自定义数据类型的数据库映射方案

基础数据类型，如String、Integer、Date、Boolean等它们可以很方便的映射到数据库：

import grails.persistence.Entity

@Entity
class MyEntity {
    String code
    String name
    static constraints = {
        code(unique: true, minSize: 4, maxSize: 4)
        name(blank: false, maxSize: 255)
    }
}

这些基础数据类型是JAVA提供的语言级的，它没有语意。

比如要表达一个身份证号码：它有长度限制：15位或18位；还有规则限制；还能从身份证号码中提取出地址、性别、出生日期、年龄等信息。这些信息用一个String是无法表达，需要用类来描述：

class IDNumber{
    String idNumber

    Address address
    InsDate birthday
    Gender gender
    IDNumber() {}
    IDNumber(val) {       
        if (val.length() == 15) {
            val = to18IdNumber(val)
        }
        if (val.length() != 18) {
            throw new IllegalArgumentException("不是身份证格式")
        }
        this.idNumber = val
        return
    }
    def getAddress() {
        if (address) return address
        else return address = parseAddress()
    }
    def getBirthday() {
        if (birthday) return birthday
        else return birthday = parseBirth()
    }
    def getGender() {
        if (gender) return gender
        else return gender = parseGender()
    }
    def parseBirth() {
        ...
    }
}

这个类里面最核心的就是String idNumber身份证号码，其他属性都是暂存的临时数据，可以从身份证号码里解析出来。如果想把这个类映射到数据库中，现在只能映射成一个table，但映射成table又不合理，最好是能映射成一列：

@grails.persistence.Entity
class PersonInfo {
    String name
    IDNumber idNumber
}

现在这样显然是不能达到这个目标的。

Hibernate提供了多种实现自定义类型的方法：

1、实现org.hibernate.usertype.UserType

2、实现org.hibernate.usertype.CompositeUserType

3、实现org.hibernate.usertype.UserCollectionType

4、实现org.hibernate.usertype.EnhanceUserType

通过实现这些接口，可以将自定义数据类型映射成数据库列。

UserType可以映射成单列，CompositeUserType可以映射成多列。

看个例子：

class MyString extends InsDataType implements UserType{
    String value

    @Override
    void buildData(val) {
        if (val instanceof MyString) {
            value = val.value
            return
        }
        if (val == null) value = null
        else if (val instanceof String) value = val
        else if (val instanceof Number) value = String.valueOf(val)
        else value = val.toString()
        return
    }

    static MyString from(val) {
        if (val instanceof MyString) return val
        MyString data = new MyString()
        data.build(val)
        return data
    }

    public String toString() {
        return value
    }

    int[] sqlTypes() {
        return [Types.VARCHAR]
    }

    Class returnedClass() {
        return MyString
    }

    boolean equals(Object x, Object y) {
        MyString mx, my
        if (x instanceof String) mx = MyString.from(x)
        if (x instanceof MyString) mx = x

        if (y instanceof String) my = MyString.from(y)
        if (y instanceof MyString) my = y
        if (mx?.value == my?.value) return true
        return false
    }

    int hashCode(Object x) {
        return ((MyString) x)?.value?.hashCode()
    }

    Object nullSafeGet(ResultSet rs, String[] names, Object owner) {
        if (rs.wasNull()) return null
//        String stringFromDb = (String) Hibernate.STRING.nullSafeGet(rs, names[0]);
        String stringFromDb = rs.getString(names[0]);
        return MyString.from(stringFromDb)
    }

    void nullSafeSet(PreparedStatement st, Object value, int index) {
        if (value == null)
            st.setNull(index, Types.VARCHAR);
        else {
            MyString myString = (MyString) value;
            st.setString(index, myString.value);
//            Hibernate.STRING.nullSafeSet(st, myString.value, index);
        }
    }

    Object deepCopy(Object value) {
        if (!value || !((MyString) value).value) return null
        return MyString.from(value)
    }

    boolean isMutable() {
        return true
    }

    Serializable disassemble(Object value) {
        return ((MyString) value).value
    }

    Object assemble(Serializable cached, Object owner) {
        return MyString.from(cached)
    }

    Object replace(Object original, Object target, Object owner) {
        return null
    }
}

这样就可以将MyString映射到数据库表中的一列了。

@grails.persistence.Entity
class MyEntity {
    MyString name
    static constraints = {
        name(nullable: true)
    }
    static mapping = {
        name(length: 10)
    }
}

数据库结构:

测试保存：

def testSave() {
        MyEntity entity = new MyEntity(name: MyString.from("hehe"))
        TestDomain.withTransaction {
            if (entity.hasErrors() || !entity.save(flush: true)) {
                println "save error:" + entity.errors
            }
        }
    }

数据库记录为：

测试查询：

MyEntity entity = MyEntity.findByName(MyString.from("hehe"))

现在操作自定义的MyString就像操作基础数据类型一样了。

 

如果一个数据类型有多个字段要存储，比如姓名分姓氏和名称。一种方法是把多个字段合并成一个字段，仍然使用UserType。另一种方法是用CompositeUserType。

class MyChineseName implements CompositeUserType {
    String familyName
    String givenName

    String[] getPropertyNames() {
        return ["familyName", "givenName"] as String[]
    }
    Type[] getPropertyTypes() {
        return [Hibernate.STRING, Hibernate.STRING] as Type[]
    }
    Object getPropertyValue(Object component, int property) {
        MyChineseName name = (MyChineseName) component;
        String result;
        switch (property) {
            case 0:
                result = name.familyName;
                break;
            case 1:
                result = name.givenName;
                break;
            default:
                throw new IllegalArgumentException("unknow property: " + property);
        }
        return result;
    }
    void setPropertyValue(Object component, int property, Object value) {
        MyChineseName name = (MyChineseName) component;
        String nameValue = (String) value;
        switch (property) {
            case 0:
                name.familyName = nameValue
                break;
            case 1:
                name.givenName = nameValue
                break;
            default:
                throw new IllegalArgumentException("unknow property: " + property);
        }
    }
    Class returnedClass() {
        return MyChineseName
    }
    boolean equals(Object x, Object y) {
        if (x == y)
            return true;
        if (x == null || y == null)
            return false;
        return x.equals(y);
    }
    int hashCode(Object x) {
        return x.hashCode()
    }
    Object nullSafeGet(ResultSet rs, String[] names, SessionImplementor session, Object owner) {
        if (rs.wasNull())
            return null;
        String firstname = rs.getString(names[0]);
        String lastname = rs.getString(names[1]);
        return new MyChineseName(familyName: firstname, givenName: lastname);
    }
    void nullSafeSet(PreparedStatement statement, Object value, int index, SessionImplementor session) {
        if (value == null)
            statement.setNull(index, Types.VARCHAR);
        else {
            MyChineseName name = (MyChineseName) value;
//            statement.setString(index, name.familyName);
//            statement.setString(index + 1, name.givenName);
            Hibernate.STRING.nullSafeSet(statement, name.familyName, index + 0);
            Hibernate.STRING.nullSafeSet(statement, name.givenName, index + 1);
        }
    }
    Object deepCopy(Object value) {
        if (value == null)
            return null;
        MyChineseName name = (MyChineseName) value;
        return new MyChineseName(familyName: name.familyName, givenName: name.givenName);
    }
    boolean isMutable() {
        return false
    }
    Serializable disassemble(Object value, SessionImplementor session) {
        return (Serializable) deepCopy(value);
    }
    Object assemble(Serializable cached, SessionImplementor session, Object owner) {
        return (Serializable) deepCopy(cached);
    }
    Object replace(Object original, Object target, SessionImplementor session, Object owner) {
        return null
    }
}

这样，MyChineseName就能够映射成两列了。如果还像上面一样定义Entity类，Hibernate仍然无法映射，必须指定type和column：

@grails.persistence.Entity
class MyEntity {
    MyChineseName name
    static constraints = {
        name(nullable: true)
    }
    static mapping = {
        name type: MyChineseName, {
            column name: "chineseFamilyName", length: 10
            column name: "chineseGivenName", length: 10
        }
    }
}

生成的数据库表结构：

测试保存：

def testSave() {
        MyEntity entity = new MyEntity(name: new MyChineseName(familyName: "泛", givenName: "华"))
        TestDomain.withTransaction {
            if (entity.hasErrors() || !entity.save(flush: true)) {
                println "save error:" + entity.errors
            }
        }
        println ToStringBuilder.reflectionToString(entity)
    }

数据库记录为：

 

这种方式的麻烦之处在于映射时需要指定type和column。如果用户不清楚它的实现方式，仍然当作普通的UserType，没有指定type和column，那么就会报错：

 

Caused by: org.hibernate.MappingException: property mapping has wrong number of columns: com.baoxian.domain.MyEntity.name type: com.baoxian.datatype.MyChineseName

 

仅仅根据这个错误描述就不太好定位了。

可以把多字段组合成一个字符串，从而映射成一个字段来解决：

class MyChineseName implements UserType {
    String familyName
    String givenName

    String toOneString() {
        return "fn:${familyName};gn:${givenName}"
    }
    MyChineseName parseString(String str) {
        def regular = /(fn|gn):([^;]*)/
        def result = str =~ regular
        def map = [:]
        result.each { map[it[1]] = it[2] }
        return new MyChineseName(familyName: map["fn"], givenName: map["gn"])
    }
    int[] sqlTypes() {
        return [Types.VARCHAR]
    }
    Class returnedClass() {
        return MyChineseName
    }
    boolean equals(Object x, Object y) {
        if (x == y)
            return true;
        if (x == null || y == null)
            return false;
        return x.equals(y);
    }
    int hashCode(Object x) {
        return x.hashCode()
    }
    Object nullSafeGet(ResultSet rs, String[] names, Object owner) {
        return parseString(rs.getString(names[0]))
    }
    void nullSafeSet(PreparedStatement st, Object value, int index) {
        if (value == null)
            st.setNull(index, Types.VARCHAR);
        else {
            MyChineseName name = (MyChineseName) value
            st.setString(index, name.toOneString())
        }
    }
    Object deepCopy(Object value) {
        if (value == null)
            return null;
        MyChineseName name = (MyChineseName) value;
        return new MyChineseName(familyName: name.familyName, givenName: name.givenName);
    }
    boolean isMutable() {
        return false
    }
    Serializable disassemble(Object value) {
        return (Serializable) deepCopy(value);
    }
    Object assemble(Serializable cached, Object owner) {
        return (Serializable) deepCopy(cached);
    }
    Object replace(Object original, Object target, Object owner) {
        return null
    }
}

生成的数据库记录为：

 

除了实现CompositeUserType能将一个对象映射成多列，还有一种方法能达到这种效果：embedded。它能将本应映射成两个table的组合成一个表。

假设有两个实体关联如下：

@grails.persistence.Entity
class MyComp {
    String name
    String code
}
@grails.persistence.Entity
class MyEntity {
    String keyName
    MyComp comp

    static constraints = {
        comp(nullable: true)
    }
}

这样，它会在数据库中映射成两个表，用ID关联起来。

因为关联表很简单，能不能组合成一张表呢？可以，用embedded：

class MyComp {
    String name
    String code
}
@grails.persistence.Entity
class MyEntity {
    String keyName
    MyComp comp

    static embedded = ['comp']
    static constraints = {
        comp(nullable: true)
    }
}

生成的表为：