A Dockerfile is a set of instructions for Docker to build an image, much like a recipe for crafting a ready-to-use environment for an application. It outlines what software to include, how to configure it, and which files to add.

1. Install the necessary dependencies

FROM ubuntu:22.04 AS base
# Install basic dependencies
RUN apt-get update && \
   apt-get install -y \
   xvfb \
   ca-certificates \
   fonts-liberation \
   libasound2 \
   libatk-bridge2.0-0 \
   libatk1.0-0 \
   libatspi2.0-0 \
   libc6 \
   libcairo2 \
   libcups2 \
   libcurl4 \
   libdbus-1-3 \
   libdrm2 \
   libexpat1 \
   libgbm1 \
   libglib2.0-0 \
   libgtk-4-1 \
   libnspr4 \
   libnss3 \
   libpango-1.0-0 \
   libu2f-udev \
   libvulkan1 \
   libx11-6 \
   libxcb1 \
   libxcomposite1 \
   libxdamage1 \
   libxext6 \
   libxfixes3 \
   libxkbcommon0 \
   libxrandr2 \
   wget \
   xdg-utils \
   openjdk-18-jre-headless \
   jq \
   curl \
   unzip \
   openssh-client \
   python3 \
   python3-pip \
   libayatana-appindicator3-dev && \
   pip install requests selenium && \
   curl --location --fail --output mlxdeb.deb "https://mlxdists.s3.eu-west-3.amazonaws.com/mlx/1.15.0/multiloginx-amd64.deb" && \
   dpkg -i mlxdeb.deb && \
   apt-get clean && \
   rm -rf /var/lib/apt/lists/*

2. Set up a virtual display

Many applications need a graphical interface to interact with. This also applies to browsers and Multilogin. By default, containers don’t have a GUI, so we need to set one for headless automation. You don't need to install a conventional desktop environment like GNOME or KDE – a virtual one will do the trick. 

For this, you can use the Xvfb (X Virtual Frame Buffer) package. Xvfb emulates a display that doesn't rely on a physical screen or windowing system, which makes it lightweight and ideal for headless setups.

Run the commands below to set up a virtual display:

ENV DISPLAY=:99
RUN mkdir /tmp/.X11-unix
RUN chown root:root /tmp/.X11-unix && \
   chmod 1777 /tmp/.X11-unix

3. Create a user

RUN useradd -m mlx-user && \
   chown -R mlx-user:mlx-user .

Before running the entrypoint file, switch to this user by adding the line USER mlx-user.

4. Copy the complementary files and set up permissions for them

COPY entrypoint.sh /entrypoint.sh
RUN chmod +x /entrypoint.sh
COPY ./env.py /app/mlx-app/
COPY ./main.py /app/mlx-app/
COPY ./mlx_functions.py /app/mlx-app/

5. Define the entrypoint.sh file to be run once the container is started

ENTRYPOINT ["/entrypoint.sh"]

An entrypoint file is a script or executable that tells Docker how to start the application inside the container. They work together to ensure the application runs correctly and consistently in the container.

1. Start virtual display in the background
2. Start Multilogin X launcher in the background by using the shortcut “mlx“
3. Wait for a few seconds for the launcher to update
4. Change the directory to where the automation script is located
5. Start the automation script

#!/bin/bash

# Start virtual display
Xvfb :99 -screen 0 1024x768x24 &

# Start Multilogin X launcher
mlx &

# Wait for a while so the launcher is properly installed
sleep 20

# Run Python automation script
cd /app/mlx-app && python3 main.py

An automation script conducts some basic scraping tasks with Multilogin X headless browser profiles. Written in Python, it employs the Selenium WebDriver for task automation.

The script performs the following actions:

1. Go to a specific website
2. Print the website title
3. Print the number of articles present on the page
4. Print the name of each article
5. Conclude the automation process and terminate the driver session

def automation(driver): # Automation function

    try:
        driver.get("https://www.scrapethissite.com/pages/") # Go to this website
        print(f'The title of the webpage is: "{driver.title}"\n') # Print the title of the website
        articles = driver.find_elements(By.CLASS_NAME, "page")
        print(f'There are {len(articles)} different articles on this page:\n') # Print how many articles are there
        for i, article in enumerate(articles): # For each article, show us what is the name of the article
             article_text = article.find_element(By.TAG_NAME, "a").text
             print(f"{i+1}. {article_text}")
        print("\nBasic automation finished.\n") # Print that the automation is finished
            
    except Exception as e: # Catch expection if any
            print(f"Something happened: {e}") # If any exception, print its details
    finally:
        driver.quit() # Quit driver session
        mlx.stop_profile(quick_profile_id) # Close browser profile