tumblrから画像ぶっこ抜くけど、もうあるやつはぶっこ抜かないやつ

tumblr_photo_backup.rb

#!/usr/bin/env ruby
# -*- coding: utf-8 -*-
#
# Tumblr photos backup
#
#
require 'rubygems'
require 'json'
require 'open-uri'
require 'fileutils'
require 'logger'

log = Logger.new(STDOUT)
log.level = Logger::INFO

abort "Usage: ruby #{$0} ACCOUNT OUTDIR [OFFSET]" if ARGV.size < 2

user   =  ARGV.shift
outdir =  ARGV.shift
offset =  ARGV.shift || 0

unless FileTest.exist? outdir
  FileUtils.mkdir_p outdir
end

REGEXP_SAVED_FILE = /tumblr\d+_(\w+).\w+/
saved_files = Dir.glob("#{outdir}/*.*").select{|e| REGEXP_SAVED_FILE =~ e}
saved_keys = saved_files.map{|e| e.match(REGEXP_SAVED_FILE)[1] }

url = "http://#{user}.tumblr.com/api/read/json"
query = {:debug => 1, :type => 'photo', :num => 50}
start = offset.to_i

while(true)
  query[:start] = start
  uri = url + '?' + query.map{|k,v| "#{URI.encode(k.to_s)}=#{URI.encode(v.to_s)}"}.join("&")
  json = JSON.load(open(uri))
  total = json['posts-total'].to_i
  posts = json["posts"]
  posts.each_with_index{|e,i| e[:i] = total - start - i}
  posts.delete_if{|e| saved_keys.include? e["reblog-key"]}

  if posts.empty?
    log.info("finish!")
    break
  else
    log.info("getting posts (%d/%d) ..." % [total - start - posts.size, total])
  end

  posts.each{|post|
    n_retry = 0
    type = 1280
    begin
      key  = post["reblog-key"]
      src = post["photo-url-#{type}"]
      dst = "#{outdir}/tumblr#{post[:i]}_#{key}#{File.extname(src)}"
      log.debug("open #{src} / reblog-key:#{key}")
      open(dst, "wb"){|f|
        open(src){|data|
          f.write(data.read)
        }
      }
      log.info("saved " + dst)
    rescue
      if n_retry < 2
        # change photo-url-1280 -> 500
        type = 500
        if n_retry == 1
          log.warn("WATING FOR RETRY...");
          sleep(10)
        end
        n_retry += 1
        retry
      end
      log.error "NOT_SAVED: " + post["photo-url-500"]
    end
  }
  start += query[:num]
end