Grab the list of LinkedIn industries.

linkedin_industries.rb

# Updated version of the following gist:
# https://gist.github.com/stevenzeiler/1895583
#
# list = Linkedin::IndustryList.new
# industry = list.industries.first
#
# industry.description
# => "Accounting"
#
# industry.code
# => 47
#
# industry.groups
# => ["corp", "fin"]

require 'nokogiri'
require 'open-uri'

module Linkedin
  class IndustryList
    attr_reader :industries

    def initialize
      @industries = scrape_linkedin
    end

    def scrape_linkedin
      industries = []
      doc = Nokogiri::HTML open('https://developer.linkedin.com/docs/reference/industry-codes')
      doc.search('.resourceTable tr').each do |tr|
        industry = Industry.new(tr.children)
        industries << industry if industry.valid?
      end

      industries
    end
  end

  class Industry
    attr_reader :code, :groups, :description

    def initialize(tds)
      return if ['Code', '', ' ', nil].include? tds.first.text

      @code = tds[0].text.to_i
      @groups = tds[2].text.split(', ')
      @description = tds[4].text
      @valid = true
    end

    def valid?
      @valid
    end
  end
end