ARROW-1883: [Python] Fix handling of metadata in to_pandas when not all columns are present

python/pyarrow/pandas_compat.py

@@ -530,19 +530,32 @@ def _add_any_metadata(table, pandas_metadata):
 
     schema = table.schema
 
+    n_index_levels = len(pandas_metadata['index_columns'])
+    n_columns = len(pandas_metadata['columns']) - n_index_levels
+
     # Add time zones
     for i, col_meta in enumerate(pandas_metadata['columns']):
-        if col_meta['pandas_type'] == 'datetimetz':
-            col = table[i]
-            converted = col.to_pandas()
-            tz = col_meta['metadata']['timezone']
-            tz_aware_type = pa.timestamp('ns', tz=tz)
-            with_metadata = pa.Array.from_pandas(converted.values,
-                                                 type=tz_aware_type)
 
-            field = pa.field(schema[i].name, tz_aware_type)
-            modified_columns[i] = pa.Column.from_array(field,
-                                                       with_metadata)
+        raw_name = col_meta['name']
+        if i >= n_columns:
+            # index columns
+            raw_name = pandas_metadata['index_columns'][i - n_columns]
+        if raw_name is None:
+            raw_name = 'None'
+
+        idx = schema.get_field_index(raw_name)
+        if idx != -1:
+            if col_meta['pandas_type'] == 'datetimetz':
+                col = table[idx]
+                converted = col.to_pandas()
+                tz = col_meta['metadata']['timezone']
+                tz_aware_type = pa.timestamp('ns', tz=tz)
+                with_metadata = pa.Array.from_pandas(converted.values,
+                                                     type=tz_aware_type)
+
+                field = pa.field(schema[idx].name, tz_aware_type)
+                modified_columns[idx] = pa.Column.from_array(field,
+                                                             with_metadata)
 
     if len(modified_columns) > 0:
         columns = []

python/pyarrow/tests/test_convert_pandas.py

@@ -1221,6 +1221,21 @@ class TestPandasConversion(object):
         assert pa.Array.from_pandas(expected,
                                     type=pa.list_(t())).equals(result)
 
+    def test_table_column_subset_metadata(self):
+        # ARROW-1883
+        df = pd.DataFrame({
+            'a': [1, 2, 3],
+            'b': pd.date_range("2017-01-01", periods=3, tz='Europe/Brussels')})
+        table = pa.Table.from_pandas(df)
+
+        table_subset = table.remove_column(1)
+        result = table_subset.to_pandas()
+        tm.assert_frame_equal(result, df[['a']])
+
+        table_subset2 = table.remove_column(1)
+        result = table_subset2.to_pandas()
+        tm.assert_frame_equal(result, df[['a']])
+
 
 def _pytime_from_micros(val):
     microseconds = val % 1000000