Hey guys! You've probably used Amazon prime video to watch a TV show or movie. In this tutorial we’ll build our own automation tool that scrapes the homepage of Amazon Prime Video for a list of movie and TV show titles and ratings with BeautifulSoup. We’ll create our own database of content on the platform and then as a final example, visualize a word cloud with our dataset.

Here is a roadmap for this project:

  • Part 1: Set-up packages for web scraping
  • Part 2: Find CSS tags to input into Python code
  • Part 3: Write Python code to import packages, prepare variables and define functions
  • Part 4: Finish Python code to trigger functions and save data to CSV file
  • Part 5: Create a word cloud visualization to analyze differences in synopsis keywords between different groups of ratings

Before we begin, I want to mention that the tutorial below can be found in a video form on our website. You can find more free courses and projects on my website, The Codex to master Python by building projects. You can find all the code for this project at my GitHub Repo here.

Part 1: Set-up packages for web scraping

I am using the Python3 environment on Mac, however the majority of these steps will work on Windows as well.  I’m also using Visual Code Studio as my IDE or editor, but you can use other IDEs.

Instructions for Installing Chrome Driver

Separate from Python, you will need the following downloads installed:

  • Google Chrome internet browser
  • ChromeDriver

You can install ChromeDriver from the https://chromedriver.chromium.org site and choose the correct version based on your Chrome browser version.  You can find your Chrome browser version by going into Google Chrome > Help > About Google Chrome and find the version there.

A close up of a sign

Description automatically generated


Once you have ChromeDriver downloaded, unzip the folder and move the chromedriver file to /usr/local/bin.

Double-click to open the chromedriver file to initialize it.

Troubleshooting: You may run into an error message below.

A screenshot of a cell phone

Description automatically generated


To resolve this, go to System Preferences > Security & Privacy > General, then click the lock to make changes.

Click Open Anyway beside where it says “chromedriver” was blocked.

Then click Open to proceed.

A screenshot of a cell phone

Description automatically generated

Click the lock again to prevent further changes.

Go back to the bin folder and double-click chromedriver again to initiate.  This time a window will pop up with the last line of text saying:

ChromeDriver was started successfully.

You can then close the window.

Python Package Installation

Make sure the following Python packages are installed.

  • pandas - will allow us to save data to a csv, text, or excel file
  • selenium – will allow us to control a webpage using CSS tags
  • bs4 – will format and parse page source data neatly
  • matplotlib – creates and exports graphs and visuals
  • wordcloud – creates word cloud visualizations

You can install the packages with the pip installer by running the following example code in the Mac Terminal (find Terminal by pressing CMD + SPACE and typing “Terminal” in the search bar on your Mac):

python3 -m pip install pandas –user

When the install finishes you should receive a message similar to below if you ran the code above:

Successfully installed numpy-1.19.1 pandas-1.1.1 python-dateutil-2.8.1 pytz-2020.1

Part 2: Find CSS tags to input into python code

First, we need to direct Python to automatically find each link for each movie or show.  To do that, we will need to find the CSS tag that holds all the individual movie and show links.

Get Movie and Show Link Tags

The site I want to gather information from is:

https://www.amazon.com/gp/video/storefront/

  1. Open this link up in your Chrome browser.
  2. Press F12 to open the Developer Tools to reveal the CSS Code.
  3. Click the top left button in the Developer Tools to enable inspecting elements.
A close up of a sign

Description automatically generated
  1. Start clicking on a movie title.  Notice it will highlight certain lines of code on the right pane.  We are looking for something like below:
A close up of a sign

Description automatically generated


We know this class holds movie links because within the tag (in the <> brackets) there is a “href=” tag that specifies the correct website links in HTML/CSS language.  The link for the element we inspected is:

/gp/video/detail/B08DKPFFBC/ref=atv_hm_hom_1_c_Yzm1i7_brws_3_6


In this case, this is a partial link that is appended to the main website, which is

https://www.amazon.com

If we combine the two, we would get

https://www.amazon.com/gp/video/detail/B08DKPFFBC/ref=atv_hm_hom_1_c_Yzm1i7_brws_3_6

Now if we test this link in the browser, you will see it directs to the correct individual site for that particular movie or show where we can find all the details we need.  

A screenshot of a cell phone

Description automatically generated


All the titles will be set up like this with /gp/video/detail/ in the link which we will input in our code.

Get Title Details Tags

Now that we have the link tags ready, we’ll have to instruct Python on where exactly on the webpage we want Python to extract information from.  Looking at a particular movie or show, we want to extract the title, ratings, and synopsis.

A screenshot of a cell phone

Description automatically generated


Inspect each part the same way as the previous steps, but this time we are looking for (Yellow highlighted means we will use this in our Python code and Blue highlighted shows what will be recorded into our csv files):

Title: <h1 class="_1GTSsh _2Q73m9" data-automation-id="title">7500</h1>

Rating: <span class="XqYSS8 …="IMDb Rating 6.3">6.3</span></span>

Synopsis: <div class="_3qsVvm _1wxob_">When terrorists try to seize control of a Berlin-Paris flight, a soft-spoken young American co-pilot struggles to save the lives of the passengers and crew while forging a surprising connection with one of the hijackers.</div>

NOTE: Some class names require full string in the quotes and some only require the first part before the space.  You might have to test out the code to see which returns the correct data.

Part 3: Write python code to import packages, prepare variables and define functions

Now that we have our programs and CSS tags set up, it’s finally time to write some python code!

Let’s start off with importing the correct packages:

from selenium import webdriver
from bs4 import BeautifulSoup
import pandas as pd
import matplotlib.pyplot as plt
from wordcloud import WordCloud

Next we’ll define our variables:

#create empty lists to be used in functions later
vidLink = []
title = []
rating = []
synopsis = []
#define CSS variables
titleClass = "h1"
titleName = "_1GTSsh _2Q73m9"
ratingClass = "span"
ratingName = "XqYSS8 AtzNiv"
synopsisClass = "div"
synopsisName = "_3qsVvm _1wxob_"
#Store URLs and Substrings known to use in code
storeFrontURL = "https://www.amazon.com/gp/video/storefront/"
vidDtlSubStr = "/gp/video/detail/"

Next let’s define our function:

This function finds the specific text to extract based on the CSS tags specified

def scrapText(lst,classType,className):
findClass = soup.find_all(classType,class_=className)
if len(findClass) == 0:
lst.append(None) #helps fill in data that can't be found
else:
for n in findClass:
if className == ratingName:
lst.append(float(n.text[-3:])) #Extract and convert rating to float
else:
lst.append(n.text)

Part 4: Finish python code to trigger functions and save data to csv file

Next we’ll put in code that initializes a Chrome browser to be controlled by Python using the selenium packages and point it to the storefront website:

#Next initialize browser to be controlled by Python below
#DO NOT CLOSE new browser window that is opened until end of code run.
# Make sure chromedriver is installed beforehand
driver = webdriver.Chrome("/usr/local/bin/chromedriver")
driver.get(storeFrontURL)

Next we’ll find the URLs of each title:

#Find all URL links on website and append them in the vidLink list
elems = driver.find_elements_by_xpath("//a[@href]")
for elem in elems:
if vidDtlSubStr in elem.get_attribute("href"):
vidLink.append(elem.get_attribute("href"))
#remove duplicates in vidLink list
vidLink = list(dict.fromkeys(vidLink))

We will then start scraping the individual titles’ data. Note that depending on performance of system, you might want to decrease the range.

#start scraping the data for each link
for i in range(0,len(vidLink)): #control for performance
driver.get(vidLink[i])
content = driver.page_source
soup = BeautifulSoup(content)
scrapText(title,titleClass,titleName)
scrapText(rating,ratingClass,ratingName)
scrapText(synopsis,synopsisClass,synopsisName)

Lastly, we will print the data to a csv file named “PrimeVid.csv”:

#Save the data into neat columns in a csv file
a = {'Title':title,'Rating':rating, 'Synopsis':synopsis}
dfA = pd.DataFrame(a)
dfA.to_csv('PrimeVid.csv', index=False, encoding='utf-8')

Now you have your csv file containing the scraped data!

Text

Description automatically generated

Part 5: Create a word cloud visualization to analyze differences in synopsis keywords between different groups of ratings

Let’s take this a step further and analyze the data we gathered.  Here we will be creating a word cloud visualization for movie or show descriptions that fall below a rating of 6, in-between 6 and 7, and high ratings of 8 and above.

First let’s split up the data frame into these 3 ratings groupings.

#Creating the wordcloud by ratings groups
# First separte dataframe into 3 different ratings groups
dfBelow6 = df.loc[(df['Rating'] < 6)]
df67 = df.loc[(df['Rating'] >= 6) & (df['Rating'] < 8)]
dfAbove8 = df.loc[(df['Rating'] >= 8)]

Next, we’ll create the word clouds and export them to an image file.

#Then write a function to create a wordcloud for each ratings group
# Wordcloud idea from https://towardsdatascience.com/simple-wordcloud-in-python-2ae54a9f58e5
def wordcloud(dataframe,filename):
# Read the whole text.
text = ' '.join(dataframe.Synopsis)
# Generate a word cloud image
wordcloud = WordCloud().generate(text)
# Display the generated image:
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.savefig(filename + '.png')
wordcloud(dfBelow6, "Below6")
wordcloud(df67, "6-7")
wordcloud(dfAbove8, "Above8")

Now you have each word cloud exported as an image file!

Below 6:

Text

Description automatically generated

6-7:

Text

Description automatically generated

Above 8:

Text

Description automatically generated


An analysis we can make out of this is that lower-rate titles tend to focus on life, while mid and higher rated titles feature new releases most likely.


That's it folks! You just built an Amazon Prime Video Scraper with Selenium. You can find all the code for this project at our GitHub Repo here. As always, if you have face any troubles building this project, join our discord and TheCodex community can help!


For those of you interested in more project walkthroughs: Every Tuesday, I release a new Python/Data Science Project tutorial. I was honestly just tired of watching webcasted lectures and YouTube videos of instructors droning on with robotic voices teaching pure theory, so I started recording my own fun and practical projects.

Want to get notified every time a new project launches?

Subscribe to get Tinker Tuesday delivered to your inbox.

    No spam. Just 1 email / project. Unsubscribe at any time.