🔄 MongoDB Aggregation Stages

Stages are the building blocks of aggregation pipelines.

📘 We saw in the previous page two methods to compose stages to effectively build a pipeline:

Using the pipeline stages methods

Using the stages classes directly

Repeating what was described previously:

💡 Each stage of the aggregation framework also has its own class in the package. And each Stage class has a mirror method in the Pipeline.

There is actually an asterisk to this. Monggregate does not yet provide an interface to all of the stages provided by MongoDB. It is a work in progress and the list of available stages will grow over time. If you want to contribute, please refer to the contributing guide.

You can see the full list of stages provided by MongoDB here.

📋 List of Available Stages In Monggregate

The following table lists the stages that are currently available in Monggregate:

$addFields
$bucket
$bucketAuto
$count
$group
$limit
$lookup
$match
$merge
$out
$project
$replaceRoot
$replaceWith
$sample
$search
$searchMeta
$set
$skip
$sort
$sortByCount
$unionWith
$unset
$unwind

🚀 Usage

🎯 monggregate aims at providing a simple and intuitive interface to the MongoDB aggregation framework.

Even though, it tries as much as possible to stick by the MongoDB aggregation framework syntax, it also tries to provide alternative ways to reproduce the syntax of other tools that new Mongo users might be more familiar with such as SQL and Pandas.

For example, in the $group stage, the MongoDB aggregation framework expects the grouping field(s) to be provided in the _id key. However, monggregate allows you to provide the grouping field(s) in the by key instead.

pipeline = Pipeline()

pipeline.group(
    by="year",
    query={
        "movie_count": {"$sum": 1},
        "movie_titles": {"$push": "$title"}
    }
)

Similarly, monggregate pipeline lookup method and Lookup class provide aliases for the orignal MongoDB arguments:

MongoDB Original Name	Monggregate Original Name	Monggregate Convenient Alias
from	from	right
localField	local_field	left_on
foreignField	foreign_field	right_on
as	as	name

ℹ️ Note: The original names of the arguments were converted to snake_case to follow the Python convention. You cannot use the camelCase version of the arguments names here.

You can therefore use any combination of arguments names from the two rightmost columns above to build your stage.

pipeline = Pipeline()
pipeline.lookup(
    right = "comments", # collection to join
    left_on = "_id",  # primary key
    right_on = "movie_id", # foreign key
    # name of the field that will contain the matching documents
    name = "related_comments" 
)

When using the stages methods, you can sometimes omit to name the argument(s). We can for example complete the previous example as follows:

pipeline = Pipeline()
pipeline.lookup(
    right = "comments", 
    left_on = "_id", 
    right_on = "movie_id",
    name = "related_comments" 
).sort(
    "movie_count"
).limit(
    10
)

🔍 The arguments names (by and value respectively) for the sort and limit stages are omitted.

💡 Note: Just as the Pipeline class provides methods for all stages, the S (Dollar) object provides methods for all MongoDB operators. They serve similar roles in different contexts - Pipeline for constructing aggregation sequences, and S for building expressions with operators. For more details on the S object, see the Operators documentation.

🛠️ Operators

You might have noticed in the grouping example how we tell Monggregate to perform operations on the groups. In the example, we used the $sum and $push operators.

🔜 For more information about operators, check the next page.