lodash sortBy then groupBy, is order maintained?

Question

I'm having trouble figuring out from the lodash documentation if my assumption about sorting and grouping is correct.

If I use sortBy, then use groupBy, do the arrays produced by groupBy maintain the sort order of items?

For example, say I have the following array:

var testArray = [[5,6],[1,3],[5,4],[5,1]]

And I would like to group these by their first element, but also have them sorted by their second element within these groups. So, in lodash I assume I can do the following:

_.chain(testArray)
  .sortBy(function (item) { return item[1]; })
  .groupBy(function (item) { return item[0]; })
  .value()

Which ends up producing what I would expect it to:

{
  1: [[1,3]]
  5: [[5,1],[5,4],[5,6]]
}

Is this just coincidence? Is there anything about how sortBy and groupBy work that ensures this ordering of the grouped arrays? The documentation says that sortBy is a stable sort, does that in the same way apply to groupBy? Is there any reason I should not assume this will work every time?

Answer 1

It's not. Here's example, where order is not retained:

const data = [
  {
    item: 'item1',
    group: 'g2'
  },   {
    item: 'item2',
    group: 'g3'
  },   {
    item: 'item3',
    group: 'g1'
  },   {
    item: 'item4',
    group: 'g2'
  },   {
    item: 'item5',
    group: 'g3'
  }
]

const groupedItems = _(data).groupBy(item => item.group).value()

In this case one would expect that group order would be: g2, g3, g1 - reality is that they are sorted g1, g2, g3.

You can re-sort them with original array though.

const groupedItems = _(data)
  .groupBy(item => item.group)
  .sortBy(group => data.indexOf(group[0]))
  .value()

This will ensure original order of items.

Answer 2

The current implementation of _.groupBy is:

// An internal function used for aggregate "group by" operations.
var group = function(behavior) {
  return function(obj, iteratee, context) {
    var result = {};
    iteratee = cb(iteratee, context);
    _.each(obj, function(value, index) {
      var key = iteratee(value, index, obj);
      behavior(result, value, key);
    });
    return result;
  };
};

// Groups the object's values by a criterion. Pass either a string attribute
// to group by, or a function that returns the criterion.
_.groupBy = group(function(result, value, key) {
  if (_.has(result, key)) result[key].push(value); else result[key] = [value];
});

Basically it iterates through each of the items in the collection in order (if the collection is array-like, which it would be after a sortBy), and pushes them to an array based on their key value.

So yes, I'm not sure if this is an "official" characteristic of _.groupBy, but it does preserve the order of array-like collections, and that's probably unlikely to change.