Lets say I have an array ['123', '456', '789']

I want to Aggregate and look through every document with the field books and only return the values that are NOT in any documents. For example if '123' is in a document, and '456' is, but '789' is not, it would return an array with ['789'] as it's not included in any books fields in any document.

.aggregate( [
{
  $match: {
    books: {
        $in: ['123', '456', '789']
    }
  }
},

I don't want the documents returned, but just the actual values that are not in any documents.

CodePudding user response：

Here's one way to scan the entire collection to look for missing book values.

db.collection.aggregate([
  {  // "explode" books array to docs with individual book values
    "$unwind": "$books"
  },
  {  // scan entire collection creating set of book values
    "$group": {
      "_id": null,
      "allBooksSet": {
        "$addToSet": "$books"  // <-- generate set of book values
      }
    }
  },
  {
    "$project": {
      "_id": 0,  // don't need this anymore
      "missing": {  // use $setDifference to find missing values
        "$setDifference": [
          [ "123", "456", "789" ],  // <-- your values go here
          "$allBooksSet"  // <-- the entire collection's set of book values
        ]
      }
    }
  }
])

Example output:

[
  {
    "missing": [ "789" ]
  }
]

Try it on mongoplayground.net.

CodePudding user response：

Based on @rickhg12hs's answer, there is another variation replacing $unwind with $reduce, which considered less costly. Two out of Three steps are the same:

db.collection.aggregate([
  {
    $group: {
      _id: null,
      allBooks: {$push: "$books"}
    }
  },
  {
    $project: {
      _id: 0,
      allBooksSet: {
        $reduce: {
          input: "$allBooks",
          initialValue: [],
          in: {$setUnion: ["$$value", "$$this"]}
        }
      }
    }
  },
  {
    $project: {
      missing: {
        $setDifference: [["123","456", "789"], "$allBooksSet"]
      }
    }
  }
])

Try it on mongoplayground.net.