def get_file_id(rec):
"""Returns a suitable '@id' of a file metadata from a status result
Prefer a present annex key, but fall back on the Git shasum that is
always around. Identify the GITSHA as such, and in a similar manner
to git-annex's style.
Any ID string is prefixed with 'datalad:' to identify it as a
DataLad-recognized ID. This prefix is defined in the main JSON-LD
context defintion.
"""
id_ = rec['key'] if 'key' in rec else 'SHA1-s{}--{}'.format(
rec['bytesize'] if 'bytesize' in rec else
0 if rec['type'] == 'symlink' else os.stat(rec['path']).st_size,
rec['gitshasum'] if 'gitshasum' in rec else Digester(
digests=['sha1'])(rec['path'])['sha1'])
return 'datalad:{}'.format(id_)
def get_mtimes_and_digests(target_path):
"""Return digests (md5) and mtimes for all the files under target_path"""
from datalad.utils import find_files
from datalad.support.digests import Digester
digester = Digester(['md5'])
# bother only with existing ones for this test, i.e. skip annexed files without content
target_files = [
f for f in find_files('.*', topdir=target_path, exclude_vcs=False, exclude_datalad=False)
if exists(f)
]
# let's leave only relative paths for easier analysis
target_files_ = [relpath(f, target_path) for f in target_files]
digests = {frel: digester(f) for f, frel in zip(target_files, target_files_)}
mtimes = {frel: os.stat(f).st_mtime for f, frel in zip(target_files, target_files_)}
return digests, mtimes
def _the_same_across_datasets(relpath, *dss):
"""Check if the file (present content or not) is identical across two datasets
Compares files by content if under git, or by checksum if under annex
Parameters
----------
*ds: Datasets
relpath: str
path within datasets
Returns
-------
bool or None
True if identical, False if not, None if cannot be decided
(e.g. different git-annex backend used)
"""
from datalad.utils import md5sum, unique
from datalad.support.exceptions import FileInGitError
from datalad.support.digests import Digester
paths = [op.join(ds.path, relpath) for ds in dss]
# The simplest check first -- exist in both and content is the same.
# Even if content is just a symlink file on windows, the same content
# condition would be correct
if all(map(op.exists, paths)) and all_same(map(md5sum, paths)):
return True
# We first need to find problematic ones which are annexed and
# have no content locally, and take their
keys = []
backends = []
presents = []
for ds in dss:
repo = ds.repo
key = None
present = True
if isinstance(repo, AnnexRepo):
annexprops = repo.get_file_annexinfo(
relpath, eval_availability=True)
if 'key' not in annexprops:
continue
key = annexprops['key']
# For now the rest (e.g. not tracked) remains an error
if not annexprops['has_content']:
present = False
backends.append(repo.get_key_backend(key))
keys.append(key)
presents.append(present)
if all(presents):
return all_same(map(md5sum, paths))
backends = unique(backends)
assert backends, "Since not all present - some must be under annex, and thus must have a backend!"
# so some files are missing!
assert not all(presents)
NeedContentError = RuntimeError
if len(backends) > 1:
# TODO: or signal otherwise somehow that we just need to get at least some
# of those files to do the check!...
raise NeedContentError(
"Following paths are missing content and have different annex "
"backends: %s. Cannot determine here if the same or not!"
% ", ".join(p for (p, b) in zip(paths, presents) if not b)
)
backend = backends[0].lower()
if backend.endswith('E'):
backend = backend[':-1']
if backend not in Digester.DEFAULT_DIGESTS:
raise NeedContentError(
"Do not know how to figure out content check for backend %s" % backend
)
checksums = [
split_ext(key).split('--', 1)[1] if key else key
for key in keys
]
thechecksum = set(
checksum
for present, checksum in zip(presents, checksums)
if present
)
if len(thechecksum) > 1:
# Different checksum (with the same backend)
return False
elif not thechecksum:
raise RuntimeError("We must have had at least one key since prior logic"
" showed that not all files have content here")
thechecksum = thechecksum[0]
if any(presents):
# We do need to extract checksum from the key and check the present
# files' content to match
digester = Digester([backend])
for present, path in zip(presents, paths):
if present and digester(path)[backend] != thechecksum:
return False
return True
return False
