前面的链接：
用php从零开始写一个编译器（一）
用PHP从零开始写一个编译器（二）
用PHP从零开始写一个编译器（三）
用PHP从零开始写一个编译器（四）
前面我们把整个编译器的算法讲解完成了。Parser类编译完成以后是一个PHP程序可用的数组。但是，为了对程序员友好，我们最后把数据封装到了Query类中。为什么要这么做呢，这是因为，对于完整的查询或者是查询语句片断，PHP程序使用时是要区别对待的。其实，任何一个开源，如果你对程序员不友好，那就是死路一条。这也就是为什么Laravel框架能在短时间内成为世界第一，因为，它真的是好用。
其实，Query类，也并没有做什么，它就是实现了QueryInterface这个接口。
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\RqlParser;
/**
* Interface QueryInterface
*
* @package ByteFerry\RqlParser
*
* @property array $container
*/
interface QueryInterface
{
/**
* @return QueryInterface
*/
public static function of();
/**
* @param array $query
*
* @return \ByteFerry\RqlParser\QueryInterface
*/
public function from($query);
/**
* @param $name
*
* @return mixed|null
*/
public function __get($name);
/**
* @param $name
*
* @return bool
*/
public function __isset($name);
/**
* @return mixed
*/
public function toArray();
}
而实现这个接口的有两个类，一个叫Fragment，它是RQL查询片段。
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\RqlParser;
use ByteFerry\RqlParser\Abstracts\BaseObject;
/**
* Class Fragment
*
* @package ByteFerry\RqlParser
*/
class Fragment extends BaseObject implements QueryInterface
{
/**
* @var array
*/
protected $container = [];
/**
* @param array $query
*
* @return \ByteFerry\RqlParser\Fragment
*/
public function from($query)
{
$this->container = $query;
return $this;
}
/**
* @param $name
*
* @return mixed|null
*/
public function __get($name)
{
if (isset($this->container[$name])) {
return $this->container[$name];
}
return null;
}
/**
* @param $name
*
* @return bool
*/
public function __isset($name)
{
return isset($this->container[$name]);
}
/**
* @return array|mixed
*/
public function toArray(){
return $this->container;
}
}
另一个则是Query，Query对象提供了
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\RqlParser;
use ByteFerry\RqlParser\Abstracts\BaseObject;
/**
* Class Query
*
* @package ByteFerry\RqlParser
*/
class Query extends BaseObject implements QueryInterface
{
/**
* @var array
*/
protected $container = [];
/**
* @param array $query
*
* @return \ByteFerry\RqlParser\Query
*/
public function from($query)
{
$this->container = $query;
return $this;
}
/**
* @param $name
*
* @return mixed|null
*/
public function __get($name){
if(isset($this->container[$name])){
return $this->container[$name];
}
return null;
}
/**
* @param $name
*
* @return bool
*/
public function __isset($name){
return isset($this->container[$name]);
}
/**
* @return array|mixed
*/
public function toArray(){
return $this->container;
}
/**
* @return mixed|null
*/
public function getOperator(){
return $this->container['operator']??null;
}
/**
* @return mixed|null
*/
public function getQueryType(){
return $this->container['query_type']??null;
}
/**
* @return string
*/
public function getResourceName(){
return $this->container['resource']??null;
}
/**
* @return array
*/
public function getColumns(){
return $this->container['columns']??null;
}
/**
* @return mixed|null
*/
public function getColumnsOperator(){
return $this->container['columns_operator']??null;
}
/**
* @return mixed|null
*/
public function getGroupBy(){
return $this->container['group_by']??null;
}
/**
* @return mixed|null
*/
public function getFilter(){
return $this->container['filter'][0]??null;
}
/**
* @return mixed|null
*/
public function getParameters(){
return $this->container['paramaters']??null;
}
/**
* @return mixed|null
*/
public function getSearch(){
return $this->container['search']??null;
}
/**
* @return mixed|null
*/
public function getSort(){
return $this->container['sort']??null;
}
/**
* @return mixed|null
*/
public function getHaving(){
return $this->container['having'][0]??null;
}
/**
* @return mixed|null
*/
public function getLimit(){
return $this->container['limit']??null;
}
项目当中，剩下的两个类，那就是BaseObject 和 ParseException，无需讲解。接下来，就是单元测试了。
为什么要单元测试？假如你写的组件，那是交给天下人用的。如果未经测试，谁愿意做你的小白鼠？使用单元测试，我可以用composer下载之后，可以通过PhpUnit命令行验证一下。为什么要验证？因为，php版本，关联组件版本可能不一样。就象一些公司，说明，有新版本了，于是，尽快升级，但是，一升级，马上就崩了。这种情况相当常见，毕境，一旦有关联组件，肯定就有兼容问题，比如，thinkphp5.0就不兼容现在的redis，redis把函数名改了，del改成delete了。这还是小事。像thinkphp5.0与thinkphp5.1直接未在composer中设置版本冲突，所以，你在用5.0时，一不小心升级了，于是崩了。再如，thinkphp5系列均不带PhpUnit，所以，当我安装了组件之后，我无法验证可不可运行。只有在开发了，代码出来以后，才开始报错，这就浪费了大量的时间。所以，一般我composer安装后，喜欢用PhpUnit验证一下之后再开发，如果不行，则换版本，或换组件。
由此来看，你把你的想法实现了，交给大家来用，这是开源精神，值得称赞，但是，实际上，github中大量的开源并不是按此规范做的。所以，那些开源基本无人问津。
那么，该怎么测呢？首先我们要安装PhpUnit 和 orchestra/testbench，为什么要安装orchestra/testbench呢？这是为了能和travisCI这个免费的持续集成工具集成方便。另外，写代码也会方便很多。
关于这两个组件的使用，不多讲了，大家还是直接看文档。基本步骤是，配置phpuint.xml，用以告诉它，怎么测。然后，继承一个TestCase，废话少说，上代码：
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\Tests;
use Orchestra\Testbench\TestCase as Orchestra;
/**
* Class TestCase
*
* @package ByteFerry\tests
*/
class TestCase extends Orchestra
{
/**
* The base URL to use while testing the application.
*
* @var string
*/
protected $baseUrl = 'http://localhost';
public function setUp(): void
{
// touch('./tests.sqlite');
parent::setUp();
}
public function tearDown(): void
{
// unlink('./tests.sqlite');
}
/**
* Define environment setup.
*
* @param \Illuminate\Foundation\Application $app
*
* @return void
*/
protected function getEnvironmentSetUp($app)
{
// Setup default database to use sqlite :memory:
$app['config']->set('database.default', 'testbench');
$app['config']->set('database.connections.testbench', [
'driver' => 'sqlite',
'database' => ':memory:',
'prefix' => '',
]);
}
}
这个类的代码基本都是默认的。接下来，要写相关测试了。 首先，我们要测RQL的所有关键字。
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\Tests\Unit;
use ByteFerry\Tests\TestCase;
use ByteFerry\RqlParser\Parser;
/**
* Class KeyTest
*
* @package ByteFerry\tests\Unit
*/
final class KeyTest extends TestCase
{
protected $test_keys = [
'aggr' =>['aggr(id,sum(amount))','{"columns":["id","sum(amount)"],"columns_operator":"aggr","group_by":["id"]}'],
'aggregate' =>['aggregate(id,sum(amount))','{"columns":["id","sum(amount)"],"columns_operator":"aggregate","group_by":["id"]}'],
'and' =>['and(gt(age,20),lt(id,300),is(name,null()))','"( age > 20 )and( id < 300 )and( name is null )"'],
'arr' =>['in(id,(1,34,2,4))','" id in (1, 34, 2, 4) "'],
'avg' =>['aggr(id,avg(amount))', '{"columns":["id","avg(amount)"],"columns_operator":"aggr","group_by":["id"]}'],
'between' =>['between(age,18,55)','" age BETWEEN 18 and 55 "'],
'cols' =>['cols(id,age)','{"columns":["id","age"],"columns_operator":"cols","group_by":[]}'],
'columns' =>['columns(id,age,name)','{"columns":["id","age","name"],"columns_operator":"columns","group_by":[]}'],
'data' =>['data(id:17,score:33)','{"data":{"id":"17","score":"33"}}'],
'empty' => ['is(name,empty())','" name is \"\" "'],
'eq' =>['eq(age,17)','" age = 17 "'],
'except' =>['except(password)','{"columns":["password"],"columns_operator":"except","group_by":[]}'],
'false' => ['filter(false())','{"filter":[0],"paramaters":[]}'],
'filter' =>['filter(eq(age,11))','{"filter":[" age = 11 "],"paramaters":{"age":"11"}}'],
'ge' =>['ge(age,35)','" age >= 35 "'],
'gt' =>['gt(age,15)','" age > 15 "'],
'having' =>['having(ge(sum(amount),100))','{"having":[" sum(amount) >= 100 "],"paramaters":{"sum(amount)":"100"}}'],
'in' =>['in(id,(1,3,5))','" id in (1, 3, 5) "'],
'is' => ['is(name,null())','" name is null "'],
'le' =>['le(age,3)','" age <= 3 "'],
'like' =>['like(name,"ad%")','" name like \"ad%\" "'],
'limit' =>['limit(1,20)','{"limit":["1","20"]}'],
'lt' =>['lt(age,15)','" age < 15 "'],
'max' =>['aggr(id,max(amount))','{"columns":["id","max(amount)"],"columns_operator":"aggr","group_by":["id"]}'],
'mean' =>['aggr(id,mean(amount))','{"columns":["id","avg(amount)"],"columns_operator":"aggr","group_by":["id"]}'],
'min' =>['aggr(id,min(amount))','{"columns":["id","min(amount)"],"columns_operator":"aggr","group_by":["id"]}'],
'ne' =>['ne(age,15)','" age <> 15 "'],
'nin' =>['nin(age,(15,25,35))','" age not in (15, 25, 35) "'],
'not' =>['is(name,not(null()))','" name is not null "'],
'null' =>['is(name,not(null()))','" name is not null "'],
'only' =>['only(id,name,age,email)','{"columns":["id","name","age","email"],"columns_operator":"only","group_by":[]}'],
'or' =>['or(eq(age,18),eq(gender,1))','"( age = 18 )or( gender = 1 )"'],
'out' =>['out(id,(1,3,5))','" id not in (1, 3, 5) "'],
'search' =>['search("jh%")','{"search":"jh%"}'],
'select' =>['select(id,name,age)','{"columns":["id","name","age"],"columns_operator":"select","group_by":[]}'],
'sort' =>['sort(+age,-name)','{"sort":[["age","ASC"],["name","DESC"]]}'],
'sum' =>['aggr(id,sum(amount))','{"columns":["id","sum(amount)"],"columns_operator":"aggr","group_by":["id"]}'],
'true' => ['eq(deleted,true())','" deleted = 1 "'],
'values' =>['values(name)','{"columns":["name"],"columns_operator":"values","group_by":[]}'],
];
/** @test */
public function testKeys(){
foreach($this->test_keys as $key => $item){
[$test_str,$compare] = $item;
$result = Parser::parse($test_str,true);
$this->assertEquals($compare,json_encode($result[0]->toArray()));
}
}
}
这里，我们使用了一个偷懒的办法，那就是，我们把编译的结果用json_encode转换为字符串，与我们相要的字符串比对，如果一致，则OK。（这样做，是让代码行数变少）
第二步，我们要测所有的完整的查询
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\Tests\Unit;
use ByteFerry\Tests\TestCase;
use ByteFerry\RqlParser\Parser;
/**
* Class SimpleQueryTest
*
* @package ByteFerry\tests\Unit
*/
final class SimpleQueryTest extends TestCase
{
public $query_keys = [
'all' =>['all(user,only(id,age))', '{"resource":"user","columns":["id","age"],"columns_operator":"only","group_by":[],"operator":"all","query_type":"Q_READ"}'],
'any' =>['any(user,cols(id,name,age))','{"resource":"user","columns":["id","name","age"],"columns_operator":"cols","group_by":[],"operator":"any","query_type":"Q_READ"}'],
'count' =>['count(user,cols(id))','{"resource":"user","columns":["id"],"columns_operator":"cols","group_by":[],"operator":"count","query_type":"Q_READ"}'],
'create' =>['create(user,data(name:jhon,age:17))','{"resource":"user","data":{"name":"jhon","age":"17"},"operator":"create","query_type":"Q_WRITE"}'],
'decrement' =>['decrement(Article,cols(read_count))','{"resource":"Article","columns":["read_count"],"columns_operator":"cols","group_by":[],"operator":"decrement","query_type":"Q_WRITE"}'],
'delete' =>['delete(User,filter(eq(id,23)))','{"resource":"User","filter":[" id = 23 "],"paramaters":{"id":"23"},"operator":"delete","query_type":"Q_WRITE"}'],
'distinct' =>['distinct(User_score,filter(gt(score,95)))','{"columns":["User_score",{"filter":[" score > 95 "],"paramaters":{"score":"95"}}],"columns_operator":"distinct","group_by":[]}'],
'exists' =>['exists(user,filter(eq(mobile,1111)))','{"resource":"user","filter":[" mobile = 1111 "],"paramaters":{"mobile":"1111"},"operator":"exists","query_type":"Q_READ"}'],
'first' =>['first(user,filter(gt(age,10)))','{"resource":"user","filter":[" age > 10 "],"paramaters":{"age":"10"},"operator":"first","query_type":"Q_READ"}'],
'increment' =>['increment(Article,cols(read_count))','{"resource":"Article","columns":["read_count"],"columns_operator":"cols","group_by":[],"operator":"increment","query_type":"Q_WRITE"}'],
'minus' =>['minus(Article,cols(read_count))','{"resource":"Article","columns":["read_count"],"columns_operator":"cols","group_by":[],"operator":"decrement","query_type":null}'],
'one' =>['one(user,filter(eq(id,3)),search(ada))','{"resource":"user","filter":[" id = 3 "],"paramaters":{"id":"3"},"search":"ada%","operator":"one","query_type":"Q_READ"}'],
'plus' =>['plus(Article,cols(read_count))','{"resource":"Article","columns":["read_count"],"columns_operator":"cols","group_by":[],"operator":"increment","query_type":null}'],
'update' =>['update(user,data(age:18),filter(eq(id,3)))','{"resource":"user","data":{"age":"18"},"filter":[" id = 3 "],"paramaters":{"id":"3"},"operator":"update","query_type":"Q_WRITE"}']
];
/** @test */
public function testQuery(){
foreach($this->query_keys as $key => $item){
[$test_str,$compare] = $item;
$result = Parser::parse($test_str);
$this->assertEquals($compare,json_encode($result[0]->toArray()));
}
}
}
第三步，我们要测QueryInterface这个接口类
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\Tests\Unit;
use ByteFerry\Tests\TestCase;
use ByteFerry\RqlParser\Parser;
use ByteFerry\RqlParser\Fragment;
/**
* Class QueryInterfaceTest
*
* @package ByteFerry\Tests\Unit
*/
final class QueryInterfaceTest extends TestCase
{
/** @test */
public function testQueryInterface(){
$rql_str= 'all(User,aggr(id,name,age,gender,address,avg(age)),filter(is(created_at, null()), search(Jhon),sort(-id,+age),having(gt(sum(amount),0)),limit(0,20)))'; //, //,
$result = Parser::parse($rql_str);
/** @var \ByteFerry\RqlParser\Query $query */
$query = $result[0];
$this->assertEquals($query->getResourceName(),'User');
$this->assertEquals($query->getColumns(),[0 => 'id',
1 => 'name',
2 => 'age',
3 => 'gender',
4 => 'address',
5 => 'avg(age)']);
$this->assertEquals($query->getColumnsOperator(),'aggr');
$this->assertEquals($query->getGroupBy(),[
0 => 'id',
1 => 'name',
2 => 'age',
3 => 'gender',
4 => 'address']);
$this->assertEquals($query->getFilter(),' created_at is null ');
$this->assertEquals($query->getParameters(),[
'created_at' => NULL,
'sum(amount)' => '0'
]);
$this->assertEquals($query->getSearch(),'Jhon%');
$this->assertEquals($query->getSort(),[
[ 0 => 'id',
1 => 'DESC'
],
[
0 => 'age',
1 => 'ASC',
]
]);
$this->assertEquals($query->getHaving(),' sum(amount) > 0 ');
$this->assertEquals($query->getLimit(),['0','20']);
$this->assertEquals($query->getOperator(),'all');
$this->assertEquals($query->getQueryType(),'Q_READ');
$this->assertEquals($query->query_type,'Q_READ');
$this->assertEquals($query->operator,'all');
$this->assertTrue(isset($query->operator));
$this->assertEquals($query->name,null);
}
/** @test */
public function testFragment(){
$rql_str= 'aggregate(id,sum(amount))'; //, //,
$result = Parser::parse($rql_str,true);
/** @var \ByteFerry\RqlParser\Fragment $query */
$query = $result[0];
$this->assertTrue($query instanceof Fragment);
$this->assertEquals($query->columns,["id","sum(amount)"]);
$this->assertEquals($query->group_by,["id"]);
$this->assertEquals($query->columns_operator,'aggregate');
$this->assertFalse(isset($query->operator));
$this->assertEquals($query->operator,null);
}
}
最后，我们还要进行异常测试。为什么呢？这是因为，出错时，你要保证抛出异常给程序员看到。同时，也只有经过异常测试，才能保证代码测试的覆盖率。
declare(strict_types=1);
/*
* This file is part of the ByteFerry/Rql-Parser package.
*
* (c) BardoQi <67158925@qq.com>
*
* For the full copyright and license information, please view the LICENSE
* file that was distributed with this source code.
*/
namespace ByteFerry\Tests\Unit;
use ByteFerry\RqlParser\Fragment;
use ByteFerry\Tests\TestCase;
use ByteFerry\RqlParser\Parser;
use ByteFerry\RqlParser\Exceptions\ParseException;
/**
* Class ExceptionTest
*
* @package ByteFerry\Tests\Unit
*/
final class ExceptionTest extends TestCase
{
/** @test */
public function testExceptions(){
$rql_str= 'aggregate(id,,sum(amount))'; //, //,
try{
$result = Parser::parse($rql_str,true);
}catch(\Exception $e){
$this->assertTrue($e instanceof ParseException);
}
$rql_str= 'aggregate(id,sum(amount)'; //, //,
try{
$result = Parser::parse($rql_str,true);
}catch(\Exception $e){
$this->assertTrue($e instanceof ParseException);
}
}
}
最后，我们要集成到TraviCI , 并且，用CodeCov检查代码测试覆盖率。 如何配，这里，我们不是单独运行，因为，本组件是无任何依赖项的，我们选择集成到Laravel，为什么不选择thinkPHP？原因相当简单，它那里连phpUnit都没有，最后出错，用户认为我错了，岂不坏了我的名声？
其实， 这些东西简单的，官方网站都有详细的文档，并且是免费的。 TraviCI配置tavis.yml如下：(是基于yml的配置)
language: php
php:
- 7.2
env:
- LARAVEL_VERSION=5.8.*
- LARAVEL_VERSION=6.*
- LARAVEL_VERSION=7.*
- LARAVEL_VERSION=8.*
before_script:
- travis_retry composer self-update
- travis_retry composer install --prefer-source --no-interaction
- if [ "$LARAVEL_VERSION" != "" ]; then composer require --dev "laravel/laravel:${LARAVEL_VERSION}" --no-update; fi;
- composer dump-autoload
script:
- ./vendor/bin/phpunit --coverage-clover coverage.xml
after_success:
- bash <(curl -s https://codecov.io/bash)
notifications:
email:
recipients:
- bardoqi@gmail.com
on_success: always
on_failure: always
测试做完了，集成完成，你可以将相关链接图标放到你的readme.md中。如下
（说明，上面是截图，这里无法正常显示动态的SHELD，只能上截图了）
这些链接告诉用户，你的代码是经过测试的，这其中，
第一个是 TravisCI 编译通过的标识
第二个是SytleCI（代码格式规范）通过的标志，
第三个，最重要的，就是测试代码覆盖率，总共覆盖了多少。我们可是99%
后面三个则是packagist.org版本与下载量，以及你的授权。这些主要是让人一眼可以看出是否可以用。
到此，我们的PHP代码编译器开发算是完成了。最后，别忘了，能够提供详细的文档，我们提供了中英文两个版本的readme,同时还提供了在线的RQL文档。因为，我们修改了一些RQL的关键字。此外，大家可以看出，此代码相当易用，因为，就是一个静态函数。为什么要这么做？越简单，大家用得越爽呀。另一方面，毕境是每个版本，目前不清楚客户会有什么配置项之类的。当然从简。
大家可以到
github.com/byteferry/r…
国内链接： gitee.com/bulo/rql-pa…
下载完整的原码。如果你觉得有用，别忘了star，既是自己收藏，也是一份鼓励。
那么，这个组件的效率如何？我们在内存仅8G, i7CPU的笔记本环境，测试下来，最长的RQL查询的编译一般在2毫秒左右。看起来，这里有10多个文件，但是，对一个懂得设计模式的人来说，每一个文件最多300行，那代码阅读是相当开心的。换个初学者，一样可以一个文件，3000行一个函数，也能完成，但未来就惨了。
所以，一些人认为，代码越多，程序越慢，实际是不对的。代码快与慢，关键是看数据结构与程序结构，不是代码多与少的问题。比如,ElasticSearch代码远比一般全文搜索的代码多得多了，为什么那么快？
同时，我们也可以看到，我们代码中，没有一个函数超过30行的。
那么，有人要问，你这是不是重造轮子？实际不是，因为，第一，目前任何一个PHP的RQL Parser均没有理想的程序结构。第二，更没有支持写查询的。第三，有些效率也不是很高。
rql-parser 是我们的第一步计划，我们是根据组件化的原理，下一步要写ApiBirdge，用来真正增强RestAPi的调用。所以，这个组件中的RQL与现在RQL标准的草案还是有些区别的。
自然，作者水平有限，也希望高手一些参与。RQL作为一项新技术，我相信未来肯定会被广泛运用到基础应用架构以及ORM等领域。作为个人，做开源的力量是有限的，所以，也希望感兴趣的企业共同参与。（全文完）
常见问题FAQ
- 免费下载或者VIP会员专享资源能否直接商用？
- 本站所有资源版权均属于原作者所有，这里所提供资源均只能用于参考学习用，请勿直接商用。若由于商用引起版权纠纷，一切责任均由使用者承担。更多说明请参考 VIP介绍。
- 提示下载完但解压或打开不了？
- 找不到素材资源介绍文章里的示例图片？
- 模板不会安装或需要功能定制以及二次开发？
发表评论
还没有评论，快来抢沙发吧！