gh-135462: Fix quadratic complexity in processing special input in HT… · python/cpython@6eb6c5d · GitHub
Skip to content

Commit 6eb6c5d

Browse files
serhiy-storchakaserhiy-storchaka
authored
gh-135462: Fix quadratic complexity in processing special input in HTMLParser (GH-135464)
End-of-file errors are now handled according to the HTML5 specs -- comments and declarations are automatically closed, tags are ignored.
1 parent 14c1d09 commit 6eb6c5d

3 files changed

Lines changed: 111 additions & 31 deletions

File tree

‎Lib/html/parser.py‎

Lines changed: 30 additions & 11 deletions

‎Lib/test/test_htmlparser.py‎

Lines changed: 77 additions & 20 deletions
Original file line numberDiff line numberDiff line change
@@ -5,6 +5,7 @@
55
import unittest
66

77
from unittest.mock import patch
8+
from test import support
89

910

1011
class EventCollector(html.parser.HTMLParser):
@@ -430,28 +431,34 @@ def test_tolerant_parsing(self):
430431
('data', '<'),
431432
('starttag', 'bc<', [('a', None)]),
432433
('endtag', 'html'),
433-
('data', '\n<img src="URL>'),
434-
('comment', '/img'),
435-
('endtag', 'html<')])
434+
('data', '\n')])
436435

437436
def test_starttag_junk_chars(self):
437+
self._run_check("<", [('data', '<')])
438+
self._run_check("<>", [('data', '<>')])
439+
self._run_check("< >", [('data', '< >')])
440+
self._run_check("< ", [('data', '< ')])
438441
self._run_check("</>", [])
442+
self._run_check("<$>", [('data', '<$>')])
439443
self._run_check("</$>", [('comment', '$')])
440444
self._run_check("</", [('data', '</')])
441-
self._run_check("</a", [('data', '</a')])
445+
self._run_check("</a", [])
446+
self._run_check("</ a>", [('endtag', 'a')])
447+
self._run_check("</ a", [('comment', ' a')])
442448
self._run_check("<a<a>", [('starttag', 'a<a', [])])
443449
self._run_check("</a<a>", [('endtag', 'a<a')])
444-
self._run_check("<!", [('data', '<!')])
445-
self._run_check("<a", [('data', '<a')])
446-
self._run_check("<a foo='bar'", [('data', "<a foo='bar'")])
447-
self._run_check("<a foo='bar", [('data', "<a foo='bar")])
448-
self._run_check("<a foo='>'", [('data', "<a foo='>'")])
449-
self._run_check("<a foo='>", [('data', "<a foo='>")])
450+
self._run_check("<!", [('comment', '')])
451+
self._run_check("<a", [])
452+
self._run_check("<a foo='bar'", [])
453+
self._run_check("<a foo='bar", [])
454+
self._run_check("<a foo='>'", [])
455+
self._run_check("<a foo='>", [])
450456
self._run_check("<a$>", [('starttag', 'a$', [])])
451457
self._run_check("<a$b>", [('starttag', 'a$b', [])])
452458
self._run_check("<a$b/>", [('startendtag', 'a$b', [])])
453459
self._run_check("<a$b >", [('starttag', 'a$b', [])])
454460
self._run_check("<a$b />", [('startendtag', 'a$b', [])])
461+
self._run_check("</a$b>", [('endtag', 'a$b')])
455462

456463
def test_slashes_in_starttag(self):
457464
self._run_check('<a foo="var"/>', [('startendtag', 'a', [('foo', 'var')])])
@@ -576,21 +583,50 @@ def test_EOF_in_charref(self):
576583
for html, expected in data:
577584
self._run_check(html, expected)
578585

579-
def test_EOF_in_comments_or_decls(self):
586+
def test_eof_in_comments(self):
580587
data = [
581-
('<!', [('data', '<!')]),
582-
('<!-', [('data', '<!-')]),
583-
('<!--', [('data', '<!--')]),
584-
('<![', [('data', '<![')]),
585-
('<![CDATA[', [('data', '<![CDATA[')]),
586-
('<![CDATA[x', [('data', '<![CDATA[x')]),
587-
('<!DOCTYPE', [('data', '<!DOCTYPE')]),
588-
('<!DOCTYPE HTML', [('data', '<!DOCTYPE HTML')]),
588+
('<!--', [('comment', '')]),
589+
('<!---', [('comment', '')]),
590+
('<!----', [('comment', '')]),
591+
('<!-----', [('comment', '-')]),
592+
('<!------', [('comment', '--')]),
593+
('<!----!', [('comment', '')]),
594+
('<!---!', [('comment', '-!')]),
595+
('<!---!>', [('comment', '-!>')]),
596+
('<!--foo', [('comment', 'foo')]),
597+
('<!--foo-', [('comment', 'foo')]),
598+
('<!--foo--', [('comment', 'foo')]),
599+
('<!--foo--!', [('comment', 'foo')]),
600+
('<!--<!--', [('comment', '<!')]),
601+
('<!--<!--!', [('comment', '<!')]),
589602
]
590603
for html, expected in data:
591604
self._run_check(html, expected)
605+
606+
def test_eof_in_declarations(self):
607+
data = [
608+
('<!', [('comment', '')]),
609+
('<!-', [('comment', '-')]),
610+
('<![', [('comment', '[')]),
611+
('<![CDATA[', [('unknown decl', 'CDATA[')]),
612+
('<![CDATA[x', [('unknown decl', 'CDATA[x')]),
613+
('<![CDATA[x]', [('unknown decl', 'CDATA[x]')]),
614+
('<![CDATA[x]]', [('unknown decl', 'CDATA[x]]')]),
615+
('<!DOCTYPE', [('decl', 'DOCTYPE')]),
616+
('<!DOCTYPE ', [('decl', 'DOCTYPE ')]),
617+
('<!DOCTYPE html', [('decl', 'DOCTYPE html')]),
618+
('<!DOCTYPE html ', [('decl', 'DOCTYPE html ')]),
619+
('<!DOCTYPE html PUBLIC', [('decl', 'DOCTYPE html PUBLIC')]),
620+
('<!DOCTYPE html PUBLIC "foo', [('decl', 'DOCTYPE html PUBLIC "foo')]),
621+
('<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN" "foo',
622+
[('decl', 'DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN" "foo')]),
623+
]
624+
for html, expected in data:
625+
self._run_check(html, expected)
626+
592627
def test_bogus_comments(self):
593-
html = ('<! not really a comment >'
628+
html = ('<!ELEMENT br EMPTY>'
629+
'<! not really a comment >'
594630
'<! not a comment either -->'
595631
'<! -- close enough -->'
596632
'<!><!<-- this was an empty comment>'
@@ -604,6 +640,7 @@ def test_bogus_comments(self):
604640
'<![CDATA]]>' # required '[' after CDATA
605641
)
606642
expected = [
643+
('comment', 'ELEMENT br EMPTY'),
607644
('comment', ' not really a comment '),
608645
('comment', ' not a comment either --'),
609646
('comment', ' -- close enough --'),
@@ -684,6 +721,26 @@ def test_convert_charrefs_dropped_text(self):
684721
('endtag', 'a'), ('data', ' bar & baz')]
685722
)
686723

724+
@support.requires_resource('cpu')
725+
def test_eof_no_quadratic_complexity(self):
726+
# Each of these examples used to take about an hour.
727+
# Now they take a fraction of a second.
728+
def check(source):
729+
parser = html.parser.HTMLParser()
730+
parser.feed(source)
731+
parser.close()
732+
n = 120_000
733+
check("<a " * n)
734+
check("<a a=" * n)
735+
check("</a " * 14 * n)
736+
check("</a a=" * 11 * n)
737+
check("<!--" * 4 * n)
738+
check("<!" * 60 * n)
739+
check("<?" * 19 * n)
740+
check("</$" * 15 * n)
741+
check("<![CDATA[" * 9 * n)
742+
check("<!doctype" * 35 * n)
743+
687744

688745
class AttributesTestCase(TestCaseBase):
689746

‎Misc/NEWS.d/next/Security/2025-06-13-15-55-22.gh-issue-135462.KBeJpc.rst‎

Lines changed: 4 additions & 0 deletions

0 commit comments

Comments
 (0)