Question

I have a document which includes a subdocument:

{
    "_id" : ObjectId("XXXXX"),
    "SearchKey" : "1234",
    "SearchTerms" : {
        "STA" : ["1"],
        "STB" : ["asdfasdf"],
        "STC" : ["another"]
    }
}

The SearchTerm elements are not fixed – sometimes we’ll have STA without STC, for example.

I can do this:

var map = function() {
    for (key in this.SearchTerms)
    {
        emit(key, 1);
    }
}

but I can’t do this:

var map = function() {
    for (var i=0; i< this.SearchTerms.length; i++)
    {
        emit(this.SearchTerms[i], 1)
    }
}

because the latter doesn’t produce any results after the reduce. Why not?

As an aside – what I need to do is count the cross-product of the search terms over all documents, that is, find the incidence of (STA and STB) and (STA and STC) and (STB and STC) in the case above. If someone knows how to do that right away, that works even better.

As always, thanks for the help

Answer 1

The key that you emit should be a composite of both keys.

    var map = function() {
        if(!this.SearchTerms) return;
        for(var i = 0 ; i < this.SearchTerms.length; i++){
            var outerKey = this.SearchTerms[i];
            for(var j = i + 1; j < this.SearchTerms.length; j++){
               var innerKey = this.SearchTerms[j];
               // assuming you don't care about counting (STA and STC) separately from (STC and STA), 
               // then order doesn't matter, lets make sure both occurences have the same key.
               var compositeKey = (outerKey < innerKey) ? outerKey+"_"+innerKey : innerKey+"_"+outerKey; 
               emit(compositeKey, 1);
            }
        }
    }