feat: automatically detect input format for template construction and collection CLI

src/cabinetry/cli/__init__.py

@@ -4,7 +4,7 @@
 import json
 import logging
 import pathlib
-from typing import Any, List, Optional, Tuple
+from typing import Any, List, Literal, Optional, Tuple
 
 import click
 import yaml
@@ -48,15 +48,36 @@ def cabinetry() -> None:
     default="uproot",
     help="backend for histogram production (default: uproot)",
 )
-def templates(config: io.TextIOWrapper, method: str) -> None:
+@click.option(
+    "--input_type",
+    type=click.Choice(["ntuple", "histogram"]),
+    help="type of input files (overrides automatic detection)",
+)
+def templates(
+    config: io.TextIOWrapper,
+    method: str,
+    input_type: Optional[Literal["ntuple", "histogram"]],
+) -> None:
     """Produces template histograms.
 
     CONFIG: path to cabinetry configuration file
     """
     _set_logging()
     cabinetry_config = yaml.safe_load(config)
     cabinetry_configuration.validate(cabinetry_config)
-    cabinetry_templates.build(cabinetry_config, method=method)
+
+    if input_type is None:
+        # detect input type
+        input_type = (
+            "ntuple"
+            if cabinetry_configuration._input_is_ntuple(cabinetry_config)
+            else "histogram"
+        )
+
+    if input_type == "ntuple":
+        cabinetry_templates.build(cabinetry_config, method=method)
+    else:
+        cabinetry_templates.collect(cabinetry_config, method=method)
 
 
 @click.command()

src/cabinetry/configuration.py

@@ -269,3 +269,19 @@ def region_dict(config: Dict[str, Any], region_name: str) -> Dict[str, Any]:
     if len(regions) > 1:
         log.error(f"found more than one region with name {region_name}")
     return regions[0]
+
+
+def _input_is_ntuple(config: Dict[str, Any]) -> bool:
+    """Checks if input files for given config are ntuples or histograms.
+
+    This relies on the "Tree" property of samples, which is required for ntuple inputs.
+    In case this property is used for histogram inputs (where it does nothing), this
+    function will mistakenly identify this as ntuple input type.
+
+    Args:
+        config (Dict[str, Any]): cabinetry configuration file
+
+    Returns:
+        bool: whether inputs are of ntuple type
+    """
+    return all([sample.get("Tree") is not None for sample in config["Samples"]])

tests/cli/test_cli.py

@@ -36,9 +36,22 @@ def test_cabinetry():
 
 
 # using autospec to catch changes in public API
+@mock.patch("cabinetry.templates.collect", autospec=True)
 @mock.patch("cabinetry.templates.build", autospec=True)
+@mock.patch(
+    "cabinetry.configuration._input_is_ntuple",
+    side_effect=[True, False],
+    autospec=True,
+)
 @mock.patch("cabinetry.configuration.validate", autospec=True)
-def test_templates(mock_validate, mock_create_histograms, cli_helpers, tmp_path):
+def test_templates(
+    mock_validate,
+    mock_is_ntuple,
+    mock_create_histograms,
+    mock_collect_histograms,
+    cli_helpers,
+    tmp_path,
+):
     config = {"General": {"Measurement": "test_config"}}
 
     config_path = str(tmp_path / "config.yml")
@@ -50,12 +63,21 @@ def test_templates(mock_validate, mock_create_histograms, cli_helpers, tmp_path)
     result = runner.invoke(cli.templates, [config_path])
     assert result.exit_code == 0
     assert mock_validate.call_args_list == [((config,), {})]
+    assert mock_is_ntuple.call_args_list == [((config,), {})]
     assert mock_create_histograms.call_args_list == [((config,), {"method": "uproot"})]
+    assert mock_collect_histograms.call_count == 0
 
-    # different method
+    # different method, histogram input via side effect
     result = runner.invoke(cli.templates, ["--method", "unknown", config_path])
     assert result.exit_code == 0
-    assert mock_create_histograms.call_args == ((config,), {"method": "unknown"})
+    assert mock_create_histograms.call_count == 1  # no new call
+    assert mock_collect_histograms.call_args == ((config,), {"method": "unknown"})
+
+    # input type specified
+    result = runner.invoke(cli.templates, ["--input_type", "ntuple", config_path])
+    assert mock_is_ntuple.call_count == 2  # no new call
+    assert mock_create_histograms.call_count == 2
+    assert mock_collect_histograms.call_count == 1  # no new call
 
 
 @mock.patch("cabinetry.templates.postprocess", autospec=True)

tests/test_configuration.py

@@ -295,3 +295,11 @@ def test_region_dict(caplog):
 
     with pytest.raises(ValueError, match="region abc not found in config"):
         configuration.region_dict(config, "abc")
+
+
+def test__input_is_ntuple():
+    config_ntuple = {"Samples": [{"Name": "", "Tree": ""}]}
+    config_histogram = {"Samples": [{"Name": ""}]}
+
+    assert configuration._input_is_ntuple(config_ntuple) is True
+    assert configuration._input_is_ntuple(config_histogram) is False