def test_plot_choose_columns(tmp_dir, scm, dvc, custom_template,
run_copy_metrics):
metric = [{"a": 1, "b": 2, "c": 3}, {"a": 2, "b": 3, "c": 4}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="init",
tag="v1",
)
props = {
"template": os.fspath(custom_template),
"fields": {"b", "c"},
"x": "b",
"y": "c",
}
plot_string = dvc.plots.show(props=props)["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{
"b": 2,
"c": 3,
"rev": "workspace"
},
{
"b": 3,
"c": 4,
"rev": "workspace"
},
]
assert plot_content["encoding"]["x"]["field"] == "b"
assert plot_content["encoding"]["y"]["field"] == "c"
def test_show_non_plot_and_plot_with_params(tmp_dir, scm, dvc,
run_copy_metrics):
metric = [
{
"first_val": 100,
"val": 2
},
{
"first_val": 200,
"val": 3
},
]
_write_json(tmp_dir, metric, "metric.json")
run_copy_metrics("metric.json",
"metric2.json",
plots_no_cache=["metric2.json"])
dvc.plots.modify("metric2.json", props={"title": "TITLE"})
result = dvc.plots.show(targets=["metric.json", "metric2.json"])
plot_content = json.loads(result["metric.json"])
plot2_content = json.loads(result["metric2.json"])
assert plot2_content["title"] == "TITLE"
assert plot_content != plot2_content
plot_content.pop("title")
plot2_content.pop("title")
assert plot_content == plot2_content
def test_plot_json_single_val(tmp_dir, scm, dvc, run_copy_metrics):
metric = [{"val": 2}, {"val": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="first run",
)
plot_string = dvc.plots.show()["metric.json"]
plot_json = json.loads(plot_string)
assert plot_json["data"]["values"] == [
{
"val": 2,
PlotData.INDEX_FIELD: 0,
"rev": "workspace"
},
{
"val": 3,
PlotData.INDEX_FIELD: 1,
"rev": "workspace"
},
]
assert plot_json["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_json["encoding"]["y"]["field"] == "val"
def test_plot_even_if_metric_missing(tmp_dir, scm, dvc, caplog,
run_copy_metrics):
tmp_dir.scm_gen("some_file", "content", commit="there is no metric")
scm.tag("v1")
metric = [{"y": 2}, {"y": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="there is metric",
tag="v2",
)
caplog.clear()
with caplog.at_level(logging.WARNING, "dvc"):
plots = dvc.plots.show(revs=["v1", "v2"], targets=["metric.json"])
assert "'metric.json' was not found at: 'v1'." in caplog.text
plot_content = json.loads(plots["metric.json"])
assert plot_content["data"]["values"] == [
{
"y": 2,
PlotData.INDEX_FIELD: 0,
"rev": "v2"
},
{
"y": 3,
PlotData.INDEX_FIELD: 1,
"rev": "v2"
},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "y"
def test_plot_default_choose_column(tmp_dir, scm, dvc, run_copy_metrics):
metric = [{"a": 1, "b": 2, "c": 3}, {"a": 2, "b": 3, "c": 4}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="init",
tag="v1",
)
plot_string = dvc.plots.show(props={"fields": {"b"}})["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{
PlotData.INDEX_FIELD: 0,
"b": 2,
"rev": "workspace"
},
{
PlotData.INDEX_FIELD: 1,
"b": 3,
"rev": "workspace"
},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "b"
def test_diff_no_file_on_target_rev(tmp_dir, scm, dvc, caplog, diff_fun,
metric_value):
with tmp_dir.branch("new_branch", new=True):
_write_json(tmp_dir, metric_value, "metric.json")
with caplog.at_level(logging.WARNING, "dvc"):
diff_fun(dvc, "metric.json", "master")
assert "'metric.json' was not found at: 'master'." in caplog.text
def test_diff_no_file_on_target_rev(tmp_dir, scm, dvc, caplog, command,
metric_value):
with tmp_dir.branch("new_branch", new=True):
_write_json(tmp_dir, metric_value, "metric.json")
with caplog.at_level(logging.WARNING, "dvc"):
assert (main(
[command, "diff", "master", "--targets", "metric.json"]) == 0)
assert "'metric.json' was not found at: 'master'." in caplog.text
def test_should_raise_on_no_template(tmp_dir, dvc, run_copy_metrics):
metric = [{"val": 2}, {"val": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="first run",
)
with pytest.raises(TemplateNotFoundError):
props = {"template": "non_existing_template.json"}
dvc.plots.show("metric.json", props=props)
def test_plots_modify_existing_template(tmp_dir, dvc, run_copy_metrics,
custom_template):
metric = [{"a": 1, "b": 2}, {"a": 2, "b": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
stage = run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
name="copy-metrics",
single_stage=False,
)
dvc.plots.modify("metric.json",
props={"template": relpath(custom_template)})
stage = stage.reload()
assert stage.outs[0].plot == {"template": relpath(custom_template)}
def test_raise_on_wrong_field(tmp_dir, scm, dvc, run_copy_metrics):
metric = [{"val": 2}, {"val": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="first run",
)
with pytest.raises(NoFieldInDataError):
dvc.plots.show("metric.json", props={"x": "no_val"})
with pytest.raises(NoFieldInDataError):
dvc.plots.show("metric.json", props={"y": "no_val"})
def test_bad_template(tmp_dir, dvc, run_copy_metrics):
metric = [{"val": 2}, {"val": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="first run",
)
tmp_dir.gen("template.json", json.dumps({"a": "b", "c": "d"}))
with pytest.raises(BadTemplateError):
props = {"template": "template.json"}
dvc.plots.show("metric.json", props=props)
def test_unset_nonexistent(tmp_dir, dvc, run_copy_metrics, custom_template):
metric = [{"a": 1, "b": 2}, {"a": 2, "b": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
name="copy-metrics",
single_stage=False,
)
with pytest.raises(PropsNotFoundError):
dvc.plots.modify(
"metric.json",
unset=["nonexistent"],
)
def test_plots_modify_should_not_change_lockfile(tmp_dir, dvc,
run_copy_metrics,
custom_template):
_write_json(tmp_dir, [{"a": 1, "b": 2}], "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
name="copy-metrics",
single_stage=False,
)
(tmp_dir / PIPELINE_LOCK).unlink()
dvc.plots.modify("metric.json",
props={"template": relpath(custom_template)})
assert not (tmp_dir / PIPELINE_LOCK).exists()
def test_show_no_repo(tmp_dir):
metric = [
{
"first_val": 100,
"val": 2
},
{
"first_val": 200,
"val": 3
},
]
_write_json(tmp_dir, metric, "metric.json")
dvc = Repo(uninitialized=True)
dvc.plots.show(["metric.json"])
def test_plot_confusion_normalized(tmp_dir, dvc, run_copy_metrics):
confusion_matrix = [
{
"predicted": "B",
"actual": "A"
},
{
"predicted": "A",
"actual": "A"
},
]
_write_json(tmp_dir, confusion_matrix, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="first run",
)
props = {
"template": "confusion_normalized",
"x": "predicted",
"y": "actual",
}
plot_string = dvc.plots.show(props=props)["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{
"predicted": "B",
"actual": "A",
"rev": "workspace"
},
{
"predicted": "A",
"actual": "A",
"rev": "workspace"
},
]
assert plot_content["spec"]["transform"][0]["groupby"] == [
"actual",
"predicted",
]
assert plot_content["spec"]["transform"][1]["groupby"] == ["rev", "actual"]
assert plot_content["spec"]["encoding"]["x"]["field"] == "predicted"
assert plot_content["spec"]["encoding"]["y"]["field"] == "actual"
def test_multiple_plots(tmp_dir, scm, dvc, run_copy_metrics):
metric1 = [
OrderedDict([("first_val", 100), ("second_val", 100), ("val", 2)]),
OrderedDict([("first_val", 200), ("second_val", 300), ("val", 3)]),
]
metric2 = [
OrderedDict([("first_val", 100), ("second_val", 100), ("val", 2)]),
OrderedDict([("first_val", 200), ("second_val", 300), ("val", 3)]),
]
_write_csv(metric1, "metric_t1.csv")
_write_json(tmp_dir, metric2, "metric_t2.json")
run_copy_metrics("metric_t1.csv",
"metric1.csv",
plots_no_cache=["metric1.csv"])
run_copy_metrics("metric_t2.json",
"metric2.json",
plots_no_cache=["metric2.json"])
assert len(dvc.plots.show().keys()) == 2
def test_show_from_subdir(tmp_dir, dvc, caplog):
subdir = tmp_dir / "subdir"
subdir.mkdir()
metric = [
{
"first_val": 100,
"val": 2
},
{
"first_val": 200,
"val": 3
},
]
_write_json(subdir, metric, "metric.json")
with subdir.chdir(), caplog.at_level(logging.INFO, "dvc"):
assert main(["plots", "show", "metric.json"]) == 0
assert f"file://{str(subdir)}" in caplog.text
assert (subdir / "plots.html").exists()
def test_diff_dirty(tmp_dir, scm, dvc, run_copy_metrics):
metric_1 = [{"y": 2}, {"y": 3}]
_write_json(tmp_dir, metric_1, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="init",
)
metric_2 = [{"y": 3}, {"y": 5}]
_write_json(tmp_dir, metric_2, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="second",
)
metric_3 = [{"y": 5}, {"y": 6}]
_write_json(tmp_dir, metric_3, "metric_t.json")
run_copy_metrics(
"metric_t.json", "metric.json", plots_no_cache=["metric.json"]
)
plot_string = dvc.plots.diff(props={"fields": {"y"}})["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{"y": 3, PlotData.INDEX_FIELD: 0, "rev": "HEAD"},
{"y": 5, PlotData.INDEX_FIELD: 1, "rev": "HEAD"},
{"y": 5, PlotData.INDEX_FIELD: 0, "rev": "workspace"},
{"y": 6, PlotData.INDEX_FIELD: 1, "rev": "workspace"},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "y"
_write_json(tmp_dir, [{"y": 7}, {"y": 8}], "metric.json")
plot_string = dvc.plots.diff(props={"fields": {"y"}})["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{"y": 3, PlotData.INDEX_FIELD: 0, "rev": "HEAD"},
{"y": 5, PlotData.INDEX_FIELD: 1, "rev": "HEAD"},
{"y": 7, PlotData.INDEX_FIELD: 0, "rev": "workspace"},
{"y": 8, PlotData.INDEX_FIELD: 1, "rev": "workspace"},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "y"
def test_show_non_plot(tmp_dir, scm, use_dvc):
metric = [
{
"first_val": 100,
"val": 2
},
{
"first_val": 200,
"val": 3
},
]
_write_json(tmp_dir, metric, "metric.json")
if use_dvc:
dvc = Repo.init()
else:
dvc = Repo(uninitialized=True)
plot_string = dvc.plots.show(targets=["metric.json"])["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{
"val": 2,
PlotData.INDEX_FIELD: 0,
"first_val": 100,
"rev": "workspace",
},
{
"val": 3,
PlotData.INDEX_FIELD: 1,
"first_val": 200,
"rev": "workspace",
},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "val"
if not use_dvc:
assert not (tmp_dir / ".dvc").exists()
def test_plot_cache_missing(tmp_dir, scm, dvc, caplog, run_copy_metrics):
metric = [{"y": 2}, {"y": 3}]
_write_json(tmp_dir, metric, "metric_t.json")
stage = run_copy_metrics(
"metric_t.json",
"metric.json",
plots=["metric.json"],
commit="there is metric",
)
scm.tag("v1")
# Make a different plot and then remove its datafile
metric = [{"y": 3}, {"y": 4}]
_write_json(tmp_dir, metric, "metric_t.json")
stage = run_copy_metrics(
"metric_t.json",
"metric.json",
plots=["metric.json"],
commit="there is an another metric",
)
scm.tag("v2")
remove(stage.outs[0].fspath)
remove(stage.outs[0].cache_path)
plots = dvc.plots.show(revs=["v1", "v2"], targets=["metric.json"])
plot_content = json.loads(plots["metric.json"])
assert plot_content["data"]["values"] == [
{
"y": 2,
PlotData.INDEX_FIELD: 0,
"rev": "v1"
},
{
"y": 3,
PlotData.INDEX_FIELD: 1,
"rev": "v1"
},
]
def test_plot_multiple_revs_default(tmp_dir, scm, dvc, run_copy_metrics):
metric_1 = [{"y": 2}, {"y": 3}]
_write_json(tmp_dir, metric_1, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="init",
tag="v1",
)
metric_2 = [{"y": 3}, {"y": 5}]
_write_json(tmp_dir, metric_2, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="second",
tag="v2",
)
metric_3 = [{"y": 5}, {"y": 6}]
_write_json(tmp_dir, metric_3, "metric_t.json")
run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="third",
)
plot_string = dvc.plots.show(revs=["HEAD", "v2", "v1"],
props={"fields": {"y"}})["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{
"y": 5,
PlotData.INDEX_FIELD: 0,
"rev": "HEAD"
},
{
"y": 6,
PlotData.INDEX_FIELD: 1,
"rev": "HEAD"
},
{
"y": 3,
PlotData.INDEX_FIELD: 0,
"rev": "v2"
},
{
"y": 5,
PlotData.INDEX_FIELD: 1,
"rev": "v2"
},
{
"y": 2,
PlotData.INDEX_FIELD: 0,
"rev": "v1"
},
{
"y": 3,
PlotData.INDEX_FIELD: 1,
"rev": "v1"
},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "y"
def test_plot_multiple_revs(tmp_dir, scm, dvc, run_copy_metrics):
templates_dir = dvc.plots.templates.templates_dir
shutil.copy(
os.path.join(templates_dir, "default.json"),
os.path.join(templates_dir, "template.json"),
)
metric_1 = [{"y": 2}, {"y": 3}]
_write_json(tmp_dir, metric_1, "metric_t.json")
stage = run_copy_metrics(
"metric_t.json",
"metric.json",
plots_no_cache=["metric.json"],
commit="init",
tag="v1",
)
metric_2 = [{"y": 3}, {"y": 5}]
_write_json(tmp_dir, metric_2, "metric_t.json")
assert dvc.reproduce(stage.addressing) == [stage]
scm.add(["metric.json", stage.path])
scm.commit("second")
scm.tag("v2")
metric_3 = [{"y": 5}, {"y": 6}]
_write_json(tmp_dir, metric_3, "metric_t.json")
assert dvc.reproduce(stage.addressing) == [stage]
scm.add(["metric.json", stage.path])
scm.commit("third")
props = {"template": "template.json"}
plot_string = dvc.plots.show(revs=["HEAD", "v2", "v1"],
props=props)["metric.json"]
plot_content = json.loads(plot_string)
assert plot_content["data"]["values"] == [
{
"y": 5,
PlotData.INDEX_FIELD: 0,
"rev": "HEAD"
},
{
"y": 6,
PlotData.INDEX_FIELD: 1,
"rev": "HEAD"
},
{
"y": 3,
PlotData.INDEX_FIELD: 0,
"rev": "v2"
},
{
"y": 5,
PlotData.INDEX_FIELD: 1,
"rev": "v2"
},
{
"y": 2,
PlotData.INDEX_FIELD: 0,
"rev": "v1"
},
{
"y": 3,
PlotData.INDEX_FIELD: 1,
"rev": "v1"
},
]
assert plot_content["encoding"]["x"]["field"] == PlotData.INDEX_FIELD
assert plot_content["encoding"]["y"]["field"] == "y"
