diff options
| author | naruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e> | 2008-06-01 18:29:08 +0000 |
|---|---|---|
| committer | naruse <naruse@b2dd03c8-39d4-4d8f-98ff-823fe69b080e> | 2008-06-01 18:29:08 +0000 |
| commit | 9c13fc7d899582fe8b0612c609595917c42d68a6 (patch) | |
| tree | e5e2eaec8bc4a474b57241fd5345f9492060e88b | |
| parent | ccb61351eab257e9b31ae6c7718febb237027ee8 (diff) | |
* enc/gb18030.c (gb18030_mbc_to_code): mask by 0x7FFFFFFF
because OnigCodePoint will be used as 32bit signed int.
Masking by 0x7FFFFFFF is ok on GB18030;
Minumum 4bytes character is 0x81308130.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@16737 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
| -rw-r--r-- | ChangeLog | 7 | ||||
| -rw-r--r-- | enc/gb18030.c | 15 |
2 files changed, 21 insertions, 1 deletions
@@ -1,3 +1,10 @@ +Mon Jun 2 03:23:25 2008 NARUSE, Yui <naruse@ruby-lang.org> + + * enc/gb18030.c (gb18030_mbc_to_code): mask by 0x7FFFFFFF + because OnigCodePoint will be used as 32bit signed int. + Masking by 0x7FFFFFFF is ok on GB18030; + Minumum 4bytes character is 0x81308130. + Sun Jun 1 22:29:35 2008 NARUSE, Yui <naruse@ruby-lang.org> * rational.c (string_to_r_internal): use rb_isdigit. diff --git a/enc/gb18030.c b/enc/gb18030.c index 231b961188..f62a2394c9 100644 --- a/enc/gb18030.c +++ b/enc/gb18030.c @@ -166,7 +166,20 @@ gb18030_mbc_enc_len(const UChar* p, const UChar* e, OnigEncoding enc ARG_UNUSED) static OnigCodePoint gb18030_mbc_to_code(const UChar* p, const UChar* end, OnigEncoding enc) { - return onigenc_mbn_mbc_to_code(enc, p, end); + int c, i, len; + OnigCodePoint n; + + len = enclen(enc, p, end); + n = (OnigCodePoint )(*p++); + if (len == 1) return n; + + for (i = 1; i < len; i++) { + if (p >= end) break; + c = *p++; + n <<= 8; n += c; + } + n &= 0x7FFFFFFF; + return n; } static int |