Skip to content

climate_ref_core.testing #

Test infrastructure for diagnostic testing.

This module provides: - TestCase and TestDataSpecification for defining test scenarios - YAML serialization for dataset catalogs (with local paths stored separately, under the cache)

TestCase #

A single test case for a diagnostic.

Test cases define scenarios for testing, with data resolved via: - requests: ESGF requests to fetch data (use ref test-cases fetch) - datasets_file: Path to a pre-built catalog YAML file

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
@frozen
class TestCase:
    """
    A single test case for a diagnostic.

    Test cases define scenarios for testing, with data resolved via:
    - `requests`: ESGF requests to fetch data (use `ref test-cases fetch`)
    - `datasets_file`: Path to a pre-built catalog YAML file
    """

    name: str
    """Name of the test case (e.g., 'default', 'short-timeseries')."""

    description: str
    """Human-readable description of what this test case covers."""

    requests: tuple[ESGFRequest, ...] | None = None
    """Optional ESGF requests to fetch data for this test case."""

    datasets_file: str | None = None
    """Path to YAML file with dataset specification (relative to package)."""

datasets_file = None class-attribute instance-attribute #

Path to YAML file with dataset specification (relative to package).

description instance-attribute #

Human-readable description of what this test case covers.

name instance-attribute #

Name of the test case (e.g., 'default', 'short-timeseries').

requests = None class-attribute instance-attribute #

Optional ESGF requests to fetch data for this test case.

TestCasePaths #

Path resolver for test case data.

Provides access to all paths within a test case directory: - catalog.yaml: Dataset metadata (tracked in git) - regression/: Regression outputs (tracked in git)

The local-paths sidecar (catalog.paths.yaml) is the exception: it lives under the dataset cache rather than in the package tree. The paths it records are machine-specific and point into the various caches (the dataset cache, the intake-esgf cache), so the sidecar belongs with those caches rather than with a checkout. It is shared by every checkout on the machine, and a fresh checkout cannot lose it.

Can be constructed from: - A diagnostic + test case name (auto-resolves provider's test-data dir) - An explicit test_data_dir + provider slug + diagnostic slug + test case name

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
@frozen
class TestCasePaths:
    """
    Path resolver for test case data.

    Provides access to all paths within a test case directory:
    - catalog.yaml: Dataset metadata (tracked in git)
    - regression/: Regression outputs (tracked in git)

    The local-paths sidecar (``catalog.paths.yaml``) is the exception: it lives under the
    dataset cache rather than in the package tree. The paths it records are machine-specific
    and point into the various caches (the dataset cache, the intake-esgf cache), so the
    sidecar belongs with those caches rather than with a checkout. It is shared by every
    checkout on the machine, and a fresh checkout cannot lose it.

    Can be constructed from:
    - A diagnostic + test case name (auto-resolves provider's test-data dir)
    - An explicit test_data_dir + provider slug + diagnostic slug + test case name
    """

    root: Path
    """The test case directory (test_data_dir / diagnostic_slug / test_case_name)."""

    provider_slug: str
    """Slug of the provider owning the diagnostic, used to namespace the paths sidecar."""

    @classmethod
    def from_diagnostic(cls, diagnostic: Diagnostic, test_case: str) -> TestCasePaths | None:
        """
        Create from a diagnostic, auto-resolving the provider's test-data directory.

        Returns None if the provider's test-data directory cannot be determined
        (e.g., not a development checkout).

        Parameters
        ----------
        diagnostic
            The diagnostic to get paths for
        test_case
            Test case name (e.g., 'default')
        """
        test_data_dir = _get_provider_test_data_dir(diagnostic)
        if test_data_dir is None:
            return None
        return cls(
            root=test_data_dir / diagnostic.slug / test_case,
            provider_slug=diagnostic.provider.slug,
        )

    @classmethod
    def from_test_data_dir(
        cls,
        test_data_dir: Path,
        diagnostic_slug: str,
        test_case: str,
        provider_slug: str,
    ) -> TestCasePaths:
        """
        Create from an explicit test data directory.

        Use this when you have a test_data_dir fixture (in tests) or
        know the base path explicitly.

        Parameters
        ----------
        test_data_dir
            Base test data directory (e.g., from test fixture)
        diagnostic_slug
            The diagnostic slug
        test_case
            Test case name (e.g., 'default')
        provider_slug
            The provider slug, used to namespace the paths sidecar
        """
        return cls(root=test_data_dir / diagnostic_slug / test_case, provider_slug=provider_slug)

    @property
    def catalog(self) -> Path:
        """Path to catalog.yaml."""
        return self.root / "catalog.yaml"

    @property
    def catalog_paths(self) -> Path:
        """
        Path to catalog.paths.yaml, under the dataset cache rather than the package tree.

        The sidecar maps each dataset to its local file, so it is a property of the machine,
        not of the checkout. Namespaced by provider because diagnostic slugs are only unique
        within a provider.
        """
        return (
            resolve_cache_dir("test-case-paths")
            / self.provider_slug
            / self.root.parent.name
            / self.root.name
            / "catalog.paths.yaml"
        )

    @property
    def regression(self) -> Path:
        """Path to regression/ directory."""
        return self.root / COMMITTED_DIRNAME

    @property
    def manifest(self) -> Path:
        """Path to manifest.json (the regression bundle manifest, tracked in git)."""
        return self.root / "manifest.json"

    @property
    def output(self) -> Path:
        """Path to the output/ directory (gitignored; holds materialised native slots)."""
        return self.root / "output"

    def output_slot(self, label: str = "latest") -> Path:
        """
        Path to a named output slot under ``output/`` (gitignored).

        A slot is a self-contained, inspectable snapshot of one execute/materialise:
        the curated native set (flat, at manifest-relative paths) plus a ``regression/``
        subdirectory holding the rebuilt committed bundle.
        ``latest`` (the default) is overwritten on every run; named slots persist so two
        runs can be diffed (e.g. ``--label before`` vs ``--label after``).

        Parameters
        ----------
        label
            Slot name. Must be a single path segment (no separators or ``..``).
        """
        return safe_path(label, self.output, label="output slot label", single_segment=True)

    @property
    def test_data_dir(self) -> Path:
        """Path to the test-data directory (parent of diagnostic slug dir)."""
        return self.root.parent.parent

    def exists(self) -> bool:
        """Check if the test case directory exists."""
        return self.root.exists()

    def create(self) -> None:
        """Create the test case directory if it doesn't exist."""
        self.root.mkdir(parents=True, exist_ok=True)

catalog property #

Path to catalog.yaml.

catalog_paths property #

Path to catalog.paths.yaml, under the dataset cache rather than the package tree.

The sidecar maps each dataset to its local file, so it is a property of the machine, not of the checkout. Namespaced by provider because diagnostic slugs are only unique within a provider.

manifest property #

Path to manifest.json (the regression bundle manifest, tracked in git).

output property #

Path to the output/ directory (gitignored; holds materialised native slots).

provider_slug instance-attribute #

Slug of the provider owning the diagnostic, used to namespace the paths sidecar.

regression property #

Path to regression/ directory.

root instance-attribute #

The test case directory (test_data_dir / diagnostic_slug / test_case_name).

test_data_dir property #

Path to the test-data directory (parent of diagnostic slug dir).

create() #

Create the test case directory if it doesn't exist.

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
272
273
274
def create(self) -> None:
    """Create the test case directory if it doesn't exist."""
    self.root.mkdir(parents=True, exist_ok=True)

exists() #

Check if the test case directory exists.

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
268
269
270
def exists(self) -> bool:
    """Check if the test case directory exists."""
    return self.root.exists()

from_diagnostic(diagnostic, test_case) classmethod #

Create from a diagnostic, auto-resolving the provider's test-data directory.

Returns None if the provider's test-data directory cannot be determined (e.g., not a development checkout).

Parameters:

Name Type Description Default
diagnostic Diagnostic

The diagnostic to get paths for

 required
test_case str

Test case name (e.g., 'default')

 required
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
@classmethod
def from_diagnostic(cls, diagnostic: Diagnostic, test_case: str) -> TestCasePaths | None:
    """
    Create from a diagnostic, auto-resolving the provider's test-data directory.

    Returns None if the provider's test-data directory cannot be determined
    (e.g., not a development checkout).

    Parameters
    ----------
    diagnostic
        The diagnostic to get paths for
    test_case
        Test case name (e.g., 'default')
    """
    test_data_dir = _get_provider_test_data_dir(diagnostic)
    if test_data_dir is None:
        return None
    return cls(
        root=test_data_dir / diagnostic.slug / test_case,
        provider_slug=diagnostic.provider.slug,
    )

from_test_data_dir(test_data_dir, diagnostic_slug, test_case, provider_slug) classmethod #

Create from an explicit test data directory.

Use this when you have a test_data_dir fixture (in tests) or know the base path explicitly.

Parameters:

Name Type Description Default
test_data_dir Path

Base test data directory (e.g., from test fixture)

 required
diagnostic_slug str

The diagnostic slug

 required
test_case str

Test case name (e.g., 'default')

 required
provider_slug str

The provider slug, used to namespace the paths sidecar

 required
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
@classmethod
def from_test_data_dir(
    cls,
    test_data_dir: Path,
    diagnostic_slug: str,
    test_case: str,
    provider_slug: str,
) -> TestCasePaths:
    """
    Create from an explicit test data directory.

    Use this when you have a test_data_dir fixture (in tests) or
    know the base path explicitly.

    Parameters
    ----------
    test_data_dir
        Base test data directory (e.g., from test fixture)
    diagnostic_slug
        The diagnostic slug
    test_case
        Test case name (e.g., 'default')
    provider_slug
        The provider slug, used to namespace the paths sidecar
    """
    return cls(root=test_data_dir / diagnostic_slug / test_case, provider_slug=provider_slug)

output_slot(label='latest') #

Path to a named output slot under output/ (gitignored).

A slot is a self-contained, inspectable snapshot of one execute/materialise: the curated native set (flat, at manifest-relative paths) plus a regression/ subdirectory holding the rebuilt committed bundle. latest (the default) is overwritten on every run; named slots persist so two runs can be diffed (e.g. --label before vs --label after).

Parameters:

Name Type Description Default
label str

Slot name. Must be a single path segment (no separators or ..).

 'latest'
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
def output_slot(self, label: str = "latest") -> Path:
    """
    Path to a named output slot under ``output/`` (gitignored).

    A slot is a self-contained, inspectable snapshot of one execute/materialise:
    the curated native set (flat, at manifest-relative paths) plus a ``regression/``
    subdirectory holding the rebuilt committed bundle.
    ``latest`` (the default) is overwritten on every run; named slots persist so two
    runs can be diffed (e.g. ``--label before`` vs ``--label after``).

    Parameters
    ----------
    label
        Slot name. Must be a single path segment (no separators or ``..``).
    """
    return safe_path(label, self.output, label="output slot label", single_segment=True)

TestDataSpecification #

Test data specification for a diagnostic.

Contains multiple named test cases for testing different input datasets.

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
@frozen
class TestDataSpecification:
    """
    Test data specification for a diagnostic.

    Contains multiple named test cases for testing different input datasets.
    """

    test_cases: tuple[TestCase, ...] = field(factory=tuple)
    """Collection of test cases for this diagnostic."""

    def get_case(self, name: str) -> TestCase:
        """
        Get a test case by name.

        Parameters
        ----------
        name
            Name of the test case to retrieve

        Returns
        -------
        TestCase
            The matching test case

        Raises
        ------
        StopIteration
            If no test case with that name exists
        """
        return next(tc for tc in self.test_cases if tc.name == name)

    def has_case(self, name: str) -> bool:
        """
        Check if a test case with the given name exists.

        Parameters
        ----------
        name
            Name of the test case to check

        Returns
        -------
        bool
            True if the test case exists
        """
        return any(tc.name == name for tc in self.test_cases)

    @property
    def case_names(self) -> list[str]:
        """Get names of all test cases."""
        return [tc.name for tc in self.test_cases]

case_names property #

Get names of all test cases.

test_cases = field(factory=tuple) class-attribute instance-attribute #

Collection of test cases for this diagnostic.

get_case(name) #

Get a test case by name.

Parameters:

Name Type Description Default
name str

Name of the test case to retrieve

 required

Returns:

Type Description
TestCase

The matching test case

Raises:

Type Description
StopIteration

If no test case with that name exists
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
def get_case(self, name: str) -> TestCase:
    """
    Get a test case by name.

    Parameters
    ----------
    name
        Name of the test case to retrieve

    Returns
    -------
    TestCase
        The matching test case

    Raises
    ------
    StopIteration
        If no test case with that name exists
    """
    return next(tc for tc in self.test_cases if tc.name == name)

has_case(name) #

Check if a test case with the given name exists.

Parameters:

Name Type Description Default
name str

Name of the test case to check

 required

Returns:

Type Description
bool

True if the test case exists
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
def has_case(self, name: str) -> bool:
    """
    Check if a test case with the given name exists.

    Parameters
    ----------
    name
        Name of the test case to check

    Returns
    -------
    bool
        True if the test case exists
    """
    return any(tc.name == name for tc in self.test_cases)

catalog_changed_since_regression(paths) #

Check if the catalog has changed since regression data was generated.

The baseline's input hash is read from manifest.json (catalog_hash), the single coupling record; there is no separate sidecar.

Returns True if: - No regression data exists (new test case) - No manifest, or the manifest records no catalog_hash (legacy regression data) - No catalog file exists - The current catalog hash differs from the one recorded in the manifest

Parameters:

Name Type Description Default
paths TestCasePaths

TestCasePaths for the test case

 required

Returns:

Type Description
bool

True if regression should be regenerated, False otherwise
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
def catalog_changed_since_regression(paths: TestCasePaths) -> bool:
    """
    Check if the catalog has changed since regression data was generated.

    The baseline's input hash is read from ``manifest.json`` (``catalog_hash``),
    the single coupling record; there is no separate sidecar.

    Returns True if:
    - No regression data exists (new test case)
    - No manifest, or the manifest records no ``catalog_hash`` (legacy regression data)
    - No catalog file exists
    - The current catalog hash differs from the one recorded in the manifest

    Parameters
    ----------
    paths
        TestCasePaths for the test case

    Returns
    -------
    :
        True if regression should be regenerated, False otherwise
    """
    if not paths.regression.exists():
        return True  # No regression data, needs to run
    if not paths.catalog.exists():
        return True  # No catalog file, needs to run
    if not paths.manifest.exists():
        return True  # No manifest, needs to run

    stored_hash = Manifest.load(paths.manifest).catalog_hash
    if stored_hash is None:
        return True  # Legacy manifest without a recorded catalog hash, needs to run

    return stored_hash != get_catalog_hash(paths.catalog)

collect_test_case_params(provider) #

Collect all diagnostic/test_case pairs from a provider for parameterized testing.

Returns a list of pytest.param objects with (diagnostic, test_case_name) tuples, each with an id of "{diagnostic.slug}/{test_case.name}".

Parameters:

Name Type Description Default
provider DiagnosticProvider

The diagnostic provider to collect test cases from

 required

Returns:

Type Description
list[ParameterSet]

List of pytest.param objects for use with @pytest.mark.parametrize
Example 
from climate_ref_core.testing import collect_test_case_params
from my_provider import provider

test_case_params = collect_test_case_params(provider)


@pytest.mark.parametrize("diagnostic,test_case_name", test_case_params)
def test_my_test(diagnostic, test_case_name): ...
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
def collect_test_case_params(provider: DiagnosticProvider) -> list[ParameterSet]:
    """
    Collect all diagnostic/test_case pairs from a provider for parameterized testing.

    Returns a list of pytest.param objects with (diagnostic, test_case_name) tuples,
    each with an id of "{diagnostic.slug}/{test_case.name}".

    Parameters
    ----------
    provider
        The diagnostic provider to collect test cases from

    Returns
    -------
    :
        List of pytest.param objects for use with @pytest.mark.parametrize

    Example
    -------
    ```python
    from climate_ref_core.testing import collect_test_case_params
    from my_provider import provider

    test_case_params = collect_test_case_params(provider)


    @pytest.mark.parametrize("diagnostic,test_case_name", test_case_params)
    def test_my_test(diagnostic, test_case_name): ...
    ```
    """
    import pytest  # noqa: PLC0415

    params: list[ParameterSet] = []
    for diagnostic in provider.diagnostics():
        if diagnostic.test_data_spec is None:
            continue
        for test_case in diagnostic.test_data_spec.test_cases:
            params.append(
                pytest.param(
                    diagnostic,
                    test_case.name,
                    id=f"{diagnostic.slug}/{test_case.name}",
                )
            )
    return params

excluded_test_case_diagnostics() #

Diagnostic identifiers to skip when fetching or running test cases.

Reads the REF_TEST_CASES_SKIP environment variable as a comma-separated list of diagnostic slugs or provider/diagnostic pairs.

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
39
40
41
42
43
44
45
46
47
def excluded_test_case_diagnostics() -> set[str]:
    """
    Diagnostic identifiers to skip when fetching or running test cases.

    Reads the ``REF_TEST_CASES_SKIP`` environment variable
    as a comma-separated list of ``diagnostic`` slugs or ``provider/diagnostic`` pairs.
    """
    raw = os.environ.get("REF_TEST_CASES_SKIP", "")
    return {token.strip() for token in raw.split(",") if token.strip()}

get_catalog_hash(path) #

Get the hash stored in an existing catalog file.

Parameters:

Name Type Description Default
path Path

Path to the catalog YAML file

 required

Returns:

Type Description
str | None

The hash string if found, None if file doesn't exist or has no hash
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
def get_catalog_hash(path: Path) -> str | None:
    """
    Get the hash stored in an existing catalog file.

    Parameters
    ----------
    path
        Path to the catalog YAML file

    Returns
    -------
    :
        The hash string if found, None if file doesn't exist or has no hash
    """
    if not path.exists():
        return None
    with open(path) as f:
        data = yaml.safe_load(f)
    if data is None:
        return None
    hash_value = data.get("_metadata", {}).get("hash")
    return str(hash_value) if hash_value is not None else None

is_test_case_excluded(provider_slug, diagnostic_slug) #

Return True if the diagnostic is excluded via REF_TEST_CASES_SKIP.

Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
50
51
52
53
def is_test_case_excluded(provider_slug: str, diagnostic_slug: str) -> bool:
    """Return True if the diagnostic is excluded via ``REF_TEST_CASES_SKIP``."""
    excluded = excluded_test_case_diagnostics()
    return diagnostic_slug in excluded or f"{provider_slug}/{diagnostic_slug}" in excluded

load_datasets_from_yaml(path, paths_file) #

Load ExecutionDatasetCollection from a YAML file.

The YAML file structure:

cmip6:
  slug_column: instance_id
  selector:
    source_id: ACCESS-ESM1-5
  datasets:
    - instance_id: CMIP6.CMIP...
      variable_id: tas
      filename: tas_Amon_ACCESS-ESM1-5_historical_r1i1p1f1_gn_185001-201412.nc
      # ... other metadata

Paths are loaded from paths_file if it exists, allowing the main catalog to be version-controlled while paths remain machine-specific. Multi-file datasets have multiple rows with paths keyed by {instance_id}::{filename}.

Parameters:

Name Type Description Default
path Path

Path to the catalog YAML file

 required
paths_file Path

Path to the local-paths sidecar (see :attr:TestCasePaths.catalog_paths). If it does not exist, the loaded datasets carry no path column.

 required
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
def load_datasets_from_yaml(path: Path, paths_file: Path) -> ExecutionDatasetCollection:
    """
    Load ExecutionDatasetCollection from a YAML file.

    The YAML file structure:

    ```yaml
    cmip6:
      slug_column: instance_id
      selector:
        source_id: ACCESS-ESM1-5
      datasets:
        - instance_id: CMIP6.CMIP...
          variable_id: tas
          filename: tas_Amon_ACCESS-ESM1-5_historical_r1i1p1f1_gn_185001-201412.nc
          # ... other metadata
    ```

    Paths are loaded from ``paths_file`` if it exists, allowing the main catalog to be
    version-controlled while paths remain machine-specific. Multi-file datasets have
    multiple rows with paths keyed by `{instance_id}::{filename}`.

    Parameters
    ----------
    path
        Path to the catalog YAML file
    paths_file
        Path to the local-paths sidecar (see :attr:`TestCasePaths.catalog_paths`).
        If it does not exist, the loaded datasets carry no ``path`` column.
    """
    with open(path) as f:
        data = yaml.safe_load(f)

    paths_map: dict[str, str] = {}
    if paths_file.exists():
        with open(paths_file) as f:
            paths_map = yaml.safe_load(f) or {}

    collections: dict[SourceDatasetType | str, DatasetCollection] = {}

    for source_type_str, source_data in data.items():
        if source_type_str == "_metadata":
            continue  # Skip metadata section
        source_type = SourceDatasetType(source_type_str)
        selector_dict = source_data.get("selector", {})
        selector: Selector = tuple(sorted(selector_dict.items()))
        datasets_list = source_data.get("datasets", [])
        slug_column = source_data.get("slug_column", "instance_id")

        # Merge paths from paths file using composite key
        for dataset in datasets_list:
            instance_id = dataset.get(slug_column)
            filename = dataset.get("filename")
            if instance_id and filename:
                # Try composite key first (new format for multi-file datasets)
                composite_key = f"{instance_id}::{filename}"
                if composite_key in paths_map:
                    dataset["path"] = paths_map[composite_key]
                elif instance_id in paths_map:
                    # Fall back to simple key for backward compatibility
                    dataset["path"] = paths_map[instance_id]
            elif instance_id and instance_id in paths_map:
                # Legacy format without filename
                dataset["path"] = paths_map[instance_id]

        collections[source_type] = DatasetCollection(
            datasets=pd.DataFrame(datasets_list),
            slug_column=slug_column,
            selector=selector,
        )

    return ExecutionDatasetCollection(collections)

save_datasets_to_yaml(datasets, path, paths_file, *, force=False) #

Save ExecutionDatasetCollection to a YAML file.

Paths are saved to a separate sidecar file to allow the main catalog to be version-controlled while paths remain machine-specific.

Multi-file datasets (e.g., time-chunked data) are stored as multiple rows, one per file. Paths are keyed by {instance_id}::{filename} to support multiple files per dataset.

By default, the catalog is only written if the content has changed (detected via hash comparison). Use force=True to always write.

The paths sidecar is regenerated on every save, even when the catalog content is unchanged. Local cache contents can change independently of the version-controlled catalog, so retaining an existing sidecar can leave missing or stale paths behind.

Parameters:

Name Type Description Default
datasets ExecutionDatasetCollection

The datasets to save

 required
path Path

Path to write the YAML file

 required
paths_file Path

Path to write the local-paths sidecar (see :attr:TestCasePaths.catalog_paths)

 required
force bool

If True, always write the catalog even if unchanged

 False

Returns:

Type Description
bool

True if the catalog was (re)written, False if the catalog was left unchanged (the paths sidecar may still have been regenerated).
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
def save_datasets_to_yaml(
    datasets: ExecutionDatasetCollection,
    path: Path,
    paths_file: Path,
    *,
    force: bool = False,
) -> bool:
    """
    Save ExecutionDatasetCollection to a YAML file.

    Paths are saved to a separate sidecar file to allow the main catalog to be
    version-controlled while paths remain machine-specific.

    Multi-file datasets (e.g., time-chunked data) are stored as multiple rows,
    one per file. Paths are keyed by `{instance_id}::{filename}` to support
    multiple files per dataset.

    By default, the catalog is only written if the content has changed
    (detected via hash comparison). Use `force=True` to always write.

    The paths sidecar is regenerated on every save, even when the catalog content is
    unchanged. Local cache contents can change independently of the version-controlled
    catalog, so retaining an existing sidecar can leave missing or stale paths behind.

    Parameters
    ----------
    datasets
        The datasets to save
    path
        Path to write the YAML file
    paths_file
        Path to write the local-paths sidecar (see :attr:`TestCasePaths.catalog_paths`)
    force
        If True, always write the catalog even if unchanged

    Returns
    -------
    :
        True if the catalog was (re)written, False if the catalog was left unchanged
        (the paths sidecar may still have been regenerated).
    """
    new_hash = datasets.hash

    data, paths_map = _serialise_datasets(datasets)

    if not force and get_catalog_hash(path) == new_hash:
        # Keep the tracked catalog byte-identical, but always refresh its machine-local
        # paths. An existing sidecar may be partial or point at an old cache location.
        _write_paths_file(paths_file, paths_map)
        logger.info(f"Catalog unchanged, refreshed paths file: {paths_file}")
        return False

    path.parent.mkdir(parents=True, exist_ok=True)
    with open(path, "w") as f:
        yaml.dump(data, f, default_flow_style=False, sort_keys=False)
    _write_paths_file(paths_file, paths_map)
    logger.info(f"Saved catalog to {path} (paths: {paths_file})")
    return True

validate_catalog_paths(path, paths_file) #

Load a catalog and check that every row resolves to a local file.

Loading remains permissive because callers also use catalogs for metadata-only operations. Fetch and execution entry points can call this function when they require usable input files.

Parameters:

Name Type Description Default
path Path

Path to the catalog YAML file.

 required
paths_file Path

Path to the machine-local paths sidecar.

 required

Returns:

Type Description
ExecutionDatasetCollection

The loaded datasets, so callers need not parse the catalog a second time.

Raises:

Type Description
DatasetResolutionError

If the sidecar is absent for a non-empty catalog, incomplete, or points to missing files.
Source code in packages/climate-ref-core/src/climate_ref_core/testing.py 
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
def validate_catalog_paths(path: Path, paths_file: Path) -> ExecutionDatasetCollection:
    """Load a catalog and check that every row resolves to a local file.

    Loading remains permissive because callers also use catalogs for metadata-only
    operations. Fetch and execution entry points can call this function when they
    require usable input files.

    Parameters
    ----------
    path
        Path to the catalog YAML file.
    paths_file
        Path to the machine-local paths sidecar.

    Returns
    -------
    :
        The loaded datasets, so callers need not parse the catalog a second time.

    Raises
    ------
    DatasetResolutionError
        If the sidecar is absent for a non-empty catalog, incomplete, or points to missing files.
    """
    from climate_ref_core.exceptions import DatasetResolutionError  # noqa: PLC0415

    try:
        datasets = load_datasets_from_yaml(path, paths_file)
    except Exception as e:
        raise DatasetResolutionError(f"Could not load catalog paths for {path}: {e}") from e

    has_rows = any(len(collection.datasets) for collection in datasets.values())
    if has_rows and not paths_file.exists():
        raise DatasetResolutionError(
            f"Paths file is missing for catalog {path}: {paths_file}. "
            "Run `ref test-cases fetch` to rebuild it."
        )

    missing_entries: list[str] = []
    missing_files: list[str] = []
    for source_type, collection in datasets.items():
        for _, dataset in collection.datasets.iterrows():
            instance_id = dataset.get(collection.slug_column)
            filename = dataset.get("filename")
            key = f"{instance_id}::{filename}" if instance_id and filename else instance_id
            resolved_path = dataset.get("path")
            label = f"{source_type.value}:{key or '<unknown dataset>'}"
            if not isinstance(resolved_path, (str, Path)) or not str(resolved_path):
                missing_entries.append(label)
            elif not Path(resolved_path).is_file():
                missing_files.append(f"{label} -> {resolved_path}")

    if missing_entries or missing_files:
        problems = []
        if missing_entries:
            problems.append(f"missing path entries: {', '.join(missing_entries)}")
        if missing_files:
            problems.append(f"files not found: {', '.join(missing_files)}")
        raise DatasetResolutionError(
            f"Catalog paths are incomplete for {path} ({'. '.join(problems)}). "
            "Run `ref test-cases fetch` to rebuild the paths file."
        )
    return datasets