How to reduce this mongodb multi update query to just one query involving nested arrays?

Working with Data
, , ,
#1

Schema:

{
   img: // array of minimum length 1 of ObjectIds or field is "missing" to indicate that array is empty
   tar: [
     { img:  // array of minimum length 1 of ObjectIds or field is "missing" to indicate that array is empty },
     { img:  // array of minimum length 1 of ObjectIds or field is "missing" to indicate that array is empty },
     ... 
   ]
}

This is my query to delete a photo, which is an ObjectId in the top level img array, and may be in the img field of some tar subdocuments:

      await mongo_tran_ses.withTransaction(async () => {
        // pulling the ObjectId matching `req.params._id` from `img`, and all `tar.img`
        await user.updateOne(
          { _id: ObjectId(req.user._id) },
          {
            $pull: {
              img: { _id: { $eq: ObjectId(req.params._id) } },
              "tar.$[].img": { $eq: ObjectId(req.params._id) },
            },
          },
          { session: mongo_tran_ses }
        );
        // removing all `tar.img` if it is now an empty array 
        await user.updateOne(
          {
            _id: ObjectId(req.user._id),
          },
          {
            $unset: {
              "tar.$[element].img": "",
            },
          },
          {
            arrayFilters: [{ "element.img": { $size: 0 } }],
            session: mongo_tran_ses,
          }
        );
        // removing `img` if it is now an empty array 
        await user.updateOne(
          {
            _id: ObjectId(req.user._id),
            img: { $size: 0 },
          },
          {
            $unset: {
              img: undefined,
            },
          },
          {
            session: mongo_tran_ses,
          }
        );
      }, tran_option);

The goal:

Perform the above with just one update query instead of the 3 above.

#2

Hi :wave: @Big_Cat_Public_Safety_Act,

Could you please elaborate on it further? Also, can you confirm if your sample document looks like the following:

{
   img: [ 6423c91b114e33a0106407be, 6423c91b114e33a0106407bg, 6423c91b114e33a0106407ba]
   tar: [
     { img:  [6423c91b114e33a0106407be] },
     { img:  [6423c91b114e33a0106407be] },
     ... 
   ]
}

If not, please share a sample document, MongoDB version, and expected result document.

Could you please explain the reason behind opting for a single query? Is it to ensure atomicity? If so, would it be possible to use transactions instead?

Without knowing your use case details, I think there will be a tradeoff between performance and complexity. I believe if you can craft a three-query solution, to me it means that you can maintain it easily. A single query solution might be a little faster, but I think a proper workload simulation would be needed to determine whether the performance gains is worth it.

Regards
Kushagra

#3

yes, your sample document is correct. It is mongodb 5.

And the motivation for one query is for faster performance. And I do understand that the query will be more complex. But I am not quite sure how much more complext.