Merge pull request #250 from microbiomedata/2180_protocol_type_migrator

Add migrator to check and add `type` slot to `Protocol` records

nmdc_schema/migrators/partials/migrator_from_10_2_0_to_11_0_0/__init__.py

@@ -18,6 +18,7 @@
     migrator_from_10_2_0_to_11_0_0_part_14,
     migrator_from_10_2_0_to_11_0_0_part_15,
     migrator_from_10_2_0_to_11_0_0_part_16,
+    migrator_from_10_2_0_to_11_0_0_part_17,
 )
 
 def get_migrator_classes() -> List[Type[MigratorBase]]:
@@ -52,4 +53,5 @@ def get_migrator_classes() -> List[Type[MigratorBase]]:
         migrator_from_10_2_0_to_11_0_0_part_14.Migrator,
         migrator_from_10_2_0_to_11_0_0_part_15.Migrator,
         migrator_from_10_2_0_to_11_0_0_part_16.Migrator,
+        migrator_from_10_2_0_to_11_0_0_part_17.Migrator,
     ]

nmdc_schema/migrators/partials/migrator_from_10_2_0_to_11_0_0/migrator_from_10_2_0_to_11_0_0_part_17.py

@@ -0,0 +1,50 @@
+from nmdc_schema.migrators.migrator_base import MigratorBase
+
+
+class Migrator(MigratorBase):
+    r"""
+    Migrates a database between two schemas.
+    """
+
+    _from_version = "XX"
+    _to_version = "FIXES_ISSUE_2180"
+
+    def upgrade(self) -> None:
+        r"""
+        Migrates the database from conforming to the original schema, to conforming to the new schema.
+
+        Should be run after migrator_from_10_2_0_to_11_0_0_part_16.py.
+        """
+        # Add a type slot on the Protocol class within the protocol_link slot on each document
+        collections_to_update = [
+            "data_generation_set",
+            "material_processing_set",
+            "collecting_biosamples_from_site_set",
+            "protocol_execution_set",
+            "storage_process_set",
+            "workflow_execution_set",
+            "study_set"
+        ] 
+
+        for collection_name in collections_to_update:
+            self.adapter.process_each_document(collection_name, [self.add_type_to_protocol_link])
+
+    def add_type_to_protocol_link(self, document: dict) -> dict:
+        r"""
+        Add a `type` field to the `Protocol` instance within the `protocol_link` field on each document.
+
+        >>> m = Migrator()
+        >>> m.add_type_to_protocol_link({'id': 123})  # no protocol_link field
+        {'id': 123}
+        >>> m.add_type_to_protocol_link({'id': 123, 'protocol_link': {'id': 456}})
+        {'id': 123, 'protocol_link': {'id': 456, 'type': 'nmdc:Protocol'}}
+        >>> m.add_type_to_protocol_link({'id': 123, 'protocol_link': {'id': 456, 'type': 'nmdc:Protocol'}}) # test: does not overwrite existing type slot
+        {'id': 123, 'protocol_link': {'id': 456, 'type': 'nmdc:Protocol'}}
+        """
+
+        self.logger.info(f"Starting migration of {document['id']}")
+        if "protocol_link" in document:
+            if "type" not in document["protocol_link"]:
+                document["protocol_link"]["type"] = "nmdc:Protocol"
+
+        return document

src/data/invalid/LibraryPreparation-missing-protocol-type.yaml

@@ -0,0 +1,12 @@
+# This record is missing the type field in the protocol_link slot
+end_date: '2017-12-12'
+has_input:
+- nmdc:procsm-12-px6qn983
+has_output:
+- nmdc:procsm-12-h63qmv56
+id: nmdc:libprp-12-1qp98z14
+processing_institution: Battelle
+start_date: 2014-08-05T18:40Z
+protocol_link:
+  name: BMI_metagenomicsSequencingSOP_v1
+type: nmdc:LibraryPreparation

src/data/valid/LibraryPreparation.yaml

@@ -0,0 +1,12 @@
+end_date: '2017-12-12'
+has_input:
+- nmdc:procsm-12-px6qn983
+has_output:
+- nmdc:procsm-12-h63qmv56
+id: nmdc:libprp-12-1qp98z14
+processing_institution: Battelle
+start_date: 2014-08-05T18:40Z
+protocol_link:
+  name: BMI_metagenomicsSequencingSOP_v1
+  type: nmdc:Protocol
+type: nmdc:LibraryPreparation