全局 - 搜索升級 - 分詞搜索

數據庫配置修改：

1. 設置ngram_token_size = 1 2. 設置innodb_ft_server_stopword_table 2.1 建立自定義中止字規則 CREATE TABLE bbavip_stopwords(value VARCHAR(30)) ENGINE = INNODB; 2.2 設置全局使用指定的中止字規則 SET GLOBAL innodb_ft_server_stopword_table = 'bbavip/bbavip_stopwords'; 3. 重建fulltext index 4. 規劃指定表的fulltext index 4.1 SET GLOBAL innodb_optimize_fulltext_only=ON; 4.2 OPTIMIZE TABLE product_base; 4.3 SET GLOBAL innodb_ft_aux_table = 'fulltextsearch/product_base'; 4.4 select * from information_schema.INNODB_FT_INDEX_TABLE where word = 'a'; 4.5 SET GLOBAL innodb_optimize_fulltext_only=OFF;

 

 

1. 合併fullTextSearch分支

2.新增Fulltext index

3. 新增類 繼承 Against 私有 靜態 抽象類

public static class MatchProductBaseInBoolean extends Against { private static String matchParamDefault = "skuid";   public MatchProductBaseInBoolean(String keyword) { super(null, keyword, matchParamDefault); }   public MatchProductBaseInBoolean(String title, String keyword) { super(title, keyword, matchParamDefault); } }

3.1 配置 matchParamDeafult 爲 要全文索引的列。多個列用","隔開，如(skuid,name,bar_code)

3.2 根據須要實現Agaist構造方法。

3.2.1 title 用來作連表查詢時使用。

3.2.2 keyword 做爲查詢關鍵詞

4. 對外暴露工廠方法

public static String buildMatchInBoolean(String matchParamList, String keyWord) { return new MatchInBoolean(matchParamList, keyWord).toString(); }   public static String buildMatchInBoolean(String tableAlias, String matchParamList, String keyWord) { return new MatchInBoolean(tableAlias, matchParamList, keyWord).toString(); }   public static String buildMatchInBooleanWithClazz(String keyWord, Class<? extends Against> clazz) { try { return clazz.getConstructor(String.class).newInstance(keyWord).toString(); } catch (InstantiationException e) { e.printStackTrace(); } catch (IllegalAccessException e) { e.printStackTrace(); } catch (InvocationTargetException e) { e.printStackTrace(); } catch (NoSuchMethodException e) { e.printStackTrace(); } return ""; }   public static String buildMatchInBooleanWithClazz(String tableAlias, String keyWord, Class<? extends Against> clazz) { try { return clazz.getConstructor(String.class, String.class).newInstance(tableAlias, keyWord).toString(); } catch (InstantiationException e) { e.printStackTrace(); } catch (IllegalAccessException e) { e.printStackTrace(); } catch (InvocationTargetException e) { e.printStackTrace(); } catch (NoSuchMethodException e) { e.printStackTrace(); } return ""; }

4.1 buildMatchInBoolean 適用於在嵌套查詢中在父查詢中使用子查詢的全文索引的狀況，或沒法定義Against衍生類的狀況

Long storeBaseId = (Long) map.get("storeBaseId"); String keyWord = (String) map.get("keyWord"); StringBuffer sql = new StringBuffer(); sql.append("SELECT c.*, e.mobile FROM \n" + "(\n" + " SELECT a.id, a.first_name, a.last_name, \n" + " a.gender, a.age , a.car_amount, a.partner_base_id,\n" + " b.car_times, b.pay_expenses,\n" + " a.channel ,\n" + " b.arrive_time, b.car_id FROM car_owner AS a INNER JOIN owner_car_relevance as b\n" + " WHERE a.id = b.car_owner_id \n" + " AND\n" + " b.store_base_id = #{storeBaseId}) as c LEFT JOIN user_car_relation as d ON c.car_id = d.car_id\n" + " LEFT JOIN car_owner_user AS e ON d.user_id = e.id\n"); if(StringUtils.isNotEmpty(keyWord)){ sql.append(FullTextUtils.buildMatchInBoolean("c", "first_name,last_name", keyWord)); } sql.append("ORDER BY id DESC\n"); return sql.toString();

4.2 buildMatchInBooleanWithClazz 適用於能夠定義Against衍生類的狀況。本質上依舊是buildMatchInBoolean方法的調用，只是經過反射處理。

因爲多用於列表查詢，異常沒有向上拋出，調用方須要確保數據的正確性。

if (StringUtils.isNotEmpty(skuid)) { WHERE(FullTextUtils.buildMatchInBooleanWithClazz(skuid, FullTextUtils.MatchProductBaseInBoolean.class)); }else { if (StringUtils.isNotEmpty(orderParam)) { ORDER_BY(orderParam); } else { ORDER_BY("id desc"); } }

5. 使用全文索引時儘可能避免其餘因素的排序，全文索引會有本身的相關性值。添加其餘因素的排序時，返回結果集的順序會被打亂。