def test_status_basics(path=None, linkpath=None, otherdir=None):
if has_symlink_capability():
# make it more complicated by default
ut.Path(linkpath).symlink_to(path, target_is_directory=True)
path = linkpath
with chpwd(path):
assert_raises(NoDatasetFound, status)
ds = Dataset(path).create()
# outcome identical between ds= and auto-discovery
with chpwd(path):
assert_raises(IncompleteResultsError, status, path=otherdir)
stat = status(result_renderer='disabled')
eq_(stat, ds.status(result_renderer='disabled'))
assert_status('ok', stat)
# we have a bunch of reports (be vague to be robust to future changes
assert len(stat) > 2
# check the composition
for s in stat:
eq_(s['status'], 'ok')
eq_(s['action'], 'status')
eq_(s['state'], 'clean')
eq_(s['type'], 'file')
assert_in('gitshasum', s)
assert_in('bytesize', s)
eq_(s['refds'], ds.path)
def test_gh2927(path=None, linkpath=None):
if has_symlink_capability():
# make it more complicated by default
Path(linkpath).symlink_to(path, target_is_directory=True)
path = linkpath
ds = Dataset(path).create()
ds.create('subds_clean')
assert_status(
'ok',
ds.create(op.join('subds_clean', 'subds_lvl1_clean'),
result_xfm=None,
return_type='list'))
def test_status_symlinked_dir_within_repo(path=None):
if not has_symlink_capability():
raise SkipTest("Can't create symlinks")
# <path>
# |-- bar -> <path>/foo
# `-- foo
# `-- f
ds = Dataset(path).create()
foo = ds.pathobj / "foo"
foo.mkdir()
(foo / "f").write_text("content")
(ds.pathobj / "bar").symlink_to(foo, target_is_directory=True)
ds.save()
bar_f = ds.pathobj / "bar" / "f"
def call():
return ds.status(path=[bar_f],
annex="availability",
on_failure="ignore",
result_renderer='disabled')
if ds.repo.git_annex_version < "8.20200522" or on_windows:
# TODO: on windows even with a recent annex -- no CommandError is
# raised, TODO
assert_result_count(call(), 0)
elif ds.repo.git_annex_version < '10.20220222':
# As of 2a8fdfc7d (Display a warning message when asked to operate on a
# file inside a symlinked directory, 2020-05-11), git-annex will error.
with assert_raises(CommandError):
call()
elif '10.20220222' <= ds.repo.git_annex_version < '10.20220322':
# No error on annex' side since 10.20220222;
# However, we'd now get something like this:
# > git annex find bar/f
# error: pathspec 'bar/f' did not match any file(s) known to git
# Did you forget to 'git add'?
#
# But exists zero until 10.20220322!
assert_result_count(call(), 0)
else:
res = call()
assert_result_count(res,
1,
status='error',
state='unknown',
path=str(bar_f))
from datalad.tests.utils_pytest import (
DEFAULT_REMOTE,
assert_equal,
assert_result_count,
assert_true,
has_symlink_capability,
skip_if,
skip_if_on_windows,
with_tempfile,
with_tree,
)
from datalad.utils import Path
@skip_if_on_windows # currently all tests re RIA/ORA don't run on windows
@skip_if(cond=not has_symlink_capability(),
msg="skip testing ephemeral clone w/o symlink capabilities")
@with_tree({'file1.txt': 'some', 'sub': {'other.txt': 'other'}})
@with_tempfile
@with_tempfile
def test_ephemeral(ds_path=None, store_path=None, clone_path=None):
dspath = Path(ds_path)
store = Path(store_path)
file_test = Path('file1.txt')
file_testsub = Path('sub') / 'other.txt'
# create the original dataset
ds = Dataset(dspath)
ds.create(force=True)
ds.save()
def test_status(_path=None, linkpath=None):
# do the setup on the real path, not the symlink, to have its
# bugs not affect this test of status()
ds = get_deeply_nested_structure(str(_path))
if has_symlink_capability():
# make it more complicated by default
ut.Path(linkpath).symlink_to(_path, target_is_directory=True)
path = linkpath
else:
path = _path
ds = Dataset(path)
if has_symlink_capability():
assert ds.pathobj != ds.repo.pathobj
# spotcheck that annex status reporting and availability evaluation
# works
assert_result_count(
ds.status(annex='all', result_renderer='disabled'),
1,
path=str(ds.pathobj / 'subdir' / 'annexed_file.txt'),
key='MD5E-s5--275876e34cf609db118f3d84b799a790.txt',
has_content=True,
objloc=str(ds.repo.pathobj / '.git' / 'annex' / 'objects' /
# hashdir is different on windows
('f33' if ds.repo.is_managed_branch() else '7p') /
('94b' if ds.repo.is_managed_branch() else 'gp') /
'MD5E-s5--275876e34cf609db118f3d84b799a790.txt' /
'MD5E-s5--275876e34cf609db118f3d84b799a790.txt'))
plain_recursive = ds.status(recursive=True, result_renderer='disabled')
# check integrity of individual reports
for res in plain_recursive:
# anything that is an "intended" symlink should be reported
# as such
if 'link2' in str(res['path']):
assert res['type'] == 'symlink', res
# every item must report its parent dataset
assert_in('parentds', res)
# bunch of smoke tests
# query of '.' is same as no path
eq_(plain_recursive,
ds.status(path='.', recursive=True, result_renderer='disabled'))
# duplicate paths do not change things
eq_(plain_recursive,
ds.status(path=['.', '.'], recursive=True, result_renderer='disabled'))
# neither do nested paths
eq_(
plain_recursive,
ds.status(path=['.', 'subds_modified'],
recursive=True,
result_renderer='disabled'))
# when invoked in a subdir of a dataset it still reports on the full thing
# just like `git status`, as long as there are no paths specified
with chpwd(op.join(path, 'directory_untracked')):
plain_recursive = status(recursive=True, result_renderer='disabled')
# should be able to take absolute paths and yield the same
# output
eq_(plain_recursive,
ds.status(path=ds.path, recursive=True, result_renderer='disabled'))
# query for a deeply nested path from the top, should just work with a
# variety of approaches
rpath = op.join('subds_modified', 'subds_lvl1_modified',
OBSCURE_FILENAME + u'_directory_untracked')
apathobj = ds.pathobj / rpath
apath = str(apathobj)
# ds.repo.pathobj will have the symlink resolved
arealpath = ds.repo.pathobj / rpath
# TODO include explicit relative path in test
for p in (rpath, apath, arealpath, None):
if p is None:
# change into the realpath of the dataset and
# query with an explicit path
with chpwd(ds.repo.path):
res = ds.status(path=op.join('.', rpath),
result_renderer='disabled')
else:
res = ds.status(path=p, result_renderer='disabled')
assert_result_count(
res,
1,
state='untracked',
type='directory',
refds=ds.path,
# path always comes out a full path inside the queried dataset
path=apath,
)
assert_result_count(ds.status(recursive=True, result_renderer='disabled'),
1,
path=apath)
# limiting recursion will exclude this particular path
assert_result_count(ds.status(recursive=True,
recursion_limit=1,
result_renderer='disabled'),
0,
path=apath)
# negative limit is unlimited limit
eq_(
ds.status(recursive=True,
recursion_limit=-1,
result_renderer='disabled'),
ds.status(recursive=True, result_renderer='disabled'))
# check integrity of individual reports with a focus on how symlinks
# are reported in annex-mode
# this is different from plain git-mode, which reports types as-is
# from the git record
for res in ds.status(recursive=True,
annex='basic',
result_renderer='disabled'):
# anything that is an "intended" symlink should be reported
# as such. In contrast, anything that is a symlink for mere
# technical reasons (annex using it for something in some mode)
# should be reported as the thing it is representing (i.e.
# a file)
if 'link2' in str(res['path']):
assert res['type'] == 'symlink', res
else:
assert res['type'] != 'symlink', res
def _test_initremote_basic(url, io, store, ds_path, link):
ds_path = Path(ds_path)
store = Path(store)
link = Path(link)
ds = Dataset(ds_path).create()
populate_dataset(ds)
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:
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_path_diff(_path=None, linkpath=None):
# do the setup on the real path, not the symlink, to have its
# bugs not affect this test of status()
ds = get_deeply_nested_structure(str(_path))
if has_symlink_capability():
# make it more complicated by default
ut.Path(linkpath).symlink_to(_path, target_is_directory=True)
path = linkpath
else:
path = _path
ds = Dataset(path)
if has_symlink_capability():
assert ds.pathobj != ds.repo.pathobj
plain_recursive = ds.diff(recursive=True,
annex='all',
result_renderer='disabled')
# check integrity of individual reports with a focus on how symlinks
# are reported
for res in plain_recursive:
# anything that is an "intended" symlink should be reported
# as such. In contrast, anything that is a symlink for mere
# technical reasons (annex using it for something in some mode)
# should be reported as the thing it is representing (i.e.
# a file)
if 'link2' in str(res['path']):
assert res['type'] == 'symlink', res
else:
assert res['type'] != 'symlink', res
# every item must report its parent dataset
assert_in('parentds', res)
# bunch of smoke tests
# query of '.' is same as no path
eq_(
plain_recursive,
ds.diff(path='.',
recursive=True,
annex='all',
result_renderer='disabled'))
# duplicate paths do not change things
eq_(
plain_recursive,
ds.diff(path=['.', '.'],
recursive=True,
annex='all',
result_renderer='disabled'))
# neither do nested paths
if not "2.24.0" <= ds.repo.git_version < "2.25.0":
# Release 2.24.0 contained a regression that was fixed with 072a231016
# (2019-12-10).
eq_(
plain_recursive,
ds.diff(path=['.', 'subds_modified'],
recursive=True,
annex='all',
result_renderer='disabled'))
# when invoked in a subdir of a dataset it still reports on the full thing
# just like `git status`, as long as there are no paths specified
with chpwd(op.join(path, 'directory_untracked')):
plain_recursive = diff(recursive=True,
annex='all',
result_renderer='disabled')
# should be able to take absolute paths and yield the same
# output
eq_(
plain_recursive,
ds.diff(path=ds.path,
recursive=True,
annex='all',
result_renderer='disabled'))
# query for a deeply nested path from the top, should just work with a
# variety of approaches
rpath = op.join('subds_modified', 'subds_lvl1_modified',
u'{}_directory_untracked'.format(OBSCURE_FILENAME))
apathobj = ds.pathobj / rpath
apath = str(apathobj)
for p in (rpath, apath, None):
if p is None:
# change into the realpath of the dataset and
# query with an explicit path
with chpwd(ds.path):
res = ds.diff(path=op.join('.', rpath),
recursive=True,
annex='all',
result_renderer='disabled')
else:
res = ds.diff(path=p,
recursive=True,
annex='all',
result_renderer='disabled')
assert_result_count(
res,
1,
state='untracked',
type='directory',
refds=ds.path,
# path always comes out a full path inside the queried dataset
path=apath,
)
assert_result_count(ds.diff(recursive=True, result_renderer='disabled'),
1,
path=apath)
# limiting recursion will exclude this particular path
assert_result_count(ds.diff(recursive=True,
recursion_limit=1,
result_renderer='disabled'),
0,
path=apath)
# negative limit is unlimited limit
eq_(
ds.diff(recursive=True, recursion_limit=-1,
result_renderer='disabled'),
ds.diff(recursive=True, result_renderer='disabled'))