Sebastian Jimenez Solution

[sebasjimenez10]

Extract a Google paintings carousel → JSON

Parses the saved Google knowledge-graph paintings carousel into { "artworks": [...] } (name, link, image, optional extensions/date), directly from the local HTML. No extra HTTP requests.

Approach

• Structural parsing, not CSS classes. Items are found as stick= search links wrapping an <img>, since Google rotates its obfuscated class names (a class-based parser returns 0 items on a newer page).
• Images resolved offline from inline _setImagesSrc base64 (matched by img id) or the data-src thumbnail URL.
• Small, focused objects: Parser, ImageResolver, Artwork, SearchResult.

Validation

Tested against 139 artworks across 3 real pages, 100% line/branch coverage: van-gogh (exact match to the SerpApi-provided example), picasso and leonardo (generated, spot-verified snapshots + independent counts).

Usage

bin/extract files/van-gogh-paintings.html prints the JSON. bundle exec rake runs the specs + linter. bin/console opens an IRB session with the classes loaded. See INSTRUCTIONS.md.

Tooling

RSpec (unit + integration), Nokogiri, SimpleCov (100% line/branch), standardrb, Rake. See DESIGN.md for the full approach.