Kouhei Sutou
null+****@clear*****
Sat May 18 15:44:27 JST 2013
Kouhei Sutou 2013-05-18 15:44:27 +0900 (Sat, 18 May 2013) New Revision: d167006e4d3c16ceef38aef5da5aeba30e2bfee4 https://github.com/groonga/groonga-normalizer-mysql/commit/d167006e4d3c16ceef38aef5da5aeba30e2bfee4 Message: Ignore U+0000 as normalized character https://gist.github.com/y-ken/eb49eaad879f47b27cec Reported by Y.Kentaro. Thanks!!! Added files: test/suite/unicode_ci/halfwidth_katakana_voiced_sound_mark.expected test/suite/unicode_ci/halfwidth_katakana_voiced_sound_mark.test test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/halfwidth_katakana_voiced_sound_mark.expected test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/halfwidth_katakana_voiced_sound_mark.test Modified files: normalizers/mysql.c Modified: normalizers/mysql.c (+5 -3) =================================================================== --- normalizers/mysql.c 2013-05-09 17:34:10 +0900 (de2f9f5) +++ normalizers/mysql.c 2013-05-18 15:44:27 +0900 (413b7e9) @@ -159,9 +159,11 @@ normalize(grn_ctx *ctx, grn_obj *string, uint32_t **normalize_table) uint32_t normalized_code; unsigned int n_bytes; normalized_code = normalize_table[page][low_code]; - n_bytes = unichar_to_utf8(normalized_code, - normalized + normalized_length_in_bytes); - normalized_length_in_bytes += n_bytes; + if (normalized_code != 0) { + n_bytes = unichar_to_utf8(normalized_code, + normalized + normalized_length_in_bytes); + normalized_length_in_bytes += n_bytes; + } } else { int i; for (i = 0; i < character_length; i++) { Added: test/suite/unicode_ci/halfwidth_katakana_voiced_sound_mark.expected (+4 -0) 100644 =================================================================== --- /dev/null +++ test/suite/unicode_ci/halfwidth_katakana_voiced_sound_mark.expected 2013-05-18 15:44:27 +0900 (685fd1f) @@ -0,0 +1,4 @@ +register normalizers/mysql +[[0,0.0,0.0],true] +normalize NormalizerMySQLUnicodeCI "beforeｹﾞafter" +[[0,0.0,0.0],{"normalized":"BEFOREけAFTER","types":[]}] Added: test/suite/unicode_ci/halfwidth_katakana_voiced_sound_mark.test (+3 -0) 100644 =================================================================== --- /dev/null +++ test/suite/unicode_ci/halfwidth_katakana_voiced_sound_mark.test 2013-05-18 15:44:27 +0900 (c9f0a5e) @@ -0,0 +1,3 @@ +register normalizers/mysql + +normalize NormalizerMySQLUnicodeCI "beforeｹﾞafter" Added: test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/halfwidth_katakana_voiced_sound_mark.expected (+4 -0) 100644 =================================================================== --- /dev/null +++ test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/halfwidth_katakana_voiced_sound_mark.expected 2013-05-18 15:44:27 +0900 (3ff6818) @@ -0,0 +1,4 @@ +register normalizers/mysql +[[0,0.0,0.0],true] +normalize NormalizerMySQLUnicodeCIExceptKanaCIKanaWithVoicedSoundMark "beforeｹﾞafter" +[[0,0.0,0.0],{"normalized":"BEFOREけAFTER","types":[]}] Added: test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/halfwidth_katakana_voiced_sound_mark.test (+4 -0) 100644 =================================================================== --- /dev/null +++ test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/halfwidth_katakana_voiced_sound_mark.test 2013-05-18 15:44:27 +0900 (c6ac08f) @@ -0,0 +1,4 @@ +register normalizers/mysql + +normalize NormalizerMySQLUnicodeCIExceptKanaCIKanaWithVoicedSoundMark \ + "beforeｹﾞafter" -------------- next part -------------- HTML����������������������������...Download