MongoDB如何仅保留嵌入式文档指定字段且避免全量投影

Absolutely! You don’t have to list every top-level field in your projection—there’s a much cleaner way to achieve this. The key is to exclude the unwanted subfields inside the item document instead of explicitly including all top-level fields. Here’s how to do it:

Using find() with projection #

MongoDB lets you target and exclude specific subfields while keeping all top-level fields (and the parent item field itself) intact. Just set the subfields you don’t want to 0:

db.yourCollection.find(
  { _id: 1 }, // Your query filter (adjust this to match your needs)
  { "item.manyfieldshere": 0, "item.evenmore": 0 } // Exclude unwanted subfields from `item`
)

This query will return your full original document, but the item subdocument will only contain the name field. All other top-level fields (_id, name, alotofstuff, morestuff) stay exactly as they are.

If item has many unwanted subfields #

If item has dozens of fields you don’t want to list individually, use the aggregation pipeline to explicitly retain only item.name while preserving all top-level fields. This uses the $$ROOT variable to keep all existing fields, then overrides the item field:

db.yourCollection.aggregate([
  { $match: { _id: 1 } }, // Your filter
  {
    $project: {
      // Keep all original fields except the old `item`
      ...$$ROOT,
      // Replace `item` with only its `name` property
      item: { name: "$item.name" }
    }
  }
])

Note: The spread operator (...$$ROOT) works in MongoDB 3.6 and later. For older versions, you’d combine $addFields and $project, but the first method is simpler if you only have a few subfields to exclude.

内容的提问来源于stack exchange，提问作者hansmaad