summaryrefslogtreecommitdiff
path: root/enc/utf_32be.c
diff options
context:
space:
mode:
Diffstat (limited to 'enc/utf_32be.c')
-rw-r--r--enc/utf_32be.c27
1 files changed, 22 insertions, 5 deletions
diff --git a/enc/utf_32be.c b/enc/utf_32be.c
index 43c07e2e8f..e05cfaf1b2 100644
--- a/enc/utf_32be.c
+++ b/enc/utf_32be.c
@@ -28,12 +28,25 @@
*/
#include "regenc.h"
+#include "iso_8859.h"
+static OnigCodePoint utf32be_mbc_to_code(const UChar* p, const UChar* end, OnigEncoding enc);
static int
-utf32be_mbc_enc_len(const UChar* p ARG_UNUSED, const OnigUChar* e ARG_UNUSED,
- OnigEncoding enc ARG_UNUSED)
+utf32be_mbc_enc_len(const UChar* p ARG_UNUSED, const OnigUChar* e,
+ OnigEncoding enc)
{
- return 4;
+ if (e < p) {
+ return ONIGENC_CONSTRUCT_MBCLEN_INVALID();
+ }
+ else if (e-p < 4) {
+ return ONIGENC_CONSTRUCT_MBCLEN_NEEDMORE(4-(int)(e-p));
+ }
+ else {
+ OnigCodePoint c = utf32be_mbc_to_code(p, e, enc);
+ if (!UNICODE_VALID_CODEPOINT_P(c))
+ return ONIGENC_CONSTRUCT_MBCLEN_INVALID();
+ return ONIGENC_CONSTRUCT_MBCLEN_CHARFOUND(4);
+ }
}
static int
@@ -126,7 +139,7 @@ utf32be_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* e
int c, v;
p += 3;
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
+ if (*p == SHARP_s && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
return TRUE;
}
@@ -186,8 +199,12 @@ OnigEncodingDefine(utf_32be, UTF_32BE) = {
onigenc_utf16_32_get_ctype_code_range,
utf32be_left_adjust_char_head,
onigenc_always_false_is_allowed_reverse_match,
+#ifdef USE_CASE_MAP_API
+ onigenc_unicode_case_map,
+#else
+ NULL,
+#endif
0,
ONIGENC_FLAG_UNICODE,
};
ENC_ALIAS("UCS-4BE", "UTF-32BE")
-