Node.js callback unexpected behaviour with Mongoose - node.js

Question

I am a nodejs beginner and I ran into some callback behaviour that I don't understand. I am using an Express router to write Mongoose objects to Mongodb using a POST request. In the body of the request I am passing in a nested json structure with two fields - jobDetails and examples. The data in jobDetails is used to create a Job Mongoose object and the data in examples is used to create several Example Mongoose objects. The Job and Example objects are linked in that a Job contains a list of Example objects in one of it's fields.
The way I tried to implement this was with callbacks in the following way. Basically I save the Job object to mongo first, then iterate over the examples - each time creating an Example object and linking it to the job via the .job field and also saving the Example object to mongo. Then in the callback to the Example object save function I updated the Job object with the new Example object and saved the updated version to mongo.
router.post('/jobs', function (req, res, next) {
var job = new Job(req.body.jobDetails);
var examples = req.body.examples;
console.log("JOB DETAILS");
console.log(req.body.jobDetails);
console.log("EXAMPLES");
console.log(req.body.examples);
//save job
job.save(function (err, job) {
console.log(err);
});
//save examples
for(i=0; i<examples.length;i++){
var eg = new Example({content: examples[i]});
eg.job=job;
eg.save(function (err, eg){
job.examples.push(eg);
job.save(function(err, job){
console.log(err);
});
console.log(err);
});
}
});
This did not perform as I would expect. Specifically, double the number of examples were actually saved to mongo with several duplicates and some missing. I understand that callbacks are asynchronous but to me that still doesn't explain why double the number of examples would be saved and some would be duplicated and some would be missing.
I eventually got it to work correctly without using callbacks at all in the following way.
router.post('/jobs', function (req, res, next) {
var job = new Job(req.body.jobDetails);
var examples = req.body.examples;
console.log("JOB DETAILS");
console.log(req.body.jobDetails);
console.log("EXAMPLES");
console.log(req.body.examples);
//save job
job.save(function (err, job) {
console.log(err);
});
//save examples
for(i=0; i<examples.length;i++){
var eg = new Example({content: examples[i]});
eg.job=job;
eg.save(function (err, eg){
console.log(err);
});
job.examples.push(eg);
job.save(function(err,job){
console.log(err);
});
}
});
And I'm not sure if this is the optimal solution either. But I would like to know why my original approach resulted in the unintended behaviour.

Answer 1

This should work..
router.post('/jobs', function(req, res, next) {
var job = new Job(req.body.jobDetails);
var examples = req.body.examples;
console.log("JOB DETAILS");
console.log(req.body.jobDetails);
console.log("EXAMPLES");
console.log(req.body.examples);
//save job
job.save(function(err, result) {
if (!err) {
//save examples
for (i = 0; i < examples.length; i++) {
var eg = new Example({
content: examples[i]
});
eg.job = job;
eg.save(function(err, eg) {
job.examples.push(eg);
job.save(function(err, job) {
if (!err)
job.examples = [];
});
console.log(err);
});
}
}
});
});

Answer 2

I would suggest you use a library like async to perform such save operations in a step by step manner. Follow this approach for better readability of code and better results
var async = require('async');
router.post('/jobs', function(req, res, next) {
var job = new Job(req.body.jobDetails);
var examples = req.body.examples;
var savedExamples = [];
console.log("JOB DETAILS");
console.log(req.body.jobDetails);
console.log("EXAMPLES");
console.log(req.body.examples);
async.eachSeries(examples, function iteratee(example, callback) {
var eg = new Example({
content: example
});
eg.job = job;
eg.save(function(err, savedEg) {
if(!err) {
savedExamples.push(savedEg);
}
callback(err)
});
}, function(err) {
if(err) {
//handle errors
}
job.examples = savedExamples;
job.save(function(err,job) {
if(err) {
//handle errors
}
//success callback
});
});
});
Using this approach you will have to call the save function for job only once after all other operations are completed. If an error is triggered at any point the whole flow is stopped. For more info regarding async library refer this!