Suruga System Utility Script - Project NA00S3

Gemfile

source :rubygems

gem 'nokogiri'
# gem 'pry'

group :build do
  gem 'maliq'
  gem 'liquid'
end

my-own-purpose-script.rb

# -*- coding: utf-8 -*-
# Licensed under MIT
require 'nokogiri'
# require 'pry'
require 'open-uri'
require 'fileutils'
require 'erb'
require 'date'

module SurugaUtils
  class NA00S3
    include Nokogiri::XML
    def initialize(uri="http://nareru-se.dengeki.com/webnovel/?pagenum=1", index=1)
      source = open(uri).read
      @doc = Nokogiri::HTML(source)
      @index = index
    end
    attr_reader :doc, :index

    def to_readable_paragraphs
      main_nodes.
        chunk{|node| (node.is_a? Element and node.name == 'br') ? nil : true }.
        map(&:last).
        map{|parts| parts.map{|node| node.text.strip }.join }
    end

    def generate_build_files(dir='./', file_name="nareru-se-%02d.md")
      unless File.directory? dir
        FileUtils.mkdir dir
      end
      file = File.join dir, (file_name % index)
      File.open(file, 'w') do |file|
        body = to_readable_paragraphs.join("\n\n")
        file.write ERB.new(TEMPLATE).result binding
      end
      STDERR.puts "generated #{file}"
    end

    def main_nodes
      doc.css("#main_frame .novel > p").children
    end

    def title
      doc.css("#main_frame > #left_frame > img").map{|img| img.attributes["alt"].value }.max_by(&:length)
    end
  end

  TEMPLATE = <<-EOT.gsub(/^ {4}/, '')
    ---
    language: 'ja'
    unique_identifier:
     - 'http://nareru-se.dengeki.com/'
    title: 'ドラマCD発売記念特別短編「立華ズ・ブートキャンプ」'
    subtitle: '萌えるSE残酷物語'
    creator: '夏海公司'
    date: '<%= Date.today.to_s %>'
    ---
    # <%= title %> - <%= index %>

    ## chapter <%= index %>

    <%= body %>
  EOT
end

Rakefile

# Licensed under MIT
PAGES = 1..7
FILE_NAME = "nareru-se-%02d.md"
MD_FILES = PAGES.map {|index| FILE_NAME % index }
XHTML_FILES = MD_FILES.map {|f| f.sub '.md', '.xhtml' }

file 'nareru.epub' => (XHTML_FILES + ["images/cover.png", "css/style.css"]) do
  sh 'maliq_gepub -o nareru.epub'
end

rule '.xhtml' => '.md' do
  sh 'maliq *.md'
end

file 'images/cover.png' do
  mkdir_p "images"
  sh "wget http://nareru-se.dengeki.com/common/img/top/bg_kv_top.png -O images/cover.png"
end

file 'css/style.css' do
  mkdir_p "css"
  open('css/style.css', 'w') do |f|
    f.write <<-EOCSS
h1, h2, h3 {
  color: #8B1A1A;
}

ol {
  list-style-type: none;
}
    EOCSS
  end
end

rule '.md' do
  require './my-own-purpose-script'
  PAGES.each do |index|
    uri = "http://nareru-se.dengeki.com/webnovel/?pagenum=#{index}"
    SurugaUtils::NA00S3.new(uri, index).generate_build_files("./", FILE_NAME)
  end
end

desc "Scrape web contents and generate epub"
task :default => 'nareru.epub'

desc "Cleanup files"
task :clean do
  (MD_FILES + XHTML_FILES + %w(nareru.epub images css)).each {|path| rm_rf path }
end