BUG: Ensure data_columns is always a list (i.e. min_itemsize can extend it)

closes pandas-dev#10381

Author: toobaz

pandas/io/pytables.py

@@ -3247,7 +3247,7 @@ def validate_data_columns(self, data_columns, min_itemsize):
         # evaluate the passed data_columns, True == use all columns
         # take only valide axis labels
         if data_columns is True:
-            data_columns = axis_labels
+            data_columns = list(axis_labels)
         elif data_columns is None:
             data_columns = []
 
@@ -4084,7 +4084,7 @@ def write(self, obj, data_columns=None, **kwargs):
             obj = DataFrame({name: obj}, index=obj.index)
             obj.columns = [name]
         return super(AppendableSeriesTable, self).write(
-            obj=obj, data_columns=obj.columns, **kwargs)
+            obj=obj, data_columns=list(obj.columns), **kwargs)
 
     def read(self, columns=None, **kwargs):
 
@@ -4185,7 +4185,7 @@ def write(self, obj, data_columns=None, **kwargs):
         if data_columns is None:
             data_columns = []
         elif data_columns is True:
-            data_columns = obj.columns[:]
+            data_columns = list(obj.columns[:])
         obj, self.levels = self.validate_multiindex(obj)
         for n in self.levels:
             if n not in data_columns:

pandas/io/tests/test_pytables.py

@@ -1338,6 +1338,16 @@ def check_col(key, name, size):
                 [[124, 'abcdefqhij'], [346, 'abcdefghijklmnopqrtsuvwxyz']])
             self.assertRaises(ValueError, store.append, 'df_new', df_new)
 
+            # min_itemsize on Series with Multiindex (GH 10381)
+            df = tm.makeMixedDataFrame().set_index(['A', 'C'])
+            store.append('ss', df['B'], min_itemsize={'index': 4})
+            tm.assert_series_equal(store.select('ss'), df['B'])
+
+            # min_itemsize with MultiIndex and data_columns=True
+            store.append('midf', df, data_columns=True,
+                         min_itemsize={'index': 4})
+            tm.assert_frame_equal(store.select('midf'), df)
+
             # with nans
             _maybe_remove(store, 'df')
             df = tm.makeTimeDataFrame()