navidrome/utils/weighted_random_chooser.go

package utils

import (
	"errors"
	"math/rand"
	"time"
)

type weightedChooser struct {
	entries     []interface{}
	weights     []int
	totalWeight int
	rng         *rand.Rand
}

func NewWeightedRandomChooser() *weightedChooser {
	src := rand.NewSource(time.Now().UTC().UnixNano())

	return &weightedChooser{
		rng: rand.New(src), // nolint:gosec
	}
}

func (w *weightedChooser) Put(value interface{}, weight int) {
	w.entries = append(w.entries, value)
	w.weights = append(w.weights, weight)
	w.totalWeight += weight
}

// GetAndRemove choose a random entry based on their weights, and removes it from the list
func (w *weightedChooser) GetAndRemove() (interface{}, error) {
	if w.totalWeight == 0 {
		return nil, errors.New("cannot choose from zero weight")
	}
	i, err := w.weightedChoice()
	if err != nil {
		return nil, err
	}
	entry := w.entries[i]
	w.Remove(i)
	return entry, nil
}

// Based on https://eli.thegreenplace.net/2010/01/22/weighted-random-generation-in-python/
func (w *weightedChooser) weightedChoice() (int, error) {
	rnd := w.rng.Intn(w.totalWeight)
	for i, weight := range w.weights {
		rnd -= weight
		if rnd < 0 {
			return i, nil
		}
	}
	return 0, errors.New("internal error - code should not reach this point")
}

func (w *weightedChooser) Remove(i int) {
	w.totalWeight -= w.weights[i]

	w.weights[i] = w.weights[len(w.weights)-1]
	w.weights = w.weights[:len(w.weights)-1]

	w.entries[i] = w.entries[len(w.entries)-1]
	w.entries[len(w.entries)-1] = nil
	w.entries = w.entries[:len(w.entries)-1]
}

func (w *weightedChooser) Size() int {
	return len(w.entries)
}