scrape_alma_form.rb

# frozen_string_literal: true

require 'nokogiri'
require 'csv'

# Get Alma publishing job data enhancement schema.
# Download data enrichment html and point this script to it.

# TODO: clean this up and refine it, if needed again

output = CSV.open('publishing_job_text.txt', 'w')

doc = File.open('Publishing Profile Details.html') { |f| Nokogiri::HTML(f) }

data = doc.search('form').map do |form|
  profile_name_label = form.xpath('./div[3]/div/div[3]/div/div[2]/div/div[1]/div/label/span')
  profile_name = profile_name_label.xpath('../../div').text
  profileid_label = form.xpath('./div[3]/div/div[3]/div/div[2]/div/div[2]/div/label/span[1]')
  profileid = profileid_label.xpath("../../span[@id='pageBeanpublishingProfileid']").text

  next unless profile_name_label.text.chomp(' ') != ''

  output << [
    profile_name_label.text.chomp(' '),
    profile_name.strip
  ]

  output << [
    profileid_label.text,
    profileid
  ]

  form.xpath(".//div[@class='marSidesAuto panel panel-default section  ']").each do |section|
    title = section.xpath(".//span[@class='sectionTitle']")

    output << ['-' * 35]
    output << [title.text.upcase]
    output << ['-' * 35]
    section.xpath('.//label/span').each do |label|
      value = label.xpath('../../div/input/@value').text

      label = label.text.to_s
      value = value
      next unless label.strip.chomp(' ') != ''

      output << [
        label.strip.chomp(' '),
        value.strip.chomp
      ]
    end
  end
end