Get a value in a reference of a lookup with MongoD

问题:

I have the structures below. I use Golang 1.9.2.

// EventBoost describes the model of a EventBoost
type EventBoost struct {
  ID          string    `bson:"_id" json:"_id" valid:"alphanum,printableascii"`
  CampaignID  string    `bson:"_campaign_id" json:"_campaign_id" valid:"alphanum,printableascii"`
  Name        string    `bson:"name" json:"name"`
  Description string    `bson:"description" json:"description"`
  Level       string    `bson:"level" json:"level"`
  EventID     string    `bson:"_event_id" json:"_event_id" valid:"alphanum,printableascii"`
  StartDate   time.Time `bson:"start_date" json:"start_date"`
  EndDate     time.Time `bson:"end_date" json:"end_date"`
  IsPublished bool      `bson:"is_published" json:"is_published"`
  CreatedBy   string    `bson:"created_by" json:"created_by"`
  CreatedAt   time.Time `bson:"created_at" json:"created_at"`
  ModifiedAt  time.Time `bson:"modified_at" json:"modified_at"`
}

// LocationBoost describes the model of a LocationBoost
type LocationBoost struct {
  ID          string    `bson:"_id" json:"_id" valid:"alphanum,printableascii"`
  CampaignID  string    `bson:"_campaign_id" json:"_campaign_id" valid:"alphanum,printableascii"`
  Name        string    `bson:"name" json:"name"`
  Description string    `bson:"description" json:"description"`
  Level       string    `bson:"level" json:"level"`
  LocationID  string    `bson:"_location_id" json:"_location_id" valid:"alphanum,printableascii"`
  StartDate   time.Time `bson:"start_date" json:"start_date"`
  EndDate     time.Time `bson:"end_date" json:"end_date"`
  IsPublished bool      `bson:"is_published" json:"is_published"`
  CreatedBy   string    `bson:"created_by" json:"created_by"`
  CreatedAt   time.Time `bson:"created_at" json:"created_at"`
  ModifiedAt  time.Time `bson:"modified_at" json:"modified_at"`
}

// Campaign describes the model of a Campaign
type Campaign struct {
    ID               string           `bson:"_id" json:"_id" valid:"alphanum,printableascii"`
    Name             string           `bson:"name" json:"name"`
    Description      string           `bson:"description" json:"description"`
    EventBoostIDs    []string         `bson:"event_boost_ids" json:"event_boost_ids"`
    LocationBoostIDs []string         `bson:"location_boost_ids" json:"location_boost_ids"`
    StartDate        time.Time        `bson:"start_date" json:"start_date"`
    EndDate          time.Time        `bson:"end_date" json:"end_date"`
    IsPublished      bool             `bson:"is_published" json:"is_published"`
    CreatedBy        string           `bson:"created_by" json:"created_by"`
    CreatedAt        time.Time        `bson:"created_at" json:"created_at"`
    ModifiedAt       time.Time        `bson:"modified_at" json:"modified_at"`
}

A Campaign (understand a marketing campaign) is made of Events or Locations that can be boosted with a level (basic or premium). A campaign has a start and a end date, so do have the boosts.

The function GetEventLevel has to return me the level of a given event.

// GetEventLevel of an event
func (dao *campaignDAO) GetEventLevel(eventID string) (string, error) {
}

If the event is boosted in an active campaign (isPublished is true), and the boost is active (isPublished is true) and the now date is between the start and end date of the boost, then my Event is boosted, so the function returns the level (basic or premium). Else, it returns "standard".

My question is : can I do this fully with Mongo ? Or do I need to perform some logic in the DAO with Golang ?

If I can do this with Mongo, what I hope, I have no idea how to do this. From what I understand, I would first need to lookup the events and the locations of the campaign, and then search in it with dates, but..

回答1:

Doing most (and the hardest part) of what you want can easily be done in MongoDB. The final step when returning "basic", "premium" or "standard" most likely can also be done, but I think it's not worth the hassle as that is trivial in Go.

In MongoDB use the Aggregation framework for this. This is available in the mgo package via the Collection.Pipe() method. You have to pass a slice to it, each element corresponds to an aggregation stage. Read this answer for more details: How to Get an Aggregate from a MongoDB Collection

Back to your example. Your GetEventLevel() method could be implemented like this:

func (dao *campaignDAO) GetEventLevel(eventID string) (string, error) {
    c := sess.DB("").C("eventboosts") // sess represents a MongoDB Session
    now := time.Now()
    pipe := c.Pipe([]bson.M{
        {
            "$match": bson.M{
                "_event_id":    eventID,            // Boost for the specific event
                "is_published": true,               // Boost is active
                "start_date":   bson.M{"$lt": now}, // now is between start and end
                "end_date":     bson.M{"$gt": now}, // now is between start and end
            },
        },
        {
            "$lookup": bson.M{
                "from":         "campaigns",
                "localField":   "_campaign_id",
                "foreignField": "_id",
                "as":           "campaign",
            },
        },
        {"$unwind": "$campaign"},
        {
            "$match": bson.M{
                "campaign.is_published": true,      // Attached campaign is active
            },
        },
    })

    var result []*EventBoost
    if err := pipe.All(&result); err != nil {
        return "", err
    }
    if len(result) == 0 {
        return "standard", nil
    }
    return result[0].Level, nil
}

If you only need at most one EventBoost (or there may not be more at the same time), use $limit stage to limit results to a single one, and use $project to only fetch the level field and nothing more.

Use this pipeline for the above mentioned simplification / optimization:

pipe := c.Pipe([]bson.M{
    {
        "$match": bson.M{
            "_event_id":    eventID,            // Boost for the specific event
            "is_published": true,               // Boost is active
            "start_date":   bson.M{"$lt": now}, // now is between start and end
            "end_date":     bson.M{"$gt": now}, // now is between start and end
        },
    },
    {
        "$lookup": bson.M{
            "from":         "campaigns",
            "localField":   "_campaign_id",
            "foreignField": "_id",
            "as":           "campaign",
        },
    },
    {"$unwind": "$campaign"},
    {
        "$match": bson.M{
            "campaign.is_published": true,      // Attached campaign is active
        },
    },
    {"$limit": 1},             // Fetch at most 1 result
    {
        "$project": bson.M{
            "_id":   0,        // We don't even need the EventBoost's ID
            "level": "$level", // We do need the level and nothing more
        },
    },
})

回答2:

Since you're only storing IDs to reference documents in other collections, rather than de-normalizing the data entirely, no, you cannot do this purely in MongoDB. MongoDB is not a relational database. What you're describing is exactly what MongoDB was designed not to do.

You'll need to perform the logic in Go; whether you do so in the DAO is up to you, but personally, I would lean toward a simple method on the type that does the logic on the fly based on the field values, e.g. Campaign.GetEventLevel or something along those lines. Having a method that fine-grained in your DAO implies to me some unusual design decisions that don't suit the MongoDB model very well. For the most part, with MongoDB, you want to retrieve you documents (objects) and work with them in your app. Trying to perform query logic in MongoDB as you would with a typical RDBMS is going to lead to frustration and poor performance.