Add more tests for group names and refs in RE (GH-91695)

Author: serhiy-storchaka

Lib/test/test_re.py

@@ -236,6 +236,16 @@ def test_symbolic_groups(self):
         re.compile(r'(?P<a>x)(?P=a)(?(a)y)')
         re.compile(r'(?P<a1>x)(?P=a1)(?(a1)y)')
         re.compile(r'(?P<a1>x)\1(?(1)y)')
+        re.compile(b'(?P<a1>x)(?P=a1)(?(a1)y)')
+        # New valid identifiers in Python 3
+        re.compile('(?P<µ>x)(?P=µ)(?(µ)y)')
+        re.compile('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)(?P=𝔘𝔫𝔦𝔠𝔬𝔡𝔢)(?(𝔘𝔫𝔦𝔠𝔬𝔡𝔢)y)')
+        # Support > 100 groups.
+        pat = '|'.join('x(?P<a%d>%x)y' % (i, i) for i in range(1, 200 + 1))
+        pat = '(?:%s)(?(200)z|t)' % pat
+        self.assertEqual(re.match(pat, 'xc8yz').span(), (0, 5))
+
+    def test_symbolic_groups_errors(self):
         self.checkPatternError(r'(?P<a>)(?P<a>)',
                                "redefinition of group name 'a' as group 2; "
                                "was group 1")
@@ -261,16 +271,22 @@ def test_symbolic_groups(self):
         self.checkPatternError(r'(?(-1))', "bad character in group name '-1'", 3)
         self.checkPatternError(r'(?(1a))', "bad character in group name '1a'", 3)
         self.checkPatternError(r'(?(a.))', "bad character in group name 'a.'", 3)
-        # New valid/invalid identifiers in Python 3
-        re.compile('(?P<µ>x)(?P=µ)(?(µ)y)')
-        re.compile('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)(?P=𝔘𝔫𝔦𝔠𝔬𝔡𝔢)(?(𝔘𝔫𝔦𝔠𝔬𝔡𝔢)y)')
         self.checkPatternError('(?P<©>x)', "bad character in group name '©'", 4)
+        self.checkPatternError('(?P=©)', "bad character in group name '©'", 4)
+        self.checkPatternError('(?(©)y)', "bad character in group name '©'", 3)
+
+    def test_symbolic_refs(self):
+        self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\g<b>', 'xx'), '')
+        self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\2', 'xx'), '')
+        self.assertEqual(re.sub(b'(?P<a1>x)', br'\g<a1>', b'xx'), b'xx')
+        # New valid identifiers in Python 3
+        self.assertEqual(re.sub('(?P<µ>x)', r'\g<µ>', 'xx'), 'xx')
+        self.assertEqual(re.sub('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)', r'\g<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>', 'xx'), 'xx')
         # Support > 100 groups.
         pat = '|'.join('x(?P<a%d>%x)y' % (i, i) for i in range(1, 200 + 1))
-        pat = '(?:%s)(?(200)z|t)' % pat
-        self.assertEqual(re.match(pat, 'xc8yz').span(), (0, 5))
+        self.assertEqual(re.sub(pat, r'\g<200>', 'xc8yzxc8y'), 'c8zc8')
 
-    def test_symbolic_refs(self):
+    def test_symbolic_refs_errors(self):
         self.checkTemplateError('(?P<a>x)', r'\g<a', 'xx',
                                 'missing >, unterminated name', 3)
         self.checkTemplateError('(?P<a>x)', r'\g<', 'xx',
@@ -288,18 +304,14 @@ def test_symbolic_refs(self):
                                 'invalid group reference 2', 1)
         with self.assertRaisesRegex(IndexError, "unknown group name 'ab'"):
             re.sub('(?P<a>x)', r'\g<ab>', 'xx')
-        self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\g<b>', 'xx'), '')
-        self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\2', 'xx'), '')
         self.checkTemplateError('(?P<a>x)', r'\g<-1>', 'xx',
                                 "bad character in group name '-1'", 3)
-        # New valid/invalid identifiers in Python 3
-        self.assertEqual(re.sub('(?P<µ>x)', r'\g<µ>', 'xx'), 'xx')
-        self.assertEqual(re.sub('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)', r'\g<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>', 'xx'), 'xx')
         self.checkTemplateError('(?P<a>x)', r'\g<©>', 'xx',
                                 "bad character in group name '©'", 3)
-        # Support > 100 groups.
-        pat = '|'.join('x(?P<a%d>%x)y' % (i, i) for i in range(1, 200 + 1))
-        self.assertEqual(re.sub(pat, r'\g<200>', 'xc8yzxc8y'), 'c8zc8')
+        self.checkTemplateError('(?P<a>x)', r'\g<㊀>', 'xx',
+                                "bad character in group name '㊀'", 3)
+        self.checkTemplateError('(?P<a>x)', r'\g<¹>', 'xx',
+                                "bad character in group name '¹'", 3)
 
     def test_re_subn(self):
         self.assertEqual(re.subn("(?i)b+", "x", "bbbb BBBB"), ('x x', 2))
@@ -561,9 +573,23 @@ def test_re_groupref_exists(self):
         pat = '(?:%s)(?(200)z)' % pat
         self.assertEqual(re.match(pat, 'xc8yz').span(), (0, 5))
 
-        self.checkPatternError(r'(?P<a>)(?(0))', 'bad group number', 10)
+    def test_re_groupref_exists_errors(self):
+        self.checkPatternError(r'(?P<a>)(?(0)a|b)', 'bad group number', 10)
+        self.checkPatternError(r'()(?(-1)a|b)',
+                               "bad character in group name '-1'", 5)
+        self.checkPatternError(r'()(?(㊀)a|b)',
+                               "bad character in group name '㊀'", 5)
+        self.checkPatternError(r'()(?(¹)a|b)',
+                               "bad character in group name '¹'", 5)
+        self.checkPatternError(r'()(?(1',
+                               "missing ), unterminated name", 5)
+        self.checkPatternError(r'()(?(1)a',
+                               "missing ), unterminated subpattern", 2)
         self.checkPatternError(r'()(?(1)a|b',
                                'missing ), unterminated subpattern', 2)
+        self.checkPatternError(r'()(?(1)a|b|c',
+                               'conditional backref with more than '
+                               'two branches', 10)
         self.checkPatternError(r'()(?(1)a|b|c)',
                                'conditional backref with more than '
                                'two branches', 10)