Conditional aggregation in array field

Working with Data
,
1

my data in a collection looks like this.

    {_id: ..., gid: '0', application: [ {packageName: 'ABC', version: '0.0.1'},  {packageName: 'DEF', version: '1.0.1'}  ]}; 
    {_id: ..., gid: '1', application: [ {packageName: 'ABC', version: '0.0.2'},  {packageName: 'DEF', version: '1.0.1'}  ]}; 
    {_id: ..., gid: '2', application: [ {packageName: 'ABC', version: '0.0.1'},  {packageName: 'DEF', version: '1.0.1'}  ]};

What i want to query is to get the count of documents with ‘ABC’ app group by it’s version.
Expected output would be

    {version: '0.0.1', count: 2}
    {version: '0.0.2', count: 1}
2

Please read Formatting code and log snippets in posts so that it is easier for us the next time you publish code or data.

By

you mean where packageName is ABC.

Try

match = { "$match" : {  "application.packageName" : "ABC" } }
unwind = { "$unwind" : { "path" : "$application" } }
group = { "$group" : { "_id" : "$application.version" , "count" : { "$sum" : 1 } } }
pipeline = [ match , unwind , match , group ]
/* pipeline = [ unwind , match , group ]
   should also work but less efficient */
db.Collection.aggregate( pipeline )
/* You will have _id rather than version in the results but a $set or
   $project can take care of that. */
> [ { _id: '0.0.2', count: 1 }, { _id: '0.0.1', count: 2 } ]

You probably could use $map or $filter to avoid $unwind but I like the simplicity of $unwind. I left out the obvious $sort to get then in the order you posted.