Shows how unicode & str types, utf-8 and ascii encodings interoperate with each other in Python 2.x

unicode_examples.py

import sys

def _helper(s, method, encoding):
    try:
        print 'print s.%s(\'%s\'): %s' % (method, encoding, getattr(s, method)(encoding))
    except UnicodeError as e:
        print 'print s.%s(\'%s\'): %s' % (method, encoding, e)


def test_encodings(s):
    print 'print s: %s' % s
    _helper(s, 'encode', 'utf-8')
    _helper(s, 'encode', 'ascii')
    _helper(s, 'encode', 'unicode-escape')
    _helper(s, 'decode', 'utf-8')
    _helper(s, 'decode', 'ascii')
    _helper(s, 'decode', 'unicode-escape')
    print '\n'

print 'your python encoding is set to: %s\n' % sys.getdefaultencoding()

packed_bytes_str = '\\xe3\\x84\\x8a'
print 's = \'%s\'' % packed_bytes_str
packed_bytes_str = '\xe3\x84\x8a'
test_encodings(packed_bytes_str)

packed_bytes_unicode = u'\\xe3\\x84\\x8a'
print 's = u\'%s\'' % packed_bytes_unicode
packed_bytes_unicode = u'\xe3\x84\x8a'
test_encodings(packed_bytes_unicode)

utf8_str = '\\u310a'
print 's = \'%s\'' % utf8_str
utf8_str = '\u310a'
test_encodings(utf8_str)

utf8_unicode = u'\\u310a'
print 's = u\'%s\'' % utf8_unicode
utf8_unicode = u'\u310a'
test_encodings(utf8_unicode)