Method test_named_unicode_escapes

Lib/test/test_re.py:863–901

(self)

                self.assertRaises(re.PatternError, re.compile, '[\\%c]' % c)

    def test_named_unicode_escapes(self):
        # test individual Unicode named escapes
        self.assertTrue(re.match(r'\N{LESS-THAN SIGN}', '<'))
        self.assertTrue(re.match(r'\N{less-than sign}', '<'))
        self.assertIsNone(re.match(r'\N{LESS-THAN SIGN}', '>'))
        self.assertTrue(re.match(r'\N{SNAKE}', '\U0001f40d'))
        self.assertTrue(re.match(r'\N{ARABIC LIGATURE UIGHUR KIRGHIZ YEH WITH '
                                 r'HAMZA ABOVE WITH ALEF MAKSURA ISOLATED FORM}',
                                 '\ufbf9'))
        self.assertTrue(re.match(r'[\N{LESS-THAN SIGN}-\N{GREATER-THAN SIGN}]',
                                 '='))
        self.assertIsNone(re.match(r'[\N{LESS-THAN SIGN}-\N{GREATER-THAN SIGN}]',
                                   ';'))

        # test errors in \N{name} handling - only valid names should pass
        self.checkPatternError(r'\N', 'missing {', 2)
        self.checkPatternError(r'[\N]', 'missing {', 3)
        self.checkPatternError(r'\N{', 'missing character name', 3)
        self.checkPatternError(r'[\N{', 'missing character name', 4)
        self.checkPatternError(r'\N{}', 'missing character name', 3)
        self.checkPatternError(r'[\N{}]', 'missing character name', 4)
        self.checkPatternError(r'\NSNAKE}', 'missing {', 2)
        self.checkPatternError(r'[\NSNAKE}]', 'missing {', 3)
        self.checkPatternError(r'\N{SNAKE',
                               'missing }, unterminated name', 3)
        self.checkPatternError(r'[\N{SNAKE]',
                               'missing }, unterminated name', 4)
        self.checkPatternError(r'[\N{SNAKE]}',
                               "undefined character name 'SNAKE]'", 1)
        self.checkPatternError(r'\N{SPAM}',
                               "undefined character name 'SPAM'", 0)
        self.checkPatternError(r'[\N{SPAM}]',
                               "undefined character name 'SPAM'", 1)
        self.checkPatternError(r'\N{KEYCAP NUMBER SIGN}',
                            "undefined character name 'KEYCAP NUMBER SIGN'", 0)
        self.checkPatternError(r'[\N{KEYCAP NUMBER SIGN}]',
                            "undefined character name 'KEYCAP NUMBER SIGN'", 1)
        self.checkPatternError(br'\N{LESS-THAN SIGN}', r'bad escape \N', 0)
        self.checkPatternError(br'[\N{LESS-THAN SIGN}]', r'bad escape \N', 1)

    def test_word_boundaries(self):
        # See http://bugs.python.org/issue10713