def _test_initremote_alias(host, ds_path, store):
ds_path = Path(ds_path)
store = Path(store)
ds = Dataset(ds_path).create()
populate_dataset(ds)
ds.save()
if host:
url = "ria+ssh://{host}{path}".format(host=host,
path=store)
else:
url = "ria+{}".format(store.as_uri())
init_opts = common_init_opts + ['url={}'.format(url)]
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
create_store(io, store, '1')
# set up the dataset with alias
create_ds_in_store(io, store, ds.id, '2', '1', 'ali')
ds.repo.init_remote('ria-remote', options=init_opts)
assert_in('ria-remote',
[cfg['name']
for uuid, cfg in ds.repo.get_special_remotes().items()]
)
assert_repo_status(ds.path)
assert_true(io.exists(store / "alias" / "ali"))
def test_push_url(storepath, dspath, blockfile):
dspath = Path(dspath)
store = Path(storepath)
blockfile = Path(blockfile)
blockfile.touch()
ds = Dataset(dspath).create()
populate_dataset(ds)
ds.save()
assert_repo_status(ds.path)
# set up store:
io = LocalIO()
store_url = "ria+{}".format(store.as_uri())
create_store(io, store, '1')
create_ds_in_store(io, store, ds.id, '2', '1')
# initremote fails with invalid url (not a ria+ URL):
invalid_url = (store.parent / "non-existent").as_uri()
init_opts = common_init_opts + [
'url={}'.format(store_url), 'push-url={}'.format(invalid_url)
]
assert_raises(CommandError,
ds.repo.init_remote,
'store',
options=init_opts)
# initremote succeeds with valid but inaccessible URL (pointing to a file
# instead of a store):
block_url = "ria+" + blockfile.as_uri()
init_opts = common_init_opts + [
'url={}'.format(store_url), 'push-url={}'.format(block_url)
]
ds.repo.init_remote('store', options=init_opts)
# but a push will fail:
assert_raises(CommandError, ds.repo.call_annex,
['copy', 'one.txt', '--to', 'store'])
# reconfigure with correct push-url:
init_opts = common_init_opts + [
'url={}'.format(store_url), 'push-url={}'.format(store_url)
]
ds.repo.enable_remote('store', options=init_opts)
# push works now:
ds.repo.call_annex(['copy', 'one.txt', '--to', 'store'])
store_uuid = ds.siblings(name='store',
return_type='item-or-list')['annex-uuid']
here_uuid = ds.siblings(name='here',
return_type='item-or-list')['annex-uuid']
known_sources = ds.repo.whereis('one.txt')
assert_in(here_uuid, known_sources)
assert_in(store_uuid, known_sources)
def test_read_access(store_path, store_url, ds_path):
ds = Dataset(ds_path).create()
populate_dataset(ds)
ds.save()
if ds.repo.is_managed_branch():
# TODO: on crippled FS copytree to populate store doesn't seem to work.
# Or may be it's just the serving via HTTP that doesn't work.
# Either way, after copytree and fsck, whereis doesn't report
# the store as an available source.
raise SkipTest("Skip on crippled FS")
files = [Path('one.txt'), Path('subdir') / 'two']
store_path = Path(store_path)
url = "ria+" + store_url
init_opts = common_init_opts + ['url={}'.format(url)]
io = LocalIO()
create_store(io, store_path, '1')
create_ds_in_store(io, store_path, ds.id, '2', '1')
ds.repo.init_remote('ora-remote', options=init_opts)
ds.repo.fsck(remote='ora-remote', fast=True)
store_uuid = ds.siblings(name='ora-remote',
return_type='item-or-list')['annex-uuid']
here_uuid = ds.siblings(name='here',
return_type='item-or-list')['annex-uuid']
# nothing in store yet:
for f in files:
known_sources = ds.repo.whereis(str(f))
assert_in(here_uuid, known_sources)
assert_not_in(store_uuid, known_sources)
annex_obj_target = str(store_path / ds.id[:3] / ds.id[3:] / 'annex' /
'objects')
shutil.rmtree(annex_obj_target)
shutil.copytree(src=str(ds.repo.dot_git / 'annex' / 'objects'),
dst=annex_obj_target)
ds.repo.fsck(remote='ora-remote', fast=True)
# all in store now:
for f in files:
known_sources = ds.repo.whereis(str(f))
assert_in(here_uuid, known_sources)
assert_in(store_uuid, known_sources)
ds.drop('.')
res = ds.get('.')
assert_equal(len(res), 2)
assert_result_count(res,
2,
status='ok',
type='file',
action='get',
message="from ora-remote...")
def test_read_access(store_path, store_url, ds_path):
ds = Dataset(ds_path).create()
populate_dataset(ds)
ds.save()
files = [Path('one.txt'), Path('subdir') / 'two']
store_path = Path(store_path)
url = "ria+" + store_url
init_opts = common_init_opts + ['url={}'.format(url)]
io = LocalIO()
create_store(io, store_path, '1')
create_ds_in_store(io, store_path, ds.id, '2', '1')
ds.repo.init_remote('ora-remote', options=init_opts)
ds.repo.fsck(remote='ora-remote', fast=True)
store_uuid = ds.siblings(name='ora-remote',
return_type='item-or-list')['annex-uuid']
here_uuid = ds.siblings(name='here',
return_type='item-or-list')['annex-uuid']
# nothing in store yet:
for f in files:
known_sources = ds.repo.whereis(str(f))
assert_in(here_uuid, known_sources)
assert_not_in(store_uuid, known_sources)
annex_obj_target = str(store_path / ds.id[:3] / ds.id[3:] / 'annex' /
'objects')
shutil.rmtree(annex_obj_target)
shutil.copytree(src=str(ds.repo.dot_git / 'annex' / 'objects'),
dst=annex_obj_target)
ds.repo.fsck(remote='ora-remote', fast=True)
# all in store now:
for f in files:
known_sources = ds.repo.whereis(str(f))
assert_in(here_uuid, known_sources)
assert_in(store_uuid, known_sources)
ds.drop('.')
res = ds.get('.')
assert_equal(len(res), 2)
assert_result_count(res,
2,
status='ok',
type='file',
action='get',
message="from ora-remote...")
def _test_gitannex(host, store, dspath):
store = Path(store)
dspath = Path(dspath)
store = Path(store)
ds = Dataset(dspath).create()
if ds.repo.is_managed_branch():
# git-annex-testremote is way too slow on crippled FS.
# Use is_managed_branch() as a proxy and skip only here
# instead of in a decorator
raise SkipTest("Test too slow on crippled FS")
populate_dataset(ds)
ds.save()
assert_repo_status(ds.path)
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
if host:
store_url = "ria+ssh://{host}{path}".format(host=host, path=store)
else:
store_url = "ria+{}".format(store.as_uri())
create_store(io, store, '1')
# TODO: Re-establish test for version 1
# version 2: dirhash
create_ds_in_store(io, store, ds.id, '2', '1')
# add special remote
init_opts = common_init_opts + ['url={}'.format(store_url)]
ds.repo.init_remote('store', options=init_opts)
from datalad.support.external_versions import external_versions
if '8.20200330' < external_versions['cmd:annex'] < '8.20200624':
# https://git-annex.branchable.com/bugs/testremote_breeds_way_too_many_instances_of_the_externals_remote/?updated
raise SkipTest(
"git-annex might lead to overwhelming number of external "
"special remote instances")
# run git-annex-testremote
# note, that we don't want to capture output. If something goes wrong we
# want to see it in test build's output log.
GitWitlessRunner(cwd=dspath).run(['git', 'annex', 'testremote', 'store'])
def _test_initremote_rewrite(host, ds_path, store):
# rudimentary repetition of test_initremote_basic, but
# with url.<base>.insteadOf config, which should not only
# be respected, but lead to the rewritten URL stored in
# git-annex:remote.log
ds_path = Path(ds_path)
store = Path(store)
ds = Dataset(ds_path).create()
populate_dataset(ds)
ds.save()
assert_repo_status(ds.path)
url = "mystore:"
init_opts = common_init_opts + ['url={}'.format(url)]
if host:
replacement = "ria+ssh://{host}{path}".format(host=host,
path=store)
else:
replacement = "ria+{}".format(store.as_uri())
ds.config.set("url.{}.insteadOf".format(replacement), url, where='local')
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
create_store(io, store, '1')
create_ds_in_store(io, store, ds.id, '2', '1')
# run initremote and check what's stored:
ds.repo.init_remote('ria-remote', options=init_opts)
assert_in('ria-remote',
[cfg['name']
for uuid, cfg in ds.repo.get_special_remotes().items()]
)
# git-annex:remote.log should have:
# - rewritten url
# - common_init_opts
# - archive_id (which equals ds id)
remote_log = ds.repo.call_git(['cat-file', 'blob', 'git-annex:remote.log'],
read_only=True)
assert_in("url={}".format(replacement), remote_log)
[assert_in(c, remote_log) for c in common_init_opts]
assert_in("archive-id={}".format(ds.id), remote_log)
def _test_initremote_basic(host, ds_path, store, link):
ds_path = Path(ds_path)
store = Path(store)
link = Path(link)
ds = Dataset(ds_path).create()
populate_dataset(ds)
ds.save()
if host:
url = "ria+ssh://{host}{path}".format(host=host,
path=store)
else:
url = "ria+{}".format(store.as_uri())
init_opts = common_init_opts + ['url={}'.format(url)]
# fails on non-existing storage location
assert_raises(CommandError,
ds.repo.init_remote, 'ria-remote', options=init_opts)
# Doesn't actually create a remote if it fails
assert_not_in('ria-remote',
[cfg['name']
for uuid, cfg in ds.repo.get_special_remotes().items()]
)
# fails on non-RIA URL
assert_raises(CommandError, ds.repo.init_remote, 'ria-remote',
options=common_init_opts + ['url={}'.format(store.as_uri())]
)
# Doesn't actually create a remote if it fails
assert_not_in('ria-remote',
[cfg['name']
for uuid, cfg in ds.repo.get_special_remotes().items()]
)
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
create_store(io, store, '1')
# still fails, since ds isn't setup in the store
assert_raises(CommandError,
ds.repo.init_remote, 'ria-remote', options=init_opts)
# Doesn't actually create a remote if it fails
assert_not_in('ria-remote',
[cfg['name']
for uuid, cfg in ds.repo.get_special_remotes().items()]
)
# set up the dataset as well
create_ds_in_store(io, store, ds.id, '2', '1')
# now should work
ds.repo.init_remote('ria-remote', options=init_opts)
assert_in('ria-remote',
[cfg['name']
for uuid, cfg in ds.repo.get_special_remotes().items()]
)
assert_repo_status(ds.path)
# git-annex:remote.log should have:
# - url
# - common_init_opts
# - archive_id (which equals ds id)
remote_log = ds.repo.call_git(['cat-file', 'blob', 'git-annex:remote.log'],
read_only=True)
assert_in("url={}".format(url), remote_log)
[assert_in(c, remote_log) for c in common_init_opts]
assert_in("archive-id={}".format(ds.id), remote_log)
# re-configure with invalid URL should fail:
assert_raises(
CommandError,
ds.repo.call_annex,
['enableremote', 'ria-remote'] + common_init_opts + [
'url=ria+file:///non-existing'])
# but re-configure with valid URL should work
if has_symlink_capability():
link.symlink_to(store)
new_url = 'ria+{}'.format(link.as_uri())
ds.repo.call_annex(
['enableremote', 'ria-remote'] + common_init_opts + [
'url={}'.format(new_url)])
# git-annex:remote.log should have:
# - url
# - common_init_opts
# - archive_id (which equals ds id)
remote_log = ds.repo.call_git(['cat-file', 'blob',
'git-annex:remote.log'],
read_only=True)
assert_in("url={}".format(new_url), remote_log)
[assert_in(c, remote_log) for c in common_init_opts]
assert_in("archive-id={}".format(ds.id), remote_log)
# we can deal with --sameas, which leads to a special remote not having a
# 'name' property, but only a 'sameas-name'. See gh-4259
try:
ds.repo.init_remote('ora2',
options=init_opts + ['--sameas', 'ria-remote'])
except CommandError as e:
if 'Invalid option `--sameas' in e.stderr:
# annex too old - doesn't know --sameas
pass
else:
raise
def _test_version_check(host, dspath, store):
dspath = Path(dspath)
store = Path(store)
ds = Dataset(dspath).create()
populate_dataset(ds)
ds.save()
assert_repo_status(ds.path)
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
if host:
store_url = "ria+ssh://{host}{path}".format(host=host,
path=store)
else:
store_url = "ria+{}".format(store.as_uri())
create_store(io, store, '1')
# TODO: Re-establish test for version 1
# version 2: dirhash
create_ds_in_store(io, store, ds.id, '2', '1')
# add special remote
init_opts = common_init_opts + ['url={}'.format(store_url)]
ds.repo.init_remote('store', options=init_opts)
ds.repo.copy_to('.', 'store')
# check version files
remote_ds_tree_version_file = store / 'ria-layout-version'
dsgit_dir, archive_dir, dsobj_dir = \
get_layout_locations(1, store, ds.id)
remote_obj_tree_version_file = dsgit_dir / 'ria-layout-version'
assert_true(remote_ds_tree_version_file.exists())
assert_true(remote_obj_tree_version_file.exists())
with open(str(remote_ds_tree_version_file), 'r') as f:
assert_equal(f.read().strip(), '1')
with open(str(remote_obj_tree_version_file), 'r') as f:
assert_equal(f.read().strip(), '2')
# Accessing the remote should not yield any output regarding versioning,
# since it's the "correct" version. Note that "fsck" is an arbitrary choice.
# We need just something to talk to the special remote.
with swallow_logs(new_level=logging.INFO) as cml:
ds.repo.fsck(remote='store', fast=True)
# TODO: For some reason didn't get cml.assert_logged to assert
# "nothing was logged"
assert not cml.out
# Now fake-change the version
with open(str(remote_obj_tree_version_file), 'w') as f:
f.write('X\n')
# Now we should see a message about it
with swallow_logs(new_level=logging.INFO) as cml:
ds.repo.fsck(remote='store', fast=True)
cml.assert_logged(level="INFO",
msg="Remote object tree reports version X",
regex=False)
# reading still works:
ds.drop('.')
assert_status('ok', ds.get('.'))
# but writing doesn't:
with open(str(Path(ds.path) / 'new_file'), 'w') as f:
f.write("arbitrary addition")
ds.save(message="Add a new_file")
# TODO: use self.annex.error in special remote and see whether we get an
# actual error result
assert_raises(CommandError,
ds.repo.copy_to, 'new_file', 'store')
# However, we can force it by configuration
ds.config.add("annex.ora-remote.store.force-write", "true", where='local')
ds.repo.copy_to('new_file', 'store')
def _test_remote_layout(host, dspath, store, archiv_store):
dspath = Path(dspath)
store = Path(store)
archiv_store = Path(archiv_store)
ds = Dataset(dspath).create()
populate_dataset(ds)
ds.save()
assert_repo_status(ds.path)
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
if host:
store_url = "ria+ssh://{host}{path}".format(host=host,
path=store)
arch_url = "ria+ssh://{host}{path}".format(host=host,
path=archiv_store)
else:
store_url = "ria+{}".format(store.as_uri())
arch_url = "ria+{}".format(archiv_store.as_uri())
create_store(io, store, '1')
# TODO: Re-establish test for version 1
# version 2: dirhash
create_ds_in_store(io, store, ds.id, '2', '1')
# add special remote
init_opts = common_init_opts + ['url={}'.format(store_url)]
ds.repo.init_remote('store', options=init_opts)
# copy files into the RIA store
ds.repo.copy_to('.', 'store')
# we should see the exact same annex object tree
dsgit_dir, archive_dir, dsobj_dir = \
get_layout_locations(1, store, ds.id)
store_objects = get_all_files(dsobj_dir)
local_objects = get_all_files(ds.pathobj / '.git' / 'annex' / 'objects')
assert_equal(len(store_objects), 2)
if not ds.repo.is_managed_branch():
# with managed branches the local repo uses hashdirlower instead
# TODO: However, with dataset layout version 1 this should therefore
# work on adjusted branch the same way
# TODO: Wonder whether export-archive-ora should account for that and
# rehash according to target layout.
assert_equal(sorted([p for p in store_objects]),
sorted([p for p in local_objects])
)
if not io.get_7z():
raise SkipTest("No 7z available in RIA store")
# we can simply pack up the content of the remote into a
# 7z archive and place it in the right location to get a functional
# archive remote
create_store(io, archiv_store, '1')
create_ds_in_store(io, archiv_store, ds.id, '2', '1')
whereis = ds.repo.whereis('one.txt')
dsgit_dir, archive_dir, dsobj_dir = \
get_layout_locations(1, archiv_store, ds.id)
ds.export_archive_ora(archive_dir / 'archive.7z')
init_opts = common_init_opts + ['url={}'.format(arch_url)]
ds.repo.init_remote('archive', options=init_opts)
# now fsck the new remote to get the new special remote indexed
ds.repo.fsck(remote='archive', fast=True)
assert_equal(len(ds.repo.whereis('one.txt')), len(whereis) + 1)
def test_read_access(store_path=None, store_url=None, ds_path=None):
ds = Dataset(ds_path).create()
populate_dataset(ds)
files = [Path('one.txt'), Path('subdir') / 'two']
store_path = Path(store_path)
url = "ria+" + store_url
init_opts = common_init_opts + ['url={}'.format(url)]
io = LocalIO()
create_store(io, store_path, '1')
create_ds_in_store(io, store_path, ds.id, '2', '1')
ds.repo.init_remote('ora-remote', options=init_opts)
fsck_results = ds.repo.fsck(remote='ora-remote', fast=True)
# Note: Failures in the special remote will show up as a success=False
# result for fsck -> the call itself would not fail.
for r in fsck_results:
if "note" in r:
# we could simply assert "note" to not be in r, but we want proper
# error reporting - content of note, not just its unexpected
# existence.
assert_equal(r["success"],
"true",
msg="git-annex-fsck failed with ORA over HTTP: %s" %
r)
assert_equal(r["error-messages"], [])
store_uuid = ds.siblings(name='ora-remote',
return_type='item-or-list',
result_renderer='disabled')['annex-uuid']
here_uuid = ds.siblings(name='here',
return_type='item-or-list',
result_renderer='disabled')['annex-uuid']
# nothing in store yet:
for f in files:
known_sources = ds.repo.whereis(str(f))
assert_in(here_uuid, known_sources)
assert_not_in(store_uuid, known_sources)
annex_obj_target = str(store_path / ds.id[:3] / ds.id[3:] / 'annex' /
'objects')
shutil.rmtree(annex_obj_target)
shutil.copytree(src=str(ds.repo.dot_git / 'annex' / 'objects'),
dst=annex_obj_target)
ds.repo.fsck(remote='ora-remote', fast=True)
# all in store now:
for f in files:
known_sources = ds.repo.whereis(str(f))
assert_in(here_uuid, known_sources)
assert_in(store_uuid, known_sources)
ds.drop('.')
res = ds.get('.')
assert_equal(len(res), 4)
assert_result_count(res,
4,
status='ok',
type='file',
action='get',
message="from ora-remote...")
# try whether the reported access URL is correct
one_url = ds.repo.whereis('one.txt',
output='full')[store_uuid]['urls'].pop()
assert_status(
'ok', ds.download_url(urls=[one_url], path=str(ds.pathobj / 'dummy')))
def _test_permission(host, storepath, dspath):
# Test whether ORA correctly revokes and obtains write permissions within
# the annex object tree. That is: Revoke after ORA pushed a key to store
# in order to allow the object tree to safely be used with an ephemeral
# clone. And on removal obtain write permissions, like annex would
# internally on a drop (but be sure to restore if something went wrong).
dspath = Path(dspath)
storepath = Path(storepath)
ds = Dataset(dspath).create()
populate_dataset(ds)
ds.save()
assert_repo_status(ds.path)
testfile = 'one.txt'
# set up store:
io = SSHRemoteIO(host) if host else LocalIO()
if host:
store_url = "ria+ssh://{host}{path}".format(host=host, path=storepath)
else:
store_url = "ria+{}".format(storepath.as_uri())
create_store(io, storepath, '1')
create_ds_in_store(io, storepath, ds.id, '2', '1')
_, _, obj_tree = get_layout_locations(1, storepath, ds.id)
assert_true(obj_tree.is_dir())
file_key_in_store = obj_tree / 'X9' / '6J' / 'MD5E-s8--7e55db001d319a94b0b713529a756623.txt' / 'MD5E-s8--7e55db001d319a94b0b713529a756623.txt'
init_opts = common_init_opts + ['url={}'.format(store_url)]
ds.repo.init_remote('store', options=init_opts)
store_uuid = ds.siblings(name='store',
return_type='item-or-list')['annex-uuid']
here_uuid = ds.siblings(name='here',
return_type='item-or-list')['annex-uuid']
known_sources = ds.repo.whereis(testfile)
assert_in(here_uuid, known_sources)
assert_not_in(store_uuid, known_sources)
assert_false(file_key_in_store.exists())
ds.repo.call_annex(['copy', testfile, '--to', 'store'])
known_sources = ds.repo.whereis(testfile)
assert_in(here_uuid, known_sources)
assert_in(store_uuid, known_sources)
assert_true(file_key_in_store.exists())
# Revoke write permissions from parent dir in-store to test whether we
# still can drop (if we can obtain the permissions). Note, that this has
# no effect on VFAT.
file_key_in_store.parent.chmod(file_key_in_store.parent.stat().st_mode
& ~stat.S_IWUSR)
# we can't directly delete; key in store should be protected
assert_raises(PermissionError, file_key_in_store.unlink)
# ORA can still drop, since it obtains permission to:
ds.repo.call_annex(['drop', testfile, '--from', 'store'])
known_sources = ds.repo.whereis(testfile)
assert_in(here_uuid, known_sources)
assert_not_in(store_uuid, known_sources)
assert_false(file_key_in_store.exists())
