def test_failed_clone(dspath):
ds = create(dspath)
res = ds.clone("http://nonexistingreallyanything.datalad.org/bla", "sub",
on_failure='ignore')
assert_status('error', res)
assert_message('Failed to clone from all attempted sources: %s',
res)
def test_notclone_known_subdataset(src, path):
# get the superdataset:
ds = clone(src, path, result_xfm='datasets', return_type='item-or-list')
# subdataset not installed:
subds = Dataset(opj(path, 'subm 1'))
assert_false(subds.is_installed())
assert_in('subm 1', ds.subdatasets(fulfilled=False, result_xfm='relpaths'))
assert_not_in('subm 1',
ds.subdatasets(fulfilled=True, result_xfm='relpaths'))
# clone is not meaningful
res = ds.clone('subm 1', on_failure='ignore')
assert_status('error', res)
assert_message('Failed to clone data from any candidate source URL: %s',
res)
# get does the job
res = ds.get(path='subm 1', get_data=False)
assert_status('ok', res)
ok_(subds.is_installed())
ok_(AnnexRepo.is_valid_repo(subds.path, allow_noninitialized=False))
# Verify that it is the correct submodule installed and not
# new repository initiated
eq_(set(subds.repo.get_indexed_files()),
{'test.dat', 'INFO.txt', 'test-annex.dat'})
assert_not_in('subm 1',
ds.subdatasets(fulfilled=False, result_xfm='relpaths'))
assert_in('subm 1', ds.subdatasets(fulfilled=True, result_xfm='relpaths'))
def test_notclone_known_subdataset(src, path):
# get the superdataset:
ds = clone(src, path,
result_xfm='datasets', return_type='item-or-list')
# subdataset not installed:
subds = Dataset(opj(path, 'subm 1'))
assert_false(subds.is_installed())
assert_in('subm 1', ds.subdatasets(fulfilled=False, result_xfm='relpaths'))
assert_not_in('subm 1', ds.subdatasets(fulfilled=True, result_xfm='relpaths'))
# clone is not meaningful
res = ds.clone('subm 1', on_failure='ignore')
assert_status('error', res)
assert_message('Failed to clone from any candidate source URL. '
'Encountered errors per each url were: %s',
res)
# get does the job
res = ds.get(path='subm 1', get_data=False)
assert_status('ok', res)
ok_(subds.is_installed())
ok_(AnnexRepo.is_valid_repo(subds.path, allow_noninitialized=False))
# Verify that it is the correct submodule installed and not
# new repository initiated
eq_(set(subds.repo.get_indexed_files()),
{'test.dat', 'INFO.txt', 'test-annex.dat'})
assert_not_in('subm 1', ds.subdatasets(fulfilled=False, result_xfm='relpaths'))
assert_in('subm 1', ds.subdatasets(fulfilled=True, result_xfm='relpaths'))
def test_failed_clone(dspath):
ds = create(dspath)
res = ds.clone("http://nonexistingreallyanything.somewhere/bla",
"sub",
on_failure='ignore')
assert_status('error', res)
assert_message('Failed to clone data from any candidate source URL: %s',
res)
def test_failed_clone(dspath):
ds = create(dspath)
res = ds.clone("http://nonexistingreallyanything.datalad.org/bla", "sub",
on_failure='ignore')
assert_status('error', res)
assert_message('Failed to clone from any candidate source URL. '
'Encountered errors per each url were: %s',
res)
def test_get_invalid_call(path, file_outside):
# no argument at all:
assert_raises(InsufficientArgumentsError, get, None)
assert_raises(InsufficientArgumentsError, get, [])
# invalid dataset:
assert_raises(ValueError, get, None, dataset=path, on_failure='ignore')
# have a plain git:
ds = Dataset(path)
ds.create(annex=False)
with open(opj(path, "some.txt"), "w") as f:
f.write("whatever")
ds.save("some.txt", to_git=True, message="Initial commit.")
# make it an annex (remove indicator file that create has placed
# in the dataset to make it possible):
(ds.pathobj / '.noannex').unlink()
AnnexRepo(path, init=True, create=True)
# call get again on a file in git:
result = ds.get("some.txt")
assert_status('notneeded', result)
# invalid source:
# yoh: but now we would need to add it to annex since clever code first
# checks what needs to be fetched at all
create_tree(path, {'annexed.dat': 'some'})
ds.save("annexed.dat")
ds.repo.drop("annexed.dat", options=['--force'])
with assert_raises(RemoteNotAvailableError) as ce:
ds.get("annexed.dat", source='MysteriousRemote')
eq_("MysteriousRemote", ce.exception.remote)
res = ds.get("NotExistingFile.txt", on_failure='ignore')
assert_status('impossible', res)
assert_message("path does not exist", res)
# path outside repo errors as with most other commands:
res = ds.get(file_outside, on_failure='ignore', result_renderer='default')
assert_in_results(res,
status='error',
message=('path not associated with dataset %s', ds))
def test_clone_datasets_root(tdir):
with chpwd(tdir):
ds = clone("///", result_xfm='datasets', return_type='item-or-list')
ok_(ds.is_installed())
eq_(ds.path, opj(tdir, get_datasets_topdir()))
# do it a second time:
res = clone("///", on_failure='ignore')
assert_message("dataset %s was already cloned from '%s'", res)
assert_status('notneeded', res)
# and a third time into an existing something, that is not a dataset:
with open(opj(tdir, 'sub', 'a_file.txt'), 'w') as f:
f.write("something")
res = clone('///', path="sub", on_failure='ignore')
assert_message(
'target path already exists and not empty, refuse to clone into target path',
res)
assert_status('error', res)
def test_get_invalid_call(path, file_outside):
# no argument at all:
assert_raises(InsufficientArgumentsError, get, None)
assert_raises(InsufficientArgumentsError, get, [])
# invalid dataset:
assert_status('impossible', get(None, dataset=path, on_failure='ignore'))
# have a plain git:
ds = Dataset(path)
ds.create(no_annex=True)
with open(opj(path, "some.txt"), "w") as f:
f.write("whatever")
ds.save("some.txt", to_git=True, message="Initial commit.")
# make it an annex (remove indicator file that create has placed
# in the dataset to make it possible):
(ds.pathobj / '.noannex').unlink()
AnnexRepo(path, init=True, create=True)
# call get again on a file in git:
result = ds.get("some.txt")
assert_status('notneeded', result)
# invalid source:
# yoh: but now we would need to add it to annex since clever code first
# checks what needs to be fetched at all
create_tree(path, {'annexed.dat': 'some'})
ds.save("annexed.dat")
ds.repo.drop("annexed.dat", options=['--force'])
with assert_raises(RemoteNotAvailableError) as ce:
ds.get("annexed.dat", source='MysteriousRemote')
eq_("MysteriousRemote", ce.exception.remote)
res = ds.get("NotExistingFile.txt", on_failure='ignore')
assert_status('impossible', res)
assert_message("path does not exist", res)
# path outside repo errors as with most other commands:
res = ds.get(file_outside, on_failure='ignore')
assert_in_results(
res, status='impossible', message='path not associated with any dataset')
def test_clone_simple_local(src, path):
origin = Dataset(path)
# now install it somewhere else
ds = clone(src,
path,
description='mydummy',
result_xfm='datasets',
return_type='item-or-list')
eq_(ds.path, path)
ok_(ds.is_installed())
if not isinstance(origin.repo, AnnexRepo):
# this means it is a GitRepo
ok_(isinstance(origin.repo, GitRepo))
# stays plain Git repo
ok_(isinstance(ds.repo, GitRepo))
ok_(not isinstance(ds.repo, AnnexRepo))
ok_(GitRepo.is_valid_repo(ds.path))
eq_(set(ds.repo.get_indexed_files()), {'test.dat', 'INFO.txt'})
ok_clean_git(path, annex=False)
else:
# must be an annex
ok_(isinstance(ds.repo, AnnexRepo))
ok_(AnnexRepo.is_valid_repo(ds.path, allow_noninitialized=False))
eq_(set(ds.repo.get_indexed_files()),
{'test.dat', 'INFO.txt', 'test-annex.dat'})
ok_clean_git(path, annex=True)
# no content was installed:
ok_(not ds.repo.file_has_content('test-annex.dat'))
uuid_before = ds.repo.uuid
eq_(ds.repo.get_description(), 'mydummy')
# installing it again, shouldn't matter:
res = clone(src, path)
assert_result_values_equal(res, 'source_url', [src])
assert_status('notneeded', res)
assert_message("dataset %s was already cloned from '%s'", res)
ok_(ds.is_installed())
if isinstance(origin.repo, AnnexRepo):
eq_(uuid_before, ds.repo.uuid)
def test_clone_datasets_root(tdir):
tdir = Path(tdir)
with chpwd(tdir):
ds = clone("///")
ok_(ds.is_installed())
eq_(ds.pathobj, tdir / get_datasets_topdir())
# do it a second time:
res = clone("///", on_failure='ignore', result_xfm=None, return_type='list')
assert_message(
"dataset %s was already cloned from '%s'",
res)
assert_status('notneeded', res)
# and a third time into an existing something, that is not a dataset:
(tdir / 'sub' / 'a_file.txt').write_text("something")
res = clone('///', path="sub", on_failure='ignore')
assert_message(
'target path already exists and not empty, refuse to clone into target path',
res)
assert_status('error', res)
def test_clone_datasets_root(tdir):
with chpwd(tdir):
ds = clone("///", result_xfm='datasets', return_type='item-or-list')
ok_(ds.is_installed())
eq_(ds.path, opj(tdir, get_datasets_topdir()))
# do it a second time:
res = clone("///", on_failure='ignore')
assert_message(
"dataset %s was already cloned from '%s'",
res)
assert_status('notneeded', res)
# and a third time into an existing something, that is not a dataset:
with open(opj(tdir, 'sub', 'a_file.txt'), 'w') as f:
f.write("something")
res = clone('///', path="sub", on_failure='ignore')
assert_message(
'target path already exists and not empty, refuse to clone into target path',
res)
assert_status('error', res)
def test_clone_simple_local(src, path):
origin = Dataset(path)
# now install it somewhere else
ds = clone(src, path, description='mydummy',
result_xfm='datasets', return_type='item-or-list')
eq_(ds.path, path)
ok_(ds.is_installed())
if not isinstance(origin.repo, AnnexRepo):
# this means it is a GitRepo
ok_(isinstance(origin.repo, GitRepo))
# stays plain Git repo
ok_(isinstance(ds.repo, GitRepo))
ok_(not isinstance(ds.repo, AnnexRepo))
ok_(GitRepo.is_valid_repo(ds.path))
eq_(set(ds.repo.get_indexed_files()),
{'test.dat', 'INFO.txt'})
ok_clean_git(path, annex=False)
else:
# must be an annex
ok_(isinstance(ds.repo, AnnexRepo))
ok_(AnnexRepo.is_valid_repo(ds.path, allow_noninitialized=False))
eq_(set(ds.repo.get_indexed_files()),
{'test.dat', 'INFO.txt', 'test-annex.dat'})
ok_clean_git(path, annex=True)
# no content was installed:
ok_(not ds.repo.file_has_content('test-annex.dat'))
uuid_before = ds.repo.uuid
eq_(ds.repo.get_description(), 'mydummy')
# installing it again, shouldn't matter:
res = clone(src, path)
assert_result_values_equal(res, 'source_url', [src])
assert_status('notneeded', res)
assert_message("dataset %s was already cloned from '%s'", res)
ok_(ds.is_installed())
if isinstance(origin.repo, AnnexRepo):
eq_(uuid_before, ds.repo.uuid)
