Support additional Pandas formats if dependencies are satisfied - see a…

…pache#3302

contrib/connectors/pandas/models.py

@@ -34,6 +34,27 @@
 from superset.utils import QueryStatus
 
 
+FORMATS = [
+    ('csv', 'csv'),
+    ('html', 'html'),
+    ('json', 'json'),
+    ('excel', 'Microsoft Excel'),
+    ('stata', 'Stata'),
+]
+
+try:
+    import tables  # NOQA
+    FORMATS.append(('hdf', 'HDF5'))
+except ImportError:
+    pass
+
+try:
+    import feather  # NOQA
+    FORMATS.append(('feather', 'Feather'))
+except ImportError:
+    pass
+
+
 class PandasDatabase(object):
     """Non-ORM object for a Pandas Source"""
 
@@ -123,13 +144,6 @@ def perm(self):
 class PandasDatasource(Model, BaseDatasource):
     """A datasource based on a Pandas DataFrame"""
 
-    FORMATS = [
-        ('csv', 'csv'),
-        ('html', 'html'),
-        ('json', 'json'),
-        ('excel', 'Microsoft Excel'),
-    ]
-
     # See http://pandas.pydata.org/pandas-docs/stable/timeseries.html#offset-aliases # NOQA
     GRAINS = OrderedDict([
         ('5 seconds', '5S'),
@@ -577,7 +591,7 @@ def process_dataframe(
             # If there is more than one DataFrame in the list then
             # concatenate them along the index
             if len(dfs) > 1:
-                df = pd.concat(dfs, axis=1)
+                df = pd.concat(dfs, axis=0)
                 query_str = 'pd.concat([{}])'.format(', '.join(query_strs))
             else:
                 df = dfs[0]
@@ -627,9 +641,10 @@ def process_dataframe(
                     ascending=ascending)
 
             # Remove metrics only added for post-aggregation filtering
-            df = df.drop(filtered_metrics, axis=1)
-            query_str += '.drop({filtered_metrics}, axis=1)'.format(
-                filtered_metrics=filtered_metrics)
+            if filtered_metrics:
+                df = df.drop(filtered_metrics, axis=1)
+                query_str += '.drop({filtered_metrics}, axis=1)'.format(
+                    filtered_metrics=filtered_metrics)
 
         elif groupby:
             # Group by without any metrics is equivalent to SELECT DISTINCT,

contrib/connectors/pandas/views.py

@@ -20,7 +20,7 @@
     get_datasource_exist_error_mgs,
 )
 
-from .models import PandasDatasource, PandasColumn, PandasMetric
+from .models import FORMATS, PandasDatasource, PandasColumn, PandasMetric
 
 
 class ChoiceTypeSelectField(SelectField):
@@ -158,8 +158,7 @@ class PandasDatasourceModelView(DatasourceModelView, DeleteMixin):  # noqa
         'link', 'changed_on_']
     add_columns = ['name', 'source_url', 'format']
     add_form_extra_fields = {
-        'format': ChoiceTypeSelectField(_('Format'),
-                                        choices=PandasDatasource.FORMATS)
+        'format': ChoiceTypeSelectField(_('Format'), choices=FORMATS)
     }
     edit_columns = [
         'name', 'source_url', 'format',
@@ -168,8 +167,7 @@ class PandasDatasourceModelView(DatasourceModelView, DeleteMixin):  # noqa
         'description', 'owner',
         'main_dttm_col', 'default_endpoint', 'offset', 'cache_timeout']
     edit_form_extra_fields = {
-        'format': ChoiceTypeSelectField(_('Format'),
-                                        choices=PandasDatasource.FORMATS)
+        'format': ChoiceTypeSelectField(_('Format'), choices=FORMATS)
     }
     show_columns = edit_columns + ['perm']
     related_views = [PandasColumnInlineView, PandasMetricInlineView]

setup.py

@@ -44,6 +44,7 @@ def get_git_sha():
     install_requires=[
         'beautifulsoup4==4.6.0',
         'boto3==1.4.4',
+        'bottleneck==1.2.1',
         'celery==3.1.25',
         'colorama==0.3.9',
         'cryptography==1.9',
@@ -63,6 +64,7 @@ def get_git_sha():
         'idna==2.5',
         'lxml==3.8.0',
         'markdown==2.6.8',
+        'numexpr==2.6.4',
         'pandas==0.20.3',
         'parsedatetime==2.0.0',
         'pydruid==0.3.1',