Merge pull request #218 from Mimino666/django-item-validation

Django item validation

scrapy/contrib/djangoitem.py

@@ -1,3 +1,5 @@
+from django.core.exceptions import ValidationError
+
 from scrapy.item import Field, Item, ItemMeta
 
 
@@ -24,10 +26,49 @@ class DjangoItem(Item):
 
     django_model = None
 
+    def __init__(self, *args, **kwargs):
+        super(DjangoItem, self).__init__(*args, **kwargs)
+        self._instance = None
+        self._errors = None
+
     def save(self, commit=True):
-        modelargs = dict((k, self.get(k)) for k in self._values
-                         if k in self._model_fields)
-        model = self.django_model(**modelargs)
         if commit:
-            model.save()
-        return model
+            self.instance.save()
+        return self.instance
+
+    def is_valid(self, exclude=None):
+        self._get_errors(exclude)
+        return not bool(self._errors)
+
+    def _get_errors(self, exclude=None):
+        if self._errors is not None:
+            return self._errors
+
+        self._errors = {}
+        if exclude is None:
+            exclude = []
+
+        try:
+            self.instance.clean_fields(exclude=exclude)
+        except ValidationError, e:
+            self._errors = e.update_error_dict(self._errors)
+
+        try:
+            self.instance.clean()
+        except ValidationError, e:
+            self._errors = e.update_error_dict(self._errors)
+
+        # uniqueness is not checked, because it is faster to check it when
+        # saving object to database. Just beware, that failed save()
+        # raises IntegrityError instead of ValidationError.
+
+        return self._errors
+    errors = property(_get_errors)
+
+    @property
+    def instance(self):
+        if self._instance is None:
+            modelargs = dict((k, self.get(k)) for k in self._values
+                             if k in self._model_fields)
+            self._instance = self.django_model(**modelargs)
+        return self._instance

scrapy/tests/test_djangoitem/__init__.py

@@ -34,7 +34,7 @@ class IdentifiedPersonItem(DjangoItem):
 
 
 class DjangoItemTest(unittest.TestCase):
-    
+
     def setUp(self):
         if not django:
             raise unittest.SkipTest("Django is not available")
@@ -74,9 +74,43 @@ class DjangoItemTest(unittest.TestCase):
         i = OverrideFieldPersonItem()
 
         i['name'] = 'John'
+        # it is not obvious that "age" should be saved also, since it was
+        # redefined in child class
+        i['age'] = '22'
         person = i.save(commit=False)
 
         self.assertEqual(person.name, 'John')
+        self.assertEqual(person.age, '22')
+
+    def test_validation(self):
+        long_name = 'z' * 300
+        i = BasePersonItem(name=long_name)
+        self.assertFalse(i.is_valid())
+        self.assertDictEqual(
+            {
+                'age': [u'This field cannot be null.'],
+                'name': [u'Ensure this value has at most 255 characters (it has 300).']
+            },
+            i.errors)
+
+        i = BasePersonItem(name='John')
+        self.assertTrue(i.is_valid(exclude=['age']))
+        self.assertDictEqual({}, i.errors)
+
+        # once the item is validated, it does not validate again
+        i['name'] = long_name
+        self.assertTrue(i.is_valid())
+
+    def test_override_validation(self):
+        i = OverrideFieldPersonItem()
+        i['name'] = 'John'
+        self.assertFalse(i.is_valid())
+
+        i = i = OverrideFieldPersonItem()
+        i['name'] = 'John'
+        i['age'] = '22'
+        self.assertTrue(i.is_valid())
+
 
     def test_default_field_values(self):
         i = BasePersonItem()