Use the referencing library

This is to avoid the `DeprecationWarning` about
*retrieving remote references being a security vulnerability*.

(We have only used this feature for referencing items in the local file
system, so the software should not have been compromised. Nevertheless
it is best to modify the code so we don't get this warning.)

src/yamlprocessor/dataprocess.py

@@ -18,19 +18,24 @@
 from contextlib import suppress
 from datetime import datetime
 from errno import ENOENT
+import json
 import logging
 import logging.config
 import os
 from pathlib import Path
 import re
 import sys
-from urllib.parse import urlparse
 
 from dateutil.parser import parse as datetimeparse
 from dateutil.relativedelta import relativedelta
 from dateutil.tz import tzlocal
 import jmespath
 import jsonschema
+try:
+    from referencing import Registry, Resource
+    from referencing.jsonschema import DRAFT202012
+except ImportError:
+    pass  # python < 3.8
 from ruamel.yaml import YAML
 from ruamel.yaml.constructor import ConstructorError
 
@@ -700,22 +705,44 @@ def validate_data(
         """
         if not schema_location:
             return
-        schema = {"$ref": schema_location}
-        if not urlparse(schema_location).scheme:
-            schema_path = Path(schema_location)
-            if schema_path.exists():
-                schema = {"$ref": schema_path.absolute().as_uri()}
-            elif self.schema_prefix:
-                schema = {"$ref": self.schema_prefix + schema_location}
         try:
-            jsonschema.validate(schema=schema, instance=data)
+            registry = Registry(retrieve=self.get_schema_file)
+            jsonschema.Draft202012Validator(
+                {
+                    '$schema': 'https://json-schema.org/draft/2020-12/schema',
+                    '$ref': schema_location,
+                },
+                registry=registry,
+            ).validate(data)
+        except NameError:
+            ref = 'file://' + str(self._get_schema_file(schema_location))
+            jsonschema.validate(schema={'$ref': ref}, instance=data)
         except jsonschema.exceptions.ValidationError as exc:
             logging.error(f'not ok {out_file_name}')
             logging.exception(exc)
             raise
         else:
             logging.info(f'ok {out_file_name}')
 
+    def get_schema_file(self, schema_location: str):
+        schema_path = self._get_schema_file(schema_location)
+        return Resource(
+            contents=json.loads(schema_path.read_text()),
+            specification=DRAFT202012)
+
+    def _get_schema_file(self, schema_location: str):
+        if schema_location.startswith('file://'):
+            schema_location = schema_location.replace('file://', '', 1)
+        schema_path = Path(schema_location)
+        if not schema_path.exists():
+            for prefix in [self.schema_prefix]:  # FIXME
+                if prefix.startswith('file://'):
+                    prefix = prefix.replace('file://', '', 1)
+                if (Path(prefix) / schema_path).exists():
+                    schema_path = Path(prefix) / schema_path
+                    break
+        return schema_path
+
 
 def main(argv=None):
     configure_basic_logging()

src/yamlprocessor/tests/test_dataprocess.py

@@ -556,6 +556,7 @@ def test_main_15(tmp_path, yaml):
 def test_main_validate_1(tmp_path, capsys, yaml):
     """Test main, YAML with JSON schema validation."""
     schema = {
+        '$schema': 'https://json-schema.org/draft/2020-12/schema',
         'type': 'object',
         'properties': {'hello': {'type': 'string'}},
         'required': ['hello'],

src/yamlprocessor/tests/test_schemaprocess.py

@@ -1,6 +1,12 @@
 import json
+from pathlib import Path
 
 import jsonschema
+try:
+    from referencing import Registry, Resource
+    from referencing.jsonschema import DRAFT202012
+except ImportError:
+    pass  # python < 3.8
 
 from ..schemaprocess import INCLUDE_SCHEMA, INCLUDE_SCHEMA_FILENAME, main
 
@@ -75,17 +81,8 @@ def test_main_one(monkeypatch, tmp_path):
         ('schema-root.json', SAMPLE_MAIN_ONE_INC),
         ('schema-a.json', SAMPLE_TESTING),
     ):
-        try:
-            jsonschema.validate(
-                schema={
-                    '$ref': (tmp_path / schema_filename).absolute().as_uri()
-                },
-                instance=sample_data,
-            )
-        except jsonschema.exceptions.ValidationError as exc:
-            assert False, f"{schema_filename} does not validate data:\n{exc}"
-        else:
-            assert True, f"{schema_filename} works OK with data"
+        assert_jsonschema_validate(
+            (tmp_path / schema_filename).absolute().as_uri(), sample_data)
 
 
 def test_main_three(monkeypatch, tmp_path):
@@ -189,14 +186,40 @@ def test_main_three(monkeypatch, tmp_path):
         ('schema-b.json', SAMPLE_EXAMINING),
         ('schema-c.json', SAMPLE_EXAMINING_NAME_ONE),
     ):
-        try:
-            jsonschema.validate(
-                schema={
-                    '$ref': (tmp_path / schema_filename).absolute().as_uri()
-                },
-                instance=sample_data,
-            )
-        except jsonschema.exceptions.ValidationError as exc:
-            assert False, f"{schema_filename} does not validate data:\n{exc}"
-        else:
-            assert True, f"{schema_filename} works OK with data"
+        assert_jsonschema_validate(
+            (tmp_path / schema_filename).absolute().as_uri(), sample_data)
+
+
+def assert_jsonschema_validate(schema_uri, sample_data):
+    """Helper to assert schema can be used to validate sample data."""
+    try:
+        registry = Registry(retrieve=get_schema_file)
+        jsonschema.Draft202012Validator(
+            {
+                '$schema': 'https://json-schema.org/draft/2020-12/schema',
+                '$ref': schema_uri
+            },
+            registry=registry,
+        ).validate(sample_data)
+    except NameError:
+        ref = 'file://' + str(_get_schema_file(schema_uri))
+        jsonschema.validate(schema={'$ref': ref}, instance=sample_data)
+    except jsonschema.exceptions.ValidationError as exc:
+        assert False, f"{schema_uri} does not validate data:\n{exc}"
+    else:
+        assert True, f"{schema_uri} works OK with data"
+
+
+def get_schema_file(schema_location: str):
+    """Helper to retrieve a local schema file as Resource."""
+    schema_path = _get_schema_file(schema_location)
+    return Resource(
+        contents=json.loads(schema_path.read_text()),
+        specification=DRAFT202012)
+
+
+def _get_schema_file(schema_location: str):
+    """Helper to retrieve a local schema file."""
+    if schema_location.startswith('file://'):
+        schema_location = schema_location.replace('file://', '', 1)
+    return Path(schema_location)