def fetch_data(repo: Repository, remote, startpoint, column, nbytes, all_):
"""Get data from REMOTE referenced by STARTPOINT (short-commit or branch).
The default behavior is to only download a single commit's data or the HEAD
commit of a branch. Please review optional arguments for other behaviors.
"""
from hangar.records.commiting import expand_short_commit_digest
from hangar.records.heads import get_branch_head_commit
from hangar.records.heads import get_staging_branch_head
from hangar.utils import parse_bytes
if startpoint is None:
branch = get_staging_branch_head(repo._env.branchenv)
commit = get_branch_head_commit(repo._env.branchenv, branch)
elif startpoint in repo.list_branches():
commit = get_branch_head_commit(repo._env.branchenv, startpoint)
else:
commit = expand_short_commit_digest(repo._env.refenv, startpoint)
click.echo(f'Fetching data for commit: {commit}')
try:
max_nbytes = parse_bytes(nbytes)
except AttributeError:
max_nbytes = None
if len(column) == 0:
column = None
commits = repo.remote.fetch_data(remote=remote,
commit=commit,
column_names=column,
max_num_bytes=max_nbytes,
retrieve_all_history=all_)
click.echo(f'completed data for commits: {commits}')
def test_check_repo_size(repo_20_filled_samples):
from hangar.utils import parse_bytes, folder_size
expected_nbytes = folder_size(repo_20_filled_samples._repo_path, recurse=True)
nbytes = repo_20_filled_samples.size_nbytes
assert expected_nbytes == nbytes
format_nbytes = repo_20_filled_samples.size_human
# account for rounding when converting int to str.
assert nbytes * 0.95 <= parse_bytes(format_nbytes) <= nbytes * 1.05
def fetch_data(ctx, remote, startpoint, aset, nbytes, all_):
"""Get data from REMOTE referenced by STARTPOINT (short-commit or branch).
The default behavior is to only download a single commit's data or the HEAD
commit of a branch. Please review optional arguments for other behaviors
"""
from hangar.records.heads import get_branch_head_commit, get_staging_branch_head
from hangar.utils import parse_bytes
P = os.getcwd()
repo = Repository(path=P)
if startpoint is None:
branch = get_staging_branch_head(repo._env.branchenv)
commit = get_branch_head_commit(repo._env.branchenv, branch)
click.echo(
f'No startpoint supplied, fetching data of HEAD: {commit} for BRANCH: {branch}'
)
elif startpoint in repo.list_branches():
commit = get_branch_head_commit(repo._env.branchenv, startpoint)
click.echo(
f'Fetching data for HEAD: {commit} of STARTPOINT BRANCH: {startpoint}'
)
else:
commit = expand_short_commit_digest(repo._env.refenv, startpoint)
click.echo(f'Fetching data for STARTPOINT HEAD: {commit}')
click.echo(f'aset argument: {aset}')
try:
max_nbytes = parse_bytes(nbytes)
click.echo(f'nbytes argument: {max_nbytes}')
except AttributeError:
max_nbytes = None
if len(aset) == 0:
aset = None
commits = repo.remote.fetch_data(remote=remote,
commit=commit,
arrayset_names=aset,
max_num_bytes=max_nbytes,
retrieve_all_history=all_)
click.echo(f'completed data for commits: {commits}')
def test_parse_bytes(arg, expected):
from hangar.utils import parse_bytes
res = parse_bytes(arg)
assert res == expected
