Remove pypfilt.summary.dtype_names_to_str

All dtype field names are already Unicode strings, and we no longer need
to check for byte strings and convert them to Unicode.

doc/api/summary.rst

@@ -63,8 +63,6 @@ The following column types are provided for convenience when defining custom
 
 .. autofunction:: pypfilt.summary.dtype_value
 
-.. autofunction:: pypfilt.summary.dtype_names_to_str
-
 The following functions are provided for converting column types in structured
 arrays.
 

src/pypfilt/summary.py

@@ -31,37 +31,6 @@ def dtype_value(value, name='value'):
     return (name, value_dtype)
 
 
-def dtype_names_to_str(dtypes, encoding='utf-8'):
-    """
-    Ensure that dtype field names are native strings, as
-    `required <https://github.com/numpy/numpy/issues/2407>`__ by NumPy.
-
-    :param dtypes: A list of fields where each field is either a string, or a
-        tuple of length 2 or 3 (see the NumPy
-        `docs <http://docs.scipy.org/doc/numpy-1.8.0/reference/arrays.dtypes.html#index-9>`__
-        for details).
-    :param encoding: The encoding for converting Unicode strings to native
-        strings.
-    :return: A list of fields, where each field name is a native string
-        (``str`` type).
-    :raises ValueError: If a name cannot be converted to a native string.
-    """
-
-    def str_name(value):
-        if isinstance(value, str):
-            return value
-        elif isinstance(value, bytes):
-            return value.decode(encoding)
-        else:
-            raise ValueError("Invalid column name '{!r}'".format(value))
-
-    # Inspect the first element to see if we have a list of strings or tuples.
-    if isinstance(dtypes[0], tuple):
-        return [(str_name(dtype[0]),) + dtype[1:] for dtype in dtypes]
-    else:
-        return [str_name(dtype) for dtype in dtypes]
-
-
 def obs_types(params, obs_list, obs_reqd=False):
     """
     Return a sorted list of ``(unit, period)`` tuples that define the unique
@@ -692,8 +661,8 @@ class HDF5(object):
             self.__tbl_dict[name] = table
             # NOTE: provide the table name here so that the table can look for
             # table-specific parameters.
-            self.__dtypes[name] = dtype_names_to_str(
-                table.dtype(self.__ctx, self.__all_obs, name))
+            self.__dtypes[name] = table.dtype(self.__ctx, self.__all_obs,
+                                              name)
 
     def load_state(self, grp):
         """
@@ -1328,8 +1297,7 @@ def obs_table(ctx, all_obs):
     value = dtype_value(obs_val)
     incomplete = ('incomplete', np.bool)
     upper = dtype_value(obs_val, name='upper_bound')
-    dtype = dtype_names_to_str([unit, period, source, date, value, incomplete,
-                                upper])
+    dtype = [unit, period, source, date, value, incomplete, upper]
     obs_df = np.empty(len(all_obs), dtype=dtype)
     for ix, o in enumerate(all_obs):
         date_enc = ctx.component['time'].to_dtype(o['date'])
@@ -1351,8 +1319,8 @@ def convert_cols(data, converters):
     :returns: A new structured array.
     """
     dt = data.dtype
-    col_names = dtype_names_to_str(dt.names)
-    conv_names = dtype_names_to_str([n for n in converters.keys()])
+    col_names = dt.names
+    conv_names = [n for n in converters.keys()]
     conv_fns = {}
     new_dtype = []
     for col in col_names: