blob: 112b5a6e98e67f8f60303cdf8a66f2ed94304621 [file] [log] [blame]
Bram Moolenaar0c0590d2017-01-28 13:48:10 +01001" Tests for Unicode manipulations
Bram Moolenaar94722c52023-01-28 19:19:03 +00002
zeertzjq6cac7702022-01-04 18:01:21 +00003source check.vim
Bram Moolenaar2912abb2019-03-29 14:16:42 +01004source view_util.vim
Yasuhiro Matsumoto2bc849f2023-01-10 16:03:08 +00005source screendump.vim
Bram Moolenaar0c0590d2017-01-28 13:48:10 +01006
7" Visual block Insert adjusts for multi-byte char
8func Test_visual_block_insert()
9 new
10 call setline(1, ["aaa", "あああ", "bbb"])
11 exe ":norm! gg0l\<C-V>jjIx\<Esc>"
Bram Moolenaarfc6cceb2022-01-20 12:22:35 +000012 call assert_equal(['axaa', ' xあああ', 'bxbb'], getline(1, '$'))
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010013 bwipeout!
14endfunc
15
Bram Moolenaar70ce8a12021-03-14 19:02:09 +010016" Test for built-in functions strchars() and strcharlen()
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010017func Test_strchars()
18 let inp = ["a", "あいa", "A\u20dd", "A\u20dd\u20dd", "\u20dd"]
19 let exp = [[1, 1, 1], [3, 3, 3], [2, 2, 1], [3, 3, 1], [1, 1, 1]]
20 for i in range(len(inp))
21 call assert_equal(exp[i][0], strchars(inp[i]))
Bram Moolenaarf6ed61e2019-09-07 19:05:09 +020022 call assert_equal(exp[i][1], inp[i]->strchars(0))
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010023 call assert_equal(exp[i][2], strchars(inp[i], 1))
24 endfor
Bram Moolenaar70ce8a12021-03-14 19:02:09 +010025
26 let exp = [1, 3, 1, 1, 1]
27 for i in range(len(inp))
28 call assert_equal(exp[i], inp[i]->strcharlen())
29 call assert_equal(exp[i], strcharlen(inp[i]))
30 endfor
31
zeertzjq8cf51372023-05-08 15:31:38 +010032 call assert_fails("call strchars('abc', 2)", ['E1023:', 'E1023:'])
33 call assert_fails("call strchars('abc', -1)", ['E1023:', 'E1023:'])
34 call assert_fails("call strchars('abc', {})", ['E728:', 'E728:'])
35 call assert_fails("call strchars('abc', [])", ['E745:', 'E745:'])
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010036endfunc
37
38" Test for customlist completion
Bram Moolenaar1e115362019-01-09 23:01:02 +010039func CustomComplete1(lead, line, pos)
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010040 return ['あ', 'い']
Bram Moolenaar1e115362019-01-09 23:01:02 +010041endfunc
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010042
Bram Moolenaar1e115362019-01-09 23:01:02 +010043func CustomComplete2(lead, line, pos)
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010044 return ['あたし', 'あたま', 'あたりめ']
Bram Moolenaar1e115362019-01-09 23:01:02 +010045endfunc
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010046
Bram Moolenaar1e115362019-01-09 23:01:02 +010047func CustomComplete3(lead, line, pos)
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010048 return ['Nこ', 'Nん', 'Nぶ']
Bram Moolenaar1e115362019-01-09 23:01:02 +010049endfunc
Bram Moolenaar0c0590d2017-01-28 13:48:10 +010050
51func Test_customlist_completion()
52 command -nargs=1 -complete=customlist,CustomComplete1 Test1 echo
53 call feedkeys(":Test1 \<C-L>\<C-B>\"\<CR>", 'itx')
54 call assert_equal('"Test1 ', getreg(':'))
55
56 command -nargs=1 -complete=customlist,CustomComplete2 Test2 echo
57 call feedkeys(":Test2 \<C-L>\<C-B>\"\<CR>", 'itx')
58 call assert_equal('"Test2 あた', getreg(':'))
59
60 command -nargs=1 -complete=customlist,CustomComplete3 Test3 echo
61 call feedkeys(":Test3 \<C-L>\<C-B>\"\<CR>", 'itx')
62 call assert_equal('"Test3 N', getreg(':'))
63
64 call garbagecollect(1)
65endfunc
66
67" Yank one 3 byte character and check the mark columns.
68func Test_getvcol()
69 new
70 call setline(1, "x\u2500x")
71 normal 0lvy
72 call assert_equal(2, col("'["))
73 call assert_equal(4, col("']"))
74 call assert_equal(2, virtcol("'["))
75 call assert_equal(2, virtcol("']"))
76endfunc
Bram Moolenaar2912abb2019-03-29 14:16:42 +010077
Bram Moolenaar9d401282019-04-06 13:18:12 +020078func Test_list2str_str2list_utf8()
79 " One Unicode codepoint
80 let s = "\u3042\u3044"
81 let l = [0x3042, 0x3044]
82 call assert_equal(l, str2list(s, 1))
83 call assert_equal(s, list2str(l, 1))
84 if &enc ==# 'utf-8'
85 call assert_equal(str2list(s), str2list(s, 1))
86 call assert_equal(list2str(l), list2str(l, 1))
87 endif
88
89 " With composing characters
90 let s = "\u304b\u3099\u3044"
91 let l = [0x304b, 0x3099, 0x3044]
92 call assert_equal(l, str2list(s, 1))
Bram Moolenaar02b31112019-08-31 22:16:38 +020093 call assert_equal(s, l->list2str(1))
Bram Moolenaar9d401282019-04-06 13:18:12 +020094 if &enc ==# 'utf-8'
95 call assert_equal(str2list(s), str2list(s, 1))
96 call assert_equal(list2str(l), list2str(l, 1))
97 endif
98
99 " Null list is the same as an empty list
100 call assert_equal('', list2str([]))
101 call assert_equal('', list2str(test_null_list()))
102endfunc
103
104func Test_list2str_str2list_latin1()
105 " When 'encoding' is not multi-byte can still get utf-8 string.
106 " But we need to create the utf-8 string while 'encoding' is utf-8.
107 let s = "\u3042\u3044"
108 let l = [0x3042, 0x3044]
109
110 let save_encoding = &encoding
111 set encoding=latin1
Bram Moolenaar94722c52023-01-28 19:19:03 +0000112
Bram Moolenaar9d401282019-04-06 13:18:12 +0200113 let lres = str2list(s, 1)
114 let sres = list2str(l, 1)
Bram Moolenaar0e05de42020-03-25 22:23:46 +0100115 call assert_equal([65, 66, 67], str2list("ABC"))
Bram Moolenaar9d401282019-04-06 13:18:12 +0200116
Bram Moolenaar08f41572020-04-20 16:50:00 +0200117 " Try converting a list to a string in latin-1 encoding
118 call assert_equal([1, 2, 3], str2list(list2str([1, 2, 3])))
119
Bram Moolenaar9d401282019-04-06 13:18:12 +0200120 let &encoding = save_encoding
121 call assert_equal(l, lres)
122 call assert_equal(s, sres)
123endfunc
124
Bram Moolenaar2912abb2019-03-29 14:16:42 +0100125func Test_screenchar_utf8()
126 new
127
Bram Moolenaar94722c52023-01-28 19:19:03 +0000128 " 1-cell, with composing characters
Bram Moolenaar2912abb2019-03-29 14:16:42 +0100129 call setline(1, ["ABC\u0308"])
130 redraw
131 call assert_equal([0x0041], screenchars(1, 1))
Bram Moolenaar196b4662019-09-06 21:34:30 +0200132 call assert_equal([0x0042], 1->screenchars(2))
Bram Moolenaar2912abb2019-03-29 14:16:42 +0100133 call assert_equal([0x0043, 0x0308], screenchars(1, 3))
134 call assert_equal("A", screenstring(1, 1))
135 call assert_equal("B", screenstring(1, 2))
136 call assert_equal("C\u0308", screenstring(1, 3))
137
zeertzjq47eec672023-06-01 20:26:55 +0100138 " 1-cell, with 6 composing characters
139 set maxcombine=6
140 call setline(1, ["ABC" .. repeat("\u0308", 6)])
141 redraw
142 call assert_equal([0x0041], screenchars(1, 1))
143 call assert_equal([0x0042], 1->screenchars(2))
144 " This should not use uninitialized memory
145 call assert_equal([0x0043] + repeat([0x0308], 6), screenchars(1, 3))
146 call assert_equal("A", screenstring(1, 1))
147 call assert_equal("B", screenstring(1, 2))
148 call assert_equal("C" .. repeat("\u0308", 6), screenstring(1, 3))
149 set maxcombine&
150
Bram Moolenaar94722c52023-01-28 19:19:03 +0000151 " 2-cells, with composing characters
Bram Moolenaar2912abb2019-03-29 14:16:42 +0100152 let text = "\u3042\u3044\u3046\u3099"
153 call setline(1, text)
154 redraw
155 call assert_equal([0x3042], screenchars(1, 1))
156 call assert_equal([0], screenchars(1, 2))
157 call assert_equal([0x3044], screenchars(1, 3))
158 call assert_equal([0], screenchars(1, 4))
159 call assert_equal([0x3046, 0x3099], screenchars(1, 5))
160
161 call assert_equal("\u3042", screenstring(1, 1))
162 call assert_equal("", screenstring(1, 2))
163 call assert_equal("\u3044", screenstring(1, 3))
164 call assert_equal("", screenstring(1, 4))
165 call assert_equal("\u3046\u3099", screenstring(1, 5))
166
Bram Moolenaar48aed082019-03-30 15:44:17 +0100167 call assert_equal([text . ' '], ScreenLines(1, 8))
Bram Moolenaar2912abb2019-03-29 14:16:42 +0100168
169 bwipe!
170endfunc
Bram Moolenaar0e05de42020-03-25 22:23:46 +0100171
Bram Moolenaar08aac3c2020-08-28 21:04:24 +0200172func Test_setcellwidths()
173 call setcellwidths([
174 \ [0x1330, 0x1330, 2],
Bram Moolenaar08aac3c2020-08-28 21:04:24 +0200175 \ [9999, 10000, 1],
Bram Moolenaarb06a6d52020-08-28 23:27:20 +0200176 \ [0x1337, 0x1339, 2],
Bram Moolenaar08aac3c2020-08-28 21:04:24 +0200177 \])
178
179 call assert_equal(2, strwidth("\u1330"))
180 call assert_equal(1, strwidth("\u1336"))
181 call assert_equal(2, strwidth("\u1337"))
182 call assert_equal(2, strwidth("\u1339"))
183 call assert_equal(1, strwidth("\u133a"))
184
K.Takata71933232023-01-20 16:00:55 +0000185 for aw in ['single', 'double']
186 exe 'set ambiwidth=' . aw
187 " Handle \u0080 to \u009F as control chars even on MS-Windows.
188 set isprint=@,161-255
189
190 call setcellwidths([])
191 " Control chars
192 call assert_equal(4, strwidth("\u0081"))
193 call assert_equal(6, strwidth("\uFEFF"))
194 " Ambiguous width chars
195 call assert_equal((aw == 'single') ? 1 : 2, strwidth("\u00A1"))
196 call assert_equal((aw == 'single') ? 1 : 2, strwidth("\u2010"))
197
198 call setcellwidths([[0x81, 0x81, 1], [0xA1, 0xA1, 1],
199 \ [0x2010, 0x2010, 1], [0xFEFF, 0xFEFF, 1]])
200 " Control chars
201 call assert_equal(4, strwidth("\u0081"))
202 call assert_equal(6, strwidth("\uFEFF"))
203 " Ambiguous width chars
204 call assert_equal(1, strwidth("\u00A1"))
205 call assert_equal(1, strwidth("\u2010"))
206
207 call setcellwidths([[0x81, 0x81, 2], [0xA1, 0xA1, 2],
208 \ [0x2010, 0x2010, 2], [0xFEFF, 0xFEFF, 2]])
209 " Control chars
210 call assert_equal(4, strwidth("\u0081"))
211 call assert_equal(6, strwidth("\uFEFF"))
212 " Ambiguous width chars
213 call assert_equal(2, strwidth("\u00A1"))
214 call assert_equal(2, strwidth("\u2010"))
215 endfor
216 set ambiwidth& isprint&
217
Bram Moolenaar08aac3c2020-08-28 21:04:24 +0200218 call setcellwidths([])
219
Bram Moolenaard83392a2022-09-01 12:22:46 +0100220 call assert_fails('call setcellwidths(1)', 'E1211:')
Bram Moolenaar08aac3c2020-08-28 21:04:24 +0200221
222 call assert_fails('call setcellwidths([1, 2, 0])', 'E1109:')
223
224 call assert_fails('call setcellwidths([[0x101]])', 'E1110:')
225 call assert_fails('call setcellwidths([[0x101, 0x102]])', 'E1110:')
226 call assert_fails('call setcellwidths([[0x101, 0x102, 1, 4]])', 'E1110:')
227 call assert_fails('call setcellwidths([["a"]])', 'E1110:')
228
229 call assert_fails('call setcellwidths([[0x102, 0x101, 1]])', 'E1111:')
230
231 call assert_fails('call setcellwidths([[0x101, 0x102, 0]])', 'E1112:')
232 call assert_fails('call setcellwidths([[0x101, 0x102, 3]])', 'E1112:')
233
234 call assert_fails('call setcellwidths([[0x111, 0x122, 1], [0x115, 0x116, 2]])', 'E1113:')
235 call assert_fails('call setcellwidths([[0x111, 0x122, 1], [0x122, 0x123, 2]])', 'E1113:')
236
237 call assert_fails('call setcellwidths([[0x33, 0x44, 2]])', 'E1114:')
zeertzjq94358a12021-10-20 11:01:15 +0100238
239 set listchars=tab:--\\u2192
240 call assert_fails('call setcellwidths([[0x2192, 0x2192, 2]])', 'E834:')
241
242 set fillchars=stl:\\u2501
243 call assert_fails('call setcellwidths([[0x2501, 0x2501, 2]])', 'E835:')
244
245 set listchars&
246 set fillchars&
247 call setcellwidths([])
Bram Moolenaar08aac3c2020-08-28 21:04:24 +0200248endfunc
249
Kota Kato66bb9ae2023-01-17 18:31:56 +0000250func Test_getcellwidths()
251 call setcellwidths([])
252 call assert_equal([], getcellwidths())
253
254 let widthlist = [
255 \ [0x1330, 0x1330, 2],
256 \ [9999, 10000, 1],
257 \ [0x1337, 0x1339, 2],
258 \]
259 let widthlistsorted = [
260 \ [0x1330, 0x1330, 2],
261 \ [0x1337, 0x1339, 2],
262 \ [9999, 10000, 1],
263 \]
264 call setcellwidths(widthlist)
265 call assert_equal(widthlistsorted, getcellwidths())
266
267 call setcellwidths([])
268endfunc
269
Yasuhiro Matsumoto2bc849f2023-01-10 16:03:08 +0000270func Test_setcellwidths_dump()
271 CheckRunVimInTerminal
272
273 let lines =<< trim END
274 call setline(1, "\ue5ffDesktop")
275 END
276 call writefile(lines, 'XCellwidths', 'D')
277 let buf = RunVimInTerminal('-S XCellwidths', {'rows': 6})
278 call VerifyScreenDump(buf, 'Test_setcellwidths_dump_1', {})
279
280 call term_sendkeys(buf, ":call setcellwidths([[0xe5ff, 0xe5ff, 2]])\<CR>")
281 call VerifyScreenDump(buf, 'Test_setcellwidths_dump_2', {})
282
283 call StopVimInTerminal(buf)
284endfunc
285
Bram Moolenaar1cbfc992020-12-02 12:37:37 +0100286func Test_print_overlong()
287 " Text with more composing characters than MB_MAXBYTES.
288 new
289 call setline(1, 'axxxxxxxxxxxxxxxxxxxxxxxxxxxxxx')
290 s/x/\=nr2char(1629)/g
291 print
292 bwipe!
293endfunc
294
Bram Moolenaar0e05de42020-03-25 22:23:46 +0100295" vim: shiftwidth=2 sts=2 expandtab