[READ-ONLY] Mirror of https://github.com/bombshell-dev/tty. Platform independent 2D layout engine for terminal applications based on Clay
ref(wcwidth): generate clang-format friendly code
Nate Moore
(May 28, 2026, 2:52 AM EDT)
6c5cba49
51fdc1ae
+56
-32
+28
-16
src/wcwidth.c
+28
-16
src/wcwidth.c
···
1
1
/* wcwidth.c - Unicode character width lookup
2
-
* Unicode 16.0 — generated by tasks/gen-wcwidth.ts on 2026-05-28
2
+
* Unicode 16.0 - generated by tasks/gen-wcwidth.ts on 2026-05-28
3
3
*
4
4
* Only zero-width (combining marks + default ignorable) and double-width
5
5
* (wide/fullwidth) codepoints are stored. Everything else defaults to
6
-
* width 1. Control characters (U+0000–U+001F, U+007F–U+009F) are handled
6
+
* width 1. Control characters (U+0000-U+001F, U+007F-U+009F) are handled
7
7
* by the fast-path checks in wcwidth() and are not in the tables.
8
8
*
9
9
* Sources:
10
-
* extracted/DerivedGeneralCategory.txt Mn/Me categories → width 0
11
-
* DerivedCoreProperties.txt Default_Ignorable_Code_Point → width 0
12
-
* EastAsianWidth.txt W/F properties → width 2
10
+
* DerivedGeneralCategory.txt: Mn/Me categories -> width 0
11
+
* DerivedCoreProperties.txt: Default_Ignorable_Code_Point -> width 0
12
+
* EastAsianWidth.txt: W/F properties -> width 2
13
13
*
14
14
* Combining (width 0) and wide (width 2) ranges are merged into a single
15
15
* sorted table so wcwidth() needs only one binary search for any codepoint.
16
16
*
17
17
* BMP coarse filter (bmp_filter):
18
18
* 64-byte bitmap, 1 bit per 128-codepoint BMP block. A 0-bit means no
19
-
* special codepoints in that block — return width 1 without searching.
19
+
* special codepoints in that block - return width 1 without searching.
20
20
*
21
21
* Packed encoding (special_small_ranges):
22
22
* Each uint32_t entry packs one Unicode range as
23
-
* bits 31–11 start codepoint (fits in 21 bits; Unicode max is U+10FFFF)
24
-
* bits 10–1 count of additional codepoints beyond start (max 1023)
23
+
* bits 31-11 start codepoint (fits in 21 bits; Unicode max is U+10FFFF)
24
+
* bits 10-1 count of additional codepoints beyond start (max 1023)
25
25
* bit 0 0 = width 0 (combining), 1 = width 2 (wide)
26
26
* Array is sorted by start so binary search operates on raw uint32_t values.
27
27
*
···
30
30
*/
31
31
32
32
#include <stdint.h>
33
+
34
+
/* clang-format off */
33
35
34
36
static const uint32_t bmp_filter[16] = {
35
37
0xfffffa40, 0x0bf7c047, 0xee40f8c3, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff,
···
111
113
};
112
114
#define SPECIAL_LARGE_COUNT 7
113
115
116
+
/* clang-format on */
117
+
114
118
static int codepoint_in_special(uint32_t codepoint) {
115
119
if (codepoint <= 0xffff) {
116
120
uint32_t block = codepoint >> 7;
117
-
if (!((bmp_filter[block >> 5] >> (block & 31u)) & 1u)) return 1;
121
+
if (!((bmp_filter[block >> 5] >> (block & 31u)) & 1u))
122
+
return 1;
118
123
}
119
124
int left = 0, right = SPECIAL_SMALL_COUNT - 1;
120
125
while (left <= right) {
121
126
int mid = (left + right) / 2;
122
127
uint32_t entry = special_small_ranges[mid];
123
128
uint32_t start = entry >> 11;
124
-
if (codepoint < start) right = mid - 1;
125
-
else if (codepoint > start + ((entry >> 1) & 0x3FF)) left = mid + 1;
126
-
else return (entry & 1) ? 2 : 0;
129
+
if (codepoint < start)
130
+
right = mid - 1;
131
+
else if (codepoint > start + ((entry >> 1) & 0x3FF))
132
+
left = mid + 1;
133
+
else
134
+
return (entry & 1) ? 2 : 0;
127
135
}
128
-
left = 0; right = SPECIAL_LARGE_COUNT - 1;
136
+
left = 0;
137
+
right = SPECIAL_LARGE_COUNT - 1;
129
138
while (left <= right) {
130
139
int mid = (left + right) / 2;
131
-
if (codepoint < special_large_starts[mid]) right = mid - 1;
132
-
else if (codepoint > special_large_starts[mid] + special_large_counts[mid]) left = mid + 1;
133
-
else return special_large_widths[mid];
140
+
if (codepoint < special_large_starts[mid])
141
+
right = mid - 1;
142
+
else if (codepoint > special_large_starts[mid] + special_large_counts[mid])
143
+
left = mid + 1;
144
+
else
145
+
return special_large_widths[mid];
134
146
}
135
147
return 1;
136
148
}
+28
-16
tasks/gen-wcwidth.ts
+28
-16
tasks/gen-wcwidth.ts
···
311
311
312
312
const output = `\
313
313
/* wcwidth.c - Unicode character width lookup
314
-
* Unicode 16.0 — generated by tasks/gen-wcwidth.ts on ${date}
314
+
* Unicode 16.0 - generated by tasks/gen-wcwidth.ts on ${date}
315
315
*
316
316
* Only zero-width (combining marks + default ignorable) and double-width
317
317
* (wide/fullwidth) codepoints are stored. Everything else defaults to
318
-
* width 1. Control characters (U+0000–U+001F, U+007F–U+009F) are handled
318
+
* width 1. Control characters (U+0000-U+001F, U+007F-U+009F) are handled
319
319
* by the fast-path checks in wcwidth() and are not in the tables.
320
320
*
321
321
* Sources:
322
-
* extracted/DerivedGeneralCategory.txt Mn/Me categories → width 0
323
-
* DerivedCoreProperties.txt Default_Ignorable_Code_Point → width 0
324
-
* EastAsianWidth.txt W/F properties → width 2
322
+
* DerivedGeneralCategory.txt: Mn/Me categories -> width 0
323
+
* DerivedCoreProperties.txt: Default_Ignorable_Code_Point -> width 0
324
+
* EastAsianWidth.txt: W/F properties -> width 2
325
325
*
326
326
* Combining (width 0) and wide (width 2) ranges are merged into a single
327
327
* sorted table so wcwidth() needs only one binary search for any codepoint.
328
328
*
329
329
* BMP coarse filter (bmp_filter):
330
330
* 64-byte bitmap, 1 bit per 128-codepoint BMP block. A 0-bit means no
331
-
* special codepoints in that block — return width 1 without searching.
331
+
* special codepoints in that block - return width 1 without searching.
332
332
*
333
333
* Packed encoding (special_small_ranges):
334
334
* Each uint32_t entry packs one Unicode range as
335
-
* bits 31–11 start codepoint (fits in 21 bits; Unicode max is U+10FFFF)
336
-
* bits 10–1 count of additional codepoints beyond start (max 1023)
335
+
* bits 31-11 start codepoint (fits in 21 bits; Unicode max is U+10FFFF)
336
+
* bits 10-1 count of additional codepoints beyond start (max 1023)
337
337
* bit 0 0 = width 0 (combining), 1 = width 2 (wide)
338
338
* Array is sorted by start so binary search operates on raw uint32_t values.
339
339
*
···
343
343
344
344
#include <stdint.h>
345
345
346
+
/* clang-format off */
347
+
346
348
static const uint32_t bmp_filter[16] = {
347
349
${formatUint32Array(Array.from(bmpFilter))}
348
350
};
···
363
365
};
364
366
#define SPECIAL_LARGE_COUNT ${largeRanges.length}
365
367
368
+
/* clang-format on */
369
+
366
370
static int codepoint_in_special(uint32_t codepoint) {
367
371
if (codepoint <= 0xffff) {
368
372
uint32_t block = codepoint >> 7;
369
-
if (!((bmp_filter[block >> 5] >> (block & 31u)) & 1u)) return 1;
373
+
if (!((bmp_filter[block >> 5] >> (block & 31u)) & 1u))
374
+
return 1;
370
375
}
371
376
int left = 0, right = SPECIAL_SMALL_COUNT - 1;
372
377
while (left <= right) {
373
378
int mid = (left + right) / 2;
374
379
uint32_t entry = special_small_ranges[mid];
375
380
uint32_t start = entry >> 11;
376
-
if (codepoint < start) right = mid - 1;
377
-
else if (codepoint > start + ((entry >> 1) & 0x3FF)) left = mid + 1;
378
-
else return (entry & 1) ? 2 : 0;
381
+
if (codepoint < start)
382
+
right = mid - 1;
383
+
else if (codepoint > start + ((entry >> 1) & 0x3FF))
384
+
left = mid + 1;
385
+
else
386
+
return (entry & 1) ? 2 : 0;
379
387
}
380
-
left = 0; right = SPECIAL_LARGE_COUNT - 1;
388
+
left = 0;
389
+
right = SPECIAL_LARGE_COUNT - 1;
381
390
while (left <= right) {
382
391
int mid = (left + right) / 2;
383
-
if (codepoint < special_large_starts[mid]) right = mid - 1;
384
-
else if (codepoint > special_large_starts[mid] + special_large_counts[mid]) left = mid + 1;
385
-
else return special_large_widths[mid];
392
+
if (codepoint < special_large_starts[mid])
393
+
right = mid - 1;
394
+
else if (codepoint > special_large_starts[mid] + special_large_counts[mid])
395
+
left = mid + 1;
396
+
else
397
+
return special_large_widths[mid];
386
398
}
387
399
return 1;
388
400
}