|
| 1 | +# |
| 2 | +#----------------------------------------------------------------------------- |
| 3 | +# wl6607 : InnoDB FULLTEXT SEARCH: CJK support (mecab parser) |
| 4 | +# Adding FTS check with mecab parser |
| 5 | +# - Table with mecab parser |
| 6 | +# - Basic FTS query ( mainly BOOLEAN MODE check) |
| 7 | +# - FTS index with single and two columns |
| 8 | +# - external parser index using alter table |
| 9 | +# |
| 10 | +#------------------------------------------------------------------------------ |
| 11 | + |
| 12 | +# restart does not work with embedded |
| 13 | +-- source include/not_embedded.inc |
| 14 | + |
| 15 | +call mtr.add_suppression("unknown variable 'loose_mecab_rc_file"); |
| 16 | +call mtr.add_suppression("Fulltext index charset 'eucjpms' doesn't match mecab charset"); |
| 17 | +call mtr.add_suppression("Mecab: param.cpp"); |
| 18 | +call mtr.add_suppression("Plugin 'mecab' init function returned error"); |
| 19 | + |
| 20 | +-- disable_query_log |
| 21 | + |
| 22 | +-- error 0, ER_CANT_INITIALIZE_UDF |
| 23 | +eval INSTALL PLUGIN mecab SONAME '$MECAB'; |
| 24 | + |
| 25 | +let $mecab_charset=`SELECT variable_value FROM INFORMATION_SCHEMA.GLOBAL_STATUS WHERE VARIABLE_NAME='mecab_charset'`; |
| 26 | +if ($mecab_charset == '') { |
| 27 | + # Restart with package dictionary. |
| 28 | + let $MYSQL_DATADIR=`select @@datadir`; |
| 29 | + let $MYSQL_BASEDIR=`select @@basedir`; |
| 30 | + |
| 31 | + let $mecabrc = $MYSQL_DATADIR/mecabrc; |
| 32 | + let $dicdir = $MYSQL_BASEDIR/lib/mecab/dic/ipadic_$ipadic_charset; |
| 33 | + |
| 34 | + -- exec echo "dicdir=$dicdir" > $mecabrc |
| 35 | + |
| 36 | + -- source include/shutdown_mysqld.inc |
| 37 | + -- exec echo "restart: --loose_mecab_rc_file=$mecabrc $MECAB_OPT --innodb_ft_min_token_size=2" >$MYSQLTEST_VARDIR/tmp/mysqld.1.expect |
| 38 | + -- enable_reconnect |
| 39 | + -- source include/wait_until_connected_again.inc |
| 40 | + -- disable_reconnect |
| 41 | + |
| 42 | + -- disable_query_log |
| 43 | + |
| 44 | + -- error 0, ER_CANT_INITIALIZE_UDF |
| 45 | + eval INSTALL PLUGIN mecab SONAME '$MECAB'; |
| 46 | + |
| 47 | + let $mecab_charset=`SELECT variable_value FROM INFORMATION_SCHEMA.GLOBAL_STATUS WHERE VARIABLE_NAME='mecab_charset'`; |
| 48 | +} |
| 49 | + |
| 50 | +if ($mecab_charset == '') { |
| 51 | + -- skip Test fail to load mecab parser, please set correct 'loose_mecab_rc_file'. |
| 52 | +} |
| 53 | + |
| 54 | +if ($mecab_charset != $mysql_charset) { |
| 55 | + UNINSTALL PLUGIN mecab; |
| 56 | + -- skip Test mecab charset mismatch. |
| 57 | +} |
| 58 | + |
| 59 | +--enable_query_log |
| 60 | + |
| 61 | +SHOW STATUS LIKE 'mecab_charset'; |
| 62 | + |
| 63 | +SET NAMES utf8; |
| 64 | +eval CREATE TABLE page ( |
| 65 | + page_id int(8) unsigned NOT NULL AUTO_INCREMENT, |
| 66 | + page_title varchar(255) CHARACTER SET $mecab_charset NOT NULL DEFAULT '', |
| 67 | + PRIMARY KEY (page_id), |
| 68 | + FULLTEXT KEY page_title (page_title) WITH PARSER mecab |
| 69 | +) ENGINE=InnoDB; |
| 70 | + |
| 71 | +INSERT INTO page VALUES |
| 72 | +(1,'レモナ'), |
| 73 | +(2,'SDレモナ'), |
| 74 | +(3,'レモナ'), |
| 75 | +(4,'データベース管理システム'), |
| 76 | +(5,'キム'), |
| 77 | +(6,'鉄道100電車'), |
| 78 | +(7,'100'), |
| 79 | +(8,'ラフィタ・カスティージョ'), |
| 80 | +(9,'ツル科_(Sibley)'), |
| 81 | +(10,'ツル'), |
| 82 | +(11,'鉄道'); |
| 83 | + |
| 84 | +SET GLOBAL innodb_ft_aux_table="test/page"; |
| 85 | +SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_CACHE; |
| 86 | +SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_TABLE; |
| 87 | + |
| 88 | +SHOW CREATE TABLE page; |
| 89 | + |
| 90 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("レモナ" IN BOOLEAN MODE); |
| 91 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("キムショウカン" IN BOOLEAN MODE); |
| 92 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("100" IN BOOLEAN MODE); |
| 93 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("ツル*" IN BOOLEAN MODE); |
| 94 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("ラフィタ・カスティージョ"); |
| 95 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("ラフィタ・カスティージョ" IN BOOLEAN MODE); |
| 96 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("+ラフィタ・カスティージョ +データベース" IN BOOLEAN MODE); |
| 97 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("+ラフィタ・カスティージョ +(データベース)" IN BOOLEAN MODE); |
| 98 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("-ラフィタ・カスティージョ +(データベース)" IN BOOLEAN MODE); |
| 99 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("ラフィタ・カスティージョ 鉄道" IN BOOLEAN MODE); |
| 100 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("<ラフィタ・カスティージョ >鉄道" IN BOOLEAN MODE); |
| 101 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST('鉄道'); |
| 102 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST('鉄道' WITH QUERY EXPANSION); |
| 103 | + |
| 104 | +eval ALTER TABLE page add column c1 varchar(100) CHARACTER SET $mecab_charset DEFAULT 'レモナ'; |
| 105 | +ALTER TABLE page ADD FULLTEXT INDEX idx1 (page_title,c1) WITH PARSER mecab; |
| 106 | +SELECT * FROM page WHERE MATCH(page_title,c1) AGAINST('レモナ'); |
| 107 | +SELECT * FROM page WHERE MATCH(page_title,c1) AGAINST('-レモナ' IN BOOLEAN MODE); |
| 108 | + |
| 109 | +DROP TABLE page; |
| 110 | + |
| 111 | +# Test charset mismatch. |
| 112 | +CREATE TABLE page ( |
| 113 | + page_id int(8) unsigned NOT NULL AUTO_INCREMENT, |
| 114 | + page_title varchar(255) CHARACTER SET eucjpms NOT NULL DEFAULT '', |
| 115 | + PRIMARY KEY (page_id) |
| 116 | +)ENGINE=InnoDB; |
| 117 | + |
| 118 | +INSERT INTO page VALUES (1,'レモナ'); |
| 119 | + |
| 120 | +# We don't return charset mismatch error because of parallel fts build. |
| 121 | +CREATE FULLTEXT INDEX ft_idx ON page(page_title) WITH PARSER mecab; |
| 122 | + |
| 123 | +SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_CACHE; |
| 124 | +SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_TABLE; |
| 125 | + |
| 126 | +--error ER_ERROR_ON_WRITE |
| 127 | +INSERT INTO page VALUES (2,'レモナ'); |
| 128 | + |
| 129 | +--error ER_ERROR_ON_WRITE |
| 130 | +SELECT * FROM page WHERE MATCH(page_title) AGAINST("レモナ"); |
| 131 | + |
| 132 | +DROP TABLE page; |
| 133 | + |
| 134 | +SET GLOBAL innodb_ft_aux_table=default; |
| 135 | + |
| 136 | +UNINSTALL PLUGIN mecab; |
0 commit comments