UnicodeCIExceptKanaCIKanaWithVoicedSoundMark: fix a wrong type bug

This bug is caused when a fullwidth character is normalized to a halfwidth character. [groonga-dev,02108] Reported by yoku. Thanks!!! [groonga-dev,02110] Fixed by Naoya Murakami. Thanks!!!
groonga · Feb 5, 2014 · 1507963 · 1507963
1 parent 153e287
commit 1507963
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 1 deletion.
diff --git a/normalizers/mysql.c b/normalizers/mysql.c
@@ -369,7 +369,7 @@ normalize(grn_ctx *ctx, grn_obj *string,
       if (current_type) {
         char *current_normalized;
         current_normalized =
-          normalized + normalized_length_in_bytes - character_length;
+          normalized + normalized_length_in_bytes - normalized_character_length;
         current_type[0] =
           grn_nfkc_char_type((unsigned char *)current_normalized);
         current_type++;

diff --git a/test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/with_types.expected b/test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/with_types.expected
@@ -0,0 +1,28 @@
+register normalizers/mysql
+[[0,0.0,0.0],true]
+normalize NormalizerMySQLUnicodeCIExceptKanaCIKanaWithVoicedSoundMark   "a1!！あアｱ亜💕 " WITH_TYPES
+[
+  [
+    0,
+    0.0,
+    0.0
+  ],
+  {
+    "normalized": "A1!!あああ亜💕 ",
+    "types": [
+      "alpha",
+      "digit",
+      "symbol",
+      "symbol",
+      "hiragana",
+      "hiragana",
+      "hiragana",
+      "kanji",
+      "others",
+      "others"
+    ],
+    "checks": [
+
+    ]
+  }
+]
diff --git a/test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/with_types.test b/test/suite/unicode_ci_except_kana_ci_kana_with_voiced_sound_mark/with_types.test
@@ -0,0 +1,4 @@
+register normalizers/mysql
+
+normalize NormalizerMySQLUnicodeCIExceptKanaCIKanaWithVoicedSoundMark \
+  "a1!！あアｱ亜💕 " WITH_TYPES