How to unset fields based on the fields' name with a "less than" condition?

user_11511043 0
Viewed 139

I am trying to unset certain fields based on the name of the field (the key).

Say I have something like this:

{
    5: "cool",
    93: "cool",
    30: "cool",
    56: "cool"
}

How would I unset all fields with a value below, say, 40.

So the result should be:

{
    5: "cool",
    30: "cool"
}

I tried using the less than command on the whole field or using the positional operator but both of those failed.

collection.update_one(
        {"_id": id},
        {"$unset": {"blacklist.$": {"$lt": 40}}}
    )

I couldn't find anything online or on the docs so I am hoping to find some help here,

Thanks!

pymongo mongodb-query mongodb
2 Answers

You can't really use $unset like this, we can still achieve this using pipelined updates - with slightly more complicated syntax.

Our approach will be to turn our root object to an array using $objectToArray, iterate over it and filter all numeric keys under a certain threshold. then finally convert back to an object and update our document, like so:

db.collection.update({},
[
  {
    $replaceRoot: {
      newRoot: {
        $arrayToObject: {
          $filter: {
            input: {
              $objectToArray: "$$ROOT"
            },
            cond: {
              $cond: [
                {
                  $regexMatch: {
                    input: "$$this.k",
                    regex: "^[0-9]+$"
                  }
                },
                {
                  $lt: [
                    {
                      $toInt: "$$this.k"
                    },
                    40
                  ]
                },
                true
              ]
            }
          }
        }
      }
    }
  }
])

Mongo Playground

user_107703700

Query

  • into [] the $$ROOT
  • filter the keys to be either "_id" or < 40
  • back to object and
  • replace root

*this removes all fields >=40 , you can change the last $lt if you don't need this

*toms answer works and its more general, if you have mixed keys that some are numbers and some are not use toms way (this is slightly simpler but supposes that all fields are numbers except the "_id")

Test code here

aggregate(
[{"$replaceRoot":
  {"newRoot":
   {"$arrayToObject":
    [{"$filter":
      {"input":{"$objectToArray":"$$ROOT"},
       "cond":
       {"$or":
        [{"$eq":["$$this.k", "_id"]},
         {"$lt":[{"$toInt":"$$this.k"}, 40]}]}}}]}}}])
user_48826920
Related