GSoC 2017: Adding Statistics and Graphs to ListenBrainz

[iliekcomputers]

Thanks for the comments, @Zas.

@Zas

Nice proposal

Thank you!

@Zas

Stats API endpoints:actually it would be a good idea to provide the stats through a json format usable by telegraf/plugins/inputs/httpjson at master · influxdata/telegraf · GitHub
Storing values on regular basis in an influxdb database makes it very easy to represent data (we use grafana + influxdb + telegraf for nodes/services metrics already).

I have been thinking of what format json the API endpoints would return and for the stats, that I have proposed, it would be something like this. Suppose we make a request to get a user’s top artists, the json would be:

{
     "user_id": "iliekcomputers",
     "artist_count": 10,
     "payload" : [
          {
              "artist_name": "Kanye West",
              "artist_msid": "msid here",
              "timeframe": "last_week",
              "listen_count": 80
          },
          ...
      ]
 }

Some discussion around the format exactly would be needed before we start with the API endpoints, but I completely agree with making the json format usable by telegraf.
I am not completely sure how to design it so that it can be used by telegraf. Some help here would be appreciated. I will add this format to the draft while more discussion on it takes place.

@Zas

Stats keyed by time:Of course, you could use iinfluxdb as a local storage and let people access to it through useful influx API.

I’m not sure how saving the data in influx would help with making stats show listen time instead of listen count. Could you please elaborate on this? It would be really helpful. The problem I came across when thinking of how to implement this is how to get listen times from listen counts. This data is not sent by the user and making requests to the MusicBrainz database would be expensive. Also, I think I may not have explained this idea well enough. I will edit the proposal to fix this.