Working in NodeJS with Mongoose, though the methods are all the same.
So I have documents like (simplified):
{_id: 1, message: 'text1'}
{_id: 2, message: 'text2', reactions: [{_id: autoGen1, user: 'bob', reaction:'👍'},
{_id: autoGen2, user: 'bob', reaction:'👎'}
{_id: autoGen3, user: 'meg', reaction:'😵'}]}
I want to run an update query that finds a message and either
Adds the reaction if it doesn't exist
Removes the reaction if it does exist
The user + reaction make a "unique, compound key".
edit
Much testing and experimentation later, I clearly misunderstood how $cond operated and its use-case. Evaluating other options. Open to suggestions.
edit2
Removed my previous, bad attempts. Post was getting too long.
Current Solution: https://mongoplayground.net/p/cgk6qgZ6l9k
Solution based on link in the comments; modified due to array contents.
I really don't like this solution. It's long. It's ugly. Adding an element probably doesn't add the _id. Not an elegant solution overall, in my opinion, unlike the tidy versions linked on SO that deal with primitive arrays and not arrays of objects.
I tried using $elemMatch and $in without success in that playground.
Related
I have a Mongo collection that has two fields, let's say "name" and "randomString".
I want to create a random string for a name, only if it doesn't exist already. So the first request for { name: "SomeName" } will result in saving e.g. { name: "someName", randomString: "abc" }. The second request will do nothing.
Is there a mongo command for this? All I could find are things like findOneAndUpdate, replaceOne etc, who all support an optional "upsert" but their behavior on match is to update, I want the behavior on match to be do nothing.
I'm not looking for an if-then solution like in this question, as I have a race condition issue - I need to be able to get multiple requests simultaneously without updating the document or failing any of the requests.
Yes there is a command for this you can do this by using $addToSet method.
For more info please go through the given link: https://docs.mongodb.com/manual/reference/operator/update/addToSet/
PS: If you still have any confusion regarding this question please feel free to comment further.
Thanks
This is the solution I found in the end:
CustomerRandomString.findOneAndUpdate(
{ name: "someName" },
{
$setOnInsert: { randomString: generateRandomString() },
},
{ upsert: true },
);
The setOnInsert operator only applies when creating a new document, which is exactly what I needed.
EDIT: per the docs, this solution requires a unique index on the field in order to fully avoid duplicates.
You can easily do it using the $exists command to check for randomString field and then use $set in an aggregation pipeline to upsert that field.
db.collection.updateMany({"name":someName,"randomString":{$exists: false}},[{$set:{"randomString":"abcd"}}],{upsert:true})
If the condition query doesn't match with any documents, then it returns null.
Note: Aggregation pipeline works in updateMany() only from MongoDB version 4.2 and above.
Isolated data validation is pretty straight forward with something like joi. But what is a good way to solve validation that depends on other models, like given the following collection:
items:
[
{_id: ".....", title: "Product 1", in_stock: 3},
{_id: ".....", title: "Product 2", in_stock: 10},
....
]
And an "order" request like:
{
items:[
{_id: "....", quantity: 3},
{_id: "....", quantity: 6},
...
]
}
Now I want to check that all the items in the order request are in stock (quantity <= in_stock of the corresponding item). What would be good way of solving this?
I myself faced the same problem
The simplest solution is to use mongoose built in validation mechanism so that you don't
need to go ahead and create a lot of schemas in your application
It acctually makes it less robust and the code gets actually prettry hard to maintain as you grow
Joi is recommended when you are using a server db connectivity engine like mongojs
coz it does not have a proper validation mechanism
mongoose on the other hand is a pretty fast and robust mongoDb wrapper that comes with all the possible CRUDs, indexing aggregations and Schema creation.( you can just simply integrate your validations using mongoose validates and make your Schema centered and make your code less redundant
Use mongoose validator property and make a function using regex to check a white list
then throw en error if user input does not meet what u force them
I'm using Node.js + mongodb. I have few documents in my collection and i want to know does my collection have any document matched my condition. Of course i can simply use
myModel.find({ myField: someValue }) and check is anything comes or not. But i want to use solution like sql provides exists keyword? Help me, please
Edit: my bad. I forget to tell that "performance first".
MongoDB's $exists actually doesn't help you very much to find out if a certain document exists in your collection. It is used for example to give you all documents that have a specific field set.
MongoDB has no native support for an sql like exists. What you can use, however, is myModel.findOne({ myField: someValue }) and then check if it is null.
To enhance performance you can tell MongoDB to only load the object id via projection, like this:
myModel.findOne({ myField: someValue }, {_id: 1})
There is an exist mechanism in mongodb, I'll demonstrate a sample below.
For example below, I'm looking for records that have tomato.consensus fields and that it's empty, so I can delete them or avoid them. In case I was looking for "tomato.consensus": Dublin, I'd change Null to Dublin, to match that.
I hope this is helpful, if not fire away any questions
tomato
----consensus
db.movieDetails.updateMany({$and: [
{"tomato.consensus": {$exists: true} },
{"tomato.consensus": null} ] },
]})
I have a sample schema like this -
Comment.add({
text:String,
url:{type:String,unique:true},
username:String,
timestamp:{type:Date,default:Date}
});
Feed.add({
url:{type:String, unique:true },
username:String,
message:{type:String,required:'{PATH} is required!'},
comments:[Comment],
timestamp:{type:Date,default:Date}
});
Now, I don't want to expose the _id fields to the outside world that's why I am not sending it to the clients anywhere.
Now, I have two important properties in my comment schema (username,url)
What I want to do is update the content of the sub document that satisfies
feed.url
comment.url
comment.username
if the comment.username is same as my client value req.user.username then update the comment.text property of that record whose url was supplied by client in req.body.url variable.
One long and time consuming approach I thought is to first find the feed with the given url and then iterating over all the subdocuments to find the document which satisfies the comment.url==req.body.url and then check if the comment.username==req.user.username if so, update the comment object.
But, I think there must be an easier way of doing this?
I already tried -
db.feeds.update({"username":"harshitladdha93#gmail.com","comments.username":"harshitladdha3#gmail.com","comments.url":"test"},{$set:{"comments.$.text":"updated text 2"}})
found from http://www.tagwith.com/question_305575_how-to-find-and-update-subdocument-within-array-based-on-parent-property
but this updates even when the comments.url or comments.usernamematches other sub documents
and I also tried
db.feeds.distinct("comments._id",{"comments.url":req.body.url})
to find the _id of document associated with the url but it returns all the _id in the subdocument
First off - you should not rely on _id not being seen by the outside world in terms of security. This is a very bad idea for a multitude of reasons (primarily REST and also the fact that it's returned by default with all your queries).
Now, to address your question, what you want is the $elemMatch operator. This says that you're looking for something where the specified sub-document within an array matches multiple queries.
E.g.
db.feeds.update({
"username":"harshitladdha93#gmail.com",
comments: {
$elemMatch: {
username: "harshitladdha3#gmail.com",
url: "test"
}
}
}, {$set: {"comments.$.text":"updated text 2"}})
If you don't use $elemMatch you're saying that you're ok with the document if any of the comments match your query - i.e. if there is a comment by user "harshitladdha3#gmail.com", and separate comment has a url "test", the document will match unless you use $elemMatch
I'd like the unique _id field in one of my models to be relatively short: 8 letters/numbers, instead of the usual Mongo _id which is much longer. Having a short unique-index like this helps elsewhere in my code, for reasons I'll skip over here. I've successfully created a schema that does the trick (randomString is a function that generates a string of the given length):
new Schema('Activities', {
'_id': { type: String, unique: true, 'default': function(){ return randomString(8); } },
// ... other definitions
}
This works well so far, but I am concerned about duplicate IDs generated from the randomString function. There are 36^8 possible IDs, so right now it is not a problem... but as the set of possible IDs fills up, I am worried about insert commands failing due to a duplicate ID.
Obviously, I could do an extra query to check if the ID was taken before doing an insert... but that makes me cry inside.
I'm sure there's a better way to be doing this, but I'm not seeing it in the documentation.
This shortid lib https://github.com/dylang/shortid is being used by Doodle or Die, seems to be battle tested.
By creating a unique index on _id you'll get an error if you try to insert a document with a duplicate key. So wrap error handling around any inserts you do that looks for the error and then generates another ID and retries the insert in that case. You could add a method to your schema that implements this enhanced save to keep things clean and DRY.