aboutsummaryrefslogtreecommitdiff
path: root/lib/unicode/gbrk
diff options
context:
space:
mode:
authorThomas Voss <mail@thomasvoss.com> 2024-04-09 18:49:27 +0200
committerThomas Voss <mail@thomasvoss.com> 2024-04-09 18:49:27 +0200
commitfc0dae9c819a4ee64f4610db1f005dfb841030c5 (patch)
treecaa7e5a358b1a2a8614bcd16be112546994d8d26 /lib/unicode/gbrk
parentbf212e80fb0f1d83902de520e9cee098187383f7 (diff)
Move unicode/gbrk.h to unicode/string.h
Diffstat (limited to 'lib/unicode/gbrk')
-rw-r--r--lib/unicode/gbrk/u8glen.c10
-rw-r--r--lib/unicode/gbrk/u8gnext.c147
2 files changed, 0 insertions, 157 deletions
diff --git a/lib/unicode/gbrk/u8glen.c b/lib/unicode/gbrk/u8glen.c
deleted file mode 100644
index 763a834..0000000
--- a/lib/unicode/gbrk/u8glen.c
+++ /dev/null
@@ -1,10 +0,0 @@
-#include "unicode/gbrk.h"
-
-size_t
-u8glen(const char8_t *s, size_t n)
-{
- size_t m = 0;
- while (u8gnext(nullptr, &s, &n))
- m++;
- return m;
-}
diff --git a/lib/unicode/gbrk/u8gnext.c b/lib/unicode/gbrk/u8gnext.c
deleted file mode 100644
index f3f7afc..0000000
--- a/lib/unicode/gbrk/u8gnext.c
+++ /dev/null
@@ -1,147 +0,0 @@
-#include <stddef.h>
-
-#include "__bsearch.h"
-#include "mbstring.h"
-#include "unicode/__gbrk.h"
-#include "unicode/gbrk.h"
-
-struct gbrk_state {
- enum {
- GB9C_NONE,
- GB9C_CNSNT,
- GB9C_LNK,
- } gb9c;
- bool gb11 : 1;
- bool gb12 : 1;
-};
-
-static bool u8isgbrk(rune, rune, struct gbrk_state *);
-
-__MLIB_DEFINE_BSEARCH(gbrk_prop, gbrk_prop_tbl, GBP_OTHER)
-
-size_t
-u8gnext(struct u8view *g, const char8_t **s, size_t *n)
-{
- int m;
- rune ch1;
- const char8_t *p;
- struct gbrk_state gs = {0};
-
- if (*n == 0)
- return 0;
-
- p = *s;
- if (g)
- g->p = p;
- p += u8tor(&ch1, p);
-
- for (;;) {
- rune ch2;
-
- if ((size_t)(p - *s) >= *n)
- ch2 = 0;
- else
- m = u8tor(&ch2, p);
- if (u8isgbrk(ch1, ch2, &gs)) {
- ptrdiff_t d = p - *s;
- *n -= d;
- *s = p;
- if (g)
- g->len = d;
- return d;
- }
-
- ch1 = ch2;
- p += m;
- }
-}
-
-bool
-u8isgbrk(rune a, rune b, struct gbrk_state *gs)
-{
- gbrk_prop ap, bp;
-
- /* GB1 & GB2 */
- if (!a || !b)
- goto do_break;
-
- /* GB3 & ASCII fast-track */
- if ((a | b) < 0x300) {
- if (a == '\r' && b == '\n')
- return false;
- goto do_break;
- }
-
- /* GB4 */
- if (a == '\r' || a == '\n' || ((ap = mlib_lookup(a)) & GBP_CTRL))
- goto do_break;
-
- /* GB5 */
- if (b == '\r' || b == '\n' || ((bp = mlib_lookup(b)) & GBP_CTRL))
- goto do_break;
-
- /* Setting flags for GB9c */
- if (ap & GBP_INDC_CNSNT)
- gs->gb9c = GB9C_CNSNT;
- else if ((ap & GBP_INDC_LNK) && gs->gb9c == GB9C_CNSNT)
- gs->gb9c = GB9C_LNK;
-
- /* GB6 */
- if ((ap & GBP_HNGL_L)
- && (bp & (GBP_HNGL_L | GBP_HNGL_V | GBP_HNGL_LV | GBP_HNGL_LVT)))
- {
- return false;
- }
-
- /* GB7 */
- if ((ap & (GBP_HNGL_LV | GBP_HNGL_V)) && (bp & (GBP_HNGL_V | GBP_HNGL_T)))
- return false;
-
- /* GB8 */
- if ((ap & (GBP_HNGL_LVT | GBP_HNGL_T)) && (bp & GBP_HNGL_T))
- return false;
-
- /* GB9 */
- if (bp & (GBP_EXT | GBP_ZWJ)) {
- if (ap & GBP_PIC)
- gs->gb11 = true;
- return false;
- }
-
- /* GB9a */
- if (bp & GBP_SM)
- return false;
-
- /* GB9b */
- if (ap & GBP_PREP)
- return false;
-
- /* GB9c */
- if ((ap & (GBP_INDC_EXT | GBP_INDC_LNK)) && (bp & GBP_INDC_CNSNT)
- && gs->gb9c == GB9C_LNK)
- {
- return false;
- }
-
- /* GB11 */
- if (gs->gb11) {
- if ((ap & GBP_EXT) && (bp & (GBP_EXT | GBP_ZWJ)))
- return false;
- if ((ap & GBP_ZWJ) && (bp & GBP_PIC))
- return false;
- }
-
- /* GB12 & GB13 */
- if (ap & GBP_RI) {
- if (gs->gb12 || !(bp & GBP_RI))
- goto do_break;
- gs->gb12 = true;
- return false;
- }
-
- /* GB999 */
-do_break:
- gs->gb9c = GB9C_NONE;
- gs->gb11 = gs->gb12 = false;
- return true;
-}