Update bundled PCRE2-library to version 10.23
Some manual changes done to the library were lost with this update. They will be added in the next commit.
This commit is contained in:
Esa Korhonen
155
pcre2/testdata/testinput10
vendored
155
pcre2/testdata/testinput10
vendored
@ -1,45 +1,7 @@
|
||||
# This set of tests is for UTF-8 support and Unicode property support, with
|
||||
# relevance only for the 8-bit library.
|
||||
|
||||
/X(\C{3})/utf
|
||||
X\x{1234}
|
||||
|
||||
/X(\C{4})/utf
|
||||
X\x{1234}YZ
|
||||
|
||||
/X\C*/utf
|
||||
XYZabcdce
|
||||
|
||||
/X\C*?/utf
|
||||
XYZabcde
|
||||
|
||||
/X\C{3,5}/utf
|
||||
Xabcdefg
|
||||
X\x{1234}
|
||||
X\x{1234}YZ
|
||||
X\x{1234}\x{512}
|
||||
X\x{1234}\x{512}YZ
|
||||
|
||||
/X\C{3,5}?/utf
|
||||
Xabcdefg
|
||||
X\x{1234}
|
||||
X\x{1234}YZ
|
||||
X\x{1234}\x{512}
|
||||
|
||||
/a\Cb/utf
|
||||
aXb
|
||||
a\nb
|
||||
|
||||
/a\C\Cb/utf
|
||||
a\x{100}b
|
||||
|
||||
/ab\Cde/utf
|
||||
abXde
|
||||
|
||||
/a\C\Cb/utf
|
||||
a\x{100}b
|
||||
** Failers
|
||||
a\x{12257}b
|
||||
# The next 4 patterns have UTF-8 errors
|
||||
|
||||
/[�]/utf
|
||||
|
||||
@ -47,7 +9,12 @@
|
||||
|
||||
/���xxx/utf
|
||||
|
||||
/��������/utf
|
||||
|
||||
# Now test subjects
|
||||
|
||||
/badutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
X\xdf
|
||||
XX\xef
|
||||
XXX\xef\x80
|
||||
@ -89,11 +56,13 @@
|
||||
\xff
|
||||
|
||||
/badutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
XX\xfb\x80\x80\x80\x80
|
||||
XX\xfd\x80\x80\x80\x80\x80
|
||||
XX\xf7\xbf\xbf\xbf
|
||||
|
||||
/shortutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
XX\xdf\=ph
|
||||
XX\xef\=ph
|
||||
XX\xef\x80\=ph
|
||||
@ -111,6 +80,7 @@
|
||||
\xfd\x80\x80\x80\x80\=ph
|
||||
|
||||
/anything/utf
|
||||
\= Expect UTF-8 errors
|
||||
X\xc0\x80
|
||||
XX\xc1\x8f
|
||||
XXX\xe0\x9f\x80
|
||||
@ -119,20 +89,57 @@
|
||||
\xfc\x83\x80\x80\x80\x80
|
||||
\xfe\x80\x80\x80\x80\x80
|
||||
\xff\x80\x80\x80\x80\x80
|
||||
\xf8\x88\x80\x80\x80
|
||||
\xf9\x87\x80\x80\x80
|
||||
\xfc\x84\x80\x80\x80\x80
|
||||
\xfd\x83\x80\x80\x80\x80
|
||||
\= Expect no match
|
||||
\xc3\x8f
|
||||
\xe0\xaf\x80
|
||||
\xe1\x80\x80
|
||||
\xf0\x9f\x80\x80
|
||||
\xf1\x8f\x80\x80
|
||||
\xf8\x88\x80\x80\x80
|
||||
\xf9\x87\x80\x80\x80
|
||||
\xfc\x84\x80\x80\x80\x80
|
||||
\xfd\x83\x80\x80\x80\x80
|
||||
\xf8\x88\x80\x80\x80\=no_utf_check
|
||||
\xf9\x87\x80\x80\x80\=no_utf_check
|
||||
\xfc\x84\x80\x80\x80\x80\=no_utf_check
|
||||
\xfd\x83\x80\x80\x80\x80\=no_utf_check
|
||||
|
||||
# Similar tests with offsets
|
||||
|
||||
/badutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
X\xdfabcd
|
||||
X\xdfabcd\=offset=1
|
||||
\= Expect no match
|
||||
X\xdfabcd\=offset=2
|
||||
|
||||
/(?<=x)badutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
X\xdfabcd
|
||||
X\xdfabcd\=offset=1
|
||||
X\xdfabcd\=offset=2
|
||||
X\xdfabcd\xdf\=offset=3
|
||||
\= Expect no match
|
||||
X\xdfabcd\=offset=3
|
||||
|
||||
/(?<=xx)badutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
X\xdfabcd
|
||||
X\xdfabcd\=offset=1
|
||||
X\xdfabcd\=offset=2
|
||||
X\xdfabcd\=offset=3
|
||||
|
||||
/(?<=xxxx)badutf/utf
|
||||
\= Expect UTF-8 errors
|
||||
X\xdfabcd
|
||||
X\xdfabcd\=offset=1
|
||||
X\xdfabcd\=offset=2
|
||||
X\xdfabcd\=offset=3
|
||||
X\xdfabc\xdf\=offset=6
|
||||
X\xdfabc\xdf\=offset=7
|
||||
\= Expect no match
|
||||
X\xdfabcd\=offset=6
|
||||
|
||||
/\x{100}/IB,utf
|
||||
|
||||
/\x{1000}/IB,utf
|
||||
@ -167,27 +174,12 @@
|
||||
|
||||
/\x{212ab}/IB,utf
|
||||
|
||||
# This one is here not because it's different to Perl, but because the way
|
||||
# the captured single-byte is displayed. (In Perl it becomes a character, and you
|
||||
# can't tell the difference.)
|
||||
|
||||
/X(\C)(.*)/utf
|
||||
X\x{1234}
|
||||
X\nabc
|
||||
|
||||
# This one is here because Perl gives out a grumbly error message (quite
|
||||
# correctly, but that messes up comparisons).
|
||||
|
||||
/a\Cb/utf
|
||||
*** Failers
|
||||
a\x{100}b
|
||||
|
||||
/[^ab\xC0-\xF0]/IB,utf
|
||||
\x{f1}
|
||||
\x{bf}
|
||||
\x{100}
|
||||
\x{1000}
|
||||
*** Failers
|
||||
\= Expect no match
|
||||
\x{c0}
|
||||
\x{f0}
|
||||
|
||||
@ -214,7 +206,6 @@
|
||||
\x{100}
|
||||
Z\x{100}
|
||||
\x{100}Z
|
||||
*** Failers
|
||||
|
||||
/[\xff]/IB,utf
|
||||
>\x{ff}<
|
||||
@ -236,21 +227,23 @@
|
||||
# This tests the stricter UTF-8 check according to RFC 3629.
|
||||
|
||||
/X/utf
|
||||
\= Expect UTF-8 errors
|
||||
\x{d800}
|
||||
\x{d800}\=no_utf_check
|
||||
\x{da00}
|
||||
\x{da00}\=no_utf_check
|
||||
\x{dfff}
|
||||
\x{dfff}\=no_utf_check
|
||||
\x{110000}
|
||||
\x{110000}\=no_utf_check
|
||||
\x{2000000}
|
||||
\x{2000000}\=no_utf_check
|
||||
\x{7fffffff}
|
||||
\= Expect no match
|
||||
\x{d800}\=no_utf_check
|
||||
\x{da00}\=no_utf_check
|
||||
\x{dfff}\=no_utf_check
|
||||
\x{110000}\=no_utf_check
|
||||
\x{2000000}\=no_utf_check
|
||||
\x{7fffffff}\=no_utf_check
|
||||
|
||||
/(*UTF8)\x{1234}/
|
||||
abcd\x{1234}pqr
|
||||
abcd\x{1234}pqr
|
||||
|
||||
/(*CRLF)(*UTF)(*BSR_UNICODE)a\Rb/I
|
||||
|
||||
@ -290,11 +283,14 @@
|
||||
|
||||
/a+/utf
|
||||
a\x{123}aa\=offset=1
|
||||
a\x{123}aa\=offset=2
|
||||
a\x{123}aa\=offset=3
|
||||
a\x{123}aa\=offset=4
|
||||
a\x{123}aa\=offset=5
|
||||
\= Expect bad offset value
|
||||
a\x{123}aa\=offset=6
|
||||
\= Expect bad UTF-8 offset
|
||||
a\x{123}aa\=offset=2
|
||||
\= Expect no match
|
||||
a\x{123}aa\=offset=5
|
||||
|
||||
/\x{1234}+/Ii,utf
|
||||
|
||||
@ -395,7 +391,6 @@
|
||||
Z\x{100}
|
||||
\x{100}
|
||||
\x{100}Z
|
||||
*** Failers
|
||||
|
||||
/[z-\x{100}]/IB,utf
|
||||
|
||||
@ -421,7 +416,7 @@
|
||||
\x{104}
|
||||
\x{105}
|
||||
\x{109}
|
||||
** Failers
|
||||
\= Expect no match
|
||||
\x{100}
|
||||
\x{10a}
|
||||
|
||||
@ -435,7 +430,7 @@
|
||||
\x{ff}
|
||||
\x{100}
|
||||
\x{101}
|
||||
** Failers
|
||||
\= Expect no match
|
||||
\x{102}
|
||||
Y
|
||||
y
|
||||
@ -445,6 +440,22 @@
|
||||
/\x{3a3}B/IBi,utf
|
||||
|
||||
/abc/utf,replace=�
|
||||
abc
|
||||
abc
|
||||
|
||||
/(?<=(a)(?-1))x/I,utf
|
||||
a\x80zx\=offset=3
|
||||
|
||||
/[\W\p{Any}]/B
|
||||
abc
|
||||
123
|
||||
|
||||
/[\W\pL]/B
|
||||
abc
|
||||
\= Expect no match
|
||||
123
|
||||
|
||||
/(*:*++++++++++++''''''''''''''''''''+''+++'+++x+++++++++++++++++++++++++++++++++++(++++++++++++++++++++:++++++%++:''''''''''''''''''''''''+++++++++++++++++++++++++++++++++++++++++++++++++++++-++++++++k+++++++''''+++'+++++++++++++++++++++++''''++++++++++++':ƿ)/utf
|
||||
|
||||
/[\s[:^ascii:]]/B,ucp
|
||||
|
||||
# End of testinput10
|
||||
|
||||
Reference in New Issue
Block a user