load File.dirname(__FILE__) + '/net/http.rb'
load File.dirname(__FILE__) + '/net/response.rb'

module HttpCrawler
  class HTTP < Net::HTTP

    # 自动获取代理，true 表示自动获取代理 、false 表示不自动获取
    attr_accessor :auto_proxy
    # 代理API的别名 主要关联 HttpCrawler::Proxy中维护的代理API
    attr_accessor :proxy_api
    # 调用自己的代理池所需要的主键 key
    attr_accessor :proxy_key
    # 请求错误后的重复最大请求次数
    attr_accessor :max_error_num

    def initialize(address, port = nil)
      super(address, port)
      @max_error_num = 2
      @error_num = 0
      @proxy_key = "default"
    end

    def http_error_sleep
      sleep(0.5)
    end

    def server_error_sleep
      sleep(3)
    end

    def proxy_api
      @proxy_api ||= "my"
    end

    @@proxy_list = []
    # 为 @http 重设代理
    def proxy(p = {})

      raise '代理设置 p_addr 不能为空' unless p["p_addr"]
      raise '代理设置 p_port 不能为空' unless p["p_port"]

      p["p_user"] ||= nil
      p["p_pass"] ||= nil

      Rails.logger.info("切换代理至 => #{p}")
      # 设为 false 否则不会启用代理
      @proxy_from_env = false

      # 初始化代理数据
      @proxy_address = p["p_addr"]
      @proxy_port = p["p_port"]
      @proxy_user = p["p_user"]
      @proxy_pass = p["p_pass"]

    end

    # 通过调用 api 获取代理或者通过自定义设置代理
    def get_proxy

      while @@proxy_list.blank?
        Rails.logger.debug("@@proxy_list 为空进行更新")
        proxy_client = HttpCrawler::Proxy.for(proxy_api)
        proxy_r = proxy_client.get_proxy(key: proxy_key)
        @@proxy_list << proxy_r.parsing
        Rails.logger.debug("@@proxy_list => #{@@proxy_list}")
        sleep(1)
      end

      p = @@proxy_list.delete_at(0)

      Rails.logger.debug("当前IP => #{@proxy_address}:#{@proxy_port},获取最新代理 => #{p}")

      unless p && p["p_addr"] && p["p_port"]
        Rails.logger.warn "无最新代理等待5秒后重新获取"
        sleep(5)
        p = get_proxy
      end

      if (@proxy_address == p["p_addr"] && @proxy_port == p["p_port"])
        Rails.logger.warn "无最新代理等待5秒后重新获取"
        sleep(5)
        p = get_proxy
      end
      p
    end

    def update_proxy(p = {})
      if p.blank?
        proxy(get_proxy)
      else
        proxy(p)
      end
    end

    # 如果自动更新代理 则更新代理返回 true，否则返回false
    def update_proxy?(p = {})
      if auto_proxy
        if p.blank?
          proxy(get_proxy)
        else
          proxy(p)
        end
        return true
      else
        return false
      end
    end


    # 重定向请求
    def get_fetch(uri_or_path, initheader = nil, dest = nil, limit = 10, &block)
      # You should choose a better exception.
      raise ArgumentError, 'too many HTTP repeated' if limit == 0
      # 更新uri_or_path
      uri_or_path = URI.encode(uri_or_path) if String === uri_or_path && CharDet.detect(uri_or_path)["encoding"] != "ascii"

      response = get(uri_or_path, initheader, dest, &block)
      case response
      when Net::HTTPSuccess then
        response
      when Net::HTTPRedirection then
        location = response['location']
        Rails.logger.warn "redirected to #{location}"
        # 传入 location 进行跳转
        get_fetch(location, initheader, dest, limit - 1, &block)
      when Net::HTTPServerError then
        Rails.logger.warn "Net::HTTPServerError  5XX to #{address}"
        server_error_sleep
        # 重新请求
        get_fetch(uri_or_path, initheader, dest, &block)
      when Net::HTTPProxyAuthenticationRequired then
        Rails.logger.warn "Net::HTTPProxyAuthenticationRequired 407 to proxy:[#{@proxy_address}:#{@proxy_port}]  =>#{address}"
        if update_proxy?
          server_error_sleep
          # 重新请求
          get_fetch(uri_or_path, initheader, dest, &block)
        else
          response.error!
        end
      else
        server_error_sleep
        response.error!
      end
    end

    # 重定向请求
    def post_fetch(uri_or_path, data, initheader = nil, dest = nil, &block)
      # 更新uri_or_path 如果 uri_or_path 是 String类型 同时 又不是 ascii编码格式就进行转码
      uri_or_path = URI.encode(uri_or_path) if String === uri_or_path && CharDet.detect(uri_or_path)["encoding"] != "ascii"
      # Rails.logger.debug "post_fetch => #{uri_or_path}"
      response = post(uri_or_path, data, initheader, dest, &block)
      case response
      when Net::HTTPSuccess then
        response
      when Net::HTTPRedirection then
        location = response['location']
        Rails.logger.warn "redirected to #{location}"
        # 传入 location 进行跳转
        get_fetch(location, initheader, dest, 9, &block)
      when Net::HTTPServerError then
        Rails.logger.warn "Net::HTTPServerError 5XX to #{address}"
        server_error_sleep
        # 重新请求
        post_fetch(uri_or_path, initheader, dest, &block)
      when Net::HTTPProxyAuthenticationRequired then
        Rails.logger.warn "Net::HTTPProxyAuthenticationRequired 407 to proxy:[#{http.proxy_address}:#{http.proxy_port}]  =>#{address}"
        if update_proxy?
          server_error_sleep
          # 重新请求
          post_fetch(uri_or_path, initheader, dest, &block)
        else
          response.error!
        end
      else
        server_error_sleep
        response.error!
      end
    end

    # def post_fetch

    #
    # 重写 发送请求的方法
    #
    def request(req, body = nil, &block)
      begin
        Rails.logger.debug("#{req.class} => #{use_ssl? ? "https://" : "http://" }#{address}:#{port}#{req.path}") if started?
        Rails.logger.debug("body => #{body}") if started? && body
        super(req, body, &block)
      rescue => error
        if started?
          # started? 是为了判断是否结束http请求，如果不添加则会处理2次异常
          raise error
        else
          # 最大错误尝试次数
          if @error_num < @max_error_num
            @error_num += 1
            http_error_sleep
            retry # 这将把控制移到 begin 的开头
          else
            # 超过最大错误限制 判断错误类型
            case error
            when Net::HTTPFatalError
              raise error
            when EOFError
              Rails.logger.warn "EOFError!"
              if update_proxy?
                proxy(get_proxy)
                http_error_sleep
                retry # 这将把控制移到 begin 的开头
              else
                raise error
              end
            when Timeout::Error
              Rails.logger.warn "请求超时!"
              if update_proxy?
                @error_num = 0
                http_error_sleep
                retry # 这将把控制移到 begin 的开头
              else
                raise error
              end
            else
              raise error
            end
          end
        end
      end # begin
    end # def request(req, body = nil, &block)
  end
end