How do I update MongoDB document fields only if they don't exist?

I have collection foo with documents like:

{site_id: 'xxx', title: {ru: 'a', en: 'b'}, content: {ru: 'a', en: 'b'}}
{site_id: 'xxx', title: {ru: 'c', de: 'd'}, content: {ru: 'c', de: 'd'}}

I need to update multiple fields which are can exists or not:

db.foo.update(
    { site_id: 'xxx'},
    { $set: {'title.de': '', 'content.de': ''}},
    {multi: true}
)

But I need something like $set which will not overwrite value if it exists.

You can add a query to your update statement:

db.foo.update({'title.de': {$exists : false}}, {$set: {'title.de': ''}})

Update

For your modified question my solution looks like this - would that work for you? (If not, why?)

db.foo.update({site_id: 'xxx', 'title.de': {$exists : false}}, {$set: {'title.de': ''}, {multi: true})
db.foo.update({site_id: 'xxx', 'content.de': {$exists : false}}, {$set: {'content.de': ''}}, {multi: true})

Starting Mongo 4.2, db.collection.update() can accept an aggregation pipeline, finally allowing the update/creation of a field based on another field:

This way, we can move field checks within the update stage rather than within the match stage, thus making it a one-pass update:

// { site_id: "xxx", title: { ru: "a", en: "b" }, content: {} }
// { site_id: "xxx", title: { ru: "c", de: "d" }, content: { ru: "c" } }
db.collection.update(
  { site_id: "xxx" },
  [{ $set: {    "title.de": { $cond: [ { $not: ["$title.de"] }, "", "$title.de" ] },
    "content.ru": { $cond: [ { $not: ["$content.ru"] }, "", "$content.ru" ] }
  }}],
  { multi: true }
)
// { site_id: "xxx", title: { ru: "a", en: "b", de: "" }, content: { ru: "" } }
// { site_id: "xxx", title: { ru: "c", de: "d"         }, content: { ru: "c" } }

• The first part { site_id: "xxx" } is the match query, filtering which documents to update.

• The second part [{ $set: { ... } }] is the update aggregation pipeline (note the squared brackets signifying the use of an aggregation pipeline). $set is a new aggregation operator and an alias of $addFields. The rest of this stage checks with $cond if the title.de exists, and if yes, then keep it as it is, or otherwise create it as an ''.

• Don't forget { multi: true }, otherwise only the first matching document will be updated.

there is a update field operator $setOnInsert which meets your requirement. Please read the documents here: https://docs.mongodb.com/manual/reference/operator/update/setOnInsert/#up._S_setOnInsert

I have a solution for one particular case, but may be it helps someone.

My case was: Update several fields, among which was a field that had to be updated only once (lets call it "Date_of_first_update").

> db.test.find();
{ "_id" : ObjectId("57f298fdeb30478a033c70e4"), "a" : "1", "b" : "2" }

First update:

> db.test.updateOne({ "_id" : ObjectId("57f298fdeb30478a033c70e4")}, 
  {$set: {a: 100, b: 200 }, $min : {'Date_of_first_update' : (new Date())  }});

Result: 'a', 'b' updated, 'Date_of_first_update' is set.

{ "_id" : ObjectId("57f298fdeb30478a033c70e4"), "a" : 100, "b" : 200, "Date_of_first_update" : ISODate("2016-10-03T**17:47:43**.570Z") }

Second update:

> db.test.updateOne({ "_id" : ObjectId("57f298fdeb30478a033c70e4")}, 
  {$set: {a: 400, b: 800 }, $min : {'Date_of_first_update' : (new Date()) }});

Result: 'a', 'b' updated, 'Date_of_first_update' left unchanged, as I needed!!!

{ "_id" : ObjectId("57f298fdeb30478a033c70e4"), "a" : 400, "b" : 800, "Date_of_first_update" : ISODate("2016-10-03T**17:47:43**.570Z") } 

If anyone encounters the problem as me:

My solution is to set _id only if the update results in insert a new on (upsert: true)

return {
    updateOne: {
        filter: {
            email: shadowUser.email,
        },
        update: {
            $set: user,
            $setOnInsert: { _id: shadowUser._id },
        },
        upsert: true,
    },
};

Even though the answer given essentially outlines the approach, you can do this sort of thing with MongoDB 2.6 or greater due to the implementation there that supports "bulk updates".

This are still, "atomically speaking", separate update statements. But you can submit them "over the wire" in one go. Which at least makes sure that the latency between the updates is much shorter as they are executed on the server:

var bulk = db.foo.initializeBulkOrderedOp();
bulk.find({ "site_id": "xxx",
    "title.de": { "$exists" false } })
    .update({ "$set": { "title.de": "" } });
bulk.find({ "site_id": "xxx", 
    "content.de": { "$exists" false } })
    .update({ "$set": { "content.de": "" } });
bulk.execute();

So that is actually one round trip to the server as everything only sends on .execute()

But in your present form (though this may not be a accurate representation of your data), you can actually "re-structure" in order to do this in a single operation. So if your documents looked like this:

{ 
    "site_id": "xxx",
    "docs": [
        { "title": "a", "content": "a", "lang": "ru" },
        { "title": "b", "content": "b", "lang": "en" }
    ]
},
{
    "site_id": "xxx",
    "docs": [
        { "title": "c", "content": "c", "lang": "ru" },
        { "title": "d", "content": "d", "lang": "de" }
    ]
}

Then the following works by the rule of $addToSet where the "set" element would be "unique":

db.foo.update(
    { "site_id": "xxx" },
    { "$addToSet": { "docs": { "title": "d", content: "d", "lang": "de" } } },
    { "multi": true }
)

Or even without the logic there and just checking for presence:

db.foo.update(
    { "site_id": "xxx", "docs.lang": { "$ne": "de" } },
    { "$push": { "docs": { "title": "", "content": "", "lang": "de" } } },
    { "multi": true }

)

Which in that last case would result in this:

{
    "_id" : ObjectId("53c936265117367f5ff2038b"),
    "site_id" : "xxx",
    "docs" : [
            {
                    "title" : "a",
                    "content" : "a",
                    "lang" : "ru"
            },
            {
                    "title" : "b",
                    "content" : "b",
                    "lang" : "en"
            },
            {
                    "title" : "",
                    "content" : "",
                    "lang" : "de"
            }
    ]
}
{
    "_id" : ObjectId("53c936265117367f5ff2038c"),
    "site_id" : "xxx",
    "docs" : [
            {
                    "title" : "c",
                    "content" : "c",
                    "lang" : "ru"
            },
            {
                    "title" : "d",
                    "content" : "d",
                    "lang" : "de"
            }
    ]
}

So the choice is there to either "handle" things differently or otherwise just change your schema to accommodate the sort of updates you want to do atomically.

@nutlike's answer does solve the issue, however, if you would like to update multiple fields on the item it will require many database operations. In short, what you want is not exactly possible.

If your doc has more updates to do than you'd like to do one at a time (any greater than 2, IMO), then you should just get the document, update the fields and then save it. This is what I do on a couple of OAuth user creating/updating routes.

As update query is deprecated use updateOne or updateMany as per your requirement.

updateMany to update all at once.

db.foo.updateMany({'title.de': {$exists : false}}, {$set: {'title.de': ''}})