bin/3ch-idx.rb

   1 #!/usr/bin/ruby
   2 # download from 3ch
   3 # http://www.esc-j.net/tech-arts/ta3dc/t1a931d9c1s9.html
   4
   5 $LOAD_PATH.unshift File.expand_path(File.dirname(__FILE__) + '/../lib')
   6 require 'sn_uploader'
   7
   8 def each_fileno(text)
   9   file_href = 'http://www.esc-j.net/tech-arts/ta3dc/upload.cgi?mode=dl&file='
  10   file_re = Regexp.new(Regexp.escape(file_href) + '(\d+)')
  11   text.scan(/href="(.+?)"/) do |href, |
  12     if md = file_re.match(href)
  13       yield md[1]
  14     end
  15   end
  16 end
  17
  18 uploader = SnUploader.new
  19 uploader.host = "www.esc-j.net"
  20 uploader.root_path = "/tech-arts/ta3dc"
  21 uploader.base_html = "t1a931d9c1s9.html"
  22 uploader.kcode = 'U'
  23 uploader.authorization = "Basic " + ["tech:mybride"].pack('m').chomp
  24 uploader.local_dir = '/Volumes/uploader/arc/3ch'
  25 body = uploader.get_base
  26 each_fileno(body) do |fileno|
  27   basename_without_extension = "TA3CH%04d" % fileno.to_i
  28   filename_re = Regexp.new('\A' + Regexp.escape(basename_without_extension) + '\.')
  29   exist = uploader.local_file_match?(filename_re)
  30   puts [ fileno, exist ? 'exist' : 'download' ].join("\t")
  31   uploader.download(fileno) unless exist
  32 end