exp show: Drop --pcp. (#9270)

Closes #9268
This commit is contained in:
David de la Iglesia Castro
2023-05-19 15:36:33 +05:45
committed by GitHub
parent cdba65f27e
commit 02f7fcae73
4 changed files with 11 additions and 182 deletions
+11 -66
View File
@@ -1,13 +1,11 @@
import argparse
import logging
import os
import re
from datetime import date, datetime
from typing import TYPE_CHECKING, Dict, Iterable
from funcy import lmap
from dvc.cli import completion
from dvc.cli.command import CmdBase
from dvc.cli.utils import append_doc_link
from dvc.commands.metrics import DEFAULT_PRECISION
@@ -59,7 +57,7 @@ def baseline_styler(typ):
return {"style": "bold"} if typ == "baseline" else {}
def show_experiments( # noqa: C901, PLR0912
def show_experiments( # noqa: C901
td: "TabularData",
headers: Dict[str, Iterable[str]],
keep=None,
@@ -67,7 +65,6 @@ def show_experiments( # noqa: C901, PLR0912
pager=True,
csv=False,
markdown=False,
pcp=False,
**kwargs,
):
if keep:
@@ -110,51 +107,23 @@ def show_experiments( # noqa: C901, PLR0912
}
)
if kwargs.get("only_changed", False) or pcp:
if kwargs.get("only_changed", False):
td.drop_duplicates("cols", ignore_empty=False)
cols_to_drop = set()
if drop is not None:
cols_to_drop = {col for col in td.keys() if re.match(drop, col)} # noqa: SIM118
if pcp:
cols_to_drop.add("Created")
td.drop(*cols_to_drop)
if pcp:
subset = {x for x in td.keys() if x != "Experiment"} # noqa: SIM118
td.dropna(
"rows",
how="all",
subset=subset,
)
td.drop_duplicates("rows", subset=subset)
if "Experiment" in td:
td.column("Experiment")[:] = [
# remove tree characters
str(x).encode("ascii", "ignore").strip().decode()
for x in td.column("Experiment")
]
out = kwargs.get("out") or "dvc_plots"
output_file = os.path.join(out, "index.html")
ui.write(
td.to_parallel_coordinates(
output_path=os.path.abspath(output_file),
color_by=kwargs.get("sort_by") or "Experiment",
)
)
if kwargs.get("open"):
return ui.open_browser(output_file)
else:
td.render(
pager=pager,
borders="horizontals",
rich_table=True,
header_styles=styles,
row_styles=row_styles,
csv=csv,
markdown=markdown,
)
td.render(
pager=pager,
borders="horizontals",
rich_table=True,
header_styles=styles,
row_styles=row_styles,
csv=csv,
markdown=markdown,
)
def _normalize_headers(names, count):
@@ -221,9 +190,6 @@ class CmdExperimentsShow(CmdBase):
csv=self.args.csv,
markdown=self.args.markdown,
only_changed=self.args.only_changed,
pcp=self.args.pcp,
out=self.args.out,
open=self.args.open,
)
return 0
@@ -342,27 +308,6 @@ def add_parser(experiments_subparsers, parent_parser):
),
metavar="<n>",
)
experiments_show_parser.add_argument(
"--parallel-coordinates-plot",
"--pcp",
dest="pcp",
action="store_true",
default=False,
help="Generate a Parallel Coordinates Plot from the tabulated output.",
)
experiments_show_parser.add_argument(
"-o",
"--out",
default=None,
help="Destination folder to save the Parallel Coordinates Plot to",
metavar="<path>",
).complete = completion.DIR
experiments_show_parser.add_argument(
"--open",
action="store_true",
default=False,
help="Open the Parallel Coordinates Plot directly in the browser.",
)
experiments_show_parser.add_argument(
"--no-fetch",
dest="fetch_running",
-19
View File
@@ -22,7 +22,6 @@ from typing import (
from funcy import reraise
if TYPE_CHECKING:
from dvc.types import StrPath
from dvc.ui.table import CellT
@@ -161,24 +160,6 @@ class TabularData(MutableSequence[Sequence["CellT"]]):
writer.writerow(row)
return buff.getvalue()
def to_parallel_coordinates(
self, output_path: "StrPath", color_by: Optional[str] = None
) -> "StrPath":
from dvc_render.html import render_html
from dvc_render.plotly import ParallelCoordinatesRenderer
render_html(
renderers=[
ParallelCoordinatesRenderer(
self.as_dict(),
color_by=color_by,
fill_value=self._fill_value,
)
],
output_file=output_path,
)
return output_path
def add_column(self, name: str) -> None:
self._columns[name] = Column([self._fill_value] * len(self))
self._keys.append(name)
-79
View File
@@ -551,85 +551,6 @@ def test_show_only_changed(tmp_dir, dvc, scm, capsys, copy_script):
assert "metrics.yaml:goobar" in cap.out
def test_show_parallel_coordinates(tmp_dir, dvc, scm, mocker, capsys, copy_script):
from dvc.commands.experiments import show
webbroser_open = mocker.patch("webbrowser.open")
show_experiments = mocker.spy(show, "show_experiments")
params_file = tmp_dir / "params.yaml"
params_data = {
"foo": 1,
"bar": 1,
}
(tmp_dir / params_file).dump(params_data)
dvc.run(
cmd="python copy.py params.yaml metrics.yaml",
metrics_no_cache=["metrics.yaml"],
params=["foo", "bar"],
name="copy-file",
deps=["copy.py"],
)
scm.add(
[
"dvc.yaml",
"dvc.lock",
"copy.py",
"params.yaml",
"metrics.yaml",
".gitignore",
]
)
scm.commit("init")
dvc.experiments.run(params=["foo=2"])
assert main(["exp", "show", "--pcp"]) == 0
kwargs = show_experiments.call_args[1]
html_text = (tmp_dir / "dvc_plots" / "index.html").read_text()
assert all(rev in html_text for rev in ["workspace", "master"])
assert "[exp-" not in html_text
assert '{"label": "metrics.yaml:foo", "values": [2.0, 1.0]}' in html_text
assert '{"label": "params.yaml:foo", "values": [2.0, 1.0]}' in html_text
assert '"line": {"color": [1, 0]' in html_text
assert '"label": "metrics.yaml:bar"' not in html_text
assert '"label": "Created"' not in html_text
assert main(["exp", "show", "--pcp", "--sort-by", "metrics.yaml:foo"]) == 0
kwargs = show_experiments.call_args[1]
html_text = (tmp_dir / "dvc_plots" / "index.html").read_text()
assert '"line": {"color": [2.0, 1.0]' in html_text
assert main(["exp", "show", "--pcp", "--out", "experiments"]) == 0
kwargs = show_experiments.call_args[1]
assert kwargs["out"] == "experiments"
assert (tmp_dir / "experiments" / "index.html").exists()
assert main(["exp", "show", "--pcp", "--open"]) == 0
webbroser_open.assert_called()
params_data = {"foo": 1, "bar": 1, "foobar": 2}
(tmp_dir / params_file).dump(params_data)
assert main(["exp", "show", "--pcp"]) == 0
html_text = (tmp_dir / "dvc_plots" / "index.html").read_text()
assert '{"label": "foobar", "values": [2.0, null, null]}' in html_text
assert main(["exp", "show", "--pcp", "--drop", "foobar"]) == 0
html_text = (tmp_dir / "dvc_plots" / "index.html").read_text()
assert '"label": "Created"' not in html_text
assert '"label": "foobar"' not in html_text
assert main(["exp", "show", "--pcp", "--drop", "Experiment"]) == 0
html_text = (tmp_dir / "dvc_plots" / "index.html").read_text()
assert '"label": "Experiment"' not in html_text
@pytest.mark.vscode
def test_show_outs(tmp_dir, dvc, scm, erepo_dir, copy_script):
params_file = tmp_dir / "params.yaml"
-18
View File
@@ -374,21 +374,3 @@ def test_drop_duplicates_invalid_axis():
with pytest.raises(ValueError, match="Invalid 'axis' value foo."):
td.drop_duplicates("foo")
def test_to_parallel_coordinates(tmp_dir, mocker):
(tmp_dir / "foo").mkdir()
td = TabularData(["categorical", "scalar"])
td.extend([["foo", "0.1"], ["bar", "2"]])
write = mocker.patch("dvc_render.html.render_html")
renderer_class = mocker.patch("dvc_render.plotly.ParallelCoordinatesRenderer")
renderer = renderer_class.return_value
td.to_parallel_coordinates(output_path="foo")
renderer_class.assert_called_with(
td.as_dict(), color_by=None, fill_value=td._fill_value
)
write.assert_called_with(renderers=[renderer], output_file="foo")