selenium 拦截网站非ajax数据，爬虫处理js

最新推荐文章于 2024-06-28 20:29:35 发布

TensorGroot

最新推荐文章于 2024-06-28 20:29:35 发布

阅读量1.9k

点赞数

分类专栏：爬虫文章标签： javascript ajax 爬虫

本文链接：https://blog.csdn.net/qq948993066/article/details/70195672

版权

爬虫专栏收录该内容

4 篇文章 0 订阅

订阅专栏

WebDriver driver = new MyPhantomJSDriver(sessionId, 48105);
		driver.findElement(By.name("username")).sendKeys(name);
		driver.findElement(By.name("idcard")).sendKeys(idCard);
		driver.findElement(By.name("cdrCondition.randpsw")).sendKeys(code);

		String submit = "jQuery.ajax({"+"\n"
           +" type: \"POST\","+"\n"
           +" dataType: \"html\","+"\n"
           +" url: \"/myTest.htm\","+"\n"
           +" data: jQuery(\"form[name='form1']\").serialize(),"+"\n"
           +" success: function (result) {"+"\n"
           +"     re=/(.|\\s|\\S)*(location.href.* error.html)(.|\\s|\\S)*/"+"\n"
           +"     if(re.test(result)){"+"\n"
           +"         window.myData=false;"+"\n"
           +"         return;"+"\n"
           +"     }"+"\n"
           +"     window.myData=true;"+"\n"
           +"     jQuery(\"body\").append(\"<iframe srcdoc='\"+result+\"'></iframe>\");"+"\n"
           +" },"+"\n"
           +" error: function(data) {"+"\n"
           +"     console.log(\"error:\"+data.responseText);"+"\n"
           +"  }"+"\n"
           +" });";
		
		WebDriverWait wait = new WebDriverWait(driver, 10);
		((RemoteWebDriver) driver).executeScript(submit);
		boolean success = (boolean)wait.until(new Function<WebDriver, Object>() {
			public Object apply(@Nullable WebDriver driver) {
				return ((RemoteWebDriver) driver).executeScript("return window.myData;");
			}
		});
		if(!success){
			return new Result(Constants.INPUTERROR, Constants.getMessage(Constants.INPUTERROR));
		}