add tests for schema to pydantic matching

src/ga4gh/vrs/_internal/models.py

@@ -295,20 +295,37 @@ class ga4gh(_ValueObject.ga4gh):
         ]
 
 
+class LengthExpression(_ValueObject):
+    """An expression of a DNA, RNA, or protein polymer of known length but unspecified sequence."""
+
+    type: Literal['ReferenceLengthExpression'] = Field(
+        'ReferenceLengthExpression', description='MUST be "ReferenceLengthExpression"'
+    )
+    length: Union[Range, int] = Field(
+        ..., description='The number of residues of the expressed sequence.'
+    )
+
+    class ga4gh(_ValueObject.ga4gh):
+        keys = [
+            'length',
+            'type'
+        ]
+
+
 class ReferenceLengthExpression(_ValueObject):
-    """An expression of a length of a sequence from a repeating reference."""
+    """An expression sequence derived from a reference."""
 
     type: Literal['ReferenceLengthExpression'] = Field(
         'ReferenceLengthExpression', description='MUST be "ReferenceLengthExpression"'
     )
     length: Union[Range, int] = Field(
-        ..., description='The number of residues in the expressed sequence.'
+        ..., description='The number of residues of the expressed sequence.'
     )
     sequence: Optional[SequenceString] = Field(
         None, description='the Sequence encoded by the Reference Length Expression.'
     )
     repeatSubunitLength: int = Field(
-        None, description='The number of residues in the repeat subunit.'
+        None, description='The number of residues of the repeat subunit.'
     )
 
     class ga4gh(_ValueObject.ga4gh):
@@ -455,25 +472,25 @@ class ga4gh(_Ga4ghIdentifiableObject.ga4gh):
         ]
 
 
-class GenotypeMember(_ValueObject):
-    """A class for expressing the count of a specific `MolecularVariation` present
-    in-trans at a genomic locus represented by a `Genotype`.
-    """
-
-    type: Literal['GenotypeMember'] = Field('GenotypeMember', description='MUST be "GenotypeMember".')
-    count: Union[Range, int] = Field(
-        ..., description='The number of copies of the `variation` at a Genotype locus.'
-    )
-    variation: Union[Allele, Haplotype] = Field(
-        ..., description='A MolecularVariation at a Genotype locus.'
-    )
-
-    class ga4gh(_Ga4ghIdentifiableObject.ga4gh):
-        keys = [
-            'type',
-            'count',
-            'variation'
-        ]
+# class GenotypeMember(_ValueObject):
+#     """A class for expressing the count of a specific `MolecularVariation` present
+#     in-trans at a genomic locus represented by a `Genotype`.
+#     """
+#
+#     type: Literal['GenotypeMember'] = Field('GenotypeMember', description='MUST be "GenotypeMember".')
+#     count: Union[Range, int] = Field(
+#         ..., description='The number of copies of the `variation` at a Genotype locus.'
+#     )
+#     variation: Union[Allele, Haplotype] = Field(
+#         ..., description='A MolecularVariation at a Genotype locus.'
+#     )
+#
+#     class ga4gh(_Ga4ghIdentifiableObject.ga4gh):
+#         keys = [
+#             'type',
+#             'count',
+#             'variation'
+#         ]
 
 
 class MolecularVariation(RootModel):
@@ -488,31 +505,31 @@ class MolecularVariation(RootModel):
     )
 
 
-class Genotype(_VariationBase):
-    """A quantified set of _in-trans_ `MolecularVariation` at a genomic locus."""
-
-    type: Literal['Genotype'] = Field(
-        'Genotype',
-        description='MUST be "Genotype"'
-    )
-    # TODO members temporarily typed as List instead of Set + validate unique items
-    members: List[GenotypeMember] = Field(
-        ...,
-        description='Each GenotypeMember in `members` describes a MolecularVariation and the count of that variation at the locus.',
-        min_length=1,
-    )
-    count: Union[Range, int] = Field(
-        ...,
-        description='The total number of copies of all MolecularVariation at this locus, MUST be greater than or equal to the sum of GenotypeMember copy counts. If greater than the total counts, this implies additional MolecularVariation that are expected to exist but are not explicitly indicated.',
-    )
-
-    class ga4gh(_Ga4ghIdentifiableObject.ga4gh):
-        prefix = 'GT'
-        keys = [
-            'count',
-            'members',
-            'type'
-        ]
+# class Genotype(_VariationBase):
+#     """A quantified set of _in-trans_ `MolecularVariation` at a genomic locus."""
+#
+#     type: Literal['Genotype'] = Field(
+#         'Genotype',
+#         description='MUST be "Genotype"'
+#     )
+#     # TODO members temporarily typed as List instead of Set + validate unique items
+#     members: List[GenotypeMember] = Field(
+#         ...,
+#         description='Each GenotypeMember in `members` describes a MolecularVariation and the count of that variation at the locus.',
+#         min_length=1,
+#     )
+#     count: Union[Range, int] = Field(
+#         ...,
+#         description='The total number of copies of all MolecularVariation at this locus, MUST be greater than or equal to the sum of GenotypeMember copy counts. If greater than the total counts, this implies additional MolecularVariation that are expected to exist but are not explicitly indicated.',
+#     )
+#
+#     class ga4gh(_Ga4ghIdentifiableObject.ga4gh):
+#         prefix = 'GT'
+#         keys = [
+#             'count',
+#             'members',
+#             'type'
+#         ]
 
 
 class SequenceExpression(RootModel):
@@ -534,7 +551,7 @@ class Location(RootModel):
 
 
 class Variation(RootModel):
-    root: Union[Allele, CopyNumberChange, CopyNumberCount, Genotype, Haplotype] = Field(
+    root: Union[Allele, CopyNumberChange, CopyNumberCount, Haplotype] = Field(
         ...,
         json_schema_extra={
             'description': 'A representation of the state of one or more biomolecules.'
@@ -548,7 +565,7 @@ class SystemicVariation(RootModel):
     sample, or homologous chromosomes.
     """
 
-    root: Union[CopyNumberChange, CopyNumberCount, Genotype] = Field(
+    root: Union[CopyNumberChange, CopyNumberCount] = Field(
         ...,
         json_schema_extra={
             'description': 'A Variation of multiple molecules in the context of a system, e.g. a genome, sample, or homologous chromosomes.'

tests/validation/test_vrs_schema.py

@@ -0,0 +1,61 @@
+"""test that VRS Python model structures match VRS Schema
+"""
+import yaml
+from pathlib import Path
+from ga4gh.vrs import models
+
+ROOT_DIR = Path(__file__).parents[2]
+VRS_SCHEMA_DIR = ROOT_DIR / 'submodules' / 'vrs' / 'schema'
+
+with open(VRS_SCHEMA_DIR / 'vrs.yaml') as vrs_yaml:
+    VRS_SCHEMA = yaml.safe_load(vrs_yaml)
+
+VRS_CONCRETE_CLASSES = set()
+VRS_PRIMITIVES = set()
+
+for vrs_class in VRS_SCHEMA['$defs']:
+    cls_def = VRS_SCHEMA['$defs'][vrs_class]
+    if 'properties' in cls_def:
+        VRS_CONCRETE_CLASSES.add(vrs_class)
+    elif cls_def.get('type') in ['array', 'int', 'str']:
+        VRS_PRIMITIVES.add(vrs_class)
+
+NOT_IMPLEMENTED = ['Adjacency', 'Haplotype']  # Use this to skip testing of not-implemented classes
+                                              # TODO: Remove this once 2.0 models at beta
+
+
+def test_schema_models_exist():
+    """test that VRS Python covers the models defined by VRS
+    """
+    for vrs_class in VRS_CONCRETE_CLASSES | VRS_PRIMITIVES:
+        if vrs_class in NOT_IMPLEMENTED:
+            continue
+        assert getattr(models, vrs_class, False)
+
+
+def test_schema_class_fields_are_valid():
+    """test that VRS Python model fields match the VRS specification
+    """
+    for vrs_class in VRS_CONCRETE_CLASSES:
+        if vrs_class in NOT_IMPLEMENTED:
+            continue
+        schema_fields = set(VRS_SCHEMA['$defs'][vrs_class]['properties'])
+        pydantic_model = getattr(models, vrs_class)
+        assert set(pydantic_model.__fields__) == schema_fields, vrs_class
+
+
+def test_model_keys_are_valid():
+    """test that digest keys on Value Objects are valid and sorted
+    """
+    for vrs_class in VRS_CONCRETE_CLASSES:
+        if vrs_class in NOT_IMPLEMENTED:
+            continue
+        if VRS_SCHEMA['$defs'][vrs_class].get('ga4ghDigest', {}).get('keys', None) is None:
+            continue
+        pydantic_model = getattr(models, vrs_class)
+        try:
+            pydantic_model_digest_keys = pydantic_model.ga4gh.keys
+        except AttributeError:
+            raise AttributeError(vrs_class)
+        assert set(pydantic_model_digest_keys) == set(VRS_SCHEMA['$defs'][vrs_class]['ga4ghDigest']['keys']), vrs_class
+        assert pydantic_model_digest_keys == sorted(pydantic_model.ga4gh.keys), vrs_class