examples/unicode_jp: document unidecode as a potential solution

unidecode() provides a solution for ASCII representation of unicode
data.  In lieu of complicating the standard MakeIdentifier, document the
alternative solution in the JP example and the routine that allows
override of MakeIdentifier.

Closes pabigot#67

examples/unicode_jp/README.txt

@@ -66,6 +66,13 @@ the bindings.
 Many thanks to Hiroaki Itoh for providing the schemas, example document, and
 romanization code.
 
+If you are interested in other languages, consider replacing the
+ConvertJPIdentifier() function in the modified pyxbgen script with one
+that uses unidecode: https://pypi.python.org/pypi/Unidecode
+
+See this comment for further details:
+https://sourceforge.net/p/pyxb/discussion/956708/thread/5246b205/#1c7f
+
 Note: Because the package depends on OpenGIS, and OpenGIS bindings are no
 longer provided in the PyXB distribution, you should generate these bindings
 first.  If they are missing, the test script will emit a warning and PyXB

pyxb/utils/utility.py

@@ -132,6 +132,10 @@ def _SetXMLIdentifierToPython (xml_identifier_to_python):
     perform that translation before the invalid characters are
     stripped.
 
+    For example, see `unidecode
+    <https://pypi.python.org/pypi/Unidecode>`_ and `this forum posting
+    <https://sourceforge.net/p/pyxb/discussion/956708/thread/5246b205/#1c7f>`_.
+
     It is not the responsibility of this callable to do anything other
     than replace whatever characters it wishes to.  All
     transformations performed by L{MakeIdentifier} will still be
@@ -143,6 +147,7 @@ def _SetXMLIdentifierToPython (xml_identifier_to_python):
     default implementation, which is L{_DefaultXMLIdentifierToPython}.
 
     @rtype: C{unicode}
+
     """
     global _XMLIdentifierToPython
     if xml_identifier_to_python is None: