CODEWITHSUNDEEP
arraysmongodbaggregation-framework
Wernfried Domscheit
Wernfried Domscheit

Reputation: 59456

How to modify value in array by position at aggregation framework

Let's say a have a simple document:

{
   array: ["a", "b", "c", "d"]
}

How to modify the second value in aggregation?

With update it is very simple:

db.collection.updateOne({},
   { $set: { "array.1": "B" } }
)

gives:

{
   array: ["a", "B", "c", "d"]
}

In aggregation framework you can use this one:

db.collection.aggregate([
   {
      $set: {
         "array": {
            $map: {
               input: "$array",
               in: {
                  $cond: {
                     if: { $eq: [{ $indexOfArray: ["$array", "$$this"] }, 1] },
                     then: "B",
                     else: "$$this"
                  }
               }
            }
         }
      }
   }
])

However, this fails when the array is not a Set, i.e. values are not unique like this one

{
   array: ["a", "b", "c", "b"]
}

Upvotes: 3

Views: 2036

Answers (3)

Wernfried Domscheit
Wernfried Domscheit

Reputation: 59456

I found a solution which requires less coding:

db.coll.aggregate([
   {
      $set: {
         array: {
            $function: {
               body: function (array, i, val) { array[i] = val; return array; },
               args: ["$array", 1, "B"],
               lang: "js"
            }
         }
      }
   }
])

When you need to update nested array, then the difference is even bigger. Let's assume a nested array as this:

{
   action: [
      { task: ['a', 'b', 'c', 'd'] },
      { task: ['aa', 'bb', 'cc', 'dd'] },
      { task: ['aaa', 'bbb', 'ccc', 'ddd'] }
   ]
}

It can be updated simply with

db.collection.aggregate([
   {
      $set: {
         action: {
            $function: {
               body: function (action, a, t, val) { action[a].task[t] = val; return action; },
               args: ["$action", 1, 2, "new CC"],
               lang: "js"
            }
         }
      }
   }
])

compared to

let arrayIndex = 1
let taskIndex = 2
db.collection.aggregate([
     {
        $set: {
           action: {
              $map: {
                 input: { $range: [0, { $size: "$action" }] },
                 as: "a",
                 in: {
                    $cond: {
                       if: { $eq: ["$$a", arrayIndex] },
                       then: {
                          $let: {
                             vars: { action: { $arrayElemAt: ["$action", "$$a"] } },
                             in: {
                                $mergeObjects: ["$$action", {
                                   task: {
                                      $map: {
                                         input: { $range: [0, { $size: "$$action.task" }] },
                                         as: "t",
                                         in: {
                                            $cond: {
                                               if: { $eq: ["$$t", taskIndex] },
                                               then: "new CC",
                                               else: { $arrayElemAt: ["$$action.task", "$$t"] }
                                            }
                                         }
                                      }
                                   }
                                }]
                             }
                          }
                       },
                       else: { $arrayElemAt: ["$action", "$$a"] }
                    }
                 }
              }
           }
        }
     }
 ])

Upvotes: 0

matthPen
matthPen

Reputation: 4343

Another solution is to $unwind your array with includeArrayIndex and preserveNullAndEmptyArrays, set value with condition, then group. But beware, although it can be useful in some situations, it can consume more server resources in some cases.

db.collection.aggregate([
  {
    $unwind: {
      path: "$array",
      includeArrayIndex: "index",
      preserveNullAndEmptyArrays: true
    }
  },
  {
    $addFields: {
      array: {
        $cond: {
          if: {
            $eq: [
              "$index",
              1
            ]
          },
          then: "B",
          else: "$array"
        }
      }
    }
  },
  {
    $group: {
      _id: "$_id",
      array: {
        $push: "$array"
      }
    }
  }
])

you can test here

Upvotes: 1

Ashh
Ashh

Reputation: 46451

You can loop over the $size(length) of the array using $range and then get the index of the element and can change the value of the index element to whatever you want.

db.collection.aggregate([
  { "$project": {
    "array": {
      "$map": {
        "input": { "$range": [0, { "$size": "$array" }] },
        "in": {
          "$cond": [
            { "$eq": ["$$this", 1] },
            "B",
            { "$arrayElemAt": ["$array", "$$this"] }
          ]
        }
      }
    }
  }}
])

MongoPlayground

Upvotes: 4

Related Questions