get_random_exp_name: Use custom random generator. (backport #9490) (#9491)

get_random_exp_name: Use custom random generator.

See https://github.com/iterative/dvclive/pull/576

Co-authored-by: Ivan Shcheklein <[email protected]>
(cherry picked from commit eb45cda5d0)

Co-authored-by: daavoo <[email protected]>

dvc/commands/stage.py

@@ -1,5 +1,6 @@
 import argparse
 import logging
+from contextlib import contextmanager
 from itertools import chain, filterfalse
 from typing import TYPE_CHECKING, Dict, Iterable, List
 
@@ -119,8 +120,22 @@ def parse_cmd(commands: List[str]) -> str:
     return " ".join(map(quote_argument, commands))
 
 
+@contextmanager
+def _disable_logging(highest_level=logging.CRITICAL):
+    previous_level = logging.root.manager.disable
+
+    logging.disable(highest_level)
+
+    try:
+        yield
+    finally:
+        logging.disable(previous_level)
+
+
 class CmdStageAdd(CmdBase):
     def run(self):
+        from dvc.repo import lock_repo
+
         kwargs = vars(self.args)
         kwargs.update(
             {
@@ -128,7 +143,15 @@ class CmdStageAdd(CmdBase):
                 "params": parse_params(self.args.params),
             }
         )
-        self.repo.stage.add(**kwargs)
+
+        with self.repo.scm_context, lock_repo(self.repo):
+            with _disable_logging(logging.INFO):
+                stage = self.repo.stage.add(**kwargs)
+            logger.info("Added stage %r in %r", stage.addressing, stage.relpath)
+            if self.args.run:
+                stage.run()
+                stage.dump(update_pipeline=False)
+
         return 0
 
 
@@ -257,6 +280,12 @@ def _add_common_args(parser):
             "This doesn't affect any DVC operations."
         ),
     )
+    parser.add_argument(
+        "--run",
+        action="store_true",
+        default=False,
+        help="Execute the stage after generating it.",
+    )
     parser.add_argument(
         "command",
         nargs=argparse.REMAINDER,

dvc/repo/experiments/ls.py

@@ -1,6 +1,6 @@
 import logging
 from collections import defaultdict
-from typing import Optional
+from typing import List, Optional, Union
 
 from dvc.repo import locked
 from dvc.repo.scm_context import scm_context
@@ -15,14 +15,17 @@ logger = logging.getLogger(__name__)
 @scm_context
 def ls(
     repo,
-    rev: Optional[str] = None,
+    rev: Optional[Union[List[str], str]] = None,
     all_commits: bool = False,
     num: int = 1,
     git_remote: Optional[str] = None,
 ):
     rev_set = None
     if not all_commits:
-        revs = iter_revs(repo.scm, [rev or "HEAD"], num)
+        rev = rev or "HEAD"
+        if isinstance(rev, str):
+            rev = [rev]
+        revs = iter_revs(repo.scm, rev, num)
         rev_set = set(revs.keys())
     ref_info_dict = exp_refs_by_baseline(repo.scm, rev_set, git_remote)
 

dvc/repo/experiments/pull.py

@@ -23,7 +23,7 @@ def pull(  # noqa: C901
     git_remote: str,
     exp_names: Union[Iterable[str], str],
     all_commits=False,
-    rev: Optional[str] = None,
+    rev: Optional[Union[List[str], str]] = None,
     num=1,
     force: bool = False,
     pull_cache: bool = False,
@@ -49,7 +49,9 @@ def pull(  # noqa: C901
                 raise UnresolvedExpNamesError(unresolved_exp_names)
 
         if rev:
-            rev_dict = iter_revs(repo.scm, [rev], num)
+            if isinstance(rev, str):
+                rev = [rev]
+            rev_dict = iter_revs(repo.scm, rev, num)
             rev_set = set(rev_dict.keys())
             ref_info_dict = exp_refs_by_baseline(repo.scm, rev_set, git_remote)
             for _, ref_info_list in ref_info_dict.items():

dvc/repo/experiments/remove.py

@@ -24,7 +24,7 @@ logger = logging.getLogger(__name__)
 def remove(  # noqa: C901, PLR0912
     repo: "Repo",
     exp_names: Union[None, str, List[str]] = None,
-    rev: Optional[str] = None,
+    rev: Optional[Union[List[str], str]] = None,
     all_commits: bool = False,
     num: int = 1,
     queue: bool = False,
@@ -60,6 +60,8 @@ def remove(  # noqa: C901, PLR0912
         if remained:
             raise UnresolvedExpNamesError(remained, git_remote=git_remote)
     elif rev:
+        if isinstance(rev, str):
+            rev = [rev]
         exp_ref_dict = _resolve_exp_by_baseline(repo, rev, num, git_remote)
         removed.extend(exp_ref_dict.keys())
         exp_ref_list.extend(exp_ref_dict.values())
@@ -85,14 +87,14 @@ def remove(  # noqa: C901, PLR0912
 
 def _resolve_exp_by_baseline(
     repo: "Repo",
-    rev: str,
+    rev: List[str],
     num: int,
     git_remote: Optional[str] = None,
 ) -> Dict[str, "ExpRefInfo"]:
     assert isinstance(repo.scm, Git)
 
     commit_ref_dict: Dict[str, "ExpRefInfo"] = {}
-    rev_dict = iter_revs(repo.scm, [rev], num)
+    rev_dict = iter_revs(repo.scm, rev, num)
     rev_set = set(rev_dict.keys())
     ref_info_dict = exp_refs_by_baseline(repo.scm, rev_set, git_remote)
     for _, ref_info_list in ref_info_dict.items():

dvc/repo/stage.py

@@ -122,16 +122,15 @@ class StageLoad:
             force=force,
             **stage_data,
         )
-        with self.repo.scm_context:
-            stage.dump(update_lock=update_lock)
-            try:
-                stage.ignore_outs()
-            except FileNotFoundError as exc:
-                ui.warn(
-                    f"Could not create .gitignore entry in {exc.filename}."
-                    " DVC will attempt to create .gitignore entry again when"
-                    " the stage is run."
-                )
+        stage.dump(update_lock=update_lock)
+        try:
+            stage.ignore_outs()
+        except FileNotFoundError as exc:
+            ui.warn(
+                f"Could not create .gitignore entry in {exc.filename}."
+                " DVC will attempt to create .gitignore entry again when"
+                " the stage is run."
+            )
 
         return stage
 

pyproject.toml

@@ -303,5 +303,5 @@ ignore-names = ["M", "SCM"]
 max-args = 10
 
 [tool.bandit]
-exclude_dirs = ["tests", "scripts"]
+exclude_dirs = ["tests", "scripts", "dvc/repo/experiments/utils.py"]
 skips = ["B101"]

tests/func/experiments/test_experiments.py

@@ -369,6 +369,12 @@ def test_list(tmp_dir, scm, dvc, exp_stage):
         baseline_a[:7]: {ref_info_a.name, ref_info_b.name}
     }
 
+    exp_list = dvc.experiments.ls(rev=[baseline_a, scm.get_rev()])
+    assert {key: set(val) for key, val in exp_list.items()} == {
+        baseline_a[:7]: {ref_info_a.name, ref_info_b.name},
+        "master": {ref_info_c.name},
+    }
+
     exp_list = dvc.experiments.ls(all_commits=True)
     assert {key: set(val) for key, val in exp_list.items()} == {
         baseline_a[:7]: {ref_info_a.name, ref_info_b.name},

tests/func/experiments/test_remote.py

@@ -251,6 +251,30 @@ def test_pull_args(tmp_dir, scm, dvc, git_downstream, exp_stage, all_, rev, resu
     assert git_downstream.tmp_dir.scm.get_ref(str(ref_info3)) == result3
 
 
+def test_pull_multi_rev(tmp_dir, scm, dvc, git_downstream, exp_stage):
+    baseline = scm.get_rev()
+
+    results = dvc.experiments.run(exp_stage.addressing, params=["foo=1"])
+    exp1 = first(results)
+    ref_info1 = first(exp_refs_by_rev(scm, exp1))
+    results = dvc.experiments.run(exp_stage.addressing, params=["foo=2"])
+    exp2 = first(results)
+    ref_info2 = first(exp_refs_by_rev(scm, exp2))
+
+    scm.commit("new_baseline")
+
+    results = dvc.experiments.run(exp_stage.addressing, params=["foo=3"])
+    exp3 = first(results)
+    ref_info3 = first(exp_refs_by_rev(scm, exp3))
+
+    downstream_exp = git_downstream.tmp_dir.dvc.experiments
+    git_downstream.tmp_dir.scm.fetch_refspecs(str(tmp_dir), ["master:master"])
+    downstream_exp.pull(git_downstream.remote, [], rev=[baseline, scm.get_rev()])
+    assert git_downstream.tmp_dir.scm.get_ref(str(ref_info1)) == exp1
+    assert git_downstream.tmp_dir.scm.get_ref(str(ref_info2)) == exp2
+    assert git_downstream.tmp_dir.scm.get_ref(str(ref_info3)) == exp3
+
+
 def test_pull_diverged(tmp_dir, scm, dvc, git_downstream, exp_stage):
     git_downstream.tmp_dir.scm_gen("foo", "foo", commit="init")
     remote_rev = git_downstream.tmp_dir.scm.get_rev()

tests/func/experiments/test_remove.py

@@ -156,3 +156,26 @@ def test_remove_experiments_by_rev(tmp_dir, scm, dvc, exp_stage):
     assert "queue2" in queue_revs
     assert scm.get_ref(new_exp_ref) is not None
     assert "queue4" in queue_revs
+
+
+def test_remove_multi_rev(tmp_dir, scm, dvc, exp_stage):
+    baseline = scm.get_rev()
+
+    results = dvc.experiments.run(exp_stage.addressing, params=["foo=1"])
+    baseline_exp_ref = first(exp_refs_by_rev(scm, first(results)))
+
+    dvc.experiments.run(
+        exp_stage.addressing, params=["foo=2"], queue=True, name="queue2"
+    )
+    scm.commit("new_baseline")
+
+    results = dvc.experiments.run(exp_stage.addressing, params=["foo=3"])
+    new_exp_ref = first(exp_refs_by_rev(scm, first(results)))
+
+    assert set(dvc.experiments.remove(rev=[baseline, scm.get_rev()])) == {
+        baseline_exp_ref.name,
+        new_exp_ref.name,
+    }
+
+    assert scm.get_ref(str(baseline_exp_ref)) is None
+    assert scm.get_ref(str(new_exp_ref)) is None

tests/func/experiments/test_utils.py

@@ -2,18 +2,19 @@ from funcy import first
 
 
 def test_generate_random_exp_name(tmp_dir, dvc, scm, exp_stage, mocker):
+    mocked_generator = mocker.MagicMock()
+    mocked_generator.choice.side_effect = [
+        0,
+        0,
+        0,
+        0,
+        1,
+        1,
+        0,
+        0,
+    ]
     mocker.patch(
-        "dvc.repo.experiments.utils.random.choice",
-        side_effect=[
-            0,
-            0,
-            0,
-            0,
-            1,
-            1,
-            0,
-            0,
-        ],
+        "dvc.repo.experiments.utils.random.Random", return_value=mocked_generator
     )
 
     ref = first(dvc.experiments.run(exp_stage.addressing, params=["foo=1"]))

tests/unit/command/test_stage.py

@@ -86,3 +86,15 @@ def test_stage_add(mocker, dvc, command, parsed_command):
         cmd=parsed_command,
         force=True,
     )
+
+
+def test_stage_add_and_run(mocker, dvc):
+    cli_args = parse_args(["stage", "add", "--run", "-n", "foo", "-o", "foo", "cmd"])
+    cmd = cli_args.func(cli_args)
+    add_mock = mocker.patch.object(cmd.repo.stage, "add")
+
+    assert cmd.run() == 0
+
+    assert called_once_with_subset(add_mock, name="foo", outs=["foo"], cmd="cmd")
+    add_mock.return_value.run.assert_called_once()
+    add_mock.return_value.dump.assert_called_once_with(update_pipeline=False)