dataeval-flow 0.2.2__tar.gz → 0.2.3__tar.gz

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/.gitignore

@@ -1,5 +1,6 @@
 # Environment
 .env
+.env.local
 .venv/
 __pycache__/
 *.pyc

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: dataeval-flow
-Version: 0.2.2
+Version: 0.2.3
 Summary: DataEval Workflows container for data evaluation
 Project-URL: Homepage, https://github.com/aria-ml/dataeval-flow
 Project-URL: Documentation, https://dataeval-flow.readthedocs.io/
@@ -19,7 +19,7 @@ Classifier: Topic :: Scientific/Engineering
 Classifier: Typing :: Typed
 Requires-Python: <3.15,>=3.10
 Requires-Dist: click>=8.0
-Requires-Dist: dataeval>=1.1.0
+Requires-Dist: dataeval>=1.1.1
 Requires-Dist: datamaite[ic,od]==0.4.1
 Requires-Dist: numpy>=1.24.2
 Requires-Dist: pydantic>=2.0
@@ -49,11 +49,9 @@ Requires-Dist: onnx>=1.15; extra == 'onnx-cu130'
 Requires-Dist: onnxruntime-gpu>=1.27; (python_version >= '3.11' and extra != 'onnx-cu126') and extra == 'onnx-cu130'
 Requires-Dist: onnxruntime>=1.20; (python_version == '3.10' and extra != 'onnx-cu126') and extra == 'onnx-cu130'
 Provides-Extra: ontology
-Requires-Dist: dataeval[ontology]==1.1.0; extra == 'ontology'
+Requires-Dist: dataeval[ontology]>=1.1.1; extra == 'ontology'
 Provides-Extra: opencv
-Requires-Dist: opencv-python-headless>=4.8.0; extra == 'opencv'
-Provides-Extra: opencv-fips
-Requires-Dist: opencv-python-headless<4.13.0,>=4.8.0; extra == 'opencv-fips'
+Requires-Dist: opencv-python-headless<4.13.0,>=4.8.0; extra == 'opencv'
 Description-Content-Type: text/markdown
 
 <!-- markdownlint-disable MD041 -->
@@ -355,10 +353,10 @@ Dataset and model paths in config files are resolved relative to the data root (
 
 A `datasets` entry selects its loader with `format`:
 
-| `format`       | Structure                                                     | Split selection                    |
-| -------------- | ------------------------------------------------------------- | ---------------------------------- |
-| `huggingface`  | HuggingFace arrow dataset or dataset dict                     | `split:` (omit for a single split) |
-| `image_folder` | Directory of images, optionally one subdirectory per class    | n/a                                |
+| `format`       | Structure                                                     | Split selection                     |
+| -------------- | ------------------------------------------------------------- | ----------------------------------- |
+| `huggingface`  | HuggingFace arrow dataset or dataset dict                     | `split:` (omit for a single split)  |
+| `image_folder` | Directory of images, optionally one subdirectory per class    | n/a                                 |
 | `coco`         | COCO images plus a JSON annotation file                       | `annotations_file:` + `images_dir:` |
 | `yolo`         | Ultralytics root — `data.yaml` plus image/label trees         | `split:` (omit to load every split) |
 
@@ -549,7 +547,7 @@ pip install dataeval-flow
 `torchvision` is not installed by default — it is imported lazily and is only needed
 for preprocessing pipelines, the torchvision dataset adapter, and the TUI's transform
 discovery. Install it together with `torch` in step 1 so both come from the same index.
-Feature extras (`onnx`, `onnx-cu126`, `onnx-cu130`, `opencv`, `opencv-fips`, `app`, `ontology`) work normally under
+Feature extras (`onnx`, `onnx-cu126`, `onnx-cu130`, `opencv`, `app`, `ontology`) work normally under
 pip and are independent of the PyTorch variant:
 
 ```bash

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/README.md

@@ -297,10 +297,10 @@ Dataset and model paths in config files are resolved relative to the data root (
 
 A `datasets` entry selects its loader with `format`:
 
-| `format`       | Structure                                                     | Split selection                    |
-| -------------- | ------------------------------------------------------------- | ---------------------------------- |
-| `huggingface`  | HuggingFace arrow dataset or dataset dict                     | `split:` (omit for a single split) |
-| `image_folder` | Directory of images, optionally one subdirectory per class    | n/a                                |
+| `format`       | Structure                                                     | Split selection                     |
+| -------------- | ------------------------------------------------------------- | ----------------------------------- |
+| `huggingface`  | HuggingFace arrow dataset or dataset dict                     | `split:` (omit for a single split)  |
+| `image_folder` | Directory of images, optionally one subdirectory per class    | n/a                                 |
 | `coco`         | COCO images plus a JSON annotation file                       | `annotations_file:` + `images_dir:` |
 | `yolo`         | Ultralytics root — `data.yaml` plus image/label trees         | `split:` (omit to load every split) |
 
@@ -491,7 +491,7 @@ pip install dataeval-flow
 `torchvision` is not installed by default — it is imported lazily and is only needed
 for preprocessing pipelines, the torchvision dataset adapter, and the TUI's transform
 discovery. Install it together with `torch` in step 1 so both come from the same index.
-Feature extras (`onnx`, `onnx-cu126`, `onnx-cu130`, `opencv`, `opencv-fips`, `app`, `ontology`) work normally under
+Feature extras (`onnx`, `onnx-cu126`, `onnx-cu130`, `opencv`, `app`, `ontology`) work normally under
 pip and are independent of the PyTorch variant:
 
 ```bash

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/pyproject.toml

@@ -19,7 +19,7 @@ classifiers = [
 ]
 dependencies = [
     "click>=8.0",
-    "dataeval>=1.1.0",
+    "dataeval>=1.1.1",
     "datamaite[ic,od]==0.4.1",
     "numpy>=1.24.2",
     "pydantic>=2.0",
@@ -51,10 +51,9 @@ onnx-cu130 = [
   "onnxruntime>=1.20; python_version == '3.10' and extra != 'onnx-cu126'",  # CUDA 13.0 is not supported on Python 3.10
   "onnxruntime-gpu>=1.27; python_version >= '3.11' and extra != 'onnx-cu126'",
 ]
-opencv = ["opencv-python-headless>=4.8.0"]  
-opencv-fips = ["opencv-python-headless>=4.8.0,<4.13.0"]  # OpenCV 4.13 and higher are not FIPS compliant
+opencv = ["opencv-python-headless>=4.8.0,<4.13.0"]  # OpenCV 4.13 and higher are not FIPS compliant
 app = ["textual>=3.0"]
-ontology = ["dataeval[ontology]==1.1.0"]
+ontology = ["dataeval[ontology]>=1.1.1"]
 
 [project.urls]
 Homepage = "https://github.com/aria-ml/dataeval-flow"
@@ -167,10 +166,6 @@ conflicts = [
     { extra = "onnx" },
     { extra = "onnx-cu126" },
     { extra = "onnx-cu130" },
-  ],
-  [
-    { extra = "opencv" },
-    { extra = "opencv-fips" },
   ]
 ]
 
@@ -332,7 +327,7 @@ ignore = [
     "C416",
 ]
 fixable = ["ALL"]
-per-file-ignores = { "tests/*" = ["ANN", "S101", "D", "SLF001", "ARG002", "PT019"], "docs/*" = ["D", "E402", "E501", "S101"], "verification/*" = ["ANN", "S101", "D", "SLF001", "ARG001", "ARG002", "PT011", "PT019"] }
+per-file-ignores = { "tests/*" = ["ANN", "S101", "D", "SLF001", "ARG002", "PT019"], "docs/*" = ["D", "E402", "E501", "S101"], "verification/*" = ["ANN", "S101", "D", "SLF001", "ARG001", "ARG002", "PT011", "PT019"], "scripts/*" = ["C901", "D", "S"] }
 
 [tool.ruff.lint.pydocstyle]
 convention = "pep257"
@@ -347,5 +342,5 @@ indent-style = "space"
 # --- Codespell ---
 
 [tool.codespell]
-skip = "./*env*,./output,./htmlcov,./.nox,*.lock,./docs/source/notebooks/*,./docker/cosign.pub,./docker/Dockerfile.*,*.onnx"
+skip = "./*env*,./output,./specs,./htmlcov,./.nox,*.lock,./docs/source/notebooks/*,./docker/cosign.pub,./docker/Dockerfile.*,*.onnx"
 ignore-words-list = ["statics"]

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/src/dataeval_flow/__main__.py

@@ -3,11 +3,12 @@
 
 import argparse
 import logging
-import os
 import sys
 from pathlib import Path
 from typing import NoReturn
 
+from dataeval_flow._env import env_bool, env_choice, env_int, env_list, env_path
+
 _logger = logging.getLogger(__name__)
 
 
@@ -41,36 +42,43 @@ def _build_parser() -> argparse.ArgumentParser:
         "-c",
         "--config",
         type=Path,
-        default=None,
-        help="Path to config file or folder. If omitted, auto-discovers YAML/JSON at the data root.",
+        default=env_path("DATAEVAL_CONFIG"),
+        help=(
+            "Path to config file or folder (default: $DATAEVAL_CONFIG). If omitted, "
+            "auto-discovers YAML/JSON at the data root."
+        ),
     )
-
-    _data_default = os.environ.get("DATAEVAL_DATA")
     parser.add_argument(
         "-d",
         "--data",
         type=Path,
-        default=Path(_data_default) if _data_default else None,
+        default=env_path("DATAEVAL_DATA"),
         help="Root directory for data files (default: $DATAEVAL_DATA or current directory)",
     )
-
-    _output_default = os.environ.get("DATAEVAL_OUTPUT")
     parser.add_argument(
         "-o",
         "--output",
         type=Path,
-        default=Path(_output_default) if _output_default else None,
+        default=env_path("DATAEVAL_OUTPUT"),
         help="Path to output directory for artifacts (default: $DATAEVAL_OUTPUT or None).",
     )
-
-    _cache_default = os.environ.get("DATAEVAL_CACHE")
     parser.add_argument(
         "-k",
         "--cache",
         type=Path,
-        default=Path(_cache_default) if _cache_default else None,
+        default=env_path("DATAEVAL_CACHE"),
         help="Directory for disk-backed computation cache (default: $DATAEVAL_CACHE or None).",
     )
+    parser.add_argument(
+        "--log-format",
+        choices=("structured", "plain"),
+        default=env_choice("DATAEVAL_LOG_FORMAT", ("structured", "plain")) or "structured",
+        help=(
+            "Console log format (default: $DATAEVAL_LOG_FORMAT, else structured). "
+            "'structured' prefixes each record with an ISO-8601 UTC timestamp and level; "
+            "'plain' prints bare messages."
+        ),
+    )
     parser.add_argument(
         "-t",
         "--task",
@@ -85,10 +93,12 @@ def _build_parser() -> argparse.ArgumentParser:
     )
     parser.add_argument(
         "--fail-on-warning",
-        action="store_true",
+        action=argparse.BooleanOptionalAction,
+        default=env_bool("DATAEVAL_FAIL_ON_WARNING") or False,
         help=(
             "Exit non-zero when a task succeeds but reports findings that breached their "
-            "health thresholds. Off by default, so a warning stays a prompt to look."
+            "health thresholds (default: $DATAEVAL_FAIL_ON_WARNING, else off). "
+            "Use --no-fail-on-warning to override the environment."
         ),
     )
 
@@ -131,14 +141,14 @@ def _build_parser() -> argparse.ArgumentParser:
         "-d",
         "--data",
         type=Path,
-        default=None,
+        default=env_path("DATAEVAL_DATA"),
         help="Root directory for data files (default: $DATAEVAL_DATA or current directory)",
     )
     app_parser.add_argument(
         "-k",
         "--cache",
         type=Path,
-        default=None,
+        default=env_path("DATAEVAL_CACHE"),
         help="Directory for disk-backed computation cache (embeddings, metadata, stats).",
     )
 
@@ -157,6 +167,7 @@ def _build_parser() -> argparse.ArgumentParser:
         type=Path,
         help="Path to a result.json written by a run",
     )
+    # Do not read DATAEVAL_OUTPUT here: encoding defaults to stdout unless -o is explicitly specified.
     encoding_parser.add_argument(
         "-o",
         "--output",
@@ -221,15 +232,37 @@ def _list_workflows(name: str | None, *, as_json: bool) -> int:
     return 0
 
 
+def apply_env_defaults(args: argparse.Namespace) -> argparse.Namespace:
+    """Apply environment variable defaults that cannot be handled by argparse defaults.
+
+    For options with ``count`` or ``append`` actions (such as ``--verbose`` and
+    ``--task``), setting defaults in argparse causes command-line arguments to
+    increment or append to the default instead of overriding it. Applying these
+    environment variables post-parsing ensures CLI arguments take precedence.
+
+    ``DATAEVAL_TASKS`` applies only to headless execution, not subcommands.
+    """
+    if getattr(args, "verbose", 0) == 0:
+        args.verbose = env_int("DATAEVAL_VERBOSITY") or 0
+    if args.command is None and getattr(args, "task", None) is None:
+        args.task = env_list("DATAEVAL_TASKS")
+    return args
+
+
 def parse_args() -> argparse.Namespace:
-    """Parse CLI arguments."""
+    """Parse CLI arguments and apply environment defaults."""
     parser = _build_parser()
-    return parser.parse_args()
+    return apply_env_defaults(parser.parse_args())
 
 
 def main() -> NoReturn:
     """CLI entry point."""
-    args = parse_args()
+    try:
+        args = parse_args()
+    except ValueError as e:
+        # Report invalid environment variable values without a traceback.
+        print(f"ERROR: {e}", file=sys.stderr)
+        sys.exit(1)
 
     if args.command == "app":
         try:
@@ -256,7 +289,7 @@ def main() -> NoReturn:
         # refused, and the "commit it" hand-off when it succeeds — is dropped and the user
         # is left with a bare exit code. At INFO because this command's whole output is
         # one artifact and one sentence saying where it went.
-        setup_logging(verbosity=max(args.verbose, 2))
+        setup_logging(verbosity=max(args.verbose, 2), log_format=args.log_format)
         sys.exit(write_encoding(args.result, args.output, args.task))
 
     if args.command == "workflows":
@@ -273,7 +306,7 @@ def main() -> NoReturn:
     # resolution are reported even before the runner configures the file log.
     from dataeval_flow._logging import setup_logging
 
-    setup_logging(verbosity=args.verbose)
+    setup_logging(verbosity=args.verbose, log_format=args.log_format)
     try:
         from dataeval_flow.runner import run
 

dataeval_flow-0.2.3/src/dataeval_flow/_env.py

@@ -0,0 +1,74 @@
+"""Environment variable parsing for CLI options.
+
+Functions return ``None`` when the variable is unset or blank. Invalid values
+raise ``ValueError``.
+"""
+
+import os
+from pathlib import Path
+
+__all__ = ["env_bool", "env_choice", "env_int", "env_list", "env_path"]
+
+_TRUE = ("1", "true", "yes", "on")
+_FALSE = ("0", "false", "no", "off")
+
+
+def _raw(name: str) -> str | None:
+    """Return the stripped value of *name*, or None when unset or blank."""
+    value = os.environ.get(name)
+    if value is None:
+        return None
+    stripped = value.strip()
+    return stripped or None
+
+
+def env_path(name: str) -> Path | None:
+    """Read *name* as a filesystem path."""
+    raw = _raw(name)
+    return Path(raw) if raw is not None else None
+
+
+def env_int(name: str) -> int | None:
+    """Read *name* as an integer, raising ValueError if invalid."""
+    raw = _raw(name)
+    if raw is None:
+        return None
+    try:
+        return int(raw)
+    except ValueError:
+        raise ValueError(f"{name} must be an integer, got {raw!r}") from None
+
+
+def env_bool(name: str) -> bool | None:
+    """Read *name* as a boolean, raising ValueError if unrecognized."""
+    raw = _raw(name)
+    if raw is None:
+        return None
+    lowered = raw.lower()
+    if lowered in _TRUE:
+        return True
+    if lowered in _FALSE:
+        return False
+    accepted = ", ".join((*_TRUE, *_FALSE))
+    raise ValueError(f"{name} must be one of {accepted}; got {raw!r}")
+
+
+def env_list(name: str) -> list[str] | None:
+    """Read *name* as a comma-separated list, raising ValueError if empty."""
+    raw = _raw(name)
+    if raw is None:
+        return None
+    items = [item.strip() for item in raw.split(",") if item.strip()]
+    if not items:
+        raise ValueError(f"{name} was set but lists no values: {raw!r}")
+    return items
+
+
+def env_choice(name: str, choices: tuple[str, ...]) -> str | None:
+    """Read *name* and validate that its value is in *choices*."""
+    raw = _raw(name)
+    if raw is None:
+        return None
+    if raw not in choices:
+        raise ValueError(f"{name} must be one of {', '.join(choices)}; got {raw!r}")
+    return raw

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/src/dataeval_flow/_logging.py

@@ -32,22 +32,31 @@ _FILE_ROLE = "_dataeval_flow_file"
 _FILE_FORMAT = "%(asctime)s [%(levelname)-5s] %(name)s: %(message)s"
 _FILE_DATEFMT = "%Y-%m-%dT%H:%M:%SZ"
 
+# Console format for structured mode (ISO-8601 UTC timestamp and log level).
+_CONSOLE_FORMAT = "%(asctime)s [%(levelname)s] %(message)s"
+_CONSOLE_DATEFMT = "%Y-%m-%dT%H:%M:%SZ"
+
 
 class _ConsoleFormatter(logging.Formatter):
-    """Clean console formatter for CLI/container output.
+    """Console formatter for CLI and container output.
 
-    Strips the library-style prefix (timestamp, logger name) so user-facing
-    output reads like plain program output.  INFO/DEBUG records render as just
-    the message; WARNING and above are tagged with ``LEVEL:`` so problems stay
-    visible.  Tracebacks (``exc_info``) are appended via the standard machinery.
+    Supports ``structured`` mode (ISO-8601 UTC timestamp and level prefix)
+    and ``plain`` mode (bare messages, with a level prefix only for WARNING
+    and above).
     """
 
-    def __init__(self) -> None:
-        super().__init__("%(message)s")
-        self._warn_formatter = logging.Formatter("%(levelname)s: %(message)s")
+    def __init__(self, log_format: str = "structured") -> None:
+        if log_format == "plain":
+            super().__init__("%(message)s")
+            self._warn_formatter: logging.Formatter | None = logging.Formatter("%(levelname)s: %(message)s")
+        else:
+            super().__init__(_CONSOLE_FORMAT, datefmt=_CONSOLE_DATEFMT)
+            # Ensure timestamps format in UTC
+            self.converter = time.gmtime
+            self._warn_formatter = None
 
     def format(self, record: logging.LogRecord) -> str:
-        if record.levelno >= logging.WARNING:
+        if self._warn_formatter is not None and record.levelno >= logging.WARNING:
             return self._warn_formatter.format(record)
         return super().format(record)
 
@@ -72,7 +81,11 @@ class LogMessage:
         return self._str
 
 
-def setup_logging(output_dir: Path | None = None, verbosity: int = 0) -> None:
+def setup_logging(
+    output_dir: Path | None = None,
+    verbosity: int = 0,
+    log_format: str = "structured",
+) -> None:
     """Configure root logger with a clean console handler and optional file log.
 
     Additive and idempotent: the console (stdout) handler and the file handler
@@ -81,10 +94,11 @@ def setup_logging(output_dir: Path | None = None, verbosity: int = 0) -> None:
     file handler once ``output_dir`` is known — without ever duplicating a
     handler.
 
-    The console handler uses :class:`_ConsoleFormatter` (bare messages, with a
-    ``LEVEL:`` prefix only for warnings and above) so CLI/container output reads
-    like plain program output.  The file handler keeps the full timestamped,
-    named format at DEBUG for diagnostics.
+    The console handler uses :class:`_ConsoleFormatter` in either
+    ``"structured"`` mode (ISO-8601 UTC timestamp and level prefix) or
+    ``"plain"`` mode (bare messages, with a level prefix only for WARNING
+    and above). The file handler retains the timestamped format at DEBUG
+    level.
 
     Parameters
     ----------
@@ -93,6 +107,9 @@ def setup_logging(output_dir: Path | None = None, verbosity: int = 0) -> None:
         handler is created and output is console-only.
     verbosity : int
         Console verbosity level (0=quiet, 1=report, 2=+INFO, 3=+DEBUG).
+    log_format : str
+        Console format: ``"structured"`` (timestamp and level prefix) or
+        ``"plain"`` (bare messages). The file handler is unaffected.
     """
     global _initialized
     _initialized = True
@@ -110,7 +127,7 @@ def setup_logging(output_dir: Path | None = None, verbosity: int = 0) -> None:
             sh.setLevel(logging.INFO)
         else:
             sh.setLevel(logging.WARNING)
-        sh.setFormatter(_ConsoleFormatter())
+        sh.setFormatter(_ConsoleFormatter(log_format))
         setattr(sh, _CONSOLE_ROLE, True)
         root.addHandler(sh)
 

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/src/dataeval_flow/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.2.2'
-__version_tuple__ = version_tuple = (0, 2, 2)
+__version__ = version = '0.2.3'
+__version_tuple__ = version_tuple = (0, 2, 3)
 
 __commit_id__ = commit_id = None

{dataeval_flow-0.2.2 → dataeval_flow-0.2.3}/src/dataeval_flow/workflows/coverage/workflow.py

@@ -10,6 +10,7 @@ Assessments are organized by the dimension they evaluate:
 """
 
 import contextlib
+import inspect
 import logging
 import warnings
 from typing import Any
@@ -20,6 +21,7 @@ from dataeval import Metadata
 from dataeval.core import label_stats
 from dataeval.protocols import AnnotatedDataset, ObjectDetectionTarget
 from pydantic import BaseModel
+from typing_extensions import get_protocol_members
 
 from dataeval_flow.binning import attach_binning
 from dataeval_flow.cache import active_cache, get_or_compute_embeddings, get_or_compute_metadata
@@ -149,12 +151,25 @@ def _run_coverage(
     )
 
 
+#: Protocol members declared by ObjectDetectionTarget.
+_OD_TARGET_MEMBERS = get_protocol_members(ObjectDetectionTarget)
+
+_MISSING = object()
+
+
 def _is_object_detection(dataset: AnnotatedDataset[Any]) -> bool:
-    """Whether the dataset's targets are detections rather than one label per image."""
+    """Check whether dataset targets implement the ObjectDetectionTarget protocol.
+
+    Uses :func:`inspect.getattr_static` to verify member presence without
+    invoking property getters, ensuring consistent behavior across Python
+    versions.
+    """
     if len(dataset) == 0:
         return False
     datum = dataset[0]
-    return isinstance(datum, tuple) and len(datum) == 3 and isinstance(datum[1], ObjectDetectionTarget)
+    if not isinstance(datum, tuple) or len(datum) != 3:
+        return False
+    return all(inspect.getattr_static(datum[1], name, _MISSING) is not _MISSING for name in _OD_TARGET_MEMBERS)
 
 
 def _crop_view(