Clean MongoDB nested array duplicated records

clean-duplicates.js

const assert = require('assert');
const { MongoClient } = require('mongodb');

MongoClient.connect('mongodb://127.0.0.1:27017/', (err, client) => {
  assert.equal(null, err);
  console.log('Connected correctly to server\n');

  async function run(arrayName) {
    const database = client.db('data-integrator');
    const orders = database.collection('orders');
    const fieldPath = `$${arrayName}`;

    const pipeline = [
      // {
      //   $match: {
      //     'externalSource.id': '63a1f9027c2a697cc587da8d',
      //   },
      // },
      {
        $unwind: {
          path: fieldPath,
        },
      },
      {
        $group: {
          _id: {
            _id: '$_id',
            [arrayName]: fieldPath,
          },
          count: {
            $sum: 1,
          },
          [arrayName]: {
            $push: fieldPath,
          },
        },
      },
      {
        $match: {
          count: {
            $ne: 1,
          },
        },
      },
      {
        $project: {
          _id: '$_id._id',
          [arrayName]: {
            $slice: [
              fieldPath,
              1,
              {
                $subtract: [
                  {
                    $size: fieldPath,
                  },
                  1,
                ],
              },
            ],
          },
        },
      },
    ];

    const cursor = orders.aggregate(pipeline);
    const batch = orders.initializeOrderedBulkOp();

    await cursor.forEach(async doc => {
      await doc[arrayName].forEach(dup => {
        batch
          .find({
            _id: doc._id,
            [arrayName]: { $elemMatch: dup },
          })
          .updateOne({
            $unset: { [`${arrayName}.$`]: '' },
          });

        batch.find({ _id: doc._id }).updateOne({
          $pull: { [arrayName]: null },
        });
      });
    });

    console.log(`${batch.length} records found for ${arrayName}\n`);

    if (batch.length > 0) {
      await batch.execute();

      console.log(JSON.stringify(batch.batches, null, 2));
    }
  }

  run('courierHistory')
    .then(run('statusHistory'))
    .then(() => client.close())
    .catch(console.dir);
});