返回数组类型字段包含确切顺序的数组的文档

Question

I'm having documents in MongoDB structured like this:

{ _id : 1, tokens : [ "one","two","three","four","five","six","seven" ] }
{ _id : 2, tokens : [ "two","three","four","one","one","five","eight" ] }
{ _id : 3, tokens : [ "six","three","four","five","one","five","nine" ] }

On average the documents contain token arrays with a length of 4500 items.

I need to do some sort of pattern matching, where I have arrays of tokens in exact order to match, ie let's say I have to find the following in exactly matching order...

["three","four","five"]

...I want my query to provide me the following documents...

{ _id : 1, tokens : [ "one","two","three","four","five","six","seven" ] }
{ _id : 3, tokens : [ "six","three","four","five","one","five","nine" ] }

Ie both documents contain the exact order of the items I had in my array to search with.

Arrays I search with may have different lengths, ranging from 1 to 15 tokens.

I'm looking for the following:

• Is this doable with MongoDB queries? I've read, and re-read and re-re-read the pretty good docs, but couldn't find a solution eg using $all.
• Is there perhaps a better way to store tokens like this to get done what I need?

Thanks for any help.

Answer 1

It's going to be slow, but you can do this with a $where operator; pairing it with an $all operator to help with performance.

db.test.find({
    tokens: {$all: ["three","four","five"]},
    $where: function() {
        var ix = -1;
        // Find each occurrence of 'three' in this doc's tokens array and return
        // true if it's followed by 'four' and 'five'.
        do {
            ix = this.tokens.indexOf('three', ix + 1);
            if (ix !== -1 && ix+2 < this.tokens.length && 
                this.tokens[ix+1] === 'four' && this.tokens[ix+2] === 'five') {
                return true;
            }
        } while (ix !== -1);
        return false;
    }
})