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 | 83557084388514631eb0e0113683071942b21848 (patch) | |
tree | e5e2eaec8bc4a474b57241fd5345f9492060e88b /enc | |
parent | a13ea5f81f90cf3a1c8e005eac3d3c265f8de9d7 (diff) | |
download | ruby-83557084388514631eb0e0113683071942b21848.tar.gz ruby-83557084388514631eb0e0113683071942b21848.tar.xz ruby-83557084388514631eb0e0113683071942b21848.zip |
* 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: http://svn.ruby-lang.org/repos/ruby/trunk@16737 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Diffstat (limited to 'enc')
-rw-r--r-- | enc/gb18030.c | 15 |
1 files changed, 14 insertions, 1 deletions
diff --git a/enc/gb18030.c b/enc/gb18030.c index 231b96118..f62a2394c 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 |