Improve error logging (#428)

* ADD hash checks to datasets

* split target on comma, allows errors on multiple targets

* FIX unit test

Author: mfeurer

openml/datasets/dataset.py

@@ -270,7 +270,10 @@ def get_data(self, target=None,
             rval.append(data)
         else:
             if isinstance(target, six.string_types):
-                target = [target]
+                if ',' in target:
+                    target = target.split(',')
+                else:
+                    target = [target]
             targets = np.array([True if column in target else False
                                 for column in attribute_names])
             if np.sum(targets) > 1:

openml/datasets/functions.py

@@ -10,7 +10,8 @@
 
 import openml.utils
 from .dataset import OpenMLDataset
-from ..exceptions import OpenMLCacheException, OpenMLServerNoResult
+from ..exceptions import OpenMLCacheException, OpenMLServerNoResult, \
+    OpenMLHashException
 from .. import config
 from .._api_calls import _perform_api_call, _read_url
 
@@ -404,12 +405,14 @@ def _get_dataset_arff(did_cache_dir, description):
     url = description['oml:url']
     arff_string = _read_url(url)
     md5 = hashlib.md5()
-    md5.update(arff_string.encode('utf8'))
+    md5.update(arff_string.encode('utf-8'))
     md5_checksum = md5.hexdigest()
     if md5_checksum != md5_checksum_fixture:
-        raise ValueError(
+        raise OpenMLHashException(
             'Checksum %s of downloaded dataset %d is unequal to the checksum '
-            '%s sent by the server.' % (md5_checksum, did, md5_checksum_fixture)
+            '%s sent by the server.' % (
+                md5_checksum, int(did), md5_checksum_fixture
+            )
         )
 
     with io.open(output_file_path, "w", encoding='utf8') as fh:

openml/exceptions.py

@@ -35,3 +35,8 @@ class OpenMLCacheException(PyOpenMLError):
     """Dataset / task etc not found in cache"""
     def __init__(self, message):
         super(OpenMLCacheException, self).__init__(message)
+
+
+class OpenMLHashException(PyOpenMLError):
+    """Locally computed hash is different than hash announced by the server."""
+    pass

tests/test_datasets/test_dataset_functions.py

@@ -17,7 +17,7 @@
 
 import openml
 from openml import OpenMLDataset
-from openml.exceptions import OpenMLCacheException, PyOpenMLError
+from openml.exceptions import OpenMLCacheException, PyOpenMLError, OpenMLHashException
 from openml.testing import TestBase
 from openml.utils import _tag_entity
 
@@ -268,7 +268,7 @@ def test__getarff_md5_issue(self):
             'oml:url': 'https://www.openml.org/data/download/61',
         }
         self.assertRaisesRegexp(
-            ValueError,
+            OpenMLHashException,
             'Checksum ad484452702105cbf3d30f8deaba39a9 of downloaded dataset 5 '
             'is unequal to the checksum abc sent by the server.',
             _get_dataset_arff,