Scraping GitHub pull requests using Ruby

scraper.rb

# determine if a user has made a pull request (or several), returning a JSON payload either way

require 'nokogiri'
require 'open-uri'
require 'json'

student = ''    # 'iandouglas'
repo = 'turingschool/portfolios'

doc = Nokogiri::HTML(open("https://github.com/#{repo}/pulls"))

# important note: this does not handle pagination!

pull_requests = Array.new
next_state = 'pr_link'
pr_link = ''
doc.xpath('//li//div//a').each do |link|
  if next_state == 'pr_link' && link.attributes['href'].value.include?('/pull/')
    pr_link = link.attributes['href'].value
    next_state = 'github_username'
  elsif next_state == 'github_username'
    next_state = 'pr_link'
    if student.empty? || (link.children[0].content.strip.downcase == student.downcase)
      pull_requests << "https://github.com#{pr_link}"
    end
    pr_link = ''
  end
end

pr = {'pr_list' => pull_requests}.to_json
puts pr