BUG: GH3493 fix Cannot append DataFrames with uint dtypes to HDFStore

RELEASE.rst

@@ -58,6 +58,7 @@ pandas 0.12.0
   - Duplicate indexes with getitem will return items in the correct order (GH3455_, GH3457_)
   - Fix sorting in a frame with a list of columns which contains datetime64[ns] dtypes (GH3461_)
   - DataFrames fetched via FRED now handle '.' as a NaN. (GH3469_)
+  - Fix issue when storing uint dtypes in an HDFStore. (GH3493_)
 
 .. _GH3164: https://github.com/pydata/pandas/issues/3164
 .. _GH3251: https://github.com/pydata/pandas/issues/3251

pandas/io/pytables.py

@@ -1284,8 +1284,17 @@ def set_atom_string(self, block, existing_col, min_itemsize, nan_rep):
     def convert_string_data(self, data, itemsize):
         return data.astype('S%s' % itemsize)
 
+    def get_atom_coltype(self):
+        """ return the PyTables column class for this column """
+        if self.kind.startswith('uint'):
+            col_name = "UInt%sCol" % self.kind[4:]
+        else:
+            col_name = "%sCol" % self.kind.capitalize()
+
+        return getattr(_tables(), col_name)
+
     def get_atom_data(self, block):
-        return getattr(_tables(), "%sCol" % self.kind.capitalize())(shape=block.shape[0])
+        return self.get_atom_coltype()(shape=block.shape[0])
 
     def set_atom_data(self, block):
         self.kind = block.dtype.name
@@ -1383,7 +1392,7 @@ def get_atom_string(self, block, itemsize):
         return _tables().StringCol(itemsize=itemsize)
 
     def get_atom_data(self, block):
-        return getattr(_tables(), "%sCol" % self.kind.capitalize())()
+        return self.get_atom_coltype()()
 
     def get_atom_datetime64(self, block):
         return _tables().Int64Col()

pandas/io/tests/test_pytables.py

@@ -458,6 +458,21 @@ def test_append(self):
             store.append('df', df)
             tm.assert_frame_equal(store['df'], df)
 
+            # uints - test storage of uints
+            uint_data = DataFrame({'u08' : Series(np.random.random_integers(0, high=255, size=5), dtype=np.uint8),
+                                   'u16' : Series(np.random.random_integers(0, high=65535, size=5), dtype=np.uint16),
+                                   'u32' : Series(np.random.random_integers(0, high=2**30, size=5), dtype=np.uint32),
+                                   'u64' : Series([2**58, 2**59, 2**60, 2**61, 2**62], dtype=np.uint64)},
+                                  index=np.arange(5))
+            _maybe_remove(store, 'uints')
+            store.append('uints', uint_data)
+            tm.assert_frame_equal(store['uints'], uint_data)
+
+            # uints - test storage of uints in indexable columns
+            _maybe_remove(store, 'uints')
+            store.append('uints', uint_data, data_columns=['u08','u16','u32']) # 64-bit indices not yet supported
+            tm.assert_frame_equal(store['uints'], uint_data)
+
     def test_append_some_nans(self):
 
         with ensure_clean(self.path) as store: