Answer a question

Concept:

Using AWS Lambda functions with Python and Selenium, I want to create a undetectable headless chrome scraper by passing a headless chrome test. I check the undetectability of my headless scraper by opening up the test and taking a screenshot. I ran this test on a Local IDE and on a Lambda server.


Implementation:

I will be using a python library called selenium-stealth and will follow their basic configuration:

stealth(driver,
        languages=["en-US", "en"],
        vendor="Google Inc.",
        platform="Win32",
        webgl_vendor="Intel Inc.",
        renderer="Intel Iris OpenGL Engine",
        fix_hairline=True,
        )

I implemented this configuration on a Local IDE as well as an AWS Lambda Server to compare the results.


Local IDE:

Found below are the test results running on a local IDE: enter image description here


Lambda Server:

When I run this on a Lambda server, both the WebGL Vendor and Renderer are blank. as shown below:

enter image description here

I even tried to manually change the WebGL Vendor/Renderer using the following JavaScript command:

driver.execute_cdp_cmd('Page.addScriptToEvaluateOnNewDocument', {"source": "WebGLRenderingContext.prototype.getParameter = function(parameter) {if (parameter === 37445) {return 'VENDOR_INPUT';}if (parameter === 37446) {return 'RENDERER_INPUT';}return getParameter(parameter);};"})

Then I thought maybe that it could be something wrong with the parameter number. I configured the command execution without the if statement, but the same thing happened: It worked on my Local IDE but had no effect on an AWS Lambda Server.

Simply Put:

Is it possible to add Vendor/Renderer on AWS Lambda? In my efforts, it seems that there is no possible way. I made sure to submit this issue on the selenium-stealth GitHub Repository.

Answers

A solution I found for the missing WebGL Vendor/Renderer was using a docker container instead of the normal Lambda layers when creating a function. Not only does the storage increase by a factor of 40X but it also solves the WebGL Vendor/Renderer problem: enter image description here

Logo

Python社区为您提供最前沿的新闻资讯和知识内容

更多推荐