#
# Author: IKARASHI Yoshinori <yoosee@gmail.com>
# License: GPLv2
# $Id: amazon-a2s.rb 4 2008-03-25 01:55:23Z yoosee $
#

=begin

This module wraps Amazon A2S (aka ECS4.0). 
About Amazon A2S, see
  http://www.amazon.com/gp/browse.html?node=3435361
  http://www.ajaxtower.jp/ecs/responsegroup/index.html (japanese)

Sample:
  aws = AmazonA2S.new
  item = aws.lookup "B0012M1BRG"
  puts "<a href='#{item.uri}'><img src='#{item.image_uri}' />#{item.title}</a>"

  items = aws.search "Ruby"
  items.each do |item|
    ...

=end

require 'open-uri'
require 'uri'
require 'timeout'
require 'net/http'
require 'zlib'
require 'rexml/document'
require 'fileutils'
require 'md5'

class AmazonItem
  attr_accessor :title, :uri, :image_uri, :manufacture, :author, :rexml
end

class AmazonA2S
  attr_accessor :associate_tag, :cache_dir, :cache_life, :xml
  AwsParameters = {
    :AWSAccessKeyId  =>  "1X7MZVP4YBE35YP64PG2",
    :AssociateTag    =>  "yooseenet-22",
    :Operation       =>  "ItemSearch",
    :SearchIndex     =>  nil,
    :Keywords        =>  nil,
    :ItemId          =>  nil,
    :ResponseGroup   =>  "Request,Small,Images", # default
    :ItemPage        =>  1,
    :Version         =>  "2007-10-29",
  }
  Host = "webservices.amazon.co.jp"
  URIBase="http://" + Host + "/onca/xml?Service=AWSECommerceService"

  def initialize
    @associate_tag = nil
    @cache_dir = "/tmp/amazon/"
    @cache_life = 3600 * 6 # 6 hours
  end

  def cache uri
    digest = Digest::MD5.hexdigest(uri)
    file = @cache_dir + "/" + digest
    if !File.exist? file || (File.mtime file - Time.now) > @cache_life
      return nil
    else
      page = ''
      File.open(file) do |f|
        page = f.read
      end
      return page
    end
    return nil
  end

  def write_cache uri, page
    digest = Digest::MD5.hexdigest(uri)
    FileUtils.mkdir_p @cache_dir unless File.directory? @cache_dir
    file = (@cache_dir + "/" + digest).gsub(/\/+/, '/')
    begin
      File.open file, 'w' do |f|
        f.puts page
      end
    rescue
    end
    return page
  end

  def fetch uri
    opts = {
      "Accept-Encoding"=>"gzip, deflate",
      "User-Agent"=>'Mozilla/6.0 AWSBot'
    }
    page, meta, content_type, status, message = nil, nil, nil, nil, nil
    begin
      page = timeout(10) {
        URI.parse(uri).open(opts) {|f|
          content_type = f.content_type
          f.read
        }
      }
    rescue OpenURI::HTTPError
      STDERR.puts "HTTP Error #{$!}"
      return ''
    rescue StandardError, TimeoutError
      STDERR.puts "Error #{$!}"
      return ''
    end

    if /\A\x1f\x8b/ =~ page
      begin
        page = Zlib::GzipReader.new(StringIO.new(page)).read || ''
      rescue Zlib::Error
        page = '(zlib decode error)'
      end
    end
    return page
  end

  def build_uri aws_parameters
    uri = URIBase
    aws_parameters[:AssociateTag] = @associate_tag if @associate_tag
    aws_parameters.each do |key, value|
      uri += "&" + key.to_s + "=" + value.to_s if value
    end
    return uri
  end

  def item_search keyword, category="Blended", response_group = "Request,Small,Images"
    aws_parameters = AwsParameters
    aws_parameters[:Operation] = "ItemSearch"
    aws_parameters[:Keywords] = URI.encode(keyword)
    aws_parameters[:SearchIndex] = category
    aws_parameters[:ResponseGroup] = response_group
    uri = build_uri aws_parameters
    puts uri
    if (page = cache uri)
      return page
    end
    page = fetch uri
    write_cache uri, page
    return page
  end

  def item_lookup asin
    aws_parameters = AwsParameters
    aws_parameters[:Operation] = "ItemLookup"
    aws_parameters[:ItemId] = asin
    uri = build_uri aws_parameters
    if (page = cache uri)
      return page
    end
    page = fetch uri
    write_cache uri, page
    return page
  end

  def items key, type=:lookup # or search
    items = Array.new
    case type
    when :search
      page = item_search key
    when :lookup
      page = item_lookup key
    else
      return
    end

    @xml = page
    rexml = REXML::Document.new page
    rexml.elements.each("*/Items/Item") do |item|
      amazonitem = AmazonItem.new
      amazonitem.rexml = item
      begin
        amazonitem.title = item.elements["ItemAttributes/Title"].text
        amazonitem.uri = item.elements["DetailPageURL"].text
        amazonitem.image_uri = item.elements["MediumImage/URL"].text if item.elements["MediumImage/URL"]
        if item.elements["ItemAttributes/Actor"]
          amazonitem.author = item.elements["ItemAttributes/Actor"].text
        end
        items.push amazonitem
      rescue
        puts "#{$!}:#{$@}"
      end
    end
    return items
  end

  def lookup asin
    return items(asin, :lookup).first
  end

  def search keyword
    return items(keyword, :search)
  end
end

if __FILE__ == $0
  aws = AmazonA2S.new

  asin = "B0012M1BRG"
  item = aws.lookup asin
  puts "<a href='#{item.uri}'><img src='#{item.image_uri}' />#{item.title}</a>"

  items = aws.search "ぼのぼの"
  items.each do |item|
    puts "<a href='#{item.uri}'><img src='#{item.image_uri}' />#{item.title}</a>"
  end
end