/[pcre]/code/trunk/testdata/testinput5
ViewVC logotype

Diff of /code/trunk/testdata/testinput5

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

revision 428 by ph10, Mon Aug 31 17:10:26 2009 UTC revision 836 by ph10, Wed Dec 28 17:16:11 2011 UTC
# Line 1  Line 1 
1  /\x{100}/8DZ  /-- This set of tests checks the API, internals, and non-Perl stuff for UTF
2        support, excluding Unicode properties. However, tests that give different
3        results in 8-bit and 16-bit modes are excluded (see tests 16 and 17). --/
4    
5  /\x{1000}/8DZ  /\x{110000}/8DZ
6    
7  /\x{10000}/8DZ  /\x{ffffffff}/8
   
 /\x{100000}/8DZ  
   
 /\x{1000000}/8DZ  
   
 /\x{4000000}/8DZ  
8    
9  /\x{7fffFFFF}/8DZ  /\x{100000000}/8
10    
11  /[\x{ff}]/8DZ  /\x{d800}/8
12    
13  /[\x{100}]/8DZ  /\x{dfff}/8
14    
15  /\x{ffffffff}/8  /\x{d7ff}/8
16    
17  /\x{100000000}/8  /\x{e000}/8
18    
19  /^\x{100}a\x{1234}/8  /^\x{100}a\x{1234}/8
20      \x{100}a\x{1234}bcd      \x{100}a\x{1234}bcd
21    
 /\x80/8DZ  
   
 /\xff/8DZ  
   
22  /\x{0041}\x{2262}\x{0391}\x{002e}/DZ8  /\x{0041}\x{2262}\x{0391}\x{002e}/DZ8
23      \x{0041}\x{2262}\x{0391}\x{002e}      \x{0041}\x{2262}\x{0391}\x{002e}
24    
 /\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  
   
25  /.{3,5}X/DZ8  /.{3,5}X/DZ8
26      \x{212ab}\x{212ab}\x{212ab}\x{861}X      \x{212ab}\x{212ab}\x{212ab}\x{861}X
27    
   
28  /.{3,5}?/DZ8  /.{3,5}?/DZ8
29      \x{212ab}\x{212ab}\x{212ab}\x{861}      \x{212ab}\x{212ab}\x{212ab}\x{861}
30    
 /-- These tests are here rather than in testinput4 because Perl 5.6 has some  
 problems with UTF-8 support, in the area of \x{..} where the value is < 255.  
 It grumbles about invalid UTF-8 strings. --/  
   
 /^[a\x{c0}]b/8  
     \x{c0}b  
   
 /^([a\x{c0}]*?)aa/8  
     a\x{c0}aaaa/  
   
 /^([a\x{c0}]*?)aa/8  
     a\x{c0}aaaa/  
     a\x{c0}a\x{c0}aaa/  
   
 /^([a\x{c0}]*)aa/8  
     a\x{c0}aaaa/  
     a\x{c0}a\x{c0}aaa/  
   
 /^([a\x{c0}]*)a\x{c0}/8  
     a\x{c0}aaaa/  
     a\x{c0}a\x{c0}aaa/  
   
 /-- --/  
   
31  /(?<=\C)X/8  /(?<=\C)X/8
32      Should produce an error diagnostic      Should produce an error diagnostic
33    
 /-- 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)(.*)/8  
     X\x{1234}  
     X\nabc  
   
34  /^[ab]/8DZ  /^[ab]/8DZ
35      bar      bar
36      *** Failers      *** Failers
# Line 102  can't tell the difference.) --/ Line 45  can't tell the difference.) --/
45      *** Failers      *** Failers
46      aaa      aaa
47    
 /[^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  
   
48  /\x{100}*(\d+|"(?1)")/8  /\x{100}*(\d+|"(?1)")/8
49      1234      1234
50      "1234"      "1234"
# Line 132  can't tell the difference.) --/ Line 55  can't tell the difference.) --/
55      *** Failers      *** Failers
56      \x{100}\x{100}abcd      \x{100}\x{100}abcd
57    
 /\x{100}/8DZ  
   
58  /\x{100}*/8DZ  /\x{100}*/8DZ
59    
60  /a\x{100}*/8DZ  /a\x{100}*/8DZ
61    
62  /ab\x{100}*/8DZ  /ab\x{100}*/8DZ
63    
 /a\x{100}\x{101}*/8DZ  
   
 /a\x{100}\x{101}+/8DZ  
   
64  /\x{100}*A/8DZ  /\x{100}*A/8DZ
65      A      A
66    
67  /\x{100}*\d(?R)/8DZ  /\x{100}*\d(?R)/8DZ
68    
 /[^\x{c4}]/DZ  
   
 /[^\x{c4}]/8DZ  
   
 /[\x{100}]/8DZ  
     \x{100}  
     Z\x{100}  
     \x{100}Z  
     *** Failers  
   
69  /[Z\x{100}]/8DZ  /[Z\x{100}]/8DZ
70      Z\x{100}      Z\x{100}
71      \x{100}      \x{100}
# Line 183  can't tell the difference.) --/ Line 90  can't tell the difference.) --/
90  /[\xFF]/DZ  /[\xFF]/DZ
91      >\xff<      >\xff<
92    
 /[\xff]/DZ8  
     >\x{ff}<  
   
93  /[^\xFF]/DZ  /[^\xFF]/DZ
94    
 /[^\xff]/8DZ  
   
95  /[Ä-Ü]/8  /[Ä-Ü]/8
96      Ö # Matches without Study      Ö # Matches without Study
97      \x{d6}      \x{d6}
# Line 206  can't tell the difference.) --/ Line 108  can't tell the difference.) --/
108      Ö <-- Same with Study      Ö <-- Same with Study
109      \x{d6}      \x{d6}
110    
 /[]/8  
   
 //8  
   
 /xxx/8  
   
 /xxx/8?DZ  
   
 /abc/8  
     ]  
       
       
     \?  
   
 /anything/8  
     \xc0\x80  
     \xc1\x8f  
     \xe0\x9f\x80  
     \xf0\x8f\x80\x80  
     \xf8\x87\x80\x80\x80  
     \xfc\x83\x80\x80\x80\x80  
     \xfe\x80\x80\x80\x80\x80  
     \xff\x80\x80\x80\x80\x80  
     \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  
     \?\xf9\x87\x80\x80\x80  
     \?\xfc\x84\x80\x80\x80\x80  
     \?\xfd\x83\x80\x80\x80\x80  
   
 /\x{100}abc(xyz(?1))/8DZ  
   
111  /[^\x{100}]abc(xyz(?1))/8DZ  /[^\x{100}]abc(xyz(?1))/8DZ
112    
113  /[ab\x{100}]abc(xyz(?1))/8DZ  /[ab\x{100}]abc(xyz(?1))/8DZ
# Line 264  can't tell the difference.) --/ Line 127  can't tell the difference.) --/
127  /\w/8  /\w/8
128      \x{100}X      \x{100}X
129    
 /a\x{1234}b/P8  
     a\x{1234}b  
   
130  /^\ሴ/8DZ  /^\ሴ/8DZ
131    
132  /\777/I  /\777/I
133    
 /\777/8I  
   \x{1ff}  
   \777  
   
134  /\x{100}*\d/8DZ  /\x{100}*\d/8DZ
135    
136  /\x{100}*\s/8DZ  /\x{100}*\s/8DZ
# Line 287  can't tell the difference.) --/ Line 143  can't tell the difference.) --/
143    
144  /\x{100}*\W/8DZ  /\x{100}*\W/8DZ
145    
 /\x{100}+\x{200}/8DZ  
   
 /\x{100}+X/8DZ  
   
 /X+\x{200}/8DZ  
   
146  /()()()()()()()()()()  /()()()()()()()()()()
147   ()()()()()()()()()()   ()()()()()()()()()()
148   ()()()()()()()()()()   ()()()()()()()()()()
# Line 304  can't tell the difference.) --/ Line 154  can't tell the difference.) --/
154    
155  /^[\QĀ\E-\QŐ\E]/BZ8  /^[\QĀ\E-\QŐ\E]/BZ8
156    
 /^[\QĀ\E-\QŐ\E/BZ8  
   
157  /^abc./mgx8<any>  /^abc./mgx8<any>
158      abc1 \x0aabc2 \x0babc3xx \x0cabc4 \x0dabc5xx \x0d\x0aabc6 \x{0085}abc7 \x{2028}abc8 \x{2029}abc9 JUNK      abc1 \x0aabc2 \x0babc3xx \x0cabc4 \x0dabc5xx \x0d\x0aabc6 \x{0085}abc7 \x{2028}abc8 \x{2029}abc9 JUNK
159    
# Line 400  can't tell the difference.) --/ Line 248  can't tell the difference.) --/
248  /.*$/8<any>  /.*$/8<any>
249      \x{1ec5}      \x{1ec5}
250    
 /-- 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}\?  
   
251  /a\Rb/I8<bsr_anycrlf>  /a\Rb/I8<bsr_anycrlf>
252      a\rb      a\rb
253      a\nb      a\nb
# Line 475  can't tell the difference.) --/ Line 306  can't tell the difference.) --/
306    
307  /(\x{de})\1/  /(\x{de})\1/
308      \x{de}\x{de}      \x{de}\x{de}
     \x{123}  
309    
310  /X/8f<any>  /X/8f<any>
311      A\x{1ec5}ABCXYZ      A\x{1ec5}ABCXYZ
312    
 /(*UTF8)\x{1234}/  
   abcd\x{1234}pqr  
   
 /(*CRLF)(*UTF8)(*BSR_UNICODE)a\Rb/I  
   
313  /Xa{2,4}b/8  /Xa{2,4}b/8
314      X\P      X\P
315      Xa\P      Xa\P
# Line 741  can't tell the difference.) --/ Line 566  can't tell the difference.) --/
566      the cat\P      the cat\P
567      the cat\P\P      the cat\P\P
568    
569  / End of testinput5 /  /abcd*/8
570        xxxxabcd\P
571        xxxxabcd\P\P
572    
573    /abcd*/i8
574        xxxxabcd\P
575        xxxxabcd\P\P
576        XXXXABCD\P
577        XXXXABCD\P\P
578    
579    /abc\d*/8
580        xxxxabc1\P
581        xxxxabc1\P\P
582    
583    /(a)bc\1*/8
584        xxxxabca\P
585        xxxxabca\P\P
586    
587    /abc[de]*/8
588        xxxxabcde\P
589        xxxxabcde\P\P
590    
591    /X\W{3}X/8
592        \PX
593    
594    /\sxxx\s/8T1
595        AB\x{85}xxx\x{a0}XYZ
596        AB\x{a0}xxx\x{85}XYZ
597    
598    /\S \S/8T1
599        \x{a2} \x{84}
600    
601    'A#хц'8x<any>BZ
602    
603    'A#хц
604      PQ'8x<any>BZ
605    
606    /a+#хaa
607      z#XX?/8x<any>BZ
608    
609    /a+#хaa
610      z#х?/8x<any>BZ
611    
612    /\g{A}xxx#bXX(?'A'123) (?'A'456)/8x<any>BZ
613    
614    /\g{A}xxx#bх(?'A'123) (?'A'456)/8x<any>BZ
615    
616    /^\cģ/8
617    
618    /(\R*)(.)/s8
619        \r\n
620        \r\r\n\n\r
621        \r\r\n\n\r\n
622    
623    /(\R)*(.)/s8
624        \r\n
625        \r\r\n\n\r
626        \r\r\n\n\r\n
627    
628    /[^\x{1234}]+/iS8I
629    
630    /[^\x{1234}]+?/iS8I
631    
632    /[^\x{1234}]++/iS8I
633    
634    /[^\x{1234}]{2}/iS8I
635    
636    //<bsr_anycrlf><bsr_unicode>
637    
638    /f.*/
639        \P\Pfor
640    
641    /f.*/s
642        \P\Pfor
643    
644    /f.*/8
645        \P\Pfor
646    
647    /f.*/8s
648        \P\Pfor
649    
650    /\x{d7ff}\x{e000}/8
651    
652    /\x{d800}/8
653    
654    /\x{dfff}/8
655    
656    /\h+/8
657        \x{1681}\x{200b}\x{1680}\x{2000}\x{202f}\x{3000}
658        \x{3001}\x{2fff}\x{200a}\x{a0}\x{2000}
659    
660    /[\h\x{e000}]+/8BZ
661        \x{1681}\x{200b}\x{1680}\x{2000}\x{202f}\x{3000}
662        \x{3001}\x{2fff}\x{200a}\x{a0}\x{2000}
663    
664    /\H+/8
665        \x{1680}\x{180e}\x{167f}\x{1681}\x{180d}\x{180f}
666        \x{2000}\x{200a}\x{1fff}\x{200b}
667        \x{202f}\x{205f}\x{202e}\x{2030}\x{205e}\x{2060}
668        \x{a0}\x{3000}\x{9f}\x{a1}\x{2fff}\x{3001}
669    
670    /[\H\x{d7ff}]+/8BZ
671        \x{1680}\x{180e}\x{167f}\x{1681}\x{180d}\x{180f}
672        \x{2000}\x{200a}\x{1fff}\x{200b}
673        \x{202f}\x{205f}\x{202e}\x{2030}\x{205e}\x{2060}
674        \x{a0}\x{3000}\x{9f}\x{a1}\x{2fff}\x{3001}
675    
676    /\v+/8
677        \x{2027}\x{2030}\x{2028}\x{2029}
678        \x09\x0e\x{84}\x{86}\x{85}\x0a\x0b\x0c\x0d
679    
680    /[\v\x{e000}]+/8BZ
681        \x{2027}\x{2030}\x{2028}\x{2029}
682        \x09\x0e\x{84}\x{86}\x{85}\x0a\x0b\x0c\x0d
683    
684    /\V+/8
685        \x{2028}\x{2029}\x{2027}\x{2030}
686        \x{85}\x0a\x0b\x0c\x0d\x09\x0e\x{84}\x{86}
687    
688    /[\V\x{d7ff}]+/8BZ
689        \x{2028}\x{2029}\x{2027}\x{2030}
690        \x{85}\x0a\x0b\x0c\x0d\x09\x0e\x{84}\x{86}
691    
692    /\R+/8<bsr_unicode>
693        \x{2027}\x{2030}\x{2028}\x{2029}
694        \x09\x0e\x{84}\x{86}\x{85}\x0a\x0b\x0c\x0d
695    
696    /-- End of testinput5 --/

Legend:
Removed from v.428  
changed lines
  Added in v.836

  ViewVC Help
Powered by ViewVC 1.1.5