I started learning performance improvements on mongoDb. And i have a question based aggregation functions. I have created a basic test collection with 3 fields product, colour and year:
{
"product" : "car",
"colour" : "red",
"year" : "2019"
}
{
"product" : "car",
"colour" : "black",
"year" : "2018"
}
{
"product" : "bike",
"colour" : "blue",
"year" : "2014"
}
{
"product" : "train",
"colour" : "black",
"year" : "2019"
}
{
"product" : "ship",
"colour" : "red",
"year" : "2018"
}
{
"product" : "car",
"colour" : "red",
"year" : "2016"
}
{
"product" : "car",
"colour" : "blue",
"year" : "2015"
}
{
"product" : "bike",
"colour" : "white",
"year" : "2014"
}
{
"product" : "train",
"colour" : "red",
"year" : "2016"
}
{
"product" : "ship",
"colour" : "red",
"year" : "2015"
}
Due to performance issue, i wish to use facet for single aggregation stage:
db.getCollection('test').aggregate([
{
"$match":{
"colour":"red"
}
},
{
"$facet": {
"cntOfReds":[
{"$group":{"_id": "$product", "count": {"$sum": 1}}}
],
"cntOfRedsIn2015":[
{"$match" :{"year": "2015"}},
{"$group": {"_id": "$year", "count": {"$sum": 1}}}
]
} }
])
How do we implement the same query on golang with mgo.v2 and set the results to a struct?
Using mgo, you have to pass an array of documents to Pipe
:
p:=collection.Pipe([]bson.M{
{"$match": bson.M{"colour":"red"}},
{"$facet": bson.M{ "cntOfReds":[]bson.M{{"$group":{"_id": "$product", "count": bson.M{"$sum":1}}}}},
"cntOfRedsIn2015": ...
Then, run the pipeline and get results:
p.All(&results)