--- embedaddon/pcre/testdata/testinput10 2012/10/09 09:19:18 1.1.1.3 +++ embedaddon/pcre/testdata/testinput10 2013/07/22 08:25:57 1.1.1.4 @@ -1026,4 +1026,376 @@ AA\P AA\P\P +/-- These are tests for extended grapheme clusters --/ + +/^\X/8+ + 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/8+ + \x{1111}\x{ae4c}\x{1111}\x{ae4c}X + +/^\X{2,4}X/8+ + \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/8+ + \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{1e9e}+/8i + \x{1e9e}\x{00df} + +/[z\x{1e9e}]+/8i + \x{1e9e}\x{00df} + +/\x{00df}+/8i + \x{1e9e}\x{00df} + +/[z\x{00df}]+/8i + \x{1e9e}\x{00df} + +/\x{1f88}+/8i + \x{1f88}\x{1f80} + +/[z\x{1f88}]+/8i + \x{1f88}\x{1f80} + +/-- Perl matches these --/ + +/\x{00b5}+/8i + \x{00b5}\x{039c}\x{03bc} + +/\x{039c}+/8i + \x{00b5}\x{039c}\x{03bc} + +/\x{03bc}+/8i + \x{00b5}\x{039c}\x{03bc} + + +/\x{00c5}+/8i + \x{00c5}\x{00e5}\x{212b} + +/\x{00e5}+/8i + \x{00c5}\x{00e5}\x{212b} + +/\x{212b}+/8i + \x{00c5}\x{00e5}\x{212b} + + +/\x{01c4}+/8i + \x{01c4}\x{01c5}\x{01c6} + +/\x{01c5}+/8i + \x{01c4}\x{01c5}\x{01c6} + +/\x{01c6}+/8i + \x{01c4}\x{01c5}\x{01c6} + + +/\x{01c7}+/8i + \x{01c7}\x{01c8}\x{01c9} + +/\x{01c8}+/8i + \x{01c7}\x{01c8}\x{01c9} + +/\x{01c9}+/8i + \x{01c7}\x{01c8}\x{01c9} + + +/\x{01ca}+/8i + \x{01ca}\x{01cb}\x{01cc} + +/\x{01cb}+/8i + \x{01ca}\x{01cb}\x{01cc} + +/\x{01cc}+/8i + \x{01ca}\x{01cb}\x{01cc} + + +/\x{01f1}+/8i + \x{01f1}\x{01f2}\x{01f3} + +/\x{01f2}+/8i + \x{01f1}\x{01f2}\x{01f3} + +/\x{01f3}+/8i + \x{01f1}\x{01f2}\x{01f3} + + +/\x{0345}+/8i + \x{0345}\x{0399}\x{03b9}\x{1fbe} + +/\x{0399}+/8i + \x{0345}\x{0399}\x{03b9}\x{1fbe} + +/\x{03b9}+/8i + \x{0345}\x{0399}\x{03b9}\x{1fbe} + +/\x{1fbe}+/8i + \x{0345}\x{0399}\x{03b9}\x{1fbe} + + +/\x{0392}+/8i + \x{0392}\x{03b2}\x{03d0} + +/\x{03b2}+/8i + \x{0392}\x{03b2}\x{03d0} + +/\x{03d0}+/8i + \x{0392}\x{03b2}\x{03d0} + + +/\x{0395}+/8i + \x{0395}\x{03b5}\x{03f5} + +/\x{03b5}+/8i + \x{0395}\x{03b5}\x{03f5} + +/\x{03f5}+/8i + \x{0395}\x{03b5}\x{03f5} + + +/\x{0398}+/8i + \x{0398}\x{03b8}\x{03d1}\x{03f4} + +/\x{03b8}+/8i + \x{0398}\x{03b8}\x{03d1}\x{03f4} + +/\x{03d1}+/8i + \x{0398}\x{03b8}\x{03d1}\x{03f4} + +/\x{03f4}+/8i + \x{0398}\x{03b8}\x{03d1}\x{03f4} + + +/\x{039a}+/8i + \x{039a}\x{03ba}\x{03f0} + +/\x{03ba}+/8i + \x{039a}\x{03ba}\x{03f0} + +/\x{03f0}+/8i + \x{039a}\x{03ba}\x{03f0} + + +/\x{03a0}+/8i + \x{03a0}\x{03c0}\x{03d6} + +/\x{03c0}+/8i + \x{03a0}\x{03c0}\x{03d6} + +/\x{03d6}+/8i + \x{03a0}\x{03c0}\x{03d6} + + +/\x{03a1}+/8i + \x{03a1}\x{03c1}\x{03f1} + +/\x{03c1}+/8i + \x{03a1}\x{03c1}\x{03f1} + +/\x{03f1}+/8i + \x{03a1}\x{03c1}\x{03f1} + + +/\x{03a3}+/8i + \x{03A3}\x{03C2}\x{03C3} + +/\x{03c2}+/8i + \x{03A3}\x{03C2}\x{03C3} + +/\x{03c3}+/8i + \x{03A3}\x{03C2}\x{03C3} + + +/\x{03a6}+/8i + \x{03a6}\x{03c6}\x{03d5} + +/\x{03c6}+/8i + \x{03a6}\x{03c6}\x{03d5} + +/\x{03d5}+/8i + \x{03a6}\x{03c6}\x{03d5} + + +/\x{03c9}+/8i + \x{03c9}\x{03a9}\x{2126} + +/\x{03a9}+/8i + \x{03c9}\x{03a9}\x{2126} + +/\x{2126}+/8i + \x{03c9}\x{03a9}\x{2126} + + +/\x{1e60}+/8i + \x{1e60}\x{1e61}\x{1e9b} + +/\x{1e61}+/8i + \x{1e60}\x{1e61}\x{1e9b} + +/\x{1e9b}+/8i + \x{1e60}\x{1e61}\x{1e9b} + + +/\x{1e9e}+/8i + \x{1e9e}\x{00df} + +/\x{00df}+/8i + \x{1e9e}\x{00df} + + +/\x{1f88}+/8i + \x{1f88}\x{1f80} + +/\x{1f80}+/8i + \x{1f88}\x{1f80} + +/\x{004b}+/8i + \x{004b}\x{006b}\x{212a} + +/\x{006b}+/8i + \x{004b}\x{006b}\x{212a} + +/\x{212a}+/8i + \x{004b}\x{006b}\x{212a} + + +/\x{0053}+/8i + \x{0053}\x{0073}\x{017f} + +/\x{0073}+/8i + \x{0053}\x{0073}\x{017f} + +/\x{017f}+/8i + \x{0053}\x{0073}\x{017f} + +/ist/8i + ikt + +/is+t/8i + iSs\x{17f}t + ikt + +/is+?t/8i + ikt + +/is?t/8i + ikt + +/is{2}t/8i + iskt + +/^\p{Xuc}/8 + $abc + @abc + `abc + \x{1234}abc + ** Failers + abc + +/^\p{Xuc}+/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^\p{Xuc}+?/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^\p{Xuc}+?\*/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^\p{Xuc}++/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^\p{Xuc}{3,5}/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^\p{Xuc}{3,5}?/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^[\p{Xuc}]/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^[\p{Xuc}]+/8 + $@`\x{a0}\x{1234}\x{e000}** + ** Failers + \x{9f} + +/^\P{Xuc}/8 + abc + ** Failers + $abc + @abc + `abc + \x{1234}abc + +/^[\P{Xuc}]/8 + abc + ** Failers + $abc + @abc + `abc + \x{1234}abc + /-- End of testinput10 --/