在GreenDao开源代码中添加FTS4全文检索功能

GreenDao并不支持全文检索，导致在大量数据中的查询效率低下。本文介绍了如何在GreenDao中添加支持全文检索的功能，希望对GreenDao粉有所帮助。
本文基于GreenDao V3.2.3 + rk3288 android7.1.2实测有效，读者有任何问题可留言沟通。

1.修改DaoGeneratorV3

1)Entity.java增加一个属性useFts，用来标识是否使用全文检索。

private boolean useFts;//lihuili add 20210310 for FTS public boolean isUseFts() {return useFts;}public void setUseFts(boolean useFts) {this.useFts = useFts;}Entity(Schema schema, String className) {this.schema = schema;this.className = className;properties = new ArrayList<Property>();propertiesPk = new ArrayList<Property>();propertiesNonPk = new ArrayList<Property>();propertyNames = new HashSet<String>();indexes = new ArrayList<Index>();multiIndexes = new ArrayList<Index>();toOneRelations = new ArrayList<ToOne>();toManyRelations = new ArrayList<ToManyBase>();incomingToManyRelations = new ArrayList<ToManyBase>();additionalImportsEntity = new TreeSet<String>();additionalImportsDao = new TreeSet<String>();interfacesToImplement = new ArrayList<String>();contentProviders = new ArrayList<ContentProvider>();constructors = true;useFts = false;}

2)模板dao.ftl增加：如果entity.useFts为true，就创建使用FTS4的虚表

<!-- lihuili add 20210310 for FTS --><#if entity.useFts>/** Creates the underlying database table. */public static void createTable(Database db, boolean ifNotExists) {String constraint = ifNotExists? "IF NOT EXISTS ": "";db.execSQL("CREATE VIRTUAL TABLE " + constraint + "\"${entity.dbName}\" USING FTS4 (" + //<#list entity.propertiesColumns as property>"\"${property.dbName}\" ${property.dbType}<#if property.constraints??> ${property.constraints} </#if><#if property_has_next>," +<#else>);");</#if> // ${property_index}: ${property.propertyName}</#list><#if entity.indexes?has_content >// Add Indexes<#list entity.indexes as index>db.execSQL("CREATE <#if index.unique>UNIQUE </#if>INDEX " + constraint + "${index.name} ON \"${entity.dbName}\"" +" (<#list index.properties as property>\"${property.dbName}\"<#if (index.propertiesOrder[property_index])??> ${index.propertiesOrder[property_index]}</#if><#sep>,</#list>);");</#list></#if>}<#else>/** Creates the underlying database table. */public static void createTable(Database db, boolean ifNotExists) {String constraint = ifNotExists? "IF NOT EXISTS ": "";db.execSQL("CREATE TABLE " + constraint + "\"${entity.dbName}\" (" + //<#list entity.propertiesColumns as property>"\"${property.dbName}\" ${property.dbType}<#if property.constraints??> ${property.constraints} </#if><#if property_has_next>," +<#else>);");</#if> // ${property_index}: ${property.propertyName}</#list><#if entity.indexes?has_content >// Add Indexes<#list entity.indexes as index>db.execSQL("CREATE <#if index.unique>UNIQUE </#if>INDEX " + constraint + "${index.name} ON \"${entity.dbName}\"" +" (<#list index.properties as property>\"${property.dbName}\"<#if (index.propertiesOrder[property_index])??> ${index.propertiesOrder[property_index]}</#if><#sep>,</#list>);");</#list></#if>         }</#if>

2.修改自动生成代码的java文件

以DaoGeneratorFaceCollectSTD.java为例：
1)创建一个15列的普通表check_list


```java
private static Entity addCheck_list(Schema schema){Entity check_list = schema.addEntity("check_list");check_list.addLongProperty("item_id").primaryKey().autoincrement();check_list.addLongProperty("file_id");check_list.addStringProperty("name");check_list.addStringProperty("identy_no");check_list.addStringProperty("col3");check_list.addStringProperty("col4");check_list.addStringProperty("col5");check_list.addStringProperty("col6");check_list.addStringProperty("col7");check_list.addStringProperty("col8");check_list.addStringProperty("col9");check_list.addStringProperty("col10");check_list.addStringProperty("col11");check_list.addStringProperty("col12");check_list.addStringProperty("col13");check_list.addStringProperty("col14");check_list.addStringProperty("col15");      return check_list;}


2)创建一个3列的虚表，这3列与普通表完全一样
关键点：需要调用setUseFts(true)，明确告诉程序创建虚表。```javaprivate static Entity addCheck_list_fts(Schema schema){Entity check_list_fts = schema.addEntity("check_list_fts");check_list_fts.setUseFts(true);//使用FTS4check_list_fts.addLongProperty("item_id").primaryKey().autoincrement();        check_list_fts.addStringProperty("name");check_list_fts.addStringProperty("identy_no");         return check_list_fts;
}

3)在main函数中调用上面封装的函数

public static void main(String[] args) throws Exception {Schema schema = new Schema(SCHEMA_VERSION_INFO_DB, "com.routon.idr.facecollect.std.dao.info");schema.enableActiveEntitiesByDefault();schema.enableKeepSectionsByDefault();addCommonParam(schema);addFaceInfo(schema);addCheckfile_info(schema);addCheck_list(schema);addCheck_list_fts(schema);new DaoGenerator().generateAll(schema, "../iDRFaceCollectSTD/src");}

4)Run As Java Application自动生成代码

3.修改DaoCoreV3

1)修改Property.java，增加支持match
注意：一定要带问号！

public class Property {...public WhereCondition match(Object value) {return new PropertyCondition(this, " match?", value);}
}

4.封装FTS查询函数

以FaceInfoCtrl.java为例：
1)函数QueryCheckListFtsByIdentyNo()调用match()构造where条件

public List<Check_list_fts> QueryCheckListFtsByIdentyNo(String identy_no){if(!TextUtils.isEmpty(identy_no)){Check_list_ftsEntityDao entityDao = new Check_list_ftsEntityDao(getDBManager());if(entityDao!=null){Check_list_ftsDao dao = entityDao.getEntityDao();QueryBuilder<Check_list_fts> builder = dao.queryBuilder();builder = builder.where(Check_list_ftsDao.Properties.Identy_no.match(identy_no));long beginTick = SystemClock.uptimeMillis();List<Check_list_fts> checkListFtss = builder.build().list();long endTick = SystemClock.uptimeMillis();Log.d(TAG, "match " + identy_no + " use " + (endTick-beginTick) + "ms");if(checkListFtss!=null && checkListFtss.size()>0){Log.d(TAG, checkListFtss.get(0).getName());}return checkListFtss;}}return null;}

5.验证测试FTS4全文检索

1)先单独运行一下添加了FTS支持的test.apk，生成一个数据库文件face_info.db, 该文件中已经自动创建了空表check_list_fts
2)创建一个插入65535条记录的sql脚本文件CHECK_LIST_FTS.sql，例：

INSERT INTO "CHECK_LIST_FTS" VALUES ('1', '夏紫寒', '642401194107180628');
INSERT INTO "CHECK_LIST_FTS" VALUES ('2', '范梦松', '220904201505271831');
INSERT INTO "CHECK_LIST_FTS" VALUES ('3', '魏孤岚', '612232196501110123');

…
3)从终端pull face_info.db到pc
4)在pc使用在navicat premium把CHECK_LIST_FTS.sql导入到CHECK_LIST_FTS表
5)从pc push face_info.db到终端
6)在test.apk中添加调用FTS查询的函数QueryCheckListFtsByIdentyNo()
7)对比测试普通查询和FTS4全文检索的效率
在65535条记录中，普通查询平均耗时190ms，FTS4全文检索平均耗时5ms！