| Bram Moolenaar | 22e4215 | 2016-04-03 14:02:02 +0200 | [diff] [blame] | 1 | " Tests for regexp in utf8 encoding |
| 2 | if !has('multi_byte') |
| 3 | finish |
| 4 | endif |
| 5 | set encoding=utf-8 |
| 6 | scriptencoding utf-8 |
| 7 | |
| 8 | func s:equivalence_test() |
| 9 | 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" |
| 10 | let groups = split(str) |
| 11 | for group1 in groups |
| 12 | for c in split(group1, '\zs') |
| 13 | " next statement confirms that equivalence class matches every |
| 14 | " character in group |
| 15 | call assert_match('^[[=' . c . '=]]*$', group1) |
| 16 | for group2 in groups |
| 17 | if group2 != group1 |
| 18 | " next statement converts that equivalence class doesn't match |
| 19 | " character in any other group |
| 20 | call assert_equal(-1, match(group2, '[[=' . c . '=]]')) |
| 21 | endif |
| 22 | endfor |
| 23 | endfor |
| 24 | endfor |
| 25 | endfunc |
| 26 | |
| 27 | func Test_equivalence_re1() |
| 28 | set re=1 |
| 29 | call s:equivalence_test() |
| 30 | endfunc |
| 31 | |
| 32 | func Test_equivalence_re2() |
| 33 | set re=2 |
| 34 | call s:equivalence_test() |
| 35 | endfunc |
| Bram Moolenaar | af98a49 | 2016-04-24 14:40:12 +0200 | [diff] [blame] | 36 | |
| 37 | func s:classes_test() |
| 38 | call assert_equal('Motörhead', matchstr('Motörhead', '[[:print:]]\+')) |
| 39 | |
| 40 | let alphachars = '' |
| 41 | let lowerchars = '' |
| 42 | let upperchars = '' |
| 43 | let alnumchars = '' |
| 44 | let printchars = '' |
| 45 | let punctchars = '' |
| 46 | let xdigitchars = '' |
| 47 | let i = 1 |
| 48 | while i <= 255 |
| 49 | let c = nr2char(i) |
| 50 | if c =~ '[[:alpha:]]' |
| 51 | let alphachars .= c |
| 52 | endif |
| 53 | if c =~ '[[:lower:]]' |
| 54 | let lowerchars .= c |
| 55 | endif |
| 56 | if c =~ '[[:upper:]]' |
| 57 | let upperchars .= c |
| 58 | endif |
| 59 | if c =~ '[[:alnum:]]' |
| 60 | let alnumchars .= c |
| 61 | endif |
| 62 | if c =~ '[[:print:]]' |
| 63 | let printchars .= c |
| 64 | endif |
| 65 | if c =~ '[[:punct:]]' |
| 66 | let punctchars .= c |
| 67 | endif |
| 68 | if c =~ '[[:xdigit:]]' |
| 69 | let xdigitchars .= c |
| 70 | endif |
| 71 | let i += 1 |
| 72 | endwhile |
| 73 | |
| 74 | call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars) |
| 75 | call assert_equal('abcdefghijklmnopqrstuvwxyzµßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', lowerchars) |
| 76 | call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ', upperchars) |
| 77 | call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars) |
| 78 | call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖ×ØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ', printchars) |
| 79 | call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars) |
| 80 | call assert_equal('0123456789ABCDEFabcdef', xdigitchars) |
| 81 | endfunc |
| 82 | |
| 83 | func Test_classes_re1() |
| 84 | set re=1 |
| 85 | call s:classes_test() |
| 86 | endfunc |
| 87 | |
| 88 | func Test_classes_re2() |
| 89 | set re=2 |
| 90 | call s:classes_test() |
| 91 | endfunc |