BUG GH9057

Have DataFrame.to_hdf() raise an error when using pytables with non-string
column types

Author: jameshiebert

pandas/io/pytables.py

@@ -257,6 +257,30 @@ def _tables():
 def to_hdf(path_or_buf, key, value, mode=None, complevel=None, complib=None,
            append=None, **kwargs):
     """ store this object, close it if we opened it """
+
+    # PyTables has some limitations that we need to check for
+    if kwargs.get('format', None) == 'table':
+        if value.columns.dtype.kind == 'S':
+            pass
+        elif value.columns.dtype.kind == 'O':
+            # PyTables doesn't support Unicode columns, but types 'O' can still
+            # contain unicode values so check for that
+            for col in value.columns:
+                try:
+                    col.encode('ascii').decode('ascii')
+                except UnicodeEncodeError:
+                    raise ValueError(
+                        "PyTables does not support UTF-8 column headers\n"
+                        "but header {} is at least not encodable to ASCII"
+                        .format(col)
+                    )
+        else:
+            raise ValueError(
+                "Storing data in HDF with PyTables requires that the column\n"
+                "types be of some string type or an object.\n"
+                "    Instead, your columns: {}\n"
+                "    are of type {}".format(value.columns, value.columns.dtype))
+
     if append:
         f = lambda store: store.append(key, value, **kwargs)
     else:

pandas/io/tests/test_pytables.py

@@ -4617,6 +4617,19 @@ def test_preserve_timedeltaindex_type(self):
             store['df'] = df
             assert_frame_equal(store['df'], df)
 
+    def test_to_hdf_with_integer_column_names(self):
+        # GH9057
+        df = DataFrame(columns=[5, 10], data=[[1, 0]])
+
+        with ensure_clean_path(self.path) as path:
+            self.assertRaises(ValueError, df.to_hdf, path, 'df', format='table', data_columns=True)
+
+    def test_to_hdf_with_unicode_column_names(self):
+        # GH9057
+        df = DataFrame(columns=[u('\u1234'), u('\u2345')], data=[[1, 0]])
+
+        with ensure_clean_path(self.path) as path:
+            self.assertRaises(ValueError, df.to_hdf, path, 'df', format='table', data_columns=True)
 
 def _test_sort(obj):
     if isinstance(obj, DataFrame):