Xkcd_download_script.rb

require 'socket'
require 'cgi'
@host = "xkcd.com"
@port = 80
path = "/"
def logic(i,errors,max,table)
    s = true
    begin
        ssocket = TCPSocket.open(@host,@port)
        ssocket.print(request ("/"+i.to_s+"/"))
        response = ssocket.read
        headers,body = response.split("\r\n\r\n",2)
        img = body[(/cs\//=~body)+3..(/[gf]" t/=~body)]
        /"/=~body
        if img.count(">") > 0
            img = img[(/src="/=~img)+33..-1]
        end
        imgfmt = img[-3..-1]
        ft = true
        ctitle = body[(/ctitle/=~body)+8..(/comicNav/=~body)-19]
        table[i] = CGI.unescapeHTML ctitle
        ctitle = CGI.escapeHTML ctitle
        ctitle = ctitle.gsub(/\'/,"'")
        begin
            funnytext = CGI.unescapeHTML(body[(/[gf]" t/=~body)+10..(/"#{Regexp.escape ctitle}"/=~body)-7])
        rescue
            errors << [i,"Mouse Over Text Not Downloaded, please download manually"]
            ft = false
        end
        puts "Downloading #{img} to #{i}.#{imgfmt} (#{i}/#{max})"
        ssocket2 = TCPSocket.open("imgs.xkcd.com",80)
        ssocket2.print(request2("/comics/"+img))
        response = ssocket2.read
        headers,imgdata = response.split("\r\n\r\n",2)
        f = File.new("./imgs/"+i.to_s+"."+imgfmt,"w")
        f.syswrite(imgdata)
        f.close
        f = File.new("./txts/"+i.to_s+".txt","w")
        f.syswrite(funnytext)
        f.close
        s = ft
    rescue
        errors << [i,"Image NOT Downloaded, please download manually"]
        s = false
    end
    return s
end
def request(path)
    return "GET #{path} HTTP/1.0\r\n\r\n"
end
def request2(path)
    return "GET #{path} HTTP/1.1\r\nHost:imgs.xkcd.com\r\n\r\n"
end
if Dir["./imgs"] == []
    Dir.mkdir("imgs")
end
if Dir["./txts"] == []
    Dir.mkdir("txts")
end
socket = TCPSocket.open(@host,@port)
socket.print(request (path))
response = socket.read
headers,body = response.split("\r\n\r\n",2)
max = body[(/Permanent\ link\ to\ this\ comic:\ http:\/\/xkcd.com\//=~body)+46..(/\/\<\/h3\>$/=~body)-1].to_i
t = 0
continue = true
errors = []
s = ""
if File.exists?("strips.db")
    f = File.new("strips.db", "r+")
    f.each_line do |line|
        s += line
    end
else
    s = "[]"
end
numt=45
table = eval s
while t <= max/numt + numt+1
    threads = []
    numt.times do |i|
        threads << Thread.new do
            i = i + t*numt + 1
            if i <= max
                    if !File.exists?("./txts/"+i.to_s+".txt") && i != 404
                        logic(i,errors,max,table)
                    end
            end
        end
    end
    threads.each{|a| a.join}
    t+=1
end
errors.each do |arr|
    v = []
    fixed = logic(arr[0],v,max,table)
    if !fixed
        puts "Error in downloading #{v[0][0]}: #{v[0][1]}"
    end
end
table[0]=max
f = File.new("strips.db", "w+")
f.syswrite(table.inspect)
f.close