Mapreduce split input string into output array - arrays

Question

I am dealing with documents like the following one:
> db.productData.find({"upc" : "XXX"}).pretty()
{
"_id" : ObjectId("538dfa3d44e19b2bcf590a77"),
"upc" : "XXX",
"productDescription" : "bla foo bar bla bla fooX barY",
"productSize" : "",
"ingredients" : "foo; bar; foo1; bar1.",
"notes" : "bla bla bla"
}
>
I would like to have a document containing, among the fields, a list/array of splitted ingredients (on the ;). I want to split the string of the original collection into an array of strings.
I would like to map only some of the input fields in the output collection.
I would like to use mapreduce on MongoDB.
I've tried many different ways moving stuff from the map function to the reduce function failing to find a proper solution.
From all the attempts I performed, now I know I need to check for null values etc, so the following one is my last attempt:
The map function:
var mapperProductData = function () {
var ingredientsSplitted = values.ingredientsString.split(';');
var objToEmit = {barcode : "", description : "", ingredients : []};
// checking for null (is this strictly necessary? why?)
if (
this.hasOwnProperty('ingredients')
&& this.hasOwnProperty('productDescription')
&& this.hasOwnProperty('upc')
) {
for (var i = 0; i < ingredientsSplitted.length; i++) {
// I want to emit a new document only when I have all the splitted strings inside the array
if (i == ingredientsSplitted.length - 1) {
objToEmit.barcode = this.upc;
objToEmit.description = this.productDescription;
objToEmit.ingredients = ingredientsSplitted;
emit(this.upc, objToEmit);
}
}
}
};
The reduce function:
var reducerNewMongoCollection = function(key, values) {
return values;
};
The map-reduce call:
db.productData.mapReduce(
mapperProductData,
reducerNewMongoCollection,
{
out : "newMongoCollection" ,
query: { "values" : {$exists: true} }
}
);
I am getting an empty collection in output (newMongoCollection is empty).
What am I doing wrong?

Answer 1

Let's start from the beginning. Your map function should look like this:
var mapperProductData = function () {
var ingredientsSplitted = this.ingredients.split(';');
var objToEmit = {
barcode : this.upc,
description : this.productDescription,
ingredients : ingredientsSplitted
};
emit(this.upc, objToEmit);
};
Your map-reduce call should be:
db.productData.mapReduce(
mapperProductData,
reducerNewMongoCollection,
{
out : "newMongoCollection",
query : {
upc : { $exists : true },
productDescription : { $exists : true },
ingredients : { $exists : true , $type : 4 }
}
}
);
The query part will filter the documents that do have relevant fields. Also the query parameter $type will match only documents where ingredients is an array. This way you don't need to do complicated checking inside your map function and the number of documents sent to map function will be lower.
The result for your test document document will look like this:
key : XXX,
value: {
"barcode" : "XXX",
"description" : "bla foo bar bla bla fooX barY",
"ingredients" : [
"foo",
" bar",
" foo1",
" bar1."
]
}