Enhance json schema validation with custom validators

relecov_tools/assets/schema_utils/custom_validators.py

@@ -0,0 +1,27 @@
+import os
+import json
+from jsonschema import ValidationError
+
+
+# TODO: ADD AN USAGE DOC HERE.
+
+
+# Disable default enum validation for amr_genes
+def ignore_enum(validator, enums, instance, schema):
+    pass
+
+
+def validate_amr_genes(validator, value, instance, schema):
+    # Load Config from File
+    amr_config = os.path.join(os.path.dirname(__file__), "conf", "amr_genes.config")
+    with open(amr_config, "r") as file:
+        amr_json = json.load(file)
+
+    amr_genes = instance.get("amr_genes", [])
+    for gene in amr_genes:
+        if gene not in amr_json.keys():
+            yield ValidationError(f"Gene '{gene}' is not annotated in any group.")
+
+
+# Map of custom validators
+available = {"amr_genes_validator": validate_amr_genes}

relecov_tools/conf/amr_genes.json

@@ -0,0 +1,12 @@
+{
+    "geneA": "group1",
+    "geneB": "group2",
+    "geneC": "group3",
+    "geneD": "group1",
+    "geneE": "group2",
+    "geneF": "group3",
+    "geneG": "group1",
+    "geneH": "group2",
+    "geneI": "group3",
+    "geneJ": "group1"
+}

relecov_tools/conf/configuration.json

@@ -400,5 +400,8 @@
                 "fastq_r2_md5"
             ]
         }
-    }
+    },
+    "json_schema_validators": [
+        "amr_genes_validator"
+    ]
 }

relecov_tools/json_validation.py

@@ -9,6 +9,8 @@
 
 import relecov_tools.utils
 from relecov_tools.config_json import ConfigJson
+import relecov_tools.assets.schema_utils.custom_validators
+
 
 log = logging.getLogger(__name__)
 stderr = rich.console.Console(
@@ -64,6 +66,42 @@ def validate_schema(self):
             stderr.print("[red] Json schema does not fulfill Draft 202012 Validation")
             sys.exit(1)
 
+    def register_custom_validators(self):
+        """Register custom JSON schema validators based on configuration."""
+
+        # Get custom validators utils
+        stderr.print("Loading configuration for custom JSON schema validators.")
+        conf = ConfigJson()
+        conf_path = os.path.join(
+            os.path.dirname(__file__), "conf", "configuration.json"
+        )
+
+        custom_validator_list = conf.get_configuration("json_schema_valiators")
+        if custom_validator_list is None:
+            stderr.print(f"No 'json_schema_validators' key found in {conf_path}")
+            return
+
+        if not len(custom_validator_list) > 0:
+            stderr.print(
+                f"No validators defined yet under 'json_schema_validators' key in {conf_path}"
+            )
+            return
+
+        # Append custom validators to Draft202012Validator
+        stderr.print("Registering custom validators.")
+        available_validators = (
+            relecov_tools.assets.schema_utils.custom_validators.available
+        )
+        for validator_name in custom_validator_list:
+            if validator_name in available_validators:
+                Draft202012Validator.VALIDATORS[validator_name] = available_validators[
+                    validator_name
+                ]
+                stderr.print(f"Custom validator '{validator_name}' successfully added.")
+            else:
+                stderr.print(f"No validator found for '{validator_name}'. Exiting.")
+                sys.exit(1)
+
     def validate_instances(self):
         """Validate data instances against a validated json schema"""
 
@@ -189,5 +227,6 @@ def validate(self):
         """Write invalid samples from metadata to excel"""
 
         self.validate_schema()
+        self.register_custom_validators()
         invalid_json = self.validate_instances()
         self.create_invalid_metadata(invalid_json, self.metadata, self.out_folder)

relecov_tools/schema/relecov_schema.json

@@ -18,7 +18,8 @@
         "sequencing_instrument_model",
         "sequencing_instrument_platform",
         "enrichment_panel",
-        "enrichment_panel_version"
+        "enrichment_panel_version",
+        "amr_genes"
     ],
     "type": "object",
     "properties": {
@@ -2899,6 +2900,18 @@
             "description": "",
             "clasification": "Submission ENA",
             "label": "Run Alias"
+        },
+        "tmp-amr_genes": { "$ref": "#/$defs/gene_name"}
+    },
+    "amr_genes_validation": true,
+    "$defs": {
+        "gene_name" : {
+            "type": "array",
+            "items": {
+                "type": "string",
+                "enum": ["geneA", "geneB", "geneC", "geneD", "geneE", "geneF", "geneG", "geneH", "geneI", "geneJ"],
+                "uniqueItems": true
+            }
         }
     }
 }