如何从MongoDB集合中获取聚合

Question

The following collection contains the ticks for the diverse exchange symbols (in this case BTCUSD and BTCEUR):

{ "_id" : ObjectId("5a08d2b956df9b2302759d1a"), "symbol" : "BTCUSD", "time" : ISODate("2013-04-13T00:00:00Z"), "open" : 112, "close" : 91.1, "high" : 130, "low" : 81.12, "volume" : 23866.6770456 }
{ "_id" : ObjectId("5a08d2b956df9b2302759d1c"), "symbol" : "BTCUSD", "time" : ISODate("2013-04-14T00:00:00Z"), "open" : 91.1, "close" : 90.171, "high" : 109, "low" : 20, "volume" : 16437.2196645 }
{ "_id" : ObjectId("5a08d2b956df9b2302759d1e"), "symbol" : "BTCEUR", "time" : ISODate("2013-04-15T00:00:00Z"), "open" : 89.86, "close" : 83.302, "high" : 104, "low" : 71.497, "volume" : 16393.12856398 }
{ "_id" : ObjectId("5a08d2b956df9b2302759d20"), "symbol" : "BTCEUR", "time" : ISODate("2013-04-16T00:00:00Z"), "open" : 84.27, "close" : 67.588, "high" : 84.48, "low" : 0.01, "volume" : 26092.5432296 }

How do I get the list of symbols actually present in the collection in Go? For instance, the result I'm looking for is:

{ "BTCUSD", "BTCEUR" }

This would let me retrieve the latest tick values from the exchange service without having to store somewhere the symbols I'm interested in.

From the mongo shell I've tried this...

db.candles.aggregate({$group: {_id:"$symbol"}})

... and here is the result:

{ "_id" : "BTCUSD" }
{ "_id" : "BTCEUR" }

What is the equivalent in Go using mgo? Is it possible to just get an array of symbols instead of a collection of "_id" : "value" pairs?

Answer 1

Using Distinct()

What you want is easiest done using collection.distinct(). In MongoDB console it would look like this:

db.candles.distinct("symbol")

And in Go, using mgo, Query.Distinct():

var sess *mgo.Session = ... // Acquire session
c := sess.DB("dbname").C("candles")

var symbols []string
err := c.Find(nil).Distinct("symbol", &symbols)
if err != nil {
    log.Printf("Error: %v", err)
    return
}
fmt.Println(symbols)

Using the Aggregation Framework

You can also do it using MongoDB aggregation, which is available via the Collection.Pipe() method. You have to pass a slice to it, each element corresponds to an aggregation stage.

The result of an aggregation is always a list of documents, so if you want the result as a slice of strings, you have to do the "conversion" manually.

var sess *mgo.Session = ... // Acquire session
c := sess.DB("dbname").C("candles")

pipe := c.Pipe([]bson.M{
    {
        "$group": bson.M{
            "_id": "$symbol",
        },
    },
})
var results = []bson.M{}
if err := pipe.All(&results); err != nil {
    log.Printf("Error: %v", err)
    return
}

// results holds {"_id": "symbol"} documents
// To get a slice of symbols:
symbols := make([]string, len(results))
for i, doc := range results {
    symbols[i] = doc["_id"].(string)
}
fmt.Println(symbols)

For more techniques how to get a slice of values result instead of documents, see Is there a way to get slice as result of Find()?