Best way to replace a particular array element/value

[Projapati]

I have some strings in an array within my document.
I need to replace a particular value from the array.

Currently I am pulling out the current value and then pushing the new
value. It works.
db.col.update({_id:0},{$pull:{"field":"old_value"}})
db.col.update({_id:0},{$push:{"field":"new_value"}})

This involves 2 update operations.

Can it be done in atomic way like a[idx] = "new_value" like we do in C/
C#?

I tried $elemMatch but the positional operator $ seems not working.

Thanks

[Gaetan Voyer-Perrault]

> Can it be done in atomic way like a[idx] = "new_value" like we do in C/C#?

No this cannot be done atomically at the moment.

There are several JIRA tickets related to such a feature:

https://jira.mongodb.org/browse/SERVER-991

https://jira.mongodb.org/browse/SERVER-1050

https://jira.mongodb.org/browse/SERVER-1947

To follow the status or vote on these, you will need to log into JIRA with your account.

You can click on the "Vote for it" or "Watch it" links on the left side of page.

If you do not have an account, you can create one for free.

- Gates

--
You received this message because you are subscribed to the Google Groups "mongodb-user" group.
To post to this group, send email to mongod...@googlegroups.com.
To unsubscribe from this group, send email to mongodb-user...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/mongodb-user?hl=en.

[Projapati]

Thanks. Voted the jiras up.

On Apr 15, 4:38 pm, Gaetan Voyer-Perrault <ga...@10gen.com> wrote:
> > Can it be done in atomic way like a[idx] = "new_value" like we do in C/C#?
>
> No this cannot be done atomically at the moment.
>
> There are several JIRA tickets related to such a feature:https://jira.mongodb.org/browse/SERVER-991

> <https://jira.mongodb.org/browse/SERVER-991>https://jira.mongodb.org/browse/SERVER-1050

[Keith Branton]

This depends on what your data looks like, but if it resembles this:

> db.a.insert({_id:0,field:["one_value","old_value","another_value"]})

> db.a.find()

{ "_id" : 0, "field" : [ "one_value", "old_value", "another_value" ] }

then this update:

> db.a.update({_id:0,field:"old_value"},{$set:{"field.$":"new_value"}})

results in this:

> db.a.find()                                                          

{ "_id" : 0, "field" : [ "one_value", "new_value", "another_value" ] }

does that help?

If not then an example of the data your are trying to manipulate would be useful.

[Projapati]

Keith,
I am pretty sure I tried this before posting with my tags & it didn't
work.
In general, your answers are great. I will retry this and let you
know.

Thanks

[Projapati]

It works! I am not sure what I missed earlier.
One side effect of this approach is that it will allow duplicate tags.
If there is a tag "abc" and user replaces "xyz" with "abc", it will
create 2 "abc" tags.

On the other hand if I do with 2 updates, the problem is not there.
db.albums.update({_id:0},{$pull:{"tags":"xyz"}})
db.albums.update({_id:0},{$addToSet:{"tags":"abc"}})

[Keith Branton]

That's true, though with that approach, unless you are only executing the second update if the first update actually pulled a value, then your second update will add the value regardless, even if the first value wasn't in the array before. 

Not sure if that's a behavior you want, but my method (which works much more like a[idx] = "new_value") wouldn't do that.

It all depends on what you really need.

[Keith Branton]

Maybe doing it like this would be better:

update({id, array has old_item, array doesn't have new_item},{$set old item to new item})

update({id, array has old_item and new_item},{$pull old_item})

it's still two updates, but the difference is that only one would actually change the document, so the other would be very fast since it wouldn't do anything. Actually you could detect if the first one does update a record and not bother running the second if it does, but you could also just run these fire and forget if you don't care to wait.

so your code would look like:

> db.a.remove()                                                                   

> db.a.insert({_id:0,field:["one_value","old_value","old_value2","another_value"]})  

> db.a.find()                                                                      

{ "_id" : 0, "field" : [ "one_value", "old_value", "old_value2", "another_value" ] }

> db.a.update({_id:0,field:{$in:["old_value"],$ne:"new_value"}},{$set:{"field.$":"new_value"}})

> db.a.update({_id:0,field:{$in:["old_value","new_value"]}}, {$pull:{field:"old_value"}})      

> db.a.find()                                                                                  

{ "_id" : 0, "field" : [ "one_value", "new_value", "old_value2", "another_value" ] }

> db.a.update({_id:0,field:{$in:["old_value2"],$ne:"new_value"}},{$set:{"field.$":"new_value"}})

> db.a.update({_id:0,field:{$in:["old_value2","new_value"]}}, {$pull:{field:"old_value2"}})     

> db.a.find()                                                                              

{ "_id" : 0, "field" : [ "one_value", "new_value", "another_value" ] }

> 

but this approach doesn't add new_value if it's not already in the list, so it depends if you want that too (I'm not sure I can see why you would want to add a tag and optionally replace an existing tag in the same operation in a tagging use case)