For granular control on the test dataset, you can write a custom dataset loader function JavaScript or TypeScript. This requires a JS/TS configuration file.

import { Config } from "empiricalrun";

const config: Config = {
  dataset: function() {
    // build dataset
    return dataset;
  },
};

export default config;

This can be used to

​
Modify source dataset

The Spider using TS example uses this loader to add an dynamic input to the dataset sample before running the tests.

import { loadDataset } from "empiricalrun";

async function datasetLoader() {
  let dataset = await loadDataset({
    path: "https://docs.google.com/spreadsheets/d/1x_p0lX2pJEyGkFoe1A9nY3q87qOJUd547f2lz99ugiM/edit#gid=1000015421"
  })
  dataset.samples = dataset.samples.map(sample => {
    // get DB schema for the mentioned database name
    sample.inputs.schema = getSchema(sample.inputs.database_name)
    return sample;
  })
  return dataset
}

export default {
  dataset: datasetLoader
};

​
Generate synthetic data

With the custom loader, you can write a synthetic data generator for your use-case.

import { DatasetSample } from "empiricalrun";

async function datasetLoader(): Promise<DatasetSample[]> {
  // Custom generator for test data
}

export default {
  dataset: datasetLoader
};
Import from fileBasics