Count number of rows and get only the last row in MongoDB - mongodb

Question

I have a collection of posts as follows:
{
"author": "Rothfuss",
"text": "Name of the Wind",
"likes": 1007,
"date": ISODate("2013-03-20T11:30:05Z")
},
{
"author": "Rothfuss",
"text": "Doors of Stone",
"likes": 1,
"date": ISODate("2051-03-20T11:30:05Z")
}
I want to get the count of each author's posts and his/her last post.
There is a SQL answer for the same question here. I try to find its MongoDB alternative.
I ended up this query so far:
db.collection.aggregate([
{
"$group": {
"_id": "$author",
"count": {
"$sum": 1
},
"lastPost": {
"$max": {
"_id": "$date",
"post": "$text"
}
}
}
}
])
which seems to work, but its different runs generate different results. It can be tested here in Mongo playground.
I don't understand how to use $max to select another property from the document containing the maximum. I am new to MongoDB, so describing the basics is also warmly appreciated.
extra question
Is it possible to limit $sum to only add posts with likes more than 100?

Answer 1

its different runs generate different results.
I don't understand how to use $max to select another property from the document containing the maximum.
The $max does not work in multiple fields, and also it is not effective in that field that having text/string value.
It will select any of the properties from a group of posts, it will different every time.
So the accurate result you can add new stage $sort before $group stage, to sort by date in descending order, and in the group stage you can select a value by $first operator,
{ $sort: { date: -1 } },
{
$group: {
_id: "$author",
count: { $sum: 1 },
date: { $first: "$date" },
post: { $first: "$text" }
}
}
Is it possible to limit $sum to only add posts with likes more than 100?
There is two meaning of your requirement, I am not sure which is you are asking but let me give both the solutions,
If you only don't want to count posts in count but you want to get it as the last post's date and text if it is.
$cond check condition if likes is greater than 100 then count 1 otherwise count 0
db.collection.aggregate([
{ $sort: { date: -1 } },
{
$group: {
_id: "$author",
count: {
$sum: {
$cond: [{ $gt: ["$likes", 100] }, 1, 0]
}
},
date: { $first: "$date" },
post: { $first: "$text" }
}
}
])
Playground
If you don't want to count and also don't want the last post if it is.
You can add a $match stage at the first stage to check greater than condition, and your final query would be,
db.collection.aggregate([
{ $match: { likes: { $gt: 100 } } },
{ $sort: { date: -1 } },
{
$group: {
_id: "$author",
count: { $sum: 1 },
date: { $first: "$date" },
post: { $first: "$text" }
}
}
])
Playground

Answer 2

Your query looks ok to me, adding a $match stage can filter out the posts if not likes > 100. (you can also do it in $sum, with $cond but there is no need here)
Query
$max accumulator can be used for documents also
Here you can see how MongoDB compares documents
mongoplayground has a problem and loses the order of fields in the documents(behaves likes they are are hashmaps when they are not) (test it in your driver also)
Test code here
db.collection.aggregate([
{
"$match": {
"likes": {
"$gt": 100
}
}
},
{
"$group": {
"_id": "$author",
"count": {
"$sum": 1
},
"lastPost": {
"$max": {
_id: "$date",
post: "$text"
}
}
}
}
])