import getCovarianceMatrix from "./getCovarianceMatrix.js"
import getMahalanobisDistance from "./getMahalanobisDistance.js"

/**
 * Computes the Mahalanobis distance from an origin object for each object in an array. The keys in the origin object are the dimensions considered (tested up to four dimensions but should work for more in theory). The function adds the key `mahaDist` in each object in the original array.
 *
 * If you pass the option `{ similarity: true }`, it will add another key `similarity` which goes from 1 to 0. The closer similarity is to 1, the closer the item is to the origin. You can also pass a precomputed matrix in options if needed: `{ matrix : precomputedMatrix }.`
 *
 * @example
 * Basic usage
 * ```js
 * // Wines
 * const data = [
 *     {'fixed acidity': 6.5, 'alcohol' : 11, ... },
 *     {'fixed acidity': 7.1, 'alcohol' : 12.2, ... },
 *     {'fixed acidity': 6.3, 'alcohol' : 10.5, ... }
 * ]
 *
 * // We want the distance from this wine. All the keys in the origin will be used as variables. They need to be present in the data, of course.
 * const origin = {'fixed acidity': 7.2, 'alcohol' : 11.3 }
 *
 * addMahalanobisDistance(origin, data)
 *
 * // We now have mahaDist in the data.
 * data.sort((a, b) => a.mahaDist - b.mahaDist)
 *
 * // And data looks like this.
 * //[
 * //  { 'fixed acidity': 7.2, alcohol: 11.3, mahaDist: 0 },
 * //  { 'fixed acidity': 7.5, alcohol: 10.5, mahaDist: 0.939 },
 * //  { 'fixed acidity': 7.3, alcohol: 11.4, mahaDist: 1.263 },
 * //  { 'fixed acidity': 6.5, alcohol: 13, mahaDist: 2.079 },
 * //  { 'fixed acidity': 7.1, alcohol: 12.2, mahaDist: 2.411 }
 * //]
 *
 * // You can also pass the option similarity if you want.
 * addMahalanobisDistance(origin, data, { similarity: true })
 *
 * // The data will have a similarity key going from 0 to 1.
 * // [
 * //  { "fixed acidity": 7.2, alcohol: 11.3, mahaDist: 0, similarity: 1 },
 * //  { "fixed acidity": 7.5, alcohol: 10.5, mahaDist: 0.939, similarity: 0.611 },
 * //  { "fixed acidity": 7.3, alcohol: 11.4, mahaDist: 1.263, similarity: 0.476 },
 * //  { "fixed acidity": 6.5, alcohol: 13, mahaDist: 2.079, similarity: 0.138 },
 * //  { "fixed acidity": 7.1, alcohol: 12.2, mahaDist: 2.412, similarity: 0 }
 * // ]
 * ```
 *
 * @param origin - The origin object with keys as dimensions and values as the reference point.
 * @param data - The array of objects to compute the Mahalanobis distance for.
 * @param options - Optional parameters.
 * @param options.similarity - If true, adds a similarity key to each object.
 * @param options.matrix - A precomputed covariance matrix.
 * @category Statistics
 */

export default function addMahalanobisDistance(
    origin: Record<string, number>,
    data: Record<string, unknown>[],
    options: { similarity?: boolean; matrix?: number[][] } = {}
): Record<string, unknown>[] {
    const variables = Object.keys(origin)
    const originArray = variables.map((v) => origin[v])
    const dataArray = data.map((d) => variables.map((v) => d[v]))

    const invertedCovarianceMatrix = Array.isArray(options.matrix)
        ? options.matrix
        : getCovarianceMatrix(
              dataArray as number[][], // getCovarianceMatrix will check the types
              {
                  invert: true,
              }
          )

    data.forEach(
        (d) =>
            (d.mahaDist = getMahalanobisDistance(
                originArray,
                variables.map((v) => d[v] as number), // types checked in getCovarianceMatrix
                invertedCovarianceMatrix
            ))
    )

    if (options.similarity) {
        const maxDist = Math.max(...data.map((d) => d.mahaDist as number))
        data.forEach(
            (d) => (d.similarity = 1 - (d.mahaDist as number) / maxDist)
        )
    }

    return data
}