I have following document:
{
_id : 21353456,
username : "xy",
text : "asdf",
comments : [
{
username : "User1",
text : "hi",
},
{
username : "User2",
text : "hi1",
},
{
username : "User3",
text : "hi2",
},
{
username : "User4",
text : "hi3",
}
]
}
Now I want to get the username, text and comments with aggregation and project. In addition I also want a boolean if the comments array contains an username with "User1". I have this, but it doesn't work.
db.posttest.aggregate(
[
{
$project:
{
username: 1,
text: 1,
comments : 1,
hasComment: { $eq: [ "comments.$.username", "User1" ] },
_id: 0
}
}
]
)
To achieve this you would need to first unwind the comments, and then use a group with a little trick. If you want to omit the _id, then you would also need to do a simple project. Here is the full aggregation pipeline:
db.posttest.aggregate([
{ $unwind : "$comments" },
{ $group : {
_id : "$_id",
username : { $first : "$username" },
text : { $first : "$text" },
comments : { $push : "$comments" },
hasComments : { $max : { $eq : [ "$comments.username", "User1" ] } }
}},
{ $project : { _id : false } }
])
An explanation is following.
First, we need to get rid of an array (comments). To do this we unwind the record; it gives us four records:
{
"_id" : 21353456,
"username" : "xy",
"text" : "asdf",
"comments" : {
"username" : "User1",
"text" : "hi"
}
},
{
"_id" : 21353456,
"username" : "xy",
"text" : "asdf",
"comments" : {
"username" : "User2",
"text" : "hi1"
}
},
{
"_id" : 21353456,
"username" : "xy",
"text" : "asdf",
"comments" : {
"username" : "User3",
"text" : "hi2"
}
},
{
"_id" : 21353456,
"username" : "xy",
"text" : "asdf",
"comments" : {
"username" : "User4",
"text" : "hi3"
}
}
Now we can group all the records into one applying a function to each field. First, we need to give criteria, the 'group by' field (or set of fields). In our case, it is simply the id: _id: "$_id".
Then, for each field, we need to make a decision on how to include it into the resulting record. We have few fields: username, text, and comments. For each four records the username and text are the same, so we can easily pick any of them, namely $first or $last.
comments, however, are different. We want to preserve all of them so that we $push each one back.
The hasComments is a bit tricky here: we need to check if at least one comment.username contains the username. We can use $eq: [...] here, it will give us some array, e.g. [true, false, false, false] or [false, false, true, false]. We would need to pick which value goes into the resulting record. In this case, we can use neither $first nor $last. However, $max will give us an appropriate result.
I know this is an old question and also has an accepted answer, but there is an easiest way than using $unwind and $group, only using $project like this:
The trick here is compare the intersection beetween comments array and desired value (in this case User1. If the interesection is greater than 0 (i.e. exists the value) then the field existsUser will be true, otherwise false.
{
"$project": {
"comments": 1,
"existsUser": {
"$cond": {
"if": {
"$gt": [
{
"$size": {
"$setIntersection": [
"$comments.username",
[
"User1"
]
]
}
},
0
]
},
"then": true,
"else": false
}
}
}
}
Example here
Related
How can I insert an value in "total" array where pid = "123"
{
"_id" : ObjectId("625bc1983016ed5208bbdf90"),
"NAME" : "XYZ",
"data" : [
{
"pid" : "123",
"total" : [ ]
},
{
"pid" : "456",
"total" : [ ]
}
]
}
The right way is:
db.collection.update({
data: {
$elemMatch: {
pid: "123"
}
}
},
{
$set: {
"data.$.buyers": "some name"
}
})
$elemMatch allows you to match more than one component within the same array element and then you can use $set to update the specific element.
You can run it here to check the same query https://mongoplayground.net/p/lveZw-8wkZ0
You can also have a look at documentation of $elemMatch in this link.
Let me know if you face any issue.
I can't make $nin work with nested arrays, can you guys spot any issue with this query?
I'm basically trying to update the status of every document under items to "closed" in case their hash field is not in a list of hashes provided.
db.getCollection('projects').update({
name: 'test',
'issues.hash': { $nin: [
'8ff28fcc9cbf10c9b690bb331e5609efbd3c526be4f536ebca02cc51bd63eac7',
'd5368ad5658ec11103796255d127d26da7f3324cdedbd124bdd5db50812d588e',
'37298229097785ebc9d419cc1a3f13e0d090a15ceb9a8e6bea3505366902556d',
'fad290f2ddd0e097e4098c3b2c3d65611406cf208a3f86924d45c7736393b44b'
]}
},
{
$set: { "issues.$.status": "closed" }
}
)
This is the data:
{
"_id" : ObjectId("611bb4d2ee06769a5f6d906d"),
"name" : "test",
"issues" : [
{
"_id" : ObjectId("611bb4d20b2fb200167aa588"),
"status" : "open",
"hash" : "8ff28fcc9cbf10c9b690bb331e5609efbd3c526be4f536ebca02cc51bd63eac7"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa589"),
"status" : "open",
"hash" : "3b83e469049e46b16d3471a188d3f5e3ddbf6b296995a71765bbf17b7289e6ea"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa58a"),
"status" : "open",
"hash" : "bef5f50628b669b9930b89cdc040361b9c8cc2b4aab3c2059c171786d38d507e"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa58b"),
"status" : "open",
"hash" : "1b4a91eb5de97d6ad7493b6e1ffa48a2a648084b4af7b37916c723533a07c37c"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa58c"),
"status" : "open",
"hash" : "bb64ba7b2612856dcd95c3ac2fad3f7368e5d463168545b12f4c869af56b55b7"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa58d"),
"status" : "open",
"hash" : "1d5fc04739b10414dea8d327998df4f200f47ce57da243bd578d4ae102f2d670"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa58e"),
"status" : "open",
"hash" : "d5368ad5658ec11103796255d127d26da7f3324cdedbd124bdd5db50812d588e"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa58f"),
"status" : "open",
"hash" : "37298229097785ebc9d419cc1a3f13e0d090a15ceb9a8e6bea3505366902556d"
},
{
"_id" : ObjectId("611bb4d20b2fb200167aa590"),
"status" : "open",
"hash" : "fad290f2ddd0e097e4098c3b2c3d65611406cf208a3f86924d45c7736393b44b"
}
]
}
And the is my result:
Updated 0 record(s) in 12ms
Thank you!
You have to use arrayFilters in this way:
db.collection.update({
"name": "test"
},
{
"$set": {
"issues.$[element].status": "closed"
}
},
{
"arrayFilters": [
{
"element.hash": {
"$nin": [
"8ff28fcc9cbf10c9b690bb331e5609efbd3c526be4f536ebca02cc51bd63eac7",
"d5368ad5658ec11103796255d127d26da7f3324cdedbd124bdd5db50812d588e",
"37298229097785ebc9d419cc1a3f13e0d090a15ceb9a8e6bea3505366902556d",
"fad290f2ddd0e097e4098c3b2c3d65611406cf208a3f86924d45c7736393b44b"
]
}
}
]
})
Example here.
Note that update query has the format: update(query, update, options) (Check the docs).
So with your find query mongo doesn't find anything. Check this example.
This is why you are telling mongo: "Give me a DOCUMENT where name is test and issues array NOT contains a field called hash with these values".
So, as mongo search by the whole document, there is no any document where hash value is not on the $nin array.
As another example to exaplain better: Check this example where hash value is 1. In this case, find query works because it matches two conditions:
There is a field name with value test
There is not any field hash into issues with values into $nin array.
You can use arrayFilters, like this:
db.collection.update({
"name": "test"
},
{
"$set": {
"issues.$[elem].status": "closed"
}
},
{
"multi": true,
"arrayFilters": [
{
"elem.hash": {
"$nin": [
"8ff28fcc9cbf10c9b690bb331e5609efbd3c526be4f536ebca02cc51bd63eac7",
"d5368ad5658ec11103796255d127d26da7f3324cdedbd124bdd5db50812d588e",
"37298229097785ebc9d419cc1a3f13e0d090a15ceb9a8e6bea3505366902556d",
"fad290f2ddd0e097e4098c3b2c3d65611406cf208a3f86924d45c7736393b44b"
]
}
}
]
})
Here is the working example: https://mongoplayground.net/p/8wZkmlBgKiq
[{
"username":"user1",
"products":[
{"productID":1,"itemCode":"CODE1"},
{"productID":2,"itemCode":"CODE1"},
{"productID":3,"itemCode":"CODE2"},
]
},
{
"username":"user2",
"products":[
{"productID":1,"itemCode":"CODE1"},
{"productID":2,"itemCode":"CODE2"},
]
}]
I want to find all the "productID" of "products" for "user1" such that "itemCode" for the product is "CODE1".
What query in mongoDB should be written to do so?
If you only need to match a single condition, then the dot notation is sufficient.
In Mongo shell:
db.col.find({"products.itemCode" : "CODE1", "username" : "user1"})
This will return all users with nested product objects having itemCode "CODE1".
Updated
Wasn't clear on your requirements at first but this should be it.
If you want each product as a separate entry, then you would need to use the aggregate framework. First split the entries in the array using $unwind, then use $match for your conditions.
db.col.aggregate(
{ $unwind: "$products" },
{ $match: { username: "user1", "products.itemCode": "CODE1" } }
);
response:
{ "_id" : ObjectId("57cdf9c0f7f7ecd0f7ef81b6"), "username" : "user1", "products" : { "productID" : 1, "itemCode" : "CODE1" } }
{ "_id" : ObjectId("57cdf9c0f7f7ecd0f7ef81b6"), "username" : "user1", "products" : { "productID" : 2, "itemCode" : "CODE1" } }
The answer to your question is
db.col.aggregate([
{ $unwind: "$products" },
{ $match: { username: "user1", "products.itemCode": CODE1 } },
{ $project: { _id: 0, "products.productID": 1 } }
]);
In my case didn't work without [ ] tags.
You need multiple filter for this like below which is nothing but AND condition (assuming your collection name is collection1)
db.collection1.find({"username":"user1", "products.itemCode" : "CODE1"})
I have a collection of documents, each has a field which is an array of subdocuments, and all subdocuments have a common field 'status'. I want to find all documents that have the same status for all subdocuments.
collection:
{
"name" : "John",
"wives" : [
{
"name" : "Mary",
"status" : "dead"
},
{
"name" : "Anne",
"status" : "alive"
}
]
},
{
"name" : "Bill",
"wives" : [
{
"name" : "Mary",
"status" : "dead"
},
{
"name" : "Anne",
"status" : "dead"
}
]
},
{
"name" : "Mohammed",
"wives" : [
{
"name" : "Jane",
"status" : "dead"
},
{
"name" : "Sarah",
"status" : "dying"
}
]
}
I want to check if all wives are dead and find only Bill.
You can use the following aggregation query to get records of person whose wives are all dead:
db.collection.aggregate(
{$project: {name:1, wives:1, size:{$size:'$wives'}}},
{$unwind:'$wives'},
{$match:{'wives.status':'dead'}},
{$group:{_id:'$_id',name:{$first:'$name'}, wives:{$push: '$wives'},size:{$first:'$size'},count:{$sum:1}}},
{$project:{_id:1, wives:1, name:1, cmp_value:{$cmp:['$size','$count']}}},
{$match:{cmp_value:0}}
)
Output:
{ "_id" : ObjectId("56d401de8b953f35aa92bfb8"), "name" : "Bill", "wives" : [ { "name" : "Mary", "status" : "dead" }, { "name" : "Anne", "status" : "dead" } ], "cmp_value" : 0 }
If you need to find records of users who has same status, then you may remove the initial match stage.
The most efficient way to handle this is always going to be to "match" on the status of "dead" as the opening query, otherwise you are processing items that cannot possibly match, and the logic really quite simply followed with $map and $allElementsTrue:
db.collection.aggregate([
{ "$match": { "wives.status": "dead" } },
{ "$redact": {
"$cond": {
"if": {
"$allElementsTrue": {
"$map": {
"input": "$wives",
"as": "wife",
"in": { "$eq": [ "$$wife.status", "dead" ] }
}
}
},
"then": "$$KEEP",
"else": "$$PRUNE"
}
}}
])
Or the same thing with $where:
db.collection.find({
"wives.status": "dead",
"$where": function() {
return this.wives.length
== this.wives.filter(function(el) {
el.status == "dead";
}).length;
}
})
Both essentially test the "status" value of all elements to make sure they match in the fastest possible way. But the aggregate pipeline with just $match and $redact should be faster. And "less" pipeline stages ( essentially each a pass through the data ) means faster as well.
Of course keeping a property on the document is always fastest, but it would involve logic to set that only where "all elements" are the same property. Which of course would typically mean inspecting the document by loading it from the server prior to each update.
I have the following mongo document, which is part of a bigger document called attributes, which also has Colour and Size
> db.attributes.find({'name': {'en-UK': 'Fabric'}}).pretty()
{
"_id" : ObjectId("543261cda14c971132fa2b91"),
"values" : [
{
"source" : [
{
"_id" : ObjectId("543261cda14c971132fa2b79"),
"name" : {
"en-UK" : "Combed Cotton"
}
},
],
"name" : [
{
"_id" : ObjectId("543261cda14c971132fa2b85"),
"name" : {
"en-UK" : "Brushed 3-ply"
}
},
{
"_id" : ObjectId("543261cda14c971132fa2b8f"),
"name" : {
"en-UK" : "Plain Weave"
}
},
{
"_id" : ObjectId("543261cda14c971132fa2b90"),
"name" : {
"en-UK" : "1x1 Rib"
}
}
]
}
],
"name" : {
"en-UK" : "Fabric"
}
}
I am trying to return the _id for a sub document and have the following:
db.attributes.aggregate([
{ '$match': {'name.en-UK': 'Fabric'} },
{ '$unwind' : '$values' },
{ '$project': { 'name' : '$values.name'} },
{ '$match': { '$and': [{"name.name.en-UK" : "1x1 Rib"} ] }}
])
What is the correct way to do this?
Also, the values of Fabric is an array with two items, source and name, but if I populate it like:
> db.attributes.find({'name': {'en-UK': 'Fabric'}}).pretty()
{
"_id" : ObjectId("543261cda14c971132fa2b91"),
"values" : {
"source" : [{ ... }]
"name": [{ ... }]
}
}
I get the following error
"errmsg" : "exception: $unwind: value at end of field path must be an array"
But if I wrap it inside a square brackets this then works, so that
> db.attributes.find({'name': {'en-UK': 'Fabric'}}).pretty()
{
"_id" : ObjectId("543261cda14c971132fa2b91"),
"values" : [{
"source" : [{ ... }],
"name": [{ ... }]
}]
}
what am I missing as values is an array of two objects, source and name each containing a list of arrays
Any advice much appreciated
What you seem to be "missing" here is that "some" of your documents do either not contain a "value" property at all or at the very least it is "not an array". This is the basic context of the error you have been given.
Fortunately there are a couple of ways to get around this. Namely, either "testing" for the presence of an array when submitting you original query. Or actually "substituting" the missing element for some kind of array when processing the pipeline.
Here are both approaches in what is effectively an redundant form since the first $match condition really sorts this out:
db.attributes.aggregate([
{ "$match": {
"name.en-UK": "Fabric",
"values.0": { "$exists": true }
}},
{ "$project": {
"name": 1,
"values": { "$ifNull": [ "$values", [] ] }
}},
{ "$unwind": "$values" },
{ "$unwind": "$values.name" },
{ "$match": { "values.name.name.en-UK" : "1x1 Rib" }}
])
So as I said. Really redundant in that the initial $match actually asks if an "initial array element" actually exists. Which kind of means that there is an array there.
The second $project phase actually uses the $ifNull operator to "fill in" a value ( or basically an empty array ) where the tested element does not exist. We tested for that anyway before, but this demonstrates the different approaches.
But the basic idea id either "avoiding" or "filling-in" where your document does not have the expected data that you want to process. Which is the cause of your error.