Test that UTF-8 in the look-ahead of (?(?=...)...) restarts the sizing parse.
authorNicholas Clark <nick@ccl4.org>
Wed, 16 Jan 2013 20:58:02 +0000 (21:58 +0100)
committerNicholas Clark <nick@ccl4.org>
Tue, 19 Mar 2013 10:53:19 +0000 (11:53 +0100)
S_reg() recurses to itself to parse various constructions used as the
conditionals in conditional matching. Look-aheads and look-behinds can turn
out to need to be sized as UTF-8, which can cause the inner S_reg() to use
the macro REQUIRE_UTF8 is used to restart the parse. Test that this is
handled correctly.

t/re/re_tests

index c41d529..7e7fc85 100644 (file)
@@ -592,6 +592,10 @@ x(~~)*(?:(?:F)?)?  x~~     y       -       -
 (?(?!a)b|a)    a       y       $&      a
 (?(?=a)b|a)    a       n       -       -
 (?(?=a)a|b)    a       y       $&      a
+(?(?!\x{100})\x{100}|b)        \x{100} n       -       -
+(?(?!\x{100})b|\x{100})        \x{100} y       $&      \x{100}
+(?(?=\x{100})b|\x{100})        \x{100} n       -       -
+(?(?=\x{100})\x{100}|b)        \x{100} y       $&      \x{100}
 (?=(a+?))(\1ab)        aaab    y       $2      aab
 ^(?=(a+?))\1ab aaab    n       -       -
 (\w+:)+        one:    y       $1      one: