近来接手了一套代码，里面有很多感觉可以改进的地方，修改后做个记录。
话不多说，直接上代码。代码的大意是对一系列对象数据按照某字段来统计另一个字段的和。原来的代码，调用了五次数据接口：
int tsjt = measureRepository.findScoreByTypeAndUserIdAndDate("tenderness", userId, date);
int tnsj = measureRepository.findScoreByTypeAndUserIdAndDate("swelling", userId, date);
int stij = measureRepository.findScoreByTypeAndUserIdAndDate("temperature", userId, date);
int leftHand = measureRepository.findScoreByCodeAndUserIdAndDate("left_grip", userId, date);
int rightHand = measureRepository.findScoreByCodeAndUserIdAndDate("right_grip", userId, date);
对应两个数据接口：
@Query( value = "select ifnull(sum(score),0) as score from pride_measured where type=:type and user_id=:userId and date=:date ", nativeQuery = true) int findScoreByTypeAndUserIdAndDate(String type, Integer userId, Date date); @Query( value = "select ifnull(sum(score),0) as score from pride_measured where code=:code and user_id=:userId and date=:date", nativeQuery = true) int findScoreByCodeAndUserIdAndDate(String code, Integer userId, Date date);Date date);
改造为只调用一次数据查询接口，然后通过java8 Stream特性来处理：
Measure measureQuery = new Measure(); measureQuery.setUserId(userId); measureQuery.setDate(date); List<Measure> list = measureService.queryMeasures(measureQuery); Map<String, IntSummaryStatistics> measuresByType = list.stream() .collect(Collectors.groupingBy(Measure::getType, Collectors.summarizingInt(Measure::getScore))); int tsjt = measuresByType.get("tenderness") == null ? 0 : (int)measuresByType.get("tenderness").getSum(); int tnsj = measuresByType.get("swelling") == null ? 0 : (int)measuresByType.get("swelling").getSum(); int stij = measuresByType.get("temperature") == null ? 0 : (int)measuresByType.get("temperature").getSum(); Map<String, Integer> measuresByCode = list.stream().filter(item -> ("grip".equals(item.getType()))) .collect(Collectors.toMap(Measure::getCode, Measure::getScore)); int leftHand = measuresByCode.get("left_grip") == null ? 0 : measuresByCode.get("left_grip"); int rightHand = measuresByCode.get("right_grip") == null ? 0 : measuresByCode.get("right_grip");
分别三次测试该方法的原来代码执行时间和现代码执行时间（其中包含无符合条件数据和有符合条件数据两种情况），原来执行时间分别为：1059ms、1281ms、1371ms，改造后执行时间分别为：928ms、922ms、1067ms，可以看到执行时间在同样情况下改造后有明显的改观。
本次改造，除了减少了访问数据库次数节省时间外，还把处理集中到了服务程序中，便于分布式扩展。