DEV Community

AWS Community ASEAN profile image Jaira Encio
Jaira Encio for AWS Community ASEAN

Posted on • Edited on

Creating an API that runs Selenium via AWS Lambda

#serverless #lambda #selenium

Being an automation tester, my job is to automate everything. As I was running my test script via terminal I realised that I’m the only who can execute the scripts I made. What if someone wants to run it? like the Devs, Project Manager, etc. It would be a tedious task to clone my repo, install libraries, and run the script. So I decided that maybe I can store my test script inside a serverless machine and make it accessible via API request.

I experimented with various AWS resources such as creating my own lambda function, checking features of API gateway, Codepipeline, etc. After several attempts, I was finally able to run my script inside. And then I just researched how to access my lambda via API.

This will result in higher production and time savings. Engineers may focus on vital work because automated testing does not require human interaction. This is like a portable testing device that anyone could execute. With fast test execution, developers get testing reports in an instant, so whenever a failure occurs, they will react to it quickly. Test automation will make it easier to update programs quickly. As a result, automated testing leads to increased team responsiveness, improved user experience, and increased customer satisfaction.

Overview

  1. Create 2 Lambda layers that has selenium and chromedriver libraries
  2. Include created lambda layers in serverless.yml of lambda then deploy

Creating Selenium Lambda Layer

Place libraries in python/lib/python3.6/site-packages/ to include them in a layer.

Download Selenium to layer directory



$ pip3.6 install -t selenium/python/lib/python3.6/site-packages selenium==3.8.0
$ cd selenium
$ zip -r python.zip python/
Enter fullscreen mode Exit fullscreen mode

Once finished, Create lambda layer then upload zip file

1. Go to AWS Console Lambda/Layers
2. Click Create Layer
3. Input the following in the layer configuration



Name: selenium
Description: Selenium layer
Upload zip file created: python.zip
Compatible runtimes: Python 3.6
Enter fullscreen mode Exit fullscreen mode

selenium-layer

4. Click Create

Note: You can user whatever version you prefer, you just need to select compatible runtime when uploading your package

 Creating Chromedriver Lambda layer

Download chrome driver



$ mkdir -p chromedriver
$ cd chromedriver
$ curl -SL https://chromedriver.storage.googleapis.com/2.37/chromedriver_linux64.zip > chromedriver.zip
$ unzip chromedriver.zip
$ rm chromedriver.zip
Enter fullscreen mode Exit fullscreen mode

Download chrome binary



$ curl -SL https://github.com/adieuadieu/serverless-chrome/releases/download/v1.0.0-41/stable-headless-chromium-amazonlinux-2017-03.zip > headless-chromium.zip
$ unzip headless-chromium.zip
$ rm headless-chromium.zip
Enter fullscreen mode Exit fullscreen mode

Compress driver and binary



$ ls
chromedriver headless-chromium
$ zip -r chromedriver.zip chromedriver headless-chromium
Enter fullscreen mode Exit fullscreen mode

Once finished, Create lambda layer then upload zip file

1. Go to AWS Console Lambda/Layers
2. Click Create Layer
3. Input the following in the layer configuration



Name: chromedriver
Description: chrome driver and binary layer
Upload zip file created: chromedriver.zip
Compatible runtimes: Python 3.6
Enter fullscreen mode Exit fullscreen mode

chrome-layer

4. Click Create

 Creating Lambda Function

To ensure that your function code has access to libraries included in layers, Lambda runtimes include paths in the '/opt' directory.

 File Structure 


 shell
── /lambda/            # lambda function
  ├── /handler.py      # source code of lambda function 
  └── /serverless.yaml # serverless config
Enter fullscreen mode Exit fullscreen mode

 Code

Copy the code below to /lambda/handler.py



from selenium import webdriver
from selenium.webdriver.chrome.options import Options

def main(event, context):
    options = Options()
    options.binary_location = '/opt/headless-chromium'
    options.add_argument('--headless')
    options.add_argument('--no-sandbox')
    options.add_argument('--single-process')
    options.add_argument('--disable-dev-shm-usage')

    driver = webdriver.Chrome('/opt/chromedriver',chrome_options=options)
    driver.get('https://www.google.com/')

    driver.close();
    driver.quit();

    response = {
        "statusCode": 200,
        "body": "Selenium Headless Chrome Initialized"
    }

    return response
Enter fullscreen mode Exit fullscreen mode

Copy the code below to /lambda/serverless.yaml.



service: selenium-lambda

provider:
  name: aws
  runtime: python3.6
  region: ap-southeast-2
  timeout: 900

functions:
  main:
    memorySize: 1000
    handler: handler.main
    events:
      - http:
          path: test
          method: get

    layers:
      - arn:aws:lambda:ap-southeast-2:{}:layer:chromedriver:2
      - arn:aws:lambda:ap-southeast-2:{}:layer:selenium:2

resources:
  Resources:
    ApiGatewayRestApi:
      Properties:
        BinaryMediaTypes:
          - "*/*"
Enter fullscreen mode Exit fullscreen mode

 Deploy Lambda Function

Go to /lambda directory



$ sls deploy
Enter fullscreen mode Exit fullscreen mode

 Output 



Serverless: Stack update finished...
Service Information
service: selenium-lambda
stage: dev
region: ap-southeast-2
stack: selenium-lambda-dev
api keys:
  None
endpoints:
  GET - https://{name}.execute-api.ap-southeast-2.amazonaws.com/dev/test
functions:
  main: selenium-lambda-dev-main
Enter fullscreen mode Exit fullscreen mode

output

You should get same response as below when accessing API



{
"statusCode": 200,
"body": "Selenium Headless Chrome Initialized"
}
Enter fullscreen mode Exit fullscreen mode

This deployment automatically creates cloudformation stack and s3 bucket.

Cloudformation stack

S3

Deprecation note:

Since python3.6 has been deprecated, you can try using the ff packages compatible with 3.9. note that if this exceeds lambda function zip file, you can try using docker image or uploading to s3 instead.



$ python3.9 -m pip install -t python/lib/python3.9/site-packages selenium==4.5.0
$ cd selenium
$ zip -r python.zip python/
Enter fullscreen mode Exit fullscreen mode

Name: selenium
Description: Selenium layer
Upload zip file created: python.zip
Compatible runtimes: Python 3.9



$ mkdir -p chromedriver
$ cd chromedriver
$ curl -SL https://chromedriver.storage.googleapis.com/107.0.5304.62/chromedriver_linux64.zip > chromedriver.zip
$ unzip chromedriver.zip
$ rm chromedriver.zip
Enter fullscreen mode Exit fullscreen mode 


$ curl -SL https://github.com/adieuadieu/serverless-chrome/releases/download/v1.0.0-57/stable-headless-chromium-amazonlinux-2.zip > headless-chromium.zip
$ unzip headless-chromium.zip
$ rm headless-chromium.zip
Enter fullscreen mode Exit fullscreen mode

References for new releases:
chromedriver release
https://chromedriver.chromium.org/downloads

serverless-chrome release
https://github.com/adieuadieu/serverless-chrome/releases

Checkout Achim's comment on how he used 3.9 using dockerimage
https://dev.to/achimgrolimund/comment/22d99

Jaira Encio Linkedin

Top comments (48)

Collapse
 
prenitwankhede profile image
prenit-wankhede •

Thanks a ton brother for simple yet elegant walk through.
I have been trying so many tutorials and ways to get it to work but no luck.

With selenium version, chromedriver version and headless-chrome version as mentioned in the post, finally got it working. Thanks a bunch !

Collapse
 
awolad profile image
Awolad Hossain •

@jairaencio It's working great. But I can't use the selenium-stealth plugin. Getting an error. Message: unknown error: Chrome failed to start: exited abnormally\n (Driver info: .....

from selenium import webdriver
from selenium.webdriver.chrome.options import Options
from selenium_stealth import stealth

def lambda_handler(event, context):
    options = Options()
    options.binary_location = '/opt/headless-chromium'    
    options.add_argument("start-maximized")
    options.add_experimental_option("excludeSwitches", ["enable-automation"])
    options.add_experimental_option('useAutomationExtension', False)

    driver = webdriver.Chrome('/opt/chromedriver',chrome_options=options)

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

    driver.get('https://quizlet.com/446134722/it-management-flash-cards/')

    driver.close();
    driver.quit();

    response = {
        "statusCode": 200,
        "body": "Selenium Headless Chrome Initialized"
    }

    return response
Enter fullscreen mode Exit fullscreen mode
Collapse
 
jairaencio profile image
Jaira Encio •

Hi @awolad I think you need to include the stealth library package in your lambda layer. Notice in my tutorial I have 2 different lambda layers for my selenium and chromedriver package. You can create another lambda layer or just simply include it in the 2 layers

Collapse
 
awolad profile image
Awolad Hossain •

@jairaencio Yes, I've added the stealth library package in the selenium lambda layer. There is no import error.

Thread Thread
 
jairaencio profile image
Jaira Encio •

Great! Always happy to help :)

Thread Thread
 
awolad profile image
Awolad Hossain •

@jairaencio Sorry, It's not solved yet. I mean the error is not related to the import the stealth package issue. Because the package is already in my lambda layer. The driver fails to load when the stealth package is used.

Thread Thread
 
jairaencio profile image
Jaira Encio • • Edited

does the error only occur when you add selenium-stealth library? Upon checking I noticed that others are experiencing issue in their local machines just by using stealth. You could try adding options.add_argument("--disable-blink-features=AutomationControlled") . Then try if it works both on your local and lambda.

Thread Thread
 
awolad profile image
Awolad Hossain •

Yes.

With the selenium-stealth default options like following:

options = Options()
options.binary_location = '/opt/headless-chromium'
options = webdriver.ChromeOptions()
options.add_argument("start-maximized")
# options.add_argument("--headless")
options.add_experimental_option("excludeSwitches", ["enable-automation"])
options.add_experimental_option('useAutomationExtension', False)

driver = webdriver.Chrome('/opt/chromedriver', chrome_options=options)

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

I'm getting error: Message: unknown error: Chrome failed to start: exited abnormally\n (Driver info: .....

By using this post options like following:

options = Options()
options.binary_location = '/opt/headless-chromium'
options.add_argument('--headless')
options.add_argument('--no-sandbox')
options.add_argument('--single-process')
options.add_argument('--disable-dev-shm-usage')

driver = webdriver.Chrome('/opt/chromedriver', chrome_options=options)

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

I'm getting error: "'WebDriver' object has no attribute 'execute_cdp_cmd'"

Thread Thread
 
jairaencio profile image
Jaira Encio •

I'm seeing this article related to "execute_cdp_cmd" error. Apparently they used pip install --pre selenium to be able to execute CDP commands github.com/SeleniumHQ/selenium/iss...

Thread Thread
 
awolad profile image
Awolad Hossain •

I also tried that but not working. I forgot to mention that. It would be helpful for us if you try with the selenium-stealth package and update this post. Because some websites we can't scrape without the selenium-stealth package. Thanks!

Collapse
 
da_shen_a7cf582bace0b4404 profile image
Da Shen •

tested working.. good article. note that Python runtime has to be 3.6. It won't work otherwise.

Collapse
 
achimgrolimund profile image
Achim Grolimund • • Edited

Hey @jairaencio
Hello everyone, thanks for the guide. but it seems to me that this method no longer works without a Dockerimage.

I keep getting the error:

{
  "errorMessage": "Message: Service /opt/chromedriver unexpectedly exited. Status code was: 127\n",
  "errorType": "WebDriverException",
  "requestId": "ef0d3b0d-ee3f-4822-ba11-9dd40920680b",
  "stackTrace": [
    "  File \"/var/task/app.py\", line 29, in main\n    driver = webdriver.Chrome(service=s, options=op)\n",
    "  File \"/opt/python/lib/python3.9/site-packages/selenium/webdriver/chrome/webdriver.py\", line 69, in __init__\n    super().__init__(DesiredCapabilities.CHROME['browserName'], \"goog\",\n",
    "  File \"/opt/python/lib/python3.9/site-packages/selenium/webdriver/chromium/webdriver.py\", line 89, in __init__\n    self.service.start()\n",
    "  File \"/opt/python/lib/python3.9/site-packages/selenium/webdriver/common/service.py\", line 98, in start\n    self.assert_process_still_running()\n",
    "  File \"/opt/python/lib/python3.9/site-packages/selenium/webdriver/common/service.py\", line 110, in assert_process_still_running\n    raise WebDriverException(\n"
  ]
}
Enter fullscreen mode Exit fullscreen mode

Python 3.9
Selenium 4.5.0
chromedriver 106.0.5249.61
headless-chromium v1.0.0-57

This is My Makefile to create all the stuff i need. At the end i do upload all what is inside the lambda Folder (Code + 2 Layers)

BOLD := \033[1m
NORMAL := \033[0m
GREEN := \033[1;32m

.DEFAULT_GOAL := help
HELP_TARGET_DEPTH ?= \#
.PHONY: help
help: # Show how to get started & what targets are available
    @printf "This is a list of all the make targets that you can run, e.g. $(BOLD)make dagger$(NORMAL) - or $(BOLD)m dagger$(NORMAL)\n\n"
    @awk -F':+ |$(HELP_TARGET_DEPTH)' '/^[0-9a-zA-Z._%-]+:+.+$(HELP_TARGET_DEPTH).+$$/ { printf "$(GREEN)%-20s\033[0m %s\n", $$1, $$3 }' $(MAKEFILE_LIST) | sort
    @echo


install: clean # Format all files (terraform fmt --recursive .)
    python3.9 -m pip install -t python/lib/python3.9/site-packages selenium==4.5.0 --upgrade
    python3.9 -m pip install -t python/lib/python3.9/site-packages wget --upgrade
    curl -SL https://chromedriver.storage.googleapis.com/106.0.5249.61/chromedriver_linux64.zip > chromedriver.zip
    curl -SL https://github.com/adieuadieu/serverless-chrome/releases/download/v1.0.0-57/stable-headless-chromium-amazonlinux-2.zip > headless-chromium.zip
    mkdir -p lambda
    zip -9 -r lambda/python.zip python
    unzip chromedriver.zip
    unzip headless-chromium.zip
    zip -9 -r lambda/chromedriver.zip chromedriver headless-chromium
    rm -f chromedriver.zip headless-chromium.zip chromedriver headless-chromium

build: # Create all layers to upload to the Lambda Function
    zip -9 -r lambda/app.zip app.py

clean:
    rm -rf lambda python
Enter fullscreen mode Exit fullscreen mode

And here is a overview of my python script:

import re
import wget

from selenium import webdriver
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.common.by import By
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.support.wait import WebDriverWait


def main(event, context):
    regex = r"[ \w-]+?(?=\.txt\.gz)"

    s = Service('/opt/chromedriver')
    op = webdriver.ChromeOptions()
    op.binary_location = '/opt/headless-chromium'
    op.add_argument('--headless')
    op.add_argument('--no-sandbox')
    op.add_argument('--disable-dev-shm-usage')
    op.add_argument('--disable-gpu')
    op.add_argument('--disable-dev-tools')
    op.add_argument("--disable-extensions")
    op.add_argument('--no-zygote')
    op.add_argument('--single-process')
    op.add_argument('--enable-logging')
    op.add_argument('--log-level=0')
    op.add_argument("--disable-notifications")
    op.add_argument('--v=99')
    driver = webdriver.Chrome(service=s, options=op)

    driver.get(
        "https://xxxxxx")

    (.....)

    driver.close()
Enter fullscreen mode Exit fullscreen mode

Is there another way to run it directly in a lambda without a docker image?

Best Regards

Collapse
 
jairaencio profile image
Jaira Encio •

Great take on this achin💪 I still havent gotten back on this but your method of using dockerimage is very useful for everyone as well 👏🏽

Collapse
 
roccolocko profile image
Rocco •

I have the configuration propose on the edited version of the article but I get the same error. I created all the version on an amazon linux using the python3.9 command like in the example but I keep getting the same error.

Did you find any solution?

Collapse
 
achimgrolimund profile image
Achim Grolimund •

Yes it needs an special tool inside the docker image. I wil post it here today as soon as im on my computer

Thread Thread
 
nadiaou profile image
Nadia-Ou •

did you find any solution, please?

Collapse
 
rajans163 profile image
rajans163 •

Hi Dear...were you able to get resolution of this issue. I am getting the same issue in my lambda.
Please share the resolution as I am completely stuck.

Collapse
 
quibski profile image
quibski •

Surely many Devs and QA will benefit from this. Hopefully a demo can be made/shown

Collapse
 
tchua profile image
tchua •

+1 to this, a demo would be great!

Collapse
 
jairaencio profile image
Jaira Encio •

uhm hahaha

Collapse
 
chris93007wq profile image
Christine John •

I'm getting the following error -
selenium.common.exceptions.WebDriverException: Message: unknown error: cannot find Chrome binary
(Driver info: chromedriver=2.37.544315(730aa6a5fdba159ac9f4c1e8cbc59bf1b5ce12b7),platform=Linux 4.14.255-276-224.499.amzn2.x86_64 x86_64)

Could someone help me please? :(

Collapse
 
awspipe profile image
awspipe •

Hi I am new to AWS lambda, so kindly apologize for any obvious questions...
I was able to create the 2 layers mentioned by you, for python and chromium.
But I have no idea how can I run serverless.yaml...
You've mentioned "sls deploy", but I don't have linux to run this.

Any other alternative to run this? Thanks

Collapse
 
jairaencio profile image
Jaira Encio •

I'm not a windows user but here is what I found codegrepper.com/code-examples/shel...

Collapse
 
chris93007wq profile image
Christine John •

@jairaencio

Collapse
 
jairaencio profile image
Jaira Encio •

Hi, this error would likely happen if the location for your chromedriver is incorrect. (lambda layer)

Collapse
 
sebifc profile image
sebifc •

Hi! I tried to run it in a Lambda function and the execution keeps running until it get the 600 seconds timeout.
I had downloaded the chromium 93 driver and the headless chrome 93 as well.

I'm using the following code:

from selenium import webdriver
from selenium.webdriver.chrome.options import Options
from webdriver_manager.chrome import ChromeDriverManager
from selenium_stealth import stealth

def lambda_handler(event, context):
    options = Options()
    options.binary_location = '/opt/headless-chromium'
    options.add_argument("start-maximized")
    #options.add_experimental_option("excludeSwitches", ["enable-automation"])
    #options.add_experimental_option('useAutomationExtension', False)
    #options.add_argument("--disable-blink-features=AutomationControlled")
    #options.add_argument('--headless')
    options.add_argument('--no-sandbox')
    options.add_argument('--single-process')
    options.add_argument('--disable-dev-shm-usage')
    options.add_argument('--disable-notifications')
    options.add_argument("--enable-javascript")

    driver = webdriver.Chrome('/opt/chromedriver',chrome_options=options)

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

    driver.get('https://trends.google.com/trends/trendingsearches/daily?geo=US')

    driver.close();
    driver.quit();

    response = {
        "statusCode": 200,
        "body": "Selenium Headless Chrome Initialized"
    }

    return response
Enter fullscreen mode Exit fullscreen mode
Collapse
 
jairaencio profile image
Jaira Encio •

I'm guessting that the timeout issue is caused by deployment package size. You would have to use container images solve it.

Collapse
 
chrisjeriel profile image
chrisjeriel •

Great work! A well-thought-out article, straightforward and concise. Looking forward more advanced implementations.

Collapse
 
cesarcastmore profile image
cesarcastmore • • Edited

Hello! .

I just have one question. When I was uploading my image to Lambda, I noticed that it required a lot of memory, and I think that could significantly increase costs. What are the differences between layers and Docker in Lambda?

I was following this documentation and managed to do it for version 3.7, but I was unsuccessful with version 3.9. Some of the comments below suggest using Docker, but I realized that using Docker requires increasing the Lambda's memory. Do you know of any alternatives to using Docker that won't consume a lot of memory and increase costs

Collapse
 
jairaencio profile image
Jaira Encio •

As of now there is nothing else other than docker to work with this. I dont recommend setting up an instance as well. We can only hope that aws improves their lambda pricing and memory.

Collapse
 
rajans163 profile image
rajans163 •

@jairaencio

I tried the same stuff for Python3.8(same chrome driver as you did for 3.9) but got the below error. Please help

START RequestId: ce98c274-1b51-46a5-968e-cdabe1e08a2a Version: $LATEST
[ERROR] WebDriverException: Message: Service /opt/chromedriver unexpectedly exited. Status code was: 127

Traceback (most recent call last):
  File "/var/task/lambda_function.py", line 13, in lambda_handler
    driver = webdriver.Chrome('/opt/chromedriver')
  File "/opt/python/selenium/webdriver/chrome/webdriver.py", line 69, in init
    super().init(DesiredCapabilities.CHROME['browserName'], "goog",
  File "/opt/python/selenium/webdriver/chromium/webdriver.py", line 89, in init
    self.service.start()
  File "/opt/python/selenium/webdriver/common/service.py", line 98, in start
    self.assert_process_still_running()
  File "/opt/python/selenium/webdriver/common/service.py", line 110, in assert_process_still_running
    raise WebDriverException(END RequestId: ce98c274-1b51-46a5-968e-cdabe1e08a2a
REPORT RequestId: ce98c274-1b51-46a5-968e-cdabe1e08a2a Duration: 610.08 ms Billed Duration: 611 ms Memory Size: 128 MB Max Memory Used: 47 MB Init Duration: 253.40 ms

Collapse
 
jairaencio profile image
Jaira Encio •

dev.to/achimgrolimund/comment/22d99 hi, you might want to follow this using dockerimage

View full discussion (48 comments)

Read next

manojchrsya profile image

How to Vertically Scale Your Amazon RDS Instance

Manoj Chaurasiya -

takatws profile image

OneOf Library in C#

Takat Wicaksono -

testifytech profile image

Types of Automation Testing: A Comprehensive Guide

Steve Wortham -

tkssharma profile image

Build standalone applications with Nestjs

tkssharma -