require 'rubygems'
require 'net/http'
require 'cgi'

class Twitter
  TWITTER_SEARCH_API_URL = 'http://search.twitter.com/search.atom'
  DEFAULT_TIMEOUT = 5
  HEADERS = { "Content-Type" => 'application/rss+xml',
              "User-Agent"   => 'twitter-search' }

  def self.search(opts)
    search_for = '#twitter'

    url = URI.parse(TWITTER_SEARCH_API_URL)
    url.query = sanitize_query(opts)

    ensure_no_location_operators(url.query)

    req  = Net::HTTP::Get.new(url.path)
    http = Net::HTTP.new(url.host, url.port)
    http.read_timeout = DEFAULT_TIMEOUT

    res = http.start { |h| h.get("#{url.path}?#{url.query}", HEADERS) }

    if res.code == '404'
      raise "Twitter responded with a 404 for your query"
    end

    self.parse_search(res.body)
  end

  def self.sanitize_query(opts)
    if opts.is_a? String
      "q=#{CGI.escape(opts)}"
    elsif opts.is_a? Hash
      "#{sanitize_query_hash(opts)}"
    else
      raise "sanitique_query expects a String or a Hash"
    end
  end

  def self.sanitize_query_hash(query_hash)
    query_hash.collect{|key, value| "#{CGI.escape(key.to_s)}=#{CGI.escape(value.to_s)}" }.join('&')
  end

  def self.ensure_no_location_operators(query_string)
    if query_string.include?("near%3A") ||
            query_string.include?("within%3A")
      raise "near: and within: are available from the Twitter Search web interface, but not the API. The API requires the geocode parameter. See dancroak/twitter-search README."
    end
  end

  def self.parse_search(body)
    doc = Hpricot.parse(body)

    entries = []

    items = (doc/:entry)
    items.each do |raw_item|
      entry = {}

      entry[:id] = raw_item.at('id').inner_html
      entry[:published] = raw_item.at('published').inner_html
      entry[:updated] = raw_item.at('updated').inner_html
      entry[:title] = raw_item.at('title').inner_html
      entry[:content] = Hpricot(raw_item.at('content').inner_html, :xhtml_strict => true).to_plain_text
      entry[:source] = raw_item.at('twitter:source').inner_html
      entry[:lang] = raw_item.at('twitter:lang').inner_html

      entry[:links] = []
      raw_item.search('link').each do |s|
        link = {}
        link[:href] = s['href']
        link[:type] = s['type']
        link[:rel] = s['rel']
        entry[:links] << link
      end

      entry[:author] = {}
      raw_item.search('author').each do |a|
        entry[:author][:name] = a.at('name').inner_html
        entry[:author][:uri] = a.at('uri').inner_html
      end

      entries << entry
    end

    entries
  end
end