OSDN Git Service

0a88e261c3139e01860c70b6fd9a799b9203e014
[tdcgexplorer/nimono-crawlers.git] / bin / tap-idx.rb
1 #!/usr/bin/ruby
2 # download from 3ch pose
3 # http://www.esc-j.net/tech-arts/pose/
4
5 $LOAD_PATH.unshift File.expand_path(File.dirname(__FILE__) + '/../lib')
6 require 'sn_uploader'
7
8 def each_filename(text)
9   file_href = 'http://www.esc-j.net/tech-arts/pose/src/'
10   file_re = Regexp.new(Regexp.escape(file_href) + '(.+?)\.html')
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/pose"
21 uploader.base_html = ""
22 uploader.local_dir = File.dirname(__FILE__) + '/../arc/tap'
23 body = uploader.get_base
24 each_filename(body) do |filename|
25   exist = uploader.local_file_exist?(filename)
26   puts [ filename, exist ? 'exist' : 'download' ].join("\t")
27   uploader.download_file(filename) unless exist
28 end