RxJava- 操作符之转换Observable
Posted 行云间
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了RxJava- 操作符之转换Observable相关的知识,希望对你有一定的参考价值。
在 RxJava- 操作符之过滤Observables了解到,RxJava过滤操作符的基本使用,主要是针对Observable列表。本篇主要针对Observable列表中对象操作符的使用。用于变换可观测序列来创建一个能够更好的满足我们需求的序列。
map
map操作符通过指定一个的Func对象,将Observables转换为一个新的Observable对象并发射,观察者将收到新的Observable处理。map操作符的流程图如下:
示例代码:
Observable.from(mLists)
.map(new Func1<Student, Student>() {
@Override
public Student call(Student student) {
StringBuilder sb = new StringBuilder();
String age = student.getAge();
sb.append("AGE - ");
sb.append(age);
student.setAge(sb.toString());
return student;
}
})
.subscribe(new Observer<Student>() {
@Override
public void onCompleted() {
}
@Override
public void onError(Throwable e) {
}
@Override
public void onNext(Student student) {
mAdaStu.addData(student);
}
});
正如你看到的,像往常一样创建我们发射的Observable之后,追加一个map调用,在其回调call方法中,将age字段添加“AGE - ”, 更新对象后发射,观察者接收到的是更新后的Observable。
flatMap
flatMap操作符用于发射的一个数据序列,而这些数据同时本身拥有发射Observable。flatMap是以铺平序列的方式,然后合并这些Observables发射的数据,最后将合并后的结果作为最终的Observable。但是,flatMap()可能交错的发送事件,最终结果的顺序可能并是不原始Observable发送时的顺序。flatMap操作符的流程图如下:
示例代码:
private void doFlatMap() {
List<String> lists = new ArrayList<>();
lists.add("Flat - A");
lists.add("Flat - B");
lists.add("Flat - C");
lists.add("Flat - D");
Observable.from(lists)
.flatMap(new Func1<String, Observable<Student>>() {
@Override
public Observable<Student> call(String s) {
return getStudentData(s);
}
})
.subscribe(new Observer<Student>() {
@Override
public void onCompleted() {
}
@Override
public void onError(Throwable e) {
}
@Override
public void onNext(Student student) {
mAdaStu.addData(student);
}
});
}
// 针对不同的s发射不同的Observable序列
private Observable<Student> getStudentData(final String s) {
List<Student> lists_0 = new ArrayList<>();
if (TextUtils.equals("Flat - A", s) || TextUtils.equals("Concat - A", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
lists_0.add(new Student(s + "12", "20", "1102"));
lists_0.add(new Student(s + "13", "20", "1103"));
} else if (TextUtils.equals("Flat - B", s) || TextUtils.equals("Concat - B", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
} else if (TextUtils.equals("Flat - C", s) || TextUtils.equals("Concat - C", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
lists_0.add(new Student(s + "12", "20", "1102"));
} else if (TextUtils.equals("Flat - D", s) || TextUtils.equals("Concat - D", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
lists_0.add(new Student(s + "12", "20", "1102"));
lists_0.add(new Student(s + "13", "20", "1103"));
lists_0.add(new Student(s + "13", "20", "1104"));
}
return Observable.from(lists_0);
}
注:
1.flatMap允许交叉。如流程图所示,这意味着flatMap()不能够保证在最终生成的Observable中源Observables确切的发射顺序。
2.当我们在处理可能有大量的Observables时,重要是记住任何一个Observables发生错误的情况,flatMap()将会触发它自己的onError()函数并放弃整个链。
concatMap
concatMap操作符功能与flatMap操作符一致,不过,它解决了flatMap()的交叉问题,提供了一种能够把发射的值连续在一起的铺平函数,而不是合并它们,
concatMap操作符的流程图如下:
示例代码:
List<String> lists = new ArrayList<>();
lists.add("Concat - A");
lists.add("Concat - B");
lists.add("Concat - C");
lists.add("Concat - D");
Observable.from(lists)
.concatMap(new Func1<String, Observable<Student>>() {
@Override
public Observable<Student> call(String s) {
return getStudentData(s);
}
})
.subscribe(new Observer<Student>() {
@Override
public void onCompleted() {
}
@Override
public void onError(Throwable e) {
}
@Override
public void onNext(Student student) {
mAdaStu.addData(student);
}
});
flatMapIterable
flatMapIterable操作符和flatMap类似,仅有的本质不同是它将源数据两两结成对并生成Iterable,而不是原始数据项和生成的Observables。flatMapIterable操作符的流程图如下:
示例代码:
private void doFlatMapIterable() {
List<String> lists = new ArrayList<>();
lists.add("FlatMapIterable - A");
lists.add("FlatMapIterable - B");
lists.add("FlatMapIterable - C");
lists.add("FlatMapIterable - D");
Observable.from(lists)
.flatMapIterable(new Func1<String, Iterable<Student>>() {
@Override
public Iterable<Student> call(String s) {
return getStudentList(s);
}
})
.subscribe(new Observer<Student>() {
@Override
public void onCompleted() {
}
@Override
public void onError(Throwable e) {
}
@Override
public void onNext(Student student) {
mAdaStu.addData(student);
}
});
}
// 针对不同的s发射不同的Observable序列
private ArrayList<Student> getStudentList(final String s) {
ArrayList<Student> lists_0 = new ArrayList<>();
if (TextUtils.equals("FlatMapIterable - A", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
lists_0.add(new Student(s + "12", "20", "1102"));
lists_0.add(new Student(s + "13", "20", "1103"));
} else if (TextUtils.equals("FlatMapIterable - B", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
} else if (TextUtils.equals("FlatMapIterable - C", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
lists_0.add(new Student(s + "11", "20", "1102"));
} else if (TextUtils.equals("FlatMapIterable - D", s)) {
lists_0.add(new Student(s + "11", "20", "1101"));
lists_0.add(new Student(s + "12", "20", "1102"));
lists_0.add(new Student(s + "13", "20", "1103"));
lists_0.add(new Student(s + "13", "20", "1104"));
}
return lists_0;
}
switchMap
switchMap操作符flatMap类似,有一点不同,每当源Observable发射一个新的数据项(Observable)时,它将取消订阅并停止监视之前那个数据项产生的Observable,并开始监视当前发射的这一个。switchMap操作符的流程图如下:
scan
scan操作符可以认为是一个sum函数,scan()函数对原始Observable发射的每一项数据都应用一个函数,计算出函数的结果值,并将该值填充回可观测序列,等待和下一次发射的数据一起使用。scan操作符的流程图如下:
示例代码:
Observable.just(1,2,3,4,5)
.scan((sum,item) -> sum + item)
.subscribe(new Subscriber<Integer>() {
@Override
public void onCompleted() {
Log.d("RXJAVA", "Sequence completed.");
}
@Override
public void onError(Throwable e) {
Log.e("RXJAVA", "Something went south!");
}
@Override
public void onNext(Integer item) {
Log.d("RXJAVA", "item is: " + item);
}
});
goupBy
goupBy操作符用于分组元素,将源Observable变换成一个发射Observables的新的Observable(分组后的)。它们中的每一个新的Observable都发射一组指定的数据。goupBy操作符的流程图如下:
Observable<GroupedObservable<String,Student>> groupedItems = Observable.from(mLists)
.groupBy(new Func1<Student,String>(){
@Override
public String call(Student student){
return student.getAge();
}
});
Observable.concat(groupedItems)
.subscribe(new Observer<Student>() {
@Override
public void onCompleted() {
}
@Override
public void onError(Throwable e) {
}
@Override
public void onNext(Student student) {
mAdaStu.addData(student);
}
});
buffer
buffer操作符将源Observable变换一个新的Observable,这个新的Observable每次发射一组列表值而不是一个一个发射。buffer操作符的流程图如下:
示例代码:
仅仅指定cout
Observable.from(mLists)
.buffer(2)
.subscribe(new Action1<List<Student>>() {
@Override
public void call(List<Student> list) {
cout++;
Log.i("123", "cout = " + cout);
Log.i("123", list.toString());
mAdaStu.addDatdLists(list);
}
});
log打印
cout = 1
[Student(name=A11, age=26, no=1101), Student(name=A12, age=21, no=1102)]
cout = 2
[Student(name=A13, age=20, no=1103), Student(name=B11, age=22, no=1201)]
cout = 3
[Student(name=B12, age=30, no=1202), Student(name=B13, age=20, no=1203)]
cout = 4
[Student(name=S10, age=28, no=1301), Student(name=E10, age=30, no=1401)]
cout = 5
[Student(name=F10, age=26, no=1501)]
查看上方的Log,可以看出将源Observable量量分组,变换为一个新的Observable,新的Observable每次发射一组列表值。
设定count及skip
示例代码:
Observable.from(mLists)
.buffer(2,3)
.subscribe(new Action1<List<Student>>() {
@Override
public void call(List<Student> list) {
cout++;
Log.i("123", "cout = " + cout);
Log.i("123", list.toString());
mAdaStu.addDatdLists(list);
}
});
//Log打印
cout = 1
[Student(name=A11, age=26, no=1101), Student(name=A12, age=21, no=1102)]
cout = 2
[Student(name=B11, age=22, no=1201), Student(name=B12, age=30, no=1202)]
cout = 3
[Student(name=S10, age=28, no=1301), Student(name=E10, age=30, no=1401)]
设定count及timespan
window
window操作符与Buffer操作符类似,但是它发射的是Observable而不是列表。看下方流程图,winow是缓存3个数据项并把它们作为一个新的Observable发射出去。window操作符的流程图如下:
示例代码:
仅仅设定count
Observable.from(mLists)
.window(2)
.subscribe(new Action1<Observable<Student>>() {
@Override
public void call(Observable<Student> studentObservable) {
studentObservable.subscribe(new Action1<Student>() {
@Override
public void call(Student student) {
mAdaStu.addData(student);
}
});
}
});
//Log打印
I: cout = 1
I: Student(name=A11, age=26, no=1101)
I: cout = 2
I: Student(name=A12, age=21, no=1102)
I: cout = 3
I: Student(name=A13, age=20, no=1103)
I: cout = 4
I: Student(name=B11, age=22, no=1201)
I: cout = 5
I: Student(name=B12, age=30, no=1202)
I: cout = 6
I: Student(name=B13, age=20, no=1203)
I: cout = 7
I: Student(name=S10, age=28, no=1301)
I: cout = 8
I: Student(name=E10, age=30, no=1401)
I: cout = 9
I: Student(name=F10, age=26, no=1501)
细看上方Log,window将源Observable按count缓存后,然后一一发射出去。
设定count及Skip
Observable.from(mLists)
.window(2,3)
.subscribe(new Action1<Observable<Student>>() {
@Override
public void call(Observable<Student> studentObservable) {
studentObservable.subscribe(new Action1<Student>() {
@Override
public void call(Student student) {
mAdaStu.addData(student);
}
});
}
});
I: cout = 1
I: Student(name=A11, age=26, no=1101)
I: cout = 2
I: Student(name=A12, age=21, no=1102)
I: cout = 3
I: Student(name=B11, age=22, no=1201)
I: cout = 4
I: Student(name=B12, age=30, no=1202)
I: cout = 5
I: Student(name=S10, age=28, no=1301)
I: cout = 6
I: Student(name=E10, age=30, no=1401)
cast
cast操作符是是map()操作符的特殊版本,它将源Observable中的每一项数据都转换为新的类型,把它变成了不同的Class。cast操作符是是map操作符的流程图如下:
以上是关于RxJava- 操作符之转换Observable的主要内容,如果未能解决你的问题,请参考以下文章