Fetch LDRFullFeed JSON and fix type

get_ldrfullfeed_json.rb

# -*- coding: utf-8 -*-
require 'open-uri'
require 'json'

LDR_FULL_FEED_URL = 'http://wedata.net/databases/LDRFullFeed/items_all.json'
OUTFILE = 'ldrfullfeed.json'

REPLACE = {
  '^http://matome\.naver\.jp/odai/[0-9]+/[0-9]+' => 'SUB',
  '^http://matome\.naver\.jp/' => 'GEN',
}

puts "Fetching siteinfo from #{LDR_FULL_FEED_URL}"
siteinfo = JSON::parse(open(LDR_FULL_FEED_URL).read)

REPLACE.each { |target_url, type|
  candidate = siteinfo.select { |info| info['data']['url'] == target_url }
  raise "検索結果が 1 件でない" unless candidate.size == 1
  puts "Replacing: #{target_url}: #{candidate[0]['data']['type']} to #{type}"
  candidate[0]['data']['type'] = type
}

open(OUTFILE, 'w') { |f| f.write JSON::dump(siteinfo) }
puts "Wrote to #{OUTFILE}"