$match value of local field (nested field) not empty string in $lookup pipeline

Question

I cant use a nested field on an lookup pipeline expression, to get only the not empty string values.

I am doing an aggregate query with 2 lookup stages, where the second lookup query depends on a nested field of the result of the first lookup. I can't really figure out what the problem is with my code.

invoice_lines_collection.aggregate([
        {'$match':{'customer':'rtv'}, 
        {'$lookup':{
            'from':'products',
            'localField':'ArticleNumber',
            'foreignField':'number',
            'as':'article_details'
            }
        },
        {$unwind:{
            'path':'$article_details',
            'preserveNullAndEmptyArrays': true
            }
        },
        {'$lookup':{
            'from':'cutomers',
            'let':{'group_id':'$article_details.group._id',
                   'customer':'$customer'},
            'pipeline':{
                {'$unwind':'$productgroups'},
                {'$match':
                      {'$expr':
                            {$and:
                                  ['$ne':['$$group_id','2'],
                                   '$eq': 
                                    ['$productgroups.id','$$group_id'],
                                    '$eq':['$name','$$customer']
                                  ],
                            }
                        }
                 }
             },
             'as':'customer_data'
         }
      }

])

invoice_lines=[
    {
       "_id" : ObjectId("5c885d21a202fc001103saf7"),
       "ShipmentNumber" : "70727320006714asda4",
       "Price" : 179.57,
        "customer" : "test"
    }
]

products = [
             {
               "_id" : ObjectId("2cv21eba4bd009f00161153b7"),
               "number": "1234",
               "group":{
                          "_id" :'',
                          "name" : '',
                  }
             },
             {
                 "_id" : ObjectId("5ca1eba4bd009f00161153b7"),
                 "number" : "2456",
                 "group" : {
                            "_id" : ObjectId("5ca29852bd009f00185553e3"),
                             "name" : "Test group",
                  }
             }
    ]
customers = [
                {
"_id" : ObjectId("5c6fd17a72ef146fcc29c6a1"),
"name" : "test",
"displayName" : "Test",
"productgroups" : [ 
    {
        "name" : "Test group",
        "id" : ObjectId("5ca29852bd009f00185553e3"),
        "markup" : 0.5
    },
     {
        "name" : "Test group 2",
        "id" : ObjectId("5ca29852bd009f0888554443"),
        "markup" : 3.0
    }
    ]
  }
]

I want to have only one productgroup and to take only the invoice lines where article belongs to a group and get the group details.

When I run the above code (converted in PHP language) I get

An object representing an expression must have exactly one field: { $ne: [ "", "$$group_id" ] ...

Answer 1

I figured it out, it was some syntax error and in the end i had to do a $match stage to take only the results where a an invoice lines has a product which is linked to a product group and my code now looks like this :

db.getCollection('invoice_lines').aggregate([
    {$match:{'customer':'rtv'}}, 
    {$lookup:{
        'from':'products',
        'localField':'ArticleNumber',
        'foreignField':'number',
        'as':'article_details'
        }
    },
    {$unwind:{
        'path':'$article_details',
        'preserveNullAndEmptyArrays': true
        }
    },
    {$lookup:{
        'from':'customers',
        'let':{'group_id':'$article_details.group._id',
               'customer':'$customer'},
        'pipeline':[
            {$unwind:'$productgroups'},
            {'$match':
                  {'$expr':
                        {$and:
                              [
                                {$ne:['$$group_id','']},
                                {$eq:['$productgroups.id','$$group_id']},
                                {$eq:['$name','$$customer']}
                              ],
                        }
                    }
             }
         ],
         'as':'customer_data'
     }
},
 {$match:{'customer_data':{$exists: true, $not: {$size: 0}}}}