise don't behave as
# expected.
if not hasattr(guid_object.referent, 'deep_url'):
sentry.log_message(
'Guid `{}` resolved to an object with no deep_url'.format(guid)
)
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
if not referent.deep_url:
raise HTTPError(http.NOT_FOUND)
if isinstance(referent, PreprintService):
return send_from_directory(
os.path.abspath(os.path.join(os.getcwd(), EXTERNAL_EMBER_APPS['preprints']['path'])),
'index.html'
)
url = _build_guid_url(urllib.unquote(referent.deep_url), suffix)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(
_build_guid_url(guid.lower(), suffix)
)
# GUID not found
raise HTTPError(http.NOT_FOUND)
def resolve_guid(guid, suffix=None):
"""Resolve GUID to corresponding URL and return result of appropriate
view function. This effectively yields a redirect without changing the
displayed URL of the page.
:param guid: GUID value (not the object)
:param suffix: String to append to GUID route
:return: Werkzeug response
"""
# Get prefix; handles API routes
prefix = request.path.split(guid)[0].rstrip('/')
# Look up GUID
guid_object = Guid.load(guid)
if guid_object:
# verify that the object is a GuidStoredObject descendant. If a model
# was once a descendant but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a redirect_mode attribute or otherwise don't behave as
# expected.
if not isinstance(guid_object.referent, GuidStoredObject):
sentry.log_message(
'Guid `{}` resolved to non-guid object'.format(guid))
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
mode = referent.redirect_mode
if mode is None:
raise HTTPError(http.NOT_FOUND)
url = referent.deep_url if mode == 'proxy' else referent.url
url = _build_guid_url(url, prefix, suffix)
# Always redirect API URLs; URL should identify endpoint being called
if prefix or mode == 'redirect':
if request.query_string:
url += '?' + request.query_string
return redirect(url)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(_build_guid_url(guid.lower(), prefix, suffix))
# GUID not found
raise HTTPError(http.NOT_FOUND)
def migrate(targets):
collections = targets.pop('collections')
assert len(targets), 'Must specify object to create new guid for'
assert len(
targets) == 1, 'Can only create new guid for one object at a time'
old_id = targets.values()[0]
node = Node.load(old_id)
new_guid = Guid.generate(referent=node)
logger.info('* Created new guid {} for node {}'.format(
new_guid._id, old_id))
logger.info('* Preparing to set references.')
fix_backrefs(node)
node.reload()
node._id = new_guid._id
node.save()
new_guid.referent = node
new_guid.save()
for collection, _id_list in collections.iteritems():
assert type(
_id_list
) == list, 'Expected type list for collection {} ids, got {}'.format(
collection, type(_id_list))
for _id in _id_list:
logger.info('** Updating {} ({})'.format(_id, collection))
doc = clean_dict(database[collection].find_one({'_id': _id}))
replacement = json.loads(
re.sub(r'\b{}\b'.format(old_id), new_guid._id,
json.dumps(doc)))
database[collection].find_and_modify({'_id': _id},
{'$set': replacement})
logger.info('*** Updated {} ({}): \n{}\n'.format(
_id, collection, replacement))
logger.info('Successfully migrate {} to {}'.format(old_id, new_guid._id))
def test_restore_file(self):
root = models.StoredFileNode(
path='root',
name='rootfolder',
is_file=False,
node=self.node,
provider='test',
materialized_path='/long/path/to',
).wrapped()
root.save()
fn = models.StoredFileNode(
parent=root._id,
path='afile',
name='name',
is_file=True,
node=self.node,
provider='test',
materialized_path='/long/path/to/name',
).wrapped()
guid = Guid.generate(fn)
before = fn.to_storage()
trashed = fn.delete(user=self.user)
restored = trashed.restore()
assert_equal(restored.to_storage(), before)
assert_equal(models.TrashedFileNode.load(trashed._id), None)
# Guid is repointed
guid.reload()
assert_equal(guid.referent, restored)
def resolve_target(node, guid):
if not guid:
return node
target = Guid.load(guid)
if target is None:
raise HTTPError(http.BAD_REQUEST)
return target.referent
def resolve_target(node, guid):
if not guid:
return node
target = Guid.load(guid)
if target is None:
raise HTTPError(http.BAD_REQUEST)
return target.referent
def test_migrate(self):
# Sanity check
for obj in self.legacy_objs:
guid_obj = Guid.load(obj._id)
assert_equal(guid_obj.referent, obj)
nobjs = OsfStorageGuidFile.find().count()
main(dry_run=False)
Guid._clear_caches()
for obj in self.legacy_objs:
guid_obj = Guid.load(obj._id)
assert_not_equal(guid_obj.referent, obj)
assert_true(isinstance(guid_obj.referent, OsfStorageGuidFile))
assert_equal(guid_obj.referent.node, self.project)
assert_equal(guid_obj.referent.path, obj.name)
assert_equal(guid_obj.referent._id, obj._id)
assert_equal(OsfStorageGuidFile.find().count(), nobjs + 3)
# Test idempotence
main(dry_run=False)
assert_equal(OsfStorageGuidFile.find().count(), nobjs + 3)
def test_migrate(self):
# Sanity check
for obj in self.legacy_objs:
guid_obj = Guid.load(obj._id)
assert_equal(guid_obj.referent, obj)
nobjs = OsfStorageGuidFile.find().count()
main(dry_run=False)
Guid._clear_caches()
for obj in self.legacy_objs:
guid_obj = Guid.load(obj._id)
assert_not_equal(guid_obj.referent, obj)
assert_true(isinstance(guid_obj.referent, OsfStorageGuidFile))
assert_equal(guid_obj.referent.node, self.project)
assert_equal(guid_obj.referent.path, obj.name)
assert_equal(guid_obj.referent._id, obj._id)
assert_equal(OsfStorageGuidFile.find().count(), nobjs + 3)
# Test idempotence
main(dry_run=False)
assert_equal(OsfStorageGuidFile.find().count(), nobjs + 3)
def resolve_guid(guid, suffix=None):
"""Load GUID by primary key, look up the corresponding view function in the
routing table, and return the return value of the view function without
changing the URL.
:param str guid: GUID primary key
:param str suffix: Remainder of URL after the GUID
:return: Return value of proxied view function
"""
# Look up GUID
guid_object = Guid.load(guid)
if guid_object:
# verify that the object is a GuidStoredObject descendant. If a model
# was once a descendant but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a redirect_mode attribute or otherwise don't behave as
# expected.
if not isinstance(guid_object.referent, GuidStoredObject):
sentry.log_message(
'Guid `{}` resolved to non-guid object'.format(guid)
)
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
if not referent.deep_url:
raise HTTPError(http.NOT_FOUND)
url = _build_guid_url(referent.deep_url, suffix)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(
_build_guid_url(guid.lower(), suffix)
)
# GUID not found
raise HTTPError(http.NOT_FOUND)
def resolve_guid(guid, suffix=None):
"""Load GUID by primary key, look up the corresponding view function in the
routing table, and return the return value of the view function without
changing the URL.
:param str guid: GUID primary key
:param str suffix: Remainder of URL after the GUID
:return: Return value of proxied view function
"""
# Look up GUID
guid_object = Guid.load(guid)
if guid_object:
# verify that the object implements a GuidStoredObject-like interface. If a model
# was once GuidStoredObject-like but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a deep_url attribute or otherwise don't behave as
# expected.
if not hasattr(guid_object.referent, 'deep_url'):
sentry.log_message(
'Guid `{}` resolved to an object with no deep_url'.format(guid)
)
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
if not referent.deep_url:
raise HTTPError(http.NOT_FOUND)
url = _build_guid_url(urllib.unquote(referent.deep_url), suffix)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(
_build_guid_url(guid.lower(), suffix)
)
# GUID not found
raise HTTPError(http.NOT_FOUND)
def migrate():
for target in targets:
guid = Guid.load(target['guid'])
good_sfn = StoredFileNode.load(target['good'])
bad_sfn = StoredFileNode.load(target['bad'])
logger.info('Repointing Guid {} referent to StoredFileNode {}'.format(target['guid'], target['good']))
guid.referent = good_sfn
guid.save()
logger.info('Removing StoredFileNode {}'.format(target['bad']))
StoredFileNode.remove_one(bad_sfn)
def restore_file(guid):
guid_obj = Guid.load(guid)
trashed_file_node = guid_obj.referent
assert isinstance(trashed_file_node, TrashedFileNode), 'Guid does not point to a trashedfilenode'
logger.info('Loaded trashedfilenode {}'.format(trashed_file_node._id))
try:
logger.info('Calling restore()')
trashed_file_node.restore()
except KeyExistsException: # File with same name exists; user most likely re-uploaded file
logger.warn('File with name {} exists. Renaming...'.format(trashed_file_node.name))
rename_file(trashed_file_node)
logger.info('Calling restore()')
trashed_file_node.restore()
return True
def migrate_legacy_obj(legacy_guid_file):
"""Create `OsfStorageGuidFile` object corresponding to provided `OsfGuidFile`
object, then set the `referent` of the `Guid` object to the newly created
record.
"""
logger.info('Migrating legacy Guid {0}'.format(legacy_guid_file._id))
storage_obj = get_or_create_storage_file(
legacy_guid_file.node,
legacy_guid_file.name,
_id=legacy_guid_file._id,
)
guid_obj = Guid.load(legacy_guid_file._id)
guid_obj.referent = storage_obj
guid_obj.save()
return storage_obj
def migrate_legacy_obj(legacy_guid_file):
"""Create `OsfStorageGuidFile` object corresponding to provided `OsfGuidFile`
object, then set the `referent` of the `Guid` object to the newly created
record.
"""
logger.info('Migrating legacy Guid {0}'.format(legacy_guid_file._id))
storage_obj = get_or_create_storage_file(
legacy_guid_file.node,
legacy_guid_file.name,
_id=legacy_guid_file._id,
)
guid_obj = Guid.load(legacy_guid_file._id)
guid_obj.referent = storage_obj
guid_obj.save()
return storage_obj
def test_restore_file(self):
root = models.StoredFileNode(
path='root',
name='rootfolder',
is_file=False,
node=self.node,
provider='test',
materialized_path='/long/path/to',
).wrapped()
root.save()
fn = models.StoredFileNode(
parent=root._id,
path='afile',
name='name',
is_file=True,
node=self.node,
provider='test',
materialized_path='/long/path/to/name',
).wrapped()
guid = Guid.generate(fn)
before = fn.to_storage()
trashed = fn.delete(user=self.user)
restored = trashed.restore()
assert_equal(
restored.to_storage(),
before
)
assert_equal(models.TrashedFileNode.load(trashed._id), None)
# Guid is repointed
guid.reload()
assert_equal(guid.referent, restored)
def resolve_guid(guid, suffix=None):
"""Load GUID by primary key, look up the corresponding view function in the
routing table, and return the return value of the view function without
changing the URL.
:param str guid: GUID primary key
:param str suffix: Remainder of URL after the GUID
:return: Return value of proxied view function
"""
try:
# Look up
guid_object = Guid.load(guid)
except KeyError as e:
if e.message == 'osfstorageguidfile': # Used when an old detached OsfStorageGuidFile object is accessed
raise HTTPError(http.NOT_FOUND)
else:
raise e
if guid_object:
# verify that the object implements a GuidStoredObject-like interface. If a model
# was once GuidStoredObject-like but that relationship has changed, it's
# possible to have referents that are instances of classes that don't
# have a deep_url attribute or otherwise don't behave as
# expected.
if not hasattr(guid_object.referent, 'deep_url'):
sentry.log_message(
'Guid `{}` resolved to an object with no deep_url'.format(
guid))
raise HTTPError(http.NOT_FOUND)
referent = guid_object.referent
if referent is None:
logger.error('Referent of GUID {0} not found'.format(guid))
raise HTTPError(http.NOT_FOUND)
if not referent.deep_url:
raise HTTPError(http.NOT_FOUND)
# Handle file `/download` shortcut with supported types.
if suffix and suffix.rstrip('/').lower() == 'download':
file_referent = None
if isinstance(referent, PreprintService) and referent.primary_file:
if not referent.is_published:
# TODO: Ideally, permissions wouldn't be checked here.
# This is necessary to prevent a logical inconsistency with
# the routing scheme - if a preprint is not published, only
# admins should be able to know it exists.
auth = Auth.from_kwargs(request.args.to_dict(), {})
if not referent.node.has_permission(
auth.user, permissions.ADMIN):
raise HTTPError(http.NOT_FOUND)
file_referent = referent.primary_file
elif isinstance(referent, BaseFileNode) and referent.is_file:
file_referent = referent
if file_referent:
# Extend `request.args` adding `action=download`.
request.args = request.args.copy()
request.args.update({'action': 'download'})
# Do not include the `download` suffix in the url rebuild.
url = _build_guid_url(urllib.unquote(file_referent.deep_url))
return proxy_url(url)
# Handle Ember Applications
if isinstance(referent, PreprintService):
if referent.provider.domain_redirect_enabled:
# This route should always be intercepted by nginx for the branded domain,
# w/ the exception of `<guid>/download` handled above.
return redirect(referent.absolute_url, http.MOVED_PERMANENTLY)
return send_from_directory(preprints_dir, 'index.html')
url = _build_guid_url(urllib.unquote(referent.deep_url), suffix)
return proxy_url(url)
# GUID not found; try lower-cased and redirect if exists
guid_object_lower = Guid.load(guid.lower())
if guid_object_lower:
return redirect(_build_guid_url(guid.lower(), suffix))
# GUID not found
raise HTTPError(http.NOT_FOUND)
