Bram Moolenaar | 22e4215 | 2016-04-03 14:02:02 +0200 | [diff] [blame] | 1 | " Tests for regexp in utf8 encoding |
Bram Moolenaar | 22e4215 | 2016-04-03 14:02:02 +0200 | [diff] [blame] | 2 | |
| 3 | func 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 |
| 20 | endfunc |
| 21 | |
| 22 | func Test_equivalence_re1() |
| 23 | set re=1 |
| 24 | call s:equivalence_test() |
Bram Moolenaar | 6bff02e | 2016-08-16 22:50:55 +0200 | [diff] [blame] | 25 | set re=0 |
Bram Moolenaar | 22e4215 | 2016-04-03 14:02:02 +0200 | [diff] [blame] | 26 | endfunc |
| 27 | |
| 28 | func Test_equivalence_re2() |
| 29 | set re=2 |
| 30 | call s:equivalence_test() |
Bram Moolenaar | 6bff02e | 2016-08-16 22:50:55 +0200 | [diff] [blame] | 31 | set re=0 |
Bram Moolenaar | 22e4215 | 2016-04-03 14:02:02 +0200 | [diff] [blame] | 32 | endfunc |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 33 | |
| 34 | func s:classes_test() |
Bram Moolenaar | 490465b | 2016-04-24 15:11:02 +0200 | [diff] [blame] | 35 | set isprint=@,161-255 |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 36 | call assert_equal('Motörhead', matchstr('Motörhead', '[[:print:]]\+')) |
| 37 | |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 38 | let alnumchars = '' |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 39 | let alphachars = '' |
| 40 | let backspacechar = '' |
| 41 | let blankchars = '' |
| 42 | let cntrlchars = '' |
| 43 | let digitchars = '' |
| 44 | let escapechar = '' |
| 45 | let graphchars = '' |
| 46 | let lowerchars = '' |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 47 | let printchars = '' |
| 48 | let punctchars = '' |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 49 | let returnchar = '' |
| 50 | let spacechars = '' |
| 51 | let tabchar = '' |
| 52 | let upperchars = '' |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 53 | let xdigitchars = '' |
Bram Moolenaar | 221cd9f | 2019-01-31 15:34:40 +0100 | [diff] [blame^] | 54 | let identchars = '' |
| 55 | let identchars1 = '' |
| 56 | let kwordchars = '' |
| 57 | let kwordchars1 = '' |
| 58 | let fnamechars = '' |
| 59 | let fnamechars1 = '' |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 60 | let i = 1 |
| 61 | while i <= 255 |
| 62 | let c = nr2char(i) |
| 63 | if c =~ '[[:alpha:]]' |
| 64 | let alphachars .= c |
| 65 | endif |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 66 | if c =~ '[[:alnum:]]' |
| 67 | let alnumchars .= c |
| 68 | endif |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 69 | if c =~ '[[:backspace:]]' |
| 70 | let backspacechar .= c |
| 71 | endif |
| 72 | if c =~ '[[:blank:]]' |
| 73 | let blankchars .= c |
| 74 | endif |
| 75 | if c =~ '[[:cntrl:]]' |
| 76 | let cntrlchars .= c |
| 77 | endif |
| 78 | if c =~ '[[:digit:]]' |
| 79 | let digitchars .= c |
| 80 | endif |
| 81 | if c =~ '[[:escape:]]' |
| 82 | let escapechar .= c |
| 83 | endif |
| 84 | if c =~ '[[:graph:]]' |
| 85 | let graphchars .= c |
| 86 | endif |
| 87 | if c =~ '[[:lower:]]' |
| 88 | let lowerchars .= c |
| 89 | endif |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 90 | if c =~ '[[:print:]]' |
| 91 | let printchars .= c |
| 92 | endif |
| 93 | if c =~ '[[:punct:]]' |
| 94 | let punctchars .= c |
| 95 | endif |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 96 | if c =~ '[[:return:]]' |
| 97 | let returnchar .= c |
| 98 | endif |
| 99 | if c =~ '[[:space:]]' |
| 100 | let spacechars .= c |
| 101 | endif |
| 102 | if c =~ '[[:tab:]]' |
| 103 | let tabchar .= c |
| 104 | endif |
| 105 | if c =~ '[[:upper:]]' |
| 106 | let upperchars .= c |
| 107 | endif |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 108 | if c =~ '[[:xdigit:]]' |
| 109 | let xdigitchars .= c |
| 110 | endif |
Bram Moolenaar | 221cd9f | 2019-01-31 15:34:40 +0100 | [diff] [blame^] | 111 | if c =~ '[[:ident:]]' |
| 112 | let identchars .= c |
| 113 | endif |
| 114 | if c =~ '\i' |
| 115 | let identchars1 .= c |
| 116 | endif |
| 117 | if c =~ '[[:keyword:]]' |
| 118 | let kwordchars .= c |
| 119 | endif |
| 120 | if c =~ '\k' |
| 121 | let kwordchars1 .= c |
| 122 | endif |
| 123 | if c =~ '[[:fname:]]' |
| 124 | let fnamechars .= c |
| 125 | endif |
| 126 | if c =~ '\f' |
| 127 | let fnamechars1 .= c |
| 128 | endif |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 129 | let i += 1 |
| 130 | endwhile |
| 131 | |
| 132 | call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars) |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 133 | call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars) |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 134 | call assert_equal("\b", backspacechar) |
| 135 | call assert_equal("\t ", blankchars) |
Bram Moolenaar | 13489b9 | 2017-03-30 22:20:29 +0200 | [diff] [blame] | 136 | 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) |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 137 | call assert_equal("0123456789", digitchars) |
| 138 | call assert_equal("\<Esc>", escapechar) |
Bram Moolenaar | 13489b9 | 2017-03-30 22:20:29 +0200 | [diff] [blame] | 139 | call assert_equal('!"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~', graphchars) |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 140 | call assert_equal('abcdefghijklmnopqrstuvwxyzµßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', lowerchars) |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 141 | call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ', printchars) |
| 142 | call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars) |
Bram Moolenaar | 0c078fc | 2017-03-29 15:31:20 +0200 | [diff] [blame] | 143 | call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ', upperchars) |
| 144 | call assert_equal("\r", returnchar) |
| 145 | call assert_equal("\t\n\x0b\f\r ", spacechars) |
| 146 | call assert_equal("\t", tabchar) |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 147 | call assert_equal('0123456789ABCDEFabcdef', xdigitchars) |
Bram Moolenaar | 221cd9f | 2019-01-31 15:34:40 +0100 | [diff] [blame^] | 148 | |
| 149 | if has('win32') |
| 150 | let identchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz
¡¢£¤¥¦§µÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ' |
| 151 | let kwordchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyzµÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 152 | elseif has('ebcdic') |
| 153 | let identchars_ok = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz¬®µº¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 154 | let kwordchars_ok = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz¬®µº¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 155 | else |
| 156 | let identchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyzµÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 157 | let kwordchars_ok = '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyzµÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 158 | endif |
| 159 | |
| 160 | if has('win32') |
| 161 | let fnamechars_ok = '!#$%+,-./0123456789:=@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]_abcdefghijklmnopqrstuvwxyz{}~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 162 | elseif has('amiga') |
| 163 | let fnamechars_ok = '$+,-./0123456789:ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 164 | elseif has('vms') |
| 165 | let fnamechars_ok = '#$%+,-./0123456789:;<>ABCDEFGHIJKLMNOPQRSTUVWXYZ[]_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 166 | elseif has('ebcdic') |
| 167 | let fnamechars_ok = '#$%+,-./=ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 168 | else |
| 169 | let fnamechars_ok = '#$%+,-./0123456789=ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ' |
| 170 | endif |
| 171 | |
| 172 | call assert_equal(identchars_ok, identchars) |
| 173 | call assert_equal(kwordchars_ok, kwordchars) |
| 174 | call assert_equal(fnamechars_ok, fnamechars) |
| 175 | |
| 176 | call assert_equal(identchars1, identchars) |
| 177 | call assert_equal(kwordchars1, kwordchars) |
| 178 | call assert_equal(fnamechars1, fnamechars) |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 179 | endfunc |
| 180 | |
| 181 | func Test_classes_re1() |
| 182 | set re=1 |
| 183 | call s:classes_test() |
Bram Moolenaar | 6bff02e | 2016-08-16 22:50:55 +0200 | [diff] [blame] | 184 | set re=0 |
Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 185 | endfunc |
| 186 | |
| 187 | func Test_classes_re2() |
| 188 | set re=2 |
| 189 | call s:classes_test() |
Bram Moolenaar | 6bff02e | 2016-08-16 22:50:55 +0200 | [diff] [blame] | 190 | set re=0 |
| 191 | endfunc |
Bram Moolenaar | 966e58e | 2017-06-05 16:54:08 +0200 | [diff] [blame] | 192 | |
| 193 | func Test_reversed_range() |
| 194 | for re in range(0, 2) |
| 195 | exe 'set re=' . re |
| 196 | call assert_fails('call match("abc def", "[c-a]")', 'E944:') |
| 197 | endfor |
| 198 | set re=0 |
| 199 | endfunc |
| 200 | |
| 201 | func Test_large_class() |
| 202 | set re=1 |
| 203 | call assert_fails('call match("abc def", "[\u3000-\u4000]")', 'E945:') |
| 204 | set re=2 |
| 205 | call assert_equal(0, 'abc def' =~# '[\u3000-\u4000]') |
| 206 | call assert_equal(1, "\u3042" =~# '[\u3000-\u4000]') |
| 207 | set re=0 |
| 208 | endfunc |