var util = require("util"); var superagent = require("superagent"); var cheerio = require("cheerio"); var async = require("async"); var fs = require('fs');

var cityOptions = { "cityId": 110, // 合肥 // 全部商区, 蜀山区, 庐阳区, 包河区, 政务区, 瑶海区, 高新区, 经开区, 滨湖新区, 其他地区, 肥西县 "regionIds": [0, 356, 355, 357, 8840, 354, 8839, 8841, 8843, 358, -922], "categoryId": 10, // 美食 "sortId": 2, // 人气最高 "threshHold": 5000 // 最多餐馆数 }; var idVisited = {}; // used to distinct shop var ratingDict = {}; // id -> ratings var posDict = {}; // id -> pos

function isVisited(id) { if (idVisited[id] != undefined) { return true; } else { idVisited[id] = true; return false; } }

function DianpingSpider(regionId, start, callback) { console.log('crawling region=', regionId, ', start =', start); var searchBase = 'http://m.api.dianping.com/searchshop.json"htmlcode">

DianpingSpider(0, 0, function (err, restaurants) { if (err) return console.err(err.stack); var concurrency = 0; var crawlMove = function (id, callback) { var delay = parseInt((Math.random() * 30000000) % 1000, 10); concurrency++; console.log('current concurrency:', concurrency, ', now crawling id=', id, ', costs(ms):', delay); parseShop(id); parseMap(id); setTimeout(function () { concurrency--; callback(null, id); }, delay); }; async.mapLimit(restaurants, 5, function (restaurant, callback) { crawlMove(restaurant.id, callback) }, function (err, ids) { console.log('crawled ids:', ids); var resultArray = []; async.until( function () { return restaurants.length === Object.keys(ratingDict).length && restaurants.length === Object.keys(posDict).length }, function (callback) { setTimeout(function () { callback(null) }, 1000) }, function (err) { restaurants.forEach(function (restaurant) { var rating = ratingDict[restaurant.id]; var pos = posDict[restaurant.id]; var result = Object.assign(restaurant, rating, pos); resultArray.push(result); }); writeAsJson(resultArray); } ); }); });


其中，parseShop与parseMap分别为解析商家详情页、商家地图页：


function parseShop(id) { var shopBase = 'http://m.dianping.com/shop/%s'; var shopUrl = util.format(shopBase, id); superagent.get(shopUrl) .end(function (err, res) { if (err) return console.err(err.stack); console.log('crawling shop:', shopUrl); var restaurant = {}; var $ = cheerio.load(res.text); var desc = $("div.shopInfoPagelet > div.desc > span"); restaurant.taste = desc.eq(0).text().split(":")[1]; restaurant.surrounding = desc.eq(1).text().split(":")[1]; restaurant.service = desc.eq(2).text().split(":")[1]; ratingDict[id] = restaurant; }); } function parseMap(id) { var mapBase = 'http://m.dianping.com/shop/%s/map'; var mapUrl = util.format(mapBase, id); superagent.get(mapUrl) .end(function (err, res) { if (err) return console.err(err.stack); console.log('crawling map:', mapUrl); var restaurant = {}; var $ = cheerio.load(res.text); var data = $("body > script").text(); var latRegex = /(.*lat:)(\d+.\d+)(.*)/; var lngRegex = /(.*lng:)(\d+.\d+)(.*)/; if(data.match(latRegex) && data.match(lngRegex)) { restaurant.latitude = latRegex.exec(data)[2]; restaurant.longitude = lngRegex.exec(data)[2]; }else { restaurant.latitude = ''; restaurant.longitude = ''; } posDict[id] = restaurant; }); }


将array的每一个商家信息，逐行写入到json文件中：


function writeAsJson(arr) { fs.writeFile( 'data.json', arr.map(function (data) { return JSON.stringify(data); }).join('\n'), function (err) { if (err) return err.stack; }) }


总结
以上就是这篇文章的全部内容，希望本文能给学习或者使用node.js的朋友们带来一定的帮助，如果有疑问大家可以留言交流。