groonga

groongaとtritonnとで、テキストの前方一致検索をした場合、検索結果が異なる場合がある。

groongaの場合：

table_create --name searchtest  --flags TABLE_HASH_KEY --key_type UInt8
table_create --name searchtest_index --flags TABLE_PAT_KEY|KEY_NORMALIZE --type ShortText --default_tokenizer TokenBigram
column_create --table searchtest --name words --flags COLUMN_SCALAR --type ShortText
column_create --table searchtest_index --name word --flags COLUMN_INDEX|WITH_SECTION|WITH_POSITION --type searchtest --source words
load --table searchtest --input_type json --output_type json
[
["_key","words"],
[1,"aki"],
[2,"akihabara"],
]
select searchtest
select searchtest --query words:@aki*
select searchtest --query words:@a*

> table_create --name searchtest  --flags TABLE_HASH_KEY --key_type UInt8
[[0,1277112400.63585,0.001694]]
> table_create --name searchtest_index --flags TABLE_PAT_KEY|KEY_NORMALIZE --type ShortText --default_tokenizer TokenBigram
[[0,1277112405.0999,0.023584]]
> column_create --table searchtest --name words --flags COLUMN_SCALAR --type ShortText
[[0,1277112407.86787,0.02668]]
> column_create --table searchtest_index --name word --flags COLUMN_INDEX|WITH_SECTION|WITH_POSITION --type searchtest --source words
[[0,1277112410.89191,0.043828]]
> load --table searchtest --input_type json --output_type json
> [
> ["_key","words"],
> [1,"aki"],
> [2,"akihabara"],
> ]
[[0,1277112415.21585,0.059942],2]
> select searchtest
[[0,1277112427.7598,0.000244],[[[2],[["_id","UInt32"],["_key","UInt8"],["words","ShortText"]],[1,1,"aki"],[2,2,"akihabara"]]]]
> select searchtest --query words:@aki*
[[0,1277112430.66782,0.000564],[[[1],[["_id","UInt32"],["_key","UInt8"],["words","ShortText"]],[1,1,"aki"]]]]
> select searchtest --query words:@a*
[[0,1277112433.45582,0.000548],[[[1],[["_id","UInt32"],["_key","UInt8"],["words","ShortText"]],[2,2,"akihabara"]]]]
>

tritonnの場合：

CREATE TABLE `searchtest` (
  `_key` int(2) NOT NULL,
  `words` text,
  PRIMARY KEY  (`_key`),
  FULLTEXT KEY `words` (`words`)
) ENGINE=MyISAM DEFAULT CHARSET=utf8;

INSERT INTO `searchtest` VALUES (1,'aki'),(2,'akihabara');
SELECT * FROM `searchtest`;
SELECT * FROM `searchtest` WHERE MATCH(`words`) AGAINST('+aki*' IN BOOLEAN MODE);
SELECT * FROM `searchtest` WHERE MATCH(`words`) AGAINST('+a*' IN BOOLEAN MODE);

mysql> CREATE TABLE `searchtest` (
    ->   `_key` int(2) NOT NULL,
    ->   `words` text,
    ->   PRIMARY KEY  (`_key`),
    ->   FULLTEXT KEY `words` (`words`)
    -> ) ENGINE=MyISAM DEFAULT CHARSET=utf8;
Query OK, 0 rows affected (0.12 sec)

mysql> INSERT INTO `searchtest` VALUES (1,'aki'),(2,'akihabara');
Query OK, 2 rows affected (0.00 sec)
Records: 2  Duplicates: 0  Warnings: 0

mysql> SELECT * FROM `searchtest`;
+------+-----------+
| _key | words     |
+------+-----------+
|    1 | aki       |
|    2 | akihabara |
+------+-----------+
2 rows in set (0.00 sec)

mysql> SELECT * FROM `searchtest` WHERE MATCH(`words`) AGAINST('+aki*' IN BOOLEAN MODE);
+------+-------+
| _key | words |
+------+-------+
|    1 | aki   |
+------+-------+
1 row in set (0.00 sec)

mysql> SELECT * FROM `searchtest` WHERE MATCH(`words`) AGAINST('+a*' IN BOOLEAN MODE);
+------+-----------+
| _key | words     |
+------+-----------+
|    1 | aki       |
|    2 | akihabara |
+------+-----------+
2 rows in set (0.00 sec)

mysql>