Tombloo Togetter extractor

TogetterToTumblr.js

Tombloo.Service.extractors.register(
  {
    name: 'Conversation - togetter',
    RE : /^http:\/\/togetter\.com\/li\//,
    ICON : 'http://togetter.com/favicon.ico',
    check : function(ctx){
      return this.RE.test(ctx.href);
    },
    extract : function (ctx) {
      var list_item = $x('//li[contains(concat(" ", @class, " "), " list_item ")][div[contains(concat(" ", @class, " "), " type_tweet ")]]', ctx.document, true);
      var twBody = new Array();
      var twText, twId, twStatusId, twPermalink
      
      for (i = 0; i < list_item.length; i++) {
        twText = $x('.//div[@class="tweet"]', list_item[i]);
        twId = $x('.//a[@class="status_name"]', list_item[i]);
        twStatusId = $x('.//div[@class="status_right"]/a[2]/@href', list_item[i])
                       .match(/twitter\.com\/[^\/]+\/status\/(\d+)/)[1];
        twPermalink = 'https://twitter.com/' + twId.textContent + '/statuses/' + twStatusId;
        if (twText && twId && twPermalink) {
          twBody.push(twId.textContent + ":" + twText.textContent.replace(/[\r\n]/g, '') + " [" + twPermalink + "]");
        }
      }

      var tgId = $x('//div[contains(concat(" ", normalize-space(@class), " "), " info_description ")]/span/a').textContent;
      var m = {
        type: 'conversation',
        item : ctx.title + " by " + tgId + " " + ctx.href,
        body: twBody.join('\n'),
        //tags: "togetter"
      };
      
      return m;
    }
  }
);