nmdc-schema 11.11.1__py3-none-any.whl → 11.12.0__py3-none-any.whl

nmdc_schema/nmdc_schema_validation_plugin.py

@@ -0,0 +1,103 @@
+from typing import Any, Optional, Iterator, Union
+
+from linkml.validator.plugins import ValidationPlugin
+from linkml.validator.report import ValidationResult, Severity
+from linkml.validator.validation_context import ValidationContext
+from linkml_runtime import SchemaView
+
+
+def _yield_quantity_value_objects(data: Any, path: Optional[list[Union[str, int]]] = None):
+    """Recursively yield QuantityValue objects from data."""
+    if path is None:
+        path = []
+    if isinstance(data, dict):
+        if data.get("type") == "nmdc:QuantityValue":
+            yield path, data
+        else:
+            # Recursively search nested dictionaries
+            for key, value in data.items():
+                yield from _yield_quantity_value_objects(value, path + [key])
+    elif isinstance(data, list):
+        # Handle lists of objects
+        for i, item in enumerate(data):
+            yield from _yield_quantity_value_objects(item, path + [i])
+
+
+class NmdcSchemaValidationPlugin(ValidationPlugin):
+    """A validation plugin which validates instances using NMDC-specific validation logic.
+
+    This plugin is designed to be used as a part of LinkML's validation framework and in conjunction
+    with the `JsonSchemaValidationPlugin` provided by LinkML. This plugin performs the following
+    additional checks:
+
+      1. Ensure that values for slots with range `QuantityValue` have a `unit` property that is in
+         agreement with the `storage_unit` annotation on the slot
+    """
+
+    def __init__(self, *args, **kwargs) -> None:
+        super().__init__(*args, **kwargs)
+        self._slot_storage_units_cache = {}
+
+    def _slot_storage_units(self, schema_view: SchemaView, slot_name: str) -> Optional[list[str]]:
+        """Get allowed storage_units for a slot."""
+        if slot_name in self._slot_storage_units_cache:
+            return self._slot_storage_units_cache[slot_name]
+
+        slot = schema_view.get_slot(slot_name)
+        if not slot or not slot.annotations:
+            self._slot_storage_units_cache[slot_name] = None
+            return None
+
+        storage_units = None
+        if "storage_units" in slot.annotations:
+            annotation_obj = slot.annotations["storage_units"]
+            if annotation_obj and hasattr(annotation_obj, "value"):
+                # Split on pipes for multiple units
+                storage_units = str(annotation_obj.value).split("|")
+
+        self._slot_storage_units_cache[slot_name] = storage_units
+        return storage_units
+
+    def process(self, instance: Any, context: ValidationContext) -> Iterator[ValidationResult]:
+        """Perform NMDC-specific schema validation on the provided instance
+
+        :param instance: The instance to validate
+        :param context: The validation context which provides a SchemaView artifact
+        :return: Iterator over validation results
+        :rtype: Iterator[ValidationResult]
+        """
+        for data_path, quantity_value in _yield_quantity_value_objects(instance):
+            # Get the `has_unit` property from the QuantityValue instance. If it is missing, yield a
+            # validation error and continue. This is slightly redundant with JSON Schema validation,
+            # but it ensures that we do not attempt further validation on the instance.
+            unit = quantity_value.get("has_unit")
+            str_data_path = '/'.join(str(p) for p in data_path)
+            if unit is None:
+                yield ValidationResult(
+                    type="nmdc-schema validation",
+                    severity=Severity.ERROR,
+                    instance=instance,
+                    instantiates=context.target_class,
+                    message=f"QuantityValue at /{str_data_path} is missing required 'has_unit' property",
+                )
+                continue
+
+            # Find the slot name by looking for the last string in the data path. If one cannot be
+            # found, skip further validation because we cannot determine the relevant slot.
+            try:
+                slot_name = next(key for key in reversed(data_path) if isinstance(key, str))
+            except StopIteration:
+                continue
+
+            storage_units = self._slot_storage_units(context.schema_view, slot_name)
+            if storage_units and unit not in storage_units:
+                yield ValidationResult(
+                    type="nmdc-schema validation",
+                    severity=Severity.ERROR,
+                    instance=instance,
+                    instantiates=context.target_class,
+                    message=(
+                        f"QuantityValue at /{str_data_path} has unit '{unit}' which is not allowed for "
+                        f"slot '{slot_name}' (allowed: {', '.join(storage_units)})"
+                    )
+                )

{nmdc_schema-11.11.1.dist-info → nmdc_schema-11.12.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nmdc_schema
-Version: 11.11.1
+Version: 11.12.0
 Summary: Schema resources for the National Microbiome Data Collaborative (NMDC)
 License: MIT
 License-File: LICENSE

{nmdc_schema-11.11.1.dist-info → nmdc_schema-11.12.0.dist-info}/RECORD

@@ -6,7 +6,7 @@ nmdc_schema/get_json_schema.py,sha256=5TFl1GCImyR5GHLK2hOz1MqqEyoRqhbukEs4x4AMYe
 nmdc_schema/get_nmdc_view.py,sha256=AwaPqb3H75ekL0hEV3SAhrUhFXVGFr1yFxQvBjPuRnY,401
 nmdc_schema/gold-to-mixs.sssom.tsv,sha256=gSo2vLjyNPfQPqZHeRKqS9uCe1XfGzzS-1_cTWxInJs,16935
 nmdc_schema/id_helpers.py,sha256=sX6skceaLBLTLBsSeOz-OwVkde4GvI5k5X1tnDVfhrY,3380
-nmdc_schema/migration_recursion.py,sha256=Tj1sfrJrZglwVC6fwaG4xvnFOt0P9AuBc9R4OHDD1Z0,5178
+nmdc_schema/migration_recursion.py,sha256=SuYl-z-ZmOhvD52VM_9BZgOtf5oDH09IES1JNoRqYFM,5302
 nmdc_schema/migrators/.docker/.env.example,sha256=eEgP5AMi0BRNdwkZ9gMYaGTg8_8vaZRv9sdmQpcmsKo,479
 nmdc_schema/migrators/.docker/docker-compose.yaml,sha256=5uzWYsBbtXzMwdQYQUEMVBDcJ0-gM7FxazSah-_QVIs,925
 nmdc_schema/migrators/.docker/mongoKeyFile,sha256=OmlML9o_bZVcVXQBCySb8IZoNmNxwqSP8lFHUBc-R4w,12
@@ -27,8 +27,8 @@ nmdc_schema/migrators/cli/README.md,sha256=DXoNj2Yhytl60_cpCJKP53p0yz2ygkPOQ_3HM
 nmdc_schema/migrators/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nmdc_schema/migrators/cli/create_migrator.py,sha256=gdey2oclbFvywazGMvfHkf2G-J3Bc3Lz8bnG3UHLajw,4491
 nmdc_schema/migrators/cli/migrator.py.template,sha256=7C-Ela_NrhkmgGY_Dz28ICF-xrT4ooolQ4kzmroJ9wc,369
-nmdc_schema/migrators/cli/run_migrator.py,sha256=LHl2S5XZCsC8aqRM4zcBADue8DneCR3eC7LhLXcgKJ8,4508
-nmdc_schema/migrators/helpers.py,sha256=AxmFh2zGBmJD7ndmVYJvRcEQjcCyNwZmDjDoFMdlf2o,4820
+nmdc_schema/migrators/cli/run_migrator.py,sha256=tICbb6TADAzkglqi87eGT6mlSNZXhMx1HW2CD1j4oe4,4362
+nmdc_schema/migrators/helpers.py,sha256=87w3O7meGeGV61_1Iy-UAik2FR4ERTmzSea7BygDcAY,5924
 nmdc_schema/migrators/migration_reporter.py,sha256=LV04b_Eu8UfTu3ZvKJi0Azyd1DHFauSR0918SyB9u-s,12284
 nmdc_schema/migrators/migrator_base.py,sha256=6GQ0uVQ27dQ3HseWHaTiTYf6QiWQFzcb6YRYVtBCyuw,3634
 nmdc_schema/migrators/migrator_from_10_0_0_to_10_1_2.py,sha256=aRKxRs5GrBin5oFrYaVSRQct90E3dsmXuTEZ6KO2oYY,2603
@@ -41,6 +41,7 @@ nmdc_schema/migrators/migrator_from_10_5_6_to_10_6_0.py,sha256=_oeQz_mXVC-vyyfUs
 nmdc_schema/migrators/migrator_from_10_6_0_to_10_7_0.py,sha256=-FduiWWIioYFQ2eST35NINok6cy6FoPuJ7-UXJPAi_E,450
 nmdc_schema/migrators/migrator_from_11_0_3_to_11_1_0.py,sha256=3hHXGts_sfre4KuIpGCUNyRMgI92_lXm_U2UStU-fdw,1184
 nmdc_schema/migrators/migrator_from_11_10_0_to_11_11_0.py,sha256=R4nUrhxRznukwKgkMQy8uuPzn-fE9AaQYNLLNsVOfTk,1354
+nmdc_schema/migrators/migrator_from_11_11_0_to_11_12_0.py,sha256=0lFUiE8SU2fhqlyN9lBg6-AuRiIw9qmOmg8IIhPmryM,1353
 nmdc_schema/migrators/migrator_from_11_1_0_to_11_2_0.py,sha256=xGaG0oilg0Xvq-P6YOZea3EPQU8fyfcKo-KZl4BhgoM,1183
 nmdc_schema/migrators/migrator_from_11_3_0_to_11_4_0.py,sha256=XDRsT9KrKlZTOGHQekESZ2XSYD2UDydF1PWTRIeqOVE,1985
 nmdc_schema/migrators/migrator_from_11_4_0_to_11_5_0.py,sha256=uiFi5USF23lpuAbkDhVufnXxLsM-Aixrj4nw7GY-XO4,357
@@ -83,6 +84,9 @@ nmdc_schema/migrators/partials/migrator_from_11_0_3_to_11_1_0/migrator_from_11_0
 nmdc_schema/migrators/partials/migrator_from_11_10_0_to_11_11_0/__init__.py,sha256=0Vyyg3SaqNsV426IFeWw1xgEDFAol5J57ggk7bDAAf4,1001
 nmdc_schema/migrators/partials/migrator_from_11_10_0_to_11_11_0/migrator_from_11_10_0_to_11_11_0_part_1.py,sha256=JTjgjW8GhpDuNQajNN2EBoDzm6OFJg9E42pR7HDqIiE,3595
 nmdc_schema/migrators/partials/migrator_from_11_10_0_to_11_11_0/migrator_from_11_10_0_to_11_11_0_part_2.py,sha256=cNgxp1XBVKSiQAAhUfLXWvEK8UyJTnveO_c_BqvlCVo,1986
+nmdc_schema/migrators/partials/migrator_from_11_11_0_to_11_12_0/__init__.py,sha256=MQg_WO1szHMiaDN7e6wEY_eo2o_BE1htJ7bF-mcfvIw,1001
+nmdc_schema/migrators/partials/migrator_from_11_11_0_to_11_12_0/migrator_from_11_11_0_to_11_12_0_part_1.py,sha256=ydGOhXXpsj6OkvLMz88w2Yzj9ibB1JinRWy-1jjSJYE,8668
+nmdc_schema/migrators/partials/migrator_from_11_11_0_to_11_12_0/migrator_from_11_11_0_to_11_12_0_part_2.py,sha256=l3CyePWflufFzSjteMv60iKpnpsaYNQIae5ICAtKZAs,4869
 nmdc_schema/migrators/partials/migrator_from_11_1_0_to_11_2_0/__init__.py,sha256=e-neqgmL6Zsy6h5PGmWPj9qIqrTm3yvNgo2l2i7Sduc,990
 nmdc_schema/migrators/partials/migrator_from_11_1_0_to_11_2_0/migrator_from_11_1_0_to_11_2_0_part_1.py,sha256=4nN_2iZqML84R9EgznIcj7FjIlRHSSM8XqLerUgMYbI,13230
 nmdc_schema/migrators/partials/migrator_from_11_1_0_to_11_2_0/migrator_from_11_1_0_to_11_2_0_part_2.py,sha256=HLGoX3KQZglHPODDO5FFHu2gjFjyBx8oGJ3D9PAkhNI,1628
@@ -102,17 +106,18 @@ nmdc_schema/migrators/partials/migrator_from_11_9_1_to_11_10_0/__init__.py,sha25
 nmdc_schema/migrators/partials/migrator_from_11_9_1_to_11_10_0/migrator_from_11_9_1_to_11_10_0_part_1.py,sha256=6MYmrB9yfJ3XFb2QabTcJeycK2tyVO3MtSItns2s9Zg,35421
 nmdc_schema/migrators/partials/migrator_from_11_9_1_to_11_10_0/migrator_from_11_9_1_to_11_10_0_part_2.py,sha256=wzard2sUTeRS7voVTfPt3_OtafTffnv0o0OyE1OBzA0,2987
 nmdc_schema/migrators/partials/migrator_from_11_9_1_to_11_10_0/migrator_from_11_9_1_to_11_10_0_part_3.py,sha256=crm2NEmn0xepEMKWHvCSkfX3G6hRn377f38rq24XlO8,3710
-nmdc_schema/nmdc-pydantic.py,sha256=T0c6vB4cui4HMuTBKboAka30kcINRpTkcMsWnv37Yro,1425683
-nmdc_schema/nmdc.py,sha256=W2SglKly9g1FEXrFiBCTsLjcQAXDO9-zv9Wa-r1FZkE,716776
-nmdc_schema/nmdc.schema.json,sha256=Q_rD8LIpepvcyUMzkVxcy4azuq4Vzq7wnxqyV_O955A,622632
+nmdc_schema/nmdc-pydantic.py,sha256=LHTNY4CkkQ5TDWY_BpfZ43BXjjs6bF61gKvB-5pnAck,1462437
+nmdc_schema/nmdc.py,sha256=QERSBlgEWWwk307GS-r_eDL7Nbu2aGHHxd9Jqdhc4G8,730020
+nmdc_schema/nmdc.schema.json,sha256=52TH3f0S0faJn5_oUH0epmdIWgQe4qpjCJNmaw3ki6c,629145
 nmdc_schema/nmdc_data.py,sha256=_wNKi5NDxuvvRsJEim2ialX7VJkDBJLRpiTOPpFHBm8,9608
-nmdc_schema/nmdc_materialized_patterns.json,sha256=MGJWCRnbofHB2992Z6PppkwJJHCqT9WAJh0zMzbvjlQ,901987
-nmdc_schema/nmdc_materialized_patterns.schema.json,sha256=JiEGqUy09djqkws28w_DR2yqvKt8srlUfqMaWsV4EvE,630915
-nmdc_schema/nmdc_materialized_patterns.yaml,sha256=BvlOiASFH0V0I8Rlxvrh91BEyE6CntkhuP-LPAeESB4,698448
+nmdc_schema/nmdc_materialized_patterns.json,sha256=7imCTMdsoo7qNwqz-bb-0IZMCsDVYlKxRJR9yMZ8yF8,937613
+nmdc_schema/nmdc_materialized_patterns.schema.json,sha256=aIl-4K7JczII1x64MZJnPJB0tyM_kBZHRtJeKYwMiE0,637428
+nmdc_schema/nmdc_materialized_patterns.yaml,sha256=VrQgI0MlWoyEY4wyIoxgVDiIASS9V6o5RLk17OSsT4Q,725518
+nmdc_schema/nmdc_schema_validation_plugin.py,sha256=rjtn1tYhngGyc1AV8vcT8BJ9a_MmHbcK3hRUf9dzPtA,4799
 nmdc_schema/nmdc_version.py,sha256=DsfEKnmN3LHR831WWELxyfZY6rnP9U8tfygamkrsaHY,2305
 nmdc_schema/validate_nmdc_json.py,sha256=PVJV2O1qQXMi206HaUKqRNLiLc164OpNYKPURSKN8_E,3148
-nmdc_schema-11.11.1.dist-info/METADATA,sha256=mdoZkUNR001uccR7KyaZUW07n_KlfDtAf9Pf_iu6URY,5899
-nmdc_schema-11.11.1.dist-info/WHEEL,sha256=M5asmiAlL6HEcOq52Yi5mmk9KmTVjY2RDPtO4p9DMrc,88
-nmdc_schema-11.11.1.dist-info/entry_points.txt,sha256=ioyD5P6op5Zy68bqg5MafLAryb73rfzyx8YXKItroC8,1531
-nmdc_schema-11.11.1.dist-info/licenses/LICENSE,sha256=ogEPNDSH0_dhiv_lT3ifVIdgIzHAqNA_SemnxUfPBJk,7048
-nmdc_schema-11.11.1.dist-info/RECORD,,
+nmdc_schema-11.12.0.dist-info/METADATA,sha256=3vl1EW8QvikZ0zNAKPjaQBmbi3lkRfqCgDFDvHVNES8,5899
+nmdc_schema-11.12.0.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
+nmdc_schema-11.12.0.dist-info/entry_points.txt,sha256=-UOl7vTEjS6eaJreIu_UenULpuKAYz8JiLeuKbEqPB8,2079
+nmdc_schema-11.12.0.dist-info/licenses/LICENSE,sha256=ogEPNDSH0_dhiv_lT3ifVIdgIzHAqNA_SemnxUfPBJk,7048
+nmdc_schema-11.12.0.dist-info/RECORD,,

{nmdc_schema-11.11.1.dist-info → nmdc_schema-11.12.0.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: poetry-core 2.2.0
+Generator: poetry-core 2.2.1
 Root-Is-Purelib: true
 Tag: py3-none-any

{nmdc_schema-11.11.1.dist-info → nmdc_schema-11.12.0.dist-info}/entry_points.txt

@@ -22,4 +22,12 @@ schema-pattern-linting=src.scripts.schema_pattern_linting:main
 schema-view-relation-graph=src.scripts.schema_view_relation_graph:cli
 scrutinize-elements=src.scripts.scrutinize_elements:process_schema_elements
 slot-range-type-reporter=src.scripts.slot_range_type_reporter:cli
+units-mongodb-analyze=units.scripts.mongodb_analyze_units:main
+units-schema-convert=units.scripts.schema_convert_to_ucum:main
+units-schema-extract=units.scripts.schema_extract_preferred_units:main
+units-schema-extract-slot-unit-pairs=units.scripts.schema_expand_storage_units:main
+units-schema-generate=units.scripts.schema_generate_yq_commands:main
+units-testdata-check=units.scripts.testdata_check_has_unit:main
+units-testdata-extract=units.scripts.testdata_extract_quantity_values:main
+units-ucum-validate=units.scripts.ucum_validate_units:main