1" Tests for regexp in utf8 encoding 2 3func s:equivalence_test() 4 let str = "AÀÁÂÃÄÅĀĂĄǍǞǠẢ BḂḆ CÇĆĈĊČ DĎĐḊḎḐ EÈÉÊËĒĔĖĘĚẺẼ FḞ GĜĞĠĢǤǦǴḠ HĤĦḢḦḨ IÌÍÎÏĨĪĬĮİǏỈ JĴ KĶǨḰḴ LĹĻĽĿŁḺ MḾṀ NÑŃŅŇṄṈ OÒÓÔÕÖØŌŎŐƠǑǪǬỎ PṔṖ Q RŔŖŘṘṞ SŚŜŞŠṠ TŢŤŦṪṮ UÙÚÛÜŨŪŬŮŰŲƯǓỦ VṼ WŴẀẂẄẆ XẊẌ YÝŶŸẎỲỶỸ ZŹŻŽƵẐẔ aàáâãäåāăąǎǟǡả bḃḇ cçćĉċč dďđḋḏḑ eèéêëēĕėęěẻẽ fḟ gĝğġģǥǧǵḡ hĥħḣḧḩẖ iìíîïĩīĭįǐỉ jĵǰ kķǩḱḵ lĺļľŀłḻ mḿṁ nñńņňʼnṅṉ oòóôõöøōŏőơǒǫǭỏ pṕṗ q rŕŗřṙṟ sśŝşšṡ tţťŧṫṯẗ uùúûüũūŭůűųưǔủ vṽ wŵẁẃẅẇẘ xẋẍ yýÿŷẏẙỳỷỹ zźżžƶẑẕ" 5 let groups = split(str) 6 for group1 in groups 7 for c in split(group1, '\zs') 8 " next statement confirms that equivalence class matches every 9 " character in group 10 call assert_match('^[[=' . c . '=]]*$', group1) 11 for group2 in groups 12 if group2 != group1 13 " next statement converts that equivalence class doesn't match 14 " character in any other group 15 call assert_equal(-1, match(group2, '[[=' . c . '=]]')) 16 endif 17 endfor 18 endfor 19 endfor 20endfunc 21 22func Test_equivalence_re1() 23 set re=1 24 call s:equivalence_test() 25 set re=0 26endfunc 27 28func Test_equivalence_re2() 29 set re=2 30 call s:equivalence_test() 31 set re=0 32endfunc 33 34func s:classes_test() 35 if has('win32') 36 set iskeyword=@,48-57,_,192-255 37 endif 38 set isprint=@,161-255 39 call assert_equal('Motörhead', matchstr('Motörhead', '[[:print:]]\+')) 40 41 let alnumchars = '' 42 let alphachars = '' 43 let backspacechar = '' 44 let blankchars = '' 45 let cntrlchars = '' 46 let digitchars = '' 47 let escapechar = '' 48 let graphchars = '' 49 let lowerchars = '' 50 let printchars = '' 51 let punctchars = '' 52 let returnchar = '' 53 let spacechars = '' 54 let tabchar = '' 55 let upperchars = '' 56 let xdigitchars = '' 57 let identchars = '' 58 let identchars1 = '' 59 let kwordchars = '' 60 let kwordchars1 = '' 61 let fnamechars = '' 62 let fnamechars1 = '' 63 let i = 1 64 while i <= 255 65 let c = nr2char(i) 66 if c =~ '[[:alpha:]]' 67 let alphachars .= c 68 endif 69 if c =~ '[[:alnum:]]' 70 let alnumchars .= c 71 endif 72 if c =~ '[[:backspace:]]' 73 let backspacechar .= c 74 endif 75 if c =~ '[[:blank:]]' 76 let blankchars .= c 77 endif 78 if c =~ '[[:cntrl:]]' 79 let cntrlchars .= c 80 endif 81 if c =~ '[[:digit:]]' 82 let digitchars .= c 83 endif 84 if c =~ '[[:escape:]]' 85 let escapechar .= c 86 endif 87 if c =~ '[[:graph:]]' 88 let graphchars .= c 89 endif 90 if c =~ '[[:lower:]]' 91 let lowerchars .= c 92 endif 93 if c =~ '[[:print:]]' 94 let printchars .= c 95 endif 96 if c =~ '[[:punct:]]' 97 let punctchars .= c 98 endif 99 if c =~ '[[:return:]]' 100 let returnchar .= c 101 endif 102 if c =~ '[[:space:]]' 103 let spacechars .= c 104 endif 105 if c =~ '[[:tab:]]' 106 let tabchar .= c 107 endif 108 if c =~ '[[:upper:]]' 109 let upperchars .= c 110 endif 111 if c =~ '[[:xdigit:]]' 112 let xdigitchars .= c 113 endif 114 if c =~ '[[:ident:]]' 115 let identchars .= c 116 endif 117 if c =~ '\i' 118 let identchars1 .= c 119 endif 120 if c =~ '[[:keyword:]]' 121 let kwordchars .= c 122 endif 123 if c =~ '\k' 124 let kwordchars1 .= c 125 endif 126 if c =~ '[[:fname:]]' 127 let fnamechars .= c 128 endif 129 if c =~ '\f' 130 let fnamechars1 .= c 131 endif 132 let i += 1 133 endwhile 134 135 call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars) 136 call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars) 137 call assert_equal("\b", backspacechar) 138 call assert_equal("\t ", blankchars) 139 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) 140 call assert_equal("0123456789", digitchars) 141 call assert_equal("\<Esc>", escapechar) 142 call assert_equal('!"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~', graphchars) 143 call assert_equal('abcdefghijklmnopqrstuvwxyzµßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', lowerchars) 144 call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ', printchars) 145 call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars) 146 call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ', upperchars) 147 call assert_equal("\r", returnchar) 148 call assert_equal("\t\n\x0b\f\r ", spacechars) 149 call assert_equal("\t", tabchar) 150 call assert_equal('0123456789ABCDEFabcdef', xdigitchars) 151 152 if has('win32') 153 let identchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz€‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”•–—˜™š›œžŸ ¡¢£¤¥¦§µÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ' 154 let kwordchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyzµÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 155 elseif has('ebcdic') 156 let identchars_ok = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz€ŒŽœž¬®µº¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 157 let kwordchars_ok = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz€ŒŽœž¬®µº¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 158 else 159 let identchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyzµÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 160 let kwordchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyzµÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 161 endif 162 163 if has('win32') 164 let fnamechars_ok = '!#$%+,-./0123456789:=@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]_abcdefghijklmnopqrstuvwxyz{}~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 165 elseif has('amiga') 166 let fnamechars_ok = '$+,-./0123456789:ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 167 elseif has('vms') 168 let fnamechars_ok = '#$%+,-./0123456789:;<>ABCDEFGHIJKLMNOPQRSTUVWXYZ[]_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 169 elseif has('ebcdic') 170 let fnamechars_ok = '#$%+,-./=ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 171 else 172 let fnamechars_ok = '#$%+,-./0123456789=ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' 173 endif 174 175 call assert_equal(identchars_ok, identchars) 176 call assert_equal(kwordchars_ok, kwordchars) 177 call assert_equal(fnamechars_ok, fnamechars) 178 179 call assert_equal(identchars1, identchars) 180 call assert_equal(kwordchars1, kwordchars) 181 call assert_equal(fnamechars1, fnamechars) 182endfunc 183 184func Test_classes_re1() 185 set re=1 186 call s:classes_test() 187 set re=0 188endfunc 189 190func Test_classes_re2() 191 set re=2 192 call s:classes_test() 193 set re=0 194endfunc 195 196func Test_reversed_range() 197 for re in range(0, 2) 198 exe 'set re=' . re 199 call assert_fails('call match("abc def", "[c-a]")', 'E944:') 200 endfor 201 set re=0 202endfunc 203 204func Test_large_class() 205 set re=1 206 call assert_fails('call match("abc def", "[\u3000-\u4000]")', 'E945:') 207 set re=2 208 call assert_equal(0, 'abc def' =~# '[\u3000-\u4000]') 209 call assert_equal(1, "\u3042" =~# '[\u3000-\u4000]') 210 set re=0 211endfunc 212 213func Test_optmatch_toolong() 214 set re=1 215 " Can only handle about 8000 characters. 216 let pat = '\\%[' .. repeat('x', 9000) .. ']' 217 call assert_fails('call match("abc def", "' .. pat .. '")', 'E339:') 218 set re=0 219endfunc 220 221" Test for regexp patterns with multi-byte support, using utf-8. 222func Test_multibyte_chars() 223 " tl is a List of Lists with: 224 " 2: test auto/old/new 0: test auto/old 1: test auto/new 225 " regexp pattern 226 " text to test the pattern on 227 " expected match (optional) 228 " expected submatch 1 (optional) 229 " expected submatch 2 (optional) 230 " etc. 231 " When there is no match use only the first two items. 232 let tl = [] 233 234 " Multi-byte character tests. These will fail unless vim is compiled 235 " with Multibyte (FEAT_MBYTE) or BIG/HUGE features. 236 call add(tl, [2, '[[:alpha:][=a=]]\+', '879 aiaãâaiuvna ', 'aiaãâaiuvna']) 237 call add(tl, [2, '[[=a=]]\+', 'ddaãâbcd', 'aãâ']) " equivalence classes 238 call add(tl, [2, '[^ม ]\+', 'มม oijasoifjos ifjoisj f osij j มมมมม abcd', 'oijasoifjos']) 239 call add(tl, [2, ' [^ ]\+', 'start มabcdม ', ' มabcdม']) 240 call add(tl, [2, '[ม[:alpha:][=a=]]\+', '879 aiaãมâมaiuvna ', 'aiaãมâมaiuvna']) 241 242 " this is not a normal "i" but 0xec 243 call add(tl, [2, '\p\+', 'ìa', 'ìa']) 244 call add(tl, [2, '\p*', 'aあ', 'aあ']) 245 246 " Test recognition of some character classes 247 call add(tl, [2, '\i\+', '&*¨xx ', 'xx']) 248 call add(tl, [2, '\f\+', '&*Ÿfname ', 'fname']) 249 250 " Test composing character matching 251 call add(tl, [2, '.ม', 'xม่x yมy', 'yม']) 252 call add(tl, [2, '.ม่', 'xม่x yมy', 'xม่']) 253 call add(tl, [2, "\u05b9", " x\u05b9 ", "x\u05b9"]) 254 call add(tl, [2, ".\u05b9", " x\u05b9 ", "x\u05b9"]) 255 call add(tl, [2, "\u05b9\u05bb", " x\u05b9\u05bb ", "x\u05b9\u05bb"]) 256 call add(tl, [2, ".\u05b9\u05bb", " x\u05b9\u05bb ", "x\u05b9\u05bb"]) 257 call add(tl, [2, "\u05bb\u05b9", " x\u05b9\u05bb ", "x\u05b9\u05bb"]) 258 call add(tl, [2, ".\u05bb\u05b9", " x\u05b9\u05bb ", "x\u05b9\u05bb"]) 259 call add(tl, [2, "\u05b9", " y\u05bb x\u05b9 ", "x\u05b9"]) 260 call add(tl, [2, ".\u05b9", " y\u05bb x\u05b9 ", "x\u05b9"]) 261 call add(tl, [2, "\u05b9", " y\u05bb\u05b9 x\u05b9 ", "y\u05bb\u05b9"]) 262 call add(tl, [2, ".\u05b9", " y\u05bb\u05b9 x\u05b9 ", "y\u05bb\u05b9"]) 263 call add(tl, [1, "\u05b9\u05bb", " y\u05b9 x\u05b9\u05bb ", "x\u05b9\u05bb"]) 264 call add(tl, [2, ".\u05b9\u05bb", " y\u05bb x\u05b9\u05bb ", "x\u05b9\u05bb"]) 265 call add(tl, [2, "a", "ca\u0300t"]) 266 call add(tl, [2, "ca", "ca\u0300t"]) 267 call add(tl, [2, "a\u0300", "ca\u0300t", "a\u0300"]) 268 call add(tl, [2, 'a\%C', "ca\u0300t", "a\u0300"]) 269 call add(tl, [2, 'ca\%C', "ca\u0300t", "ca\u0300"]) 270 call add(tl, [2, 'ca\%Ct', "ca\u0300t", "ca\u0300t"]) 271 272 " Test \Z 273 call add(tl, [2, 'ú\Z', 'x']) 274 call add(tl, [2, 'יהוה\Z', 'יהוה', 'יהוה']) 275 call add(tl, [2, 'יְהוָה\Z', 'יהוה', 'יהוה']) 276 call add(tl, [2, 'יהוה\Z', 'יְהוָה', 'יְהוָה']) 277 call add(tl, [2, 'יְהוָה\Z', 'יְהוָה', 'יְהוָה']) 278 call add(tl, [2, 'יְ\Z', 'וְיַ', 'יַ']) 279 call add(tl, [2, "ק\u200d\u05b9x\\Z", "xק\u200d\u05b9xy", "ק\u200d\u05b9x"]) 280 call add(tl, [2, "ק\u200d\u05b9x\\Z", "xק\u200dxy", "ק\u200dx"]) 281 call add(tl, [2, "ק\u200dx\\Z", "xק\u200d\u05b9xy", "ק\u200d\u05b9x"]) 282 call add(tl, [2, "ק\u200dx\\Z", "xק\u200dxy", "ק\u200dx"]) 283 call add(tl, [2, "\u05b9\\Z", "xyz"]) 284 call add(tl, [2, "\\Z\u05b9", "xyz"]) 285 call add(tl, [2, "\u05b9\\Z", "xy\u05b9z", "y\u05b9"]) 286 call add(tl, [2, "\\Z\u05b9", "xy\u05b9z", "y\u05b9"]) 287 call add(tl, [1, "\u05b9\\+\\Z", "xy\u05b9z\u05b9 ", "y\u05b9z\u05b9"]) 288 call add(tl, [1, "\\Z\u05b9\\+", "xy\u05b9z\u05b9 ", "y\u05b9z\u05b9"]) 289 290 " Combining different tests and features 291 call add(tl, [2, '[^[=a=]]\+', 'ddaãâbcd', 'dd']) 292 293 " Run the tests 294 for t in tl 295 let re = t[0] 296 let pat = t[1] 297 let text = t[2] 298 let matchidx = 3 299 for engine in [0, 1, 2] 300 if engine == 2 && re == 0 || engine == 1 && re == 1 301 continue 302 endif 303 let ®expengine = engine 304 try 305 let l = matchlist(text, pat) 306 catch 307 call assert_report('Error ' . engine . ': pat: \"' . pat . 308 \ '\", text: \"' . text . 309 \ '\", caused an exception: \"' . v:exception . '\"') 310 endtry 311 " check the match itself 312 if len(l) == 0 && len(t) > matchidx 313 call assert_report('Error ' . engine . ': pat: \"' . pat . 314 \ '\", text: \"' . text . 315 \ '\", did not match, expected: \"' . t[matchidx] . '\"') 316 elseif len(l) > 0 && len(t) == matchidx 317 call assert_report('Error ' . engine . ': pat: \"' . pat . 318 \ '\", text: \"' . text . '\", match: \"' . l[0] . 319 \ '\", expected no match') 320 elseif len(t) > matchidx && l[0] != t[matchidx] 321 call assert_report('Error ' . engine . ': pat: \"' . pat . 322 \ '\", text: \"' . text . '\", match: \"' . l[0] . 323 \ '\", expected: \"' . t[matchidx] . '\"') 324 else 325 " Test passed 326 endif 327 if len(l) > 0 328 " check all the nine submatches 329 for i in range(1, 9) 330 if len(t) <= matchidx + i 331 let e = '' 332 else 333 let e = t[matchidx + i] 334 endif 335 if l[i] != e 336 call assert_report('Error ' . engine . ': pat: \"' . pat . 337 \ '\", text: \"' . text . '\", submatch ' . i . 338 \ ': \"' . l[i] . '\", expected: \"' . e . '\"') 339 endif 340 endfor 341 unlet i 342 endif 343 endfor 344 endfor 345 set regexpengine& 346endfunc 347 348" check that 'ambiwidth' does not change the meaning of \p 349func Test_ambiwidth() 350 set regexpengine=1 ambiwidth=single 351 call assert_equal(0, match("\u00EC", '\p')) 352 set regexpengine=1 ambiwidth=double 353 call assert_equal(0, match("\u00EC", '\p')) 354 set regexpengine=2 ambiwidth=single 355 call assert_equal(0, match("\u00EC", '\p')) 356 set regexpengine=2 ambiwidth=double 357 call assert_equal(0, match("\u00EC", '\p')) 358 set regexpengine& ambiwidth& 359endfunc 360 361func Run_regexp_ignore_case() 362 call assert_equal('iIİ', substitute('iIİ', '\([iIİ]\)', '\1', 'g')) 363 364 call assert_equal('iIx', substitute('iIİ', '\c\([İ]\)', 'x', 'g')) 365 call assert_equal('xxİ', substitute('iIİ', '\(i\c\)', 'x', 'g')) 366 call assert_equal('iIx', substitute('iIİ', '\(İ\c\)', 'x', 'g')) 367 call assert_equal('iIx', substitute('iIİ', '\c\(\%u0130\)', 'x', 'g')) 368 call assert_equal('iIx', substitute('iIİ', '\c\([\u0130]\)', 'x', 'g')) 369 call assert_equal('iIx', substitute('iIİ', '\c\([\u012f-\u0131]\)', 'x', 'g')) 370endfunc 371 372func Test_regexp_ignore_case() 373 set regexpengine=1 374 call Run_regexp_ignore_case() 375 set regexpengine=2 376 call Run_regexp_ignore_case() 377 set regexpengine& 378endfunc 379 380" Tests for regexp with multi-byte encoding and various magic settings 381func Run_regexp_multibyte_magic() 382 let text =<< trim END 383 1 a aa abb abbccc 384 2 d dd dee deefff 385 3 g gg ghh ghhiii 386 4 j jj jkk jkklll 387 5 m mm mnn mnnooo 388 6 x ^aa$ x 389 7 (a)(b) abbaa 390 8 axx [ab]xx 391 9 หม่x อมx 392 a อมx หม่x 393 b ちカヨは 394 c x ¬€x 395 d 天使x 396 e ������y 397 f ������z 398 g a啷bb 399 j 0123❤x 400 k combinations 401 l äö üᾱ̆́ 402 END 403 404 new 405 call setline(1, text) 406 exe 'normal /a*b\{2}c\+/e' .. "\<CR>x" 407 call assert_equal('1 a aa abb abbcc', getline('.')) 408 exe 'normal /\Md\*e\{2}f\+/e' .. "\<CR>x" 409 call assert_equal('2 d dd dee deeff', getline('.')) 410 set nomagic 411 exe 'normal /g\*h\{2}i\+/e' .. "\<CR>x" 412 call assert_equal('3 g gg ghh ghhii', getline('.')) 413 exe 'normal /\mj*k\{2}l\+/e' .. "\<CR>x" 414 call assert_equal('4 j jj jkk jkkll', getline('.')) 415 exe 'normal /\vm*n{2}o+/e' .. "\<CR>x" 416 call assert_equal('5 m mm mnn mnnoo', getline('.')) 417 exe 'normal /\V^aa$/' .. "\<CR>x" 418 call assert_equal('6 x aa$ x', getline('.')) 419 set magic 420 exe 'normal /\v(a)(b)\2\1\1/e' .. "\<CR>x" 421 call assert_equal('7 (a)(b) abba', getline('.')) 422 exe 'normal /\V[ab]\(\[xy]\)\1' .. "\<CR>x" 423 call assert_equal('8 axx ab]xx', getline('.')) 424 425 " search for multi-byte without composing char 426 exe 'normal /ม' .. "\<CR>x" 427 call assert_equal('9 หม่x อx', getline('.')) 428 429 " search for multi-byte with composing char 430 exe 'normal /ม่' .. "\<CR>x" 431 call assert_equal('a อมx หx', getline('.')) 432 433 " find word by change of word class 434 exe 'normal /ち\<カヨ\>は' .. "\<CR>x" 435 call assert_equal('b カヨは', getline('.')) 436 437 " Test \%u, [\u] and friends 438 " c 439 exe 'normal /\%u20ac' .. "\<CR>x" 440 call assert_equal('c x ¬x', getline('.')) 441 " d 442 exe 'normal /[\u4f7f\u5929]\+' .. "\<CR>x" 443 call assert_equal('d 使x', getline('.')) 444 " e 445 exe 'normal /\%U12345678' .. "\<CR>x" 446 call assert_equal('e y', getline('.')) 447 " f 448 exe 'normal /[\U1234abcd\u1234\uabcd]' .. "\<CR>x" 449 call assert_equal('f z', getline('.')) 450 " g 451 exe 'normal /\%d21879b' .. "\<CR>x" 452 call assert_equal('g abb', getline('.')) 453 454 " j Test backwards search from a multi-byte char 455 exe "normal /x\<CR>x?.\<CR>x" 456 call assert_equal('j 012❤', getline('.')) 457 " k 458 let @w=':%s#comb[i]nations#œ̄ṣ́m̥̄ᾱ̆́#g' 459 @w 460 call assert_equal('k œ̄ṣ́m̥̄ᾱ̆́', getline(18)) 461 462 close! 463endfunc 464 465func Test_regexp_multibyte_magic() 466 set regexpengine=1 467 call Run_regexp_multibyte_magic() 468 set regexpengine=2 469 call Run_regexp_multibyte_magic() 470 set regexpengine& 471endfunc 472 473" Test for 7.3.192 474" command ":s/ \?/ /g" splits multi-byte characters into bytes 475func Test_split_multibyte_to_bytes() 476 new 477 call setline(1, 'l äö üᾱ̆́') 478 s/ \?/ /g 479 call assert_equal(' l ä ö ü ᾱ̆́', getline(1)) 480 close! 481endfunc 482 483" Test for matchstr() with multibyte characters 484func Test_matchstr_multibyte() 485 new 486 call assert_equal('ב', matchstr("אבגד", ".", 0, 2)) 487 call assert_equal('בג', matchstr("אבגד", "..", 0, 2)) 488 call assert_equal('א', matchstr("אבגד", ".", 0, 0)) 489 call assert_equal('ג', matchstr("אבגד", ".", 4, -1)) 490 close! 491endfunc 492 493" Test for 7.4.636 494" A search with end offset gets stuck at end of file. 495func Test_search_with_end_offset() 496 new 497 call setline(1, ['', 'dog(a', 'cat(']) 498 exe "normal /(/e+" .. "\<CR>" 499 normal "ayn 500 call assert_equal("a\ncat(", @a) 501 close! 502endfunc 503 504" vim: shiftwidth=2 sts=2 expandtab 505