|
 |
04bfb0 |
From 0c311b7c345769239f38d0139ea7738feec5ca4d Mon Sep 17 00:00:00 2001
|
|
|
04bfb0 |
From: Karl Williamson <khw@cpan.org>
|
|
|
04bfb0 |
Date: Sat, 2 Nov 2019 13:59:38 -0600
|
|
|
04bfb0 |
Subject: [PATCH] toke.c: Fix bug tr/// upgrading to UTF-8 in middle
|
|
|
04bfb0 |
MIME-Version: 1.0
|
|
|
04bfb0 |
Content-Type: text/plain; charset=UTF-8
|
|
|
04bfb0 |
Content-Transfer-Encoding: 8bit
|
|
|
04bfb0 |
|
|
|
04bfb0 |
Consider tr/\x{ff}-\x{100}/AB/.
|
|
|
04bfb0 |
|
|
|
04bfb0 |
While parsing, the code keeps an offset from the beginning of the output
|
|
|
04bfb0 |
to the beginning of the second number in the range. This is purely for
|
|
|
04bfb0 |
speed so that it wouldn't have to re-find the beginning of that value,
|
|
|
04bfb0 |
when it already knew it.
|
|
|
04bfb0 |
|
|
|
04bfb0 |
But the example above shows the folly of this shortcut. The second
|
|
|
04bfb0 |
number in the range causes the output to be upgraded to UTF-8, which
|
|
|
04bfb0 |
makes that offset invalid in general. Change to re-find the beginning.
|
|
|
04bfb0 |
|
|
|
04bfb0 |
Signed-off-by: Petr Písař <ppisar@redhat.com>
|
|
|
04bfb0 |
---
|
|
|
04bfb0 |
t/op/tr.t | 12 +++++++++++-
|
|
|
04bfb0 |
toke.c | 4 +++-
|
|
|
04bfb0 |
2 files changed, 14 insertions(+), 2 deletions(-)
|
|
|
04bfb0 |
|
|
|
04bfb0 |
diff --git a/t/op/tr.t b/t/op/tr.t
|
|
|
04bfb0 |
index 47d603d4fd..25125c5bc7 100644
|
|
|
04bfb0 |
--- a/t/op/tr.t
|
|
|
04bfb0 |
+++ b/t/op/tr.t
|
|
|
04bfb0 |
@@ -13,7 +13,7 @@ BEGIN {
|
|
|
04bfb0 |
|
|
|
04bfb0 |
use utf8;
|
|
|
04bfb0 |
|
|
|
04bfb0 |
-plan tests => 301;
|
|
|
04bfb0 |
+plan tests => 304;
|
|
|
04bfb0 |
|
|
|
04bfb0 |
# Test this first before we extend the stack with other operations.
|
|
|
04bfb0 |
# This caused an asan failure due to a bad write past the end of the stack.
|
|
|
04bfb0 |
@@ -1145,4 +1145,14 @@ for ("", nullrocow) {
|
|
|
04bfb0 |
'RT #133880 illegal \N{}');
|
|
|
04bfb0 |
}
|
|
|
04bfb0 |
|
|
|
04bfb0 |
+{
|
|
|
04bfb0 |
+ my $c = "\xff";
|
|
|
04bfb0 |
+ my $d = "\x{104}";
|
|
|
04bfb0 |
+ eval '$c =~ tr/\x{ff}-\x{104}/\x{100}-\x{105}/';
|
|
|
04bfb0 |
+ is($@, "", 'tr/\x{ff}-\x{104}/\x{100}-\x{105}/ compiled');
|
|
|
04bfb0 |
+ is($c, "\x{100}", 'ff -> 100');
|
|
|
04bfb0 |
+ eval '$d =~ tr/\x{ff}-\x{104}/\x{100}-\x{105}/';
|
|
|
04bfb0 |
+ is($d, "\x{105}", '104 -> 105');
|
|
|
04bfb0 |
+}
|
|
|
04bfb0 |
+
|
|
|
04bfb0 |
1;
|
|
|
04bfb0 |
diff --git a/toke.c b/toke.c
|
|
|
04bfb0 |
index 2995737af2..28f305c62c 100644
|
|
|
04bfb0 |
--- a/toke.c
|
|
|
04bfb0 |
+++ b/toke.c
|
|
|
04bfb0 |
@@ -3044,7 +3044,7 @@ S_scan_const(pTHX_ char *start)
|
|
|
04bfb0 |
* 'offset_to_max' is the offset in 'sv' at which the character
|
|
|
04bfb0 |
* (the range's maximum end point) before 'd' begins.
|
|
|
04bfb0 |
*/
|
|
|
04bfb0 |
- char * max_ptr = SvPVX(sv) + offset_to_max;
|
|
|
04bfb0 |
+ char * max_ptr;
|
|
|
04bfb0 |
char * min_ptr;
|
|
|
04bfb0 |
IV range_min;
|
|
|
04bfb0 |
IV range_max; /* last character in range */
|
|
|
04bfb0 |
@@ -3056,6 +3056,8 @@ S_scan_const(pTHX_ char *start)
|
|
|
04bfb0 |
IV real_range_max = 0;
|
|
|
04bfb0 |
#endif
|
|
|
04bfb0 |
/* Get the code point values of the range ends. */
|
|
|
04bfb0 |
+ max_ptr = (d_is_utf8) ? (char *) utf8_hop( (U8*) d, -1) : d - 1;
|
|
|
04bfb0 |
+ offset_to_max = max_ptr - SvPVX_const(sv);
|
|
|
04bfb0 |
if (d_is_utf8) {
|
|
|
04bfb0 |
/* We know the utf8 is valid, because we just constructed
|
|
|
04bfb0 |
* it ourselves in previous loop iterations */
|
|
|
04bfb0 |
--
|
|
|
04bfb0 |
2.21.0
|
|
|
04bfb0 |
|