Node.js crawler for http://www.tehcute.com

tehcute.js

var request = require('request');
var cheerio = require('cheerio');
var util = require('util');

var baseUrl = 'http://www.tehcute.com';
var options = {
  'url': '',
  'method': 'GET',
  'headers': {
    'User-Agent': 'Mozilla/5.0 (Windows NT 5.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/31.0.1650.63 Safari/537.36',
    'Connection': 'Keep-Alive',
    'Host': 'www.tehcute.com'
  },
};

function isFullUrl(url) {
  return /^https?:\/\//.test(url);
}

function formatUrl(url) {
  if (!url) {
    return '';
  }
  if (isFullUrl(url)) {
    return url;
  } else {
    if (url.charAt(0) === '/') {
      return util.format('http://www.tehcute.com%s', url);
    } else {
      return util.format('http://www.tehcute.com/%s', url);
    }
  }
}

function crawlDetail(entry) {
  options.url = entry.url;
  request(options, function(err, res, body) {
    console.log(util.format('Current detail: %s', entry.url));
    if (err || res.statusCode != 200) {
      console.log(util.inspect({
        'err': err,
        'res': res
      }, {depth: null}));
      return;
    }
    var $ = cheerio.load(body);
    var eImg = $('#left-content img').first();
    //var eDate = null;
    var eTitle = $('#pic-info h2').first();
    if ((eImg.length !== 1) ||
        //(eDate.length !== 1) ||
        (eTitle.length !== 1) ||
        !eImg.attr('src') ||
        //!eDate.text().trim() ||
        !eTitle.text().trim()) {
      console.log('Invalid eImg, eDate or eTitle');
      console.log(util.inspect({
        'eImg': eImg,
        //'eDate': eDate,
        'eTitle': eTitle
      }, {depth: null}));
      return;
    }
    var title = eTitle.text().trim() || entry.title || $('title').text().trim();
    //var date = eDate.text().trim();
    var src = formatUrl(eImg.attr('src'));
    var alt = eImg.attr('alt') || eImg.attr('title') || title;
    if (!title /*|| !date */|| !isFullUrl(src)) {
      console.log('Invalid title, date or src');
      console.dir({
        'title': title,
        //'date': date,
        'src': src
      });
      return;
    }
    var img = {
      'src': src,
      'alt': alt,
      'thumbnail': entry.thumbnailSrc
    };
    var tags = ['gif'];
    $('#tags a').each(function() {
      tags.push($(this).text().trim());
    });
    var obj = {
      'title': title,
      //'date': date,
      'created': new Date(),
      'tags': tags,
      'imgs': [img],
      'link': entry.url
    };
    console.dir(obj);
  });
}

function crawlPager(url) {
  options.url = url;
  request(options, function(err, res, body) {
    console.log(util.format('Current pager: %s', url));
    if (err || res.statusCode != 200) {
      console.log(util.inspect({
        'err': err,
        'res': res
      }, {depth: null}));
      return;
    }
    var $ = cheerio.load(body);
    $('#main div.thumbs div.thumb-cell').each(function() {
      var eA = $(this).find('a').first();
      var eImg = $(this).find('img').first();
      var link = formatUrl($(eA).attr('href'));
      var thumbnailSrc = formatUrl($(eImg).attr('src'));
      var title = $(eA).text().trim() || $(eImg).attr('alt') || $(eA).attr('title');
      var entry = {
        'url': link,
        'thumbnailSrc': thumbnailSrc,
        'title': title
      };
      if (isFullUrl(link)) {
        setTimeout(crawlDetail, 1000, entry);
      } else {
        console.log('Invalid link:' + link);
      }
    });

    var nextUrl = '';
    $('#picnav li a').each(function() {
      if ($(this).text().trim() === 'NEXT SET') {
        nextUrl = formatUrl($(this).attr('href'));
      }
    });
    if (isFullUrl(nextUrl)) {
      setTimeout(crawlPager, 1000, nextUrl);
    } else {
      console.log(util.format('Last pager: %s', url));
    } 
  });
}

function main() {
  crawlPager('http://www.tehcute.com/thumbs/');
}

main();