/[pcre]/code/branches/pcre16/testdata/testinput16
ViewVC logotype

Diff of /code/branches/pcre16/testdata/testinput16

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 810 by ph10, Mon Dec 19 13:34:10 2011 UTC revision 814 by ph10, Wed Dec 21 12:05:24 2011 UTC
# Line 1  Line 1 
1  /-- This set of tests is for UTF-16 support, and is relevant only to the 16-bit  /-- This set of tests is run only with the 8-bit library when Unicode property
2      library. There are some non-UTF 16-bit tests as well (it doesn't seem      support is available. It starts with tests of the POSIX interface, because
3      worth setting up another test file just for this case). --/      that is supported only with the 8-bit library. --/
4    
5  /xxx/8?DZSS  /\w/P
6        +++\x{c2}
7    
8  /abc/8  /\w/WP
9      ]      +++\x{c2}
   
 /X(\C{3})/8  
     X\x{11234}Y  
   
 /X(\C{4})/8  
     X\x{11234}YZ  
   
 /X\C*/8  
     XYZabcdce  
   
 /X\C*?/8  
     XYZabcde  
   
 /X\C{3,5}/8  
     Xabcdefg  
     X\x{11234}Y  
     X\x{11234}YZ  
     X\x{11234}\x{512}  
     X\x{11234}\x{512}YZ  
     X\x{11234}\x{512}\x{11234}Z  
   
 /X\C{3,5}?/8  
     Xabcdefg  
     X\x{11234}Y  
     X\x{11234}YZ  
     X\x{11234}\x{512}YZ  
     *** Failers  
     X\x{11234}  
   
 /a\Cb/  
     aXb  
     a\nb  
   
 /a\Cb/8  
     aXb  
     a\nb  
   
 /a\C\Cb/8  
     a\x{12257}b  
     ** Failers  
     a\x{100}b  
   
 /ab\Cde/8  
     abXde  
   
 /-- Check maximum non-UTF character size --/  
   
 /\x{ffff}/  
   
 /\x{10000}/  
   
 /\x{100}/8DZ  
   
 /\x{1000}/8DZ  
   
 /\x{10000}/8DZ  
   
 /\x{100000}/8DZ  
   
 /\x{10ffff}/8DZ  
   
 /[\x{ff}]/8DZ  
   
 /[\x{100}]/8DZ  
   
 /\x80/8DZ  
   
 /\xff/8DZ  
   
 /\x{D55c}\x{ad6d}\x{C5B4}/DZ8  
     \x{D55c}\x{ad6d}\x{C5B4}  
   
 /\x{65e5}\x{672c}\x{8a9e}/DZ8  
     \x{65e5}\x{672c}\x{8a9e}  
   
 /\x{80}/DZ8  
   
 /\x{084}/DZ8  
   
 /\x{104}/DZ8  
   
 /\x{861}/DZ8  
   
 /\x{212ab}/DZ8  
   
 /-- 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.) --/  
10    
11  /X(\C)(.*)/8  /A\x{391}\x{10427}\x{ff3a}\x{1fb0}/8iDZ
     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/8  
     *** Failers  
     a\x{100}b  
   
 /[^ab\xC0-\xF0]/8SDZ  
     \x{f1}  
     \x{bf}  
     \x{100}  
     \x{1000}  
     *** Failers  
     \x{c0}  
     \x{f0}  
   
 /Ā{3,4}/8SDZ  
   \x{100}\x{100}\x{100}\x{100\x{100}  
   
 /(\x{100}+|x)/8SDZ  
   
 /(\x{100}*a|x)/8SDZ  
   
 /(\x{100}{0,2}a|x)/8SDZ  
   
 /(\x{100}{1,2}a|x)/8SDZ  
   
 /\x{100}/8DZ  
   
 /a\x{100}\x{101}*/8DZ  
   
 /a\x{100}\x{101}+/8DZ  
   
 /[^\x{c4}]/DZ  
   
 /[\x{100}]/8DZ  
     \x{100}  
     Z\x{100}  
     \x{100}Z  
     *** Failers  
   
 /[\xff]/DZ8  
     >\x{ff}<  
   
 /[^\xff]/8DZ  
   
 /\x{100}abc(xyz(?1))/8DZ  
   
 /\777/8I  
   \x{1ff}  
   \777  
   
 /\x{100}+\x{200}/8DZ  
   
 /\x{100}+X/8DZ  
   
 /^[\QĀ\E-\QŐ\E/BZ8  
   
 /-- This tests the stricter UTF-8 check according to RFC 3629. --/  
   
 /X/8  
     \x{0}\x{d7ff}\x{e000}\x{10ffff}  
     \x{d800}  
     \x{d800}\?  
     \x{da00}  
     \x{da00}\?  
     \x{dfff}  
     \x{dfff}\?  
     \x{110000}  
     \x{110000}\?  
     \x{2000000}  
     \x{2000000}\?  
     \x{7fffffff}  
     \x{7fffffff}\?  
   
 /(*UTF16)\x{11234}/  
   abcd\x{11234}pqr  
   
 /(*CRLF)(*UTF16)(*BSR_UNICODE)a\Rb/I  
   
 /\h/SI8  
     ABC\x{09}  
     ABC\x{20}  
     ABC\x{a0}  
     ABC\x{1680}  
     ABC\x{180e}  
     ABC\x{2000}  
     ABC\x{202f}  
     ABC\x{205f}  
     ABC\x{3000}  
   
 /\v/SI8  
     ABC\x{0a}  
     ABC\x{0b}  
     ABC\x{0c}  
     ABC\x{0d}  
     ABC\x{85}  
     ABC\x{2028}  
   
 /\h*A/SI8  
     CDBABC  
   
 /\v+A/SI8  
   
 /\s?xxx\s/8SI  
   
 /\sxxx\s/I8ST1  
     AB\x{85}xxx\x{a0}XYZ  
     AB\x{a0}xxx\x{85}XYZ  
   
 /\S \S/I8ST1  
     \x{a2} \x{84}  
     A Z  
   
 /a+/8  
     a\x{123}aa\>1  
     a\x{123}aa\>2  
     a\x{123}aa\>3  
     a\x{123}aa\>4  
     a\x{123}aa\>5  
     a\x{123}aa\>6  
   
 /\x{1234}+/iS8I  
12    
13  /\x{1234}+?/iS8I  /A\x{391}\x{10427}\x{ff3a}\x{1fb0}/8DZ
14    
15  /\x{1234}++/iS8I  /AB\x{1fb0}/8DZ
16    
17  /\x{1234}{2}/iS8I  /AB\x{1fb0}/8DZi
18    
19  /[^\x{c4}]/8DZ  /\x{401}\x{420}\x{421}\x{422}\x{423}\x{424}\x{425}\x{426}\x{427}\x{428}\x{429}\x{42a}\x{42b}\x{42c}\x{42d}\x{42e}\x{42f}/8iSI
20        \x{401}\x{420}\x{421}\x{422}\x{423}\x{424}\x{425}\x{426}\x{427}\x{428}\x{429}\x{42a}\x{42b}\x{42c}\x{42d}\x{42e}\x{42f}
21        \x{451}\x{440}\x{441}\x{442}\x{443}\x{444}\x{445}\x{446}\x{447}\x{448}\x{449}\x{44a}\x{44b}\x{44c}\x{44d}\x{44e}\x{44f}
22    
23  /X+\x{200}/8DZ  /[ⱥ]/8iBZ
24    
25  /\R/SI8  /[^ⱥ]/8iBZ
26    
27  /-- End of testinput16 --/  /-- End of testinput16 --/

Legend:
Removed from v.810  
changed lines
  Added in v.814

  ViewVC Help
Powered by ViewVC 1.1.5