summaryrefslogtreecommitdiff
path: root/enc
diff options
context:
space:
mode:
authorNobuyoshi Nakada <[email protected]>2024-09-20 16:14:07 +0900
committerNobuyoshi Nakada <[email protected]>2024-09-20 16:23:55 +0900
commitaa2662d5906cb3d7b55b42539b39976f5a32856a (patch)
tree7c4f4ee12ae1ccd413cee8d6c775a36609bea726 /enc
parent75a8c937e75a0b57fbdbc3d72a2429f5799a42e4 (diff)
Simplify offset calculations of `tbl0208`
Notes
Notes: Merged: https://github.com/ruby/ruby/pull/5696
Diffstat (limited to 'enc')
-rw-r--r--enc/trans/iso2022.trans38
1 files changed, 25 insertions, 13 deletions
diff --git a/enc/trans/iso2022.trans b/enc/trans/iso2022.trans
index e77cf0fc30..a25a4a12a1 100644
--- a/enc/trans/iso2022.trans
+++ b/enc/trans/iso2022.trans
@@ -436,15 +436,24 @@ rb_cp50221_encoder = {
/* JIS0201 to JIS0208 conversion table */
enum {tbl0208_num = 0xDF - 0xA1 + 1};
-static const char tbl0208[2 * tbl0208_num] =
- "\x21\x23\x21\x56\x21\x57\x21\x22\x21\x26\x25\x72\x25\x21\x25\x23" \
- "\x25\x25\x25\x27\x25\x29\x25\x63\x25\x65\x25\x67\x25\x43\x21\x3C" \
- "\x25\x22\x25\x24\x25\x26\x25\x28\x25\x2A\x25\x2B\x25\x2D\x25\x2F" \
- "\x25\x31\x25\x33\x25\x35\x25\x37\x25\x39\x25\x3B\x25\x3D\x25\x3F" \
- "\x25\x41\x25\x44\x25\x46\x25\x48\x25\x4A\x25\x4B\x25\x4C\x25\x4D" \
- "\x25\x4E\x25\x4F\x25\x52\x25\x55\x25\x58\x25\x5B\x25\x5E\x25\x5F" \
- "\x25\x60\x25\x61\x25\x62\x25\x64\x25\x66\x25\x68\x25\x69\x25\x6A" \
- "\x25\x6B\x25\x6C\x25\x6D\x25\x6F\x25\x73\x21\x2B\x21\x2C";
+static const char tbl0208[tbl0208_num][2] = {
+ "\x21\x23", "\x21\x56", "\x21\x57", "\x21\x22",
+ "\x21\x26", "\x25\x72", "\x25\x21", "\x25\x23",
+ "\x25\x25", "\x25\x27", "\x25\x29", "\x25\x63",
+ "\x25\x65", "\x25\x67", "\x25\x43", "\x21\x3C",
+ "\x25\x22", "\x25\x24", "\x25\x26", "\x25\x28",
+ "\x25\x2A", "\x25\x2B", "\x25\x2D", "\x25\x2F",
+ "\x25\x31", "\x25\x33", "\x25\x35", "\x25\x37",
+ "\x25\x39", "\x25\x3B", "\x25\x3D", "\x25\x3F",
+ "\x25\x41", "\x25\x44", "\x25\x46", "\x25\x48",
+ "\x25\x4A", "\x25\x4B", "\x25\x4C", "\x25\x4D",
+ "\x25\x4E", "\x25\x4F", "\x25\x52", "\x25\x55",
+ "\x25\x58", "\x25\x5B", "\x25\x5E", "\x25\x5F",
+ "\x25\x60", "\x25\x61", "\x25\x62", "\x25\x64",
+ "\x25\x66", "\x25\x68", "\x25\x69", "\x25\x6A",
+ "\x25\x6B", "\x25\x6C", "\x25\x6D", "\x25\x6F",
+ "\x25\x73", "\x21\x2B", "\x21\x2C"
+};
static ssize_t
fun_so_cp50220_encoder(void *statep, const unsigned char *s, size_t l,
@@ -455,17 +464,19 @@ fun_so_cp50220_encoder(void *statep, const unsigned char *s, size_t l,
if (sp[0] == G0_JISX0201_KATAKANA && sp[2]) {
int c = sp[2] & 0x7F;
- const char *p = tbl0208 + (c - 0x21) * 2;
+ const char *p = tbl0208[c - 0x21];
sp[2] = 0;
o = iso2022jp_put_state(sp, o, sp[1], G0_JISX0208_1983);
sp[0] = G0_JISX0208_1983;
*o++ = *p++;
if (l == 2 && s[0] == 0x8E) {
if (s[1] == 0xDE) {
+ /* VOICED SOUND MARK */
*o++ = *p + 1;
return o - output0;
}
else if (s[1] == 0xDF && (0x4A <= c && c <= 0x4E)) {
+ /* SEMI-VOICED SOUND MARK */
*o++ = *p + 2;
return o - output0;
}
@@ -477,7 +488,8 @@ fun_so_cp50220_encoder(void *statep, const unsigned char *s, size_t l,
if ((0xA1 <= s[1] && s[1] <= 0xB5) ||
(0xC5 <= s[1] && s[1] <= 0xC9) ||
(0xCF <= s[1] && s[1] <= 0xDF)) {
- const char *p = tbl0208 + (s[1] - 0xA1) * 2;
+ /* May not be followed by a sound mark */
+ const char *p = tbl0208[s[1] - 0xA1];
o = iso2022jp_put_state(sp, o, *sp, G0_JISX0208_1983);
*o++ = *p++;
*o++ = *p;
@@ -491,6 +503,7 @@ fun_so_cp50220_encoder(void *statep, const unsigned char *s, size_t l,
return o - output0;
}
+ /* Katakana that may be followed by a sound mark */
sp[2] = s[1];
sp[1] = sp[0];
sp[0] = G0_JISX0201_KATAKANA;
@@ -512,7 +525,7 @@ finish_cp50220_encoder(void *statep, unsigned char *o, size_t osize)
if (sp[0] == G0_JISX0201_KATAKANA && sp[2]) {
int c = sp[2] & 0x7F;
- const char *p = tbl0208 + (c - 0x21) * 2;
+ const char *p = tbl0208[c - 0x21];
o = iso2022jp_put_state(sp, o, sp[1], G0_JISX0208_1983);
sp[0] = G0_JISX0208_1983;
*o++ = *p++;
@@ -549,4 +562,3 @@ TRANS_INIT(iso2022)
rb_register_transcoder(&rb_cp50220_encoder);
rb_register_transcoder(&rb_cp50221_encoder);
}
-