api: addition of FormatChecker support

AUTHORS.rst

@@ -46,6 +46,7 @@ Invenio-Records is a metadata storage module.
 - Grzegorz Szpura <grzegorz.szpura@cern.ch>
 - Guillaume Lastecoueres <PX9e@gmx.fr>
 - Henning Weiler <henning.weiler@cern.ch>
+- Jacopo Notarstefano <jacopo.notarstefano@cern.ch>
 - Jaime García <jaime.garcia.llopis@cern.ch>
 - Jan Aage Lavik <jan.age.lavik@cern.ch>
 - Jan Iwaszkiewicz <jan.iwaszkiewicz@cern.ch>

docs/conf.py

@@ -328,4 +328,7 @@
 
 
 # Example configuration for intersphinx: refer to the Python standard library.
-intersphinx_mapping = {'https://docs.python.org/': None}
+intersphinx_mapping = {
+    'https://docs.python.org/': None,
+    'https://python-jsonschema.readthedocs.io/en/latest': None,
+}

invenio_records/api.py

@@ -75,10 +75,35 @@ def updated(self):
         """Get last updated timestamp."""
         return self.model.updated if self.model else None
 
-    def validate(self):
-        """Validate record according to schema defined in ``$schema`` key."""
+    def validate(self, **kwargs):
+        """Validate record according to schema defined in ``$schema`` key.
+
+        This method accepts arbitrary kwargs to pass the ``format_checker``
+        keyword argument to the underlying JSON Schema implementation.
+
+        A ``format_checker`` is an instance of class
+        :class:`jsonschema.FormatChecker` containing business logic to validate
+        arbitrary formats. For example:
+
+        >>> from jsonschema import FormatChecker
+        >>> from jsonschema.validators import validate
+        >>> checker = FormatChecker()
+        >>> checker.checks('foo')(lambda el: el.startswith('foo'))
+        <function <lambda> at ...>
+        >>> validate('foo', {'format': 'foo'}, format_checker=checker)
+
+        returns ``None``, which means that the validation was successful, while
+
+        >>> validate('bar', {'format': 'foo'}, format_checker=checker)
+        Traceback (most recent call last):
+        ...
+        ValidationError: 'bar' is not a 'foo'
+        ...
+
+        raises a :class:`jsonschema.exceptions.ValidationError`.
+        """
         if '$schema' in self and self['$schema'] is not None:
-            return _records_state.validate(self, self['$schema'])
+            return _records_state.validate(self, self['$schema'], **kwargs)
         return True
 
     def replace_refs(self):
@@ -94,8 +119,8 @@ class Record(RecordBase):
     """Define API for metadata creation and manipulation."""
 
     @classmethod
-    def create(cls, data, id_=None):
-        """Create a record instance and store it in database.
+    def create(cls, data, id_=None, **kwargs):
+        r"""Create a record instance and store it in database.
 
         Procedure followed:
 
@@ -111,15 +136,22 @@ def create(cls, data, id_=None):
 
         :param data: Dict with record metadata.
         :param id_: Force the UUID for the record.
+        :param \**kwargs: See below.
         :returns: A new Record instance.
+
+        :Keyword Arguments:
+          * **format_checker** --
+            An instance of class :class:`jsonschema.FormatChecker`, which
+            contains validation rules for formats. See
+            :func:`~invenio_records.api.RecordBase.validate` for details.
         """
         from .models import RecordMetadata
         with db.session.begin_nested():
             record = cls(data)
 
             before_record_insert.send(record)
 
-            record.validate()
+            record.validate(**kwargs)
 
             record.model = RecordMetadata(id=id_, json=record)
 

invenio_records/ext.py

@@ -46,15 +46,17 @@ def __init__(self, app, entry_point_group=None):
         self.ref_resolver_cls = ref_resolver_factory(self.resolver)
         self.loader_cls = json_loader_factory(self.resolver)
 
-    def validate(self, data, schema):
+    def validate(self, data, schema, **kwargs):
         """Validate data using schema with ``JSONResolver``."""
         if not isinstance(schema, dict):
             schema = {'$ref': schema}
-        return validate(data, schema,
-                        types=self.app.config.get(
-                            "RECORDS_VALIDATION_TYPES", {}
-                        ),
-                        resolver=self.ref_resolver_cls.from_schema(schema))
+        return validate(
+            data,
+            schema,
+            resolver=self.ref_resolver_cls.from_schema(schema),
+            types=self.app.config.get('RECORDS_VALIDATION_TYPES', {}),
+            **kwargs
+        )
 
     def replace_refs(self, data):
         """Replace the JSON reference objects with ``JsonRef``."""

tests/test_api.py

@@ -33,6 +33,8 @@
 import pytest
 from jsonresolver import JSONResolver
 from jsonresolver.contrib.jsonref import json_loader_factory
+from jsonschema import FormatChecker
+from jsonschema.exceptions import ValidationError
 from sqlalchemy.orm.exc import NoResultFound
 
 from invenio_records import Record
@@ -321,3 +323,26 @@ def test_record_dump(app, db):
         record_dump = record.dumps()
         record_dump['foo']['bar'] = 'Spam'
         assert record_dump['foo']['bar'] != record['foo']['bar']
+
+
+def test_validate_with_format(app, db):
+    """Test that validation can accept custom format rules."""
+    with app.app_context():
+        checker = FormatChecker()
+        checker.checks('foo')(lambda el: el.startswith('foo'))
+        record = Record.create({
+            'bar': 'foo',
+            '$schema': {
+                'properties': {
+                    'bar': {'format': 'foo'}
+                }
+            }
+        })
+
+        assert record.validate(format_checker=checker) is None
+
+        record['bar'] = 'bar'
+
+        with pytest.raises(ValidationError) as excinfo:
+            record.validate(format_checker=checker)
+        assert "'bar' is not a 'foo'" in str(excinfo.value)