protobuf实例-PHP版

最新推荐文章于 2024-05-29 09:55:32 发布

weixin_33961829

最新推荐文章于 2024-05-29 09:55:32 发布

阅读量584

点赞数

文章标签： php python json

原文链接：https://my.oschina.net/mickelfeng/blog/116410

版权

2019独角兽企业重金招聘Python工程师标准>>>

protobuf实例-PHP版

protobuf简介
protobuf是google提供的一个开源序列化框架，类似于XML，JSON这样的数据表示语言，其最大的特点是基于二进制，因此比传统的XML表示高效短小得多。虽然是二进制数据格式，但并没有因此变得复杂，开发人员通过按照一定的语法定义结构化的消息格式，然后送给命令行工具，工具将自动生成相关的类，可以支持php、java、c++、python等语言环境。通过将这些类包含在项目中，可以很轻松的调用相关方法来完成业务消息的序列化与反序列化工作。

protobuf在google中是一个比较核心的基础库，作为分布式运算涉及到大量的不同业务消息的传递，如何高效简洁的表示、操作这些业务消息在google这样的大规模应用中是至关重要的。而protobuf这样的库正好是在效率、数据大小、易用性之间取得了很好的平衡。

更多信息可参考官方文档
http://code.google.com/p/protobuf/

http://code.google.com/p/pb4php/(这是PHP类库)以下是具体操作例子

[php-protobuf](https://github.com/allegro/php-protobuf) PHP Protobuf - Google's Protocol Buffers for PHP

<?php
require_once('parser/pb_parser.php'); $test = new PBParser(); $test->parse('./test.proto');
 
require_once('message/pb_message.php');
require_once('pb_proto_test.php');
$string = file_get_contents('./example/test.pb');
 
$book = new AddressBook();
$person = $book->add_person();
$person->set_name('Kordulla');
$person->set_surname('Nikolai');
 
$phone_number = $person->add_phone();
$phone_number->set_number('49');
$phone_number->set_type(Person_PhoneType::WORK);
 
$phone_number = $person->add_phone();
$phone_number->set_number('171');
$phone_number->set_type(Person_PhoneType::MOBILE);
 
// serialize
$string = $book->SerializeToString();
//echo $string;
// write it to disk
file_put_contents('adressbook.pb', $string);
$string = file_get_contents('./adressbook.pb');
// Just read it
$book = new AddressBook();
$book->parseFromString($string);
 
var_dump($book->person_size());
$person = $book->person(0);
var_dump($person->name());
var_dump($person->surname());
var_dump($person->phone(0)->number());
var_dump($person->phone(0)->type());
var_dump($person->phone(1)->number());
var_dump($person->phone(1)->type());

输出

int(1) string(8) "Kordulla" string(7) "Nikolai" string(2) "49" int(2) string(3) "171" int(0)
“adressbook.pb” 是生成的二进制文件基本结构一个字节类型+ 字节长度
从以上操作和类库源代码来看打包速度可能慢很多。空间节省倒是非常好。符合 protobuf 定义：效率、数据大小、易用性之间的平衡。

在网上也可以搜索到xml、json、protobuf 对比的性能测试。

http://blog.hucde.com/2011/07/20/180

PHP Protobuf - Google's Protocol Buffers for PHP

Overview

Protocol Buffers are a way of encoding structured data in an efficient yet extensible format. It might be used in file formats and RPC protocols.

PHP Protobuf is Google's Protocol Buffers implementation for PHP with a goal to provide high performance, including a protoc plugin to generate PHP classes from .proto files. The heavy-lifting (a parsing and a serialization) is done by a PHP extension.

Requirements

PHP 7.0 or above (for PHP 5 support refer to php5 branch)
Pear's Console_CommandLine (for the protoc plugin)
Google's protoc compiler version 2.6 or above

Getting started

Installation

Clone the source code

git clone https://github.com/allegro/php-protobuf

Go to the source code directory
```
cd php-protobuf
```
Build and install the PHP extension (follow instructions at php.net)
Install protoc plugin dependencies
```
composer install
```

Usage

Assume you have a file foo.proto

message Foo
{
    required int32 bar = 1;
    optional string baz = 2;
    repeated float spam = 3;
}

Compile foo.proto
```
php protoc-gen-php.php foo.proto
```

Create Foo message and populate it with some data

require_once 'Foo.php';

$foo = new Foo();
$foo->setBar(1);
$foo->setBaz('two');
$foo->appendSpam(3.0);
$foo->appendSpam(4.0);

Serialize a message to a string
```
$packed = $foo->serializeToString();
```

Parse a message from a string

$parsedFoo = new Foo();
try {
    $parsedFoo->parseFromString($packed);
} catch (Exception $ex) {
    die('Oops.. there is a bug in this example, ' . $ex->getMessage());
}

Let's see what we parsed out

$parsedFoo->dump();

It should produce output similar to the following:

Foo {
  1: bar => 1
  2: baz => 'two'
  3: spam(2) =>
    [0] => 3
    [1] => 4
}

If you would like you can reset an object to its initial state
```
$parsedFoo->reset();
```

Guide

Compilation

PHP Protobuf comes with Google's protoc compiler plugin. You can run in directly:

php protoc-gen-php.php -o output_dir foo.proto

or pass it to the protoc:

protoc --plugin=protoc-gen-allegrophp=protoc-gen-php.php --allegrophp_out=output_dir foo.proto

On Windows use protoc-gen-php.bat instead.

Command line options

-o out, --out=out - the destination directory for generated files (defaults to the current directory).
-I proto_path, --proto_path=proto_path - the directory in which to search for imports.
--protoc=protoc - the protoc compiler executable path.
-D define, --define=define - define a generator option (i.e. -Dnamespace='Foo\Bar\Baz').

Generator options

namespace - the namespace to be used by the generated PHP classes.

Message class

The classes generated during the compilation are PSR-0 compliant (each class is put into it's own file). If namespace generator option is not defined then a package name (if present) is used to create a namespace. If the package name is not set then a class is put into global space.

PHP Protobuf module implements ProtobufMessage class which encapsulates the protocol logic. A message compiled from a proto file extends this class providing message field descriptors. Based on these descriptors ProtobufMessage knows how to parse and serialize a message of a given type.

For each field a set of accessors is generated. The set of methods is different for single value fields (required / optional) and multi-value fields (repeated).

required / optional

  get{FIELD}()        // return field value
  set{FIELD}($value)  // set field value to $value

repeated

  append{FIELD}($value)       // append $value value to field
  clear{FIELD}()              // empty field
  get{FIELD}()                // return array of field values
  getAt{FIELD}($index)        // return field value at $index index
  getCount{FIELD}()           // return number of field values
  getIterator{FIELD}($index)  // return ArrayIterator for field values

{FIELD} is a camel cased field name.

Enum

PHP does not natively support enum type. Hence enum is represented by the PHP integer type. For convenience enum is compiled to a class with set of constants corresponding to its possible values.

Type mapping

The range of available build-in PHP types poses some limitations. PHP does not support 64-bit positive integer type. Note that parsing big integer values might result in getting unexpected results.

Protocol Buffers types map to PHP types as follows (x86_64):

| Protocol Buffers | PHP    |
| ---------------- | ------ |
| double           | float  |
| float            |        |
| ---------------- | ------ |
| int32            | int    |
| int64            |        |
| uint32           |        |
| uint64           |        |
| sint32           |        |
| sint64           |        |
| fixed32          |        |
| fixed64          |        |
| sfixed32         |        |
| sfixed64         |        |
| ---------------- | ------ |
| bool             | bool   |
| ---------------- | ------ |
| string           | string |
| bytes            |        |

Protocol Buffers types map to PHP types as follows (x86):

| Protocol Buffers | PHP                         |
| ---------------- | --------------------------- |
| double           | float                       |
| float            |                             |
| ---------------- | --------------------------- |
| int32            | int                         |
| uint32           |                             |
| sint32           |                             |
| fixed32          |                             |
| sfixed32         |                             |
| ---------------- | --------------------------- |
| int64            | if val <= PHP_INT_MAX       |
| uint64           | then value is stored as int |
| sint64           | otherwise as double         |
| fixed64          |                             |
| sfixed64         |                             |
| ---------------- | --------------------------- |
| bool             | bool                        |
| ---------------- | --------------------------- |
| string           | string                      |
| bytes            |                             |

Not set value is represented by null type. To unset value just set its value to null.

Parsing

To parse message create a message class instance and call its parseFromString method passing it a serialized message. The errors encountered are signaled by throwing Exception. Exception message provides detailed explanation. Required fields not set are silently ignored.

$packed = /* serialized FooMessage */;
$foo = new FooMessage();

try {
    $foo->parseFromString($packed);
} catch (Exception $ex) {
    die('Parse error: ' . $e->getMessage());
}

$foo->dump(); // see what you got

Serialization

To serialize a message call serializeToString method. It returns a string containing protobuf-encoded message. The errors encountered are signaled by throwing Exception. Exception message provides detailed explanation. A required field not set triggers an error.

$foo = new FooMessage()
$foo->setBar(1);

try {
    $packed = $foo->serializeToString();
} catch (Exception $ex) {
    die 'Serialize error: ' . $e->getMessage();
}

/* do some cool stuff with protobuf-encoded $packed */

Debugging

There might be situations you need to investigate what an actual content of a given message is. What var_dump gives on a message instance is somewhat obscure.

The ProtobufMessage class comes with dump method which prints out a message content to the standard output. It takes one optional argument specifying whether you want to dump only set fields (by default it dumps only set fields). Pass falseas an argument to dump all fields. Format it produces is similar to var_dump.

Alternatively you can use printDebugString() method which produces output in protocol buffers text format.

IDE Helper and Auto-Complete Support

To integrate this extension with your IDE (PhpStorm, Eclipse etc.) and get auto-complete support, simply include stubs\ProtobufMessage.php anywhere under your project root.