def tearDown(self):
self._popd()
try:
remove(self._root_dir)
except OSError as exc:
# We ignore this under Windows with a warning because it happened
# to be really hard to trace all not properly closed files.
#
# Best guess so far is that gitpython is the culprit:
# it opens files and uses __del__ to close them, which can happen
# late in current pythons. TestGitFixture and TestDvcFixture try
# to close that and it works on most of the tests, but not all.
# Repos and thus git repos are created all over the dvc ;)
if os.name == "nt" and exc.winerror == 32:
warnings.warn("Failed to remove test dir: " + str(exc))
else:
raise
def get(url, path, out=None, rev=None):
out = resolve_output(path, out)
if Stage.is_valid_filename(out):
raise GetDVCFileError()
# Creating a directory right beside the output to make sure that they
# are on the same filesystem, so we could take the advantage of
# reflink and/or hardlink. Not using tempfile.TemporaryDirectory
# because it will create a symlink to tmpfs, which defeats the purpose
# and won't work with reflink/hardlink.
dpath = os.path.dirname(os.path.abspath(out))
tmp_dir = os.path.join(dpath, "." + str(shortuuid.uuid()))
try:
try:
with external_repo(cache_dir=tmp_dir, url=url, rev=rev) as repo:
# Try any links possible to avoid data duplication.
#
# Not using symlink, because we need to remove cache after we
# are done, and to make that work we would have to copy data
# over anyway before removing the cache, so we might just copy
# it right away.
#
# Also, we can't use theoretical "move" link type here, because
# the same cache file might be used a few times in a directory.
repo.cache.local.cache_types = ["reflink", "hardlink", "copy"]
output = repo.find_out_by_relpath(path)
if output.use_cache:
_get_cached(repo, output, out)
return
# Non-cached output, fall through and try to copy from git.
except (NotDvcRepoError, NoOutputInExternalRepoError):
# Not a DVC repository or, possibly, path is not tracked by DVC.
# Fall through and try to copy from git.
pass
if os.path.isabs(path):
raise FileNotFoundError
repo_dir = cached_clone(url, rev=rev)
fs_copy(os.path.join(repo_dir, path), out)
except (OutputNotFoundError, FileNotFoundError):
raise PathMissingError(path, url)
finally:
remove(tmp_dir)
def test_pull_no_rev_lock(erepo_dir, tmp_dir, dvc):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "contents", commit="create foo")
stage = dvc.imp(os.fspath(erepo_dir), "foo", "foo_imported")
assert "rev" not in stage.deps[0].def_repo
stage.deps[0].def_repo.pop("rev_lock")
Dvcfile(dvc, stage.path).dump(stage)
remove(stage.outs[0].cache_path)
(tmp_dir / "foo_imported").unlink()
dvc.pull([stage.path])
assert (tmp_dir / "foo_imported").is_file()
assert (tmp_dir / "foo_imported").read_text() == "contents"
def get(url, path, out=None, rev=None):
out = resolve_output(path, out)
if Stage.is_valid_filename(out):
raise GetDVCFileError()
# Creating a directory right beside the output to make sure that they
# are on the same filesystem, so we could take the advantage of
# reflink and/or hardlink. Not using tempfile.TemporaryDirectory
# because it will create a symlink to tmpfs, which defeats the purpose
# and won't work with reflink/hardlink.
dpath = os.path.dirname(os.path.abspath(out))
tmp_dir = os.path.join(dpath, "." + str(shortuuid.uuid()))
try:
with external_repo(cache_dir=tmp_dir, url=url, rev=rev) as repo:
# Try any links possible to avoid data duplication.
#
# Not using symlink, because we need to remove cache after we are
# done, and to make that work we would have to copy data over
# anyway before removing the cache, so we might just copy it
# right away.
#
# Also, we can't use theoretical "move" link type here, because
# the same cache file might be used a few times in a directory.
repo.cache.local.cache_types = ["reflink", "hardlink", "copy"]
try:
output = repo.find_out_by_relpath(path)
except OutputNotFoundError:
output = None
if output and output.use_cache:
_get_cached(repo, output, out)
else:
# Either an uncached out with absolute path or a user error
if os.path.isabs(path):
raise FileNotFoundError
_copy(os.path.join(repo.root_dir, path), out)
except (OutputNotFoundError, FileNotFoundError):
raise PathMissingError(path, url)
except NotDvcRepoError:
raise UrlNotDvcRepoError(url)
finally:
remove(tmp_dir)
def test_push_incomplete_dir(tmp_dir, dvc, mocker, local_remote):
(stage, ) = tmp_dir.dvc_gen({"dir": {"foo": "foo", "bar": "bar"}})
remote = dvc.cloud.get_remote("upstream")
cache = dvc.cache.local
dir_hash = stage.outs[0].checksum
used = stage.get_used_cache(remote=remote)
# remove one of the cache files for directory
file_hashes = list(used.child_keys(cache.tree.scheme, dir_hash))
remove(cache.tree.hash_to_path_info(file_hashes[0]))
dvc.push()
assert not remote.tree.exists(remote.tree.hash_to_path_info(dir_hash))
assert not remote.tree.exists(remote.tree.hash_to_path_info(
file_hashes[0]))
assert remote.tree.exists(remote.tree.hash_to_path_info(file_hashes[1]))
def test_pull_non_workspace(tmp_dir, scm, dvc, erepo_dir):
with erepo_dir.chdir():
erepo_dir.dvc_gen("foo", "master content", commit="create foo")
with erepo_dir.branch("branch", new=True):
erepo_dir.dvc_gen("foo", "branch content", commit="modify foo")
stage = dvc.imp(os.fspath(erepo_dir), "foo", "foo_imported", rev="branch")
tmp_dir.scm_add([stage.relpath], commit="imported branch")
scm.tag("ref-to-branch")
# Overwrite via import
dvc.imp(os.fspath(erepo_dir), "foo", "foo_imported", rev="master")
remove(stage.outs[0].cache_path)
dvc.fetch(all_tags=True)
assert os.path.exists(stage.outs[0].cache_path)
def test_git_ssh(tmp_dir, scm, server):
from dulwich.repo import Repo as DulwichRepo
from sshfs import SSHFileSystem
from dvc.utils.fs import remove
from tests.remotes.ssh import TEST_SSH_KEY_PATH, TEST_SSH_USER
fs = SSHFileSystem(
host=server.host,
port=server.port,
username=TEST_SSH_USER,
client_keys=[TEST_SSH_KEY_PATH],
)
server._ssh.execute("git init --bare test-repo.git")
url = f"ssh://{TEST_SSH_USER}@{server.host}:{server.port}/~/test-repo.git"
tmp_dir.scm_gen("foo", "foo", commit="init")
rev = scm.get_rev()
scm.push_refspec(
url,
"refs/heads/master",
"refs/heads/master",
force=True,
key_filename=TEST_SSH_KEY_PATH,
)
assert (
rev.encode("ascii")
== fs.open("test-repo.git/refs/heads/master").read().strip()
)
remove(tmp_dir / ".git")
remove(tmp_dir / "foo")
DulwichRepo.init(str(tmp_dir))
scm.fetch_refspecs(
url,
["refs/heads/master"],
force=True,
key_filename=TEST_SSH_KEY_PATH,
)
assert rev == scm.get_ref("refs/heads/master")
scm.checkout("master")
assert "foo" == (tmp_dir / "foo").read_text()
def test_subrepos_are_ignored(tmp_dir, erepo_dir):
subrepo = erepo_dir / "dir" / "subrepo"
make_subrepo(subrepo, erepo_dir.scm)
with erepo_dir.chdir():
erepo_dir.dvc_gen("dir/foo", "foo", commit="foo")
erepo_dir.scm_gen("dir/bar", "bar", commit="bar")
with subrepo.chdir():
subrepo.dvc_gen({"file": "file"}, commit="add files on subrepo")
with external_repo(os.fspath(erepo_dir)) as repo:
repo.dvcfs.get(
"dir",
os.fspath(tmp_dir / "out"),
)
expected_files = {"foo": "foo", "bar": "bar", ".gitignore": "/foo\n"}
assert (tmp_dir / "out").read_text() == expected_files
# clear cache to test saving to cache
cache_dir = tmp_dir / repo.odb.local.cache_dir
remove(cache_dir)
clean_staging()
makedirs(cache_dir)
staging, _, obj = stage(
repo.odb.local,
"dir",
repo.dvcfs,
"md5",
ignore=repo.dvcignore,
)
transfer(
staging,
repo.odb.local,
{obj.hash_info},
shallow=False,
hardlink=True,
)
assert set(cache_dir.glob("??/*")) == {
cache_dir / "e1" / "d9e8eae5374860ae025ec84cfd85c7",
cache_dir / "e1" / "d9e8eae5374860ae025ec84cfd85c7.dir",
cache_dir / "37" / "b51d194a7513e45b56f6524f2d51f2",
cache_dir / "94" / "7d2b84e5aa88170e80dff467a5bfb6",
cache_dir / "ac" / "bd18db4cc2f85cedef654fccc4a4d8",
}
def checkout_exp(self, rev, **kwargs):
"""Checkout an experiment to the user's workspace."""
from git.exc import GitCommandError
from dvc.repo.checkout import checkout as dvc_checkout
baseline_rev = self._check_baseline(rev)
self._scm_checkout(rev)
branch = self._get_branch_containing(rev)
m = self.BRANCH_RE.match(branch) if branch else None
if m and m.group("checkpoint"):
kwargs.update({"allow_missing": True, "quiet": True})
tmp = tempfile.NamedTemporaryFile(delete=False).name
self.scm.repo.head.commit.diff(baseline_rev,
patch=True,
full_index=True,
binary=True,
output=tmp)
dirty = self.repo.scm.is_dirty(untracked_files=True)
if dirty:
logger.debug("Stashing workspace changes.")
self.repo.scm.repo.git.stash("push", "--include-untracked")
try:
if os.path.getsize(tmp):
logger.debug("Patching local workspace")
self.repo.scm.repo.git.apply(tmp, reverse=True)
need_checkout = True
else:
need_checkout = False
except GitCommandError:
raise DvcException("failed to apply experiment changes.")
finally:
remove(tmp)
if dirty:
self._unstash_workspace()
args_file = os.path.join(self.repo.tmp_dir, self.PACKED_ARGS_FILE)
if os.path.exists(args_file):
remove(args_file)
if need_checkout:
dvc_checkout(self.repo, **kwargs)
def _pull_cached(self, out, path_info, dest):
with self.state:
tmp = PathInfo(tmp_fname(dest))
src = tmp / path_info.relative_to(out.path_info)
out.path_info = tmp
# Only pull unless all needed cache is present
if out.changed_cache(filter_info=src):
self.cloud.pull(out.get_used_cache(filter_info=src))
failed = out.checkout(filter_info=src)
move(src, dest)
remove(tmp)
if failed:
raise FileNotFoundError
def _stash_exp(self, *args, **kwargs):
"""Stash changes from the current (parent) workspace as an experiment.
"""
rev = self.scm.get_rev()
tmp = tempfile.NamedTemporaryFile(delete=False).name
try:
self.repo.scm.repo.git.diff(patch=True, output=tmp)
if os.path.getsize(tmp):
logger.debug("Patching experiment workspace")
self.scm.repo.git.apply(tmp)
else:
raise UnchangedExperimentError(rev)
finally:
remove(tmp)
self._pack_args(*args, **kwargs)
msg = f"{self.STASH_MSG_PREFIX}{rev}"
self.scm.repo.git.stash("push", "-m", msg)
return self.scm.resolve_rev("stash@{0}")
def test_missing_cache(tmp_dir, dvc, run_copy_metrics):
tmp_dir.gen("metrics_t.yaml", "1.1")
run_copy_metrics(
"metrics_t.yaml",
"metrics.yaml",
metrics=["metrics.yaml"],
)
# This one should be skipped
stage = run_copy_metrics(
"metrics_t.yaml",
"metrics2.yaml",
metrics=["metrics2.yaml"],
)
remove(stage.outs[0].fspath)
remove(stage.outs[0].cache_path)
assert dvc.metrics.show() == {"": {"metrics.yaml": 1.1}}
def test_diff_no_cache(tmp_dir, scm, dvc):
tmp_dir.dvc_gen({"dir": {"file": "file content"}}, commit="first")
scm.tag("v1")
tmp_dir.dvc_gen(
{"dir": {"file": "modified file content"}}, commit="second"
)
scm.tag("v2")
remove(dvc.cache.local.cache_dir)
# invalidate_dir_info to force cache loading
dvc.cache.local._dir_info = {}
diff = dvc.diff("v1", "v2")
assert diff["added"] == []
assert diff["deleted"] == []
assert first(diff["modified"])["path"] == os.path.join("dir", "")
def test_pipeline_file_target_ops(tmp_dir, dvc, run_copy, local_remote):
path = local_remote.url
tmp_dir.dvc_gen("foo", "foo")
run_copy("foo", "bar", single_stage=True)
tmp_dir.dvc_gen("lorem", "lorem")
run_copy("lorem", "lorem2", name="copy-lorem-lorem2")
tmp_dir.dvc_gen("ipsum", "ipsum")
run_copy("ipsum", "baz", name="copy-ipsum-baz")
outs = ["foo", "bar", "lorem", "ipsum", "baz", "lorem2"]
remove(dvc.stage_cache.cache_dir)
dvc.push()
outs = ["foo", "bar", "lorem", "ipsum", "baz", "lorem2"]
# each one's a copy of other, hence 3
assert len(recurse_list_dir(path)) == 3
clean(outs, dvc)
assert set(dvc.pull(["dvc.yaml"])["added"]) == {"lorem2", "baz"}
clean(outs, dvc)
assert set(dvc.pull()["added"]) == set(outs)
# clean everything in remote and push
from tests.dir_helpers import TmpDir
clean(TmpDir(path).iterdir())
dvc.push(["dvc.yaml:copy-ipsum-baz"])
assert len(recurse_list_dir(path)) == 1
clean(TmpDir(path).iterdir())
dvc.push(["dvc.yaml"])
assert len(recurse_list_dir(path)) == 2
with pytest.raises(StageNotFound):
dvc.push(["dvc.yaml:StageThatDoesNotExist"])
with pytest.raises(StageNotFound):
dvc.pull(["dvc.yaml:StageThatDoesNotExist"])
def get(url, path, out=None, rev=None):
out = resolve_output(path, out)
if Stage.is_valid_filename(out):
raise GetDVCFileError()
# Creating a directory right beside the output to make sure that they
# are on the same filesystem, so we could take the advantage of
# reflink and/or hardlink. Not using tempfile.TemporaryDirectory
# because it will create a symlink to tmpfs, which defeats the purpose
# and won't work with reflink/hardlink.
dpath = os.path.dirname(os.path.abspath(out))
tmp_dir = os.path.join(dpath, "." + str(shortuuid.uuid()))
try:
with external_repo(cache_dir=tmp_dir, url=url, rev=rev) as repo:
# Note: we need to replace state, because in case of getting DVC
# dependency on CIFS or NFS filesystems, sqlite-based state
# will be unable to obtain lock
repo.state = StateNoop()
# Try any links possible to avoid data duplication.
#
# Not using symlink, because we need to remove cache after we are
# done, and to make that work we would have to copy data over
# anyway before removing the cache, so we might just copy it
# right away.
#
# Also, we can't use theoretical "move" link type here, because
# the same cache file might be used a few times in a directory.
repo.cache.local.cache_types = ["reflink", "hardlink", "copy"]
o = repo.find_out_by_relpath(path)
with repo.state:
repo.cloud.pull(o.get_used_cache())
o.path_info = PathInfo(os.path.abspath(out))
with o.repo.state:
o.checkout()
except NotDvcRepoError:
raise UrlNotDvcRepoError(url)
except OutputNotFoundError:
raise OutputNotFoundError(path)
finally:
remove(tmp_dir)
def test_rename_multiple_files_same_hashes(tmp_dir, scm, dvc):
"""Test diff by renaming >=2 instances of file with same hashes.
DVC should be able to detect that they are renames, and should not include
them in either of the `added` or the `deleted` section.
"""
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "subdir": {"foo": "foo"}}}, commit="commit #1"
)
remove(tmp_dir / "dir")
# changing foo and subdir/foo to bar and subdir/bar respectively
tmp_dir.dvc_gen(
{"dir": {"bar": "foo", "subdir": {"bar": "foo"}}}, commit="commit #2"
)
assert dvc.diff("HEAD~") == {
"added": [],
"deleted": [],
"modified": [
{
"hash": {
"new": "31b36b3ea5f4485e27f10578c47183b0.dir",
"old": "c7684c8b3b0d28cf80d5305e2d856bfc.dir",
},
"path": os.path.join("dir", ""),
}
],
"not in cache": [],
"renamed": [
{
"hash": "acbd18db4cc2f85cedef654fccc4a4d8",
"path": {
"new": os.path.join("dir", "bar"),
"old": os.path.join("dir", "foo"),
},
},
{
"hash": "acbd18db4cc2f85cedef654fccc4a4d8",
"path": {
"new": os.path.join("dir", "subdir", "bar"),
"old": os.path.join("dir", "subdir", "foo"),
},
},
],
}
def test_subrepos_are_ignored(tmp_dir, erepo_dir):
subrepo = erepo_dir / "dir" / "subrepo"
make_subrepo(subrepo, erepo_dir.scm)
with erepo_dir.chdir():
erepo_dir.dvc_gen("dir/foo", "foo", commit="foo")
erepo_dir.scm_gen("dir/bar", "bar", commit="bar")
with subrepo.chdir():
subrepo.dvc_gen({"file": "file"}, commit="add files on subrepo")
with external_repo(os.fspath(erepo_dir)) as repo:
repo.repo_tree.download(
PathInfo(repo.root_dir) / "dir",
PathInfo(tmp_dir / "out"),
follow_subrepos=False,
)
expected_files = {"foo": "foo", "bar": "bar", ".gitignore": "/foo\n"}
assert (tmp_dir / "out").read_text() == expected_files
# clear cache to test saving to cache
cache_dir = tmp_dir / repo.cache.local.cache_dir
remove(cache_dir)
makedirs(cache_dir)
expected_hash = HashInfo("md5", "e1d9e8eae5374860ae025ec84cfd85c7.dir")
assert (
repo.repo_tree.get_hash(
os.path.join(repo.root_dir, "dir"), follow_subrepos=False
)
== expected_hash
)
repo.cache.local.save(
PathInfo(repo.root_dir) / "dir",
repo.repo_tree,
expected_hash,
link=False,
)
assert set(cache_dir.glob("*/*")) == {
cache_dir / "e1" / "d9e8eae5374860ae025ec84cfd85c7.dir",
cache_dir / "37" / "b51d194a7513e45b56f6524f2d51f2",
cache_dir / "94" / "7d2b84e5aa88170e80dff467a5bfb6",
cache_dir / "ac" / "bd18db4cc2f85cedef654fccc4a4d8",
}
def test_checkout_targets_deps(tmp_dir, scm, dvc, exp_stage):
from dvc.utils.fs import remove
tmp_dir.dvc_gen({"foo": "foo", "bar": "bar"}, commit="add files")
stage = dvc.stage.add(
cmd="python copy.py params.yaml metrics.yaml",
metrics_no_cache=["metrics.yaml"],
params=["foo"],
name="copy-file",
deps=["copy.py", "foo"],
force=True,
)
remove("foo")
remove("bar")
dvc.experiments.run(stage.addressing, params=["foo=2"])
assert (tmp_dir / "foo").exists()
assert (tmp_dir / "foo").read_text() == "foo"
assert not (tmp_dir / "bar").exists()
def _unprotect_file(path):
if System.is_symlink(path) or System.is_hardlink(path):
logger.debug("Unprotecting '{}'".format(path))
tmp = os.path.join(os.path.dirname(path), "." + uuid())
# The operations order is important here - if some application
# would access the file during the process of copyfile then it
# would get only the part of file. So, at first, the file should be
# copied with the temporary name, and then original file should be
# replaced by new.
copyfile(path, tmp, name="Unprotecting '{}'".format(relpath(path)))
remove(path)
os.rename(tmp, path)
else:
logger.debug("Skipping copying for '{}', since it is not "
"a symlink or a hardlink.".format(path))
os.chmod(path, os.stat(path).st_mode | stat.S_IWRITE)
def _reproduce(self, executors: dict, jobs: Optional[int] = 1) -> dict:
"""Run dvc repro for the specified ExperimentExecutors in parallel.
Returns dict containing successfully executed experiments.
"""
result = {}
with ProcessPoolExecutor(max_workers=jobs) as workers:
futures = {}
for rev, executor in executors.items():
future = workers.submit(
executor.reproduce,
executor.dvc_dir,
cwd=executor.dvc.root_dir,
**executor.repro_kwargs,
)
futures[future] = (rev, executor)
for future in as_completed(futures):
rev, executor = futures[future]
exc = future.exception()
if exc is None:
exp_hash = future.result()
logger.debug(f"ran exp based on {executor.baseline_rev}")
self._scm_checkout(executor.baseline_rev)
self._collect_output(executor.baseline_rev, executor)
remove(self.args_file)
try:
exp_rev = self._commit(exp_hash)
except UnchangedExperimentError:
logger.debug(
"Experiment '%s' identical to baseline '%s'",
rev,
executor.baseline_rev,
)
exp_rev = executor.baseline_rev
logger.info("Reproduced experiment '%s'.", exp_rev[:7])
result[rev] = {exp_rev: exp_hash}
else:
logger.exception("Failed to reproduce experiment '%s'",
rev)
executor.cleanup()
return result
def test_api_missing_local_cache_exists_on_remote(
tmp_dir,
scm,
dvc,
as_external,
remote,
files,
to_read,
):
tmp_dir.dvc_gen(files, commit="DVC track files")
dvc.push()
# Remove cache to make foo missing
remove(dvc.cache.local.cache_dir)
remove(first(files))
repo_url = f"file://{tmp_dir}" if as_external else None
file_content = get_in(files, to_read.split(os.sep))
assert api.read(to_read, repo=repo_url) == file_content
def apply(repo: "Repo", rev: str, force: bool = True, **kwargs):
from scmrepo.exceptions import SCMError as _SCMError
from dvc.repo.checkout import checkout as dvc_checkout
from dvc.scm import GitMergeError, RevError, resolve_rev
exps = repo.experiments
try:
exp_rev = resolve_rev(repo.scm, rev)
exps.check_baseline(exp_rev)
except (RevError, BaselineMismatchError) as exc:
raise InvalidExpRevError(rev) from exc
stash_rev = exp_rev in exps.stash_revs
if not stash_rev and not exps.get_branch_by_rev(exp_rev,
allow_multiple=True):
raise InvalidExpRevError(exp_rev)
# NOTE: we don't use scmrepo's stash_workspace() here since we need
# finer control over the merge behavior when we unstash everything
with _apply_workspace(repo, rev, force):
try:
repo.scm.merge(exp_rev, commit=False, squash=True)
except _SCMError as exc:
raise GitMergeError(str(exc), scm=repo.scm)
repo.scm.reset()
if stash_rev:
args_path = os.path.join(repo.tmp_dir, BaseExecutor.PACKED_ARGS_FILE)
if os.path.exists(args_path):
remove(args_path)
dvc_checkout(repo, **kwargs)
repo.scm.set_ref(EXEC_APPLY, exp_rev)
logger.info(
"Changes for experiment '%s' have been applied to your current "
"workspace.",
rev,
)
def get(url, path, out=None, rev=None, jobs=None):
import shortuuid
from dvc.dvcfile import is_valid_filename
from dvc.external_repo import external_repo
out = resolve_output(path, out)
if is_valid_filename(out):
raise GetDVCFileError()
# Creating a directory right beside the output to make sure that they
# are on the same filesystem, so we could take the advantage of
# reflink and/or hardlink. Not using tempfile.TemporaryDirectory
# because it will create a symlink to tmpfs, which defeats the purpose
# and won't work with reflink/hardlink.
dpath = os.path.dirname(os.path.abspath(out))
tmp_dir = os.path.join(dpath, "." + str(shortuuid.uuid()))
# Try any links possible to avoid data duplication.
#
# Not using symlink, because we need to remove cache after we
# are done, and to make that work we would have to copy data
# over anyway before removing the cache, so we might just copy
# it right away.
#
# Also, we can't use theoretical "move" link type here, because
# the same cache file might be used a few times in a directory.
cache_types = ["reflink", "hardlink", "copy"]
try:
with external_repo(url=url,
rev=rev,
cache_dir=tmp_dir,
cache_types=cache_types) as repo:
from_info = PathInfo(repo.root_dir) / path
to_info = PathInfo(out)
repo.repo_tree.download(from_info,
to_info,
jobs=jobs,
follow_subrepos=False)
finally:
remove(tmp_dir)
def test_show_dir_plots(tmp_dir, dvc, run_copy_metrics):
subdir = tmp_dir / "subdir"
subdir.mkdir()
metric = [
{
"first_val": 100,
"val": 2
},
{
"first_val": 200,
"val": 3
},
]
fname = "file.json"
_write_json(tmp_dir, metric, fname)
p1 = os.path.join("subdir", "p1.json")
p2 = os.path.join("subdir", "p2.json")
tmp_dir.dvc.run(
cmd=(f"mkdir subdir && python copy.py {fname} {p1} && "
f"python copy.py {fname} {p2}"),
deps=[fname],
single_stage=False,
plots=["subdir"],
name="copy_double",
)
result = dvc.plots.show(targets=["subdir"])
p1_content = json.loads(result[p1])
p2_content = json.loads(result[p2])
assert p1_content == p2_content
result = dvc.plots.show(targets=[p1])
assert set(result.keys()) == {p1}
remove(dvc.odb.local.cache_dir)
remove(subdir)
with pytest.raises(NoMetricsParsedError):
dvc.plots.show()
def test_open_external(tmp_dir, erepo_dir, cloud):
erepo_dir.add_remote(config=cloud.config)
with erepo_dir.chdir():
erepo_dir.dvc_gen("version", "master", commit="add version")
with erepo_dir.branch("branch", new="True"):
# NOTE: need file to be other size for Mac
erepo_dir.dvc_gen("version", "branchver", commit="add version")
erepo_dir.dvc.push(all_branches=True)
# Remove cache to force download
remove(erepo_dir.dvc.cache.local.cache_dir)
# Using file url to force clone to tmp repo
repo_url = f"file://{erepo_dir}"
with api.open("version", repo=repo_url) as fd:
assert fd.read() == "master"
assert api.read("version", repo=repo_url, rev="branch") == "branchver"
def test_open_external(remote_url, erepo_dir):
_set_remote_url_and_commit(erepo_dir.dvc, remote_url)
with erepo_dir.chdir():
erepo_dir.dvc_gen("version", "master", commit="add version")
with erepo_dir.branch("branch", new="True"):
# NOTE: need file to be other size for Mac
erepo_dir.dvc_gen("version", "branchver", commit="add version")
erepo_dir.dvc.push(all_branches=True)
# Remove cache to force download
remove(erepo_dir.dvc.cache.local.cache_dir)
# Using file url to force clone to tmp repo
repo_url = "file://{}".format(erepo_dir)
with api.open("version", repo=repo_url) as fd:
assert fd.read() == "master"
assert api.read("version", repo=repo_url, rev="branch") == "branchver"
def test_no_cache_entry(tmp_dir, scm, dvc):
tmp_dir.dvc_gen("file", "first", commit="add a file")
tmp_dir.dvc_gen({"dir": {"1": "1", "2": "2"}})
tmp_dir.dvc_gen("file", "second")
remove(tmp_dir / ".dvc" / "cache")
(tmp_dir / ".dvc" / "tmp" / "state").unlink()
dir_checksum = "5fb6b29836c388e093ca0715c872fe2a.dir"
assert dvc.diff() == {
"added": [{"path": os.path.join("dir", ""), "hash": dir_checksum}],
"deleted": [],
"modified": [
{
"path": "file",
"hash": {"old": digest("first"), "new": digest("second")},
}
],
}
def test_collect_non_existing_dir(tmp_dir, dvc, run_copy_metrics):
subdir = tmp_dir / "subdir"
subdir.mkdir()
metric = [{"first_val": 100, "val": 2}, {"first_val": 200, "val": 3}]
subdir_metric = [{"y": 101, "x": 3}, {"y": 202, "x": 4}]
pname = "source.json"
(tmp_dir / pname).dump_json(metric, sort_keys=True)
sname = "subdir_source.json"
(tmp_dir / sname).dump_json(subdir_metric, sort_keys=True)
p1 = os.path.join("subdir", "p1.json")
p2 = os.path.join("subdir", "p2.json")
subdir_stage = tmp_dir.dvc.run(
cmd=(
f"mkdir subdir && python copy.py {sname} {p1} && "
f"python copy.py {sname} {p2}"
),
deps=[sname],
single_stage=False,
plots=["subdir"],
name="copy_double",
)
run_copy_metrics(
pname,
"plot.json",
plots=["plot.json"],
commit="there is metric",
)
remove(subdir_stage.outs[0].cache_path)
remove(subdir_stage.outs[0].fs_path)
result = dvc.plots.show()
assert "error" in result["workspace"]["data"]["subdir"]
# make sure others gets loaded
assert result["workspace"]["data"]["plot.json"]["data"] == metric
def test_pull_imported_stage_from_subrepos(tmp_dir, dvc, erepo_dir, is_dvc,
files):
subrepo = erepo_dir / "subrepo"
make_subrepo(subrepo, erepo_dir.scm)
gen = subrepo.dvc_gen if is_dvc else subrepo.scm_gen
with subrepo.chdir():
gen(files, commit="files in subrepo")
key = first(files)
path = os.path.join("subrepo", key)
dvc.imp(os.fspath(erepo_dir), path, out="out")
# clean everything
remove(dvc.odb.local.cache_dir)
remove("out")
makedirs(dvc.odb.local.cache_dir)
stats = dvc.pull(["out.dvc"])
expected = [f"out{os.sep}"] if isinstance(files[key], dict) else ["out"]
assert stats["added"] == expected
assert (tmp_dir / "out").read_text() == files[key]
def test_branch_config(tmp_dir, scm):
tmp_dir.scm_gen("foo", "foo", commit="init")
scm.checkout("branch", create_new=True)
dvc = Repo.init()
with dvc.config.edit() as conf:
conf["remote"]["branch"] = {"url": "/some/path"}
scm.add([".dvc"])
scm.commit("init dvc")
scm.checkout("master")
remove(".dvc")
# sanity check
with pytest.raises(NotDvcRepoError):
Repo()
with pytest.raises(NotDvcRepoError):
Repo(scm=scm, rev="master")
dvc = Repo(scm=scm, rev="branch")
assert dvc.config["remote"]["branch"]["url"] == "/some/path"
