deps: bump dvc-data to >=0.42.3

Makes for a much simpler resource management elsewhere.

dvc/commands/experiments/clean.py

@@ -14,7 +14,7 @@ class CmdExperimentsClean(CmdBase):
 
 
 def add_parser(experiments_subparsers, parent_parser):
-    EXPERIMENTS_CLEAN_HELP = "Cleanup dvc exp internal tempfiles."
+    EXPERIMENTS_CLEAN_HELP = "Cleanup experiments temporary internal files."
     experiments_clean_parser = experiments_subparsers.add_parser(
         "clean",
         parents=[parent_parser],

dvc/commands/experiments/diff.py

@@ -52,7 +52,7 @@ class CmdExperimentsDiff(CmdBase):
 
 
 def add_parser(experiments_subparsers, parent_parser):
-    EXPERIMENTS_DIFF_HELP = "Show changes between experiments in the DVC repository."
+    EXPERIMENTS_DIFF_HELP = "Show changes between experiments."
 
     experiments_diff_parser = experiments_subparsers.add_parser(
         "diff",

dvc/commands/experiments/gc.py

@@ -78,7 +78,8 @@ class CmdExperimentsGC(CmdBase):
 def add_parser(experiments_subparsers, parent_parser):
     EXPERIMENTS_GC_HELP = "Garbage collect unneeded experiments."
     EXPERIMENTS_GC_DESCRIPTION = (
-        "Removes all experiments which are not derived from the specifiedGit revisions."
+        "Removes all experiments which are not derived"
+        " from the specified Git revisions."
     )
     experiments_gc_parser = experiments_subparsers.add_parser(
         "gc",

dvc/commands/experiments/save.py

@@ -35,7 +35,7 @@ class CmdExperimentsSave(CmdBase):
 
 
 def add_parser(experiments_subparsers, parent_parser):
-    EXPERIMENTS_SAVE_HELP = "Save current workspace as a dvc experiment."
+    EXPERIMENTS_SAVE_HELP = "Save current workspace as an experiment."
     save_parser = experiments_subparsers.add_parser(
         "save",
         parents=[parent_parser],

dvc/commands/plots.py

@@ -19,7 +19,7 @@ def _show_json(renderers, split=False):
     from dvc.render.convert import to_json
 
     result = {renderer.name: to_json(renderer, split) for renderer in renderers}
-    ui.write_json(result)
+    ui.write_json(result, highlight=False)
 
 
 def _adjust_vega_renderers(renderers):

dvc/config_schema.py

@@ -1,8 +1,21 @@
+import logging
 import os
 from urllib.parse import urlparse
 
-from funcy import walk_values
-from voluptuous import All, Any, Coerce, Invalid, Lower, Optional, Range, Schema
+from funcy import once, walk_values
+from voluptuous import (
+    REMOVE_EXTRA,
+    All,
+    Any,
+    Coerce,
+    Invalid,
+    Lower,
+    Optional,
+    Range,
+    Schema,
+)
+
+logger = logging.getLogger(__name__)
 
 Bool = All(
     Lower,
@@ -65,6 +78,27 @@ class RelPath(str):
     pass
 
 
+class FeatureSchema(Schema):
+    def __init__(self, schema, required=False):
+        super().__init__(schema, required=required, extra=REMOVE_EXTRA)
+
+    @staticmethod
+    @once
+    def _log_deprecated(keys):
+        # only run this once per session
+        message = "%s config option%s unsupported"
+        paths = ", ".join(f"'feature.{key}'" for key in keys)
+        pluralize = " is" if len(keys) == 1 else "s are"
+        logger.warning(message, paths, pluralize)
+
+    def __call__(self, data):
+        ret = super().__call__(data)
+        extra_keys = data.keys() - ret.keys()
+        if extra_keys:
+            self._log_deprecated(sorted(extra_keys))
+        return ret
+
+
 REMOTE_COMMON = {
     "url": str,
     "checksum_jobs": All(Coerce(int), Range(1)),
@@ -262,12 +296,12 @@ SCHEMA = {
         },
     },
     # section for experimental features
-    "feature": {
-        Optional("machine", default=False): Bool,
-        # enabled by default. It's of no use, kept for backward compatibility.
-        Optional("data_index_cache", default=False): Bool,
-        Optional("parametrization", default=True): Bool,
-    },
+    # only specified keys are validated, others get logged and then ignored/removed
+    "feature": FeatureSchema(
+        {
+            Optional("machine", default=False): Bool,
+        },
+    ),
     "plots": {
         "html_template": str,
         Optional("auto_open", default=False): Bool,

dvc/fs/__init__.py

@@ -121,7 +121,7 @@ def _get_cloud_fs(repo_config, **kwargs):
 
     remote_conf = get_fs_config(repo_config, **kwargs)
     try:
-        remote_conf = SCHEMA["remote"][str](remote_conf)
+        remote_conf = SCHEMA["remote"][str](remote_conf)  # type: ignore[index]
     except Invalid as exc:
         raise RepoConfigError(str(exc)) from None
 

dvc/repo/__init__.py

@@ -397,17 +397,17 @@ class Repo:
         self._reset()
 
     @property
-    def data_index(self) -> Optional["DataIndex"]:
+    def data_index(self) -> "DataIndex":
         from dvc_data.index import DataIndex
 
-        if not self.index_db_dir:
-            return None
-
         if self._data_index is None:
-            index_dir = os.path.join(self.index_db_dir, "index", "data")
-            os.makedirs(index_dir, exist_ok=True)
+            if self.index_db_dir:
+                index_dir = os.path.join(self.index_db_dir, "index", "data")
+                os.makedirs(index_dir, exist_ok=True)
 
-            self._data_index = DataIndex.open(os.path.join(index_dir, "db.db"))
+                self._data_index = DataIndex.open(os.path.join(index_dir, "db.db"))
+            else:
+                self._data_index = DataIndex()
 
         return self._data_index
 

dvc/repo/index.py

@@ -354,42 +354,34 @@ class Index:
 
     @cached_property
     def data(self) -> "Dict[str, DataIndex]":
-        from dvc_data.index import DataIndex
-
         prefix: "DataIndexKey"
         loaded = False
 
         index = self.repo.data_index
-        if index is None:
-            index = DataIndex()
-
         prefix = ("tree", self.data_tree.hash_info.value)
         if index.has_node(prefix):
             loaded = True
 
-        try:
-            if not loaded:
-                _load_data_from_outs(index, prefix, self.outs)
-                index.commit()
+        if not loaded:
+            _load_data_from_outs(index, prefix, self.outs)
+            index.commit()
 
-            by_workspace = {}
-            by_workspace["repo"] = index.view((*prefix, "repo"))
-            by_workspace["local"] = index.view((*prefix, "local"))
+        by_workspace = {}
+        by_workspace["repo"] = index.view((*prefix, "repo"))
+        by_workspace["local"] = index.view((*prefix, "local"))
 
-            for out in self.outs:
-                if not out.use_cache:
-                    continue
+        for out in self.outs:
+            if not out.use_cache:
+                continue
 
-                ws, key = out.index_key
-                if ws not in by_workspace:
-                    by_workspace[ws] = index.view((*prefix, ws))
+            ws, key = out.index_key
+            if ws not in by_workspace:
+                by_workspace[ws] = index.view((*prefix, ws))
 
-                data_index = by_workspace[ws]
-                _load_storage_from_out(data_index.storage_map, key, out)
+            data_index = by_workspace[ws]
+            _load_storage_from_out(data_index.storage_map, key, out)
 
-            return by_workspace
-        finally:
-            index.close()
+        return by_workspace
 
     @staticmethod
     def _hash_targets(

dvc/ui/__init__.py

@@ -17,7 +17,9 @@ from dvc.utils.objects import cached_property
 
 if TYPE_CHECKING:
     from rich.console import Console as RichConsole
+    from rich.console import JustifyMethod, OverflowMethod
     from rich.status import Status
+    from rich.style import Style
     from rich.text import Text as RichText
 
     from dvc.progress import Tqdm
@@ -105,8 +107,9 @@ class Console:
         if indent is None and self.isatty():
             indent = 2
 
-        console = self.error_console if stderr else self.rich_console
-        return console.print_json(
+        from rich.json import JSON
+
+        json = JSON.from_data(
             data=data,
             indent=indent,
             highlight=bool(highlight),
@@ -117,6 +120,50 @@ class Console:
             default=default,
             sort_keys=sort_keys,
         )
+        if not highlight:
+            return self.write(json.text, stderr=stderr)
+        return self.rich_print(json, stderr=stderr, soft_wrap=True)
+
+    def rich_print(
+        self,
+        *objects: Any,
+        sep: str = " ",
+        end: str = "\n",
+        stderr: bool = False,
+        style: Optional[Union[str, "Style"]] = None,
+        justify: Optional["JustifyMethod"] = None,
+        overflow: Optional["OverflowMethod"] = None,
+        no_wrap: Optional[bool] = None,
+        emoji: Optional[bool] = None,
+        markup: Optional[bool] = None,
+        highlight: Optional[bool] = None,
+        width: Optional[int] = None,
+        height: Optional[int] = None,
+        crop: bool = True,
+        soft_wrap: Optional[bool] = None,
+        new_line_start: bool = False,
+    ) -> None:
+        if stderr:
+            console = self.error_console
+        else:
+            console = self.rich_console
+        return console.print(
+            *objects,
+            sep=sep,
+            end=end,
+            style=style,
+            justify=justify,
+            overflow=overflow,
+            no_wrap=no_wrap,
+            emoji=emoji,
+            markup=markup,
+            highlight=highlight,
+            width=width,
+            height=height,
+            crop=crop,
+            soft_wrap=soft_wrap,
+            new_line_start=new_line_start,
+        )
 
     def write(
         self,
@@ -142,10 +189,21 @@ class Console:
         with Tqdm.external_write_mode(file=file):
             # if we are inside pager context, send the output to rich's buffer
             if styled or self._paginate.get():
-                console = self.error_console if stderr else self.rich_console
                 if styled:
-                    return console.print(*objects, sep=sep, end=end)
-                return console.out(*objects, sep=sep, end=end, highlight=False)
+                    return self.rich_print(*objects, sep=sep, end=end, stderr=stderr)
+                return self.rich_print(
+                    sep.join(str(_object) for _object in objects),
+                    style=None,
+                    highlight=False,
+                    emoji=False,
+                    markup=False,
+                    no_wrap=True,
+                    overflow="ignore",
+                    crop=False,
+                    sep=sep,
+                    end=end,
+                    stderr=stderr,
+                )
 
             values = (self.formatter.format(obj, style) for obj in objects)
             return print(*values, sep=sep, end=end, file=file)

pyproject.toml

@@ -34,7 +34,7 @@ dependencies = [
     "configobj>=5.0.6",
     "distro>=1.3",
     "dpath<3,>=2.1.0",
-    "dvc-data>=0.42.1,<0.43",
+    "dvc-data>=0.42.3,<0.43",
     "dvc-http",
     "dvc-render>=0.1.2",
     "dvc-studio-client>=0.5.0,<1",

tests/unit/test_config.py

@@ -1,3 +1,4 @@
+import logging
 import os
 import textwrap
 
@@ -94,3 +95,22 @@ def test_load_configob_error(tmp_dir, dvc, mocker):
     with pytest.raises(ConfigError):
         with config.edit():
             pass
+
+
+def test_feature_section_supports_arbitrary_values(caplog):
+    with caplog.at_level(logging.WARNING, logger="dvc.config_schema"):
+        data = Config.validate(
+            {
+                "feature": {
+                    "random_key_1": "random_value_1",
+                    "random_key_2": 42,
+                }
+            }
+        )
+
+    assert "random_key_1" not in data
+    assert "random_key_2" not in data
+    assert (
+        "'feature.random_key_1', 'feature.random_key_2' "
+        "config options are unsupported"
+    ) in caplog.text