cache: check-missing: add type, oid name and oid

Per https://github.com/iterative/dvc/pull/9857#issuecomment-1683185006

.pre-commit-config.yaml

@@ -32,7 +32,7 @@ repos:
     hooks:
       - id: isort
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: 'v0.0.282'
+    rev: 'v0.0.285'
     hooks:
       - id: ruff
         args: [--fix, --exit-non-zero-on-fix]

dvc/commands/experiments/save.py

@@ -27,10 +27,6 @@ class CmdExperimentsSave(CmdBase):
         else:
             name = self.repo.experiments.get_exact_name([ref])[ref]
             ui.write(f"Experiment has been saved as: {name}")
-            ui.write(
-                "\nTo promote an experiment to a Git branch run:\n\n"
-                "\tdvc exp branch <exp> <branch>\n"
-            )
 
         return 0
 

dvc/commands/ls/__init__.py

@@ -11,7 +11,18 @@ from dvc.ui import ui
 logger = logging.getLogger(__name__)
 
 
-def _prettify(entries, with_color=False):
+def _format_entry(entry, fmt):
+    from dvc.utils.humanize import naturalsize
+
+    size = entry.get("size")
+    if size is None:
+        size = ""
+    else:
+        size = naturalsize(size)
+    return size, fmt(entry)
+
+
+def show_entries(entries, with_color=False, with_size=False):
     if with_color:
         ls_colors = LsColors()
         fmt = ls_colors.format
@@ -20,7 +31,12 @@ def _prettify(entries, with_color=False):
         def fmt(entry):
             return entry["path"]
 
-    return [fmt(entry) for entry in entries]
+    if with_size:
+        ui.table([_format_entry(entry, fmt) for entry in entries])
+        return
+
+    # NOTE: this is faster than ui.table for very large number of entries
+    ui.write("\n".join(fmt(entry) for entry in entries))
 
 
 class CmdList(CmdBaseNoRepo):
@@ -41,8 +57,7 @@ class CmdList(CmdBaseNoRepo):
             if self.args.json:
                 ui.write_json(entries)
             elif entries:
-                entries = _prettify(entries, with_color=True)
-                ui.write("\n".join(entries))
+                show_entries(entries, with_color=True, with_size=self.args.size)
             return 0
         except DvcException:
             logger.exception("failed to list '%s'", self.args.url)
@@ -106,6 +121,11 @@ def add_parser(subparsers, parent_parser):
             "specified by '--remote') in the target repository."
         ),
     )
+    list_parser.add_argument(
+        "--size",
+        action="store_true",
+        help="Show sizes.",
+    )
     list_parser.add_argument(
         "path",
         nargs="?",

dvc/commands/ls_url.py

@@ -3,9 +3,8 @@ import logging
 
 from dvc.cli.command import CmdBaseNoRepo
 from dvc.cli.utils import append_doc_link
-from dvc.ui import ui
 
-from .ls import _prettify
+from .ls import show_entries
 
 logger = logging.getLogger(__name__)
 
@@ -16,8 +15,7 @@ class CmdListUrl(CmdBaseNoRepo):
 
         entries = Repo.ls_url(self.args.url, recursive=self.args.recursive)
         if entries:
-            entries = _prettify(entries, with_color=True)
-            ui.write("\n".join(entries))
+            show_entries(entries, with_color=True, with_size=self.args.size)
         return 0
 
 
@@ -40,4 +38,9 @@ def add_parser(subparsers, parent_parser):
         action="store_true",
         help="Recursively list files.",
     )
+    lsurl_parser.add_argument(
+        "--size",
+        action="store_true",
+        help="Show sizes.",
+    )
     lsurl_parser.set_defaults(func=CmdListUrl)

dvc/commands/remote.py

@@ -30,8 +30,8 @@ class CmdRemoteAdd(CmdRemote):
         with self.config.edit(self.args.level) as conf:
             if self.args.name in conf["remote"] and not self.args.force:
                 raise ConfigError(
-                    "remote '{}' already exists. Use `-f|--force` to "
-                    "overwrite it.".format(self.args.name)
+                    f"remote '{self.args.name}' already exists. Use `-f|--force` to "
+                    "overwrite it."
                 )
 
             conf["remote"][self.args.name] = {"url": self.args.url}

dvc/config.py

@@ -99,6 +99,7 @@ class Config(dict):
     ):  # pylint: disable=super-init-not-called
         from dvc.fs import LocalFileSystem
 
+        dvc_dir = os.fspath(dvc_dir) if dvc_dir else None
         self.dvc_dir = dvc_dir
         self.wfs = LocalFileSystem()
         self.fs = fs or self.wfs

dvc/dependency/repo.py

@@ -1,3 +1,4 @@
+from copy import deepcopy
 from typing import TYPE_CHECKING, Dict, Optional, Union
 
 from voluptuous import Any, Required
@@ -24,12 +25,12 @@ class RepoDependency(Dependency):
             Required(PARAM_URL): str,
             PARAM_REV: str,
             PARAM_REV_LOCK: str,
-            PARAM_CONFIG: str,
+            PARAM_CONFIG: Any(str, dict),
             PARAM_REMOTE: Any(str, dict),
         }
     }
 
-    def __init__(self, def_repo: Dict[str, str], stage: "Stage", *args, **kwargs):
+    def __init__(self, def_repo: Dict[str, Any], stage: "Stage", *args, **kwargs):
         self.def_repo = def_repo
         super().__init__(stage, *args, **kwargs)
 
@@ -107,17 +108,17 @@ class RepoDependency(Dependency):
 
         rem = self.def_repo.get("remote")
         if isinstance(rem, dict):
-            remote = None  # type: ignore[unreachable]
+            remote = None
             remote_config = rem
         else:
             remote = rem
             remote_config = None
 
-        conf = self.def_repo.get("config")
-        if conf:
-            config = Config.load_file(conf)
+        conf = self.def_repo.get("config", {})
+        if isinstance(conf, dict):
+            config = deepcopy(conf)
         else:
-            config = {}
+            config = Config.load_file(conf)
 
         config["cache"] = self.repo.config["cache"]
         config["cache"]["dir"] = self.repo.cache.local_cache_dir

dvc/exceptions.py

@@ -196,7 +196,7 @@ class ETagMismatchError(DvcException):
     def __init__(self, etag, cached_etag):
         super().__init__(
             "ETag mismatch detected when copying file to cache! "
-            "(expected: '{}', actual: '{}')".format(etag, cached_etag)
+            f"(expected: '{etag}', actual: '{cached_etag}')"
         )
 
 

dvc/fs/dvc.py

@@ -139,7 +139,7 @@ class _DVCFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
             self.repo_factory = repo_factory
 
         def _getcwd():
-            relparts = ()
+            relparts: Tuple[str, ...] = ()
             assert repo is not None
             if repo.fs.path.isin(repo.fs.path.getcwd(), repo.root_dir):
                 relparts = repo.fs.path.relparts(repo.fs.path.getcwd(), repo.root_dir)
@@ -163,6 +163,7 @@ class _DVCFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
             self._datafss[key] = DataFileSystem(index=repo.index.data["repo"])
 
     def _get_key(self, path: "StrPath") -> Key:
+        path = os.fspath(path)
         parts = self.repo.fs.path.relparts(path, self.repo.root_dir)
         if parts == (os.curdir,):
             return ()
@@ -261,15 +262,12 @@ class _DVCFileSystem(AbstractFileSystem):  # pylint:disable=abstract-method
         try:
             return self.repo.fs.open(fs_path, mode=mode)
         except FileNotFoundError:
-            repo, dvc_fs, subkey = self._get_subrepo_info(key)
+            _, dvc_fs, subkey = self._get_subrepo_info(key)
             if not dvc_fs:
                 raise
 
         dvc_path = _get_dvc_path(dvc_fs, subkey)
-        kw = {}
-        if kwargs.get("cache_remote_stream", False):
-            kw["cache_odb"] = repo.cache.local
-        return dvc_fs.open(dvc_path, mode=mode, **kw)
+        return dvc_fs.open(dvc_path, mode=mode, cache=kwargs.get("cache", False))
 
     def isdvc(self, path, **kwargs) -> bool:
         """Is this entry dvc-tracked?"""

dvc/ignore.py

@@ -148,7 +148,9 @@ class DvcIgnorePatterns(DvcIgnore):
         return bool(self.pattern_list)
 
 
-CheckIgnoreResult = namedtuple("CheckIgnoreResult", ["file", "match", "patterns"])
+CheckIgnoreResult = namedtuple(  # noqa: PYI024
+    "CheckIgnoreResult", ["file", "match", "patterns"]
+)
 
 
 def _no_match(path):

dvc/output.py

@@ -1142,7 +1142,7 @@ class Output:
             if not force and not prompt.confirm(msg.format(self.fs_path)):
                 raise CollectCacheError(  # noqa: B904
                     "unable to fully collect used cache"
-                    " without cache for directory '{}'".format(self)
+                    f" without cache for directory '{self}'"
                 )
             return None
 
@@ -1171,19 +1171,15 @@ class Output:
 
         if not self.hash_info:
             msg = (
-                "Output '{}'({}) is missing version info. "
+                f"Output '{self}'({self.stage}) is missing version info. "
                 "Cache for it will not be collected. "
-                "Use `dvc repro` to get your pipeline up to date.".format(
-                    self, self.stage
-                )
+                "Use `dvc repro` to get your pipeline up to date."
             )
             if self.exists:
                 msg += (
                     "\n"
-                    "You can also use `dvc commit {stage.addressing}` "
-                    "to associate existing '{out}' with {stage}.".format(
-                        out=self, stage=self.stage
-                    )
+                    f"You can also use `dvc commit {self.stage.addressing}` "
+                    f"to associate existing '{self}' with {self.stage}."
                 )
             logger.warning(msg)
             return {}

dvc/pathspec_math.py

@@ -8,7 +8,7 @@ from pathspec.util import normalize_file
 
 from dvc.utils import relpath
 
-PatternInfo = namedtuple("PatternInfo", ["patterns", "file_info"])
+PatternInfo = namedtuple("PatternInfo", ["patterns", "file_info"])  # noqa: PYI024
 
 
 def _not_ignore(rule):

dvc/repo/brancher.py

@@ -57,11 +57,11 @@ def brancher(  # noqa: E302
 
     from dvc.fs import LocalFileSystem
 
-    repo_root_parts = ()
+    repo_root_parts: Tuple[str, ...] = ()
     if self.fs.path.isin(self.root_dir, self.scm.root_dir):
         repo_root_parts = self.fs.path.relparts(self.root_dir, self.scm.root_dir)
 
-    cwd_parts = ()
+    cwd_parts: Tuple[str, ...] = ()
     if self.fs.path.isin(self.fs.path.getcwd(), self.scm.root_dir):
         cwd_parts = self.fs.path.relparts(self.fs.path.getcwd(), self.scm.root_dir)
 
@@ -144,11 +144,11 @@ def switch(repo: "Repo", rev: str) -> Iterator[str]:
     if rev != "workspace":
         rev = resolve_rev(repo.scm, rev)
 
-    repo_root_parts = ()
+    repo_root_parts: Tuple[str, ...] = ()
     if repo.fs.path.isin(repo.root_dir, repo.scm.root_dir):
         repo_root_parts = repo.fs.path.relparts(repo.root_dir, repo.scm.root_dir)
 
-    cwd_parts = ()
+    cwd_parts: Tuple[str, ...] = ()
     if repo.fs.path.isin(repo.fs.path.getcwd(), repo.scm.root_dir):
         cwd_parts = repo.fs.path.relparts(repo.fs.path.getcwd(), repo.scm.root_dir)
 

dvc/repo/cache.py

@@ -26,6 +26,14 @@ def check_missing(repo, rev=None, max_size=None, types=None):
             continue
 
         if not fs.exists(path):
-            ret.append(os.path.join(*entry.key))
+            typ = "directory" if (entry.meta and entry.meta.isdir) else "file"
+            ret.append(
+                (
+                    typ,
+                    entry.hash_info.name,
+                    entry.hash_info.value,
+                    os.path.join(*entry.key),
+                )
+            )
 
     return ret

dvc/repo/checkout.py

@@ -81,7 +81,7 @@ def _check_can_delete(
         if cache_fs.exists(cache_path):
             continue
 
-        entry_paths.append(fs.path.join(path, *entry.key))
+        entry_paths.append(fs.path.join(path, *(entry.key or ())))
 
     if not entry_paths:
         return

dvc/repo/experiments/__init__.py

@@ -210,10 +210,6 @@ class Experiments:
             )
         else:
             ui.write("Experiment results have been applied to your workspace.")
-        ui.write(
-            "\nTo promote an experiment to a Git branch run:\n\n"
-            "\tdvc exp branch <exp> <branch>\n"
-        )
 
     def new(
         self,

dvc/repo/experiments/executor/base.py

@@ -129,7 +129,6 @@ class BaseExecutor(ABC):
 
     PACKED_ARGS_FILE = "repro.dat"
     WARN_UNTRACKED = False
-    QUIET = False
     INFOFILE_EXT = ".run"
     DEFAULT_LOCATION: str = "workspace"
 
@@ -602,8 +601,7 @@ class BaseExecutor(ABC):
             info.status = TaskStatus.RUNNING
             if infofile is not None:
                 info.dump_json(infofile)
-            if cls.QUIET:
-                dvc.scm_context.quiet = cls.QUIET
+            dvc.scm_context.quiet = True
             old_cwd = os.getcwd()
 
             for path in copy_paths or []:

dvc/repo/experiments/executor/local.py

@@ -65,7 +65,6 @@ class TempDirExecutor(BaseLocalExecutor):
     # debugging user code), and suppress other DVC hints (like `git add`
     # suggestions) that are not applicable outside of workspace runs
     WARN_UNTRACKED = True
-    QUIET = True
     DEFAULT_LOCATION = "tempdir"
 
     @retry(180, errors=LockError, timeout=1)

dvc/repo/imp.py

@@ -13,16 +13,28 @@ def imp(
     if rev is not None:
         erepo["rev"] = rev
 
+    if remote and remote_config and isinstance(config, str):
+        raise ValueError(
+            "Can't specify config path together with both remote and remote_config"
+        )
+
     if config is not None:
         erepo["config"] = config
 
     if remote is not None and remote_config is not None:
         conf = erepo.get("config") or {}
+
+        core = conf.get("core") or {}
+        core["remote"] = remote
+
         remotes = conf.get("remote") or {}
         remote_conf = remotes.get(remote) or {}
         remote_conf.update(remote_config)
         remotes[remote] = remote_conf
+
+        conf["core"] = core
         conf["remote"] = remotes
+
         erepo["config"] = conf
     elif remote is not None:
         erepo["remote"] = remote

dvc/repo/ls.py

@@ -109,6 +109,7 @@ def _ls(
             "isout": dvc_info.get("isout", False),
             "isdir": info["type"] == "directory",
             "isexec": info.get("isexec", False),
+            "size": info.get("size"),
         }
 
     return ret

dvc/repo/ls_url.py

@@ -20,6 +20,7 @@ def ls_url(url, *, config=None, recursive=False):
             ls_info = {
                 "path": fs.path.relpath(info["name"], fs_path),
                 "isdir": info["type"] == "directory",
+                "size": info.get("size"),
             }
             ret.append(ls_info)
 

dvc/repo/plots/__init__.py

@@ -46,8 +46,8 @@ SUPPORTED_IMAGE_EXTENSIONS = ImageRenderer.EXTENSIONS
 class PlotMetricTypeError(DvcException):
     def __init__(self, file):
         super().__init__(
-            "'{}' - file type error\n"
-            "Only JSON, YAML, CSV and TSV formats are supported.".format(file)
+            f"'{file}' - file type error\n"
+            "Only JSON, YAML, CSV and TSV formats are supported."
         )
 
 
@@ -212,7 +212,7 @@ class Plots:
             short_rev = "workspace"
             if rev := getattr(self.repo.fs, "rev", None):
                 short_rev = rev[:7]
-            _resolve_data_sources(data, short_rev, cache_remote_stream=True)
+            _resolve_data_sources(data, short_rev, cache=True)
             result.update(data)
 
         errored = errored_revisions(result)
@@ -281,9 +281,7 @@ def _is_plot(out: "Output") -> bool:
     return bool(out.plot)
 
 
-def _resolve_data_sources(
-    plots_data: Dict, rev: str, cache_remote_stream: bool = False
-):
+def _resolve_data_sources(plots_data: Dict, rev: str, cache: bool = False):
     from dvc.progress import Tqdm
 
     values = list(plots_data.values())
@@ -298,7 +296,7 @@ def _resolve_data_sources(
     def resolve(value):
         data_source = value.pop("data_source")
         assert callable(data_source)
-        value.update(data_source(cache_remote_stream=cache_remote_stream))
+        value.update(data_source(cache=cache))
 
     if not to_resolve:
         return
@@ -448,6 +446,7 @@ def _resolve_definitions(
     definitions: "DictStrAny",
     onerror: Optional[Callable[[Any], Any]] = None,
 ):
+    config_path = os.fspath(config_path)
     config_dir = fs.path.dirname(config_path)
     result: Dict[str, Dict] = {}
     for plot_id, plot_props in definitions.items():

dvc/stage/exceptions.py

@@ -98,7 +98,7 @@ class StageNameUnspecified(DvcException):
     def __init__(self, file):
         super().__init__(
             "Stage name not provided."
-            "Please specify the name as: `{}:stage_name`".format(file.relpath)
+            f"Please specify the name as: `{file.relpath}:stage_name`"
         )
 
 

dvc/testing/tmp_dir.py

@@ -115,8 +115,8 @@ class TmpDir(pathlib.Path):
     def _require(self, name):
         if not hasattr(self, name):
             raise TypeError(
-                "Can't use {name} for this temporary dir. "
-                'Did you forget to use "{name}" fixture?'.format(name=name)
+                f"Can't use {name} for this temporary dir. "
+                f'Did you forget to use "{name}" fixture?'
             )
 
     # Bootstrapping methods

dvc/utils/humanize.py

@@ -31,3 +31,11 @@ def truncate_text(text: str, max_length: int, with_ellipsis: bool = True) -> str
         return text[: max_length - 1] + ELLIPSIS
 
     return text[:max_length]
+
+
+def naturalsize(value: float, base: int = 1024) -> str:
+    from tqdm import tqdm
+
+    if value < base:
+        return f"{value:.0f}"
+    return tqdm.format_sizeof(value, divisor=base)

pyproject.toml

@@ -34,7 +34,7 @@ dependencies = [
     "configobj>=5.0.6",
     "distro>=1.3",
     "dpath<3,>=2.1.0",
-    "dvc-data>=2.14.0,<2.15.0",
+    "dvc-data>=2.15.0,<2.16.0",
     "dvc-http>=2.29.0",
     "dvc-render>=0.3.1,<1",
     "dvc-studio-client>=0.9.2,<1",
@@ -75,7 +75,7 @@ gdrive = ["dvc-gdrive==2.20"]
 gs = ["dvc-gs==2.22.1"]
 hdfs = ["dvc-hdfs==2.19"]
 lint = [
-    "mypy==1.4.1",
+    "mypy==1.5.1",
     "pylint==2.17.5",
     "types-colorama",
     "types-psutil",
@@ -102,7 +102,7 @@ tests = [
     "flaky",
     "pytest<8,>=7",
     "pytest-cov>=4.1.0",
-    "pytest-docker>=1,<2",
+    "pytest-docker>=1,<3",
     "pytest-lazy-fixture",
     "pytest-mock",
     "pytest-test-utils",

tests/func/test_import.py

@@ -658,3 +658,75 @@ def test_parameterized_repo(tmp_dir, dvc, scm, erepo_dir, paths):
         "url": os.fspath(erepo_dir),
         "rev_lock": erepo_dir.scm.get_rev(),
     }
+
+
+@pytest.mark.parametrize(
+    "options, def_repo",
+    [
+        ({"config": "myconfig"}, {"config": "myconfig"}),
+        ({"remote": "myremote"}, {"remote": "myremote"}),
+        ({"remote_config": {"key": "value"}}, {"remote": {"key": "value"}}),
+        (
+            {
+                "remote": "myremote",
+                "remote_config": {"key": "value"},
+            },
+            {
+                "config": {
+                    "core": {"remote": "myremote"},
+                    "remote": {
+                        "myremote": {"key": "value"},
+                    },
+                },
+            },
+        ),
+        (
+            {
+                "remote": "myremote",
+                "remote_config": {"key": "value"},
+                "config": {"otherkey": "othervalue"},
+            },
+            {
+                "config": {
+                    "core": {"remote": "myremote"},
+                    "remote": {
+                        "myremote": {"key": "value"},
+                    },
+                    "otherkey": "othervalue",
+                },
+            },
+        ),
+    ],
+)
+def test_import_configs(tmp_dir, scm, dvc, erepo_dir, options, def_repo):
+    with erepo_dir.chdir():
+        erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
+
+    (tmp_dir / "myconfig").touch()
+
+    stage = dvc.imp(
+        os.fspath(erepo_dir), "foo", "foo_imported", no_exec=True, **options
+    )
+    assert stage.deps[0].def_repo == {
+        "url": os.fspath(erepo_dir),
+        **def_repo,
+    }
+
+
+def test_import_invalid_configs(tmp_dir, scm, dvc, erepo_dir):
+    with erepo_dir.chdir():
+        erepo_dir.dvc_gen("foo", "foo content", commit="create foo")
+
+    with pytest.raises(
+        ValueError,
+        match="Can't specify config path together with both remote and remote_config",
+    ):
+        dvc.imp(
+            os.fspath(erepo_dir),
+            "foo",
+            "foo_imported",
+            no_exec=True,
+            config="myconfig",
+            remote="myremote",
+            remote_config={"key": "value"},
+        )

tests/func/test_init.py

@@ -113,10 +113,7 @@ def test_init_when_ignored_by_git(tmp_dir, scm, caplog):
     with caplog.at_level(logging.ERROR, logger="dvc"):
         assert main(["init"]) == 1
     assert (
-        "{dvc_dir} is ignored by your SCM tool. \n"
+        f"{tmp_dir / DvcRepo.DVC_DIR} is ignored by your SCM tool. \n"
         "Make sure that it's tracked, "
-        "for example, by adding '!.dvc' to .gitignore.".format(
-            dvc_dir=tmp_dir / DvcRepo.DVC_DIR
-        )
-        in caplog.text
+        "for example, by adding '!.dvc' to .gitignore." in caplog.text
     )

tests/func/test_ls.py

@@ -460,7 +460,7 @@ def test_ls_shows_pipeline_tracked_outs(tmp_dir, dvc, scm, run_copy):
     match_files(files, ((("bar",), True),))
 
 
-def test_ls_granular(erepo_dir):
+def test_ls_granular(erepo_dir, M):
     with erepo_dir.chdir():
         erepo_dir.dvc_gen(
             {
@@ -475,15 +475,21 @@ def test_ls_granular(erepo_dir):
 
     entries = Repo.ls(os.fspath(erepo_dir), os.path.join("dir", "subdir"))
     assert entries == [
-        {"isout": True, "isdir": False, "isexec": False, "path": "bar"},
-        {"isout": True, "isdir": False, "isexec": False, "path": "foo"},
+        {"isout": True, "isdir": False, "isexec": False, "path": "bar", "size": 3},
+        {"isout": True, "isdir": False, "isexec": False, "path": "foo", "size": 3},
     ]
 
     entries = Repo.ls(os.fspath(erepo_dir), "dir")
     assert entries == [
-        {"isout": True, "isdir": False, "isexec": False, "path": "1"},
-        {"isout": True, "isdir": False, "isexec": False, "path": "2"},
-        {"isout": True, "isdir": True, "isexec": False, "path": "subdir"},
+        {"isout": True, "isdir": False, "isexec": False, "path": "1", "size": 1},
+        {"isout": True, "isdir": False, "isexec": False, "path": "2", "size": 1},
+        {
+            "isout": True,
+            "isdir": True,
+            "isexec": False,
+            "path": "subdir",
+            "size": M.instance_of(int),
+        },
     ]
 
 
@@ -508,14 +514,14 @@ def test_ls_target(erepo_dir, use_scm):
         return Repo.ls(os.fspath(erepo_dir), path)
 
     assert _ls(os.path.join("dir", "1")) == [
-        {"isout": isout, "isdir": False, "isexec": False, "path": "1"}
+        {"isout": isout, "isdir": False, "isexec": False, "path": "1", "size": 1}
     ]
     assert _ls(os.path.join("dir", "subdir", "foo")) == [
-        {"isout": isout, "isdir": False, "isexec": False, "path": "foo"}
+        {"isout": isout, "isdir": False, "isexec": False, "path": "foo", "size": 3}
     ]
     assert _ls(os.path.join("dir", "subdir")) == [
-        {"isdir": False, "isexec": 0, "isout": isout, "path": "bar"},
-        {"isdir": False, "isexec": 0, "isout": isout, "path": "foo"},
+        {"isdir": False, "isexec": 0, "isout": isout, "path": "bar", "size": 3},
+        {"isdir": False, "isexec": 0, "isout": isout, "path": "foo", "size": 3},
     ]
 
 
@@ -559,7 +565,7 @@ def test_subrepo(dvc_top_level, erepo):
     assert _list_files(subrepo, "dvc_dir") == {"lorem"}
 
 
-def test_broken_symlink(tmp_dir, dvc):
+def test_broken_symlink(tmp_dir, dvc, M):
     from dvc.fs import system
 
     tmp_dir.gen("file", "content")
@@ -575,17 +581,19 @@ def test_broken_symlink(tmp_dir, dvc):
             "isdir": False,
             "isexec": False,
             "path": ".dvcignore",
+            "size": M.instance_of(int),
         },
         {
             "isout": False,
             "isdir": False,
             "isexec": False,
             "path": "link",
+            "size": 0,
         },
     ]
 
 
-def test_ls_broken_dir(tmp_dir, dvc):
+def test_ls_broken_dir(tmp_dir, dvc, M):
     from dvc_data.index import DataIndexDirError
 
     tmp_dir.dvc_gen(
@@ -606,13 +614,43 @@ def test_ls_broken_dir(tmp_dir, dvc):
 
     entries = Repo.ls(os.fspath(tmp_dir))
     assert entries == [
-        {"isdir": False, "isexec": False, "isout": False, "path": ".dvcignore"},
-        {"isdir": True, "isexec": False, "isout": True, "path": "broken"},
-        {"isdir": False, "isexec": False, "isout": False, "path": "broken.dvc"},
-        {"isdir": True, "isexec": False, "isout": True, "path": "dir"},
-        {"isdir": False, "isexec": False, "isout": False, "path": "dir.dvc"},
-        {"isdir": False, "isexec": False, "isout": True, "path": "foo"},
-        {"isdir": False, "isexec": False, "isout": False, "path": "foo.dvc"},
+        {
+            "isdir": False,
+            "isexec": False,
+            "isout": False,
+            "path": ".dvcignore",
+            "size": M.instance_of(int),
+        },
+        {"isdir": True, "isexec": False, "isout": True, "path": "broken", "size": 3},
+        {
+            "isdir": False,
+            "isexec": False,
+            "isout": False,
+            "path": "broken.dvc",
+            "size": M.instance_of(int),
+        },
+        {
+            "isdir": True,
+            "isexec": False,
+            "isout": True,
+            "path": "dir",
+            "size": M.instance_of(int),
+        },
+        {
+            "isdir": False,
+            "isexec": False,
+            "isout": False,
+            "path": "dir.dvc",
+            "size": M.instance_of(int),
+        },
+        {"isdir": False, "isexec": False, "isout": True, "path": "foo", "size": 3},
+        {
+            "isdir": False,
+            "isexec": False,
+            "isout": False,
+            "path": "foo.dvc",
+            "size": M.instance_of(int),
+        },
     ]
 
     with pytest.raises(DataIndexDirError):

tests/unit/command/test_diff.py

@@ -45,10 +45,10 @@ def test_default(mocker, capsys, dvc):
         "    file\n"
         "\n"
         "Renamed:\n"
-        "    data{sep}file_old -> data{sep}file_new\n"
+        f"    data{os.path.sep}file_old -> data{os.path.sep}file_new\n"
         "\n"
         "files summary: 1 added, 1 renamed"
-    ).format(sep=os.path.sep) in capsys.readouterr()[0]
+    ) in capsys.readouterr()[0]
 
 
 def test_show_hash(mocker, capsys, dvc):
@@ -252,14 +252,14 @@ def test_show_markdown(capsys):
         "|--------------|----------------------|\n"
         "| added        | file                 |\n"
         "| deleted      | zoo                  |\n"
-        "| deleted      | data{sep}                |\n"
-        "| deleted      | data{sep}foo             |\n"
-        "| deleted      | data{sep}bar             |\n"
+        f"| deleted      | data{os.path.sep}                |\n"
+        f"| deleted      | data{os.path.sep}foo             |\n"
+        f"| deleted      | data{os.path.sep}bar             |\n"
         "| renamed      | file_old -> file_new |\n"
         "| modified     | file                 |\n"
         "| not in cache | file2                |\n"
         "\n"
-    ).format(sep=os.path.sep)
+    )
 
 
 def test_show_markdown_with_hash(capsys):
@@ -289,14 +289,14 @@ def test_show_markdown_with_hash(capsys):
         "|--------------|--------------------|----------------------|\n"
         "| added        | 00000000           | file                 |\n"
         "| deleted      | 22222              | zoo                  |\n"
-        "| deleted      | XXXXXXXX           | data{sep}                |\n"
-        "| deleted      | 11111111           | data{sep}foo             |\n"
-        "| deleted      | 00000000           | data{sep}bar             |\n"
+        f"| deleted      | XXXXXXXX           | data{os.path.sep}                |\n"
+        f"| deleted      | 11111111           | data{os.path.sep}foo             |\n"
+        f"| deleted      | 00000000           | data{os.path.sep}bar             |\n"
         "| renamed      | 11111111           | file_old -> file_new |\n"
         "| modified     | AAAAAAAA..BBBBBBBB | file                 |\n"
         "| not in cache | 12345678           | file2                |\n"
         "\n"
-    ).format(sep=os.path.sep)
+    )
 
 
 def test_show_markdown_hide_missing(capsys):
@@ -321,13 +321,13 @@ def test_show_markdown_hide_missing(capsys):
         "|----------|----------------------|\n"
         "| added    | file                 |\n"
         "| deleted  | zoo                  |\n"
-        "| deleted  | data{sep}                |\n"
-        "| deleted  | data{sep}foo             |\n"
-        "| deleted  | data{sep}bar             |\n"
+        f"| deleted  | data{os.path.sep}                |\n"
+        f"| deleted  | data{os.path.sep}foo             |\n"
+        f"| deleted  | data{os.path.sep}bar             |\n"
         "| renamed  | file_old -> file_new |\n"
         "| modified | file                 |\n"
         "\n"
-    ).format(sep=os.path.sep)
+    )
 
 
 def test_hide_missing(mocker, capsys, dvc):

tests/unit/command/test_help.py

@@ -25,18 +25,18 @@ def command_tuples():
 
     main_parser = get_main_parser()
     recurse_parser(main_parser)
+
+    # the no. of commands will usually go up,
+    # but if we ever remove commands and drop below, adjust the magic number accordingly
+    assert len(commands) >= 96
     return sorted(commands)
 
 
-def pytest_generate_tests(metafunc):
-    def ids(values):
-        if values:
-            return "-".join(values)
-        return "dvc"
-
-    metafunc.parametrize("command_tuples", command_tuples(), ids=ids)
+def ids(values):
+    return "-".join(values) or "dvc"
 
 
+@pytest.mark.parametrize("command_tuples", command_tuples(), ids=ids)
 def test_help(caplog, capsys, command_tuples):
     with caplog.at_level(logging.INFO), pytest.raises(SystemExit) as e:
         main([*command_tuples, "--help"])

tests/unit/repo/test_scm_context.py

@@ -100,13 +100,14 @@ def test_scm_context_on_no_files_to_track(caplog, scm_context, autostage, quiet)
     assert not caplog.text
 
 
-def test_scm_context_remind_to_track(caplog, scm_context):
-    with scm_context() as context:
+@pytest.mark.parametrize("quiet", [True, False])
+def test_scm_context_remind_to_track(caplog, scm_context, quiet):
+    with scm_context(quiet=quiet) as context:
         context.track_file("foo")
         context.track_file("lorem ipsum")
         assert context.files_to_track == {"foo", "lorem ipsum"}
 
-    if isinstance(scm_context.scm, NoSCM):
+    if quiet or isinstance(scm_context.scm, NoSCM):
         assert not caplog.text
     else:
         assert "To track the changes with git, run:" in caplog.text

tests/unit/utils/test_utils.py

@@ -80,10 +80,8 @@ def test_resolve_output(inp, out, is_dir, expected, mocker):
         ["dvc.yaml:name", ("dvc.yaml", "name"), None],
         [":name", ("dvc.yaml", "name"), None],
         ["stage.dvc", ("stage.dvc", None), None],
-        ["dvc.yaml:name", ("dvc.yaml", "name"), None],
         ["../models/stage.dvc", ("../models/stage.dvc", None), "def"],
         [":name", ("default", "name"), "default"],
-        [":name", ("default", "name"), "default"],
         ["something.dvc:name", ("something.dvc", "name"), None],
         ["../something.dvc:name", ("../something.dvc", "name"), None],
         ["file", (None, "file"), None],