Identifier Preprocessing

fairscape_models/activity.py

@@ -1,10 +1,10 @@
-from pydantic import BaseModel, Field, ConfigDict, model_validator
-from typing import Optional, List, Union
+from pydantic import Field, ConfigDict, field_validator
+from typing import Optional, List, Union, Any
 
-from fairscape_models.fairscape_base import IdentifierValue
+from fairscape_models.fairscape_base import IdentifierValue, Identifier, extractGUID
 from fairscape_models._version import __version__
 
-class Activity(BaseModel):
+class Activity(Identifier):
     """Base class for Activity types (Computation, Annotation, Experiment)"""
     guid: str = Field(alias="@id")
     name: str
@@ -20,3 +20,9 @@ class Activity(BaseModel):
     fairscapeVersion: str = __version__
 
     model_config = ConfigDict(extra="allow", populate_by_name=True)
+
+#   TODO extract guids from all subfields
+#    @field_validator('generated', mode='before')
+#    @classmethod
+#    def extract_guid_generated(cls, value: Any) -> Any:
+#        pass

fairscape_models/biochem_entity.py

@@ -1,10 +1,10 @@
-from pydantic import BaseModel, Field, ConfigDict, model_validator
+from pydantic import Field, ConfigDict, model_validator
 from typing import Optional, List, Union
 
-from fairscape_models.fairscape_base import IdentifierValue, IdentifierPropertyValue
+from fairscape_models.fairscape_base import IdentifierValue, IdentifierPropertyValue, Identifier
 from fairscape_models._version import __version__
 
-class BioChemEntity(BaseModel):
+class BioChemEntity(Identifier):
     """ Pydantic model for the Schema.org BioChemEntity datatype
 
     This class can apply to Protiens, Genes, Chemical Entities, or Biological Samples

fairscape_models/digital_object.py

@@ -1,12 +1,12 @@
-from pydantic import BaseModel, Field, ConfigDict, model_validator
+from pydantic import Field, ConfigDict, field_validator
 from typing import Optional, List, Union
 
-from fairscape_models.fairscape_base import IdentifierValue
+from fairscape_models.fairscape_base import IdentifierValue, Identifier
 from fairscape_models._version import __version__
 
-class DigitalObject(BaseModel):
+class DigitalObject(Identifier):
     """Base class for DigitalObject types (Dataset, Software, MLModel)"""
-    guid: str = Field(alias="@id")
+    guid: str = Field(alias="@id", pattern="^ark:[0-9]{5}/.+$")
     name: str
     metadataType: Optional[Union[List[str], str]] = Field(default=['prov:Entity', "https://w3id.org/EVI#DigitalObject"], alias="@type")
     author: Union[str, IdentifierValue, List[Union[str, IdentifierValue]]]
@@ -28,3 +28,5 @@ class DigitalObject(BaseModel):
     wasAttributedTo: Optional[List[Union[str, IdentifierValue]]] = Field(default=[], alias="prov:wasAttributedTo")
 
     model_config = ConfigDict(extra="allow", populate_by_name=True)
+
+#   TODO extract guids from all subfields

fairscape_models/fairscape_base.py

@@ -2,15 +2,18 @@
     BaseModel, 
     ConfigDict,
     Field,
-    BeforeValidator
+    BeforeValidator,
+    field_validator
 )
 from pydantic.networks import AnyUrl
 from typing import (
     List,
     Optional,
     Dict,
-    Union
+    Union,
+    Any
 )
+import re
 from typing_extensions import Annotated
 from enum import Enum
 
@@ -67,6 +70,20 @@
     }
 }
 
+def extractGUID(inputString: str | None) -> str|None:
+    """
+    Given an input ARK extract the normalized ARK, if validation fails return the input.
+    """
+    try:
+        match = re.search(
+            pattern="ark:[0-9]{5}/.+$",
+            string=inputString
+        )
+        return match.group()
+    except AttributeError:
+        return inputString
+
+
 class ClassType(str, Enum):
     DATASET = 'Dataset'
     SOFTWARE = 'Software'
@@ -123,16 +140,55 @@ class IdentifierPropertyValue(BaseModel):
 
 
 class Identifier(BaseModel):
+    """     
+    The Base Model for any Metadata element in FAIRSCAPE.
+
+    Every instance must have a GUID in the form of an ARK (archival resource key), 
+    a metadata type (https://www.w3.org/TR/json-ld/#specifying-the-type), and a name specified as a string.
+    Every model must have these attributes, and may have any other attributes as specified by the `ConfigDict(extra='allow')`.
+
+    For the guid property, preprocessing is preformed by the field validator `Identifier.extract_guid`.
+    This method preforms a regex search to find the identifier within the passed value. 
+    As ARKs may be specified as full IRIS or URLs pointing to several different resolvers, arks are stripped.
+    The guid for all fairscape_models clases should follow the regex `"ark:[0-9]{5}/.+$"`.
+
+    This guid preprocessing is also preformed on isPartOf.
+    """
     model_config = ConfigDict(extra='allow')
     guid: str = Field(
         title="guid",
-        alias="@id"
+        alias="@id",
+        pattern=IdentifierPattern
     )
-    metadataType: ValidatedClassType = Field(
+    metadataType: Optional[Union[List[str], str]] = Field(
         title="metadataType",
         alias="@type"
     )
     name: str = Field(...)
+    isPartOf: Optional[Union[List[str], str]] = Field(default=[])
+
+    @field_validator('guid', mode='before')
+    @classmethod
+    def extract_guid(cls, value: Any)-> Any:
+        """ 
+        Extract the ARK from the guid field, runs before validation against regex.
+        """
+        return extractGUID(value)
+
+
+    @field_validator('isPartOf', mode='before')
+    @classmethod
+    def extract_guid_is_part_of(cls, value: Any)-> Any:
+        """
+        Extract GUID from isPartOf Properties, normalizing the form of the ark.
+        """
+        if value:
+            if isinstance(value, str):
+                return extractGUID(value)
+            if isinstance(value, list):
+                return [extractGUID(elem) for elem in value]
+        else:
+            return value
 
 
 class FairscapeBaseModel(Identifier):

fairscape_models/rocrate.py

@@ -430,6 +430,8 @@ def cleanIdentifiers(self):
         def cleanGUID(metadata):
             """ Clean metadata guid property from full urls to ark:{NAAN}/{postfix}
             """
+            # TODO should be removed 
+
             if hasattr(metadata, 'guid') and isinstance(metadata.guid, str) and "http" in metadata.guid:
                 # old metadata parsing
                 #metadata.guid = urllib.parse.urlparse(metadata.guid).path.lstrip('/')