blob: d2259835ca44d10e46eba1d1ce3f1cb138a153fd [file] [log] [blame]
Bram Moolenaar22e42152016-04-03 14:02:02 +02001" Tests for regexp in utf8 encoding
2if !has('multi_byte')
3 finish
4endif
Bram Moolenaar22e42152016-04-03 14:02:02 +02005
6func s:equivalence_test()
7 let str = "AÀÁÂÃÄÅĀĂĄǍǞǠẢ BḂḆ CÇĆĈĊČ DĎĐḊḎḐ EÈÉÊËĒĔĖĘĚẺẼ F GĜĞĠĢǤǦǴḠ HĤĦḢḦḨ IÌÍÎÏĨĪĬĮİǏỈ JĴ KĶǨḰḴ LĹĻĽĿŁḺ MḾṀ NÑŃŅŇṄṈ OÒÓÔÕÖØŌŎŐƠǑǪǬỎ PṔṖ Q RŔŖŘṘṞ SŚŜŞŠṠ TŢŤŦṪṮ UÙÚÛÜŨŪŬŮŰŲƯǓỦ V WŴẀẂẄẆ XẊẌ YÝŶŸẎỲỶỸ ZŹŻŽƵẐẔ aàáâãäåāăąǎǟǡả bḃḇ cçćĉċč dďđḋḏḑ eèéêëēĕėęěẻẽ f gĝğġģǥǧǵḡ hĥħḣḧḩẖ iìíîïĩīĭįǐỉ jĵǰ kķǩḱḵ lĺļľŀłḻ mḿṁ nñńņňʼnṅṉ oòóôõöøōŏőơǒǫǭỏ pṕṗ q rŕŗřṙṟ sśŝşšṡ tţťŧṫṯẗ uùúûüũūŭůűųưǔủ v wŵẁẃẅẇẘ xẋẍ yýÿŷẏẙỳỷỹ zźżžƶẑẕ"
8 let groups = split(str)
9 for group1 in groups
10 for c in split(group1, '\zs')
11 " next statement confirms that equivalence class matches every
12 " character in group
13 call assert_match('^[[=' . c . '=]]*$', group1)
14 for group2 in groups
15 if group2 != group1
16 " next statement converts that equivalence class doesn't match
17 " character in any other group
18 call assert_equal(-1, match(group2, '[[=' . c . '=]]'))
19 endif
20 endfor
21 endfor
22 endfor
23endfunc
24
25func Test_equivalence_re1()
26 set re=1
27 call s:equivalence_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +020028 set re=0
Bram Moolenaar22e42152016-04-03 14:02:02 +020029endfunc
30
31func Test_equivalence_re2()
32 set re=2
33 call s:equivalence_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +020034 set re=0
Bram Moolenaar22e42152016-04-03 14:02:02 +020035endfunc
Bram Moolenaaraf98a492016-04-24 14:40:12 +020036
37func s:classes_test()
Bram Moolenaar490465b2016-04-24 15:11:02 +020038 set isprint=@,161-255
Bram Moolenaaraf98a492016-04-24 14:40:12 +020039 call assert_equal('Motörhead', matchstr('Motörhead', '[[:print:]]\+'))
40
41 let alphachars = ''
42 let lowerchars = ''
43 let upperchars = ''
44 let alnumchars = ''
45 let printchars = ''
46 let punctchars = ''
47 let xdigitchars = ''
48 let i = 1
49 while i <= 255
50 let c = nr2char(i)
51 if c =~ '[[:alpha:]]'
52 let alphachars .= c
53 endif
54 if c =~ '[[:lower:]]'
55 let lowerchars .= c
56 endif
57 if c =~ '[[:upper:]]'
58 let upperchars .= c
59 endif
60 if c =~ '[[:alnum:]]'
61 let alnumchars .= c
62 endif
63 if c =~ '[[:print:]]'
64 let printchars .= c
65 endif
66 if c =~ '[[:punct:]]'
67 let punctchars .= c
68 endif
69 if c =~ '[[:xdigit:]]'
70 let xdigitchars .= c
71 endif
72 let i += 1
73 endwhile
74
75 call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars)
76 call assert_equal('abcdefghijklmnopqrstuvwxyzµßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', lowerchars)
77 call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ', upperchars)
78 call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars)
79 call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬­®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ', printchars)
80 call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars)
81 call assert_equal('0123456789ABCDEFabcdef', xdigitchars)
82endfunc
83
84func Test_classes_re1()
85 set re=1
86 call s:classes_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +020087 set re=0
Bram Moolenaaraf98a492016-04-24 14:40:12 +020088endfunc
89
90func Test_classes_re2()
91 set re=2
92 call s:classes_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +020093 set re=0
94endfunc