def test_remove_stage_preserves_comment(tmp_dir, dvc, run_copy):
tmp_dir.gen(
"dvc.yaml",
textwrap.dedent("""\
stages:
generate-foo:
cmd: "echo foo > foo"
# This copies 'foo' text to 'foo' file.
outs:
- foo
copy-foo-bar:
cmd: "python copy.py foo bar"
deps:
- foo
outs:
- bar"""),
)
dvc.reproduce(PIPELINE_FILE)
dvc_file = Dvcfile(dvc, PIPELINE_FILE)
assert dvc_file.exists()
assert (tmp_dir / PIPELINE_LOCK).exists()
assert (tmp_dir / "foo").exists()
assert (tmp_dir / "bar").exists()
dvc_file.remove_stage(dvc_file.stages["copy-foo-bar"])
assert ("# This copies 'foo' text to 'foo' file."
in (tmp_dir / PIPELINE_FILE).read_text())
def run(self, fname=None, no_exec=False, single_stage=False, **kwargs):
from dvc.stage import PipelineStage, Stage, create_stage
from dvc.dvcfile import Dvcfile, PIPELINE_FILE
if not kwargs.get("cmd"):
raise InvalidArgumentError("command is not specified")
stage_cls = PipelineStage
path = PIPELINE_FILE
stage_name = kwargs.get("name")
if stage_name and single_stage:
raise InvalidArgumentError(
"`-n|--name` is incompatible with `--single-stage`")
if not stage_name and not single_stage:
raise InvalidArgumentError("`-n|--name` is required")
if single_stage:
kwargs.pop("name", None)
stage_cls = Stage
path = fname or _get_file_path(kwargs)
else:
if not is_valid_name(stage_name):
raise InvalidStageName
params = parse_params(kwargs.pop("params", []))
stage = create_stage(stage_cls,
repo=self,
path=path,
params=params,
**kwargs)
if stage is None:
return None
dvcfile = Dvcfile(self, stage.path)
if dvcfile.exists():
if kwargs.get("overwrite", True):
dvcfile.remove_stage(stage)
elif stage_cls != PipelineStage:
raise StageFileAlreadyExistsError(dvcfile.relpath)
elif stage_name and stage_name in dvcfile.stages:
raise DuplicateStageName(stage_name, dvcfile)
try:
self.check_modified_graph([stage])
except OutputDuplicationError as exc:
raise OutputDuplicationError(exc.output, set(exc.stages) - {stage})
if no_exec:
stage.ignore_outs()
else:
stage.run(
no_commit=kwargs.get("no_commit", False),
run_cache=kwargs.get("run_cache", True),
)
dvcfile.dump(stage, update_pipeline=True, no_lock=no_exec)
return stage
def test_remove_stage_removes_dvcfiles_if_no_stages_left(
tmp_dir, dvc, run_copy):
tmp_dir.gen("foo", "foo")
run_copy("foo", "bar", name="run_copy")
dvc_file = Dvcfile(dvc, PIPELINE_FILE)
assert dvc_file.exists()
assert (tmp_dir / PIPELINE_LOCK).exists()
assert (tmp_dir / "foo").exists()
dvc_file.remove_stage(dvc_file.stages["run_copy"])
assert not dvc_file.exists()
assert not (tmp_dir / PIPELINE_LOCK).exists()
def test_remove_stage_on_lockfile_format_error(tmp_dir, dvc, run_copy):
tmp_dir.gen("foo", "foo")
stage = run_copy("foo", "bar", name="copy-foo-bar")
dvc_file = Dvcfile(dvc, stage.path)
lock_file = dvc_file._lockfile
data = dvc_file._load()[0]
lock_data = lock_file.load()
lock_data["gibberish"] = True
data["gibberish"] = True
dump_yaml(lock_file.relpath, lock_data)
with pytest.raises(StageFileFormatError):
dvc_file.remove_stage(stage)
lock_file.remove()
dvc_file.dump(stage, update_pipeline=False)
dump_yaml(dvc_file.relpath, data)
with pytest.raises(StageFileFormatError):
dvc_file.remove_stage(stage)
def test_remove_stage_dvcfiles(tmp_dir, dvc, run_copy):
tmp_dir.gen("foo", "foo")
stage = run_copy("foo", "bar", single_stage=True)
dvc_file = Dvcfile(dvc, stage.path)
assert dvc_file.exists()
dvc_file.remove_stage(stage)
assert not dvc_file.exists()
# re-check to see if it fails if there's no stage entry
dvc_file.remove_stage(stage)
dvc_file.remove(force=True)
# should not fail when there's no file at all.
dvc_file.remove_stage(stage)
def test_remove_stage(tmp_dir, dvc, run_copy):
tmp_dir.gen("foo", "foo")
stage = run_copy("foo", "bar", name="copy-foo-bar")
stage2 = run_copy("bar", "foobar", name="copy-bar-foobar")
dvc_file = Dvcfile(dvc, PIPELINE_FILE)
assert dvc_file.exists()
assert {"copy-bar-foobar",
"copy-foo-bar"} == set(dvc_file._load()[0]["stages"].keys())
dvc_file.remove_stage(stage)
assert ["copy-bar-foobar"] == list(dvc_file._load()[0]["stages"].keys())
# sanity check
stage2.reload()
# re-check to see if it fails if there's no stage entry
dvc_file.remove_stage(stage)
dvc_file.remove(force=True)
# should not fail when there's no file at all.
dvc_file.remove_stage(stage)
