teken_wcwidth.h (6d3296f16a06bcaa49918799e683936711dcf9c9) teken_wcwidth.h (4b9aa38ef0e5bedcdd90b6627cc1c215037a1121)
1/*
2 * Markus Kuhn -- 2007-05-26 (Unicode 5.0)
3 *
4 * Permission to use, copy, modify, and distribute this software
5 * for any purpose and without fee is hereby granted. The author
6 * disclaims all warranties with regard to this software.
7 *
8 * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c
9 */
10
11#define TEKEN_UTF8_INVALID_CODEPOINT -1
12
13struct interval {
14 teken_char_t first;
15 teken_char_t last;
16};
17
18/* auxiliary function for binary search in interval table */
1/*
2 * Markus Kuhn -- 2007-05-26 (Unicode 5.0)
3 *
4 * Permission to use, copy, modify, and distribute this software
5 * for any purpose and without fee is hereby granted. The author
6 * disclaims all warranties with regard to this software.
7 *
8 * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c
9 */
10
11#define TEKEN_UTF8_INVALID_CODEPOINT -1
12
13struct interval {
14 teken_char_t first;
15 teken_char_t last;
16};
17
18/* auxiliary function for binary search in interval table */
19static int bisearch(teken_char_t ucs, const struct interval *table, int max) {
19static inline int bisearch(teken_char_t ucs, const struct interval *table, int max) {
20 int min = 0;
21 int mid;
22
23 if (ucs < table[0].first || ucs > table[max].last)
24 return 0;
25 while (max >= min) {
26 mid = (min + max) / 2;
27 if (ucs > table[mid].last)
28 min = mid + 1;
29 else if (ucs < table[mid].first)
30 max = mid - 1;
31 else
32 return 1;
33 }
34
35 return 0;
36}
37
20 int min = 0;
21 int mid;
22
23 if (ucs < table[0].first || ucs > table[max].last)
24 return 0;
25 while (max >= min) {
26 mid = (min + max) / 2;
27 if (ucs > table[mid].last)
28 min = mid + 1;
29 else if (ucs < table[mid].first)
30 max = mid - 1;
31 else
32 return 1;
33 }
34
35 return 0;
36}
37
38static int teken_wcwidth(teken_char_t ucs)
38static inline int teken_wcwidth(teken_char_t ucs)
39{
40 /* sorted list of non-overlapping intervals of non-spacing characters */
41 /* generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c" */
42 static const struct interval combining[] = {
43 { 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 },
44 { 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
45 { 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 },
46 { 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 },

--- 72 unchanged lines hidden (view full) ---

119 (ucs >= 0x30000 && ucs <= 0x3fffd)));
120}
121
122/*
123 * Converts an UTF-8 byte sequence to a codepoint as specified in
124 * https://datatracker.ietf.org/doc/html/rfc3629#section-3 . The function
125 * expects the 'bytes' array to start with the leading character.
126 */
39{
40 /* sorted list of non-overlapping intervals of non-spacing characters */
41 /* generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c" */
42 static const struct interval combining[] = {
43 { 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 },
44 { 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
45 { 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 },
46 { 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 },

--- 72 unchanged lines hidden (view full) ---

119 (ucs >= 0x30000 && ucs <= 0x3fffd)));
120}
121
122/*
123 * Converts an UTF-8 byte sequence to a codepoint as specified in
124 * https://datatracker.ietf.org/doc/html/rfc3629#section-3 . The function
125 * expects the 'bytes' array to start with the leading character.
126 */
127static teken_char_t
127static inline teken_char_t
128teken_utf8_bytes_to_codepoint(uint8_t bytes[4], int nbytes)
129{
130
131 /* Check for malformed characters. */
132 if (__bitcount(bytes[0] & 0xf0) != nbytes)
133 return (TEKEN_UTF8_INVALID_CODEPOINT);
134
135 switch (nbytes) {

--- 13 unchanged lines hidden --- 		128teken_utf8_bytes_to_codepoint(uint8_t bytes[4], int nbytes)
129{
130
131 /* Check for malformed characters. */
132 if (__bitcount(bytes[0] & 0xf0) != nbytes)
133 return (TEKEN_UTF8_INVALID_CODEPOINT);
134
135 switch (nbytes) {

--- 13 unchanged lines hidden ---