def test_when_params_is_omitted_in_interactive_mode(tmp_dir, scm, dvc):
(tmp_dir / "params.yaml").dump({"foo": 1})
inp = io.StringIO("python script.py\nscript.py\ndata\nmodels\nn")
init(
dvc, interactive=True, stream=inp, defaults=CmdExperimentsInit.DEFAULTS
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"default": {
"cmd": "python script.py",
"deps": ["data", "script.py"],
"metrics": [{"metrics.json": {"cache": False}}],
"outs": ["models"],
"plots": [{"plots": {"cache": False}}],
}
}
}
assert not (tmp_dir / "dvc.lock").exists()
scm._reset()
assert scm.is_tracked("dvc.yaml")
assert not scm.is_tracked("params.yaml")
assert scm.is_tracked(".gitignore")
assert scm.is_ignored("models")
def test_init_with_type_live_and_models_plots_provided(
tmp_dir, dvc, interactive, inp
):
(tmp_dir / "params.yaml").dump({"foo": 1})
init(
dvc,
type="live",
interactive=interactive,
stream=inp,
defaults=CmdExperimentsInit.DEFAULTS,
overrides={"cmd": "cmd", "metrics": "m", "plots": "p"},
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"live": {
"cmd": "cmd",
"deps": ["data", "src"],
"live": {"dvclive": {"html": True, "summary": True}},
"metrics": [{"m": {"cache": False}}],
"outs": [{"models": {"checkpoint": True}}],
"params": ["foo"],
"plots": [{"p": {"cache": False}}],
}
}
}
def test_init_with_no_defaults_interactive(tmp_dir, dvc):
inp = io.StringIO(
"python script.py\n"
"script.py\n"
"data\n"
"model\n"
"n\n"
"metric\n"
"n\n"
)
init(
dvc,
defaults={},
overrides={"cmd": "python script.py"},
interactive=True,
stream=inp,
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"default": {
"cmd": "python script.py",
"deps": ["python script.py", "script.py"],
"metrics": [{"metric": {"cache": False}}],
"outs": ["data"],
}
}
}
def test_init_interactive_params_validation(tmp_dir, dvc, capsys):
tmp_dir.gen({"data": {"foo": "foo"}})
(tmp_dir / "params.yaml").dump({"foo": 1})
inp = io.StringIO(
"python script.py\nscript.py\ndata\nmodels\nparams.json\ndata\n"
)
init(
dvc, stream=inp, interactive=True, defaults=CmdExperimentsInit.DEFAULTS
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"default": {
"cmd": "python script.py",
"deps": ["data", "script.py"],
"metrics": [{"metrics.json": {"cache": False}}],
"outs": ["models"],
"params": ["foo"],
"plots": [{"plots": {"cache": False}}],
}
}
}
out, err = capsys.readouterr()
assert (
"Path to a parameters file [params.yaml, n to omit]: "
"'params.json' does not exist. "
"Please retry with an existing parameters file.\n"
"Path to a parameters file [params.yaml, n to omit]: "
"'data' is a directory. "
"Please retry with an existing parameters file.\n"
"Path to a parameters file [params.yaml, n to omit]:"
) in err
assert not out
def test_init_interactive_when_no_path_prompts_need_to_be_asked(
tmp_dir, dvc, extra_overrides, inp
):
"""When we pass everything that's required of, it should not prompt us."""
(tmp_dir / "params.yaml").dump({"foo": 1})
init(
dvc,
interactive=True,
defaults=CmdExperimentsInit.DEFAULTS,
overrides={**CmdExperimentsInit.DEFAULTS, **extra_overrides},
stream=inp, # we still need to confirm
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"default": {
"cmd": "cmd",
"deps": ["data", "src"],
"live": {"dvclive": {"html": True, "summary": True}},
"metrics": [{"metrics.json": {"cache": False}}],
# we specify `live` through `overrides`,
# so it creates checkpoint-based output.
"outs": [{"models": {"checkpoint": True}}],
"params": ["foo"],
"plots": [{"plots": {"cache": False}}],
}
}
}
def test_init_with_no_defaults_non_interactive(tmp_dir, scm, dvc):
init(dvc, defaults={}, overrides={"cmd": "python script.py"})
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {"train": {"cmd": "python script.py"}}
}
scm._reset()
assert not (tmp_dir / "dvc.lock").exists()
assert scm.is_tracked("dvc.yaml")
def test_with_a_custom_name(tmp_dir, dvc):
init(dvc, name="custom", overrides={"cmd": "cmd"})
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"custom": {
"cmd": "cmd"
}
}
}
def test_init_with_live_and_metrics_plots_provided(tmp_dir, dvc, interactive,
inp):
(tmp_dir / "params.yaml").dump({"foo": 1})
init(
dvc,
interactive=interactive,
stream=inp,
defaults=CmdExperimentsInit.DEFAULTS,
overrides={
"cmd": "cmd",
"live": "live",
"metrics": "metrics.json",
"plots": "plots",
},
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"train": {
"cmd":
"cmd",
"deps": ["data", "src"],
"metrics": [
{
"live.json": {
"cache": False
}
},
{
"metrics.json": {
"cache": False
}
},
],
"outs": ["models"],
"params": [{
"params.yaml": None
}],
"plots": [
{
os.path.join("live", "scalars"): {
"cache": False
}
},
{
"plots": {
"cache": False
}
},
],
}
}
}
assert (tmp_dir / "src").is_dir()
assert (tmp_dir / "data").is_dir()
def test_creates_params_file_by_default(tmp_dir, dvc, interactive):
init(
dvc,
interactive=interactive,
defaults=CmdExperimentsInit.DEFAULTS,
overrides={"cmd": "cmd"},
stream=io.StringIO(""),
)
assert (tmp_dir / "params.yaml").is_file()
assert (tmp_dir / "params.yaml").parse() == {}
def test_when_stage_already_exists_with_same_name(tmp_dir, dvc, interactive):
(tmp_dir / "dvc.yaml").dump({"stages": {"train": {"cmd": "test"}}})
with pytest.raises(DuplicateStageName) as exc:
init(
dvc,
interactive=interactive,
overrides={"cmd": "true"},
defaults=CmdExperimentsInit.DEFAULTS,
)
assert (str(exc.value) == "Stage 'train' already exists in 'dvc.yaml'. "
"Use '--force' to overwrite.")
def test_when_stage_force_if_already_exists(tmp_dir, dvc):
(tmp_dir / "params.yaml").dump({"foo": 1})
(tmp_dir / "dvc.yaml").dump({"stages": {"train": {"cmd": "test"}}})
init(
dvc,
force=True,
overrides={"cmd": "true"},
defaults=CmdExperimentsInit.DEFAULTS,
)
d = (tmp_dir / "dvc.yaml").parse()
assert d["stages"]["train"]["cmd"] == "true"
def test_abort_confirmation(tmp_dir, dvc):
(tmp_dir / "param").dump({"foo": 1})
inp = io.StringIO("./script\nscript\ndata\nmodel\nparam\nmetric\nplt\nn")
with pytest.raises(DvcException) as exc:
init(
dvc,
interactive=True,
defaults=CmdExperimentsInit.DEFAULTS,
stream=inp,
)
assert str(exc.value) == "Aborting ..."
assert not (tmp_dir / "dvc.yaml").exists()
assert not (tmp_dir / "dvc.lock").exists()
def test_init_interactive_live(tmp_dir, scm, dvc, interactive, overrides, inp,
capsys):
overrides["live"] = "dvclive"
(tmp_dir / "params.yaml").dump({"foo": {"bar": 1}})
init(
dvc,
interactive=interactive,
defaults=CmdExperimentsInit.DEFAULTS,
overrides=overrides,
stream=inp,
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"train": {
"cmd": "python script.py",
"deps": ["data", "script.py"],
"metrics": [{
"dvclive.json": {
"cache": False
}
}],
"outs": ["models"],
"params": [{
"params.yaml": None
}],
"plots": [{
os.path.join("dvclive", "scalars"): {
"cache": False
}
}],
}
}
}
assert not (tmp_dir / "dvc.lock").exists()
assert (tmp_dir / "script.py").read_text() == ""
assert (tmp_dir / "data").is_dir()
scm._reset()
assert scm.is_tracked("dvc.yaml")
assert scm.is_tracked("params.yaml")
assert scm.is_tracked(".gitignore")
assert scm.is_ignored("models")
out, err = capsys.readouterr()
assert not out
if interactive:
assert "'script.py' does not exist, the file will be created." in err
assert "'data' does not exist, the directory will be created." in err
def run(self):
from dvc.command.stage import parse_cmd
cmd = parse_cmd(self.args.command)
if not self.args.interactive and not cmd:
raise InvalidArgumentError("command is not specified")
from dvc.repo.experiments.init import init
defaults = {}
if not self.args.explicit:
config = self.repo.config["exp"]
defaults.update({**self.DEFAULTS, **config})
cli_args = compact({
"cmd": cmd,
"code": self.args.code,
"data": self.args.data,
"models": self.args.models,
"metrics": self.args.metrics,
"params": self.args.params,
"plots": self.args.plots,
"live": self.args.live,
})
initialized_stage = init(
self.repo,
name=self.args.name,
type=self.args.type,
defaults=defaults,
overrides=cli_args,
interactive=self.args.interactive,
force=self.args.force,
)
text = ui.rich_text.assemble(
"\n" if self.args.interactive else "",
"Created ",
(self.args.name, "bright_blue"),
" stage in ",
("dvc.yaml", "green"),
".",
)
if not self.args.run:
text.append_text(
ui.rich_text.assemble(
" To run, use ",
('"dvc exp run"', "green"),
".\nSee ",
(self.EXP_LINK, "repr.url"),
".",
))
ui.write(text, styled=True)
if self.args.run:
return self.repo.experiments.run(
targets=[initialized_stage.addressing])
return 0
def test_gen_output_dirs(tmp_dir, dvc):
init(
dvc,
defaults=CmdExperimentsInit.DEFAULTS,
overrides={
"cmd": "cmd",
"models": "models/predict.h5",
"metrics": "eval/scores.json",
"plots": "eval/plots",
"live": "eval/live",
},
)
assert (tmp_dir / "models").is_dir()
assert (tmp_dir / "eval").is_dir()
assert (tmp_dir / "eval/plots").is_dir()
assert (tmp_dir / "eval/live").is_dir()
assert not (tmp_dir / "models/predict.h5").exists()
assert not (tmp_dir / "eval/scores.json").exists()
def test_init_interactive_when_no_path_prompts_need_to_be_asked(
tmp_dir, dvc, extra_overrides, inp):
"""When we pass everything that's required of, it should not prompt us."""
(tmp_dir / "params.yaml").dump({"foo": 1})
init(
dvc,
interactive=True,
defaults=CmdExperimentsInit.DEFAULTS,
overrides={
**CmdExperimentsInit.DEFAULTS,
**extra_overrides
},
stream=inp,
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"train": {
"cmd": "cmd",
"deps": ["data", "src"],
"metrics": [
{
"metrics.json": {
"cache": False
}
},
],
"outs": ["models"],
"params": [{
"params.yaml": None
}],
"plots": [
{
"plots": {
"cache": False
}
},
],
}
}
}
assert (tmp_dir / "src").is_dir()
assert (tmp_dir / "data").is_dir()
def test_init_interactive_default(
tmp_dir, scm, dvc, interactive, overrides, inp, capsys
):
(tmp_dir / "params.yaml").dump({"foo": {"bar": 1}})
init(
dvc,
interactive=interactive,
defaults=CmdExperimentsInit.DEFAULTS,
overrides=overrides,
stream=inp,
)
assert (tmp_dir / "dvc.yaml").parse() == {
"stages": {
"train": {
"cmd": "python script.py",
"deps": ["data", "script.py"],
"metrics": [{"metrics.json": {"cache": False}}],
"outs": ["models"],
"params": ["foo"],
"plots": [{"plots": {"cache": False}}],
}
}
}
assert not (tmp_dir / "dvc.lock").exists()
scm._reset()
assert scm.is_tracked("dvc.yaml")
assert scm.is_tracked("params.yaml")
assert scm.is_tracked(".gitignore")
assert scm.is_ignored("models")
out, err = capsys.readouterr()
if interactive:
assert "'script.py' does not exist in the workspace." in err
assert "'data' does not exist in the workspace." in err
assert not out
def run(self):
from dvc.command.stage import parse_cmd
cmd = parse_cmd(self.args.cmd)
if not self.args.interactive and not cmd:
raise InvalidArgumentError("command is not specified")
from dvc.repo.experiments.init import init
global_defaults = {
"code": self.CODE,
"data": self.DATA,
"models": self.MODELS,
"metrics": self.DEFAULT_METRICS,
"params": self.DEFAULT_PARAMS,
"plots": self.PLOTS,
"live": self.DVCLIVE,
}
defaults = {}
if not self.args.explicit:
config = {} # TODO
defaults.update({**global_defaults, **config})
cli_args = compact({
"cmd": cmd,
"code": self.args.code,
"data": self.args.data,
"models": self.args.models,
"metrics": self.args.metrics,
"params": self.args.params,
"plots": self.args.plots,
"live": self.args.live,
})
initialized_stage = init(
self.repo,
name=self.args.name,
type=self.args.type,
defaults=defaults,
overrides=cli_args,
interactive=self.args.interactive,
force=self.args.force,
)
if self.args.run:
return self.repo.experiments.run(
targets=[initialized_stage.addressing])
return 0
def run(self):
from dvc.commands.stage import parse_cmd
cmd = parse_cmd(self.args.command)
if not self.args.interactive and not cmd:
raise InvalidArgumentError("command is not specified")
from dvc.repo.experiments.init import init
defaults = {}
if not self.args.explicit:
config = self.repo.config["exp"]
defaults.update({**self.DEFAULTS, **config})
cli_args = compact({
"cmd": cmd,
"code": self.args.code,
"data": self.args.data,
"models": self.args.models,
"metrics": self.args.metrics,
"params": self.args.params,
"plots": self.args.plots,
"live": self.args.live,
})
initialized_stage, initialized_deps, initialized_out_dirs = init(
self.repo,
name=self.args.name,
type=self.args.type,
defaults=defaults,
overrides=cli_args,
interactive=self.args.interactive,
force=self.args.force,
)
self._post_init_display(initialized_stage, initialized_deps,
initialized_out_dirs)
if self.args.run:
self.repo.experiments.run(targets=[initialized_stage.addressing])
return 0
