1 require 'open-uri'
 2 require 'iconv'
 3 require 'ftools'
 4 
 5 url = "http://anime.xunlei.com/Book/category/589"
 6 targetPrefix = "http://images.mh.xunlei.com/origin/"
 7 
 8 replaceToolFileName = "replace.rb"
 9 urlFileName = "urls.lst"
10 
11 class Volumn
12     attr_reader :name, :pageCount, :url
13     def initialize(name, pageCount, url)
14         @name = name
15         @pageCount = pageCount
16         @url = url
17     end
18 end
19 
20 
21 
22 volumnContent =  open(url).read
23 target_re = /<title>(.*)<\/title>/
24 m = target_re.match(volumnContent)
25 title = Iconv.conv("gbk","utf-8",m[1])
26 Dir.mkdir(title) if !File.directory?(title)
27 
28 target_re = /<a href="(http:\/\/images.anime.xunlei.com\/book\/segment\/\d+\/\d+.html)"\D+>(\S+)<\/a>\((\d+)\S+\)/
29 m = target_re.match(volumnContent)
30 vList = Array.new
31 while m != nil
32     v = Volumn.new(Iconv.conv("gbk", "utf-8", m[2]), m[3], m[1])
33     vList << v
34     puts v.name
35     m = target_re.match(m.post_match)
36 end
37 
38 vList.reverse
39 
40 for v in vList
41     # create volumn directory
42     puts "Creating volumn directory for [#{v.name}]..."
43     puts
44     volumnDir = "#{title}/#{v.name}"
45     Dir.mkdir(volumnDir) if !File.exist?(volumnDir)
46     
47     #copy replace.rb
48     puts "Copying replace tool to #{volumnDir}"
49     puts
50     File.copy(replaceToolFileName, "#{volumnDir}/#{replaceToolFileName}")
51     
52     #get image urls
53     puts "Retriving image urls from page for volumn [#{v.name}]"
54     puts
55     volumnPageHtml = open(v.url).read
56     target_re = /images_arr\[(\d+)\] = \'(\w+\/\w+\.[a-zA-Z]{3,4})\'/
57     m = target_re.match(volumnPageHtml)
58     #Output urls
59     puts "Outputing image urls for volumn [#{v.name}]"
60     file = File.new("#{volumnDir}/#{urlFileName}", 'w')
61     while m != nil
62         imgUrl = "#{targetPrefix}#{m[2]}"
63         file.puts imgUrl
64         m  = target_re.match(m.post_match)
65     end
66     file.close
67     puts
68     puts
69 end
70     
71