Hey all I m not seeing some oddities when performing a geose typesense #community-help

Hey all! I'm not seeing some oddities when perform...

Adam Al-dbhany

02/20/2025, 10:49 PM

Hey all! I'm not seeing some oddities when performing a geosearch via a Join between two collections. I have a small sample set to test accuracy yet seeing two different results, regardless if I'm using single or multi-search. When using the Javascript SDK, no results are found. When making the request directly, I'm left with "Failed to join on `vendor_addresses_20250220_223920654Z`: No reference field found." Notes • Each collections were created with different names (ex:

vendors_YYYYMMDD

) but have aliases configured with the below collection names (

vendors

and

vendor_addresses

) • I have painstakingly verified that each geopoint is being stored and queried in Lat/Long order. Collections Snipped of each collection

Copy code

// Entity collection
{
    name: 'vendors',
    fields: [
        {
            name: 'id',
            type: 'string',
            facet: false
        }, {
            name: 'overview',
            type: 'string',
            facet: false,
            index: true,
        }, {
            name: 'updatedAt',
            type: 'int64',
            facet: false,
            index: true
        }],
    default_sorting_field: 'updatedAt',
}
// Vendor Address Collection
{
    name: 'vendor_addresses',
    fields: [
        {
            name: 'id',
            type: 'string',
            facet: false,
        },
        {
            name: 'vendorId',
            type: 'string',
            reference: "vendors.id",
        },
        {
            name: 'geolocation',
            type: 'geopoint',
            index: true,
        }
    ],
    default_sorting_field: 'vendorId',
}

Would really appreciate any and all help here!

Adam Al-dbhany

02/20/2025, 10:51 PM

Example multi-search payload that returns no results:

Copy code

{
  "searches": [
    {
      "collection": "vendors",
      "query_by": "services,overview,name",
      "limit": 10,
      "prioritize_exact_match": false,
      "filter_by": "$vendor_addresses(geolocation:(39.11553139999999, -94.62678729999999, 200.00 mi)) && enabled:true",
      "facet_strategy": "exhaustive",
      "q": "roofing service",
      "query_by_weights": "3,2,1",
      "vector_query": "someEmbeddings:([], queries:[roofing, roofing services, roof repair, roof replacement], query_weights:[0.4, 0.3, 0.2, 0.1], alpha:0.75)",
      "include_fields": "id,domain",
      "exclude_fields": "pageSummaryEmbeddings",
      "prefix": false,
      "sort_by": "_vector_distance:asc,_text_match:desc"
    }
  ]
}

Adam Al-dbhany

02/20/2025, 10:52 PM

I've also tried filters like below with similar/no results: •

$vendor_addresses(vendorId:* && geolocation:(39.11553139999999, -94.62678729999999, 200.00 mi))

•

$vendor_addresses(id:* && geolocation:(39.11553139999999, -94.62678729999999, 200.00 mi))

Jason Bosco

02/20/2025, 11:01 PM

CC: @Harpreet Sangar

thankyou 1

Adam Al-dbhany

02/20/2025, 11:04 PM

Additionally, I've tried this with both v27.1 and v28

Harpreet Sangar

02/21/2025, 4:18 AM

Hi @Adam Al-dbhany Can you check if

geolocation:(39.11553139999999, -94.62678729999999, 200.00 mi)

filter matches documents in

vendor_addresses

collections? Also, let me know if this search includes the related vendor.

Copy code

"collection": "vendor_addresses",
"q": "*",
"filter_by": "geolocation:(39.11553139999999, -94.62678729999999, 200.00 mi)",
"include_fields": "$vendors(*)"

I'm trying to figure out if the references are present or not.

Adam Al-dbhany

02/21/2025, 4:29 PM

Querying the

vendor_addresses

with the geolocation filter does give the results I would expect. I can test out the

include_fields

once I'm freed up

✅ 1

Open in Slack

Previous Next