def test_disconnected_stage(tmp_dir, dvc):
tmp_dir.dvc_gen({"base": "base"})
dvc.add("base")
dvc.run(
deps=["base"],
outs=["derived1"],
cmd="echo derived1 > derived1",
single_stage=True,
)
dvc.run(
deps=["base"],
outs=["derived2"],
cmd="echo derived2 > derived2",
single_stage=True,
)
final_stage = dvc.run(
deps=["derived1"],
outs=["final"],
cmd="echo final > final",
single_stage=True,
)
command = CmdPipelineShow([])
# Need to test __build_graph directly
nodes, edges, is_tree = command._build_graph(final_stage.path,
commands=False,
outs=True)
assert set(nodes) == {"final", "derived1", "base"}
assert edges == [("final", "derived1"), ("derived1", "base")]
assert is_tree is True
def test_split_pipeline(tmp_dir, scm, dvc):
tmp_dir.scm_gen("git_dep1", "git_dep1")
tmp_dir.scm_gen("git_dep2", "git_dep2")
tmp_dir.dvc_gen("data", "source file content")
dvc.run(
deps=["git_dep1", "data"],
outs=["data_train", "data_valid"],
cmd="echo train >> data_train && echo valid >> data_valid",
single_stage=True,
)
stage = dvc.run(
deps=["git_dep2", "data_train", "data_valid"],
outs=["result"],
cmd="echo result >> result",
single_stage=True,
)
command = CmdPipelineShow([])
nodes, edges, _ = command._build_graph(stage.path,
commands=False,
outs=True)
assert set(nodes) == {"data", "data_train", "data_valid", "result"}
assert set(edges) == {
("result", "data_train"),
("result", "data_valid"),
("data_train", "data"),
("data_valid", "data"),
}
def test_pipeline_list_show_multistage(tmp_dir, dvc, run_copy, caplog):
tmp_dir.gen("foo", "foo")
run_copy("foo", "bar", name="copy-foo-bar")
run_copy("bar", "foobar", single_stage=True)
command = CmdPipelineShow([])
caplog.clear()
with caplog.at_level(logging.INFO, "dvc"):
command._show("foobar.dvc", False, False, False)
output = caplog.text.splitlines()
assert "copy-foo-bar" in output[0]
assert "foobar.dvc" in output[1]
caplog.clear()
with caplog.at_level(logging.INFO, "dvc"):
command._show("copy-foo-bar", False, False, False)
assert "copy-foo-bar" in caplog.text
assert "foobar.dvc" not in caplog.text
command = CmdPipelineList([])
caplog.clear()
with caplog.at_level(logging.INFO, "dvc"):
command.run()
assert "copy-foo-bar" in caplog.text
assert "foobar.dvc" in caplog.text
assert "1 pipelines in total"
def test_pipeline_ascii_multistage(tmp_dir, dvc, run_copy):
tmp_dir.gen("foo", "foo")
run_copy("foo", "bar", name="copy-foo-bar")
run_copy("bar", "foobar", single_stage=True)
command = CmdPipelineShow([])
nodes, edges, _ = command._build_graph("foobar.dvc")
assert set(nodes) == {"copy-foo-bar", "foobar.dvc"}
assert set(edges) == {
("foobar.dvc", "copy-foo-bar"),
}
nodes, *_ = command._build_graph("copy-foo-bar")
assert set(nodes) == {"copy-foo-bar"}
def test_pipeline_ascii_multistage(tmp_dir, dvc, run_copy):
tmp_dir.gen("foo", "foo")
run_copy("foo", "bar", name="copy-foo-bar")
run_copy("bar", "foobar")
command = CmdPipelineShow([])
nodes, edges, is_tree = command._build_graph("foobar.dvc")
assert set(nodes) == {"pipelines.yaml:copy-foo-bar", "foobar.dvc"}
assert set(edges) == {
("foobar.dvc", "pipelines.yaml:copy-foo-bar"),
}
nodes, edges, is_tree = command._build_graph("pipelines.yaml:copy-foo-bar")
assert set(nodes) == {"pipelines.yaml:copy-foo-bar"}
def test_pipeline_multi_outputs_stages(dvc):
dvc.run(
outs=["alice", "bob"],
cmd="echo alice>alice && echo bob>bob",
single_stage=True,
)
dvc.run(
deps=["alice"],
outs=["mary", "mike"],
cmd="echo mary>mary && echo mike>mike",
single_stage=True,
)
stage = dvc.run(
deps=["mary"],
outs=["carol"],
cmd="echo carol>carol",
single_stage=True,
)
command = CmdPipelineShow([])
nodes, edges, _ = command._build_graph(stage.path, outs=True)
assert set(nodes) == {"alice", "mary", "carol"}
assert set(edges) == {("carol", "mary"), ("mary", "alice")}