Merge branch 'develop' into list_datasets_docsting

Author: amueller

.gitignore

@@ -73,6 +73,3 @@ target/
 # IDE
 .idea
 *.swp
-
-# Other
-*.pkl

doc/conf.py

@@ -63,8 +63,8 @@
 
 # General information about the project.
 project = u'OpenML'
-copyright = u'2014-2016, Matthias Feurer, Andreas Müller, Farzan Majdani, ' \
-            u'Joaquin Vanschoren and Pieter Gijsbers'
+copyright = u'2014-2017, Matthias Feurer, Andreas Müller, Farzan Majdani, ' \
+            u'Joaquin Vanschoren, Jan van Rijn and Pieter Gijsbers'
 
 # The version info for the project you're documenting, acts as replacement for
 # |version| and |release|, also used in various other places throughout the

doc/usage.rst

@@ -513,7 +513,7 @@ We can now also inspect the flow object which was automatically created:
 .. code:: python
 
     >>> flow = openml.flows.get_flow(run.flow_id)
-    >>> pprint(vars(flow), depth=2)
+    >>> pprint(vars(flow), depth=2)                             # doctest: +SKIP
     {'binary_format': None,
      'binary_md5': None,
      'binary_url': None,
@@ -522,7 +522,7 @@ We can now also inspect the flow object which was automatically created:
      'custom_name': None,
      'dependencies': 'sklearn==0.18.2\nnumpy>=1.6.1\nscipy>=0.9',
      'description': 'Automatically created scikit-learn flow.',
-     'external_version': 'openml==0.6.0dev,sklearn==0.18.2',
+     'external_version': 'openml==0.6.0,sklearn==0.18.2',
      'flow_id': 7245,
      'language': 'English',
      'model': RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',

openml/__version__.py

@@ -1,4 +1,4 @@
 """Version information."""
 
 # The following line *must* be the last in the module, exactly as formatted:
-__version__ = "0.6.0dev"
+__version__ = "0.6.0"

openml/datasets/dataset.py

@@ -205,7 +205,7 @@ def get_data(self, target=None, target_dtype=int, include_row_id=False,
 
         path = self.data_pickle_file
         if not os.path.exists(path):
-            raise ValueError("Cannot find a ndarray file for dataset %s at "
+            raise ValueError("Cannot find a pickle file for dataset %s at "
                              "location %s " % (self.name, path))
         else:
             with open(path, "rb") as fh:
@@ -425,4 +425,4 @@ def _data_features_supported(self):
                 if self.features[idx].data_type not in ['numeric', 'nominal']:
                     return False
             return True
-        return True
+        return True

openml/datasets/functions.py

@@ -179,7 +179,7 @@ def list_datasets(offset=None, size=None, tag=None):
 def _list_datasets(api_call):
     # TODO add proper error handling here!
     xml_string = _perform_api_call(api_call)
-    datasets_dict = xmltodict.parse(xml_string)
+    datasets_dict = xmltodict.parse(xml_string, force_list=('oml:dataset',))
 
     # Minimalistic check if the XML is useful
     assert type(datasets_dict['oml:data']['oml:dataset']) == list, \
@@ -337,9 +337,6 @@ def _get_dataset_description(did_cache_dir, dataset_id):
     description = xmltodict.parse(dataset_xml)[
         "oml:data_set_description"]
 
-    with io.open(description_file, "w", encoding='utf8') as fh:
-        fh.write(dataset_xml)
-
     return description
 
 
@@ -418,7 +415,7 @@ def _get_dataset_features(did_cache_dir, dataset_id):
         with io.open(features_file, "w", encoding='utf8') as fh:
             fh.write(features_xml)
 
-    features = xmltodict.parse(features_xml)["oml:data_features"]
+    features = xmltodict.parse(features_xml, force_list=('oml:feature',))["oml:data_features"]
 
     return features
 
@@ -454,7 +451,7 @@ def _get_dataset_qualities(did_cache_dir, dataset_id):
         with io.open(qualities_file, "w", encoding='utf8') as fh:
             fh.write(qualities_xml)
 
-    qualities = xmltodict.parse(qualities_xml)['oml:data_qualities']
+    qualities = xmltodict.parse(qualities_xml, force_list=('oml:quality',))['oml:data_qualities']
 
     return qualities
 
@@ -479,13 +476,17 @@ def _create_dataset_cache_directory(dataset_id):
     str
         Path of the created dataset cache directory.
     """
-    dataset_cache_dir = os.path.join(config.get_cache_directory(), "datasets",
-                                     str(dataset_id))
-    try:
-        os.makedirs(dataset_cache_dir)
-    except (OSError, IOError):
-        # TODO add debug information!
+    dataset_cache_dir = os.path.join(
+        config.get_cache_directory(),
+        "datasets",
+        str(dataset_id),
+    )
+    if os.path.exists(dataset_cache_dir) and os.path.isdir(dataset_cache_dir):
         pass
+    elif os.path.exists(dataset_cache_dir) and not os.path.isdir(dataset_cache_dir):
+        raise ValueError('Dataset cache dir exists but is not a directory!')
+    else:
+        os.makedirs(dataset_cache_dir)
     return dataset_cache_dir
 
 
@@ -498,13 +499,10 @@ def _remove_dataset_cache_dir(did_cache_dir):
     ----------
     """
     try:
-        os.rmdir(did_cache_dir)
+        shutil.rmtree(did_cache_dir)
     except (OSError, IOError):
-        try:
-            shutil.rmtree(did_cache_dir)
-        except (OSError, IOError):
-            raise ValueError('Cannot remove faulty dataset cache directory %s.'
-                             'Please do this manually!' % did_cache_dir)
+        raise ValueError('Cannot remove faulty dataset cache directory %s.'
+                         'Please do this manually!' % did_cache_dir)
 
 
 def _create_dataset_from_description(description, features, qualities, arff_file):

openml/evaluations/functions.py

@@ -61,21 +61,17 @@ def _list_evaluations(api_call):
 
     xml_string = _perform_api_call(api_call)
 
-    evals_dict = xmltodict.parse(xml_string)
+    evals_dict = xmltodict.parse(xml_string, force_list=('oml:evaluation',))
     # Minimalistic check if the XML is useful
     if 'oml:evaluations' not in evals_dict:
         raise ValueError('Error in return XML, does not contain "oml:evaluations": %s'
                          % str(evals_dict))
 
-    if isinstance(evals_dict['oml:evaluations']['oml:evaluation'], list):
-        evals_list = evals_dict['oml:evaluations']['oml:evaluation']
-    elif isinstance(evals_dict['oml:evaluations']['oml:evaluation'], dict):
-        evals_list = [evals_dict['oml:evaluations']['oml:evaluation']]
-    else:
-        raise TypeError()
+    assert type(evals_dict['oml:evaluations']['oml:evaluation']) == list, \
+        type(evals_dict['oml:evaluations'])
 
     evals = dict()
-    for eval_ in evals_list:
+    for eval_ in evals_dict['oml:evaluations']['oml:evaluation']:
         run_id = int(eval_['oml:run_id'])
         array_data = None
         if 'oml:array_data' in eval_:

openml/flows/functions.py

@@ -107,7 +107,7 @@ def flow_exists(name, external_version):
 def _list_flows(api_call):
     # TODO add proper error handling here!
     xml_string = _perform_api_call(api_call)
-    flows_dict = xmltodict.parse(xml_string)
+    flows_dict = xmltodict.parse(xml_string, force_list=('oml:flow',))
 
     # Minimalistic check if the XML is useful
     assert type(flows_dict['oml:flows']['oml:flow']) == list, \

openml/runs/functions.py

@@ -583,9 +583,6 @@ def get_run(run_id):
 
     run = _create_run_from_xml(run_xml)
 
-    with io.open(run_file, "w", encoding='utf8') as fh:
-        fh.write(run_xml)
-
     return run
 
 
@@ -680,10 +677,18 @@ def _create_run_from_xml(xml):
                          'description XML' % run_id)
 
     if 'predictions' not in files:
-        # JvR: actually, I am not sure whether this error should be raised.
-        # a run can consist without predictions. But for now let's keep it
-        raise ValueError('No prediction files for run %d in run '
-                         'description XML' % run_id)
+        task = openml.tasks.get_task(task_id)
+        if task.task_type_id == 8:
+            raise NotImplementedError(
+                'Subgroup discovery tasks are not yet supported.'
+            )
+        else:
+            # JvR: actually, I am not sure whether this error should be raised.
+            # a run can consist without predictions. But for now let's keep it
+            # Matthias: yes, it should stay as long as we do not really handle
+            # this stuff
+            raise ValueError('No prediction files for run %d in run '
+                             'description XML' % run_id)
 
     tags = openml.utils.extract_xml_tags('oml:tag', run)
 
@@ -702,14 +707,17 @@ def _create_run_from_xml(xml):
 
 
 def _create_trace_from_description(xml):
-    result_dict = xmltodict.parse(xml)['oml:trace']
+    result_dict = xmltodict.parse(xml, force_list=('oml:trace_iteration',))['oml:trace']
 
     run_id = result_dict['oml:run_id']
     trace = dict()
 
     if 'oml:trace_iteration' not in result_dict:
         raise ValueError('Run does not contain valid trace. ')
 
+    assert type(result_dict['oml:trace_iteration']) == list, \
+        type(result_dict['oml:trace_iteration'])
+
     for itt in result_dict['oml:trace_iteration']:
         repeat = int(itt['oml:repeat'])
         fold = int(itt['oml:fold'])
@@ -857,7 +865,7 @@ def _list_runs(api_call):
 
     xml_string = _perform_api_call(api_call)
 
-    runs_dict = xmltodict.parse(xml_string)
+    runs_dict = xmltodict.parse(xml_string, force_list=('oml:run',))
     # Minimalistic check if the XML is useful
     if 'oml:runs' not in runs_dict:
         raise ValueError('Error in return XML, does not contain "oml:runs": %s'
@@ -872,15 +880,11 @@ def _list_runs(api_call):
                          '"http://openml.org/openml": %s'
                          % str(runs_dict))
 
-    if isinstance(runs_dict['oml:runs']['oml:run'], list):
-        runs_list = runs_dict['oml:runs']['oml:run']
-    elif isinstance(runs_dict['oml:runs']['oml:run'], dict):
-        runs_list = [runs_dict['oml:runs']['oml:run']]
-    else:
-        raise TypeError()
+    assert type(runs_dict['oml:runs']['oml:run']) == list, \
+        type(runs_dict['oml:runs'])
 
     runs = dict()
-    for run_ in runs_list:
+    for run_ in runs_dict['oml:runs']['oml:run']:
         run_id = int(run_['oml:run_id'])
         run = {'run_id': run_id,
                'task_id': int(run_['oml:task_id']),

openml/setups/functions.py

@@ -116,7 +116,7 @@ def _list_setups(api_call):
 
     xml_string = openml._api_calls._perform_api_call(api_call)
 
-    setups_dict = xmltodict.parse(xml_string)
+    setups_dict = xmltodict.parse(xml_string, force_list=('oml:setup',))
     # Minimalistic check if the XML is useful
     if 'oml:setups' not in setups_dict:
         raise ValueError('Error in return XML, does not contain "oml:setups": %s'
@@ -131,15 +131,11 @@ def _list_setups(api_call):
                          '"http://openml.org/openml": %s'
                          % str(setups_dict))
 
-    if isinstance(setups_dict['oml:setups']['oml:setup'], list):
-        setups_list = setups_dict['oml:setups']['oml:setup']
-    elif isinstance(setups_dict['oml:setups']['oml:setup'], dict):
-        setups_list = [setups_dict['oml:setups']['oml:setup']]
-    else:
-        raise TypeError()
+    assert type(setups_dict['oml:setups']['oml:setup']) == list, \
+        type(setups_dict['oml:setups'])
 
     setups = dict()
-    for setup_ in setups_list:
+    for setup_ in setups_dict['oml:setups']['oml:setup']:
         # making it a dict to give it the right format
         current = _create_setup_from_xml({'oml:setup_parameters': setup_})
         setups[current.setup_id] = current