tenseijingo

#!/usr/bin/env ruby
# coding: utf-8
# desc: get today's Tenseijingo column
# author: eric
require 'date'
require 'moji'

Encoding.default_internal = "UTF-8"

# variables
file = Dir.home + '/.sync/diudiudui/tensei/tensei.txt'

today = Date.today
yesterday = today.prev_day

cache_dir = Dir.home + '/.sync/diudiudui/tensei/cache/'
cache_file = cache_dir + "tensei-#{today.strftime '%y%m%d'}.txt"
cache_file_yesterday = cache_dir + 
                       "tensei-#{yesterday.strftime '%y%m%d'}.txt"

title_date = Moji.han_to_zen "#{today.year}年　#{today.mon}月#{today.day}日"

page_url = 'http://www.asahi.com/paper/column.html'
element = '#MainInner .ArticleText'

# main
puts "tenseijingo downloader"

if File.exist? cache_file
  puts "today's job is already done."
  exit
end

# get today's content
require 'nokogiri'
require 'open-uri'

begin
  puts "fetching web content..."
  page = Nokogiri::HTML open(page_url, 'r:utf-8')
rescue
  p $!
  puts "error occured, exit."
  exit
else
  puts "content got"
end

content = page.css(element).text.strip
success_message = "column is downloaded."

# check newspaper holiday
if File.exist? cache_file_yesterday
  content_yesterday = File.read( cache_file_yesterday, 
                                 { encoding: 'utf-8' }
                               ).chomp

  if content == content_yesterday
    content = "【休刊日】"
    success_message = "oops, newspaper holiday :)"
  end
end

# write cache
File.open(cache_file, 'w:utf-8') { |f| f.puts content }

# add to file
File.open file, 'a:utf-8' do |f|
  f.puts
  f.puts "［＃５字下げ］#{title_date}［＃「#{title_date}」は中見出し］"
  f.puts
  f.puts content
end

# done
puts success_message