blob: f618870315f9bdb0f2f219512ee7ab6351c32b4b [file] [log] [blame]
Bram Moolenaar22e42152016-04-03 14:02:02 +02001" Tests for regexp in utf8 encoding
Bram Moolenaar22e42152016-04-03 14:02:02 +02002
3func s:equivalence_test()
4 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"
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()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +020025 set re=0
Bram Moolenaar22e42152016-04-03 14:02:02 +020026endfunc
27
28func Test_equivalence_re2()
29 set re=2
30 call s:equivalence_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +020031 set re=0
Bram Moolenaar22e42152016-04-03 14:02:02 +020032endfunc
Bram Moolenaaraf98a492016-04-24 14:40:12 +020033
34func s:classes_test()
Bram Moolenaar490465b2016-04-24 15:11:02 +020035 set isprint=@,161-255
Bram Moolenaaraf98a492016-04-24 14:40:12 +020036 call assert_equal('Motörhead', matchstr('Motörhead', '[[:print:]]\+'))
37
Bram Moolenaaraf98a492016-04-24 14:40:12 +020038 let alnumchars = ''
Bram Moolenaar0c078fc2017-03-29 15:31:20 +020039 let alphachars = ''
40 let backspacechar = ''
41 let blankchars = ''
42 let cntrlchars = ''
43 let digitchars = ''
44 let escapechar = ''
45 let graphchars = ''
46 let lowerchars = ''
Bram Moolenaaraf98a492016-04-24 14:40:12 +020047 let printchars = ''
48 let punctchars = ''
Bram Moolenaar0c078fc2017-03-29 15:31:20 +020049 let returnchar = ''
50 let spacechars = ''
51 let tabchar = ''
52 let upperchars = ''
Bram Moolenaaraf98a492016-04-24 14:40:12 +020053 let xdigitchars = ''
54 let i = 1
55 while i <= 255
56 let c = nr2char(i)
57 if c =~ '[[:alpha:]]'
58 let alphachars .= c
59 endif
Bram Moolenaaraf98a492016-04-24 14:40:12 +020060 if c =~ '[[:alnum:]]'
61 let alnumchars .= c
62 endif
Bram Moolenaar0c078fc2017-03-29 15:31:20 +020063 if c =~ '[[:backspace:]]'
64 let backspacechar .= c
65 endif
66 if c =~ '[[:blank:]]'
67 let blankchars .= c
68 endif
69 if c =~ '[[:cntrl:]]'
70 let cntrlchars .= c
71 endif
72 if c =~ '[[:digit:]]'
73 let digitchars .= c
74 endif
75 if c =~ '[[:escape:]]'
76 let escapechar .= c
77 endif
78 if c =~ '[[:graph:]]'
79 let graphchars .= c
80 endif
81 if c =~ '[[:lower:]]'
82 let lowerchars .= c
83 endif
Bram Moolenaaraf98a492016-04-24 14:40:12 +020084 if c =~ '[[:print:]]'
85 let printchars .= c
86 endif
87 if c =~ '[[:punct:]]'
88 let punctchars .= c
89 endif
Bram Moolenaar0c078fc2017-03-29 15:31:20 +020090 if c =~ '[[:return:]]'
91 let returnchar .= c
92 endif
93 if c =~ '[[:space:]]'
94 let spacechars .= c
95 endif
96 if c =~ '[[:tab:]]'
97 let tabchar .= c
98 endif
99 if c =~ '[[:upper:]]'
100 let upperchars .= c
101 endif
Bram Moolenaaraf98a492016-04-24 14:40:12 +0200102 if c =~ '[[:xdigit:]]'
103 let xdigitchars .= c
104 endif
105 let i += 1
106 endwhile
107
108 call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alphachars)
Bram Moolenaaraf98a492016-04-24 14:40:12 +0200109 call assert_equal('0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz', alnumchars)
Bram Moolenaar0c078fc2017-03-29 15:31:20 +0200110 call assert_equal("\b", backspacechar)
111 call assert_equal("\t ", blankchars)
Bram Moolenaar13489b92017-03-30 22:20:29 +0200112 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 Moolenaar0c078fc2017-03-29 15:31:20 +0200113 call assert_equal("0123456789", digitchars)
114 call assert_equal("\<Esc>", escapechar)
Bram Moolenaar13489b92017-03-30 22:20:29 +0200115 call assert_equal('!"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~', graphchars)
Bram Moolenaar0c078fc2017-03-29 15:31:20 +0200116 call assert_equal('abcdefghijklmnopqrstuvwxyzµßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', lowerchars)
Bram Moolenaaraf98a492016-04-24 14:40:12 +0200117 call assert_equal(' !"#$%&''()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ ¡¢£¤¥¦§¨©ª«¬­®¯°±²³´µ¶·¸¹º»¼½¾¿ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖרÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõö÷øùúûüýþÿ', printchars)
118 call assert_equal('!"#$%&''()*+,-./:;<=>?@[\]^_`{|}~', punctchars)
Bram Moolenaar0c078fc2017-03-29 15:31:20 +0200119 call assert_equal('ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ', upperchars)
120 call assert_equal("\r", returnchar)
121 call assert_equal("\t\n\x0b\f\r ", spacechars)
122 call assert_equal("\t", tabchar)
Bram Moolenaaraf98a492016-04-24 14:40:12 +0200123 call assert_equal('0123456789ABCDEFabcdef', xdigitchars)
124endfunc
125
126func Test_classes_re1()
127 set re=1
128 call s:classes_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +0200129 set re=0
Bram Moolenaaraf98a492016-04-24 14:40:12 +0200130endfunc
131
132func Test_classes_re2()
133 set re=2
134 call s:classes_test()
Bram Moolenaar6bff02e2016-08-16 22:50:55 +0200135 set re=0
136endfunc
Bram Moolenaar966e58e2017-06-05 16:54:08 +0200137
138func Test_reversed_range()
139 for re in range(0, 2)
140 exe 'set re=' . re
141 call assert_fails('call match("abc def", "[c-a]")', 'E944:')
142 endfor
143 set re=0
144endfunc
145
146func Test_large_class()
147 set re=1
148 call assert_fails('call match("abc def", "[\u3000-\u4000]")', 'E945:')
149 set re=2
150 call assert_equal(0, 'abc def' =~# '[\u3000-\u4000]')
151 call assert_equal(1, "\u3042" =~# '[\u3000-\u4000]')
152 set re=0
153endfunc