“A man got to have a code!” - Omar Little
See the project website for documentation and APIs.
As our teams and programs grow, the variety and volume of data also grows. Success will turn your simple data models into complex ones! Whether your application is storing data to disk or transmitting it over a network, the structure and interpretation of that data should be clear. Consumers work best with data they understand!
Schemas describe and document data models. If you have data, you should have a schema.
Google's Protocol Buffers are built around a great schema language:
-
It's cross platform and language independent. Whatever programming language you use, you'll be able to use proto schemas with your application.
-
Proto schemas are backwards-compatible and future-proof. You can evolve your schema as your application loses old features and gains new ones.
-
It's focused. Proto schemas describe your data models. That's it.
Here's a sample message definition:
syntax = "proto2";
package squareup.dinosaurs;
option java_package = "com.squareup.dinosaurs";
import "squareup/geology/period.proto";
message Dinosaur {
// Common name of this dinosaur, like "Stegosaurus".
optional string name = 1;
// URLs with images of this dinosaur.
repeated string picture_urls = 2;
optional squareup.geology.Period period = 5;
}
And here's an enum definition:
syntax = "proto2";
package squareup.geology;
option java_package = "com.squareup.geology";
enum Period {
// 145.5 million years ago — 66.0 million years ago.
CRETACEOUS = 1;
// 201.3 million years ago — 145.0 million years ago.
JURASSIC = 2;
// 252.17 million years ago — 201.3 million years ago.
TRIASSIC = 3;
}
This schema language is Protocol Buffers' best feature. You might even use it purely for documentation purposes, such as to describe a JSON API.
Protocol Buffers also defines a compact binary encoding of messages that conform to the schema. This
encoding is fast to encode, fast to decode, small to transmit, and small to store. The binary
encoding uses numeric tags from the schema, like the 5
for period
above.
For example, let's encode this dinosaur:
{
name: "Stegosaurus",
period: JURASSIC
}
The encoded value is just 15 bytes:
Hex Description
0a tag: name(1), field encoding: LENGTH_DELIMITED(2). 1 << 3 | 2
0b "Stegosaurus".length()
53 'S'
74 't'
65 'e'
67 'g'
6f 'o'
73 's'
61 'a'
75 'u'
72 'r'
75 'u'
73 's'
28 tag: period(5), field encoding: VARINT(0). 5 << 3 | 0
02 JURASSIC(2)
The Protocol Buffers schema language and binary encoding are both defined by Google. Wire is an independent implementation from Square that's specifically designed for Android and Java.
For each message type defined in the schema, Wire generates an immutable model class and its builder. The generated code looks like code you'd write by hand: it's documented, formatted, and simple. Wire's APIs should feel at home to programmers who like Effective Java.
That said, there are some interesting design decisions in Wire:
-
Wire messages declare
public final
fields instead of the usual getter methods. This cuts down on both code generated and code executed. Less code is particularly beneficial for Android programs. -
Wire avoids case mapping. A field declared as
picture_urls
in a schema yields a Java fieldpicture_urls
and not the conventionalpictureUrls
camel case. Though the name feels awkward at first, it's fantastic whenever you usegrep
or more sophisticated search tools. No more mapping when navigating between schema, Java source code, and data. It also provides a gentle reminder to calling code that proto messages are a bit special. -
Primitive types are always boxed. If a field is absent, its value is
null
. This is used for naturally optional fields, such as a dinosaur whose period is unknown. A field may also be null due to schema evolution: if tomorrow we add acarnivore
boolean to our message definition, today's data won’t have a value for that field.
Here's the compact generated code for the Dinosaur
message defined above:
// Code generated by Wire protocol buffer compiler, do not edit.
// Source file: squareup/dinosaurs/dinosaur.proto at 9:1
package com.squareup.dinosaurs;
import com.squareup.geology.Period;
import com.squareup.wire.Message;
import com.squareup.wire.ProtoAdapter;
import com.squareup.wire.WireField;
import java.util.List;
import okio.ByteString;
public final class Dinosaur extends Message<Dinosaur, Dinosaur.Builder> {
public static final ProtoAdapter<Dinosaur> ADAPTER = ProtoAdapter.newMessageAdapter(Dinosaur.class);
private static final long serialVersionUID = 0L;
public static final String DEFAULT_NAME = "";
public static final Period DEFAULT_PERIOD = Period.CRETACEOUS;
/**
* Common name of this dinosaur, like "Stegosaurus".
*/
@WireField(
tag = 1,
adapter = "com.squareup.wire.ProtoAdapter#STRING"
)
public final String name;
/**
* URLs with images of this dinosaur.
*/
@WireField(
tag = 2,
adapter = "com.squareup.wire.ProtoAdapter#STRING",
label = WireField.Label.REPEATED
)
public final List<String> picture_urls;
@WireField(
tag = 5,
adapter = "com.squareup.geology.Period#ADAPTER"
)
public final Period period;
public Dinosaur(String name, List<String> picture_urls, Period period) {
this(name, picture_urls, period, ByteString.EMPTY);
}
public Dinosaur(String name, List<String> picture_urls, Period period, ByteString unknownFields) {
super(unknownFields);
this.name = name;
this.picture_urls = immutableCopyOf("picture_urls", picture_urls);
this.period = period;
}
@Override
public Builder newBuilder() {
Builder builder = new Builder();
builder.name = name;
builder.picture_urls = copyOf("picture_urls", picture_urls);
builder.period = period;
builder.addUnknownFields(unknownFields());
return builder;
}
@Override
public boolean equals(Object other) {
if (other == this) return true;
if (!(other instanceof Dinosaur)) return false;
Dinosaur o = (Dinosaur) other;
return equals(unknownFields(), o.unknownFields())
&& equals(name, o.name)
&& equals(picture_urls, o.picture_urls)
&& equals(period, o.period);
}
@Override
public int hashCode() {
int result = super.hashCode;
if (result == 0) {
result = unknownFields().hashCode();
result = result * 37 + (name != null ? name.hashCode() : 0);
result = result * 37 + (picture_urls != null ? picture_urls.hashCode() : 1);
result = result * 37 + (period != null ? period.hashCode() : 0);
super.hashCode = result;
}
return result;
}
public static final class Builder extends com.squareup.wire.Message.Builder<Dinosaur, Builder> {
public String name;
public List<String> picture_urls;
public Period period;
public Builder() {
picture_urls = newMutableList();
}
/**
* Common name of this dinosaur, like "Stegosaurus".
*/
public Builder name(String name) {
this.name = name;
return this;
}
/**
* URLs with images of this dinosaur.
*/
public Builder picture_urls(List<String> picture_urls) {
checkElementsNotNull(picture_urls);
this.picture_urls = picture_urls;
return this;
}
public Builder period(Period period) {
this.period = period;
return this;
}
@Override
public Dinosaur build() {
return new Dinosaur(name, picture_urls, period, buildUnknownFields());
}
}
}
The Java code to create and access proto models is compact and readable:
Dinosaur stegosaurus = new Dinosaur.Builder()
.name("Stegosaurus")
.period(Period.JURASSIC)
.build();
System.out.println("My favorite dinosaur existed in the " + stegosaurus.period + " period.");
Each type has a corresponding ProtoAdapter
that can encode a message to bytes and decode bytes
back into a message.
Dinosaur stegosaurus = ...
byte[] stegosaurusBytes = Dinosaur.ADAPTER.encode(stegosaurus);
byte[] tyrannosaurusBytes = ...
Dinosaur tyrannosaurus = Dinosaur.ADAPTER.decode(tyrannosaurusBytes);
When accessing a field, use Wire.get()
to replace null values with the corresponding default:
Period period = Wire.get(stegosaurus.period, Dinosaur.DEFAULT_PERIOD);
This is equivalent to the following:
Period period = stegosaurus.period != null ? stegosaurus.period : Dinosaur.DEFAULT_PERIOD;
Since version 3.0.0, Wire can generate Kotlin code. See Wire Compiler & Gradle Plugin to learn how to configure your build.
Kotlin is a pragmatic and expressive programming language that makes it easy to model data. Here's how we used Kotlin to model Protocol Buffers messages:
-
Messages feel like
data
classes, but in fact they're not. Compiler still generatesequals()
,hashCode()
,toString()
andcopy()
for you. Wire does not generatecomponentN()
functions though, we believe that destructuring declarations are not a good fit for Protocol Buffers: a change in the schema that removes or adds a field might lead to a situation when your destructuring declaration still compiles but now describes a completely different subset of fields, rendering your code incorrect. -
copy()
is a substitute for theBuilder
, which is not used anymore. If your program relies on theBuilder
to be present, you may generate code in Java interoperability mode - Wire Compiler & Gradle Plugin explains how that works. -
Fields are generated as properties. While this is idiomatic in Kotlin, Java code will now have to access fields using getters. If your program relies on accessing fields directly, use Java interoperability mode - the compiler will generate
@JvmField
annotations for each field. -
The nullability of each field's type depends on its label:
required
,repeated
andmap
fields get non-nullable types, whereasoptional
fields are of nullable types. -
With the exception of
required
fields, each field has a default value:- null for
optional
fields, emptyList()
forrepeated
fields,emptyMap()
formap
fields.
- null for
Here's the same Dinosaur
message in Kotlin:
// Code generated by Wire protocol buffer compiler, do not edit.
// Source file: squareup/dinosaurs/dinosaur.proto
package com.squareup.dinosaurs
import com.squareup.geology.Period
import com.squareup.wire.FieldEncoding
import com.squareup.wire.Message
import com.squareup.wire.ProtoAdapter
import com.squareup.wire.ProtoReader
import com.squareup.wire.ProtoWriter
import com.squareup.wire.WireField
import kotlin.Any
import kotlin.AssertionError
import kotlin.Boolean
import kotlin.Deprecated
import kotlin.DeprecationLevel
import kotlin.Int
import kotlin.Nothing
import kotlin.String
import kotlin.collections.List
import kotlin.hashCode
import kotlin.jvm.JvmField
import okio.ByteString
class Dinosaur(
/**
* Common name of this dinosaur, like "Stegosaurus".
*/
@field:WireField(
tag = 1,
adapter = "com.squareup.wire.ProtoAdapter#STRING"
)
val name: String? = null,
/**
* URLs with images of this dinosaur.
*/
@field:WireField(
tag = 2,
adapter = "com.squareup.wire.ProtoAdapter#STRING",
label = WireField.Label.REPEATED
)
val picture_urls: List<String> = emptyList(),
@field:WireField(
tag = 5,
adapter = "com.squareup.geology.Period#ADAPTER"
)
val period: Period? = null,
unknownFields: ByteString = ByteString.EMPTY
) : Message<Dinosaur, Nothing>(ADAPTER, unknownFields) {
@Deprecated(
message = "Shouldn't be used in Kotlin",
level = DeprecationLevel.HIDDEN
)
override fun newBuilder(): Nothing {
throw AssertionError()
}
override fun equals(other: Any?): Boolean {
if (other === this) return true
if (other !is Dinosaur) return false
return unknownFields == other.unknownFields
&& name == other.name
&& picture_urls == other.picture_urls
&& period == other.period
}
override fun hashCode(): Int {
var result = super.hashCode
if (result == 0) {
result = name.hashCode()
result = result * 37 + picture_urls.hashCode()
result = result * 37 + period.hashCode()
super.hashCode = result
}
return result
}
override fun toString(): String {
val result = mutableListOf<String>()
if (name != null) result += """name=$name"""
if (picture_urls.isNotEmpty()) result += """picture_urls=$picture_urls"""
if (period != null) result += """period=$period"""
return result.joinToString(prefix = "Dinosaur{", separator = ", ", postfix = "}")
}
fun copy(
name: String? = this.name,
picture_urls: List<String> = this.picture_urls,
period: Period? = this.period,
unknownFields: ByteString = this.unknownFields
): Dinosaur = Dinosaur(name, picture_urls, period, unknownFields)
companion object {
@JvmField
val ADAPTER: ProtoAdapter<Dinosaur> = object : ProtoAdapter<Dinosaur>(
FieldEncoding.LENGTH_DELIMITED,
Dinosaur::class
) {
override fun encodedSize(value: Dinosaur): Int =
ProtoAdapter.STRING.encodedSizeWithTag(1, value.name) +
ProtoAdapter.STRING.asRepeated().encodedSizeWithTag(2, value.picture_urls) +
Period.ADAPTER.encodedSizeWithTag(5, value.period) +
value.unknownFields.size
override fun encode(writer: ProtoWriter, value: Dinosaur) {
ProtoAdapter.STRING.encodeWithTag(writer, 1, value.name)
ProtoAdapter.STRING.asRepeated().encodeWithTag(writer, 2, value.picture_urls)
Period.ADAPTER.encodeWithTag(writer, 5, value.period)
writer.writeBytes(value.unknownFields)
}
override fun decode(reader: ProtoReader): Dinosaur {
var name: String? = null
val picture_urls = mutableListOf<String>()
var period: Period? = null
val unknownFields = reader.forEachTag { tag ->
when (tag) {
1 -> name = ProtoAdapter.STRING.decode(reader)
2 -> picture_urls.add(ProtoAdapter.STRING.decode(reader))
5 -> period = Period.ADAPTER.decode(reader)
else -> reader.readUnknownField(tag)
}
}
return Dinosaur(
name = name,
picture_urls = picture_urls,
period = period,
unknownFields = unknownFields
)
}
override fun redact(value: Dinosaur): Dinosaur = value.copy(
unknownFields = ByteString.EMPTY
)
}
}
}
Creating and accessing proto models is easy:
val stegosaurus = Dinosaur(
name = "Stegosaurus",
period = Period.JURASSIC
)
println("My favorite dinosaur existed in the ${stegosaurus.period} period.")
Here's how you can modify the object to add extra fields:
val stegosaurus = stegosaurus.copy(
picture_urls = listOf("https://www.flickr.com/photos/tags/Stegosaurus/")
)
println("Here are some photos of ${stegosaurus.name}: ${stegosaurus.picture_urls}")
Wire's compiler is available via a Maven plugin. Put .proto
sources in your project's
src/main/proto
directory, then use the plugin to generate .java
files. The plugin will
automatically add the generated Java code to your project's source roots.
<build>
<plugins>
<plugin>
<groupId>com.squareup.wire</groupId>
<artifactId>wire-maven-plugin</artifactId>
<version>${project.version}</version>
<executions>
<execution>
<phase>generate-sources</phase>
<goals>
<goal>generate-sources</goal>
</goals>
<configuration>
<includes>
<!-- proto package names to generate code for -->
<include>squareup.dinosaurs.*</include>
<include>squareup.geology.*</include>
</includes>
</configuration>
</execution>
</executions>
</plugin>
</plugins>
</build>
Wire can read .proto
files from the local file system and from within .jar
files.
The compiler can optionally prune your schema to a subset of root types and their transitive dependencies. This is useful when sharing a schema between projects: a Java service and Android app may each use a subset of a larger shared schema.
If you don't use Maven, the compiler also has a command line interface. Just substitute
wire-compiler-VERSION-jar-with-dependencies.jar
with the path to your jar. Download the latest precompiled jar.
% java -jar wire-compiler-VERSION-jar-with-dependencies.jar \
--proto_path=src/main/proto \
--java_out=out \
squareup/dinosaurs/dinosaur.proto \
squareup/geology/period.proto
Writing com.squareup.dinosaurs.Dinosaur to out
Writing com.squareup.geology.Period to out
Supplying the --android
flag to the compiler causes Wire messages to implement Parcelable
.
If you use Proguard, then you need to add keep
rules. The simplest option is to tell Proguard not
to touch the Wire runtime library and your generated protocol buffers (of course these simple rules
will miss opportunities to shrink and optimize the code):
-keep class com.squareup.wire.** { *; }
-keep class com.yourcompany.yourgeneratedcode.** { *; }
The wire-runtime
package contains runtime support libraries that must be included in applications
that use Wire-generated code.
With Maven:
<dependency>
<groupId>com.squareup.wire</groupId>
<artifactId>wire-runtime</artifactId>
<version>2.2.0</version>
</dependency>
With Gradle:
compile 'com.squareup.wire:wire-runtime:2.2.0'
With Maven:
<dependency>
<groupId>com.squareup.wire</groupId>
<artifactId>wire-runtime</artifactId>
<version>3.0.0-rc01</version>
</dependency>
With Gradle:
api 'com.squareup.wire:wire-runtime:3.0.0-rc01'
Snapshots of the development version are available in Sonatype's snapshots
repository.
Wire does not support:
- Groups - they are skipped when parsing binary input data
Wire supports custom options on messages and fields. Other custom options are ignored. Pass
--excludes=google.protobuf.*
to the compiler to omit options from the generated code.
See Google's excellent documentation on the structure and syntax of proto schemas.