Question

I am writing a node js application. I am using request and cheerio to load a set of URLs and get a bunch of information for the site, now let’s assume all I am trying to get is the title:

var urls = {"url_1", "url_2", "url_3",...,"url_n"};   
for(var i=0; i<urls.length; i++)
{
     getDOMTitle(urls[i],function(error,title){
         if(error)
            console.log("Error while getting title for " + urls[i]);
         else
            console.log("The title for " + urls[i] + " is " + title);
     });
}     

This is how my getDOMTitle method looks:

function getDOMTitle(urlReq,callback)
{
      var request = require('request');
      var cheerio = require('cheerio');
      request({url:urlReq},function(error, response, doc){
           var $ = cheerio.load(doc);
           if(error)
           {
              callback(true,null);
           }
           else
           {
              $('title', 'head').each(function (i, elem) {
                 var title = $(this).text();
                 callback(false,title);
               });     
           }
      }
}

In the case where the module throws an uncaught exception, how do I handle that situation?
I have tried adding the following:

  process.on('uncaughtException', function (err) {
    console.error(err);
    console.log("Node NOT Exiting...");
    callback(true,null);
  });

When I do that, I get an error saying I cannot set the headers once they have been sent. If I remove the callback from the process error handling, I do not see that error but the client spins for a long time because I assume we are never calling the callback.

How can I solve this?

Also, I have read somewhere that you can catch uncaught exceptions at the application level so you don’t have to replicate the code to catch it in every method, is that possible? and if it is and the method that threw the exception is expected to callback with some information, how can that be achieved?

Thank you,

Answer 1

To answer your stated question, using an uncaught exception handler as a general error-trapping mechanism is commonly regarded as poor design. It’s a false economy to use it to handle anything other than non-recoverable situations where you just need to do some cleanup before exiting.

You’ve got some problems in your example code. In your for loop, all the callbacks are going to report that they were working with the very last URL in your array because they’re all referring to the same copy of i, which will be at its highest value by the time any of them execute. You need to use a helper function or an immediate function invocation to give each callback a private copy of i.

In getDOMTitle the error callback should be callback(error) and the code in your loop should include the returned value in the error message. The success callback should use null as its first parameter, though this is just a matter of convention.