cbi_scraper.rb

# CBI Scraper (Offline)
# It scraps comic book issues from the website - comicsbackissues.com - and print them to a single text file

require 'oga'
require 'pry'

FILENAME = "cbi_batman.html"

class ComicBook
  attr_accessor :year, :title, :issue_no, :writer, :penciller, :storyline

  def initialize(year = 0, title = "", issue_no = 1986, writer = "", penciller = "", storyline = "")
    @year = year
    @title = title
    @issue_no = issue_no
    @writer = writer
    @penciller = penciller
    @storyline = storyline
  end

  def to_a_single_line_string
    "#{title} ##{issue_no} (#{year}) (#{writer} & #{penciller})"
  end
end

class CBIParser
  def self.parse(filename, range_from = 0)
    comicbook_list = []

    handle = File.open(filename)
    document = Oga.parse_html(handle)
    rows = document.css("tr")

    # :year, :title, :issue_no, :writer, :penciller, :storyline

    rows.each_with_index do |row, row_index|
      if row_index > range_from
        new_comicbook = ComicBook.new

        row.children.each_with_index do |field, field_index|
          if field_index > 0 && field_index < 6
            if field_index == 1
              new_comicbook.year = field.text.to_i
            elsif field_index == 2
              new_comicbook.title = field.text
            elsif field_index == 3
              new_comicbook.issue_no = field.text.to_i
            elsif field_index == 4
              new_comicbook.writer = field.text.split(', ').first
              new_comicbook.penciller = field.text.split(', ').last
            else field_index == 5
              new_comicbook.storyline = field.text
            end
          end
        end

        comicbook_list << new_comicbook
      end
    end

    comicbook_list
  end
end

list = CBIParser.parse(FILENAME, 2071)
outf = File.new("output.txt", "w")

list.each do |issue|
  outf.puts issue.to_a_single_line_string
end

outf.close