BUG: fix reading pre-0.14.1 pickles of containers with one block and dup items

doc/source/v0.15.0.txt

@@ -191,6 +191,8 @@ Bug Fixes
 
 - Bug in pickles contains ``DateOffset`` may raise ``AttributeError`` when ``normalize`` attribute is reffered internally (:issue:`7748`)
 
+- Bug in pickle deserialization that failed for pre-0.14.1 containers with dup items trying to avoid ambiguity
+  when matching block and manager items, when there's only one block there's no ambiguity (:issue:`7794`)
 
 
 - Bug in ``is_superperiod`` and ``is_subperiod`` cannot handle higher frequencies than ``S`` (:issue:`7760`, :issue:`7772`, :issue:`7803`)

pandas/core/internals.py

@@ -2271,10 +2271,23 @@ def unpickle_block(values, mgr_locs):
             ax_arrays, bvalues, bitems = state[:3]
 
             self.axes = [_ensure_index(ax) for ax in ax_arrays]
+
+            if len(bitems) == 1 and self.axes[0].equals(bitems[0]):
+                # This is a workaround for pre-0.14.1 pickles that didn't
+                # support unpickling multi-block frames/panels with non-unique
+                # columns/items, because given a manager with items ["a", "b",
+                # "a"] there's no way of knowing which block's "a" is where.
+                #
+                # Single-block case can be supported under the assumption that
+                # block items corresponded to manager items 1-to-1.
+                all_mgr_locs = [slice(0, len(bitems[0]))]
+            else:
+                all_mgr_locs = [self.axes[0].get_indexer(blk_items)
+                                for blk_items in bitems]
+
             self.blocks = tuple(
-                unpickle_block(values,
-                               self.axes[0].get_indexer(items))
-                for values, items in zip(bvalues, bitems))
+                unpickle_block(values, mgr_locs)
+                for values, mgr_locs in zip(bvalues, all_mgr_locs))
 
         self._post_setstate()
 

pandas/io/tests/generate_legacy_pickles.py

@@ -1,6 +1,7 @@
 """ self-contained to write legacy pickle files """
 from __future__ import print_function
 
+
 def _create_sp_series():
 
     import numpy as np
@@ -53,6 +54,7 @@ def _create_sp_frame():
 def create_data():
     """ create the pickle data """
 
+    from distutils.version import LooseVersion
     import numpy as np
     import pandas
     from pandas import (Series,TimeSeries,DataFrame,Panel,
@@ -92,13 +94,23 @@ def create_data():
                                 index=MultiIndex.from_tuples(tuple(zip(*[['bar','bar','baz','baz','baz'],
                                                                        ['one','two','one','two','three']])),
                                                              names=['first','second'])),
-                 dup = DataFrame(np.arange(15).reshape(5, 3).astype(np.float64),
-                                 columns=['A', 'B', 'A']))
+                 dup=DataFrame(np.arange(15).reshape(5, 3).astype(np.float64),
+                               columns=['A', 'B', 'A']))
     panel = dict(float = Panel(dict(ItemA = frame['float'], ItemB = frame['float']+1)),
                  dup = Panel(np.arange(30).reshape(3, 5, 2).astype(np.float64),
                              items=['A', 'B', 'A']))
 
+    if LooseVersion(pandas.__version__) >= '0.14.1':
+        # Pre-0.14.1 versions generated non-unpicklable mixed-type frames and
+        # panels if their columns/items were non-unique.
+        mixed_dup_df = DataFrame(data)
+        mixed_dup_df.columns = list("ABCDA")
+
+        mixed_dup_panel = Panel(dict(ItemA=frame['float'], ItemB=frame['int']))
+        mixed_dup_panel.items = ['ItemA', 'ItemA']
 
+        frame['mixed_dup'] = mixed_dup_df
+        panel['mixed_dup'] = mixed_dup_panel
 
     return dict( series = series,
                  frame = frame,