RxPaper is a RxJava wrapper for the cool paper library, a fast NoSQL data storage for Android that lets you save/restore Java objects by using efficient Kryo serialization and handling data structure changes automatically.
For the RxJava 2 version please go to RxPaper2 made by Pakoito.
- Update internal Kryo serializer to 4.0. The data format is changed, but Paper supports backward data compatibility automatically;
- Now 58% less methods count : 4037;
- Depends on data structure you may experience faster reading but slower writing.
repositories {
jcenter()
maven { url "https://jitpack.io" }
}
dependencies {
compile 'com.cesarferreira.rxpaper:rxpaper:0.5.0'
}
Init the library in your Application class
public class SampleApplication extends Application {
@Override
public void onCreate() {
super.onCreate();
RxPaper.init(this);
}
}
Save data object. Your custom classes must have no-arg constructor. Paper creates separate data file for each key.
RxPaper.book()
.write(key, value)
.subscribe(success -> /* all good */ );
I'm serious: Your custom classes must have no-arg constructor.
Read data objects. Paper instantiates exactly the classes which has been used in saved data. The limited backward and forward compatibility is supported. See Handle data class changes.
Use default values if object doesn't exist in the storage.
RxPaper.book()
.read(key, defaultPersonValue)
.subscribe(person -> /* all good */ );
Delete data for one key.
RxPaper.book()
.delete(key)
.subscribe();
Completely destroys Paper storage.
RxPaper.book()
.destroy()
.subscribe();
Check if a key is persisted
RxPaper.book()
.exists(key)
.subscribe(success -> /* all good */);
Returns all keys for objects in the book.
RxPaper.book()
.getAllKeys()
.subscribe(allKeys -> /* all good */);
You can create custom Book book separate storage using
RxPaper.book("custom-book")...;
Any changes in one book doesn't affect to others books.
Don't forget to specify which threads you want to use before subscribing to any data manipulation, or else it'll run on the UI thread.
...
.subscribeOn(Schedulers.io())
.observeOn(AndroidSchedulers.mainThread())
.subscribe(...
Class fields which has been removed will be ignored on restore and new fields will have their default values. For example, if you have following data class saved in Paper storage:
class Person {
public String firstName; // Cesar
public String middleName; // Costa
}
And then you realized you need to change the class like:
class Person {
public String firstName; // Cesar
// public String middleName; removed field, who cares about middle names
public String lastName; // New field
}
Then on restore the middleName field will be ignored and new lastName field will have its default value null.
Use transient keyword for fields which you want to exclude from saving process.
public transient String tempId = "default"; // Won't be saved
- Keep data classes:
-keep class my.package.data.model.** { *; }
alternatively you can implement Serializable in all your data classes and keep all of them using:
-keep class * implements java.io.Serializable { *; }
- Keep library classes and its dependencies
-keep class io.paperdb.** { *; }
-keep class com.esotericsoftware.** { *; }
-dontwarn com.esotericsoftware.**
-keep class de.javakaffee.kryoserializers.** { *; }
-dontwarn de.javakaffee.kryoserializers.**
Paper is based on the following assumptions:
- Saved data on mobile are relatively small;
- Random file access on flash storage is very fast.
So each data object is saved in separate file and write/read operations write/read whole file.
The Kryo is used for object graph serialization and to provide data compatibility support.
Running Benchmark on Nexus 4, in ms:
Benchmark | Paper | Hawk |
---|---|---|
Read/write 500 contacts | 187 | 447 |
Write 500 contacts | 108 | 221 |
Read 500 contacts | 79 | 155 |