diff options
Diffstat (limited to 'enc')
| -rw-r--r-- | enc/utf_32be.c | 18 | ||||
| -rw-r--r-- | enc/utf_32le.c | 18 |
2 files changed, 30 insertions, 6 deletions
diff --git a/enc/utf_32be.c b/enc/utf_32be.c index 995c9d8ed5..17841e52a4 100644 --- a/enc/utf_32be.c +++ b/enc/utf_32be.c @@ -30,11 +30,23 @@ #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 diff --git a/enc/utf_32le.c b/enc/utf_32le.c index e255f0e246..18b798f102 100644 --- a/enc/utf_32le.c +++ b/enc/utf_32le.c @@ -30,11 +30,23 @@ #include "regenc.h" #include "iso_8859.h" +static OnigCodePoint utf32le_mbc_to_code(const UChar* p, const UChar* end, OnigEncoding enc); static int -utf32le_mbc_enc_len(const UChar* p ARG_UNUSED, const OnigUChar* e ARG_UNUSED, - OnigEncoding enc ARG_UNUSED) +utf32le_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 = utf32le_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 |