FIX: Support commons-ip use of NOTVALID

- added `convert_status` validator to `MetadataResults` class to convert commons-ip `NOTVALID` status to `INVALID`;
- moved checksum algorithm ID validation to the `PackageDetails` class;
- added test and test data for status conversion; and
- fixed type hinting for validation methods to `Any`.

eark_validator/model/package_details.py

@@ -43,6 +43,25 @@ class PackageDetails(BaseModel):
     contentinformationtype: str = ''
     checksums: List[Checksum] = []
 
+    # Validator to add a hyphen to the SHA checksum algorithm IDs generated by commons-ip
+    @model_validator(mode='before')
+    @classmethod
+    def convert_checksum_ids(cls, data: Any) -> Any:
+        incoming_checksums = data.get('checksums', [])
+        if isinstance(incoming_checksums, list):
+            # If the details are a dict type then it's a commons-ip set
+            checksums : list[Checksum] = []
+            # Loop through the checksums
+            for checksum in incoming_checksums:
+                alg_name = checksum.get('algorithm')
+                if alg_name and alg_name.startswith('SHA') and '-' not in alg_name:
+                    # If it's a SHA checksum alg ID without a hyphen, add one
+                    alg_name = f'{alg_name[:3]}-{alg_name[3:]}'
+                checksums.append(Checksum(algorithm=alg_name, value=checksum.get('value')))
+            data['checksums'] = checksums
+        # Return the reps for further validation.
+        return data
+
 class Representation(BaseModel):
     mets: Optional[MetsFile] = None
     name: Optional[str] = ''
@@ -52,9 +71,10 @@ class InformationPackage(BaseModel):
     details: Optional[PackageDetails] = None
     representations: List[Representation] = []
 
+    # Validator to convert the commons-ip representations dict to a list of representations
     @model_validator(mode='before')
     @classmethod
-    def convert_representations_dict(cls, data: Any) -> list[Representation]:
+    def convert_representations_dict(cls, data: Any) -> Any:
         representations = data.get('representations')
         if isinstance(representations, dict):
             # If this is a dict type then it's a commons-ip type, coerce to list
@@ -64,19 +84,3 @@ def convert_representations_dict(cls, data: Any) -> list[Representation]:
             data['representations'] = reps
         # Return the reps for further validation.
         return data
-
-    @model_validator(mode='before')
-    @classmethod
-    def convert_checksum_ids(cls, data: Any) -> list[Representation]:
-        details = data.get('details', {})
-        if isinstance(details, dict):
-            incoming_checksums = details.get('checksums', [])
-            checksums : list[Checksum] = []
-            for checksum in incoming_checksums:
-                alg_name = checksum.get('algorithm')
-                if alg_name and alg_name.startswith('SHA') and '-' not in alg_name:
-                    alg_name = f'{alg_name[:3]}-{alg_name[3:]}'
-                checksums.append(Checksum(algorithm=alg_name, value=checksum.get('value')))
-            data['details']['checksums'] = checksums
-        # Return the reps for further validation.
-        return data

eark_validator/model/validation_report.py

@@ -30,10 +30,10 @@
 """
 
 from enum import Enum, unique
-from typing import List, Optional
+from typing import Any, List, Optional
 import uuid
 
-from pydantic import BaseModel, ConfigDict, Field
+from pydantic import BaseModel, ConfigDict, Field, model_validator
 
 from .package_details import InformationPackage
 from .specifications import Level
@@ -123,6 +123,15 @@ class MetadataResults(BaseModel):
     status: MetadataStatus = MetadataStatus.UNKNOWN
     messages: List[Result] = []
 
+    # Validator to convert commons-ip status from NOTVALID to INVALID
+    @model_validator(mode='before')
+    @classmethod
+    def convert_status(cls, data: Any) -> Any:
+        status = data.get('status')
+        if status and status == 'NOTVALID':
+            data['status'] = 'INVALID'
+        return data
+
 class MetatdataResultSet(BaseModel):
     model_config = ConfigDict(populate_by_name=True)
     schema_results: MetadataResults = Field(validation_alias='schemaResults')

tests/resources/json/commons-ip-invalid.json

@@ -0,0 +1,79 @@
+{
+  "package" : {
+    "details" : {
+      "name" : "minimal_IP_with_schemas.zip",
+      "checksums" : [ {
+        "algorithm" : "SHA1",
+        "value" : "54BBE654FE332B51569BAF21338BC811CAD2AF66"
+      } ]
+    },
+    "profile" : {
+      "type" : "CSIP",
+      "name" : "minimal_IP_with_schemas.zip",
+      "version" : "CSIP-"
+    },
+    "representations" : {
+      "name" : "minimal_IP_with_schemas.zip"
+    }
+  },
+  "uid" : "e11da9f2-e59e-4f23-b7f8-5ac86632231f",
+  "structure" : {
+    "status" : "WellFormed",
+    "messages" : [ {
+      "ruleId" : "CSIPSTR7",
+      "location" : "",
+      "message" : "If descriptive metadata are available should include inside metadata/descriptive ",
+      "severity" : "Warn"
+    }, {
+      "ruleId" : "CSIPSTR8",
+      "location" : "",
+      "message" : "If any other metadata are available, they MAY be included in separate sub-folders, for example an additional folder named other. ",
+      "severity" : "Info"
+    }, {
+      "ruleId" : "CSIPSTR12",
+      "location" : "",
+      "message" : "The recommended best practice is to always have a METS.xml in the representation folder. ",
+      "severity" : "Warn"
+    }, {
+      "ruleId" : "CSIPSTR13",
+      "location" : "",
+      "message" : "The representation folder SHOULD include a sub-folder named metadata which MAY include all metadata about the specific representation. ",
+      "severity" : "Warn"
+    }, {
+      "ruleId" : "CSIPSTR16",
+      "location" : "",
+      "message" : "There is no documentation folder in the representation folder minimal_IP_with_schemas/representations/rep1. ",
+      "severity" : "Warn"
+    } ]
+  },
+  "metadata" : {
+    "schemaResults" : {
+      "status" : "VALID",
+      "messages" : [ ]
+    },
+    "schematronResults" : {
+      "status" : "NOTVALID",
+      "messages" : [ {
+        "ruleId" : "CSIP31",
+        "location" : "mets/amdSec",
+        "message" : "You have administrative files in the metadata/folder, you must have mets/amdSec in Root METS.xml ",
+        "severity" : "Warn"
+      }, {
+        "ruleId" : "CSIP45",
+        "location" : "mets/amdSec/rightsMD",
+        "message" : "Individual representations should state their specific rights in their representation METS file (Root METS.xml) ",
+        "severity" : "Info"
+      }, {
+        "ruleId" : "CSIP66",
+        "location" : "mets/fileSec/fileGrp/file",
+        "message" : "You have files in SIP that are not referenced in Root METS.xml ",
+        "severity" : "Error"
+      }, {
+        "ruleId" : "CSIP80",
+        "location" : "mets/structMap",
+        "message" : "Must have one structMap with the mets/structMap[@LABEL='CSIP'] in Root METS.xml doens't appear mets/structMap[@LABEL='CSIP']. ",
+        "severity" : "Error"
+      } ]
+    }
+  }
+}

tests/rules_test.py

@@ -41,6 +41,7 @@
 TEST_RES_XML = TEST_RES + '.xml'
 TEST_RES_JSON = TEST_RES + '.json'
 COMMONS_IP_JSON = str(files(JSON).joinpath('commons-ip-report.json'))
+COMMONS_IP_INVALID_JSON = str(files(JSON).joinpath('commons-ip-invalid.json'))
 PERSON_PATH = str(files(SCHEMATRON).joinpath('person.xml'))
 NOT_FOUND_PATH = str(files(SCHEMATRON).joinpath('not-found.xml'))
 EMPTY_FILE_PATH = str(files(TEST_RES).joinpath('empty.file'))
@@ -235,6 +236,13 @@ def test_deserialise_commons_ip_report(self):
         result: ValidationReport = ValidationReport.model_validate_json(contents)
         self.assertIsNotNone(result)
 
+    def test_deserialise_commons_ip_invalid(self):
+        file_name = COMMONS_IP_INVALID_JSON
+        with open(file_name, 'r', encoding='utf-8') as _f:
+            contents = _f.read()
+        result: ValidationReport = ValidationReport.model_validate_json(contents)
+        self.assertIsNotNone(result)
+
 class SeverityTest(str, Enum):
     NOT_SEV = 'NOT_SEV'