Merge pull request #16 from scrapy-plugins/feature/improve_properties_extraction

Added support for fields allOf, anyOf, oneOf on top level

Author: Víctor Ruiz

scrapy_jsonschema/item.py

@@ -59,9 +59,10 @@ class JsonSchemaMeta(ABCMeta):
         JSON_SCHEMA_DRAFT_7: draft7_format_checker,
     }
 
+    combination_schemas_keywords = ['allOf', 'anyOf', 'oneOf']
+
     def __new__(mcs, class_name, bases, attrs):
         cls = super(JsonSchemaMeta, mcs).__new__(mcs, class_name, bases, attrs)
-
         fields = {}
         schema = attrs.get('jsonschema', {})
         if cls.merge_schema:
@@ -77,7 +78,7 @@ def __new__(mcs, class_name, bases, attrs):
             )
         cls.validator = cls._get_validator(schema)
         cls.validator.check_schema(schema)
-        for k in schema['properties']:
+        for k in cls.get_top_level_property_names(schema):
             fields[k] = Field()
         cls.fields = cls.fields.copy()
         cls.fields.update(fields)
@@ -90,6 +91,16 @@ def __new__(mcs, class_name, bases, attrs):
         ]
         return cls
 
+    @classmethod
+    def get_top_level_property_names(cls, schema):
+        for field in schema.get('properties', {}):
+            yield field
+
+        for keyword in cls.combination_schemas_keywords:
+            for subschema in schema.get(keyword, []):
+                for field in subschema.get('properties', {}):
+                    yield field
+
     @classmethod
     def _get_validator(cls, schema):
         draft_version = schema.get('$schema')

tests/test_item.py

@@ -16,6 +16,102 @@
     merge_schema_new,
 )
 
+class SchemaWithCombiningKeywords(TestCase):
+
+    schema_allOf = {
+      "$schema": JSON_SCHEMA_DRAFT_7,
+        "allOf": [
+            {
+                "properties": {"foo": {"type": "string"}},
+                "required": ["foo"],
+            },
+            {
+                "properties": {"bar": {"type": "number"}},
+                "required": ["bar"]
+            }
+        ]
+    }
+
+    schema_anyOf = {
+      "$schema": JSON_SCHEMA_DRAFT_7,
+        "anyOf": [
+            {
+                "properties": {"foo": {"type": "string"}},
+                "required": ["foo"],
+            },
+            {
+                "properties": {"bar": {"type": "number"}},
+                "required": ["bar"]
+            }
+        ]
+    }
+
+    schema_oneOf = {
+      "$schema": JSON_SCHEMA_DRAFT_7,
+        "oneOf": [
+            {
+                "properties": {"foo": {"type": "string"}},
+                "required": ["foo"],
+            },
+            {
+                "properties": {"bar": {"type": "number"}},
+                "required": ["bar"]
+            }
+        ]
+    }
+
+    def test_all_of_schema(self):
+
+        class allOfItem(JsonSchemaItem):
+            jsonschema = self.schema_allOf
+
+        item = allOfItem()
+        item['foo'] = 'foo'
+        item['bar'] = 2
+        self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+        with pytest.raises(AssertionError):
+            item = allOfItem()
+            item['foo'] = 'foo'
+            self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+
+    def test_any_of_schema(self):
+
+        class anyOfItem(JsonSchemaItem):
+            jsonschema = self.schema_anyOf
+
+        item = anyOfItem()
+        item['foo'] = 'foo'
+        item['bar'] = 2
+        self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+        item = anyOfItem()
+        item['foo'] = 'foo'
+        self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+        item = anyOfItem()
+        item['bar'] = 2
+        self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+    def test_one_of_schema(self):
+
+        class oneOfItem(JsonSchemaItem):
+            jsonschema = self.schema_oneOf
+
+        item = oneOfItem()
+        item['foo'] = 'foo'
+        self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+        item = oneOfItem()
+        item['bar'] = 2
+        self.assertFalse(list(item.validator.iter_errors(dict(item))))
+
+        with pytest.raises(AssertionError):
+            item = oneOfItem()
+            item['foo'] = 'foo'
+            item['bar'] = 2
+            self.assertFalse(list(item.validator.iter_errors(dict(item))))
 
 class ValidSchemaTestCase(TestCase):