Geosearch default sorting (asc) it's not handled during data loading

[DavideArena]

Describe the bug

Hello, i was trying the geosearch feature of Orama (and it's awesome!) and i was searching how the sorting systems works in this scenario. At first point, seems like there's no sorting, and sorting of data retrieved depends on documents sort inside database (first inserted will be put as first result, second inserted as second result and so on...).

I started to investigate inside code, and i found here (searchByRadius function) a default sorting. So i didn't understand why results don't maintain this sorting.

I figured out the problem, it is the scoring associated with the result. When you use the location search system, and you don't specify other terms, the scoring system assign default 0 value to records, and then the score sorting lose the default sorting made initially by the searchByRadius function.

I don't know if one possible way could be to attach the distance calculated by geosearch feature as score of docs, and use that as sorting parameter.

To Reproduce

1. Init db with following schema:

{ entityId: 'string', location: 'geopoint' }

2. Put three elements inside

insert(db, {
      entityId: 'entityFar',
      location: {
        lat: 43.773034,
        lon: 11.252929,
      },
    })

insert(db, {
      entityId: 'entityNear',
      location: {
        lat: 43.772974,
        lon: 11.25428,
      },
    })

insert(db, {
      entityId: 'entityVeryFar',
      location: {
        lat:  43.773404,
        lon: 11.252167,
      },
    })

3. Search with geoseach

  await search(db, {
    where: {
      location: {
        radius: {
          coordinates: {
            lat: 43.773081,
            lon: 11.255533,
          },
          unit: 'm',
          value: 3000,
          inside: true,
        },
      },
    },
  });

Expected behavior

Expected

{
   "elapsed":{
      "raw":4898041,
      "formatted":"4ms"
   },
   "hits":[
      {
         "id":"21819145-3",
         "score":0,
         "document":{
            "entityId":"EntityNear",
            "location":{
               "lat":43.772974,
               "lon":11.25428
            }
         }
      },
      {
         "id":"21819145-2",
         "score":0,
         "document":{
            "entityId":"entityFar",
            "location":{
               "lat":43.773034,
               "lon":11.252929
            }
         }
      },
      {
         "id":"21819145-4",
         "score":0,
         "document":{
            "entityId":"entityVeryFar",
            "location":{
               "lat":43.773404,
               "lon":11.252167
            }
         }
      }
   ],
   "count":3
}

Got

{
   "elapsed":{
      "raw":4898041,
      "formatted":"4ms"
   },
   "hits":[
      {
         "id":"21819145-2",
         "score":0,
         "document":{
            "entityId":"entityFar",
            "location":{
               "lat":43.773034,
               "lon":11.252929
            }
         }
      },
      {
         "id":"21819145-3",
         "score":0,
         "document":{
            "entityId":"EntityNear",
            "location":{
               "lat":43.772974,
               "lon":11.25428
            }
         }
      },
      {
         "id":"21819145-4",
         "score":0,
         "document":{
            "entityId":"entityVeryFar",
            "location":{
               "lat":43.773404,
               "lon":11.252167
            }
         }
      }
   ],
   "count":3
}

Environment Info

OS: MacOS Ventura 13.1
Node: 18.14.1
Orama version: 2.0.0-beta.5

Affected areas

Search

Additional context

No response

[debiff]

@micheleriva do you think this is a bug that is worth to be explored? If yes, I'd like to try!

[micheleriva]

@debiff go ahead :)

[debiff]

Quick question: is it correct that by applying filters (with where) we expect and ideally obtain the documents sorted? There is a reason why is not used the sortBy parameter? As a user, it seems to me more intuitive and less misleading

[DavideArena]

@debiff thank you for the tip. I've tried as described here (https://docs.orama.com/open-source/usage/search/sorting#custom-sort) with a custom sorting, using the score as sorting value. Seems working! I didn't expect it will work because on result points i got all docs with score 0 (but maybe during the search the score has a value).

Using orama 3.0.4

const db = create({
    schema: {
      entityId: "string",
      location: "geopoint",
    },
  });

  insert(db, {
    entityId: "entityVeryVeryFar",
    location: {
      lat: 44.773404,
      lon: 12.252167,
    },
  });
  insert(db, {
    entityId: "entityFar",
    location: {
      lat: 43.773034,
      lon: 11.252929,
    },
  });
  insert(db, {
    entityId: "entityNear",
    location: {
      lat: 43.772974,
      lon: 11.25428,
    },
  });
  insert(db, {
    entityId: "entityVeryFar",
    location: {
      lat: 43.773404,
      lon: 11.252167,
    },
  });

  const points = await search(db, {
    where: {
      location: {
        radius: {
          coordinates: {
            lat: 43.773081,
            lon: 11.255533,
          },
          unit: "m",
          value: 1000000,
          inside: true,
        },
      },
    },
    sortBy: (a, b) => {
      // Sorting by score
      return a[1] - b[1];
    },
  });

  console.log(JSON.stringify(points, null, 2));

Result

{
  "elapsed": {
    "raw": 468208,
    "formatted": "468μs"
  },
  "hits": [
    {
      "id": "16635519-3",
      "score": 0,
      "document": {
        "entityId": "entityNear",
        "location": {
          "lat": 43.772974,
          "lon": 11.25428
        }
      }
    },
    {
      "id": "16635519-2",
      "score": 0,
      "document": {
        "entityId": "entityFar",
        "location": {
          "lat": 43.773034,
          "lon": 11.252929
        }
      }
    },
    {
      "id": "16635519-4",
      "score": 0,
      "document": {
        "entityId": "entityVeryFar",
        "location": {
          "lat": 43.773404,
          "lon": 11.252167
        }
      }
    },
    {
      "id": "16635519-1",
      "score": 0,
      "document": {
        "entityId": "entityVeryVeryFar",
        "location": {
          "lat": 44.773404,
          "lon": 12.252167
        }
      }
    }
  ],
  "count": 4
}

Probabily what i was expecting was that this behaviour was the default out of the box, but probably the sort has an effort and it's correct to be applied only if needed. Dunno at this point if this issue should be opened.