Sort descending String with Number in Mongo DB - node.js

Question

I have currently a DB with two fields. Only one of them matter for the purpose of this question.
Imagine a DB with a single String field (let's call it "Tags"), and the following pattern: [a-z]*[0-9]*, like:
test129
test130
some43
some44
some45
...
My application needs to generate new "Tags", given the prepend "identifier" (like test or some).
So let's say I input test as the prepend name, and 100 as the number of "Tags" to generate.
He finds the LAST tag with the prepended name test on it.
Parses the number after the prepended name.
Sum +1 on that number, and generate 100 tags with the sequence.
Output in this specific case, would be: test131, test132, ..., test230.
I implemented this, and it was working just great with Mongoose. However, when I tried to generate a "Tag" from a already existent Tag with more than 1000, I found that the first step was a flaw. It was returning, let's say test999 instead of test1200, and causing the iteration to start from 999, and getting errors since it needs to be unique.
This is because, sorting a String differs from sorting a Number. I know the problem, but how can I solve this in a simple way, without having to create extra fields?
UPDATE: Part of the code where I find the tag:
lastAliasNumber: function (next){
console.log('process.lastAliasNumber');
// Skip if prefix is not set (tags already have name)
if(!prefix) return next();
// Build RegExp to find tags with the prefix given
var regexp = new RegExp('^'+prefix+'[0-9]+$', 'i');
Models.Tag
.findOne()
.where({
alias: regexp
})
.sort('-alias')
.exec(function (err, tag){
if(err) return next(err);
// Remove prefix and try parsing number
var lastId = 100;
if(tag){
// Remove prefix
var number = tag.alias.toLowerCase().replace(prefix, '');
// Get number from it
number = parseInt(number);
if(number) lastId = number;
}
console.log('lastAliasNumber', lastId);
next(null, lastId);
});
},

Answer 1

There is no ready way to do this kind of sorting within MongoDB. As your field is a string field, it will be sorted by the rules of string sorting and there is no way to do variable type sorting on one field.
Your best bet (assuming you cannot simply use an integer type and wish to keep only one field) would be to work out the theoretical maximum number of entries and pad your strings with the relevant leading number of 0's accordingly.
EG. assuming a maximum of 1,000,000 entries your strings would be:
test000999
test001200
test000131
Another option would be to have these entries become whole subdocuments with two distinct datatypes.
Consider my quick example documents below
> db.bar.insert({x:{text:"test",num:1}})
WriteResult({ "nInserted" : 1 })
> db.bar.insert({x:{text:"test",num:100}})
WriteResult({ "nInserted" : 1 })
> db.bar.insert({x:{text:"test",num:2}})
WriteResult({ "nInserted" : 1 })
> db.bar.insert({x:{text:"sweet",num:2}})
WriteResult({ "nInserted" : 1 })
> db.bar.insert({x:{text:"sweet",num:1}})
WriteResult({ "nInserted" : 1 })
> db.bar.find().sort({x:1})
{ "_id" : ObjectId("55fa469d695632545d3aff1f"), "x" : { "text" : "sweet", "num" : 1 } }
{ "_id" : ObjectId("55fa469b695632545d3aff1e"), "x" : { "text" : "sweet", "num" : 2 } }
{ "_id" : ObjectId("55fa468a695632545d3aff1b"), "x" : { "text" : "test", "num" : 1 } }
{ "_id" : ObjectId("55fa4695695632545d3aff1d"), "x" : { "text" : "test", "num" : 2 } }
{ "_id" : ObjectId("55fa468f695632545d3aff1c"), "x" : { "text" : "test", "num" : 100 } }
> db.bar.find().sort({x:-1})
{ "_id" : ObjectId("55fa468f695632545d3aff1c"), "x" : { "text" : "test", "num" : 100 } }
{ "_id" : ObjectId("55fa4695695632545d3aff1d"), "x" : { "text" : "test", "num" : 2 } }
{ "_id" : ObjectId("55fa468a695632545d3aff1b"), "x" : { "text" : "test", "num" : 1 } }
{ "_id" : ObjectId("55fa469b695632545d3aff1e"), "x" : { "text" : "sweet", "num" : 2 } }
{ "_id" : ObjectId("55fa469d695632545d3aff1f"), "x" : { "text" : "sweet", "num" : 1 } }