We have documents that were originally stored using the StringObjectIdGenerator type for _id. Newer data will be saved using the default ObjectIdGenerator. We would like to be able to migrate the existing data by converting the _id to the new data type. Is this possible?
3 Answers
Yes, it's possible to change data type, we need to find all documents with string id, using the iterator we can clone a document by changing its _id, after cloning we just need to delete the old document.
Below is for a small document, for large collection you may need to use db.collection.initializeUnorderedBulkOp() for bulk insert/delete
db.i.find({_id : {$type : 2}}). //find all string _id
forEach(function(d){
var id = ObjectId(d._id); //_id to ObjectId
var oldId = d._id; // _id
d._id = id;
db.i.insert(d); // clone doc with new Id
db.i.remove({ _id : oldId }); // delete old doc
}
)
Example
> db.i.insertMany([{},{},{},{_id : "4a5ec389fee4c182f509f3ba"}, {_id : "4b5ec389fee4c182f509f3ba"}])
{
"acknowledged" : true,
"insertedIds" : [
ObjectId("5a5ec9edfee4c182f509f3c9"),
ObjectId("5a5ec9edfee4c182f509f3ca"),
ObjectId("5a5ec9edfee4c182f509f3cb"),
"4a5ec389fee4c182f509f3ba",
"4b5ec389fee4c182f509f3ba"
]
}
> db.i.find()
{ "_id" : ObjectId("5a5ec9edfee4c182f509f3c9") }
{ "_id" : ObjectId("5a5ec9edfee4c182f509f3ca") }
{ "_id" : ObjectId("5a5ec9edfee4c182f509f3cb") }
{ "_id" : "4a5ec389fee4c182f509f3ba" }
{ "_id" : "4b5ec389fee4c182f509f3ba" }
> db.i.find({_id : {$type : 2}}).
... forEach(function(d){
... var id = ObjectId(d._id);
... var oldId = d._id;
... d._id = id;
... db.i.insert(d);
... db.i.remove({ _id : oldId });
... }
... )
> db.i.find()
{ "_id" : ObjectId("5a5ec9edfee4c182f509f3c9") }
{ "_id" : ObjectId("5a5ec9edfee4c182f509f3ca") }
{ "_id" : ObjectId("5a5ec9edfee4c182f509f3cb") }
{ "_id" : ObjectId("4a5ec389fee4c182f509f3ba") }
{ "_id" : ObjectId("4b5ec389fee4c182f509f3ba") }
>
1 Comment
Udit Bhardwaj
the remove statement should be above the insert statement, as inserting before removing existing document on indexed collection can result in duplicate index error i.e. unique index.