How to filter array inside mongoDB document, so it will return a document with filtered array inside.

MongoDb-FilteredArray-Source.js

/* 1 */
{
    "_id" : "joe",
    "name" : "Joe Bookreader",
    "addresses" : [ 
        {
            "street" : "123 Fake Street",
            "city" : "Faketon",
            "state" : "MA",
            "zip" : "12345"
        }, 
        {
            "street" : "1 Some Other Street",
            "city" : "Boston",
            "state" : "MA",
            "zip" : "12345"
        }
    ]
}

/* 2 */
{
    "_id" : "driver",
    "name" : "Driver X",
    "addresses" : [ 
        {
            "street" : "123 Fake Street 12222",
            "city" : "Faketon",
            "state" : "MA",
            "zip" : "11111"
        }, 
        {
            "street" : "1 Some Other Street 34",
            "city" : "Boston",
            "state" : "MA",
            "zip" : "12345"
        }
    ]
}

MongoDb-FilteredArray.js

db.getCollection('sample').aggregate(
  {
    "$match": {
      "addresses.zip" : "11111"
       }
  },
  {
    "$unwind" :"$addresses"
  }, 
  {
    "$match": {
       "addresses.zip" : "11111"
    }
  },
  {
    "$group" : {
        "_id": "$_id",
        "addresses" :{
             "$addToSet" : "$addresses"
        }
    }
  )

This aggregate functions performs following steps to produce the filtered array:

1. Step 1 -> Filter the document, where addresses.zip = "11111"
2. Step 2 -> Pipe the result to $unwind by addresses, where each Document will contains only individual addresses.
3. Step 3 -> Pipe the result to filter, where addresses.zip = "11111".
4. Step 4 -> Pipe the result to group based on _id , and reconstruct the $addresses

So what's going on?
To have better understanding, you can remove the pipe process (Step 2-4) and incrementally adding it to the aggregate during your understanding process.

Step 1 - is a filter to document in the collection. The result is all documents where any of zip is "11111". Using sample data, it returns 2 document.
Step 2 - unwind process will extract each individual addresses into different documents. Using sample data, it returns 4 documents. Each document contains single address.
Step 3 - is a filter to document in the in-memory document. These documents now only has single address each - therefore re-applying same filter as Step 1, will remove un-match address from the result.
Step 4 - is a grouping process. In-memory document contains filtered document, with single addresses each - it is also filtered after Step 3 was applied. Grouping will construct original presentation of the document - with filtered array of addresses.

http://stackoverflow.com/questions/33531808/mongodb-c-sharp-aggregation-unwind-groupby