I was über lazy at first, so took libs from SM. But actually it's quite easy to compile, so let's update to latest version \o/.
		
			
				
	
	
		
			367 lines
		
	
	
		
			4.9 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
	
	
			
		
		
	
	
			367 lines
		
	
	
		
			4.9 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
	
	
/-- This set of tests is for UTF-8 support but not Unicode property support,
 | 
						||
    and is relevant only to the 8-bit library. --/
 | 
						||
 | 
						||
< forbid W
 | 
						||
 | 
						||
/X(\C{3})/8
 | 
						||
    X\x{1234}
 | 
						||
 | 
						||
/X(\C{4})/8
 | 
						||
    X\x{1234}YZ
 | 
						||
    
 | 
						||
/X\C*/8
 | 
						||
    XYZabcdce
 | 
						||
    
 | 
						||
/X\C*?/8
 | 
						||
    XYZabcde
 | 
						||
    
 | 
						||
/X\C{3,5}/8
 | 
						||
    Xabcdefg   
 | 
						||
    X\x{1234} 
 | 
						||
    X\x{1234}YZ
 | 
						||
    X\x{1234}\x{512}  
 | 
						||
    X\x{1234}\x{512}YZ
 | 
						||
 | 
						||
/X\C{3,5}?/8
 | 
						||
    Xabcdefg   
 | 
						||
    X\x{1234} 
 | 
						||
    X\x{1234}YZ
 | 
						||
    X\x{1234}\x{512}  
 | 
						||
 | 
						||
/a\Cb/8
 | 
						||
    aXb
 | 
						||
    a\nb
 | 
						||
    
 | 
						||
/a\C\Cb/8 
 | 
						||
    a\x{100}b 
 | 
						||
 | 
						||
/ab\Cde/8
 | 
						||
    abXde
 | 
						||
 | 
						||
/a\C\Cb/8 
 | 
						||
    a\x{100}b
 | 
						||
    ** Failers 
 | 
						||
    a\x{12257}b
 | 
						||
 | 
						||
/[Ã]/8
 | 
						||
 | 
						||
/Ã/8
 | 
						||
 | 
						||
/ÃÃÃxxx/8
 | 
						||
 | 
						||
/ÃÃÃxxx/8?DZSSO
 | 
						||
 | 
						||
/badutf/8
 | 
						||
    \xdf
 | 
						||
    \xef
 | 
						||
    \xef\x80
 | 
						||
    \xf7
 | 
						||
    \xf7\x80
 | 
						||
    \xf7\x80\x80
 | 
						||
    \xfb
 | 
						||
    \xfb\x80
 | 
						||
    \xfb\x80\x80
 | 
						||
    \xfb\x80\x80\x80
 | 
						||
    \xfd
 | 
						||
    \xfd\x80
 | 
						||
    \xfd\x80\x80
 | 
						||
    \xfd\x80\x80\x80
 | 
						||
    \xfd\x80\x80\x80\x80
 | 
						||
    \xdf\x7f
 | 
						||
    \xef\x7f\x80
 | 
						||
    \xef\x80\x7f
 | 
						||
    \xf7\x7f\x80\x80
 | 
						||
    \xf7\x80\x7f\x80
 | 
						||
    \xf7\x80\x80\x7f
 | 
						||
    \xfb\x7f\x80\x80\x80
 | 
						||
    \xfb\x80\x7f\x80\x80
 | 
						||
    \xfb\x80\x80\x7f\x80
 | 
						||
    \xfb\x80\x80\x80\x7f
 | 
						||
    \xfd\x7f\x80\x80\x80\x80
 | 
						||
    \xfd\x80\x7f\x80\x80\x80
 | 
						||
    \xfd\x80\x80\x7f\x80\x80
 | 
						||
    \xfd\x80\x80\x80\x7f\x80
 | 
						||
    \xfd\x80\x80\x80\x80\x7f
 | 
						||
    \xed\xa0\x80
 | 
						||
    \xc0\x8f
 | 
						||
    \xe0\x80\x8f
 | 
						||
    \xf0\x80\x80\x8f
 | 
						||
    \xf8\x80\x80\x80\x8f
 | 
						||
    \xfc\x80\x80\x80\x80\x8f
 | 
						||
    \x80
 | 
						||
    \xfe
 | 
						||
    \xff
 | 
						||
 | 
						||
/badutf/8
 | 
						||
    \xfb\x80\x80\x80\x80
 | 
						||
    \xfd\x80\x80\x80\x80\x80
 | 
						||
    \xf7\xbf\xbf\xbf
 | 
						||
 | 
						||
/shortutf/8
 | 
						||
    \P\P\xdf
 | 
						||
    \P\P\xef
 | 
						||
    \P\P\xef\x80
 | 
						||
    \P\P\xf7
 | 
						||
    \P\P\xf7\x80
 | 
						||
    \P\P\xf7\x80\x80
 | 
						||
    \P\P\xfb
 | 
						||
    \P\P\xfb\x80
 | 
						||
    \P\P\xfb\x80\x80
 | 
						||
    \P\P\xfb\x80\x80\x80
 | 
						||
    \P\P\xfd
 | 
						||
    \P\P\xfd\x80
 | 
						||
    \P\P\xfd\x80\x80
 | 
						||
    \P\P\xfd\x80\x80\x80
 | 
						||
    \P\P\xfd\x80\x80\x80\x80
 | 
						||
 | 
						||
/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}/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.) --/
 | 
						||
    
 | 
						||
/X(\C)(.*)/8
 | 
						||
    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} 
 | 
						||
 | 
						||
/Ä€{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
 | 
						||
 | 
						||
/a\x{1234}b/P8
 | 
						||
    a\x{1234}b
 | 
						||
 | 
						||
/\777/8I
 | 
						||
  \x{1ff}
 | 
						||
  \777 
 | 
						||
  
 | 
						||
/\x{100}+\x{200}/8DZ
 | 
						||
 | 
						||
/\x{100}+X/8DZ
 | 
						||
 | 
						||
/^[\QÄ€\E-\QÅ<51>\E/BZ8
 | 
						||
 | 
						||
/-- This tests the stricter UTF-8 check according to RFC 3629. --/ 
 | 
						||
    
 | 
						||
/X/8
 | 
						||
    \x{d800}
 | 
						||
    \x{d800}\?
 | 
						||
    \x{da00}
 | 
						||
    \x{da00}\?
 | 
						||
    \x{dfff}
 | 
						||
    \x{dfff}\?
 | 
						||
    \x{110000}    
 | 
						||
    \x{110000}\?    
 | 
						||
    \x{2000000} 
 | 
						||
    \x{2000000}\? 
 | 
						||
    \x{7fffffff} 
 | 
						||
    \x{7fffffff}\? 
 | 
						||
 | 
						||
/(*UTF8)\x{1234}/
 | 
						||
  abcd\x{1234}pqr
 | 
						||
 | 
						||
/(*CRLF)(*UTF)(*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
 | 
						||
 | 
						||
/\x{1234}+?/iS8I
 | 
						||
 | 
						||
/\x{1234}++/iS8I
 | 
						||
 | 
						||
/\x{1234}{2}/iS8I
 | 
						||
 | 
						||
/[^\x{c4}]/8DZ
 | 
						||
 | 
						||
/X+\x{200}/8DZ
 | 
						||
 | 
						||
/\R/SI8
 | 
						||
 | 
						||
/\777/8DZ
 | 
						||
 | 
						||
/\w+\x{C4}/8BZ
 | 
						||
    a\x{C4}\x{C4}
 | 
						||
 | 
						||
/\w+\x{C4}/8BZT1
 | 
						||
    a\x{C4}\x{C4}
 | 
						||
    
 | 
						||
/\W+\x{C4}/8BZ
 | 
						||
    !\x{C4}
 | 
						||
 
 | 
						||
/\W+\x{C4}/8BZT1
 | 
						||
    !\x{C4}
 | 
						||
 | 
						||
/\W+\x{A1}/8BZ
 | 
						||
    !\x{A1}
 | 
						||
 
 | 
						||
/\W+\x{A1}/8BZT1
 | 
						||
    !\x{A1}
 | 
						||
 | 
						||
/X\s+\x{A0}/8BZ
 | 
						||
    X\x20\x{A0}\x{A0}
 | 
						||
 | 
						||
/X\s+\x{A0}/8BZT1
 | 
						||
    X\x20\x{A0}\x{A0}
 | 
						||
 | 
						||
/\S+\x{A0}/8BZ
 | 
						||
    X\x{A0}\x{A0}
 | 
						||
 | 
						||
/\S+\x{A0}/8BZT1
 | 
						||
    X\x{A0}\x{A0}
 | 
						||
 | 
						||
/\x{a0}+\s!/8BZ
 | 
						||
    \x{a0}\x20!
 | 
						||
 | 
						||
/\x{a0}+\s!/8BZT1
 | 
						||
    \x{a0}\x20!
 | 
						||
 | 
						||
/A/8
 | 
						||
  \x{ff000041}
 | 
						||
  \x{7f000041} 
 | 
						||
 | 
						||
/(*UTF8)abc/9
 | 
						||
 | 
						||
/abc/89
 | 
						||
 | 
						||
/-- End of testinput15 --/
 |