使用 php 驱动 selenium

网上看到 selenium 的教程,大多是 python 或者 java 驱动的,搞得我这种 php 出身的人好尴尬。 虽然 java 和 python 都会,既然是开发,那还是挑自己拿手的先解决任务吧。于是就有了这篇文章。本文只涉及 selenium webdriver,基于 chrome 浏览器,分布式的就先不看了。

文章基本上是 SeleniumFacebook 的文档简化后的翻译。

  1. 介绍
  2. Hello World
  3. 参考手册
  4. 窗口控制
  5. 寻找元素
  6. 等待
  7. Ajax
  8. 截图

WebDriver介绍

Selenium 2 一个主要的新功能就是移植了 WebDriver 的 Api。

WebDriver 项目的出发点是提供一个简单快捷的编程接口,来代替 Selenium-RC 的 Api。同事,Selenium WebDriver 对动态网页的支持更好,而且它的接口是面向对象的api,对浏览器支持更加完善。

WebDriver 与 Selenium-RC 驱动浏览器的区别

Selenium-Driver 直接调用浏览器底层支持,不同的浏览器有不同的方式。 而 Selenium-RC 在不同浏览器间的调用方式是一样的,都是以 javascript 驱动。

开始开发 - php版本

php 驱动 Selenium 可以参考以下三个项目的使用手册。

我使用的是 Facebook 版本的,所以直接看第三个。

在项目中使用 facebook/php-webdriver 很简单,在composer.json 的 require 中加入下面一行即可:

   "facebook/webdriver": "^1.3"

hello world

use Facebook\WebDriver\Remote\DesiredCapabilities;
use Facebook\WebDriver\Remote\RemoteWebDriver;
use Facebook\WebDriver\WebDriverBy;

public function test(){
    $host = 'http://localhost:4444/wd/hub'; // this is the default
    $capabilities = DesiredCapabilities::chrome();
    $driver = RemoteWebDriver::create($host, $capabilities, 5000);
    $baseUrl = 'http://www.bilibili.com/';
    $driver->get($baseUrl);
}

参考手册

打开浏览器

# start an instance of firefox with selenium-webdriver

$browser_type = 'firefox'
$host = 'http://localhost:4444/wd/hub'

$capabilities = array(\WebDriverCapabilityType::BROWSER_NAME => $browser_type);
$driver = RemoteWebDriver::create($host, $capabilities);

$browser_type
# :firefox => firefox
# :chrome  => chrome
# :ie      => iexplore

打开页面

https://php-webdriver.github.io/php-webdriver/master/Facebook/WebDriver/WebDriver.html

$driver->get('http://google.com');
$driver->navigate()->to('http://google.com');

元素操作

参考api: https://php-webdriver.github.io/php-webdriver/master/Facebook/WebDriver/WebDriverElement.html

  1. 获得元素内容

     $result = $driver->findElement(WebDriverBy::id('signin'))->getText();
    
  2. 触发鼠标划过事件

     $element = $driver->findElement(WebDriverBy::id('some_id'));
     $driver->getMouse()->mouseMove( $element->getCoordinates() );
    
  3. 点击元素(link, checkbox, etc.)

     $driver->findElement(WebDriverBy::id('signin'))->click();
    
  4. 文字输入框写入文字

     $driver->findElement(WebDriverBy::id("element id"))->sendKeys("I want to send this");
    
  5. 文字框清除文字

     $driver->findElement(WebDriverBy::id("element id"))->clear();
    
  6. 查看元素是否可见

     $element = $driver->findElement(WebDriverBy::id('element id'));
     if ($element->isDisplayed()) {
         // do something...
     }
    
  7. 刷新页面

     $driver->navigate()->refresh();
    

    或者使用js方法,具体看后文。

  8. 是否可见

     $driver->findElement(WebDriverBy::id('element'))->isDisplayed();
    
  9. 获取attribute

     $driver->findElement(WebDriverBy::id('element'))->getAttribute('class');
    

警告框、确认框和提示框操作

  1. 警告框

     $driver->switchTo()->alert()->dismiss(); // dismiss
     $driver->switchTo()->alert()->getText(); // get alert text, works for confirmations and prompts
    
  2. 确认框

     $driver->switchTo()->alert()->accept(); // accept
     $driver->switchTo()->alert()->dismiss(); // dismiss
    
  3. 提示框

     $driver->switchTo()->alert()->sendKeys('Test'); // enter text
     $driver->switchTo()->alert()->accept(); // submit
    
     $driver->switchTo()->alert()->dismiss(); // dismiss
    
  4. 选择框

    $select = new WebDriverSelect($driver->findElement(WebDriverBy::xpath('xxx')));
    $select->selectByValue("zh-CN");
    

窗口

  1. 最大化窗口

    $driver->manage()->window()->maximize();

运行js

  1. 同步脚本

    js 脚本会完整地运行在闭包中。所以访问全局变量的时候必须把变量名打全,比如:window.document

     $sScriptResult = $session->executeScript('return window.document.location.hostname',array());
     $sScriptResult now holds the value of the current document hostname
    
  2. 异步脚本

    必须告诉服务器需要等待的时长,否则服务器会抛出超时异常错误。

     // wait at most 5 seconds before giving up with a timeout exception
     $session->timeouts()->async_script(array('ms'=>5000));
    

    异步脚本也是运行在闭包中。

    • 简单的例子

        $sResult = $session->executeAsyncScript('arguments[arguments.length-1]("done");', array());
      
    • 复杂的例子

    In the example below, we poll the global window.MY_STUFF_DONE value at regular intervals, waiting for it to exist with a non-false value. Once we see it, we return back to the calling php-webdriver code with the value “done”.

     // define the javascript code to execute. This just checks at a periodic
     // interval to see if your page created the window.MY_STUFF_DONE variable
     $sJavascript = <<<END_JAVASCRIPT
    
     var callback = arguments[arguments.length-1], // webdriver async script callback
         nIntervalId; // setInterval id to stop polling
    
     function checkDone() {
       if( window.MY_STUFF_DONE ) {
         window.clearInterval(nIntervalId); // stop polling
         callback("done"); // return "done" to PHP code
       }
     }
    
     nIntervalId = window.setInterval( checkDone, 50 ); // start polling
     END_JAVASCRIPT;
    
     $sResult = $session->executeAsyncScript($sJavascript,array());
     $sResult == "done"
    
  # By name
  $driver->manage()->deleteCookieNamed('CookieName');

  # Or all of them
  $driver->manage()->deleteAllCookies();

窗口控制

  1. 获得当前窗口句柄

     $handle = $session->getWindowHandle();
    
  2. 获得所有窗口句柄array

     $handles = $session->getWindowHandles(); //now can iterate through array to get desired handle
    
  3. 切换窗口

     $driver->switchTo()->window($handle);
    
  4. 切换Frame

     //Find an iframe
     $iframe = $session->findElement(WebDriverBy::tagName('iframe'));  //Get the iframe id
     $frameId = $iframe->getAttribute('id');
    

    //Switch the driver to the iframe $session->switchTo()->frame($iframe);

    //Go back to the main document $session->switchTo()->defaultContent();

    frame() 方法需要的是一个 element,所以用 WebDriverBy::xpath 也可以的。

     driver.switchTo().parentFrame();
     driver.switchTo().defaultContent();
    

寻找元素

  • Css选择器 - WebDriverBy::cssSelector(‘h1.foo > small’)
  • Xpath - WebDriverBy::xpath(‘(//hr)[1]/following-sibling::div[2]’)
  • Id - WebDriverBy::id(‘heading’)
  • Class - WebDriverBy::className(‘warning’)
  • input属性 - WebDriverBy::name(‘email’)
  • Tag - WebDriverBy::tagName(‘h1’)
  • 文字 - WebDriverBy::linkText(‘Sign in here’)
  • Partial link text - WebDriverBy::partialLinkText(‘Sign in’)

在下面的例子中 $driver 是 RemoteWebDriver 的实例

单个元素

$element = $driver->findElement(WebDriverBy::cssSelector('div.header'));
// $element will be instance of RemoteWebElement

$headerText = $element->getText();

多个元素

$elements = $driver->findElements(WebDriverBy::cssSelector('ul.foo > li'));
// $elements is now array - containing instances of RemoteWebElement (or empty, if no element is found)

foreach ($elements as $element) {
    var_dump($element->getText());
}

等待元素

$element = $driver->wait()->until(
    WebDriverExpectedCondition::presenceOfElementLocated(WebDriverBy::cssSelector('div.bar'))
);

$elements = $driver->wait()->until(
    WebDriverExpectedCondition::presenceOfAllElementsLocatedBy(WebDriverBy::cssSelector('ul > li'))
);

这几个方法也有同样的效果 visibilityOfElementLocated, visibilityOf, stalenessOf, elementToBeClickable.

等待

// Wait for the page tile to be 'My Page'. 

// Default wait (= 30 sec)
$driver->wait()->until(
  WebDriverExpectedCondition::titleIs('My Page')
);

// Wait for at most 10s and retry every 500ms if it the title is not correct.
$driver->wait(10, 500)->until(
  WebDriverExpectedCondition::titleIs('My Page')
);

title

titleIs()
titleContains()
titleMatches()

URL

urlIs()
urlContains()
urlMatches()

元素

presenceOfElementLocated()
presenceOfAllElementsLocatedBy()
elementTextIs()
elementTextContains()
elementTextMatches()
textToBePresentInElementValue()

元素可见性

visibilityOfElementLocated()
visibilityOf()
invisibilityOfElementLocated()
invisibilityOfElementWithText()

Frames, 窗口, 警告

frameToBeAvailableAndSwitchToIt()
elementToBeClickable()
alertIsPresent()
numberOfWindowsToBe()

其它

stalenessOf()
refreshed()
not()
elementToBeSelected()
elementSelectionStateToBe()

自定义条件

until方法

$driver->wait()->until(
    function () use ($driver) {
        $elements = $driver->findElements(WebDriverBy::cssSelector('li.foo'));

        return count($elements) > 5;
    },
    'Error locating more than five elements'
);

确定的等待

$driver->manage()->timeouts()->implicitlyWait(10);

Ajax

$submitButton = $driver->findElement(WebDriverBy::id('Submit'));
$submitButton->click();
waitForAjax($driver); // Default is jquery same with waitForAjax($driver, 'jquery');
//waitForAjax($driver, 'prototype');
//waitForAjax($driver, 'dojo');
$anotherButton = $driver->findElement(WebDriverBy::id('secondButton'));

关于 waitForAjax() 有两种实现,可以选择其中一张。

function waitForAjax($driver, $framework='jquery')
{
    // javascript framework
    switch($framework){
        case 'jquery':
            $code = "return jQuery.active;"; break;
        case 'prototype':
            $code = "return Ajax.activeRequestCount;"; break;
        case 'dojo':
            $code = "return dojo.io.XMLHTTPTransport.inFlight.length;"; break;
        default:
            throw new Exception('Not supported framework');
    }

    do {
        sleep(2);
    } while ($driver->executeScript($code));
}

function waitForAjax($driver, $framework='jquery')
{
    // javascript framework
    switch($framework){
        case 'jquery':
            $code = "return jQuery.active;"; break;
        case 'prototype':
            $code = "return Ajax.activeRequestCount;"; break;
        case 'dojo':
            $code = "return dojo.io.XMLHTTPTransport.inFlight.length;"; break;
        default:
            throw new Exception('Not supported framework');
    }

    // wait for at most 30s, retry every 2000ms (2s)
    $driver->wait(30, 2000)->until(
        function ($driver, $code) {
            return !$driver->executeScript($code);
        }
    );
}    

其他的实现方法看这里 http://agilesoftwaretesting.com/?p=111

截图

使用

$full_screenshot = TakeScreenshot();
$screenshot_of_element = TakeScreenshot($this->driver->findElement(WebDriverBy::xpath("//img[@class='test']"));

实现

public function TakeScreenshot($element=null) {
    // Change the Path to your own settings
    $screenshot = $this->TempDirectoryPath . time() . ".png";

    // Change the driver instance
    $this->driver->takeScreenshot($screenshot);
    if(!file_exists($screenshot)) {
        throw new Exception('Could not save screenshot');
    }
    
    if( ! (bool) $element) {
        return $screenshot;
    }
    
    $element_screenshot = $this->TempDirectoryPath . time() . ".png"; // Change the path here as well
    
    $element_width = $element->getSize()->getWidth();
    $element_height = $element->getSize()->getHeight();
    
    $element_src_x = $element->getLocation()->getX();
    $element_src_y = $element->getLocation()->getY();
    
    // Create image instances
    $src = imagecreatefrompng($screenshot);
    $dest = imagecreatetruecolor($element_width, $element_height);

    // Copy
    imagecopy($dest, $src, 0, 0, $element_src_x, $element_src_y, $element_width, $element_height);
    
    imagepng($dest, $element_screenshot);
    
    // unlink($screenshot); // unlink function might be restricted in mac os x.
    
    if( ! file_exists($element_screenshot)) {
        throw new Exception('Could not save element screenshot');
    }
    
    return $element_screenshot;
}

参考资料


使用 snmp 协议监控服务器 usr/bin/ld cannot find lxxx 问题