I'm using findAndModify command to maintain a list of undo/redo commands. I'm using "fields" tag to specify what the "before" values were. This is then used to build undo command. ie:
cmd:
{
findAndModify : "aaa",
query : { _id: ObjectId('5215f7d1fe789bb17427bde9') },
update : { "$set" : { "v1" : 200 } },
fields : { v1 : 1, _id : 0 }
}
built result:
"Do" : { "$set" : { "v1" : 200 } },
"Undo" : { "$set" : { v1" : 100 } }
However, I cannot make this work for arrays. I do a set command like this:
{ $set : "myArrayVar.3" : 100 }
I've tried using projections like:
{ "myArrayVar.3" : 1 }
{ "myArrayVar.$" : 1 }
but both return just empty array braces:
{ myArrayVar : [] }
What am I missing?
You can't use numeric array indexes in projections, but you can use $slice instead:
fields: { myArrayVar: { $slice: [2, 1] } }
That would include just the third element (skip 2, take 1).
Related
Have a field of type "String" that contain representation of an object/document
" {"a":35,b:[1,2,3,4]}"
I know is a strange construct but i can't change it.
my goal would be to extract for example the value of "a".
As the document represented by the string are nested and repeated a regex doesnt fit.
So how can i convert in a mongo db aggregation/query this String to object so that i can process it in a following aggregation step?
(could extract string with python make a dict and extract infos, but i'd like to stay inside the aggregation pipeline and so having better performance)
In 4.4 this works
db.target.aggregate([{$project: {
X: "$AD_GRAPHIC",
Y : {
$function : {
body: function(jsonString) {
return JSON.parse(jsonString)
},
args: [ "$AD_GRAPHIC"],
lang: "js"
}
}
}
}])
Basically use the $function operator to invoke the JSON parser. (assumes you have enabled Javascript)
Results
{ "_id" : ObjectId("60093dc8f2c829000e38a8d0"), "X" : "{\"alias\":\"MEDIA_DIR\",\"path\":\"modem.jpg\"}", "Y" : { "alias" : "MEDIA_DIR", "path" : "modem.jpg" } }
{ "_id" : ObjectId("60093dc8f2c829000e38a8d1"), "X" : "{\"alias\":\"MEDIA_DIR\",\"path\":\"monitor.jpg\"}", "Y" : { "alias" : "MEDIA_DIR", "path" : "monitor.jpg" } }
{ "_id" : ObjectId("60093dc8f2c829000e38a8d2"), "X" : "{\"alias\":\"MEDIA_DIR\",\"path\":\"mousepad.jpg\"}", "Y" : { "alias" : "MEDIA_DIR", "path" : "mousepad.jpg" } }
{ "_id" : ObjectId("60093dc8f2c829000e38a8d3"), "X" : "{\"alias\":\"MEDIA_DIR\",\"path\":\"keyboard.jpg\"}", "Y" : { "alias" : "MEDIA_DIR", "path" : "keyboard.jpg" } }
>
There's no native way in the MongoDB engine to parse a blob of JSON from a field. However, I'd recommend just doing it client-side in your language of choice and then if required save it back.
Alternatively, if your data is too big and still needs to aggregate it you could use regex and project out the required fields from the JSON to then use them later to filter etc...
For example if we insert the following document:
> db.test.insertOne({ name: 'test', blob: '{"a":35,b:[1,2,3,4]}' })
{
"acknowledged" : true,
"insertedId" : ObjectId("5ed9fe21b5d91941c9e85cdb")
}
We can then just project out the array with some regex:
db.test.aggregate([
{ $addFields: { b: { $regexFind: { input: "$blob", regex: /\[(((\d+,*))+)\]/ } } } },
{ $addFields: { b: { $split: [ { $arrayElemAt: [ "$b.captures", 0 ] }, "," ] } } }
]);
{
"_id" : ObjectId("5ed9fe21b5d91941c9e85cdb"),
"name" : "test",
"blob" : "{\"a\":35,b:[1,2,3,4]}",
"b" : [
"1",
"2",
"3",
"4"
]
}
This means we can do some filtering, sorting and any of the other aggregation stages.
You could just use JSON.parse()
For example
db.getCollection('system').find({
a: JSON.parse('{"a":35,b:[1,2,3,4]}').a
})
I have the following document structure in a MongoDB collection :
{
"A" : [ {
"B" : [ { ... } ]
} ]
}
I'd like to update this to :
{
"A" : [ {
"B" : [ { ... } ],
"x" : [],
"y" : { ... }
} ]
}
In other words, I want the "x" and "y" fields to be added to the first element of the "A" array without loosing "B".
Ok as there is only one object in A array you could simply do as below :
Sample Collection Data :
{
"_id" : ObjectId("5e7c3cadc16b5679b4aeec26"),
A:[
{
B: [{ abc: 1 }]
}
]
}
Query :
/** Insert new fields into 'A' array's first object by index 0 */
db.collection.updateOne(
{ "_id" : ObjectId("5e7c3f77c16b5679b4af4caf") },
{ $set: { "A.0.x": [] , "A.0.y" : {abcInY :1 }} }
)
Output :
{
"_id" : ObjectId("5e7c3cadc16b5679b4aeec26"),
"A" : [
{
"B" : [
{
"abc" : 1
}
],
"x" : [],
"y" : {
"abcInY" : 1.0
}
}
]
}
Or Using positional operator $ :
db.collection.updateOne(
{ _id: ObjectId("5e7c3cadc16b5679b4aeec26") , 'A': {$exists : true}},
{ $set: { "A.$.x": [] , "A.$.y" : {abcInY :1 }} }
)
Note : Result will be the same, but functionally when positional operator is used fields x & y are inserted to first object of A array only when A field exists in that documents, if not this positional query would not insert anything (Optionally you can check A is an array condition as well if needed). But when you do updates using index 0 as like in first query if A doesn't exist in document then update would create an A field which is an object & insert fields inside it (Which might cause data inconsistency across docs with two types of A field) - Check below result of 1st query when A doesn't exists.
{
"_id" : ObjectId("5e7c3f77c16b5679b4af4caf"),
"noA" : 1,
"A" : {
"0" : {
"x" : [],
"y" : {
"abcInY" : 1.0
}
}
}
}
However, I think I was able to get anothe#whoami Thanks for the suggestion, I think your first solution should work. However, I think I was able to get another solution to this though I'm not sure if its better or worse (performance wise?) than what you have here. My solution is:
db.coll.update( { "_id" : ObjectId("5e7c4eb3a74cce7fd94a3fe7") }, [ { "$addFields" : { "A" : { "x" : [ 1, 2, 3 ], "y" : { "abc" } } } } ] )
The issue with this is that if "A" has more than one array entry then this will update all elements under "A" which is not something I want. Just out of curiosity is there a way of limiting this solution to only the first entry in "A"?
I'm working on mongo, in this time in need to remove a specific element from array nested in another one, the structure is like the following:
{
"_id" : ObjectId("5e616314946b6d3ac4ed8252"),
"uid" : "5db069478556622b4a0adca5",
"areas" : [
{
"name" : "mexico",
"elements" : [
"23452345",
"24454675"
],
},
{
"name" : "usa",
"elemets" : [
"123123123",
"1234334"
],
}
]
}
So I need a query that lets me remove any element in "elements" property.
I tried with this query but I couldn't reach it.
db.areas.update(
{ },
{ $pull:
{ areas: {
$elemMatch: {
elements: { $eq : "123123123" }
}
}
}
},
{ multi:false }
);
This just said
WriteResult({ "nMatched" : 1, "nUpserted" : 0, "nModified" : 0 })
Here is the update for removing the specific array element in a sub-array (elements).
Note that to remove (pull) the specified element (elements value "24454675"), you need to specify the outer-array's element matching condition (note that there are two elements in the outer-array field areas).
db.areas.update(
{ "areas.name": "mexico" },
{ $pull: { "areas.$.elements": "24454675" } }
)
NOTE: If you don't specify the { "areas.name": "mexico" } condition, the whole sub-document of the areas array will be removed.
I'm trying to update an existing document in a MongoDb. There are many explanations how to do this if you want to update or add key/value pairs on the first level. But in my use-case, I need to create with the first updateOne (with upsert option set) a document with the following structure:
{
"_id" : "1234",
"raw" : {
"meas" : {
"meas1" : {
"data" : "blabla"
}
}
}
}
In the second command, I need to add - in the same document - a "meas2" field at the level of "meas1". My desired output is:
{
"_id" : "1234",
"raw" : {
"meas" : {
"meas1" : {
"data" : "blabla"
},
"meas2" : {
"data" : "foo"
}
}
}
}
I played with statements like
updateOne({"_id":"1234"},{$set:{"raw":{"meas":{"meas2":{"data":"foo"}}}}}, {"upsert":true})
and also with $push, both variants with insert - here only the document and also insertOne, but nothing produces the desired output. Is there a MongoDb expert who could give a hint ? ... I'm sure this functionality exists... Thanks in advance!
When you update {$set: {"raw":{"meas":{"meas2":{"data":"foo"}}}} you're not adding "mesa2" to "meas" but rather you're overriting "raw" completely.
In order to change / add one field in a document refer to it with dot notations.
The command you want is updateOne({"_id": "1234"}, {$set: {"raw.meas.mesa2": { "data" : "foo" }}}, {"upsert":"true"})
You need to understand the below concept first
Set Fields in Embedded Documents, with details document check at official documentation of mongo
For your problem, just look at the below execution on the mongo shell:
> db.st4.insert({
... "_id" : "1234",
... "raw" : {
... "meas" : {
... "meas1" : {
... "data" : "blabla"
... }
... }
... }
... })
WriteResult({ "nInserted" : 1 })
> db.st4.find()
{ "_id" : "1234", "raw" : { "meas" : { "meas1" : { "data" : "blabla" } } } }
>
> // Below query will replace the raw document with {"meas":{"meas2":{"data":"foo"}}}, will not add
> //db.st4.updateOne({"_id":"1234"},{$set:{"raw":{"meas":{"meas2":{"data":"foo"}}}}}, {"upsert":true})
>// By using the dot operator, you actually write the values inside the documents i.e you are replacing or adding at raw.meas.mesa2 i.e inside the document of mesa2.
> db.st4.updateOne({"_id":"1234"},{$set: {"raw.meas.mesa2": { "data" : "foo" }}}, {"upsert":"true"})
{ "acknowledged" : true, "matchedCount" : 1, "modifiedCount" : 1 }
> db.st4.find().pretty()
{
"_id" : "1234",
"raw" : {
"meas" : {
"meas1" : {
"data" : "blabla"
},
"mesa2" : {
"data" : "foo"
}
}
}
}
>
I'm able to decrease a value in a nested array, but I want to check that the value is greater than zero (so never go to the negative numbers). For example, I have this:
{
"_id" : ObjectId("5a3b0bd69c0000c2a1d839af"),
"slots" : [
{
"id" : "V2qlAEk7Wp0tWwlyWSfX7KRZ",
"number" : 5.0
},
.......
.......
{
"id" : "VfB4f8G1KcgRA8qx0aby5nI0",
"number" : 0.0
}]
}
{
"_id" : ObjectId("5a3b0bd69c0000c2a1d839ag"),
"slots" : [
{
"id" : "V2qlAEEESbrEB4bwberbResbd",
"number" : 10.0
},
.......
.......
{
"id" : "DFwseEb5enRbfsbre54rtFfds",
"number" : 1.0
}]
}
If I want to decrease the number value of the first document [id=5a3b0bd69c0000c2a1d839af] of the slots.id = V2qlAEk7Wp0tWwlyWSfX7KRZ, I use:
db.getCollection('schedulers').update(
{
"_id" : ObjectId("5a3b0bd69c0000c2a1d839af"),
"slots.id" :"V2qlAEk7Wp0tWwlyWSfX7KRZ"
},
{
"$inc":{"slots.$.number":-1}
})
But I don't know how to check if the number is greater than 0 before decrease the value. I tried to use The filtered positional operator $[<identifier>], but I have an error:
db.getCollection('schedulers').update(
{
"_id" : ObjectId("5a3b0bd69c0000c2a1d839af"),
"slots.id" :"V2qlAEk7Wp0tWwlyWSfX7KRZ"
},
{
"$inc":{"slots.$[elm].number":-1}
},
{
arrayFilters: [ { "elm.number": {"$gt" : 0}} ]
})
The error is:
cannot use the part (slots of slots.$[elm].number) to traverse the element....
Also, apply the condition in the first part of the query doesn't fix the problem, cause mongo traverses all the array, so also if the number is zero it simply move to another element in the array until it finds number > 0 and when it finds it, it starts to decrease the number value of another element completely ignoring the two id conditions [because now the $ then point to another element of the array]:
db.getCollection('schedulers').update(
{
"_id" : ObjectId("5a3b0bd69c0000c2a1d839af"),
"slots.id" :"V2qlAEk7Wp0tWwlyWSfX7KRZ",
"slots.number":{"$gt" : 0}
},
{
"$inc":{"slots.$.number":-1}
})
I think the correct way to obtain what I want is to use the arrayFilters but I don't understand where is the problem.
Fixed using:
db.getCollection('schedulers').update(
{
"_id" : ObjectId("5a3b0bd69c0000c2a1d839af"),
"slots" :
{
"$elemMatch":
{
"id":"V2qlAEk7Wp0tWwlyWSfX7KRZ",
"number":
{
"$gt" : 0
}
}
}
},
{
"$inc":
{
"slots.$.number":-1
}
})
Thanks