forked from aniani/vim
7d40b8a532
Problem: Tests 44 and 99 are old style. Solution: Convert to new style tests. (Yegappan Lakshmanan, closes #6536)
505 lines
19 KiB
VimL
505 lines
19 KiB
VimL
" Tests for regexp in utf8 encoding
|
||
|
||
func s:equivalence_test()
|
||
let str = "A<><41><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>廕<EFBFBD> B廎<42><E5BB8E> C<><43><EFBFBD><EFBFBD><EFBFBD> D<><44>廎𢈈<E5BB8E>廎<EFBFBD> E<><45><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>廕筵獐 F廎<46> G<><47><EFBFBD>罌エオジ廎<E382B8> H臚藻廎Ｓ蒂廎<E89282> I<><49><EFBFBD><EFBFBD>蘑蘆蘇蠔襤<E8A094>廙<EFBFBD> J譬 K譯ガ廎唸葩 L贏躉躅醴<E8BA85>廎<EFBFBD> M廎擔<E5BB8E> N<><4E><EFBFBD><EFBFBD>廜<EFBFBD><E5BB9C> O<><4F><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>ギグ廙<E382B0> P廜竢<E5BB9C> Q R<><52><EFBFBD>廜嘣<E5BB9C> S<><53><EFBFBD><EFBFBD>廜<EFBFBD> T籠聾臟廜花僥 U<><55><EFBFBD><EFBFBD>襯讀贗躓酈鑑⑸<E99191>廙<EFBFBD> V廜<56> W霽廕<E99CBD>廕<EFBFBD><E5BB95>廕<EFBFBD> X廕𢈈<E5BB95> Y<>韃顫廕墰輔廙嗤遘 Z饕驍鬚ⅰ廕靟<E5BB95> a<>獺璽瓊瓣疇<E793A3><E79687><EFBFBD><EFBFBD><EFBFBD>ゥ廕<E382A5> b廎<62><E5BB8E> c癟<63><E7999F><EFBFBD><EFBFBD> d<><64>廎𦘒<E5BB8E>廎<EFBFBD> e癡矇礙禱<E7A499><E7A6B1><EFBFBD><EFBFBD><EFBFBD>廕頗瑤 f廎<66> g<><67>纂耀ォカス廎<E382B9> h艦藹廎€葷廎抨<E5BB8E> i穫穩簾簿藺蘋蘊蠕<E8988A>廙<EFBFBD> j警ゴ k譟キ廎晨葭 l贍躁躂<E8BA81><E8BA82>廎<EFBFBD> m廎蹄<E5BB8E> n簽<6E><E7B0BD><EFBFBD><EFBFBD>廜<EFBFBD><E5BB9C> o簷籀繫繭繹繪<E7B9B9><E7B9AA><EFBFBD>①<EFBFBD>クケ廙<E382B1> p廜鐧<E5BB9C> q r<><72><EFBFBD>廜䟭<E5BB9C> s<><73><EFBFBD>禳廜<E7A6B3> t籟聽襲廜後僖廕<E58396> u羅繳羶羹觼贖躑轡鑄鑒⑹<E99192>廙<EFBFBD> v廜<76> w霾廕<E99CBE><E5BB95>廕<EFBFBD><E5BB95>廕<EFBFBD> x廕𦘒<E5BB95> y羸藩韁廕𥕥<E5BB95>廙喔遠廙<E981A0> z驕髒鱉ⅱ廕𤒹<E5BB95>"
|
||
let groups = split(str)
|
||
for group1 in groups
|
||
for c in split(group1, '\zs')
|
||
" next statement confirms that equivalence class matches every
|
||
" character in group
|
||
call assert_match('^[[=' . c . '=]]*$', group1)
|
||
for group2 in groups
|
||
if group2 != group1
|
||
" next statement converts that equivalence class doesn't match
|
||
" character in any other group
|
||
call assert_equal(-1, match(group2, '[[=' . c . '=]]'))
|
||
endif
|
||
endfor
|
||
endfor
|
||
endfor
|
||
endfunc
|
||
|
||
func Test_equivalence_re1()
|
||
set re=1
|
||
call s:equivalence_test()
|
||
set re=0
|
||
endfunc
|
||
|
||
func Test_equivalence_re2()
|
||
set re=2
|
||
call s:equivalence_test()
|
||
set re=0
|
||
endfunc
|
||
|
||
func s:classes_test()
|
||
if has('win32')
|
||
set iskeyword=@,48-57,_,192-255
|
||
endif
|
||
set isprint=@,161-255
|
||
call assert_equal('Mot繹rhead', matchstr('Mot繹rhead', '[[:print:]]\+'))
|
||
|
||
let alnumchars = ''
|
||
let alphachars = ''
|
||
let backspacechar = ''
|
||
let blankchars = ''
|
||
let cntrlchars = ''
|
||
let digitchars = ''
|
||
let escapechar = ''
|
||
let graphchars = ''
|
||
let lowerchars = ''
|
||
let printchars = ''
|
||
let punctchars = ''
|
||
let returnchar = ''
|
||
let spacechars = ''
|
||
let tabchar = ''
|
||
let upperchars = ''
|
||
let xdigitchars = ''
|
||
let identchars = ''
|
||
let identchars1 = ''
|
||
let kwordchars = ''
|
||
let kwordchars1 = ''
|
||
let fnamechars = ''
|
||
let fnamechars1 = ''
|
||
let i = 1
|
||
while i <= 255
|
||
let c = nr2char(i)
|
||
if c =~ '[[:alpha:]]'
|
||
let alphachars .= c
|
||
endif
|
||
if c =~ '[[:alnum:]]'
|
||
let alnumchars .= c
|
||
endif
|
||
if c =~ '[[:backspace:]]'
|
||
let backspacechar .= c
|
||
endif
|
||
if c =~ '[[:blank:]]'
|
||
let blankchars .= c
|
||
endif
|
||
if c =~ '[[:cntrl:]]'
|
||
let cntrlchars .= c
|
||
endif
|
||
if c =~ '[[:digit:]]'
|
||
let digitchars .= c
|
||
endif
|
||
if c =~ '[[:escape:]]'
|
||
let escapechar .= c
|
||
endif
|
||
if c =~ '[[:graph:]]'
|
||
let graphchars .= c
|
||
endif
|
||
if c =~ '[[:lower:]]'
|
||
let lowerchars .= c
|
||
endif
|
||
if c =~ '[[:print:]]'
|
||
let printchars .= c
|
||
endif
|
||
if c =~ '[[:punct:]]'
|
||
let punctchars .= c
|
||
endif
|
||
if c =~ '[[:return:]]'
|
||
let returnchar .= c
|
||
endif
|
||
if c =~ '[[:space:]]'
|
||
let spacechars .= c
|
||
endif
|
||
if c =~ '[[:tab:]]'
|
||
let tabchar .= c
|
||
endif
|
||
if c =~ '[[:upper:]]'
|
||
let upperchars .= c
|
||
endif
|
||
if c =~ '[[:xdigit:]]'
|
||
let xdigitchars .= c
|
||
endif
|
||
if c =~ '[[:ident:]]'
|
||
let identchars .= c
|
||
endif
|
||
if c =~ '\i'
|
||
let identchars1 .= c
|
||
endif
|
||
if c =~ '[[:keyword:]]'
|
||
let kwordchars .= c
|
||
endif
|
||
if c =~ '\k'
|
||
let kwordchars1 .= c
|
||
endif
|
||
if c =~ '[[:fname:]]'
|
||
let fnamechars .= c
|
||
endif
|
||
if c =~ '\f'
|
||
let fnamechars1 .= c
|
||
endif
|
||
let i += 1
|
||
endwhile
|
||
|
||
call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars)
|
||
call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars)
|
||
call assert_equal("\b", backspacechar)
|
||
call assert_equal("\t ", blankchars)
|
||
call assert_equal("\x01\x02\x03\x04\x05\x06\x07\b\t\n\x0b\f\r\x0e\x0f\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1a\e\x1c\x1d\x1e\x1f\x7f", cntrlchars)
|
||
call assert_equal("0123456789", digitchars)
|
||
call assert_equal("\<Esc>", escapechar)
|
||
call assert_equal('!"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~', graphchars)
|
||
call assert_equal('abcdefghijklmnopqrstuvwxyz繕<7A><E7B995>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繪羅繳羶羹羸臘藩', lowerchars)
|
||
call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~<7E>癒瞽瞿瞻瞼礎禮穡穢穠竄竅簫簧簪簞簣簡糧織繕繞繚繡繒繙罈翹翻職聶<E881B7><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩', printchars)
|
||
call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars)
|
||
call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZ<59><5A><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>', upperchars)
|
||
call assert_equal("\r", returnchar)
|
||
call assert_equal("\t\n\x0b\f\r ", spacechars)
|
||
call assert_equal("\t", tabchar)
|
||
call assert_equal('0123456789ABCDEFabcdef', xdigitchars)
|
||
|
||
if has('win32')
|
||
let identchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz<79><7A><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>癒瞽瞿瞻瞼礎禮繕<E7A6AE><E7B995><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繪羅繳羶羹羸臘藩'
|
||
let kwordchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz繕<7A><E7B995><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
elseif has('ebcdic')
|
||
let identchars_ok = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz<79><7A><EFBFBD><EFBFBD><EFBFBD><EFBFBD>竅簧繕繙聶<E7B999><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
let kwordchars_ok = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz<79><7A><EFBFBD><EFBFBD><EFBFBD><EFBFBD>竅簧繕繙聶<E7B999><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
else
|
||
let identchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz繕<7A><E7B995><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
let kwordchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz繕<7A><E7B995><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
endif
|
||
|
||
if has('win32')
|
||
let fnamechars_ok = '!#$%+,-./0123456789:=@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]_abcdefghijklmnopqrstuvwxyz{}~<7E>癒瞽瞿瞻瞼礎禮穡穢穠竄竅簫簧簪簞簣簡糧織繕繞繚繡繒繙罈翹翻職聶<E881B7><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
elseif has('amiga')
|
||
let fnamechars_ok = '$+,-./0123456789:ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~<7E>癒瞽瞿瞻瞼礎禮穡穢穠竄竅簫簧簪簞簣簡糧織繕繞繚繡繒繙罈翹翻職聶<E881B7><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
elseif has('vms')
|
||
let fnamechars_ok = '#$%+,-./0123456789:;<>ABCDEFGHIJKLMNOPQRSTUVWXYZ[]_abcdefghijklmnopqrstuvwxyz~<7E>癒瞽瞿瞻瞼礎禮穡穢穠竄竅簫簧簪簞簣簡糧織繕繞繚繡繒繙罈翹翻職聶<E881B7><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
elseif has('ebcdic')
|
||
let fnamechars_ok = '#$%+,-./=ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~<7E>癒瞽瞿瞻瞼礎禮穡穢穠竄竅簫簧簪簞簣簡糧織繕繞繚繡繒繙罈翹翻職聶<E881B7><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
else
|
||
let fnamechars_ok = '#$%+,-./0123456789=ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~<7E>癒瞽瞿瞻瞼礎禮穡穢穠竄竅簫簧簪簞簣簡糧織繕繞繚繡繒繙罈翹翻職聶<E881B7><E881B6><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD><EFBFBD>獺璽瓊瓣疇疆癟癡矇礙禱穫穩簾簿簸簽簷籀繫繭繹繩繪羅繳羶羹羸臘藩'
|
||
endif
|
||
|
||
call assert_equal(identchars_ok, identchars)
|
||
call assert_equal(kwordchars_ok, kwordchars)
|
||
call assert_equal(fnamechars_ok, fnamechars)
|
||
|
||
call assert_equal(identchars1, identchars)
|
||
call assert_equal(kwordchars1, kwordchars)
|
||
call assert_equal(fnamechars1, fnamechars)
|
||
endfunc
|
||
|
||
func Test_classes_re1()
|
||
set re=1
|
||
call s:classes_test()
|
||
set re=0
|
||
endfunc
|
||
|
||
func Test_classes_re2()
|
||
set re=2
|
||
call s:classes_test()
|
||
set re=0
|
||
endfunc
|
||
|
||
func Test_reversed_range()
|
||
for re in range(0, 2)
|
||
exe 'set re=' . re
|
||
call assert_fails('call match("abc def", "[c-a]")', 'E944:')
|
||
endfor
|
||
set re=0
|
||
endfunc
|
||
|
||
func Test_large_class()
|
||
set re=1
|
||
call assert_fails('call match("abc def", "[\u3000-\u4000]")', 'E945:')
|
||
set re=2
|
||
call assert_equal(0, 'abc def' =~# '[\u3000-\u4000]')
|
||
call assert_equal(1, "\u3042" =~# '[\u3000-\u4000]')
|
||
set re=0
|
||
endfunc
|
||
|
||
func Test_optmatch_toolong()
|
||
set re=1
|
||
" Can only handle about 8000 characters.
|
||
let pat = '\\%[' .. repeat('x', 9000) .. ']'
|
||
call assert_fails('call match("abc def", "' .. pat .. '")', 'E339:')
|
||
set re=0
|
||
endfunc
|
||
|
||
" Test for regexp patterns with multi-byte support, using utf-8.
|
||
func Test_multibyte_chars()
|
||
" tl is a List of Lists with:
|
||
" 2: test auto/old/new 0: test auto/old 1: test auto/new
|
||
" regexp pattern
|
||
" text to test the pattern on
|
||
" expected match (optional)
|
||
" expected submatch 1 (optional)
|
||
" expected submatch 2 (optional)
|
||
" etc.
|
||
" When there is no match use only the first two items.
|
||
let tl = []
|
||
|
||
" Multi-byte character tests. These will fail unless vim is compiled
|
||
" with Multibyte (FEAT_MBYTE) or BIG/HUGE features.
|
||
call add(tl, [2, '[[:alpha:][=a=]]\+', '879 aia瓊璽aiuvna ', 'aia瓊璽aiuvna'])
|
||
call add(tl, [2, '[[=a=]]\+', 'dda瓊璽bcd', 'a瓊璽']) " equivalence classes
|
||
call add(tl, [2, '[^鉊<> ]\+', '鉊﹤腹 oijasoifjos ifjoisj f osij j 鉊﹤腹鉊﹤腹鉊<E885B9> abcd', 'oijasoifjos'])
|
||
call add(tl, [2, ' [^ ]\+', 'start 鉊｛bcd鉊<64> ', ' 鉊｛bcd鉊<64>'])
|
||
call add(tl, [2, '[鉊︴:alpha:][=a=]]\+', '879 aia瓊鉊￣Ｒ腹aiuvna ', 'aia瓊鉊￣Ｒ腹aiuvna'])
|
||
|
||
" this is not a normal "i" but 0xec
|
||
call add(tl, [2, '\p\+', '穫a', '穫a'])
|
||
call add(tl, [2, '\p*', 'a<><61>', 'a<><61>'])
|
||
|
||
" Test recognition of some character classes
|
||
call add(tl, [2, '\i\+', '&*穡xx ', 'xx'])
|
||
call add(tl, [2, '\f\+', '&*<2A>fname ', 'fname'])
|
||
|
||
" Test composing character matching
|
||
call add(tl, [2, '.鉊<>', 'x鉊﹤<E9898A>x y鉊『', 'y鉊<79>'])
|
||
call add(tl, [2, '.鉊﹤<E9898A>', 'x鉊﹤<E9898A>x y鉊『', 'x鉊﹤<E9898A>'])
|
||
call add(tl, [2, "\u05b9", " x\u05b9 ", "x\u05b9"])
|
||
call add(tl, [2, ".\u05b9", " x\u05b9 ", "x\u05b9"])
|
||
call add(tl, [2, "\u05b9\u05bb", " x\u05b9\u05bb ", "x\u05b9\u05bb"])
|
||
call add(tl, [2, ".\u05b9\u05bb", " x\u05b9\u05bb ", "x\u05b9\u05bb"])
|
||
call add(tl, [2, "\u05bb\u05b9", " x\u05b9\u05bb ", "x\u05b9\u05bb"])
|
||
call add(tl, [2, ".\u05bb\u05b9", " x\u05b9\u05bb ", "x\u05b9\u05bb"])
|
||
call add(tl, [2, "\u05b9", " y\u05bb x\u05b9 ", "x\u05b9"])
|
||
call add(tl, [2, ".\u05b9", " y\u05bb x\u05b9 ", "x\u05b9"])
|
||
call add(tl, [2, "\u05b9", " y\u05bb\u05b9 x\u05b9 ", "y\u05bb\u05b9"])
|
||
call add(tl, [2, ".\u05b9", " y\u05bb\u05b9 x\u05b9 ", "y\u05bb\u05b9"])
|
||
call add(tl, [1, "\u05b9\u05bb", " y\u05b9 x\u05b9\u05bb ", "x\u05b9\u05bb"])
|
||
call add(tl, [2, ".\u05b9\u05bb", " y\u05bb x\u05b9\u05bb ", "x\u05b9\u05bb"])
|
||
call add(tl, [2, "a", "ca\u0300t"])
|
||
call add(tl, [2, "ca", "ca\u0300t"])
|
||
call add(tl, [2, "a\u0300", "ca\u0300t", "a\u0300"])
|
||
call add(tl, [2, 'a\%C', "ca\u0300t", "a\u0300"])
|
||
call add(tl, [2, 'ca\%C', "ca\u0300t", "ca\u0300"])
|
||
call add(tl, [2, 'ca\%Ct', "ca\u0300t", "ca\u0300t"])
|
||
|
||
" Test \Z
|
||
call add(tl, [2, '繳\Z', 'x'])
|
||
call add(tl, [2, '<27><><EFBFBD><EFBFBD>\Z', '<27><><EFBFBD><EFBFBD>', '<27><><EFBFBD><EFBFBD>'])
|
||
call add(tl, [2, '<27>眥<EFBFBD><E79CA5>硌<EFBFBD>\Z', '<27><><EFBFBD><EFBFBD>', '<27><><EFBFBD><EFBFBD>'])
|
||
call add(tl, [2, '<27><><EFBFBD><EFBFBD>\Z', '<27>眥<EFBFBD><E79CA5>硌<EFBFBD>', '<27>眥<EFBFBD><E79CA5>硌<EFBFBD>'])
|
||
call add(tl, [2, '<27>眥<EFBFBD><E79CA5>硌<EFBFBD>\Z', '<27>眥<EFBFBD><E79CA5>硌<EFBFBD>', '<27>眥<EFBFBD><E79CA5>硌<EFBFBD>'])
|
||
call add(tl, [2, '<27>眥\Z', '<27>眥<EFBFBD>硊', '<27>硊'])
|
||
call add(tl, [2, "蚹\u200d\u05b9x\\Z", "x蚹\u200d\u05b9xy", "蚹\u200d\u05b9x"])
|
||
call add(tl, [2, "蚹\u200d\u05b9x\\Z", "x蚹\u200dxy", "蚹\u200dx"])
|
||
call add(tl, [2, "蚹\u200dx\\Z", "x蚹\u200d\u05b9xy", "蚹\u200d\u05b9x"])
|
||
call add(tl, [2, "蚹\u200dx\\Z", "x蚹\u200dxy", "蚹\u200dx"])
|
||
call add(tl, [2, "\u05b9\\Z", "xyz"])
|
||
call add(tl, [2, "\\Z\u05b9", "xyz"])
|
||
call add(tl, [2, "\u05b9\\Z", "xy\u05b9z", "y\u05b9"])
|
||
call add(tl, [2, "\\Z\u05b9", "xy\u05b9z", "y\u05b9"])
|
||
call add(tl, [1, "\u05b9\\+\\Z", "xy\u05b9z\u05b9 ", "y\u05b9z\u05b9"])
|
||
call add(tl, [1, "\\Z\u05b9\\+", "xy\u05b9z\u05b9 ", "y\u05b9z\u05b9"])
|
||
|
||
" Combining different tests and features
|
||
call add(tl, [2, '[^[=a=]]\+', 'dda瓊璽bcd', 'dd'])
|
||
|
||
" Run the tests
|
||
for t in tl
|
||
let re = t[0]
|
||
let pat = t[1]
|
||
let text = t[2]
|
||
let matchidx = 3
|
||
for engine in [0, 1, 2]
|
||
if engine == 2 && re == 0 || engine == 1 && re == 1
|
||
continue
|
||
endif
|
||
let ®expengine = engine
|
||
try
|
||
let l = matchlist(text, pat)
|
||
catch
|
||
call assert_report('Error ' . engine . ': pat: \"' . pat .
|
||
\ '\", text: \"' . text .
|
||
\ '\", caused an exception: \"' . v:exception . '\"')
|
||
endtry
|
||
" check the match itself
|
||
if len(l) == 0 && len(t) > matchidx
|
||
call assert_report('Error ' . engine . ': pat: \"' . pat .
|
||
\ '\", text: \"' . text .
|
||
\ '\", did not match, expected: \"' . t[matchidx] . '\"')
|
||
elseif len(l) > 0 && len(t) == matchidx
|
||
call assert_report('Error ' . engine . ': pat: \"' . pat .
|
||
\ '\", text: \"' . text . '\", match: \"' . l[0] .
|
||
\ '\", expected no match')
|
||
elseif len(t) > matchidx && l[0] != t[matchidx]
|
||
call assert_report('Error ' . engine . ': pat: \"' . pat .
|
||
\ '\", text: \"' . text . '\", match: \"' . l[0] .
|
||
\ '\", expected: \"' . t[matchidx] . '\"')
|
||
else
|
||
" Test passed
|
||
endif
|
||
if len(l) > 0
|
||
" check all the nine submatches
|
||
for i in range(1, 9)
|
||
if len(t) <= matchidx + i
|
||
let e = ''
|
||
else
|
||
let e = t[matchidx + i]
|
||
endif
|
||
if l[i] != e
|
||
call assert_report('Error ' . engine . ': pat: \"' . pat .
|
||
\ '\", text: \"' . text . '\", submatch ' . i .
|
||
\ ': \"' . l[i] . '\", expected: \"' . e . '\"')
|
||
endif
|
||
endfor
|
||
unlet i
|
||
endif
|
||
endfor
|
||
endfor
|
||
set regexpengine&
|
||
endfunc
|
||
|
||
" check that 'ambiwidth' does not change the meaning of \p
|
||
func Test_ambiwidth()
|
||
set regexpengine=1 ambiwidth=single
|
||
call assert_equal(0, match("\u00EC", '\p'))
|
||
set regexpengine=1 ambiwidth=double
|
||
call assert_equal(0, match("\u00EC", '\p'))
|
||
set regexpengine=2 ambiwidth=single
|
||
call assert_equal(0, match("\u00EC", '\p'))
|
||
set regexpengine=2 ambiwidth=double
|
||
call assert_equal(0, match("\u00EC", '\p'))
|
||
set regexpengine& ambiwidth&
|
||
endfunc
|
||
|
||
func Run_regexp_ignore_case()
|
||
call assert_equal('iI襤', substitute('iI襤', '\([iI襤]\)', '\1', 'g'))
|
||
|
||
call assert_equal('iIx', substitute('iI襤', '\c\([襤]\)', 'x', 'g'))
|
||
call assert_equal('xx襤', substitute('iI襤', '\(i\c\)', 'x', 'g'))
|
||
call assert_equal('iIx', substitute('iI襤', '\(襤\c\)', 'x', 'g'))
|
||
call assert_equal('iIx', substitute('iI襤', '\c\(\%u0130\)', 'x', 'g'))
|
||
call assert_equal('iIx', substitute('iI襤', '\c\([\u0130]\)', 'x', 'g'))
|
||
call assert_equal('iIx', substitute('iI襤', '\c\([\u012f-\u0131]\)', 'x', 'g'))
|
||
endfunc
|
||
|
||
func Test_regexp_ignore_case()
|
||
set regexpengine=1
|
||
call Run_regexp_ignore_case()
|
||
set regexpengine=2
|
||
call Run_regexp_ignore_case()
|
||
set regexpengine&
|
||
endfunc
|
||
|
||
" Tests for regexp with multi-byte encoding and various magic settings
|
||
func Run_regexp_multibyte_magic()
|
||
let text =<< trim END
|
||
1 a aa abb abbccc
|
||
2 d dd dee deefff
|
||
3 g gg ghh ghhiii
|
||
4 j jj jkk jkklll
|
||
5 m mm mnn mnnooo
|
||
6 x ^aa$ x
|
||
7 (a)(b) abbaa
|
||
8 axx [ab]xx
|
||
9 鉊徇腹鉆ū 鉊冢腹x
|
||
a 鉊冢腹x 鉊徇腹鉆ū
|
||
b <20>～<EFBFBD><EFBD9E>具<EFBFBD>
|
||
c x 竅<EFBFBD>洪
|
||
d 憭拐蝙x
|
||
e <20><>烵y
|
||
f <20><><EFBFBD>z
|
||
g a<EFBFBD>搓b
|
||
j 0123<EFBFBD>巳
|
||
k combinations
|
||
l 瓣繹 羹帢<EFBFBD><EFBFBD><EFBFBD>
|
||
END
|
||
|
||
new
|
||
call setline(1, text)
|
||
exe 'normal /a*b\{2}c\+/e' .. "\<CR>x"
|
||
call assert_equal('1 a aa abb abbcc', getline('.'))
|
||
exe 'normal /\Md\*e\{2}f\+/e' .. "\<CR>x"
|
||
call assert_equal('2 d dd dee deeff', getline('.'))
|
||
set nomagic
|
||
exe 'normal /g\*h\{2}i\+/e' .. "\<CR>x"
|
||
call assert_equal('3 g gg ghh ghhii', getline('.'))
|
||
exe 'normal /\mj*k\{2}l\+/e' .. "\<CR>x"
|
||
call assert_equal('4 j jj jkk jkkll', getline('.'))
|
||
exe 'normal /\vm*n{2}o+/e' .. "\<CR>x"
|
||
call assert_equal('5 m mm mnn mnnoo', getline('.'))
|
||
exe 'normal /\V^aa$/' .. "\<CR>x"
|
||
call assert_equal('6 x aa$ x', getline('.'))
|
||
set magic
|
||
exe 'normal /\v(a)(b)\2\1\1/e' .. "\<CR>x"
|
||
call assert_equal('7 (a)(b) abba', getline('.'))
|
||
exe 'normal /\V[ab]\(\[xy]\)\1' .. "\<CR>x"
|
||
call assert_equal('8 axx ab]xx', getline('.'))
|
||
|
||
" search for multi-byte without composing char
|
||
exe 'normal /鉊<>' .. "\<CR>x"
|
||
call assert_equal('9 鉊徇腹鉆ū 鉊軍', getline('.'))
|
||
|
||
" search for multi-byte with composing char
|
||
exe 'normal /鉊﹤<E9898A>' .. "\<CR>x"
|
||
call assert_equal('a 鉊冢腹x 鉊咦', getline('.'))
|
||
|
||
" find word by change of word class
|
||
exe 'normal /<2F>﹏<<3C>怒<EFBFBD>\><3E><>' .. "\<CR>x"
|
||
call assert_equal('b <20>怒<EFBFBD><E68092><EFBFBD>', getline('.'))
|
||
|
||
" Test \%u, [\u] and friends
|
||
" c
|
||
exe 'normal /\%u20ac' .. "\<CR>x"
|
||
call assert_equal('c x 竅x', getline('.'))
|
||
" d
|
||
exe 'normal /[\u4f7f\u5929]\+' .. "\<CR>x"
|
||
call assert_equal('d 雿篛', getline('.'))
|
||
" e
|
||
exe 'normal /\%U12345678' .. "\<CR>x"
|
||
call assert_equal('e y', getline('.'))
|
||
" f
|
||
exe 'normal /[\U1234abcd\u1234\uabcd]' .. "\<CR>x"
|
||
call assert_equal('f z', getline('.'))
|
||
" g
|
||
exe 'normal /\%d21879b' .. "\<CR>x"
|
||
call assert_equal('g abb', getline('.'))
|
||
|
||
" j Test backwards search from a multi-byte char
|
||
exe "normal /x\<CR>x?.\<CR>x"
|
||
call assert_equal('j 012<31><32>', getline('.'))
|
||
" k
|
||
let @w=':%s#comb[i]nations#<23><>廜␌<E5BB9C>怚<EFBFBD>帢<EFBFBD><E5B8A2><EFBFBD>#g'
|
||
@w
|
||
call assert_equal('k <20><>廜␌<E5BB9C>怚<EFBFBD>帢<EFBFBD><E5B8A2><EFBFBD>', getline(18))
|
||
|
||
close!
|
||
endfunc
|
||
|
||
func Test_regexp_multibyte_magic()
|
||
set regexpengine=1
|
||
call Run_regexp_multibyte_magic()
|
||
set regexpengine=2
|
||
call Run_regexp_multibyte_magic()
|
||
set regexpengine&
|
||
endfunc
|
||
|
||
" Test for 7.3.192
|
||
" command ":s/ \?/ /g" splits multi-byte characters into bytes
|
||
func Test_split_multibyte_to_bytes()
|
||
new
|
||
call setline(1, 'l 瓣繹 羹帢<E7BEB9><E5B8A2><EFBFBD>')
|
||
s/ \?/ /g
|
||
call assert_equal(' l 瓣 繹 羹 帢<><E5B8A2><EFBFBD>', getline(1))
|
||
close!
|
||
endfunc
|
||
|
||
" Test for matchstr() with multibyte characters
|
||
func Test_matchstr_multibyte()
|
||
new
|
||
call assert_equal('<27>', matchstr("<22><><EFBFBD><EFBFBD>", ".", 0, 2))
|
||
call assert_equal('<27><>', matchstr("<22><><EFBFBD><EFBFBD>", "..", 0, 2))
|
||
call assert_equal('<27>', matchstr("<22><><EFBFBD><EFBFBD>", ".", 0, 0))
|
||
call assert_equal('<27>', matchstr("<22><><EFBFBD><EFBFBD>", ".", 4, -1))
|
||
close!
|
||
endfunc
|
||
|
||
" Test for 7.4.636
|
||
" A search with end offset gets stuck at end of file.
|
||
func Test_search_with_end_offset()
|
||
new
|
||
call setline(1, ['', 'dog(a', 'cat('])
|
||
exe "normal /(/e+" .. "\<CR>"
|
||
normal "ayn
|
||
call assert_equal("a\ncat(", @a)
|
||
close!
|
||
endfunc
|
||
|
||
" vim: shiftwidth=2 sts=2 expandtab
|