def test_copy_file_errors(dspath1, dspath2, nondspath):
ds1 = Dataset(dspath1)
# nothing given
assert_raises(ValueError, copy_file)
# no target directory given
assert_raises(ValueError, ds1.copy_file, 'somefile')
# using multiple sources and --specs-from
assert_raises(ValueError, ds1.copy_file, ['1', '2', '3'], specs_from='-')
# trying to copy to a dir that is not in a dataset
ds1.create()
assert_status(
'error',
ds1.copy_file('somepath', target_dir=nondspath, on_failure='ignore'))
# copy into a dataset that is not in the reference dataset
ds2 = Dataset(dspath2).create()
assert_status(
'error',
ds1.copy_file('somepath', target_dir=dspath2, on_failure='ignore'))
# attempt to copy from a directory, but no recursion is enabled.
# use no reference ds to excercise a different code path
assert_status('impossible',
copy_file([nondspath, dspath1], on_failure='ignore'))
# attempt to copy a file that doesn't exist
assert_status('impossible',
copy_file(['funky', dspath1], on_failure='ignore'))
# attempt to copy a file without a destination given
assert_raises(ValueError, copy_file, 'somepath')
assert_status('impossible',
copy_file(specs_from=['somepath'], on_failure='ignore'))
def test_copy_file_recursion(srcdir, destdir):
src_ds = Dataset(srcdir).create(force=True)
src_ds.save()
dest_ds = Dataset(destdir).create()
copy_file([src_ds.pathobj / 'subdir', dest_ds.pathobj], recursive=True)
# structure is mirrored
ok_file_has_content(dest_ds.pathobj / 'subdir' / 'file1', '123')
ok_file_has_content(dest_ds.pathobj / 'subdir' / 'file2', 'abc')
def test_copy_file_into_nonannex(workdir):
workdir = Path(workdir)
src_ds = Dataset(workdir / 'src').create()
(src_ds.pathobj / 'present.txt').write_text('123')
(src_ds.pathobj / 'gone.txt').write_text('abc')
src_ds.save()
src_ds.drop('gone.txt', check=False)
# destination has no annex
dest_ds = Dataset(workdir / 'dest').create(annex=False)
# no issue copying a file that has content
copy_file([src_ds.pathobj / 'present.txt', dest_ds.pathobj])
ok_file_has_content(dest_ds.pathobj / 'present.txt', '123')
# but cannot handle a dropped file, no chance to register
# availability info in an annex
assert_status(
'impossible',
copy_file([src_ds.pathobj / 'gone.txt', dest_ds.pathobj],
on_failure='ignore'))
def test_copy_file_datalad_specialremote(workdir, webdir, weburl):
workdir = Path(workdir)
src_ds = Dataset(workdir / 'src').create()
# enable datalad special remote
src_ds.repo.init_remote(DATALAD_SPECIAL_REMOTE, [
'encryption=none', 'type=external',
'externaltype={}'.format(DATALAD_SPECIAL_REMOTE), 'autoenable=true'
])
# put files into the dataset by URL
src_ds.download_url('/'.join((weburl, 'webfile1')), path='myfile1.txt')
src_ds.download_url('/'.join((weburl, 'webfile2')), path='myfile2.txt')
# approx test that the file is known to a remote
# that is not the web remote
assert_in_results(
src_ds.repo.whereis('myfile1.txt', output='full').values(),
here=False,
description='[{}]'.format(DATALAD_SPECIAL_REMOTE),
)
# now a new dataset
dest_ds = Dataset(workdir / 'dest').create()
# no special remotes
eq_(dest_ds.repo.get_special_remotes(), {})
copy_file([src_ds.pathobj / 'myfile1.txt', dest_ds.pathobj])
# we have an special remote in the destination dataset now
assert_in_results(
dest_ds.repo.get_special_remotes().values(),
externaltype=DATALAD_SPECIAL_REMOTE,
)
# and it works
dest_ds.drop('myfile1.txt')
dest_ds.repo.get('myfile1.txt', remote='datalad')
ok_file_has_content(dest_ds.pathobj / 'myfile1.txt', '123')
# now replace file in dest with a different content at the same path
copy_file(
[src_ds.pathobj / 'myfile2.txt', dest_ds.pathobj / 'myfile1.txt'])
dest_ds.drop('myfile1.txt')
dest_ds.repo.get('myfile1.txt', remote='datalad')
# no gets the "same path" but yields different content
ok_file_has_content(dest_ds.pathobj / 'myfile1.txt', 'abc')
def test_copy_file_specs_from(srcdir, destdir):
srcdir = Path(srcdir)
destdir = Path(destdir)
files = [p for p in srcdir.glob('**/*') if not p.is_dir()]
# plain list of absolute path objects
r_srcabs, res = _check_copy_file_specs_from(srcdir, destdir / 'srcabs',
files)
# same, but with relative paths
with chpwd(srcdir):
r_srcrel, res = _check_copy_file_specs_from(
srcdir, destdir / 'srcrel', [p.relative_to(srcdir) for p in files])
# same, but as strings
r_srcabs_str, res = _check_copy_file_specs_from(srcdir,
destdir / 'srcabs_str',
[str(p) for p in files])
with chpwd(srcdir):
r_srcrel_str, res = _check_copy_file_specs_from(
srcdir, destdir / 'srcrel_str',
[str(p.relative_to(srcdir)) for p in files])
# same, but with src/dest pairs
r_srcdestabs_str, res = _check_copy_file_specs_from(
srcdir, destdir / 'srcdestabs_str', [
'{}\0{}'.format(str(p), str(destdir / 'srcdestabs_str' / p.name))
for p in files
])
# all methods lead to the same dataset structure
for a, b in ((r_srcabs, r_srcrel), (r_srcabs, r_srcabs_str),
(r_srcabs, r_srcrel_str), (r_srcabs, r_srcdestabs_str)):
eq_(*[
sorted(
r
for r in d.status(result_xfm='relpaths', result_renderer=None))
for d in (a, b)
])
# fail on destination outside of the dest repo
res = copy_file(specs_from=[
'{}\0{}'.format(str(p),
str(destdir / 'srcdest_wrong' / p.relative_to(srcdir)))
for p in files
],
on_failure='ignore')
assert_status('error', res)