我有一个实体Activity
,可以是parent
或child
。
class Activity {
long modificationDate;
Activity parentActivity;
Set<Activity> subActivities;
boolean active;
}
我必须按照modification date
对数据库中先前加载的活动进行排序,但仍然是grouped by
他们的父活动。
规则:子E1
活动可以更新,而其父P1
不是。{因此,E1
的修改日期可能大于P1
如果孩子被更新,its parent has to rise
列表顶部及其所有孩子
示例:活动|修改日期
E1 | 01 june
P2 | 01 june
P1 | 01 june
P3 | 02 june
E1 | 10 july
排序后,应该给出
P1 | 01 june
E1 | 10 july
E1 | 01 june
P3 | 02 june
P2 | 01 june
Java基础
Set<Activity> actives = getRepository().findActiveActivities().stream()
.sorted(Comparator.comparing(Activity::getModificationDate).reversed())
.collect(Collectors.toCollection(LinkedHashSet::new));
actives = actives.stream().sorted((Activity a, Activity b) -> {
//sort logic
}).collect(Collectors.toCollection(LinkedHashSet::new));
我真的不知道是否可以通过sql请求执行,希望有人能为我提供解决方案。
修改
这不是一个学校项目。我正在提供提供数据库模型的休息服务。 人们认为我是个骗子:link
此时,http://localhost/activity
上的get请求提供了未设置的set活动,如json。
我的网站调用此Web服务也有未分类的活动,活动表完全不可读。
答案 0 :(得分:8)
所以我的想法如下。由于parent
活动包含一组所有子项,因此我们可以使用它来获取按最近更新的子项排序的父项列表。
List<Activity> collect = list.stream().filter(activity -> activity.getParent() == null)
.sorted((o1, o2) -> Long.compare(o2.children.stream().max(Comparator.comparing(Activity::getUpdateDate)).orElse(o2)
.getUpdateDate(), o1.children.stream().max(Comparator.comparing(Activity::getUpdateDate)).orElse(o1).getUpdateDate()))
.collect(Collectors.toCollection(LinkedList::new));
现在创建一个List
,其中将存储结果。
List<Activity> result = collect.stream().flatMap(set ->
Stream.concat(Stream.of(set), set.getChildren().stream()
.sorted((o1, o2) -> Long.compare(o2.getUpdateDate(), o1.getUpdateDate())))
).collect(Collectors.toCollection(LinkedList::new);
您将获得所需的List
个活动,首先是父级,然后是按updateDate等排序的子级。
更新
如果树有多个级别,则必须递归获取所有子级,然后对它们进行排序。
获取Activity
public static Stream<Activity> getChildren(Activity activity) {
List<Activity> activities = new LinkedList<>();
recursiveCall(activities, activity);
return activities.stream();
}
public static void recursiveCall(List<Activity> activities, Activity activity) {
if(activity.children.isEmpty()) {
return;
}
activity.getChildren().forEach(child -> {
activities.add(child);
recursiveCall(activities, child);
});
}
并像这样使用它:
list.stream().filter(activity -> activity.getParent() == null)
.flatMap(YourClass::getChildren)
.sorted...
在收集流中的所有结果时执行相同的操作。
答案 1 :(得分:5)
我将开始使用SQL方法,稍后将继续使用Java。 (SQL是我的爱)。
我真的不知道是否可以通过sql请求执行 ,希望有人能为我提供解决方案。
这可以使用SQL方法解决。这是第一个非优化的例子:
<强>模式强>
create table Activity(
name varchar(2),
modificationDate int,
parentActivity varchar(2)
);
insert into activity values('E1', 1, 'P1' );
insert into activity values('P2', 1, null );
insert into activity values('P1', 1, null );
insert into activity values('P3', 2, null );
insert into activity values('E1', 10, 'P1' );
查询(版本1玩具示例)
SELECT *
FROM
( SELECT a1.*,
(SELECT max(modificationDate)
FROM Activity a2
GROUP BY a2.parentActivity
HAVING a2.parentActivity = a1.name) AS childMod,
1 AS isParent
FROM Activity a1
WHERE a1.parentActivity IS NULL
UNION SELECT a3.*,
(SELECT max(modificationDate)
FROM Activity a4
GROUP BY a4.name
HAVING a3.name = a4.name), 0 AS isParent
FROM Activity a3
WHERE a3.parentActivity IS NOT NULL ) AS TEMP
ORDER BY childMod DESC,
isParent DESC,
temp.modificationDate DESC;
通过删除一些低效的子查询,而不是使用连接的子查询(别名临时表)和/或一元自连接,可以改进上述内容。
结果:http://i.prntscr.com/hnTeXtQzR5_BRui3TXJFSw.png
查询(最终版本2玩具示例)
select
*
FROM
(
select name, modificationDate, childMaxMod, 1 as isParent from
Activity a1 left outer join
(select ParentActivity, max(modificationDate) as childMaxMod from Activity group by ParentActivity) as a2
on a1.name = a2.parentActivity
WHERE a1.parentActivity IS NULL
UNION
select name, modificationDate, childMaxMod, 0 as isParent from
Activity a1 inner join
(select ParentActivity, max(modificationDate) as childMaxMod from Activity group by ParentActivity) as a2
on a1.parentActivity = a2.parentActivity
) as Temp
order by childMaxMod desc, isParent desc, modificationDate desc;
Luca的解决方案相当不错,纯粹是一个java 8功能解决方案。 我发布了自己的替代方案,但它使用了命令式的java编程。
活动类
import lombok.AllArgsConstructor;
import lombok.Builder;
import lombok.Data;
import java.util.HashSet;
import java.util.Objects;
import java.util.Set;
@Builder
@Data
@AllArgsConstructor
class Activity {
long modificationDate;
String name;
Activity parentActivity;
Set<Activity> subActivities = new HashSet();
boolean active;
public String debugString;
public int hashCode(){
return Objects.hash(modificationDate, name, subActivities, active);
}
}
代码 - 设置部分
public static void main(String[] args) {
//SpringApplication.run(MneServiceApplication.class, args);
ArrayList<Activity> set = new ArrayList<Activity>();
Activity p1 = Activity.builder().modificationDate(0).name("P1").subActivities(new HashSet<>()).build();
Activity e1 = Activity.builder().modificationDate(10).name("E1").subActivities(new HashSet<>()).build();
Activity e1_2 = Activity.builder().modificationDate(01).name("E1").subActivities(new HashSet<>()).build();
Activity p3 = Activity.builder().modificationDate(02).name("P3").subActivities(new HashSet<>()).build();
Activity p2 = Activity.builder().modificationDate(01).name("P2").subActivities(new HashSet<>()).build();
p1.getSubActivities().add(e1);
p1.getSubActivities().add(e1_2);
e1.setParentActivity(p1);
e1_2.setParentActivity(p1);
set.add(p1);
set.add(e1);
set.add(e1_2);
set.add(p3);
set.add(p2);
代码 - 逻辑(续)
Set<Activity> actives = set.stream()
.sorted(Comparator.comparing(Activity::getModificationDate).reversed())
.collect(Collectors.toCollection(LinkedHashSet::new));
HashMap<Activity, Optional> maxChildModifications = new HashMap<Activity, Optional>();
actives = actives.stream().sorted((Activity a, Activity b) -> {
DecimalFormat decimalFormat = new DecimalFormat("0000000000000000");
String comparisonValue_A = getComparisonString2(maxChildModifications, a, decimalFormat);
String comparisonValue_B = getComparisonString2(maxChildModifications, b, decimalFormat);
a.debugString = comparisonValue_A;
b.debugString = comparisonValue_B;
return comparisonValue_B.compareTo(comparisonValue_A);
}).collect(Collectors.toCollection(LinkedHashSet::new));
for(Activity activity:actives){
System.out.println(activity.getName() + " " + activity.getDebugString());
}
}
private static String getComparisonString2(HashMap<Activity, Optional> maxChildModificationsLookup, Activity activity, DecimalFormat decimalFormat) {
//Is a Parent or a Child
Activity lookupActivity = activity.getParentActivity();;
if(lookupActivity == null){ //It is a parent
lookupActivity = activity;
}
String comparisonValue_B = "";
Long isParent_B = 0L;
Long maxChildMod_B = 0L;
Optional<Activity> maxChildActivityOptional_B = getMaxChildModActivity(lookupActivity, maxChildModificationsLookup);
if(maxChildActivityOptional_B.isPresent()) {
maxChildMod_B = maxChildActivityOptional_B.get().getModificationDate();
}
if(activity.getSubActivities().size() > 0){
isParent_B = 1L; //Or use some other comparison system than the concatenated string
}else{
isParent_B = 0L;
}
comparisonValue_B = String.valueOf( decimalFormat.format(maxChildMod_B)).concat("-").concat(String.valueOf(isParent_B));
comparisonValue_B = comparisonValue_B .concat("-").concat(decimalFormat.format(activity.getModificationDate()));
return comparisonValue_B;
}
private static Optional<Activity> getMaxChildModActivity(Activity a, HashMap<Activity, Optional> maxChildModificationsLookup) {
Optional<Activity> result = maxChildModificationsLookup.get(a);
if(result!=null){
return result;
}
result = a.getSubActivities().stream().max((Activity a2, Activity b2) ->
{
return new Long(a2.getModificationDate()).compareTo(b2.getModificationDate());
}
);
maxChildModificationsLookup.put(a, result);
return result;
}
<强>结果:强>
P1 0000000000000010-1-0000000000000000
E1 0000000000000010-0-0000000000000010
E1 0000000000000010-0-0000000000000001
P3 0000000000000000-0-0000000000000002
P2 0000000000000000-0-0000000000000001
显然,我的SQL解决方案和Luca的Java 8解决方案比上面的java实现更简洁和优雅(尽管可读性可能需要进行一些格式更改)。