Merge pull request crs4#155 from kikkomep/fix/issue-128

feat: allow context-prefixed terms in validation

Author: kikkomep

rocrate_validator/profiles/ro-crate/must/0_file_descriptor_format.py

@@ -14,10 +14,10 @@
 
 from typing import Any
 
-from rocrate_validator.utils import log as logging
 from rocrate_validator.models import ValidationContext
 from rocrate_validator.requirements.python import (PyFunctionCheck, check,
                                                    requirement)
+from rocrate_validator.utils import log as logging
 from rocrate_validator.utils.http import HttpRequester
 
 # set up logging
@@ -339,9 +339,27 @@ def add_unexpected_key(k: str, u_keys: dict) -> None:
         # If the entity is a dictionary, check each key
         if isinstance(entity, dict):
             for k, v in entity.items():
-                if k not in context_keys and k not in SKIP_KEYS:
+                # If the key is in the skip keys, skip it
+                if k in SKIP_KEYS:
+                    logger.debug(f"Key {k} is a reserved JSON-LD keyword, skipping")
+
+                # If the key is not in the context keys,
+                # it can be used in compacted format only if it is a valid compact IRI
+                # with a prefix that is in the context
+                elif k not in context_keys:
                     logger.debug(f"Key {k} not in context keys")
-                    add_unexpected_key(k, unexpected_keys)
+
+                    # Try to get the prefix of the compact IRI, if it has one
+                    prefix = k.split(":", 1)[0] if ":" in k else None
+                    logger.debug(f"Checking prefix {prefix} of key {k}")
+                    # If the key does not have a prefix (no colon) or the prefix is not in the context keys,
+                    # it cannot be used as a key in compacted format
+                    if prefix is None or prefix not in context_keys:
+                        logger.debug(
+                            f"Key {k} does not have a valid prefix in context keys, adding to unexpected keys")
+                        add_unexpected_key(k, unexpected_keys)
+
+                # If the value is a dictionary or a list, check its keys recursively
                 if isinstance(v, (dict, list)):
                     self.__check_entity_keys__(v, context_keys, unexpected_keys)
 

tests/data/crates/valid/rocrate-with-custom-terms/ro-crate-metadata.json

@@ -2,7 +2,10 @@
     "@context": [
         "https://w3id.org/ro/crate/1.1/context",
         "https://w3id.org/ro/terms/workflow-run/context",
-        { "myCustomTerm": "https://example.org#myCustomTerm" }
+        { 
+            "myCustomTerm": "https://example.org#myCustomTerm",
+            "rdfs": "http://www.w3.org/2000/01/rdf-schema#"
+        }
     ],
     "@graph": [
         {
@@ -14,7 +17,9 @@
             "about": {
                 "@id": "./"
             },
-            "myCustomTerm": "custom term value"
+            "myCustomTerm": "custom term value",
+            "rdfs:label": "RO-Crate with custom terms",
+            "rdfs:comment": "This is a comment using a term from the rdfs namespace"
         },
         {
             "@id": "./",