Code challenge solution

[nurtoltor]

Solution

I built a small carousel scraper using Nokogiri and regex. It finds the best Knowledge Graph carousel container by looking for data-attrid sections and links that include stick=, then extracts fields for each item. I prioritised semantic HTML (role, aria-label, alt, title) over class names. It also skips "show more" items and only includes images already present in the HTML (data:image, encrypted-tbn, or knowledgecard icons).

The output is a hash where the key matches the search results selected tab (e.g., artworks, cast, albums). If no tab is selected, it defaults to results.

Structure

• lib/carousel_scraper.rb: Orchestrates the extraction and chooses the correct carousel scope.
• lib/carousel_item_extractor.rb: Extracts name, extensions, link, and image from a single item link.

I tested against 3 other result pages to find common patterns:

• "David Bowie albums" search: files/david-bowie-albums.html
• "George Orwell books" search: files/george-orwell-books.html
• "Lord of the Rings cast" search: files/lord-of-the-rings-cast.html

How to run

Install dependencies:

bundle install

Run with the default Van Gogh paintings HTML (outputs to files/van-gogh-paintings-expected-array.json):

ruby main.rb

Run with a specific HTML file (outputs JSON to the same directory):

ruby main.rb files/david-bowie-albums.html
ruby main.rb files/george-orwell-books.html
ruby main.rb files/lord-of-the-rings-cast.html

Run the tests:

bundle exec rspec