aboutsummaryrefslogtreecommitdiff
path: root/lib/unicode
diff options
context:
space:
mode:
authorThomas Voss <mail@thomasvoss.com> 2024-05-14 23:59:18 +0200
committerThomas Voss <mail@thomasvoss.com> 2024-05-14 23:59:29 +0200
commitc6d98138dd766f5ffe7cdab3a7f944a8668d367c (patch)
treed2285566881385fb478609c7dcfbd4b6e51656db /lib/unicode
parent1aeb7e2b426e7a94cdd4f83c4337f44c0f5a2ca8 (diff)
Don’t use NFD quickcheck
Diffstat (limited to 'lib/unicode')
-rw-r--r--lib/unicode/string/u8norm_nfd.c38
1 files changed, 3 insertions, 35 deletions
diff --git a/lib/unicode/string/u8norm_nfd.c b/lib/unicode/string/u8norm_nfd.c
index 8f142e1..efff81a 100644
--- a/lib/unicode/string/u8norm_nfd.c
+++ b/lib/unicode/string/u8norm_nfd.c
@@ -5,9 +5,6 @@
#include "unicode/prop.h"
#include "unicode/string.h"
-#include <stdio.h>
-
-static size_t quickchk_spn(struct u8view);
static void decomp(char8_t *, size_t *, size_t, rune);
/* Computed using a gen/scale-norm.c */
@@ -36,18 +33,9 @@ u8norm_nfd(size_t *dstn, struct u8view src, alloc_fn alloc, void *ctx)
size_t bufsz = src.len * NFD_SCALE;
uint8_t *dst = alloc(ctx, nullptr, 0, src.len, NFD_SCALE, alignof(char8_t));
- /* Copy over the initial codepoints that are already in NFD; if the entire
- string is in NFD then just return it immediately */
- size_t spn = quickchk_spn(src);
- memcpy(dst, src.p, spn);
- *dstn = spn;
- if (spn == src.len)
- return dst;
- VSHFT(&src, spn);
-
- rune ch;
- while (u8next(&ch, &src) != 0)
- decomp(dst, dstn, bufsz, ch);
+ *dstn = 0;
+ for (rune ch; u8next(&ch, &src) != 0; decomp(dst, dstn, bufsz, ch))
+ ;
return alloc(ctx, dst, src.len, *dstn, 1, alignof(char8_t));
}
@@ -104,23 +92,3 @@ out:
}
#undef WRITE
-
-size_t
-quickchk_spn(struct u8view src)
-{
- rune ch;
- size_t spn = 0;
- enum uprop_ccc prv, cur;
-
- prv = cur = CCC_NR;
-
- for (int w; w = u8next(&ch, &src); spn += w) {
- if (uprop_get_nfd_qc(ch) == NFD_QC_N)
- break;
- if ((cur = uprop_get_ccc(ch)) < prv)
- break;
- prv = cur;
- }
-
- return spn;
-}