1 /* $Id: wcwidth.c,v 1.30 2010/09/06 17:40:13 tom Exp $ */
2 /* @XTermId: wcwidth.c,v 1.21 2007/06/13 00:14:29 tom Exp @ */
3
4 /*
5 * This is an implementation of wcwidth() and wcswidth() (defined in
6 * IEEE Std 1002.1-2001) for Unicode.
7 *
8 * http://www.opengroup.org/onlinepubs/007904975/functions/wcwidth.html
9 * http://www.opengroup.org/onlinepubs/007904975/functions/wcswidth.html
10 *
11 * In fixed-width output devices, Latin characters all occupy a single
12 * "cell" position of equal width, whereas ideographic CJK characters
13 * occupy two such cells. Interoperability between terminal-line
14 * applications and (teletype-style) character terminals using the
15 * UTF-8 encoding requires agreement on which character should advance
16 * the cursor by how many cell positions. No established formal
17 * standards exist at present on which Unicode character shall occupy
18 * how many cell positions on character terminals. These routines are
19 * a first attempt of defining such behavior based on simple rules
20 * applied to data provided by the Unicode Consortium.
21 *
22 * For some graphical characters, the Unicode standard explicitly
23 * defines a character-cell width via the definition of the East Asian
24 * FullWidth (F), Wide (W), Half-width (H), and Narrow (Na) classes.
25 * In all these cases, there is no ambiguity about which width a
26 * terminal shall use. For characters in the East Asian Ambiguous (A)
27 * class, the width choice depends purely on a preference of backward
28 * compatibility with either historic CJK or Western practice.
29 * Choosing single-width for these characters is easy to justify as
30 * the appropriate long-term solution, as the CJK practice of
31 * displaying these characters as double-width comes from historic
32 * implementation simplicity (8-bit encoded characters were displayed
33 * single-width and 16-bit ones double-width, even for Greek,
34 * Cyrillic, etc.) and not any typographic considerations.
35 *
36 * Much less clear is the choice of width for the Not East Asian
37 * (Neutral) class. Existing practice does not dictate a width for any
38 * of these characters. It would nevertheless make sense
39 * typographically to allocate two character cells to characters such
40 * as for instance EM SPACE or VOLUME INTEGRAL, which cannot be
41 * represented adequately with a single-width glyph. The following
42 * routines at present merely assign a single-cell width to all
43 * neutral characters, in the interest of simplicity. This is not
44 * entirely satisfactory and should be reconsidered before
45 * establishing a formal standard in this area. At the moment, the
46 * decision which Not East Asian (Neutral) characters should be
47 * represented by double-width glyphs cannot yet be answered by
48 * applying a simple rule from the Unicode database content. Setting
49 * up a proper standard for the behavior of UTF-8 character terminals
50 * will require a careful analysis not only of each Unicode character,
51 * but also of each presentation form, something the author of these
52 * routines has avoided to do so far.
53 *
54 * http://www.unicode.org/unicode/reports/tr11/
55 *
56 * Markus Kuhn -- 2007-05-25 (Unicode 5.0)
57 *
58 * Permission to use, copy, modify, and distribute this software
59 * for any purpose and without fee is hereby granted. The author
60 * disclaims all warranties with regard to this software.
61 *
62 * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c
63 */
64
65 #include <estruct.h>
66 #include <edef.h>
67 #include <wcwidth.h>
68
69 #if OPT_LOCALE
70 #include <locale.h>
71 #ifdef HAVE_WCWIDTH
72 #include <wchar.h> /* prototype for wcwidth() */
73 #endif
74 #endif
75
76 struct interval {
77 unsigned long first;
78 unsigned long last;
79 };
80
81 /* auxiliary function for binary search in interval table */
82 static int
bisearch(unsigned long ucs,const struct interval * table,int max)83 bisearch(unsigned long ucs, const struct interval *table, int max)
84 {
85 int min = 0;
86 int mid;
87
88 if (ucs >= table[0].first && ucs <= table[max].last) {
89 while (max >= min) {
90 mid = (min + max) / 2;
91 if (ucs > table[mid].last)
92 min = mid + 1;
93 else if (ucs < table[mid].first)
94 max = mid - 1;
95 else
96 return 1;
97 }
98 }
99
100 return 0;
101 }
102
103 /* The following two functions define the column width of an ISO 10646
104 * character as follows:
105 *
106 * - The null character (U+0000) has a column width of 0.
107 *
108 * - Other C0/C1 control characters and DEL will lead to a return
109 * value of -1.
110 *
111 * - Non-spacing and enclosing combining characters (general
112 * category code Mn or Me in the Unicode database) have a
113 * column width of 0.
114 *
115 * - SOFT HYPHEN (U+00AD) has a column width of 1.
116 *
117 * - Other format characters (general category code Cf in the Unicode
118 * database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
119 *
120 * - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
121 * have a column width of 0.
122 *
123 * - Spacing characters in the East Asian Wide (W) or East Asian
124 * Full-width (F) category as defined in Unicode Technical
125 * Report #11 have a column width of 2.
126 *
127 * - All remaining characters (including all printable
128 * ISO 8859-1 and WGL4 characters, Unicode control characters,
129 * etc.) have a column width of 1.
130 *
131 * This implementation assumes that wchar_t characters are encoded
132 * in ISO 10646.
133 */
134
135 int
mk_wcwidth(wchar_t ucs)136 mk_wcwidth(wchar_t ucs)
137 {
138 unsigned long cmp = (unsigned long) ucs;
139 /* *INDENT-OFF* */
140 /* sorted list of non-overlapping intervals of non-spacing characters */
141 /* generated by
142 * uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c
143 */
144 static const struct interval combining[] = {
145 { 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 },
146 { 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
147 { 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 },
148 { 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 },
149 { 0x06D6, 0x06E4 }, { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED },
150 { 0x070F, 0x070F }, { 0x0711, 0x0711 }, { 0x0730, 0x074A },
151 { 0x07A6, 0x07B0 }, { 0x07EB, 0x07F3 }, { 0x0901, 0x0902 },
152 { 0x093C, 0x093C }, { 0x0941, 0x0948 }, { 0x094D, 0x094D },
153 { 0x0951, 0x0954 }, { 0x0962, 0x0963 }, { 0x0981, 0x0981 },
154 { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD },
155 { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 }, { 0x0A3C, 0x0A3C },
156 { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D },
157 { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 }, { 0x0ABC, 0x0ABC },
158 { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 }, { 0x0ACD, 0x0ACD },
159 { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 }, { 0x0B3C, 0x0B3C },
160 { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 }, { 0x0B4D, 0x0B4D },
161 { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 }, { 0x0BC0, 0x0BC0 },
162 { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 }, { 0x0C46, 0x0C48 },
163 { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 }, { 0x0CBC, 0x0CBC },
164 { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD },
165 { 0x0CE2, 0x0CE3 }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
166 { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
167 { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
168 { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
169 { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
170 { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
171 { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
172 { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
173 { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
174 { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x135F, 0x135F },
175 { 0x1712, 0x1714 }, { 0x1732, 0x1734 }, { 0x1752, 0x1753 },
176 { 0x1772, 0x1773 }, { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD },
177 { 0x17C6, 0x17C6 }, { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD },
178 { 0x180B, 0x180D }, { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 },
179 { 0x1927, 0x1928 }, { 0x1932, 0x1932 }, { 0x1939, 0x193B },
180 { 0x1A17, 0x1A18 }, { 0x1B00, 0x1B03 }, { 0x1B34, 0x1B34 },
181 { 0x1B36, 0x1B3A }, { 0x1B3C, 0x1B3C }, { 0x1B42, 0x1B42 },
182 { 0x1B6B, 0x1B73 }, { 0x1DC0, 0x1DCA }, { 0x1DFE, 0x1DFF },
183 { 0x200B, 0x200F }, { 0x202A, 0x202E }, { 0x2060, 0x2063 },
184 { 0x206A, 0x206F }, { 0x20D0, 0x20EF }, { 0x302A, 0x302F },
185 { 0x3099, 0x309A }, { 0xA806, 0xA806 }, { 0xA80B, 0xA80B },
186 { 0xA825, 0xA826 }, { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F },
187 { 0xFE20, 0xFE23 }, { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB },
188 { 0x10A01, 0x10A03 }, { 0x10A05, 0x10A06 }, { 0x10A0C, 0x10A0F },
189 { 0x10A38, 0x10A3A }, { 0x10A3F, 0x10A3F }, { 0x1D167, 0x1D169 },
190 { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD },
191 { 0x1D242, 0x1D244 }, { 0xE0001, 0xE0001 }, { 0xE0020, 0xE007F },
192 { 0xE0100, 0xE01EF }
193 };
194 /* *INDENT-ON* */
195
196 /* test for 8-bit control characters */
197 if (cmp == 0)
198 return 0;
199 if (cmp < 32 || (cmp >= 0x7f && cmp < 0xa0))
200 return -1;
201
202 /* binary search in table of non-spacing characters */
203 if (bisearch(cmp, combining,
204 (int) (sizeof(combining) / sizeof(struct interval) - 1)))
205 return 0;
206
207 /* if we arrive here, cmp is not a combining or C0/C1 control character */
208
209 return 1 +
210 (cmp >= 0x1100 &&
211 (cmp <= 0x115f || /* Hangul Jamo init. consonants */
212 cmp == 0x2329 || cmp == 0x232a ||
213 (cmp >= 0x2e80 && cmp <= 0xa4cf &&
214 cmp != 0x303f) || /* CJK ... Yi */
215 (cmp >= 0xac00 && cmp <= 0xd7a3) || /* Hangul Syllables */
216 (cmp >= 0xf900 && cmp <= 0xfaff) || /* CJK Compatibility Ideographs */
217 (cmp >= 0xfe10 && cmp <= 0xfe19) || /* Vertical forms */
218 (cmp >= 0xfe30 && cmp <= 0xfe6f) || /* CJK Compatibility Forms */
219 (cmp >= 0xff00 && cmp <= 0xff60) || /* Fullwidth Forms */
220 (cmp >= 0xffe0 && cmp <= 0xffe6) ||
221 (cmp >= 0x20000 && cmp <= 0x2fffd) ||
222 (cmp >= 0x30000 && cmp <= 0x3fffd)));
223 }
224
225 int
mk_wcswidth(const wchar_t * pwcs,size_t n)226 mk_wcswidth(const wchar_t *pwcs, size_t n)
227 {
228 int w, width = 0;
229
230 for (; *pwcs && n-- > 0; pwcs++)
231 if ((w = mk_wcwidth(*pwcs)) < 0)
232 return -1;
233 else
234 width += w;
235
236 return width;
237 }
238
239 /*
240 * The following functions are the same as mk_wcwidth() and
241 * mk_wcwidth_cjk(), except that spacing characters in the East Asian
242 * Ambiguous (A) category as defined in Unicode Technical Report #11
243 * have a column width of 2. This variant might be useful for users of
244 * CJK legacy encodings who want to migrate to UCS without changing
245 * the traditional terminal character-width behaviour. It is not
246 * otherwise recommended for general use.
247 */
248 int
mk_wcwidth_cjk(wchar_t ucs)249 mk_wcwidth_cjk(wchar_t ucs)
250 {
251 /* sorted list of non-overlapping intervals of East Asian Ambiguous
252 * characters, generated by
253 *
254 * uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf \
255 * +E000..F8FF \
256 * +F0000..FFFFD \
257 * +100000..10FFFD c
258 *
259 * "WIDTH-A" is a file extracted from EastAsianWidth.txt by selecting
260 * only those with width "A", and omitting:
261 *
262 * 0xAD
263 * all lines with "COMBINING"
264 *
265 * (uniset does not recognize the range expressions in WIDTH-A).
266 */
267 /* *INDENT-OFF* */
268 static const struct interval ambiguous[] = {
269 { 0x00A1, 0x00A1 }, { 0x00A4, 0x00A4 }, { 0x00A7, 0x00A8 },
270 { 0x00AA, 0x00AA }, { 0x00AE, 0x00AE }, { 0x00B0, 0x00B4 },
271 { 0x00B6, 0x00BA }, { 0x00BC, 0x00BF }, { 0x00C6, 0x00C6 },
272 { 0x00D0, 0x00D0 }, { 0x00D7, 0x00D8 }, { 0x00DE, 0x00E1 },
273 { 0x00E6, 0x00E6 }, { 0x00E8, 0x00EA }, { 0x00EC, 0x00ED },
274 { 0x00F0, 0x00F0 }, { 0x00F2, 0x00F3 }, { 0x00F7, 0x00FA },
275 { 0x00FC, 0x00FC }, { 0x00FE, 0x00FE }, { 0x0101, 0x0101 },
276 { 0x0111, 0x0111 }, { 0x0113, 0x0113 }, { 0x011B, 0x011B },
277 { 0x0126, 0x0127 }, { 0x012B, 0x012B }, { 0x0131, 0x0133 },
278 { 0x0138, 0x0138 }, { 0x013F, 0x0142 }, { 0x0144, 0x0144 },
279 { 0x0148, 0x014B }, { 0x014D, 0x014D }, { 0x0152, 0x0153 },
280 { 0x0166, 0x0167 }, { 0x016B, 0x016B }, { 0x01CE, 0x01CE },
281 { 0x01D0, 0x01D0 }, { 0x01D2, 0x01D2 }, { 0x01D4, 0x01D4 },
282 { 0x01D6, 0x01D6 }, { 0x01D8, 0x01D8 }, { 0x01DA, 0x01DA },
283 { 0x01DC, 0x01DC }, { 0x0251, 0x0251 }, { 0x0261, 0x0261 },
284 { 0x02C4, 0x02C4 }, { 0x02C7, 0x02C7 }, { 0x02C9, 0x02CB },
285 { 0x02CD, 0x02CD }, { 0x02D0, 0x02D0 }, { 0x02D8, 0x02DB },
286 { 0x02DD, 0x02DD }, { 0x02DF, 0x02DF }, { 0x0391, 0x03A1 },
287 { 0x03A3, 0x03A9 }, { 0x03B1, 0x03C1 }, { 0x03C3, 0x03C9 },
288 { 0x0401, 0x0401 }, { 0x0410, 0x044F }, { 0x0451, 0x0451 },
289 { 0x2010, 0x2010 }, { 0x2013, 0x2016 }, { 0x2018, 0x2019 },
290 { 0x201C, 0x201D }, { 0x2020, 0x2022 }, { 0x2024, 0x2027 },
291 { 0x2030, 0x2030 }, { 0x2032, 0x2033 }, { 0x2035, 0x2035 },
292 { 0x203B, 0x203B }, { 0x203E, 0x203E }, { 0x2074, 0x2074 },
293 { 0x207F, 0x207F }, { 0x2081, 0x2084 }, { 0x20AC, 0x20AC },
294 { 0x2103, 0x2103 }, { 0x2105, 0x2105 }, { 0x2109, 0x2109 },
295 { 0x2113, 0x2113 }, { 0x2116, 0x2116 }, { 0x2121, 0x2122 },
296 { 0x2126, 0x2126 }, { 0x212B, 0x212B }, { 0x2153, 0x2154 },
297 { 0x215B, 0x215E }, { 0x2160, 0x216B }, { 0x2170, 0x2179 },
298 { 0x2190, 0x2199 }, { 0x21B8, 0x21B9 }, { 0x21D2, 0x21D2 },
299 { 0x21D4, 0x21D4 }, { 0x21E7, 0x21E7 }, { 0x2200, 0x2200 },
300 { 0x2202, 0x2203 }, { 0x2207, 0x2208 }, { 0x220B, 0x220B },
301 { 0x220F, 0x220F }, { 0x2211, 0x2211 }, { 0x2215, 0x2215 },
302 { 0x221A, 0x221A }, { 0x221D, 0x2220 }, { 0x2223, 0x2223 },
303 { 0x2225, 0x2225 }, { 0x2227, 0x222C }, { 0x222E, 0x222E },
304 { 0x2234, 0x2237 }, { 0x223C, 0x223D }, { 0x2248, 0x2248 },
305 { 0x224C, 0x224C }, { 0x2252, 0x2252 }, { 0x2260, 0x2261 },
306 { 0x2264, 0x2267 }, { 0x226A, 0x226B }, { 0x226E, 0x226F },
307 { 0x2282, 0x2283 }, { 0x2286, 0x2287 }, { 0x2295, 0x2295 },
308 { 0x2299, 0x2299 }, { 0x22A5, 0x22A5 }, { 0x22BF, 0x22BF },
309 { 0x2312, 0x2312 }, { 0x2460, 0x24E9 }, { 0x24EB, 0x254B },
310 { 0x2550, 0x2573 }, { 0x2580, 0x258F }, { 0x2592, 0x2595 },
311 { 0x25A0, 0x25A1 }, { 0x25A3, 0x25A9 }, { 0x25B2, 0x25B3 },
312 { 0x25B6, 0x25B7 }, { 0x25BC, 0x25BD }, { 0x25C0, 0x25C1 },
313 { 0x25C6, 0x25C8 }, { 0x25CB, 0x25CB }, { 0x25CE, 0x25D1 },
314 { 0x25E2, 0x25E5 }, { 0x25EF, 0x25EF }, { 0x2605, 0x2606 },
315 { 0x2609, 0x2609 }, { 0x260E, 0x260F }, { 0x2614, 0x2615 },
316 { 0x261C, 0x261C }, { 0x261E, 0x261E }, { 0x2640, 0x2640 },
317 { 0x2642, 0x2642 }, { 0x2660, 0x2661 }, { 0x2663, 0x2665 },
318 { 0x2667, 0x266A }, { 0x266C, 0x266D }, { 0x266F, 0x266F },
319 { 0x273D, 0x273D }, { 0x2776, 0x277F }, { 0xE000, 0xF8FF },
320 { 0xFFFD, 0xFFFD }, { 0xF0000, 0xFFFFD }, { 0x100000, 0x10FFFD }
321 };
322 /* *INDENT-ON* */
323
324 /* binary search in table of non-spacing characters */
325 if (bisearch((unsigned long) ucs, ambiguous,
326 (int) (sizeof(ambiguous) / sizeof(struct interval) - 1)))
327 return 2;
328
329 return mk_wcwidth(ucs);
330 }
331
332 int
mk_wcswidth_cjk(const wchar_t * pwcs,size_t n)333 mk_wcswidth_cjk(const wchar_t *pwcs, size_t n)
334 {
335 int w, width = 0;
336
337 for (; *pwcs && n-- > 0; pwcs++)
338 if ((w = mk_wcwidth_cjk(*pwcs)) < 0)
339 return -1;
340 else
341 width += w;
342
343 return width;
344 }
345
346 /*
347 */
348 int
vl_wcwidth(int code)349 vl_wcwidth(int code)
350 {
351 int result = 1;
352 #ifdef HAVE_WCWIDTH
353 if (okCTYPE2(vl_wide_enc) && code > 255) {
354 char *save = setlocale(LC_CTYPE, vl_wide_enc.locale);
355 result = wcwidth(code);
356 setlocale(LC_CTYPE, save);
357 }
358 #else
359 result = mk_wcwidth((wchar_t) code);
360 #endif
361 return result;
362 }
363