web

web / lib/services/schema-inference-service

lib/services/schema-inference-service

Provides on-demand schema inference for datasets.

This service analyzes existing events in a dataset and generates a schema by sampling events in batches. It’s designed for datasets that weren’t created through the import pipeline (e.g., seeding, direct API creation).

The service reuses the existing ProgressiveSchemaBuilder for schema detection and SchemaVersioningService for creating schema versions.

Classes

SchemaInferenceService

Service for inferring schemas from existing event data.

Constructors

Constructor

new SchemaInferenceService(): SchemaInferenceService

Returns

SchemaInferenceService

Methods

inferSchemaFromEvents()

static inferSchemaFromEvents(payload, datasetId, options?): Promise<SchemaInferenceResult>

Generate a schema from existing events in a dataset.

Parameters

payload

BasePayload

datasetId

number

options?

SchemaInferenceOptions = {}

Returns

Promise<SchemaInferenceResult>

getLatestSchema()

static getLatestSchema(payload, datasetId, req?): Promise<DatasetSchema | null>

Get the latest schema for a dataset, or null if none exists.

Parameters

payload

BasePayload

datasetId

number

req?

PayloadRequest

Returns

Promise<DatasetSchema | null>

Interfaces

SchemaInferenceOptions

Properties

sampleSize?

optional sampleSize: number

Maximum number of events to sample (default: 500)

batchSize?

optional batchSize: number

Number of events to process per batch (default: 100)

forceRegenerate?

optional forceRegenerate: boolean

Generate schema even if one already exists and is fresh (default: false)

req?

optional req: PayloadRequest

Payload request for context passing

SchemaInferenceResult

Properties

generated

generated: boolean

Whether a schema was generated

schema

schema: DatasetSchema | null

The generated or existing schema, if any

message

message: string

Message describing the result

eventsSampled?

optional eventsSampled: number

Number of events sampled