fetch_dp_nb_shops.js

// Generated by CoffeeScript 1.3.3
(function() {
  var $, jsdom, nb_search_url, page_size, parse_db_search_result, request;

  jsdom = require("jsdom");

  $ = require("jQuery");

  request = require("request");

  nb_search_url = "http://www.dianping.com/search/keyword/11/0_";

  page_size = 15;

  exports.fetch_shops = function(title, callback) {
    var shops;
    if (title === "") {
      callback("lack of arg: 'title'!");
      return;
    }
    shops = [];
    return request({
      url: nb_search_url + encodeURIComponent(title.trim()),
      headers: {
        "User-Agent": "Mozilla/5.0 (X11; U; Linux i686; zh-CN; rv:1.9.1.2) Gecko/20090803 Fedora/3.5.2-2.fc11 Firefox/3.5.2"
      }
    }, function(error, response, body) {
      var jq_doc, searchList, shop, _i, _len;
      if (!error && response.statusCode === 200) {
        console.log("search title:" + title);
        jq_doc = $(body);
        searchList = $(jq_doc.find("#searchList dl")[0]).children("dd");
        if (!searchList || searchList.length === 0) {
          callback("page scraping error: no shop list elements!");
          return;
        }
        for (_i = 0, _len = searchList.length; _i < _len; _i++) {
          shop = searchList[_i];
          shop = parse_db_search_result(shop);
          shops.push(shop);
        }
        return callback(null, shops);
      } else {
        return callback(error);
      }
    });
  };

  parse_db_search_result = function(shop) {
    var address_row, jq_shop, rate_class, rate_span, result, shopid_text;
    result = {};
    jq_shop = $(shop);
    rate_span = $(jq_shop.find(".remark")[0]).children("li").children("span");
    rate_class = rate_span.attr('class');
    result.rate = parseFloat(rate_class.substr(rate_class.indexOf('irr-star') + 8)) / 10;
    result.average = parseInt($(jq_shop.find(".average")[0]).text().substr(1));
    address_row = $(jq_shop.find(".address")[0]).text().substr(3).split("  ");
    result.address = address_row[0];
    result.phone = address_row[1];
    result.shop_name = jq_shop.find(".shopname a").attr("title");
    result.title = jq_shop.find(".shopname a").text();
    shopid_text = jq_shop.find(".shopname a").attr("href");
    result.fetch_id = shopid_text.substr(shopid_text.indexOf("shop/") + 5);
    return result;
  };

}).call(this);