#community-help

Defining Nested Fields in Typesense Schema

TLDR Ivy wanted to define a list of strings in a list of objects, and Jason suggested omitting the explicit field type for "products.ingredients". The schema generated the desired field type.

5mo

Join the chat

May 09, 2023 (5 months ago)

Photo of md5-2f39998c05c5409f8c3521187ec34d4e

Ivy

09:20 PM

Hi all, I have a question about nested objects!

I'd like to have a schema like this:

{
    storeName: "Best Slice Shop"
    products: [{
       id: 123,
       name: pizza,
       ingredients: ["Tomatoes", "Garlic", ... ]
    }
]

But it looks like I can only define types on one layer deep (ref https://github.com/typesense/typesense/issues/227#issuecomment-1364072388), e.g.

{
    "enable_nested_fields": true,
    "name": "store",
    "fields": [
        {
            "name": "products",
            "type": "object[]"
        },
        {
            "name": "products.id",
            "type": "int[]"
        },
        {
            "name": "products.name",
            "type": "string[]"
        },
        {
            "name": "products.ingredients",
            "type": "string[][]" <-- doesn't work
        }
    ]
}

Is there a way to define a field to be list of strings on an list of objects field?

Thank you!!

Photo of md5-8813087cccc512313602b6d9f9ece19f

Jason

09:22 PM

Interesting! Could you try this and see what happens:

{
            "name": "products.ingredients",
            "type": "string[]"
        }

09:23

Jason

09:23 PM

Hang on

09:23

Jason

09:23 PM

Let’s try something else

09:24

Jason

09:24 PM

Could you use this schema:

{
    "enable_nested_fields": true,
    "name": "store",
    "fields": [
        {
            "name": "products",
            "type": "object[]"
    },
    ]
}

then index one of your documents and then look at the updated collection schema which Typesense generates and check the field type for products.ingredients?

Ivy

09:41 PM

Ok it looks like it's string[]! Interesting. Thank you!

    {
      "facet": false,
      "index": true,
      "infix": false,
      "locale": "",
      "name": "products.ingredients",
      "optional": true,
      "sort": false,
      "type": "string[]"
    },```

[ { "id": "123", "ingredients": [ "Tomatoes", "Garlic" ], "name": "pizza" } ]```

Jason

09:42 PM

Btw, on a side note, the most generic field definition takes precedence in the collection schema. So if you have both products and products.ingredients explicitly listed, then products will take precedence and index all nested fields

Ivy

09:46 PM

Ah ok that makes sense and clarifies things, thank you!

Typesense

Lightning-fast, open source search engine for everyone | Knowledge Base powered by Struct.AI

Indexed 2779 threads (79% resolved)

Join Our Community Manage subscription

Similar Threads

Indexing Nested Fields in Arrays

Amrit was trying to index nested fields in arrays but didn't use a flat format. Kishore Nallan suggested using dot notation and updating the type to int32[].

4mo

Resolving Issues with Reference Field in v0.26.0.rc11

Shaun reported problems applying the reference field in `v0.26.0.rc11`. The issue made the tested field disappear from UI, and an attempt to fix it resulted in an error. Harpreet suggested re-indexing data and promised to consider improving the feature in future updates. Harpreet is currently working on a related feature - sorting by reference field.

1mo

Typesense Nested Fields and Sorting Issue

Amrit faced a problem with search response not returning nested objects. Kishore Nallan tried to spot the issue in the schema, but a proper resolution was not reached.

5mo

Troubleshooting Typesense Collection with Nested Objects

Oskar was encountering problems in creating collections in Typesense with nested objects. Jason advised to ensure that the nested fields are defined as an array. Oskar was able to resolve the issue by rectifying the schema.

Defining Nested JSON Schema for Querying

Greg struggled in defining a schema for querying a JSON object with nested fields. Jason offered solutions, with the final resolution involving the use of "object" as a field type for the contributors in the schema.

13mo