8000 Add more tests for group names and refs in RE (GH-91695) · python/cpython@7407008 · GitHub
[go: up one dir, main page]

Skip to content

Commit 7407008

Browse files
Add more tests for group names and refs in RE (GH-91695)
1 parent 16fc573 commit 7407008

File tree

1 file changed

+41
-15
lines changed

1 file changed

+41
-15
lines changed

Lib/test/test_re.py

Lines changed: 41 additions & 15 deletions
Original file line numberDiff line numberDiff line change
@@ -236,6 +236,16 @@ def test_symbolic_groups(self):
236236
re.compile(r'(?P<a>x)(?P=a)(?(a)y)')
237237
re.compile(r'(?P<a1>x)(?P=a1)(?(a1)y)')
238238
re.compile(r'(?P<a1>x)\1(?(1)y)')
239+
re.compile(b'(?P<a1>x)(?P=a1)(?(a1)y)')
240+
# New valid identifiers in Python 3
241+
re.compile('(?P<µ>x)(?P=µ)(?(µ)y)')
242+
re.compile('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)(?P=𝔘𝔫𝔦𝔠𝔬𝔡𝔢)(?(𝔘𝔫𝔦𝔠𝔬𝔡𝔢)y)')
243+
# Support > 100 groups.
244+
pat = '|'.join('x(?P<a%d>%x)y' % (i, i) for i in range(1, 200 + 1))
245+
pat = '(?:%s)(?(200)z|t)' % pat
246+
self.assertEqual(re.match(pat, 'xc8yz').span(), (0, 5))
247+
248+
def test_symbolic_groups_errors(self):
239249
self.checkPatternError(r'(?P<a>)(?P<a>)',
240250
"redefinition of group name 'a' as group 2; "
241251
"was group 1")
@@ -261,16 +271,22 @@ def test_symbolic_groups(self):
261271
self.checkPatternError(r'(?(-1))', "bad character in group name '-1'", 3)
262272
self.checkPatternError(r'(?(1a))', "bad character in group name '1a'", 3)
263273
self.checkPatternError(r'(?(a.))', "bad character in group name 'a.'", 3)
264-
# New valid/invalid identifiers in Python 3
265-
re.compile('(?P<µ>x)(?P=µ)(?(µ)y)')
266-
re.compile('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)(?P=𝔘𝔫𝔦𝔠𝔬𝔡𝔢)(?(𝔘𝔫𝔦𝔠𝔬𝔡𝔢)y)')
267274
self.checkPatternError('(?P<©>x)', "bad character in group name '©'", 4)
275+
self.checkPatternError('(?P=©)', "bad character in group name '©'", 4)
276+
self.checkPatternError('(?(©)y)', "bad character in group name '©'", 3)
277+
278+
def test_symbolic_refs(self):
279+
self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\g<b>', 'xx'), '')
280+
self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\2', 'xx'), '')
281+
self.assertEqual(re.sub(b'(?P<a1>x)', br'\g<a1>', b'xx'), b'xx')
282+
# New valid identifiers in Python 3
283+
self.assertEqual(re.sub('(?P<µ>x)', r'\g<µ>', 'xx'), 'xx')
284+
self.assertEqual(re.sub('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)', r'\g<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>', 'xx'), 'xx')
268285
# Support > 100 groups.
269286
pat = '|'.join('x(?P<a%d>%x)y' % (i, i) for i in range(1, 200 + 1))
270-
pat = '(?:%s)(?(200)z|t)' % pat
271-
self.assertEqual(re.match(pat, 'xc8yz').span(), (0, 5))
287+
self.assertEqual(re.sub(pat, r'\g<200>', 'xc8yzxc8y'), 'c8zc8')
272288

273-
def test_symbolic_refs(self):
289+
def test_symbolic_refs_errors(self):
274290
self.checkTemplateError('(?P<a>x)', r'\g<a', 'xx',
275291
'missing >, unterminated name', 3)
276292
self.checkTemplateError('(?P<a>x)', r'\g<', 'xx',
@@ -288,18 +304,14 @@ def test_symbolic_refs(self):
288304
'invalid group reference 2', 1)
289305
with self.assertRaisesRegex(IndexError, "unknown group name 'ab'"):
290306
re.sub('(?P<a>x)', r'\g<ab>', 'xx')
291-
self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\g<b>', 'xx'), '')
292-
self.assertEqual(re.sub('(?P<a>x)|(?P<b>y)', r'\2', 'xx'), '')
293307
self.checkTemplateError('(?P<a>x)', r'\g<-1>', 'xx',
294308
"bad character in group name '-1'", 3)
295-
# New valid/invalid identifiers in Python 3
296-
self.assertEqual(re.sub('(?P<µ>x)', r'\g<µ>', 'xx'), 'xx')
297-
self.assertEqual(re.sub('(?P<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>x)', r'\g<𝔘𝔫𝔦𝔠𝔬𝔡𝔢>', 'xx'), 'xx')
298309
self.checkTemplateError('(?P<a>x)', r'\g<©>', 'xx',
299310
"bad character in group name '©'", 3)
300-
# Support > 100 groups.
301-
pat = '|'.join('x(?P<a%d>%x)y' % (i, i) for i in range(1, 200 + 1))
302-
self.assertEqual(re.sub(pat, r'\g<200>', 'xc8yzxc8y'), 'c8zc8')
311+
self.checkTemplateError('(?P<a>x)', r'\g<㊀>', 'xx',
312+
"bad character in group name '㊀'", 3)
313+
self.checkTemplateError('(?P<a>x)', r'\g<¹>', 'xx',
314+
"bad character in group name '¹'", 3)
303315

304316
def test_re_subn(self):
305317
self.assertEqual(re.subn("(?i)b+", "x", "bbbb BBBB"), ('x x', 2))
@@ -561,9 +573,23 @@ def test_re_groupref_exists(self):
561573
pat = '(?:%s)(?(200)z)' % pat
562574
self.assertEqual(re.match(pat, 'xc8yz').span(), (0, 5))
563575

564-
self.checkPatternError(r'(?P<a>)(?(0))', 'bad group number', 10)
576+
def test_re_groupref_exists_errors(self):
577+
self.checkPatternError(r'(?P<a>)(?(0)a|b)', 'bad group number', 10)
578+
self.checkPatternError(r'()(?(-1)a|b)',
579+
"bad character in group name '-1'", 5)
580+
self.checkPatternError(r'()(?(㊀)a|b)',
581+
"bad character in group name '㊀'", 5)
582+
self.checkPatternError(r'()(?(¹)a|b)',
583+
"bad character in group name '¹'", 5)
584+
self.checkPatternError(r'()(?(1',
585+
"missing ), unterminated name", 5)
586+
self.checkPatternError(r'()(?(1)a',
587+
"missing ), unterminated subpattern", 2)
565588
self.checkPatternError(r'()(?(1)a|b',
566589
'missing ), unterminated subpattern', 2)
590+
self.checkPatternError(r'()(?(1)a|b|c',
591+
'conditional backref with more than '
592+
'two branches', 10)
567593
self.checkPatternError(r'()(?(1)a|b|c)',
568594
'conditional backref with more than '
569595
'two branches', 10)

0 commit comments

Comments
 (0)
0