2609 lines
52 KiB
Plaintext
2609 lines
52 KiB
Plaintext
# This set of tests is for UTF support, including Unicode properties. The
|
||
# Unicode tests are all compatible with all versions of Perl >= 5.10, but
|
||
# some of the property tests may differ because of different versions of
|
||
# Unicode in use by PCRE2 and Perl.
|
||
|
||
# WARNING: Use only / as the pattern delimiter. Although pcre2test supports
|
||
# a number of delimiters, all those other than / give problems with the
|
||
# perltest.sh script.
|
||
|
||
#newline_default lf anycrlf any
|
||
#perltest
|
||
|
||
/a.b/utf
|
||
acb
|
||
a\x7fb
|
||
a\x{100}b
|
||
\= Expect no match
|
||
a\nb
|
||
|
||
/a(.{3})b/utf
|
||
a\x{4000}xyb
|
||
a\x{4000}\x7fyb
|
||
a\x{4000}\x{100}yb
|
||
\= Expect no match
|
||
a\x{4000}b
|
||
ac\ncb
|
||
|
||
/a(.*?)(.)/
|
||
a\xc0\x88b
|
||
|
||
/a(.*?)(.)/utf
|
||
a\x{100}b
|
||
|
||
/a(.*)(.)/
|
||
a\xc0\x88b
|
||
|
||
/a(.*)(.)/utf
|
||
a\x{100}b
|
||
|
||
/a(.)(.)/
|
||
a\xc0\x92bcd
|
||
|
||
/a(.)(.)/utf
|
||
a\x{240}bcd
|
||
|
||
/a(.?)(.)/
|
||
a\xc0\x92bcd
|
||
|
||
/a(.?)(.)/utf
|
||
a\x{240}bcd
|
||
|
||
/a(.??)(.)/
|
||
a\xc0\x92bcd
|
||
|
||
/a(.??)(.)/utf
|
||
a\x{240}bcd
|
||
|
||
/a(.{3})b/utf
|
||
a\x{1234}xyb
|
||
a\x{1234}\x{4321}yb
|
||
a\x{1234}\x{4321}\x{3412}b
|
||
\= Expect no match
|
||
a\x{1234}b
|
||
ac\ncb
|
||
|
||
/a(.{3,})b/utf
|
||
a\x{1234}xyb
|
||
a\x{1234}\x{4321}yb
|
||
a\x{1234}\x{4321}\x{3412}b
|
||
axxxxbcdefghijb
|
||
a\x{1234}\x{4321}\x{3412}\x{3421}b
|
||
\= Expect no match
|
||
a\x{1234}b
|
||
|
||
/a(.{3,}?)b/utf
|
||
a\x{1234}xyb
|
||
a\x{1234}\x{4321}yb
|
||
a\x{1234}\x{4321}\x{3412}b
|
||
axxxxbcdefghijb
|
||
a\x{1234}\x{4321}\x{3412}\x{3421}b
|
||
\= Expect no match
|
||
a\x{1234}b
|
||
|
||
/a(.{3,5})b/utf
|
||
a\x{1234}xyb
|
||
a\x{1234}\x{4321}yb
|
||
a\x{1234}\x{4321}\x{3412}b
|
||
axxxxbcdefghijb
|
||
a\x{1234}\x{4321}\x{3412}\x{3421}b
|
||
axbxxbcdefghijb
|
||
axxxxxbcdefghijb
|
||
\= Expect no match
|
||
a\x{1234}b
|
||
axxxxxxbcdefghijb
|
||
|
||
/a(.{3,5}?)b/utf
|
||
a\x{1234}xyb
|
||
a\x{1234}\x{4321}yb
|
||
a\x{1234}\x{4321}\x{3412}b
|
||
axxxxbcdefghijb
|
||
a\x{1234}\x{4321}\x{3412}\x{3421}b
|
||
axbxxbcdefghijb
|
||
axxxxxbcdefghijb
|
||
\= Expect no match
|
||
a\x{1234}b
|
||
axxxxxxbcdefghijb
|
||
|
||
/^[a\x{c0}]/utf
|
||
\= Expect no match
|
||
\x{100}
|
||
|
||
/(?<=aXb)cd/utf
|
||
aXbcd
|
||
|
||
/(?<=a\x{100}b)cd/utf
|
||
a\x{100}bcd
|
||
|
||
/(?<=a\x{100000}b)cd/utf
|
||
a\x{100000}bcd
|
||
|
||
/(?:\x{100}){3}b/utf
|
||
\x{100}\x{100}\x{100}b
|
||
\= Expect no match
|
||
\x{100}\x{100}b
|
||
|
||
/\x{ab}/utf
|
||
\x{ab}
|
||
\xc2\xab
|
||
\= Expect no match
|
||
\x00{ab}
|
||
|
||
/(?<=(.))X/utf
|
||
WXYZ
|
||
\x{256}XYZ
|
||
\= Expect no match
|
||
XYZ
|
||
|
||
/[^a]+/g,utf
|
||
bcd
|
||
\x{100}aY\x{256}Z
|
||
|
||
/^[^a]{2}/utf
|
||
\x{100}bc
|
||
|
||
/^[^a]{2,}/utf
|
||
\x{100}bcAa
|
||
|
||
/^[^a]{2,}?/utf
|
||
\x{100}bca
|
||
|
||
/[^a]+/gi,utf
|
||
bcd
|
||
\x{100}aY\x{256}Z
|
||
|
||
/^[^a]{2}/i,utf
|
||
\x{100}bc
|
||
|
||
/^[^a]{2,}/i,utf
|
||
\x{100}bcAa
|
||
|
||
/^[^a]{2,}?/i,utf
|
||
\x{100}bca
|
||
|
||
/\x{100}{0,0}/utf
|
||
abcd
|
||
|
||
/\x{100}?/utf
|
||
abcd
|
||
\x{100}\x{100}
|
||
|
||
/\x{100}{0,3}/utf
|
||
\x{100}\x{100}
|
||
\x{100}\x{100}\x{100}\x{100}
|
||
|
||
/\x{100}*/utf
|
||
abce
|
||
\x{100}\x{100}\x{100}\x{100}
|
||
|
||
/\x{100}{1,1}/utf
|
||
abcd\x{100}\x{100}\x{100}\x{100}
|
||
|
||
/\x{100}{1,3}/utf
|
||
abcd\x{100}\x{100}\x{100}\x{100}
|
||
|
||
/\x{100}+/utf
|
||
abcd\x{100}\x{100}\x{100}\x{100}
|
||
|
||
/\x{100}{3}/utf
|
||
abcd\x{100}\x{100}\x{100}XX
|
||
|
||
/\x{100}{3,5}/utf
|
||
abcd\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}XX
|
||
|
||
/\x{100}{3,}/utf
|
||
abcd\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}XX
|
||
|
||
/(?<=a\x{100}{2}b)X/utf,aftertext
|
||
Xyyya\x{100}\x{100}bXzzz
|
||
|
||
/\D*/utf
|
||
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
|
||
|
||
/\D*/utf
|
||
\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}\x{100}
|
||
|
||
/\D/utf
|
||
1X2
|
||
1\x{100}2
|
||
|
||
/>\S/utf
|
||
> >X Y
|
||
> >\x{100} Y
|
||
|
||
/\d/utf
|
||
\x{100}3
|
||
|
||
/\s/utf
|
||
\x{100} X
|
||
|
||
/\D+/utf
|
||
12abcd34
|
||
\= Expect no match
|
||
1234
|
||
|
||
/\D{2,3}/utf
|
||
12abcd34
|
||
12ab34
|
||
\= Expect no match
|
||
1234
|
||
12a34
|
||
|
||
/\D{2,3}?/utf
|
||
12abcd34
|
||
12ab34
|
||
\= Expect no match
|
||
1234
|
||
12a34
|
||
|
||
/\d+/utf
|
||
12abcd34
|
||
|
||
/\d{2,3}/utf
|
||
12abcd34
|
||
1234abcd
|
||
\= Expect no match
|
||
1.4
|
||
|
||
/\d{2,3}?/utf
|
||
12abcd34
|
||
1234abcd
|
||
\= Expect no match
|
||
1.4
|
||
|
||
/\S+/utf
|
||
12abcd34
|
||
\= Expect no match
|
||
\ \
|
||
|
||
/\S{2,3}/utf
|
||
12abcd34
|
||
1234abcd
|
||
\= Expect no match
|
||
\ \
|
||
|
||
/\S{2,3}?/utf
|
||
12abcd34
|
||
1234abcd
|
||
\= Expect no match
|
||
\ \
|
||
|
||
/>\s+</utf,aftertext
|
||
12> <34
|
||
|
||
/>\s{2,3}</utf,aftertext
|
||
ab> <cd
|
||
ab> <ce
|
||
\= Expect no match
|
||
ab> <cd
|
||
|
||
/>\s{2,3}?</utf,aftertext
|
||
ab> <cd
|
||
ab> <ce
|
||
\= Expect no match
|
||
ab> <cd
|
||
|
||
/\w+/utf
|
||
12 34
|
||
\= Expect no match
|
||
+++=*!
|
||
|
||
/\w{2,3}/utf
|
||
ab cd
|
||
abcd ce
|
||
\= Expect no match
|
||
a.b.c
|
||
|
||
/\w{2,3}?/utf
|
||
ab cd
|
||
abcd ce
|
||
\= Expect no match
|
||
a.b.c
|
||
|
||
/\W+/utf
|
||
12====34
|
||
\= Expect no match
|
||
abcd
|
||
|
||
/\W{2,3}/utf
|
||
ab====cd
|
||
ab==cd
|
||
\= Expect no match
|
||
a.b.c
|
||
|
||
/\W{2,3}?/utf
|
||
ab====cd
|
||
ab==cd
|
||
\= Expect no match
|
||
a.b.c
|
||
|
||
/[\x{100}]/utf
|
||
\x{100}
|
||
Z\x{100}
|
||
\x{100}Z
|
||
|
||
/[Z\x{100}]/utf
|
||
Z\x{100}
|
||
\x{100}
|
||
\x{100}Z
|
||
|
||
/[\x{100}\x{200}]/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
|
||
/[\x{100}-\x{200}]/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{111}cd
|
||
|
||
/[z-\x{200}]/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{111}cd
|
||
abzcd
|
||
ab|cd
|
||
|
||
/[Q\x{100}\x{200}]/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
Q?
|
||
|
||
/[Q\x{100}-\x{200}]/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{111}cd
|
||
Q?
|
||
|
||
/[Qz-\x{200}]/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{111}cd
|
||
abzcd
|
||
ab|cd
|
||
Q?
|
||
|
||
/[\x{100}\x{200}]{1,3}/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{200}\x{100}\x{200}\x{100}cd
|
||
|
||
/[\x{100}\x{200}]{1,3}?/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{200}\x{100}\x{200}\x{100}cd
|
||
|
||
/[Q\x{100}\x{200}]{1,3}/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{200}\x{100}\x{200}\x{100}cd
|
||
|
||
/[Q\x{100}\x{200}]{1,3}?/utf
|
||
ab\x{100}cd
|
||
ab\x{200}cd
|
||
ab\x{200}\x{100}\x{200}\x{100}cd
|
||
|
||
/(?<=[\x{100}\x{200}])X/utf
|
||
abc\x{200}X
|
||
abc\x{100}X
|
||
\= Expect no match
|
||
X
|
||
|
||
/(?<=[Q\x{100}\x{200}])X/utf
|
||
abc\x{200}X
|
||
abc\x{100}X
|
||
abQX
|
||
\= Expect no match
|
||
X
|
||
|
||
/(?<=[\x{100}\x{200}]{3})X/utf
|
||
abc\x{100}\x{200}\x{100}X
|
||
\= Expect no match
|
||
abc\x{200}X
|
||
X
|
||
|
||
/[^\x{100}\x{200}]X/utf
|
||
AX
|
||
\x{150}X
|
||
\x{500}X
|
||
\= Expect no match
|
||
\x{100}X
|
||
\x{200}X
|
||
|
||
/[^Q\x{100}\x{200}]X/utf
|
||
AX
|
||
\x{150}X
|
||
\x{500}X
|
||
\= Expect no match
|
||
\x{100}X
|
||
\x{200}X
|
||
QX
|
||
|
||
/[^\x{100}-\x{200}]X/utf
|
||
AX
|
||
\x{500}X
|
||
\= Expect no match
|
||
\x{100}X
|
||
\x{150}X
|
||
\x{200}X
|
||
|
||
/[z-\x{100}]/i,utf
|
||
z
|
||
Z
|
||
\x{100}
|
||
\= Expect no match
|
||
\x{102}
|
||
y
|
||
|
||
/[\xFF]/
|
||
>\xff<
|
||
|
||
/[\xff]/utf
|
||
>\x{ff}<
|
||
|
||
/[^\xFF]/
|
||
XYZ
|
||
|
||
/[^\xff]/utf
|
||
XYZ
|
||
\x{123}
|
||
|
||
/^[ac]*b/utf
|
||
\= Expect no match
|
||
xb
|
||
|
||
/^[ac\x{100}]*b/utf
|
||
\= Expect no match
|
||
xb
|
||
|
||
/^[^x]*b/i,utf
|
||
\= Expect no match
|
||
xb
|
||
|
||
/^[^x]*b/utf
|
||
\= Expect no match
|
||
xb
|
||
|
||
/^\d*b/utf
|
||
\= Expect no match
|
||
xb
|
||
|
||
/(|a)/g,utf
|
||
catac
|
||
a\x{256}a
|
||
|
||
/^\x{85}$/i,utf
|
||
\x{85}
|
||
|
||
/^ሴ/utf
|
||
ሴ
|
||
|
||
/^\ሴ/utf
|
||
ሴ
|
||
|
||
/(?s)(.{1,5})/utf
|
||
abcdefg
|
||
ab
|
||
|
||
/a*\x{100}*\w/utf
|
||
a
|
||
|
||
/\S\S/g,utf
|
||
A\x{a3}BC
|
||
|
||
/\S{2}/g,utf
|
||
A\x{a3}BC
|
||
|
||
/\W\W/g,utf
|
||
+\x{a3}==
|
||
|
||
/\W{2}/g,utf
|
||
+\x{a3}==
|
||
|
||
/\S/g,utf
|
||
\x{442}\x{435}\x{441}\x{442}
|
||
|
||
/[\S]/g,utf
|
||
\x{442}\x{435}\x{441}\x{442}
|
||
|
||
/\D/g,utf
|
||
\x{442}\x{435}\x{441}\x{442}
|
||
|
||
/[\D]/g,utf
|
||
\x{442}\x{435}\x{441}\x{442}
|
||
|
||
/\W/g,utf
|
||
\x{2442}\x{2435}\x{2441}\x{2442}
|
||
|
||
/[\W]/g,utf
|
||
\x{2442}\x{2435}\x{2441}\x{2442}
|
||
|
||
/[\S\s]*/utf
|
||
abc\n\r\x{442}\x{435}\x{441}\x{442}xyz
|
||
|
||
/[\x{41f}\S]/g,utf
|
||
\x{442}\x{435}\x{441}\x{442}
|
||
|
||
/.[^\S]./g,utf
|
||
abc def\x{442}\x{443}xyz\npqr
|
||
|
||
/.[^\S\n]./g,utf
|
||
abc def\x{442}\x{443}xyz\npqr
|
||
|
||
/[[:^alnum:]]/g,utf
|
||
+\x{2442}
|
||
|
||
/[[:^alpha:]]/g,utf
|
||
+\x{2442}
|
||
|
||
/[[:^ascii:]]/g,utf
|
||
A\x{442}
|
||
|
||
/[[:^blank:]]/g,utf
|
||
A\x{442}
|
||
|
||
/[[:^cntrl:]]/g,utf
|
||
A\x{442}
|
||
|
||
/[[:^digit:]]/g,utf
|
||
A\x{442}
|
||
|
||
/[[:^graph:]]/g,utf
|
||
\x19\x{e01ff}
|
||
|
||
/[[:^lower:]]/g,utf
|
||
A\x{422}
|
||
|
||
/[[:^print:]]/g,utf
|
||
\x{19}\x{e01ff}
|
||
|
||
/[[:^punct:]]/g,utf
|
||
A\x{442}
|
||
|
||
/[[:^space:]]/g,utf
|
||
A\x{442}
|
||
|
||
/[[:^upper:]]/g,utf
|
||
a\x{442}
|
||
|
||
/[[:^word:]]/g,utf
|
||
+\x{2442}
|
||
|
||
/[[:^xdigit:]]/g,utf
|
||
M\x{442}
|
||
|
||
/[^ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞĀĂĄĆĈĊČĎĐĒĔĖĘĚĜĞĠĢĤĦĨĪĬĮİIJĴĶĹĻĽĿŁŃŅŇŊŌŎŐŒŔŖŘŚŜŞŠŢŤŦŨŪŬŮŰŲŴŶŸŹŻŽƁƂƄƆƇƉƊƋƎƏƐƑƓƔƖƗƘƜƝƟƠƢƤƦƧƩƬƮƯƱƲƳƵƷƸƼDŽLJNJǍǏǑǓǕǗǙǛǞǠǢǤǦǨǪǬǮDZǴǶǷǸǺǼǾȀȂȄȆȈȊȌȎȐȒȔȖȘȚȜȞȠȢȤȦȨȪȬȮȰȲȺȻȽȾɁΆΈΉΊΌΎΏΑΒΓΔΕΖΗΘΙΚΛΜΝΞΟΠΡΣΤΥΦΧΨΩΪΫϒϓϔϘϚϜϞϠϢϤϦϨϪϬϮϴϷϹϺϽϾϿЀЁЂЃЄЅІЇЈЉЊЋЌЍЎЏАБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЪЫЬЭЮЯѠѢѤѦѨѪѬѮѰѲѴѶѸѺѼѾҀҊҌҎҐҒҔҖҘҚҜҞҠҢҤҦҨҪҬҮҰҲҴҶҸҺҼҾӀӁӃӅӇӉӋӍӐӒӔӖӘӚӜӞӠӢӤӦӨӪӬӮӰӲӴӶӸԀԂԄԆԈԊԌԎԱԲԳԴԵԶԷԸԹԺԻԼԽԾԿՀՁՂՃՄՅՆՇՈՉՊՋՌՍՎՏՐՑՒՓՔՕՖႠႡႢႣႤႥႦႧႨႩႪႫႬႭႮႯႰႱႲႳႴႵႶႷႸႹႺႻႼႽႾႿჀჁჂჃჄჅḀḂḄḆḈḊḌḎḐḒḔḖḘḚḜḞḠḢḤḦḨḪḬḮḰḲḴḶḸḺḼḾṀṂṄṆṈṊṌṎṐṒṔṖṘṚṜṞṠṢṤṦṨṪṬṮṰṲṴṶṸṺṼṾẀẂẄẆẈẊẌẎẐẒẔẠẢẤẦẨẪẬẮẰẲẴẶẸẺẼẾỀỂỄỆỈỊỌỎỐỒỔỖỘỚỜỞỠỢỤỦỨỪỬỮỰỲỴỶỸἈἉἊἋἌἍἎἏἘἙἚἛἜἝἨἩἪἫἬἭἮἯἸἹἺἻἼἽἾἿὈὉὊὋὌὍὙὛὝὟὨὩὪὫὬὭὮὯᾸᾹᾺΆῈΈῊΉῘῙῚΊῨῩῪΎῬῸΌῺΏabcdefghijklmnopqrstuvwxyzªµºßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿāăąćĉċčďđēĕėęěĝğġģĥħĩīĭįıijĵķĸĺļľŀłńņňʼnŋōŏőœŕŗřśŝşšţťŧũūŭůűųŵŷźżžſƀƃƅƈƌƍƒƕƙƚƛƞơƣƥƨƪƫƭưƴƶƹƺƽƾƿdžljnjǎǐǒǔǖǘǚǜǝǟǡǣǥǧǩǫǭǯǰdzǵǹǻǽǿȁȃȅȇȉȋȍȏȑȓȕȗșțȝȟȡȣȥȧȩȫȭȯȱȳȴȵȶȷȸȹȼȿɀɐɑɒɓɔɕɖɗɘəɚɛɜɝɞɟɠɡɢɣɤɥɦɧɨɩɪɫɬɭɮɯɰɱɲɳɴɵɶɷɸɹɺɻɼɽɾɿʀʁʂʃʄʅʆʇʈʉʊʋʌʍʎʏʐʑʒʓʔʕʖʗʘʙʚʛʜʝʞʟʠʡʢʣʤʥʦʧʨʩʪʫʬʭʮʯΐάέήίΰαβγδεζηθικλμνξοπρςστυφχψωϊϋόύώϐϑϕϖϗϙϛϝϟϡϣϥϧϩϫϭϯϰϱϲϳϵϸϻϼабвгдежзийклмнопрстуфхцчшщъыьэюяѐёђѓєѕіїјљњћќѝўџѡѣѥѧѩѫѭѯѱѳѵѷѹѻѽѿҁҋҍҏґғҕҗҙқҝҟҡңҥҧҩҫҭүұҳҵҷҹһҽҿӂӄӆӈӊӌӎӑӓӕӗәӛӝӟӡӣӥӧөӫӭӯӱӳӵӷӹԁԃԅԇԉԋԍԏաբգդեզէըթժիլխծկհձղճմյնշոչպջռսվտրցւփքօֆևᴀᴁᴂᴃᴄᴅᴆᴇᴈᴉᴊᴋᴌᴍᴎᴏᴐᴑᴒᴓᴔᴕᴖᴗᴘᴙᴚᴛᴜᴝᴞᴟᴠᴡᴢᴣᴤᴥᴦᴧᴨᴩᴪᴫᵢᵣᵤᵥᵦᵧᵨᵩᵪᵫᵬᵭᵮᵯᵰᵱᵲᵳᵴᵵᵶᵷᵹᵺᵻᵼᵽᵾᵿᶀᶁᶂᶃᶄᶅᶆᶇᶈᶉᶊᶋᶌᶍᶎᶏᶐᶑᶒᶓᶔᶕᶖᶗᶘᶙᶚḁḃḅḇḉḋḍḏḑḓḕḗḙḛḝḟḡḣḥḧḩḫḭḯḱḳḵḷḹḻḽḿṁṃṅṇṉṋṍṏṑṓṕṗṙṛṝṟṡṣṥṧṩṫṭṯṱṳṵṷṹṻṽṿẁẃẅẇẉẋẍẏẑẓẕẖẗẘẙẚẛạảấầẩẫậắằẳẵặẹẻẽếềểễệỉịọỏốồổỗộớờởỡợụủứừửữựỳỵỷỹἀἁἂἃἄἅἆἇἐἑἒἓἔἕἠἡἢἣἤἥἦἧἰἱἲἳἴἵἶἷὀὁὂὃὄὅὐὑὒὓὔὕὖὗὠὡὢὣὤὥὦὧὰάὲέὴήὶίὸόὺύὼώᾀᾁᾂᾃᾄᾅᾆᾇᾐᾑᾒᾓᾔᾕᾖᾗᾠᾡᾢᾣᾤᾥᾦᾧᾰᾱᾲᾳᾴᾶᾷιῂῃῄῆῇῐῑῒΐῖῗῠῡῢΰῤῥῦῧῲῳῴῶῷⲁⲃⲅⲇⲉⲋⲍⲏⲑⲓⲕⲗⲙⲛⲝⲟⲡⲣⲥⲧⲩⲫⲭⲯⲱⲳⲵⲷⲹⲻⲽⲿⳁⳃⳅⳇⳉⳋⳍⳏⳑⳓⳕⳗⳙⳛⳝⳟⳡⳣⳤⴀⴁⴂⴃⴄⴅⴆⴇⴈⴉⴊⴋⴌⴍⴎⴏⴐⴑⴒⴓⴔⴕⴖⴗⴘⴙⴚⴛⴜⴝⴞⴟⴠⴡⴢⴣⴤⴥfffiflffifflſtstﬓﬔﬕﬖﬗ\d_^]/utf
|
||
|
||
/^[^d]*?$/
|
||
abc
|
||
|
||
/^[^d]*?$/utf
|
||
abc
|
||
|
||
/^[^d]*?$/i
|
||
abc
|
||
|
||
/^[^d]*?$/i,utf
|
||
abc
|
||
|
||
/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/utf
|
||
|
||
/^[a\x{c0}]b/utf
|
||
\x{c0}b
|
||
|
||
/^([a\x{c0}]*?)aa/utf
|
||
a\x{c0}aaaa/
|
||
|
||
/^([a\x{c0}]*?)aa/utf
|
||
a\x{c0}aaaa/
|
||
a\x{c0}a\x{c0}aaa/
|
||
|
||
/^([a\x{c0}]*)aa/utf
|
||
a\x{c0}aaaa/
|
||
a\x{c0}a\x{c0}aaa/
|
||
|
||
/^([a\x{c0}]*)a\x{c0}/utf
|
||
a\x{c0}aaaa/
|
||
a\x{c0}a\x{c0}aaa/
|
||
|
||
/A*/g,utf
|
||
AAB\x{123}BAA
|
||
|
||
/(abc)\1/i,utf
|
||
\= Expect no match
|
||
abc
|
||
|
||
/(abc)\1/utf
|
||
\= Expect no match
|
||
abc
|
||
|
||
/a(*:a\x{1234}b)/utf,mark
|
||
abc
|
||
|
||
/a(*:a£b)/utf,mark
|
||
abc
|
||
|
||
# Noncharacters
|
||
|
||
/./utf
|
||
\x{fffe}
|
||
\x{ffff}
|
||
\x{1fffe}
|
||
\x{1ffff}
|
||
\x{2fffe}
|
||
\x{2ffff}
|
||
\x{3fffe}
|
||
\x{3ffff}
|
||
\x{4fffe}
|
||
\x{4ffff}
|
||
\x{5fffe}
|
||
\x{5ffff}
|
||
\x{6fffe}
|
||
\x{6ffff}
|
||
\x{7fffe}
|
||
\x{7ffff}
|
||
\x{8fffe}
|
||
\x{8ffff}
|
||
\x{9fffe}
|
||
\x{9ffff}
|
||
\x{afffe}
|
||
\x{affff}
|
||
\x{bfffe}
|
||
\x{bffff}
|
||
\x{cfffe}
|
||
\x{cffff}
|
||
\x{dfffe}
|
||
\x{dffff}
|
||
\x{efffe}
|
||
\x{effff}
|
||
\x{ffffe}
|
||
\x{fffff}
|
||
\x{10fffe}
|
||
\x{10ffff}
|
||
\x{fdd0}
|
||
\x{fdd1}
|
||
\x{fdd2}
|
||
\x{fdd3}
|
||
\x{fdd4}
|
||
\x{fdd5}
|
||
\x{fdd6}
|
||
\x{fdd7}
|
||
\x{fdd8}
|
||
\x{fdd9}
|
||
\x{fdda}
|
||
\x{fddb}
|
||
\x{fddc}
|
||
\x{fddd}
|
||
\x{fdde}
|
||
\x{fddf}
|
||
\x{fde0}
|
||
\x{fde1}
|
||
\x{fde2}
|
||
\x{fde3}
|
||
\x{fde4}
|
||
\x{fde5}
|
||
\x{fde6}
|
||
\x{fde7}
|
||
\x{fde8}
|
||
\x{fde9}
|
||
\x{fdea}
|
||
\x{fdeb}
|
||
\x{fdec}
|
||
\x{fded}
|
||
\x{fdee}
|
||
\x{fdef}
|
||
|
||
/^\d*\w{4}/utf
|
||
1234
|
||
\= Expect no match
|
||
123
|
||
|
||
/^[^b]*\w{4}/utf
|
||
aaaa
|
||
\= Expect no match
|
||
aaa
|
||
|
||
/^[^b]*\w{4}/i,utf
|
||
aaaa
|
||
\= Expect no match
|
||
aaa
|
||
|
||
/^\x{100}*.{4}/utf
|
||
\x{100}\x{100}\x{100}\x{100}
|
||
\= Expect no match
|
||
\x{100}\x{100}\x{100}
|
||
|
||
/^\x{100}*.{4}/i,utf
|
||
\x{100}\x{100}\x{100}\x{100}
|
||
\= Expect no match
|
||
\x{100}\x{100}\x{100}
|
||
|
||
/^a+[a\x{200}]/utf
|
||
aa
|
||
|
||
/^.\B.\B./utf
|
||
\x{10123}\x{10124}\x{10125}
|
||
|
||
/^#[^\x{ffff}]#[^\x{ffff}]#[^\x{ffff}]#/utf
|
||
#\x{10000}#\x{100}#\x{10ffff}#
|
||
|
||
# Unicode property support tests
|
||
|
||
/^\pC\pL\pM\pN\pP\pS\pZ</utf
|
||
\x7f\x{c0}\x{30f}\x{660}\x{66c}\x{f01}\x{1680}<
|
||
\np\x{300}9!\$ <
|
||
\= Expect no match
|
||
ap\x{300}9!\$ <
|
||
|
||
/^\PC/utf
|
||
X
|
||
\= Expect no match
|
||
\x7f
|
||
|
||
/^\PL/utf
|
||
9
|
||
\= Expect no match
|
||
\x{c0}
|
||
|
||
/^\PM/utf
|
||
X
|
||
\= Expect no match
|
||
\x{30f}
|
||
|
||
/^\PN/utf
|
||
X
|
||
\= Expect no match
|
||
\x{660}
|
||
|
||
/^\PP/utf
|
||
X
|
||
\= Expect no match
|
||
\x{66c}
|
||
|
||
/^\PS/utf
|
||
X
|
||
\= Expect no match
|
||
\x{f01}
|
||
|
||
/^\PZ/utf
|
||
X
|
||
\= Expect no match
|
||
\x{1680}
|
||
|
||
/^\p{Cc}/utf
|
||
\x{017}
|
||
\x{09f}
|
||
\= Expect no match
|
||
\x{0600}
|
||
|
||
/^\p{Cf}/utf
|
||
\x{601}
|
||
\= Expect no match
|
||
\x{09f}
|
||
|
||
/^\p{Cn}/utf
|
||
\x{e0000}
|
||
\= Expect no match
|
||
\x{09f}
|
||
|
||
/^\p{Co}/utf
|
||
\x{f8ff}
|
||
\= Expect no match
|
||
\x{09f}
|
||
|
||
/^\p{Ll}/utf
|
||
a
|
||
\= Expect no match
|
||
Z
|
||
\x{e000}
|
||
|
||
/^\p{Lm}/utf
|
||
\x{2b0}
|
||
\= Expect no match
|
||
a
|
||
|
||
/^\p{Lo}/utf
|
||
\x{1bb}
|
||
\x{3400}
|
||
\x{3401}
|
||
\x{4d00}
|
||
\x{4db4}
|
||
\x{4db5}
|
||
\x{4db6}
|
||
\= Expect no match
|
||
a
|
||
\x{2b0}
|
||
|
||
/^\p{Lt}/utf
|
||
\x{1c5}
|
||
\= Expect no match
|
||
a
|
||
\x{2b0}
|
||
|
||
/^\p{Lu}/utf
|
||
A
|
||
\= Expect no match
|
||
\x{2b0}
|
||
|
||
/^\p{Mc}/utf
|
||
\x{903}
|
||
\= Expect no match
|
||
X
|
||
\x{300}
|
||
|
||
/^\p{Me}/utf
|
||
\x{488}
|
||
\= Expect no match
|
||
X
|
||
\x{903}
|
||
\x{300}
|
||
|
||
/^\p{Mn}/utf
|
||
\x{300}
|
||
\= Expect no match
|
||
X
|
||
\x{903}
|
||
|
||
/^\p{Nd}+/utf
|
||
0123456789\x{660}\x{661}\x{662}\x{663}\x{664}\x{665}\x{666}\x{667}\x{668}\x{669}\x{66a}
|
||
\x{6f0}\x{6f1}\x{6f2}\x{6f3}\x{6f4}\x{6f5}\x{6f6}\x{6f7}\x{6f8}\x{6f9}\x{6fa}
|
||
\x{966}\x{967}\x{968}\x{969}\x{96a}\x{96b}\x{96c}\x{96d}\x{96e}\x{96f}\x{970}
|
||
\= Expect no match
|
||
X
|
||
|
||
/^\p{Nl}/utf
|
||
\x{16ee}
|
||
\= Expect no match
|
||
X
|
||
\x{966}
|
||
|
||
/^\p{No}/utf
|
||
\x{b2}
|
||
\x{b3}
|
||
\= Expect no match
|
||
X
|
||
\x{16ee}
|
||
|
||
/^\p{Pc}/utf
|
||
\x5f
|
||
\x{203f}
|
||
\= Expect no match
|
||
X
|
||
-
|
||
\x{58a}
|
||
|
||
/^\p{Pd}/utf
|
||
-
|
||
\x{58a}
|
||
\= Expect no match
|
||
X
|
||
\x{203f}
|
||
|
||
/^\p{Pe}/utf
|
||
)
|
||
]
|
||
}
|
||
\x{f3b}
|
||
\= Expect no match
|
||
X
|
||
\x{203f}
|
||
(
|
||
[
|
||
{
|
||
\x{f3c}
|
||
|
||
/^\p{Pf}/utf
|
||
\x{bb}
|
||
\x{2019}
|
||
\= Expect no match
|
||
X
|
||
\x{203f}
|
||
|
||
/^\p{Pi}/utf
|
||
\x{ab}
|
||
\x{2018}
|
||
\= Expect no match
|
||
X
|
||
\x{203f}
|
||
|
||
/^\p{Po}/utf
|
||
!
|
||
\x{37e}
|
||
\= Expect no match
|
||
X
|
||
\x{203f}
|
||
|
||
/^\p{Ps}/utf
|
||
(
|
||
[
|
||
{
|
||
\x{f3c}
|
||
\= Expect no match
|
||
X
|
||
)
|
||
]
|
||
}
|
||
\x{f3b}
|
||
|
||
/^\p{Sk}/utf
|
||
\x{2c2}
|
||
\= Expect no match
|
||
X
|
||
\x{9f2}
|
||
|
||
/^\p{Sm}+/utf
|
||
+<|~\x{ac}\x{2044}
|
||
\= Expect no match
|
||
X
|
||
\x{9f2}
|
||
|
||
/^\p{So}/utf
|
||
\x{a6}
|
||
\x{482}
|
||
\= Expect no match
|
||
X
|
||
\x{9f2}
|
||
|
||
/^\p{Zl}/utf
|
||
\x{2028}
|
||
\= Expect no match
|
||
X
|
||
\x{2029}
|
||
|
||
/^\p{Zp}/utf
|
||
\x{2029}
|
||
\= Expect no match
|
||
X
|
||
\x{2028}
|
||
|
||
/\p{Nd}+(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}+?(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}{2,}(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}{2,}?(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}*(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}*?(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}{2}(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}{2,3}(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}{2,3}?(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}?(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}??(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}*+(..)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}*+(...)/utf
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/\p{Nd}*+(....)/utf
|
||
\= Expect no match
|
||
\x{660}\x{661}\x{662}ABC
|
||
|
||
/(?<=A\p{Nd})XYZ/utf
|
||
A2XYZ
|
||
123A5XYZPQR
|
||
ABA\x{660}XYZpqr
|
||
\= Expect no match
|
||
AXYZ
|
||
XYZ
|
||
|
||
/(?<!\pL)XYZ/utf
|
||
1XYZ
|
||
AB=XYZ..
|
||
XYZ
|
||
\= Expect no match
|
||
WXYZ
|
||
|
||
/[\P{Nd}]+/utf
|
||
abcd
|
||
\= Expect no match
|
||
1234
|
||
|
||
/\D+/utf
|
||
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
|
||
\= Expect no match
|
||
11111111111111111111111111111111111111111111111111111111111111111111111
|
||
|
||
/\P{Nd}+/utf
|
||
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
|
||
\= Expect no match
|
||
11111111111111111111111111111111111111111111111111111111111111111111111
|
||
|
||
/[\D]+/utf
|
||
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
|
||
\= Expect no match
|
||
11111111111111111111111111111111111111111111111111111111111111111111111
|
||
|
||
/[\P{Nd}]+/utf
|
||
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
|
||
\= Expect no match
|
||
11111111111111111111111111111111111111111111111111111111111111111111111
|
||
|
||
/[\D\P{Nd}]+/utf
|
||
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
|
||
\= Expect no match
|
||
11111111111111111111111111111111111111111111111111111111111111111111111
|
||
|
||
/\pL/utf
|
||
a
|
||
A
|
||
|
||
/\pL/i,utf
|
||
a
|
||
A
|
||
|
||
/\p{Lu}/utf
|
||
A
|
||
aZ
|
||
\= Expect no match
|
||
abc
|
||
|
||
/\p{Ll}/utf
|
||
a
|
||
Az
|
||
\= Expect no match
|
||
ABC
|
||
|
||
/A\x{391}\x{10427}\x{ff3a}\x{1fb0}/utf
|
||
A\x{391}\x{10427}\x{ff3a}\x{1fb0}
|
||
\= Expect no match
|
||
a\x{391}\x{10427}\x{ff3a}\x{1fb0}
|
||
A\x{3b1}\x{10427}\x{ff3a}\x{1fb0}
|
||
A\x{391}\x{1044F}\x{ff3a}\x{1fb0}
|
||
A\x{391}\x{10427}\x{ff5a}\x{1fb0}
|
||
A\x{391}\x{10427}\x{ff3a}\x{1fb8}
|
||
|
||
/A\x{391}\x{10427}\x{ff3a}\x{1fb0}/i,utf
|
||
A\x{391}\x{10427}\x{ff3a}\x{1fb0}
|
||
a\x{391}\x{10427}\x{ff3a}\x{1fb0}
|
||
A\x{3b1}\x{10427}\x{ff3a}\x{1fb0}
|
||
A\x{391}\x{1044F}\x{ff3a}\x{1fb0}
|
||
A\x{391}\x{10427}\x{ff5a}\x{1fb0}
|
||
A\x{391}\x{10427}\x{ff3a}\x{1fb8}
|
||
|
||
/\x{391}+/i,utf
|
||
\x{391}\x{3b1}\x{3b1}\x{3b1}\x{391}
|
||
|
||
/\x{391}{3,5}(.)/i,utf
|
||
\x{391}\x{3b1}\x{3b1}\x{3b1}\x{391}X
|
||
|
||
/\x{391}{3,5}?(.)/i,utf
|
||
\x{391}\x{3b1}\x{3b1}\x{3b1}\x{391}X
|
||
|
||
/[\x{391}\x{ff3a}]/i,utf
|
||
\x{391}
|
||
\x{ff3a}
|
||
\x{3b1}
|
||
\x{ff5a}
|
||
|
||
/^(\X*)C/utf
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}C
|
||
|
||
/^(\X*?)C/utf
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}C
|
||
|
||
/^(\X*)(.)/utf
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}C
|
||
|
||
/^(\X*?)(.)/utf
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}
|
||
A\x{300}\x{301}\x{302}BCA\x{300}\x{301}C
|
||
|
||
/^\X(.)/utf
|
||
\= Expect no match
|
||
A\x{300}\x{301}\x{302}
|
||
|
||
/^\X{2,3}(.)/utf
|
||
A\x{300}\x{301}B\x{300}X
|
||
A\x{300}\x{301}B\x{300}C\x{300}\x{301}
|
||
A\x{300}\x{301}B\x{300}C\x{300}\x{301}X
|
||
A\x{300}\x{301}B\x{300}C\x{300}\x{301}DA\x{300}X
|
||
|
||
/^\X{2,3}?(.)/utf
|
||
A\x{300}\x{301}B\x{300}X
|
||
A\x{300}\x{301}B\x{300}C\x{300}\x{301}
|
||
A\x{300}\x{301}B\x{300}C\x{300}\x{301}X
|
||
A\x{300}\x{301}B\x{300}C\x{300}\x{301}DA\x{300}X
|
||
|
||
/^\X/utf
|
||
A
|
||
A\x{300}BC
|
||
A\x{300}\x{301}\x{302}BC
|
||
\x{300}
|
||
|
||
/^\p{Han}+/utf
|
||
\x{2e81}\x{3007}\x{2f804}\x{31a0}
|
||
\= Expect no match
|
||
\x{2e7f}
|
||
|
||
/^\P{Katakana}+/utf
|
||
\x{3105}
|
||
\= Expect no match
|
||
\x{30ff}
|
||
|
||
/^[\p{Arabic}]/utf
|
||
\x{06e9}
|
||
\x{060b}
|
||
\= Expect no match
|
||
X\x{06e9}
|
||
|
||
/^[\P{Yi}]/utf
|
||
\x{2f800}
|
||
\= Expect no match
|
||
\x{a014}
|
||
\x{a4c6}
|
||
|
||
/^\p{Any}X/utf
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
\= Expect no match
|
||
X
|
||
|
||
/^\P{Any}X/utf
|
||
\= Expect no match
|
||
AX
|
||
|
||
/^\p{Any}?X/utf
|
||
XYZ
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
\= Expect no match
|
||
ABXYZ
|
||
|
||
/^\P{Any}?X/utf
|
||
XYZ
|
||
\= Expect no match
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
ABXYZ
|
||
|
||
/^\p{Any}+X/utf
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
\= Expect no match
|
||
XYZ
|
||
|
||
/^\P{Any}+X/utf
|
||
\= Expect no match
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
XYZ
|
||
|
||
/^\p{Any}*X/utf
|
||
XYZ
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
|
||
/^\P{Any}*X/utf
|
||
XYZ
|
||
\= Expect no match
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
|
||
/^[\p{Any}]X/utf
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
\= Expect no match
|
||
X
|
||
|
||
/^[\P{Any}]X/utf
|
||
\= Expect no match
|
||
AX
|
||
|
||
/^[\p{Any}]?X/utf
|
||
XYZ
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
\= Expect no match
|
||
ABXYZ
|
||
|
||
/^[\P{Any}]?X/utf
|
||
XYZ
|
||
\= Expect no match
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
ABXYZ
|
||
|
||
/^[\p{Any}]+X/utf
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
\= Expect no match
|
||
XYZ
|
||
|
||
/^[\P{Any}]+X/utf
|
||
\= Expect no match
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
XYZ
|
||
|
||
/^[\p{Any}]*X/utf
|
||
XYZ
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
|
||
/^[\P{Any}]*X/utf
|
||
XYZ
|
||
\= Expect no match
|
||
AXYZ
|
||
\x{1234}XYZ
|
||
A\x{1234}XYZ
|
||
|
||
/^\p{Any}{3,5}?/utf
|
||
abcdefgh
|
||
\x{1234}\n\r\x{3456}xyz
|
||
|
||
/^\p{Any}{3,5}/utf
|
||
abcdefgh
|
||
\x{1234}\n\r\x{3456}xyz
|
||
|
||
/^\P{Any}{3,5}?/utf
|
||
\= Expect no match
|
||
abcdefgh
|
||
\x{1234}\n\r\x{3456}xyz
|
||
|
||
/^\p{L&}X/utf
|
||
AXY
|
||
aXY
|
||
\x{1c5}XY
|
||
\= Expect no match
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
!XY
|
||
|
||
/^[\p{L&}]X/utf
|
||
AXY
|
||
aXY
|
||
\x{1c5}XY
|
||
\= Expect no match
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
!XY
|
||
|
||
/^\p{L&}+X/utf
|
||
AXY
|
||
aXY
|
||
AbcdeXyz
|
||
\x{1c5}AbXY
|
||
abcDEXypqreXlmn
|
||
\= Expect no match
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
!XY
|
||
|
||
/^[\p{L&}]+X/utf
|
||
AXY
|
||
aXY
|
||
AbcdeXyz
|
||
\x{1c5}AbXY
|
||
abcDEXypqreXlmn
|
||
\= Expect no match
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
!XY
|
||
|
||
/^\p{L&}+?X/utf
|
||
AXY
|
||
aXY
|
||
AbcdeXyz
|
||
\x{1c5}AbXY
|
||
abcDEXypqreXlmn
|
||
\= Expect no match
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
!XY
|
||
|
||
/^[\p{L&}]+?X/utf
|
||
AXY
|
||
aXY
|
||
AbcdeXyz
|
||
\x{1c5}AbXY
|
||
abcDEXypqreXlmn
|
||
\= Expect no match
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
!XY
|
||
|
||
/^\P{L&}X/utf
|
||
!XY
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
\= Expect no match
|
||
\x{1c5}XY
|
||
AXY
|
||
|
||
/^[\P{L&}]X/utf
|
||
!XY
|
||
\x{1bb}XY
|
||
\x{2b0}XY
|
||
\= Expect no match
|
||
\x{1c5}XY
|
||
AXY
|
||
|
||
/^(\p{Z}[^\p{C}\p{Z}]+)*$/
|
||
\xa0!
|
||
|
||
/^[\pL](abc)(?1)/
|
||
AabcabcYZ
|
||
|
||
/([\pL]=(abc))*X/
|
||
L=abcX
|
||
|
||
/^\p{Balinese}\p{Cuneiform}\p{Nko}\p{Phags_Pa}\p{Phoenician}/utf
|
||
\x{1b00}\x{12000}\x{7c0}\x{a840}\x{10900}
|
||
|
||
# Check property support in non-UTF mode
|
||
|
||
/\p{L}{4}/
|
||
123abcdefg
|
||
123abc\xc4\xc5zz
|
||
|
||
/\X{1,3}\d/
|
||
\= Expect no match
|
||
\x8aBCD
|
||
|
||
/\X?\d/
|
||
\= Expect no match
|
||
\x8aBCD
|
||
|
||
/\P{L}?\d/
|
||
\= Expect no match
|
||
\x8aBCD
|
||
|
||
/[\PPP\x8a]{1,}\x80/
|
||
A\x80
|
||
|
||
/^[\p{Arabic}]/utf
|
||
\x{604}
|
||
\x{60e}
|
||
\x{656}
|
||
\x{657}
|
||
\x{658}
|
||
\x{659}
|
||
\x{65a}
|
||
\x{65b}
|
||
\x{65c}
|
||
\x{65d}
|
||
\x{65e}
|
||
\x{65f}
|
||
\x{66a}
|
||
\x{6e9}
|
||
\x{6ef}
|
||
\x{6fa}
|
||
|
||
/^\p{Cyrillic}/utf
|
||
\x{1d2b}
|
||
|
||
/^\p{Common}/utf
|
||
\x{2116}
|
||
\x{1D183}
|
||
|
||
/^\p{Inherited}/utf
|
||
\x{200c}
|
||
\= Expect no match
|
||
\x{64a}
|
||
\x{656}
|
||
|
||
/^\p{Shavian}/utf
|
||
\x{10450}
|
||
\x{1047f}
|
||
|
||
/^\p{Deseret}/utf
|
||
\x{10400}
|
||
\x{1044f}
|
||
|
||
/^\p{Osmanya}/utf
|
||
\x{10480}
|
||
\x{1049d}
|
||
\x{104a0}
|
||
\x{104a9}
|
||
\= Expect no match
|
||
\x{1049e}
|
||
\x{1049f}
|
||
\x{104aa}
|
||
|
||
/\p{Carian}\p{Cham}\p{Kayah_Li}\p{Lepcha}\p{Lycian}\p{Lydian}\p{Ol_Chiki}\p{Rejang}\p{Saurashtra}\p{Sundanese}\p{Vai}/utf
|
||
\x{102A4}\x{AA52}\x{A91D}\x{1C46}\x{10283}\x{1092E}\x{1C6B}\x{A93B}\x{A8BF}\x{1BA0}\x{A50A}====
|
||
|
||
/\x{a77d}\x{1d79}/i,utf
|
||
\x{a77d}\x{1d79}
|
||
\x{1d79}\x{a77d}
|
||
|
||
/\x{a77d}\x{1d79}/utf
|
||
\x{a77d}\x{1d79}
|
||
\= Expect no match
|
||
\x{1d79}\x{a77d}
|
||
|
||
/(A)\1/i,utf
|
||
AA
|
||
Aa
|
||
aa
|
||
aA
|
||
|
||
/(\x{10a})\1/i,utf
|
||
\x{10a}\x{10a}
|
||
\x{10a}\x{10b}
|
||
\x{10b}\x{10b}
|
||
\x{10b}\x{10a}
|
||
|
||
# The next two tests are for property support in non-UTF mode
|
||
|
||
/(?:\p{Lu}|\x20)+/
|
||
\x41\x20\x50\xC2\x54\xC9\x20\x54\x4F\x44\x41\x59
|
||
|
||
/[\p{Lu}\x20]+/
|
||
\x41\x20\x50\xC2\x54\xC9\x20\x54\x4F\x44\x41\x59
|
||
|
||
/\p{Avestan}\p{Bamum}\p{Egyptian_Hieroglyphs}\p{Imperial_Aramaic}\p{Inscriptional_Pahlavi}\p{Inscriptional_Parthian}\p{Javanese}\p{Kaithi}\p{Lisu}\p{Meetei_Mayek}\p{Old_South_Arabian}\p{Old_Turkic}\p{Samaritan}\p{Tai_Tham}\p{Tai_Viet}/utf
|
||
\x{10b00}\x{a6ef}\x{13007}\x{10857}\x{10b78}\x{10b58}\x{a980}\x{110c1}\x{a4ff}\x{abc0}\x{10a7d}\x{10c48}\x{0800}\x{1aad}\x{aac0}
|
||
|
||
/^\w+/utf,ucp
|
||
Az_\x{aa}\x{c0}\x{1c5}\x{2b0}\x{3b6}\x{1d7c9}\x{2fa1d}1\x{660}\x{bef}\x{16ee}
|
||
|
||
/^[[:xdigit:]]*/utf,ucp
|
||
1a\x{660}\x{bef}\x{16ee}
|
||
|
||
/^\d+/utf,ucp
|
||
1\x{660}\x{bef}\x{16ee}
|
||
|
||
/^[[:digit:]]+/utf,ucp
|
||
1\x{660}\x{bef}\x{16ee}
|
||
|
||
/^>\s+/utf,ucp
|
||
>\x{20}\x{a0}\x{1680}\x{2028}\x{2029}\x{202f}\x{9}\x{b}
|
||
|
||
/^>\pZ+/utf,ucp
|
||
>\x{20}\x{a0}\x{1680}\x{2028}\x{2029}\x{202f}\x{9}\x{b}
|
||
|
||
/^>[[:space:]]*/utf,ucp
|
||
>\x{20}\x{a0}\x{1680}\x{2028}\x{2029}\x{202f}\x{9}\x{b}
|
||
|
||
/^>[[:blank:]]*/utf,ucp
|
||
>\x{20}\x{a0}\x{1680}\x{2000}\x{202f}\x{9}\x{b}\x{2028}
|
||
|
||
/^[[:alpha:]]*/utf,ucp
|
||
Az\x{aa}\x{c0}\x{1c5}\x{2b0}\x{3b6}\x{1d7c9}\x{2fa1d}
|
||
|
||
/^[[:alnum:]]*/utf,ucp
|
||
Az\x{aa}\x{c0}\x{1c5}\x{2b0}\x{3b6}\x{1d7c9}\x{2fa1d}1\x{660}\x{bef}\x{16ee}
|
||
|
||
/^[[:cntrl:]]*/utf,ucp
|
||
\x{0}\x{09}\x{1f}\x{7f}\x{9f}
|
||
|
||
/^[[:graph:]]*/utf,ucp
|
||
A\x{a1}\x{a0}
|
||
|
||
/^[[:print:]]*/utf,ucp
|
||
A z\x{a0}\x{a1}
|
||
|
||
/^[[:punct:]]*/utf,ucp
|
||
.+\x{a1}\x{a0}
|
||
|
||
/\p{Zs}*?\R/
|
||
\= Expect no match
|
||
a\xFCb
|
||
|
||
/\p{Zs}*\R/
|
||
\= Expect no match
|
||
a\xFCb
|
||
|
||
/ⱥ/i,utf
|
||
ⱥ
|
||
Ⱥx
|
||
Ⱥ
|
||
|
||
/[ⱥ]/i,utf
|
||
ⱥ
|
||
Ⱥx
|
||
Ⱥ
|
||
|
||
/Ⱥ/i,utf
|
||
Ⱥ
|
||
ⱥ
|
||
|
||
# These are tests for extended grapheme clusters
|
||
|
||
/^\X/utf,aftertext
|
||
G\x{34e}\x{34e}X
|
||
\x{34e}\x{34e}X
|
||
\x04X
|
||
\x{1100}X
|
||
\x{1100}\x{34e}X
|
||
\x{1b04}\x{1b04}X
|
||
*These match up to the roman letters
|
||
\x{1111}\x{1111}L,L
|
||
\x{1111}\x{1111}\x{1169}L,L,V
|
||
\x{1111}\x{ae4c}L, LV
|
||
\x{1111}\x{ad89}L, LVT
|
||
\x{1111}\x{ae4c}\x{1169}L, LV, V
|
||
\x{1111}\x{ae4c}\x{1169}\x{1169}L, LV, V, V
|
||
\x{1111}\x{ae4c}\x{1169}\x{11fe}L, LV, V, T
|
||
\x{1111}\x{ad89}\x{11fe}L, LVT, T
|
||
\x{1111}\x{ad89}\x{11fe}\x{11fe}L, LVT, T, T
|
||
\x{ad89}\x{11fe}\x{11fe}LVT, T, T
|
||
*These match just the first codepoint (invalid sequence)
|
||
\x{1111}\x{11fe}L, T
|
||
\x{ae4c}\x{1111}LV, L
|
||
\x{ae4c}\x{ae4c}LV, LV
|
||
\x{ae4c}\x{ad89}LV, LVT
|
||
\x{1169}\x{1111}V, L
|
||
\x{1169}\x{ae4c}V, LV
|
||
\x{1169}\x{ad89}V, LVT
|
||
\x{ad89}\x{1111}LVT, L
|
||
\x{ad89}\x{1169}LVT, V
|
||
\x{ad89}\x{ae4c}LVT, LV
|
||
\x{ad89}\x{ad89}LVT, LVT
|
||
\x{11fe}\x{1111}T, L
|
||
\x{11fe}\x{1169}T, V
|
||
\x{11fe}\x{ae4c}T, LV
|
||
\x{11fe}\x{ad89}T, LVT
|
||
*Test extend and spacing mark
|
||
\x{1111}\x{ae4c}\x{0711}L, LV, extend
|
||
\x{1111}\x{ae4c}\x{1b04}L, LV, spacing mark
|
||
\x{1111}\x{ae4c}\x{1b04}\x{0711}\x{1b04}L, LV, spacing mark, extend, spacing mark
|
||
*Test CR, LF, and control
|
||
\x0d\x{0711}CR, extend
|
||
\x0d\x{1b04}CR, spacingmark
|
||
\x0a\x{0711}LF, extend
|
||
\x0a\x{1b04}LF, spacingmark
|
||
\x0b\x{0711}Control, extend
|
||
\x09\x{1b04}Control, spacingmark
|
||
*There are no Prepend characters, so we can't test Prepend, CR
|
||
|
||
/^(?>\X{2})X/utf,aftertext
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
|
||
/^\X{2,4}X/utf,aftertext
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
|
||
/^\X{2,4}?X/utf,aftertext
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}\x{1111}\x{ae4c}X
|
||
|
||
/\X*Z/utf,no_start_optimize
|
||
\= Expect no match
|
||
A\x{300}
|
||
|
||
/\X*(.)/utf,no_start_optimize
|
||
A\x{1111}\x{ae4c}\x{1169}
|
||
|
||
# --------------------------------------------
|
||
|
||
/\x{1e9e}+/i,utf
|
||
\x{1e9e}\x{00df}
|
||
|
||
/[z\x{1e9e}]+/i,utf
|
||
\x{1e9e}\x{00df}
|
||
|
||
/\x{00df}+/i,utf
|
||
\x{1e9e}\x{00df}
|
||
|
||
/[z\x{00df}]+/i,utf
|
||
\x{1e9e}\x{00df}
|
||
|
||
/\x{1f88}+/i,utf
|
||
\x{1f88}\x{1f80}
|
||
|
||
/[z\x{1f88}]+/i,utf
|
||
\x{1f88}\x{1f80}
|
||
|
||
# Check a reference with more than one other case
|
||
|
||
/^(\x{00b5})\1{2}$/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
# Characters with more than one other case; test in classes
|
||
|
||
/[z\x{00b5}]+/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
/[z\x{039c}]+/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
/[z\x{03bc}]+/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
/[z\x{00c5}]+/i,utf
|
||
\x{00c5}\x{00e5}\x{212b}
|
||
|
||
/[z\x{00e5}]+/i,utf
|
||
\x{00c5}\x{00e5}\x{212b}
|
||
|
||
/[z\x{212b}]+/i,utf
|
||
\x{00c5}\x{00e5}\x{212b}
|
||
|
||
/[z\x{01c4}]+/i,utf
|
||
\x{01c4}\x{01c5}\x{01c6}
|
||
|
||
/[z\x{01c5}]+/i,utf
|
||
\x{01c4}\x{01c5}\x{01c6}
|
||
|
||
/[z\x{01c6}]+/i,utf
|
||
\x{01c4}\x{01c5}\x{01c6}
|
||
|
||
/[z\x{01c7}]+/i,utf
|
||
\x{01c7}\x{01c8}\x{01c9}
|
||
|
||
/[z\x{01c8}]+/i,utf
|
||
\x{01c7}\x{01c8}\x{01c9}
|
||
|
||
/[z\x{01c9}]+/i,utf
|
||
\x{01c7}\x{01c8}\x{01c9}
|
||
|
||
/[z\x{01ca}]+/i,utf
|
||
\x{01ca}\x{01cb}\x{01cc}
|
||
|
||
/[z\x{01cb}]+/i,utf
|
||
\x{01ca}\x{01cb}\x{01cc}
|
||
|
||
/[z\x{01cc}]+/i,utf
|
||
\x{01ca}\x{01cb}\x{01cc}
|
||
|
||
/[z\x{01f1}]+/i,utf
|
||
\x{01f1}\x{01f2}\x{01f3}
|
||
|
||
/[z\x{01f2}]+/i,utf
|
||
\x{01f1}\x{01f2}\x{01f3}
|
||
|
||
/[z\x{01f3}]+/i,utf
|
||
\x{01f1}\x{01f2}\x{01f3}
|
||
|
||
/[z\x{0345}]+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/[z\x{0399}]+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/[z\x{03b9}]+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/[z\x{1fbe}]+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/[z\x{0392}]+/i,utf
|
||
\x{0392}\x{03b2}\x{03d0}
|
||
|
||
/[z\x{03b2}]+/i,utf
|
||
\x{0392}\x{03b2}\x{03d0}
|
||
|
||
/[z\x{03d0}]+/i,utf
|
||
\x{0392}\x{03b2}\x{03d0}
|
||
|
||
/[z\x{0395}]+/i,utf
|
||
\x{0395}\x{03b5}\x{03f5}
|
||
|
||
/[z\x{03b5}]+/i,utf
|
||
\x{0395}\x{03b5}\x{03f5}
|
||
|
||
/[z\x{03f5}]+/i,utf
|
||
\x{0395}\x{03b5}\x{03f5}
|
||
|
||
/[z\x{0398}]+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/[z\x{03b8}]+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/[z\x{03d1}]+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/[z\x{03f4}]+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/[z\x{039a}]+/i,utf
|
||
\x{039a}\x{03ba}\x{03f0}
|
||
|
||
/[z\x{03ba}]+/i,utf
|
||
\x{039a}\x{03ba}\x{03f0}
|
||
|
||
/[z\x{03f0}]+/i,utf
|
||
\x{039a}\x{03ba}\x{03f0}
|
||
|
||
/[z\x{03a0}]+/i,utf
|
||
\x{03a0}\x{03c0}\x{03d6}
|
||
|
||
/[z\x{03c0}]+/i,utf
|
||
\x{03a0}\x{03c0}\x{03d6}
|
||
|
||
/[z\x{03d6}]+/i,utf
|
||
\x{03a0}\x{03c0}\x{03d6}
|
||
|
||
/[z\x{03a1}]+/i,utf
|
||
\x{03a1}\x{03c1}\x{03f1}
|
||
|
||
/[z\x{03c1}]+/i,utf
|
||
\x{03a1}\x{03c1}\x{03f1}
|
||
|
||
/[z\x{03f1}]+/i,utf
|
||
\x{03a1}\x{03c1}\x{03f1}
|
||
|
||
/[z\x{03a3}]+/i,utf
|
||
\x{03A3}\x{03C2}\x{03C3}
|
||
|
||
/[z\x{03c2}]+/i,utf
|
||
\x{03A3}\x{03C2}\x{03C3}
|
||
|
||
/[z\x{03c3}]+/i,utf
|
||
\x{03A3}\x{03C2}\x{03C3}
|
||
|
||
/[z\x{03a6}]+/i,utf
|
||
\x{03a6}\x{03c6}\x{03d5}
|
||
|
||
/[z\x{03c6}]+/i,utf
|
||
\x{03a6}\x{03c6}\x{03d5}
|
||
|
||
/[z\x{03d5}]+/i,utf
|
||
\x{03a6}\x{03c6}\x{03d5}
|
||
|
||
/[z\x{03c9}]+/i,utf
|
||
\x{03c9}\x{03a9}\x{2126}
|
||
|
||
/[z\x{03a9}]+/i,utf
|
||
\x{03c9}\x{03a9}\x{2126}
|
||
|
||
/[z\x{2126}]+/i,utf
|
||
\x{03c9}\x{03a9}\x{2126}
|
||
|
||
/[z\x{1e60}]+/i,utf
|
||
\x{1e60}\x{1e61}\x{1e9b}
|
||
|
||
/[z\x{1e61}]+/i,utf
|
||
\x{1e60}\x{1e61}\x{1e9b}
|
||
|
||
/[z\x{1e9b}]+/i,utf
|
||
\x{1e60}\x{1e61}\x{1e9b}
|
||
|
||
# Perl 5.12.4 gets these wrong, but 5.15.3 is OK
|
||
|
||
/[z\x{004b}]+/i,utf
|
||
\x{004b}\x{006b}\x{212a}
|
||
|
||
/[z\x{006b}]+/i,utf
|
||
\x{004b}\x{006b}\x{212a}
|
||
|
||
/[z\x{212a}]+/i,utf
|
||
\x{004b}\x{006b}\x{212a}
|
||
|
||
/[z\x{0053}]+/i,utf
|
||
\x{0053}\x{0073}\x{017f}
|
||
|
||
/[z\x{0073}]+/i,utf
|
||
\x{0053}\x{0073}\x{017f}
|
||
|
||
/[z\x{017f}]+/i,utf
|
||
\x{0053}\x{0073}\x{017f}
|
||
|
||
# --------------------------------------
|
||
|
||
/(ΣΆΜΟΣ) \1/i,utf
|
||
ΣΆΜΟΣ ΣΆΜΟΣ
|
||
ΣΆΜΟΣ σάμος
|
||
σάμος σάμος
|
||
σάμος σάμοσ
|
||
σάμος ΣΆΜΟΣ
|
||
|
||
/(σάμος) \1/i,utf
|
||
ΣΆΜΟΣ ΣΆΜΟΣ
|
||
ΣΆΜΟΣ σάμος
|
||
σάμος σάμος
|
||
σάμος σάμοσ
|
||
σάμος ΣΆΜΟΣ
|
||
|
||
/(ΣΆΜΟΣ) \1*/i,utf
|
||
ΣΆΜΟΣ\x20
|
||
ΣΆΜΟΣ ΣΆΜΟΣσάμοςσάμος
|
||
|
||
# Perl matches these
|
||
|
||
/\x{00b5}+/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
/\x{039c}+/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
/\x{03bc}+/i,utf
|
||
\x{00b5}\x{039c}\x{03bc}
|
||
|
||
|
||
/\x{00c5}+/i,utf
|
||
\x{00c5}\x{00e5}\x{212b}
|
||
|
||
/\x{00e5}+/i,utf
|
||
\x{00c5}\x{00e5}\x{212b}
|
||
|
||
/\x{212b}+/i,utf
|
||
\x{00c5}\x{00e5}\x{212b}
|
||
|
||
|
||
/\x{01c4}+/i,utf
|
||
\x{01c4}\x{01c5}\x{01c6}
|
||
|
||
/\x{01c5}+/i,utf
|
||
\x{01c4}\x{01c5}\x{01c6}
|
||
|
||
/\x{01c6}+/i,utf
|
||
\x{01c4}\x{01c5}\x{01c6}
|
||
|
||
|
||
/\x{01c7}+/i,utf
|
||
\x{01c7}\x{01c8}\x{01c9}
|
||
|
||
/\x{01c8}+/i,utf
|
||
\x{01c7}\x{01c8}\x{01c9}
|
||
|
||
/\x{01c9}+/i,utf
|
||
\x{01c7}\x{01c8}\x{01c9}
|
||
|
||
|
||
/\x{01ca}+/i,utf
|
||
\x{01ca}\x{01cb}\x{01cc}
|
||
|
||
/\x{01cb}+/i,utf
|
||
\x{01ca}\x{01cb}\x{01cc}
|
||
|
||
/\x{01cc}+/i,utf
|
||
\x{01ca}\x{01cb}\x{01cc}
|
||
|
||
|
||
/\x{01f1}+/i,utf
|
||
\x{01f1}\x{01f2}\x{01f3}
|
||
|
||
/\x{01f2}+/i,utf
|
||
\x{01f1}\x{01f2}\x{01f3}
|
||
|
||
/\x{01f3}+/i,utf
|
||
\x{01f1}\x{01f2}\x{01f3}
|
||
|
||
|
||
/\x{0345}+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/\x{0399}+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/\x{03b9}+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
/\x{1fbe}+/i,utf
|
||
\x{0345}\x{0399}\x{03b9}\x{1fbe}
|
||
|
||
|
||
/\x{0392}+/i,utf
|
||
\x{0392}\x{03b2}\x{03d0}
|
||
|
||
/\x{03b2}+/i,utf
|
||
\x{0392}\x{03b2}\x{03d0}
|
||
|
||
/\x{03d0}+/i,utf
|
||
\x{0392}\x{03b2}\x{03d0}
|
||
|
||
|
||
/\x{0395}+/i,utf
|
||
\x{0395}\x{03b5}\x{03f5}
|
||
|
||
/\x{03b5}+/i,utf
|
||
\x{0395}\x{03b5}\x{03f5}
|
||
|
||
/\x{03f5}+/i,utf
|
||
\x{0395}\x{03b5}\x{03f5}
|
||
|
||
|
||
/\x{0398}+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/\x{03b8}+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/\x{03d1}+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
/\x{03f4}+/i,utf
|
||
\x{0398}\x{03b8}\x{03d1}\x{03f4}
|
||
|
||
|
||
/\x{039a}+/i,utf
|
||
\x{039a}\x{03ba}\x{03f0}
|
||
|
||
/\x{03ba}+/i,utf
|
||
\x{039a}\x{03ba}\x{03f0}
|
||
|
||
/\x{03f0}+/i,utf
|
||
\x{039a}\x{03ba}\x{03f0}
|
||
|
||
|
||
/\x{03a0}+/i,utf
|
||
\x{03a0}\x{03c0}\x{03d6}
|
||
|
||
/\x{03c0}+/i,utf
|
||
\x{03a0}\x{03c0}\x{03d6}
|
||
|
||
/\x{03d6}+/i,utf
|
||
\x{03a0}\x{03c0}\x{03d6}
|
||
|
||
|
||
/\x{03a1}+/i,utf
|
||
\x{03a1}\x{03c1}\x{03f1}
|
||
|
||
/\x{03c1}+/i,utf
|
||
\x{03a1}\x{03c1}\x{03f1}
|
||
|
||
/\x{03f1}+/i,utf
|
||
\x{03a1}\x{03c1}\x{03f1}
|
||
|
||
|
||
/\x{03a3}+/i,utf
|
||
\x{03A3}\x{03C2}\x{03C3}
|
||
|
||
/\x{03c2}+/i,utf
|
||
\x{03A3}\x{03C2}\x{03C3}
|
||
|
||
/\x{03c3}+/i,utf
|
||
\x{03A3}\x{03C2}\x{03C3}
|
||
|
||
|
||
/\x{03a6}+/i,utf
|
||
\x{03a6}\x{03c6}\x{03d5}
|
||
|
||
/\x{03c6}+/i,utf
|
||
\x{03a6}\x{03c6}\x{03d5}
|
||
|
||
/\x{03d5}+/i,utf
|
||
\x{03a6}\x{03c6}\x{03d5}
|
||
|
||
|
||
/\x{03c9}+/i,utf
|
||
\x{03c9}\x{03a9}\x{2126}
|
||
|
||
/\x{03a9}+/i,utf
|
||
\x{03c9}\x{03a9}\x{2126}
|
||
|
||
/\x{2126}+/i,utf
|
||
\x{03c9}\x{03a9}\x{2126}
|
||
|
||
|
||
/\x{1e60}+/i,utf
|
||
\x{1e60}\x{1e61}\x{1e9b}
|
||
|
||
/\x{1e61}+/i,utf
|
||
\x{1e60}\x{1e61}\x{1e9b}
|
||
|
||
/\x{1e9b}+/i,utf
|
||
\x{1e60}\x{1e61}\x{1e9b}
|
||
|
||
|
||
/\x{1e9e}+/i,utf
|
||
\x{1e9e}\x{00df}
|
||
|
||
/\x{00df}+/i,utf
|
||
\x{1e9e}\x{00df}
|
||
|
||
|
||
/\x{1f88}+/i,utf
|
||
\x{1f88}\x{1f80}
|
||
|
||
/\x{1f80}+/i,utf
|
||
\x{1f88}\x{1f80}
|
||
|
||
# Perl 5.12.4 gets these wrong, but 5.15.3 is OK
|
||
|
||
/\x{004b}+/i,utf
|
||
\x{004b}\x{006b}\x{212a}
|
||
|
||
/\x{006b}+/i,utf
|
||
\x{004b}\x{006b}\x{212a}
|
||
|
||
/\x{212a}+/i,utf
|
||
\x{004b}\x{006b}\x{212a}
|
||
|
||
|
||
/\x{0053}+/i,utf
|
||
\x{0053}\x{0073}\x{017f}
|
||
|
||
/\x{0073}+/i,utf
|
||
\x{0053}\x{0073}\x{017f}
|
||
|
||
/\x{017f}+/i,utf
|
||
\x{0053}\x{0073}\x{017f}
|
||
|
||
/^\p{Any}*\d{4}/utf
|
||
1234
|
||
\= Expect no match
|
||
123
|
||
|
||
/^\X*\w{4}/utf
|
||
1234
|
||
\= Expect no match
|
||
123
|
||
|
||
/^A\s+Z/utf,ucp
|
||
A\x{2005}Z
|
||
A\x{85}\x{2005}Z
|
||
|
||
/^A[\s]+Z/utf,ucp
|
||
A\x{2005}Z
|
||
A\x{85}\x{2005}Z
|
||
|
||
/^[[:graph:]]+$/utf,ucp
|
||
Letter:ABC
|
||
Mark:\x{300}\x{1d172}\x{1d17b}
|
||
Number:9\x{660}
|
||
Punctuation:\x{66a},;
|
||
Symbol:\x{6de}<>\x{fffc}
|
||
Cf-property:\x{ad}\x{600}\x{601}\x{602}\x{603}\x{604}\x{6dd}\x{70f}
|
||
\x{200b}\x{200c}\x{200d}\x{200e}\x{200f}
|
||
\x{202a}\x{202b}\x{202c}\x{202d}\x{202e}
|
||
\x{2060}\x{2061}\x{2062}\x{2063}\x{2064}
|
||
\x{206a}\x{206b}\x{206c}\x{206d}\x{206e}\x{206f}
|
||
\x{feff}
|
||
\x{fff9}\x{fffa}\x{fffb}
|
||
\x{110bd}
|
||
\x{1d173}\x{1d174}\x{1d175}\x{1d176}\x{1d177}\x{1d178}\x{1d179}\x{1d17a}
|
||
\x{e0001}
|
||
\x{e0020}\x{e0030}\x{e0040}\x{e0050}\x{e0060}\x{e0070}\x{e007f}
|
||
\= Expect no match
|
||
\x{09}
|
||
\x{0a}
|
||
\x{1D}
|
||
\x{20}
|
||
\x{85}
|
||
\x{a0}
|
||
\x{1680}
|
||
\x{2028}
|
||
\x{2029}
|
||
\x{202f}
|
||
\x{2065}
|
||
\x{3000}
|
||
\x{e0002}
|
||
\x{e001f}
|
||
\x{e0080}
|
||
|
||
/^[[:print:]]+$/utf,ucp
|
||
Space: \x{a0}
|
||
\x{1680}\x{2000}\x{2001}\x{2002}\x{2003}\x{2004}\x{2005}
|
||
\x{2006}\x{2007}\x{2008}\x{2009}\x{200a}
|
||
\x{202f}\x{205f}
|
||
\x{3000}
|
||
Letter:ABC
|
||
Mark:\x{300}\x{1d172}\x{1d17b}
|
||
Number:9\x{660}
|
||
Punctuation:\x{66a},;
|
||
Symbol:\x{6de}<>\x{fffc}
|
||
Cf-property:\x{ad}\x{600}\x{601}\x{602}\x{603}\x{604}\x{6dd}\x{70f}
|
||
\x{200b}\x{200c}\x{200d}\x{200e}\x{200f}
|
||
\x{202a}\x{202b}\x{202c}\x{202d}\x{202e}
|
||
\x{202f}
|
||
\x{2060}\x{2061}\x{2062}\x{2063}\x{2064}
|
||
\x{206a}\x{206b}\x{206c}\x{206d}\x{206e}\x{206f}
|
||
\x{feff}
|
||
\x{fff9}\x{fffa}\x{fffb}
|
||
\x{110bd}
|
||
\x{1d173}\x{1d174}\x{1d175}\x{1d176}\x{1d177}\x{1d178}\x{1d179}\x{1d17a}
|
||
\x{e0001}
|
||
\x{e0020}\x{e0030}\x{e0040}\x{e0050}\x{e0060}\x{e0070}\x{e007f}
|
||
\= Expect no match
|
||
\x{09}
|
||
\x{1D}
|
||
\x{85}
|
||
\x{2028}
|
||
\x{2029}
|
||
\x{2065}
|
||
\x{e0002}
|
||
\x{e001f}
|
||
\x{e0080}
|
||
|
||
/^[[:punct:]]+$/utf,ucp
|
||
\$+<=>^`|~
|
||
!\"#%&'()*,-./:;?@[\\]_{}
|
||
\x{a1}\x{a7}
|
||
\x{37e}
|
||
\= Expect no match
|
||
abcde
|
||
|
||
/^[[:^graph:]]+$/utf,ucp
|
||
\x{09}\x{0a}\x{1D}\x{20}\x{85}\x{a0}\x{1680}
|
||
\x{2028}\x{2029}\x{202f}\x{2065}
|
||
\x{3000}\x{e0002}\x{e001f}\x{e0080}
|
||
\= Expect no match
|
||
Letter:ABC
|
||
Mark:\x{300}\x{1d172}\x{1d17b}
|
||
Number:9\x{660}
|
||
Punctuation:\x{66a},;
|
||
Symbol:\x{6de}<>\x{fffc}
|
||
Cf-property:\x{ad}\x{600}\x{601}\x{602}\x{603}\x{604}\x{6dd}\x{70f}
|
||
\x{200b}\x{200c}\x{200d}\x{200e}\x{200f}
|
||
\x{202a}\x{202b}\x{202c}\x{202d}\x{202e}
|
||
\x{2060}\x{2061}\x{2062}\x{2063}\x{2064}
|
||
\x{206a}\x{206b}\x{206c}\x{206d}\x{206e}\x{206f}
|
||
\x{feff}
|
||
\x{fff9}\x{fffa}\x{fffb}
|
||
\x{110bd}
|
||
\x{1d173}\x{1d174}\x{1d175}\x{1d176}\x{1d177}\x{1d178}\x{1d179}\x{1d17a}
|
||
\x{e0001}
|
||
\x{e0020}\x{e0030}\x{e0040}\x{e0050}\x{e0060}\x{e0070}\x{e007f}
|
||
|
||
/^[[:^print:]]+$/utf,ucp
|
||
\x{09}\x{1D}\x{85}\x{2028}\x{2029}\x{2065}
|
||
\x{e0002}\x{e001f}\x{e0080}
|
||
\= Expect no match
|
||
Space: \x{a0}
|
||
\x{1680}\x{2000}\x{2001}\x{2002}\x{2003}\x{2004}\x{2005}
|
||
\x{2006}\x{2007}\x{2008}\x{2009}\x{200a}
|
||
\x{202f}\x{205f}
|
||
\x{3000}
|
||
Letter:ABC
|
||
Mark:\x{300}\x{1d172}\x{1d17b}
|
||
Number:9\x{660}
|
||
Punctuation:\x{66a},;
|
||
Symbol:\x{6de}<>\x{fffc}
|
||
Cf-property:\x{ad}\x{600}\x{601}\x{602}\x{603}\x{604}\x{6dd}\x{70f}
|
||
\x{200b}\x{200c}\x{200d}\x{200e}\x{200f}
|
||
\x{202a}\x{202b}\x{202c}\x{202d}\x{202e}
|
||
\x{202f}
|
||
\x{2060}\x{2061}\x{2062}\x{2063}\x{2064}
|
||
\x{206a}\x{206b}\x{206c}\x{206d}\x{206e}\x{206f}
|
||
\x{feff}
|
||
\x{fff9}\x{fffa}\x{fffb}
|
||
\x{110bd}
|
||
\x{1d173}\x{1d174}\x{1d175}\x{1d176}\x{1d177}\x{1d178}\x{1d179}\x{1d17a}
|
||
\x{e0001}
|
||
\x{e0020}\x{e0030}\x{e0040}\x{e0050}\x{e0060}\x{e0070}\x{e007f}
|
||
|
||
/^[[:^punct:]]+$/utf,ucp
|
||
abcde
|
||
\= Expect no match
|
||
\$+<=>^`|~
|
||
!\"#%&'()*,-./:;?@[\\]_{}
|
||
\x{a1}\x{a7}
|
||
\x{37e}
|
||
|
||
/[RST]+/i,utf,ucp
|
||
Ss\x{17f}
|
||
|
||
/[R-T]+/i,utf,ucp
|
||
Ss\x{17f}
|
||
|
||
/[q-u]+/i,utf,ucp
|
||
Ss\x{17f}
|
||
|
||
/^s?c/im,utf
|
||
scat
|
||
|
||
# The next four tests are for repeated caseless back references when the
|
||
# code unit length of the matched text is different to that of the original
|
||
# group in the UTF-8 case.
|
||
|
||
/^(\x{23a})\1*(.)/i,utf
|
||
\x{23a}\x{23a}\x{23a}\x{23a}
|
||
\x{23a}\x{2c65}\x{2c65}\x{2c65}
|
||
\x{23a}\x{23a}\x{2c65}\x{23a}
|
||
|
||
/^(\x{23a})\1*(..)/i,utf
|
||
\x{23a}\x{2c65}\x{2c65}\x{2c65}
|
||
\x{23a}\x{23a}\x{2c65}\x{23a}
|
||
|
||
/^(\x{23a})\1*(...)/i,utf
|
||
\x{23a}\x{2c65}\x{2c65}\x{2c65}
|
||
\x{23a}\x{23a}\x{2c65}\x{23a}
|
||
|
||
/^(\x{23a})\1*(....)/i,utf
|
||
\= Expect no match
|
||
\x{23a}\x{2c65}\x{2c65}\x{2c65}
|
||
\x{23a}\x{23a}\x{2c65}\x{23a}
|
||
|
||
/[A-`]/i,utf
|
||
abcdefghijklmno
|
||
|
||
/[\S\V\H]/utf
|
||
|
||
/[^\p{Any}]*+x/utf
|
||
x
|
||
|
||
/[[:punct:]]/utf,ucp
|
||
\x{b4}
|
||
|
||
/[[:^ascii:]]/utf,ucp
|
||
\x{100}
|
||
\x{200}
|
||
\x{300}
|
||
\x{37e}
|
||
\= Expect no match
|
||
aa
|
||
99
|
||
|
||
/[[:^ascii:]\w]/utf,ucp
|
||
aa
|
||
99
|
||
gg
|
||
\x{100}
|
||
\x{200}
|
||
\x{300}
|
||
\x{37e}
|
||
|
||
/[\w[:^ascii:]]/utf,ucp
|
||
aa
|
||
99
|
||
gg
|
||
\x{100}
|
||
\x{200}
|
||
\x{300}
|
||
\x{37e}
|
||
|
||
/[^[:ascii:]\W]/utf,ucp
|
||
\x{100}
|
||
\x{200}
|
||
\= Expect no match
|
||
aa
|
||
99
|
||
gg
|
||
\x{37e}
|
||
|
||
/[^[:^ascii:]\d]/utf,ucp
|
||
a
|
||
~
|
||
\a
|
||
\x{7f}
|
||
\= Expect no match
|
||
0
|
||
\x{389}
|
||
\x{20ac}
|
||
|
||
/(?=.*b)\pL/
|
||
11bb
|
||
|
||
/(?(?=.*b)(?=.*b)\pL|.*c)/
|
||
11bb
|
||
|
||
/^\x{123}+?$/utf,no_auto_possess
|
||
\x{123}\x{123}\x{123}
|
||
|
||
/^\x{123}+?$/i,utf,no_auto_possess
|
||
\x{123}\x{122}\x{123}
|
||
\= Expect no match
|
||
\x{123}\x{124}\x{123}
|
||
|
||
/\N{U+1234}/utf
|
||
\x{1234}
|
||
|
||
/[\N{U+1234}]/utf
|
||
\x{1234}
|
||
|
||
# Test the full list of Unicode "Pattern White Space" characters that are to
|
||
# be ignored by /x. The pattern lines below may show up oddly in text editors
|
||
# or when listed to the screen. Note that characters such as U+2002, which are
|
||
# matched as space by \h and \v are *not* "Pattern White Space".
|
||
|
||
/A
B/x,utf
|
||
AB
|
||
|
||
/A B/x,utf
|
||
A\x{2002}B
|
||
\= Expect no match
|
||
AB
|
||
|
||
# -------
|
||
|
||
/[^\x{100}-\x{ffff}]*[\x80-\xff]/utf
|
||
\x{99}\x{99}\x{99}
|
||
|
||
/[^\x{100}-\x{ffff}ABC]*[\x80-\xff]/utf
|
||
\x{99}\x{99}\x{99}
|
||
|
||
/[^\x{100}-\x{ffff}]*[\x80-\xff]/i,utf
|
||
\x{99}\x{99}\x{99}
|
||
|
||
# Script run tests
|
||
|
||
/^(*script_run:.{4})/utf
|
||
abcd Latin x4
|
||
\x{2e80}\x{2fa1d}\x{3041}\x{30a1} Han Han Hiragana Katakana
|
||
\x{3041}\x{30a1}\x{3007}\x{3007} Hiragana Katakana Han Han
|
||
\x{30a1}\x{3041}\x{3007}\x{3007} Katakana Hiragana Han Han
|
||
\x{1100}\x{2e80}\x{2e80}\x{1101} Hangul Han Han Hangul
|
||
\x{2e80}\x{3105}\x{2e80}\x{3105} Han Bopomofo Han Bopomofo
|
||
\x{02ea}\x{2e80}\x{2e80}\x{3105} Bopomofo-Sk Han Han Bopomofo
|
||
\x{3105}\x{2e80}\x{2e80}\x{3105} Bopomofo Han Han Bopomofo
|
||
\x{0300}cd! Inherited Latin Latin Common
|
||
\x{0391}12\x{03a9} Greek Common-digits Greek
|
||
\x{0400}12\x{fe2f} Cyrillic Common-digits Cyrillic
|
||
\x{0531}12\x{fb17} Armenian Common-digits Armenian
|
||
\x{0591}12\x{fb4f} Hebrew Common-digits Hebrew
|
||
\x{0600}12\x{1eef1} Arabic Common-digits Arabic
|
||
\x{0600}\x{0660}\x{0669}\x{1eef1} Arabic Arabic-digits Arabic
|
||
\x{0700}12\x{086a} Syriac Common-digits Syriac
|
||
\x{1200}12\x{ab2e} Ethiopic Common-digits Ethiopic
|
||
\x{1680}12\x{169c} Ogham Common-digits Ogham
|
||
\x{3041}12\x{3041} Hiragana Common-digits Hiragana
|
||
\x{0980}\x{09e6}\x{09e7}\x{0993} Bengali Bengali-digits Bengali
|
||
!cde Common Latin Latin Latin
|
||
A..B Latin Common Common Latin
|
||
0abc Ascii-digit Latin Latin Latin
|
||
1\x{0700}\x{0700}\x{0700} Ascii-digit Syriac x 3
|
||
\x{1A80}\x{1A80}\x{1a40}\x{1a41} Tai Tham Hora digits, letters
|
||
\= Expect no match
|
||
a\x{370}bcd Latin Greek Latin Latin
|
||
\x{1100}\x{02ea}\x{02ea}\x{02ea} Hangul Bopomofo x3
|
||
\x{02ea}\x{02ea}\x{02ea}\x{1100} Bopomofo x3 Hangul
|
||
\x{1100}\x{2e80}\x{3041}\x{1101} Hangul Han Hiragana Hangul
|
||
\x{0391}\x{09e6}\x{09e7}\x{03a9} Greek Bengali digits Greek
|
||
\x{0600}7\x{0669}\x{1eef1} Arabic ascii-digit Arabic-digit Arabic
|
||
\x{0600}\x{0669}7\x{1eef1} Arabic Arabic-digit ascii-digit Arabic
|
||
A5\x{ff19}B Latin Common-ascii/notascii-digits Latin
|
||
\x{0300}cd\x{0391} Inherited Latin Latin Greek
|
||
!cd\x{0391} Common Latin Latin Greek
|
||
\x{1A80}\x{1A90}\x{1a40}\x{1a41} Tai Tham Hora digit, Tham digit, letters
|
||
A\x{1d7ce}\x{1d7ff}B Common fancy-common-2-sets-digits Common
|
||
\x{2e80}\x{3105}\x{2e80}\x{30a1} Han Bopomofo Han Katakana
|
||
|
||
/^(*sr:.{4}|..)/utf
|
||
\x{2e80}\x{3105}\x{2e80}\x{30a1} Han Bopomofo Han Katakana
|
||
|
||
/^(*atomic_script_run:.{4}|..)/utf
|
||
\= Expect no match
|
||
\x{2e80}\x{3105}\x{2e80}\x{30a1} Han Bopomofo Han Katakana
|
||
|
||
/^(*asr:.*)/utf
|
||
\= Expect no match
|
||
\x{2e80}\x{3105}\x{2e80}\x{30a1} Han Bopomofo Han Katakana
|
||
|
||
/^(?>(*sr:.*))/utf
|
||
\x{2e80}\x{3105}\x{2e80}\x{30a1} Han Bopomofo Han Katakana
|
||
|
||
/^(*sr:.*)/utf
|
||
\x{2e80}\x{3105}\x{2e80}\x{30a1} Han Bopomofo Han Katakana
|
||
\x{10fffd}\x{10fffd}\x{10fffd} Private use (Unknown)
|
||
|
||
/^(*sr:\x{2e80}*)/utf
|
||
\x{2e80}\x{2e80}\x{3105} Han Han Bopomofo
|
||
|
||
/^(*sr:\x{2e80}*)\x{2e80}/utf
|
||
\x{2e80}\x{2e80}\x{3105} Han Han Bopomofo
|
||
|
||
/^(*sr:.*)Test/utf
|
||
Test script run on an empty string
|
||
|
||
/^(*sr:(.{2})){2}/utf
|
||
\x{0600}7\x{0669}\x{1eef1} Arabic ascii-digit Arabic-digit Arabic
|
||
\x{1A80}\x{1A80}\x{1a40}\x{1a41} Tai Tham Hora digits, letters
|
||
\x{1A80}\x{1a40}\x{1A90}\x{1a41} Tai Tham Hora digit, letter, Tham digit, letter
|
||
\= Expect no match
|
||
\x{1100}\x{2e80}\x{3041}\x{1101} Hangul Han Hiragana Hangul
|
||
|
||
/^(*sr:\S*)/utf
|
||
\x{1cf4}\x{20f0}\x{900}\x{11305} [Dev,Gran,Kan] [Dev,Gran,Lat] Dev Gran
|
||
\x{1cf4}\x{20f0}\x{11305}\x{900} [Dev,Gran,Kan] [Dev,Gran,Lat] Gran Dev
|
||
\x{1cf4}\x{20f0}\x{900}ABC [Dev,Gran,Kan] [Dev,Gran,Lat] Dev Lat
|
||
\x{1cf4}\x{20f0}ABC [Dev,Gran,Kan] [Dev,Gran,Lat] Lat
|
||
\x{20f0}ABC [Dev,Gran,Lat] Lat
|
||
XYZ\x{20f0}ABC Lat [Dev,Gran,Lat] Lat
|
||
\x{a36}\x{a33}\x{900} [Dev,...] [Dev,...] Dev
|
||
\x{3001}\x{2e80}\x{3041}\x{30a1} [Bopo, Han, etc] Han Hira Kata
|
||
\x{3001}\x{30a1}\x{2e80}\x{3041} [Bopo, Han, etc] Kata Han Hira
|
||
\x{3001}\x{3105}\x{2e80}\x{1101} [Bopo, Han, etc] Bopomofo Han Hangul
|
||
\x{3105}\x{3001}\x{2e80}\x{1101} Bopomofo [Bopo, Han, etc] Han Hangul
|
||
\x{3031}\x{3041}\x{30a1}\x{2e80} [Hira Kata] Hira Kata Han
|
||
\x{060c}\x{06d4}\x{0600}\x{10d00}\x{0700} [Arab Rohg Syrc Thaa] [Arab Rohg] Arab Rohg Syrc
|
||
\x{060c}\x{06d4}\x{0700}\x{0600}\x{10d00} [Arab Rohg Syrc Thaa] [Arab Rohg] Syrc Arab Rohg
|
||
\x{2e80}\x{3041}\x{3001}\x{3031}\x{2e80} Han Hira [Bopo, Han, etc] [Hira Kata] Han
|
||
|
||
/(?<!)(*sr:)/
|
||
|
||
/(?<!X(*sr:B)C)/
|
||
|
||
/(?<=abc(?=X(*sr:BCY)Z)XBCYZ)./
|
||
abcXBCYZ!
|
||
|
||
/(?<=abc(?=X(*sr:BXY)CCC)XBXYCCC)./
|
||
abcXBXYCCC!
|
||
|
||
/^(*sr:\S*)/utf
|
||
\x{10d00}\x{10d00}\x{06d4} Rohingya Rohingya Arabic-full-stop
|
||
\x{06d4}\x{10d00}\x{10d00} Arabic-full-stop Rohingya Rohingya
|
||
\x{10d00}\x{10d00}\x{0363} Rohingya Rohingya Inherited-extend-Latin
|
||
\x{0363}\x{10d00}\x{10d00} Inherited-extend-Latin Rohingya Rohingya
|
||
AB\x{0363} Latin Latin Inherited-extend-Latin
|
||
\x{0363}AB Inherited-extend-Latin Latin Latin
|
||
AB\x{1cf7} Latin Latin Common-extended-Beng
|
||
\x{1cf7}AB Common-extend-Beng Latin Latin
|
||
\x{1cf7}\x{0993} Common-extend-Beng Bengali
|
||
A\x{1abe}BC Test enclosing mark
|
||
\x{0370}\x{1abe}\x{0371} Which can occur with any script (Greek here)
|
||
\x{3001}\x{adf9}\x{3001} [.. Hangul ..] Hangul [.. Hangul ..]
|
||
\x{3400}\x{3001}XXX Han [Han etc.]
|
||
\x{3400}\x{1cd5} Han [Bengali Devanagari]
|
||
\x{ac01}\x{3400} Hangul [.. Hangul ..]
|
||
\x{ac01}\x{1cd5} Hangul [Bengali Devanagari]
|
||
\x{102e0}\x{06d4}\x{1ee4d} [Arabic Coptic] [Arab Rohingya] Arabic
|
||
\x{102e0}\x{06d4}\x{2cc9} [Arabic Coptic] [Arab Rohingya] Coptic
|
||
\x{102e0}\x{06d4}\x{10d30} [Arabic Coptic] [Arab Rohingya] Rohingya
|
||
|
||
# Test loop breaking for empty string match
|
||
|
||
/^(*sr:A|)*BCD/utf
|
||
AABCD
|
||
ABCD
|
||
BCD
|
||
|
||
# The use of (*ACCEPT) breaks script run checking
|
||
|
||
/^(*sr:.*(*ACCEPT)ZZ)/utf
|
||
\x{1100}\x{2e80}\x{3041}\x{1101} Hangul Han Hiragana Hangul
|
||
|
||
# -------
|
||
|
||
# Test group names containing non-ASCII letters and digits
|
||
|
||
/(?'ABáC'...)\g{ABáC}/utf
|
||
abcabcdefg
|
||
|
||
/(?'XʰABC'...)/utf
|
||
xyzpq
|
||
|
||
/(?'XאABC'...)/utf
|
||
12345
|
||
|
||
/(?'XᾈABC'...)/utf
|
||
%^&*(...
|
||
|
||
/(?'𐨐ABC'...)/utf
|
||
abcde
|
||
|
||
/^(?'אABC'...)(?&אABC)(?P=אABC)/utf
|
||
123123123456
|
||
|
||
/^(?'אABC'...)(?&אABC)/utf
|
||
123123123456
|
||
|
||
/\X*/
|
||
\xF3aaa\xE4\xEA\xEB\xFEa
|
||
|
||
/Я/i,utf
|
||
\x{42f}
|
||
\x{44f}
|
||
|
||
/(?=Я)/i,utf
|
||
\x{42f}
|
||
\x{44f}
|
||
|
||
# -----------------------------------------------------------------------------
|
||
# Tests for bidi control and bidi class properties, not yet supported by JIT.
|
||
|
||
/\p{ bidi_control }/utf
|
||
-->\x{202c}<--
|
||
|
||
/\p{bidicontrol}+/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/\p{bidic}+?/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/\p{bidi_control}++/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/[\p{bidi_c}]/utf
|
||
-->\x{202c}<--
|
||
|
||
/[\p{bidicontrol}]+/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/[\p{bidicontrol}]+?/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/[\p{bidicontrol}]++/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/[\p{bidicontrol}<>]+/utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/\P{bidicontrol}+/g,utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/\p{^bidicontrol}+/g,utf
|
||
-->\x{061c}\x{200e}\x{200f}\x{202a}\x{202b}\x{202c}\x{202d}<--
|
||
-->\x{2066}\x{2067}\x{2068}\x{2069}<--
|
||
|
||
/\p{bidi class = al}/utf
|
||
-->\x{061D}<--
|
||
|
||
/\p{bc = al}+/utf
|
||
-->\x{061D}\x{061e}\x{061f}<--
|
||
|
||
/\p{bidi_class : AL}+?/utf
|
||
-->\x{061D}\x{061e}\x{061f}<--
|
||
|
||
/\p{Bidi_Class : AL}++/utf
|
||
-->\x{061D}\x{061e}\x{061f}<--
|
||
|
||
/\p{b_c = aN}+/utf
|
||
-->\x{061D}\x{0602}\x{0604}\x{061f}<--
|
||
|
||
/\p{bidi class = B}+/utf
|
||
-->\x{0a}\x{0d}\x{01c}\x{01e}\x{085}\x{2029}<--
|
||
|
||
/\p{bidi class:BN}+/utf
|
||
-->\x{0}\x{08}\x{200c}\x{fffe}\x{dfffe}\x{10ffff}<--
|
||
|
||
/\p{bidiclass:cs}+/utf
|
||
-->,.\x{060c}\x{ff1a}<--
|
||
|
||
/\p{bidiclass:En}+/utf
|
||
-->09\x{b2}\x{2074}\x{1fbf9}<--
|
||
|
||
/\p{bidiclass:es}+/utf
|
||
==>+-\x{207a}\x{ff0d}<==
|
||
|
||
/\p{bidiclass:et}+/utf
|
||
-->#\{24}%\x{a2}\x{A838}\x{1e2ff}<--
|
||
|
||
/\p{bidiclass:FSI}+/utf
|
||
-->\x{2068}<--
|
||
|
||
/\p{bidi class:L}+/utf
|
||
-->ABC<--
|
||
|
||
/\P{bidi class:L}+/utf
|
||
-->ABC<--
|
||
|
||
/\p{bidi class:LRE}+\p{bidiclass=lri}*\p{bidiclass:lro}/utf
|
||
-->\x{202a}\x{2066}\x{202d}<--
|
||
|
||
/\p{bidi class:NSM}+/utf
|
||
-->\x{9bc}\x{a71}\x{e31}<--
|
||
|
||
/\p{bidi class:ON}+/utf
|
||
-->\x{21}'()*;@\x{384}\x{2039}<=-
|
||
|
||
/\p{bidiclass:pdf}\p{bidiclass:pdi}/utf
|
||
-->\x{202c}\x{2069}<--
|
||
|
||
/\p{bidi class:R}+/utf
|
||
-->\x{590}\x{5c6}\x{200f}\x{10805}<--
|
||
|
||
/\p{bidi class:RLE}+\p{bidi class:RLI}*\p{bidi class:RLO}+/utf
|
||
-->\x{202b}\x{2067}\x{202e}<--
|
||
|
||
/\p{bidi class:S}+\p{bidiclass:WS}+/utf
|
||
-->\x{9}\x{b}\x{1f} \x{c} \x{2000} \x{3000}<--
|
||
|
||
# -----------------------------------------------------------------------------
|
||
|
||
# End of testinput4
|