Merge pull request #5 from vitorbaptista/improve-exception-handling

[#4] Use exceptions to represent errors, not strings

README.md

@@ -22,6 +22,6 @@ The `schema` can be a json string, python object, or a schema id corresponding w
 
 `valid` is a boolean determining whether the datapackage validates against the schema.
 
-`errors` is an array of error string messages. Empty if `valid` is True.
+`errors` is a list of exceptions found during validation. Empty if `valid` is True.
 
 [Data Package Profiles]: https://github.com/dataprotocols/registry

datapackage_validate/exceptions.py

@@ -0,0 +1,14 @@
+class DataPackageValidateException(Exception):
+    pass
+
+
+class SchemaError(DataPackageValidateException):
+    pass
+
+
+class ValidationError(DataPackageValidateException):
+    pass
+
+
+class RegistryError(DataPackageValidateException):
+    pass

datapackage_validate/validation.py

@@ -10,6 +10,7 @@
 import requests
 
 from . import compat
+from . import exceptions
 
 
 def _get_schema_url_from_registry(id, registry):
@@ -34,7 +35,8 @@ def validate(datapackage, schema='base'):
 
     `valid` - a boolean to determine whether the datapackage validates against
     the schema.
-    `errors` - an array of error string messages. Empty if `valid` is True.
+    `errors` - a list of exceptions found during validation. Empty if `valid`
+    is True.
     '''
 
     valid = False
@@ -48,9 +50,12 @@ def validate(datapackage, schema='base'):
         try:
             datapackage_obj = json.loads(datapackage)
         except ValueError as e:
-            errors.append('Invalid JSON: {0}'.format(e))
+            errors.append(exceptions.DataPackageValidateException(e))
     elif not (isinstance(datapackage, dict) or isinstance(datapackage, list)):
-        errors.append('Invalid Data Package: not a string or object')
+        msg = 'Data Package must be a dict or JSON string, but was a \'{0}\''
+        dp_type = type(datapackage).__name__
+        error = exceptions.DataPackageValidateException(msg.format(dp_type))
+        errors.append(error)
     else:
         datapackage_obj = datapackage
 
@@ -66,20 +71,20 @@ def validate(datapackage, schema='base'):
             try:
                 registry = datapackage_registry.get()
             except requests.HTTPError as e:
-                errors.append('Registry Error: {0}'.format(e))
+                errors.append(exceptions.RegistryError(e))
             else:
                 schema_url = _get_schema_url_from_registry(schema, registry)
                 if schema_url is None:
-                    errors.append(
-                        'Registry Error: no schema with id "{0}"'
-                        .format(schema))
+                    msg = 'No schema with id \'{0}\''.format(e)
+                    errors.append(exceptions.RegistryError(msg))
                 else:
                     try:
                         schema_obj = _fetch_schema_obj_from_url(schema_url)
                     except requests.HTTPError as e:
-                        errors.append('Registry Error: {0}'.format(e))
+                        errors.append(exceptions.RegistryError(e))
     elif not isinstance(schema, dict):
-        errors.append('Invalid Schema: not a string or object')
+        msg = 'Schema must be a string or dict'
+        errors.append(exceptions.SchemaError(msg))
     else:
         schema_obj = schema
 
@@ -88,7 +93,7 @@ def validate(datapackage, schema='base'):
         try:
             jsonschema.validate(datapackage_obj, schema_obj)
         except jsonschema.ValidationError as e:
-            errors.append('Schema ValidationError: {0}'.format(e.message))
+            errors.append(exceptions.ValidationError(e.message))
         else:
             valid = True
 

tests/test_validate.py

@@ -9,10 +9,12 @@
 import unittest
 
 from nose.tools import (assert_true,
-                        assert_false)
+                        assert_false,
+                        assert_is_instance)
 import httpretty
 
 import datapackage_validate
+import datapackage_validate.exceptions as exceptions
 
 
 REGISTRY_BACKEND_URL = \
@@ -73,7 +75,7 @@ def test_invalid_json_string(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Invalid JSON:' in errors[0])
+        assert_is_instance(errors[0], exceptions.DataPackageValidateException)
 
     @httpretty.activate
     def test_invalid_json_not_string(self):
@@ -92,8 +94,7 @@ def test_invalid_json_not_string(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Invalid Data Package: not a string or object'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.DataPackageValidateException)
 
     @httpretty.activate
     def test_valid_json_obj(self):
@@ -135,12 +136,8 @@ def test_invalid_json_obj(self):
 
         assert_false(valid)
         assert_true(errors)
-        # Py2 returns u'name', Py3 doesn't. Test for either.
-        assert_true("Schema ValidationError: u'name' is a required property"
-                    in errors[0]
-                    or
-                    "Schema ValidationError: 'name' is a required property"
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.ValidationError)
+        assert_true("'name' is a required property" in str(errors[0]))
 
 
 class TestValidateWithSchemaAsArgument(unittest.TestCase):
@@ -172,8 +169,7 @@ def test_schema_as_wrong_object_type(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Invalid Schema: not a string or object'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.SchemaError)
 
     def test_schema_as_dict(self):
         '''Pass schema as python dict to validate()'''
@@ -212,8 +208,7 @@ def test_schema_as_invalid_id(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Registry Error: no schema with id "not-a-valid-id"'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.RegistryError)
 
     @httpretty.activate
     def test_schema_404_raises_error(self):
@@ -228,9 +223,7 @@ def test_schema_404_raises_error(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Registry Error: 404 Client Error: '
-                    'Not Found for url: https://example.com/base_schema.json'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.RegistryError)
 
     @httpretty.activate
     def test_schema_500_raises_error(self):
@@ -245,10 +238,7 @@ def test_schema_500_raises_error(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Registry Error: 500 Server Error: '
-                    'Internal Server Error for url: '
-                    'https://example.com/base_schema.json'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.RegistryError)
 
     @httpretty.activate
     def test_registry_404_raises_error(self):
@@ -260,9 +250,7 @@ def test_registry_404_raises_error(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Registry Error: 404 Client Error: Not Found for url: '
-                    'https://rawgit.com/dataprotocols/registry/master/registry.csv'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.RegistryError)
 
     @httpretty.activate
     def test_registry_500_raises_error(self):
@@ -274,7 +262,4 @@ def test_registry_500_raises_error(self):
 
         assert_false(valid)
         assert_true(errors)
-        assert_true('Registry Error: 500 Server Error: '
-                    'Internal Server Error for url: '
-                    'https://rawgit.com/dataprotocols/registry/master/registry.csv'
-                    in errors[0])
+        assert_is_instance(errors[0], exceptions.RegistryError)