实现类似mysql的主键递增功能， 并在此基础上做了增强， 可对于不同分类， 不同主键前缀自动递增。

表结构设计

t_sequence

字段

描述

ID

数据库主键

SEQ_PREFIX

生成的主键前缀

SEQ_TYPE

生成的主键类型

SEQ_NUM

生成的主键的最新序号

SEQ_GEN_DATE

主键最新更新时间

SEQ_CATEGORY

主键分类

SEQ_CHAR_WIDTH

主键数字部分的最大宽度， 如果seq_num的位数小于这个宽度， 则最终生成的主键前面补0

上代码：SequenceGenerator.java

public final class SequenceGenerator {

private static final Log logger = LogFactory.getLog(SequenceGenerator.class);

private static final String DEFAULT_SEQ_CATEGORY = "common";

private static final Integer DEFAULT_SEQ_CHAR_WIDTH=4;

private static final String SEQ_TYPE_NUMERIC="INT";

private static final String SEQ_TYPE_CHAR="VARCHAR2";

private final ReentrantLock lock = new ReentrantLock();

private static class SequenceGeneratorHolder{

private static SequenceGenerator instance = new SequenceGenerator();

}

private SequenceGenerator(){}

public static SequenceGenerator getInstance(){

return SequenceGeneratorHolder.instance;

}

public Integer getId(SequenceModel seqCriteria){

final ReentrantLock lock = this.lock;

lock.lock();

try{

SequenceService sequenceService = ApplicationContextProvider.getBean(SequenceService.class);

return sequenceService.getSeqId(seqCriteria); //查询数据库在最新的Seq_num基础上加1

}finally {

lock.unlock();

}

}

/**

* 默认生成不带前缀的字符串id

*

* @return

*/

public String generateStringId(){

return generateStringId(DEFAULT_SEQ_CATEGORY,DEFAULT_SEQ_CHAR_WIDTH);

}

/**

* 根据类别生成字符串ID, 使用默认字符宽度为4， 并且带前缀和category 一致

* @param category

* @return

*/

public String generateStringId(String category){

return generateStringId(category,category,DEFAULT_SEQ_CHAR_WIDTH);

}

public String generateStringId(String prefix,String category){

return generateStringId(prefix,category,DEFAULT_SEQ_CHAR_WIDTH);

}

/**

* generate varchar type id without prefix

* @param category

* @return

*/

public String generateStringId(String category, Integer charWidth){

return generateStringId(null,category,charWidth);

}

/**

*

* @param prefix

* @param category

* @param charWidth

* @return

*/

public String generateStringId(String prefix,String category, Integer charWidth){

SequenceModel seqCriteria = new SequenceModel();

seqCriteria.setSeqCategory(category);

seqCriteria.setSeqPrefix(prefix);

seqCriteria.setSeqCharWidth(charWidth);

seqCriteria.setSeqType(SEQ_TYPE_CHAR);

Integer result = getId(seqCriteria);

String numStr = StrUtil.padPre(String.valueOf(result),charWidth,'0');

String id = StrUtil.nullToEmpty(prefix)+numStr;

return id;

}

public Integer generateNumericId(){

return generateNumericId(DEFAULT_SEQ_CATEGORY);

}

/**

* generate long type id

* @return

*/

public Integer generateNumericId(String category){

SequenceModel seqCriteria = new SequenceModel();

seqCriteria.setSeqCategory(category);

seqCriteria.setSeqType(SEQ_TYPE_NUMERIC);

Integer result = getId(seqCriteria);

return result;

}

}

SequenceModel.java  ---表结构对象封装

@Data

public class SequenceModel {

private Integer id;

private String seqPrefix;

private String seqType;

private Integer seqNum;

private String seqCategory;

private Integer seqCharWidth;

private Date seqGenDate;

}

SequenceService.java

这里才是整个生成策略的核心， 查询数据库的时候需要加悲观锁， 锁住整行数据， 查询，插入， 更新需要在同一个事务内完成， 否则容易发生死锁。

使用Mybatis操作数据库

@Service

public class SequenceService {

@Autowired

private SequenceMapper sequenceMapper;

@Transactional(transactionManager = "businessDataSourceTM") // 这行不能少

public Integer getSeqId(SequenceModel seqCriteria){

try {

SequenceModel seqResult = sequenceMapper.findSequence(seqCriteria);

AtomicInteger atomicInteger = new AtomicInteger();

if(seqResult == null){

// 新建

seqResult = seqCriteria;

seqResult.setSeqNum(atomicInteger.incrementAndGet());

seqResult.setSeqGenDate(DateUtil.nowDate());

sequenceMapper.insertSequenceNum(seqResult);

}else{

//更新

Integer seqNum = seqResult.getSeqNum();

atomicInteger.set(seqNum);

seqResult.setSeqNum(atomicInteger.incrementAndGet());

seqResult.setSeqGenDate(DateUtil.nowDate());

sequenceMapper.updateSequenceNum(seqResult);

}

return atomicInteger.get();

} catch (Exception e) {

log.error("generate Id occurs error.",e);

return 0;

}

}

}

sequence-generator.xml -- mybatis SQL配置文件

select

seq.ID,

seq.SEQ_PREFIX,

seq.SEQ_TYPE,

seq.SEQ_NUM,

seq.SEQ_CHAR_WIDTH,

seq.SEQ_GEN_DATE,

seq.SEQ_CATEGORY

from t_sequence seq

where 1=1

and seq.SEQ_CATEGORY = #{seqCriteria.seqCategory}

and seq.SEQ_PREFIX = #{seqCriteria.seqPrefix}

for update

update t_sequence set seq_num=#{seq.seqNum}, seq_gen_date=#{seq.seqGenDate,jdbcType=TIMESTAMP} where id=#{seq.id}

SELECT CZBANK_COMMON_SEQ.NEXTVAL as ID from DUAL

insert into t_sequence (

id,

SEQ_PREFIX,

SEQ_TYPE,

SEQ_NUM,

SEQ_CHAR_WIDTH,

SEQ_GEN_DATE,

SEQ_CATEGORY

) VALUES (

#{id},

#{seq.seqPrefix,jdbcType=VARCHAR},

#{seq.seqType,jdbcType=VARCHAR},

#{seq.seqNum},

#{seq.seqCharWidth,jdbcType=INTEGER},

#{seq.seqGenDate},

#{seq.seqCategory}

)

SequenceMapper.java

@Mapper

public interface SequenceMapper {

SequenceModel findSequence(@Param("seqCriteria") SequenceModel seqCriteria);

void updateSequenceNum(@Param("seq") SequenceModel seq);

void insertSequenceNum(@Param("seq") SequenceModel seq);

}