workloads/segmentation/async-task.js

/*
 * Copyright (C) 2018 Apple Inc. All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY
 * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL APPLE INC. OR
 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
 * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
 */

var Statistics = new (function () {

    this.min = function (values) {
        return Math.min.apply(Math, values);
    }

    this.max = function (values) {
        return Math.max.apply(Math, values);
    }

    this.sum = function (values) {
        return values.length ? values.reduce(function (a, b) { return a + b; }) : 0;
    }

    this.mean = function (values) {
        return this.sum(values) / values.length;
    }

    this.median = function (values) {
        return values.sort(function (a, b) { return a - b; })[Math.floor(values.length / 2)];
    }

    this.squareSum = function (values) {
        return values.length ? values.reduce(function (sum, value) { return sum + value * value;}, 0) : 0;
    }

    // With sum and sum of squares, we can compute the sample standard deviation in O(1).
    // See https://rniwa.com/2012-11-10/sample-standard-deviation-in-terms-of-sum-and-square-sum-of-samples/
    this.sampleStandardDeviation = function (numberOfSamples, sum, squareSum) {
        if (numberOfSamples < 2)
            return 0;
        return Math.sqrt(squareSum / (numberOfSamples - 1) - sum * sum / (numberOfSamples - 1) / numberOfSamples);
    }

    this.supportedConfidenceIntervalProbabilities = function () {
        var supportedProbabilities = [];
        for (var probability in tDistributionByOneSidedProbability)
            supportedProbabilities.push(oneSidedToTwoSidedProbability(probability).toFixed(2));
        return supportedProbabilities
    }

    this.supportedOneSideTTestProbabilities = function () {
        return Object.keys(tDistributionByOneSidedProbability);
    }

    // Computes the delta d s.t. (mean - d, mean + d) is the confidence interval with the specified probability in O(1).
    this.confidenceIntervalDelta = function (probability, numberOfSamples, sum, squareSum) {
        var oneSidedProbability = twoSidedToOneSidedProbability(probability);
        if (!(oneSidedProbability in tDistributionByOneSidedProbability)) {
            throw 'We only support ' + this.supportedConfidenceIntervalProbabilities().map(function (probability)
            { return probability * 100 + '%'; } ).join(', ') + ' confidence intervals.';
        }
        if (numberOfSamples - 2 < 0)
            return NaN;
        var deltas = tDistributionByOneSidedProbability[oneSidedProbability];
        var degreesOfFreedom = numberOfSamples - 1;
        if (degreesOfFreedom > deltas.length)
            throw 'We only support up to ' + deltas.length + ' degrees of freedom';

        // d = c * S/sqrt(numberOfSamples) where c ~ t-distribution(degreesOfFreedom) and S is the sample standard deviation.
        return deltas[degreesOfFreedom - 1] * this.sampleStandardDeviation(numberOfSamples, sum, squareSum) / Math.sqrt(numberOfSamples);
    }

    this.confidenceInterval = function (values, probability) {
        var sum = this.sum(values);
        var mean = sum / values.length;
        var delta = this.confidenceIntervalDelta(probability || 0.95, values.length, sum, this.squareSum(values));
        return [mean - delta, mean + delta];
    }

    // Welch's t-test (http://en.wikipedia.org/wiki/Welch%27s_t_test)
    this.testWelchsT = function (values1, values2, probability) {
        return this.computeWelchsT(values1, 0, values1.length, values2, 0, values2.length, probability).significantlyDifferent;
    }

    this.probabilityRangeForWelchsT = function (values1, values2) {
        var result = this.computeWelchsT(values1, 0, values1.length, values2, 0, values2.length);
        if (isNaN(result.t) || isNaN(result.degreesOfFreedom))
            return {t: NaN, degreesOfFreedom:NaN, range: [null, null]};

        var lowerBound = null;
        var upperBound = null;
        for (var probability in tDistributionByOneSidedProbability) {
            var twoSidedProbability = oneSidedToTwoSidedProbability(probability);
            if (result.t > tDistributionByOneSidedProbability[probability][Math.round(result.degreesOfFreedom - 1)])
                lowerBound = twoSidedProbability;
            else if (lowerBound) {
                upperBound = twoSidedProbability;
                break;
            }
        }
        return {t: result.t, degreesOfFreedom: result.degreesOfFreedom, range: [lowerBound, upperBound]};
    }

    this.computeWelchsT = function (values1, startIndex1, length1, values2, startIndex2, length2, probability) {
        var stat1 = sampleMeanAndVarianceForValues(values1, startIndex1, length1);
        var stat2 = sampleMeanAndVarianceForValues(values2, startIndex2, length2);
        var sumOfSampleVarianceOverSampleSize = stat1.variance / stat1.size + stat2.variance / stat2.size;
        var t = Math.abs((stat1.mean - stat2.mean) / Math.sqrt(sumOfSampleVarianceOverSampleSize));

        // http://en.wikipedia.org/wiki/Welch-Satterthwaite_equation
        var degreesOfFreedom = sumOfSampleVarianceOverSampleSize * sumOfSampleVarianceOverSampleSize
            / (stat1.variance * stat1.variance / stat1.size / stat1.size / stat1.degreesOfFreedom
                + stat2.variance * stat2.variance / stat2.size / stat2.size / stat2.degreesOfFreedom);
        var minT = tDistributionByOneSidedProbability[twoSidedToOneSidedProbability(probability || 0.8)][Math.round(degreesOfFreedom - 1)];
        return {
            t: t,
            degreesOfFreedom: degreesOfFreedom,
            significantlyDifferent: t > minT,
        };
    }

    this.findRangesForChangeDetectionsWithWelchsTTest = function (values, segmentations, oneSidedPossibility=0.99) {
        if (!values.length)
            return [];

        const selectedRanges = [];
        const twoSidedFromOneSidedPossibility = 2 * oneSidedPossibility - 1;

        for (let i = 1; i + 2 < segmentations.length; i += 2) {
            let found = false;
            const previousMean = segmentations[i].value;
            const currentMean = segmentations[i + 1].value;
            console.assert(currentMean != previousMean);
            const currentChangePoint = segmentations[i].seriesIndex;
            const start = segmentations[i - 1].seriesIndex;
            const end = segmentations[i + 2].seriesIndex;

            for (let leftEdge = currentChangePoint - 2, rightEdge = currentChangePoint + 2; leftEdge >= start && rightEdge <= end; leftEdge--, rightEdge++) {
                const result = this.computeWelchsT(values, leftEdge, currentChangePoint - leftEdge, values, currentChangePoint, rightEdge - currentChangePoint, twoSidedFromOneSidedPossibility);
                if (result.significantlyDifferent) {
                    selectedRanges.push({
                        startIndex: leftEdge,
                        endIndex: rightEdge - 1,
                        segmentationStartValue: previousMean,
                        segmentationEndValue: currentMean
                    });
                    found = true;
                    break;
                }
            }
            if (!found && Statistics.debuggingTestingRangeNomination)
                console.log('Failed to find a testing range at', currentChangePoint, 'changing from', previousMean, 'to', currentMean);
        }

        return selectedRanges;
    };

    function sampleMeanAndVarianceForValues(values, startIndex, length) {
        var sum = 0;
        for (var i = 0; i < length; i++)
            sum += values[startIndex + i];
        var squareSum = 0;
        for (var i = 0; i < length; i++)
            squareSum += values[startIndex + i] * values[startIndex + i];
        var sampleMean = sum / length;
        // FIXME: Maybe we should be using the biased sample variance.
        var unbiasedSampleVariance = (squareSum - sum * sum / length) / (length - 1);
        return {
            mean: sampleMean,
            variance: unbiasedSampleVariance,
            size: length,
            degreesOfFreedom: length - 1,
        }
    }

    this.movingAverage = function (values, backwardWindowSize, forwardWindowSize) {
        var averages = new Array(values.length);
        // We use naive O(n^2) algorithm for simplicy as well as to avoid accumulating round-off errors.
        for (var i = 0; i < values.length; i++) {
            var sum = 0;
            var count = 0;
            for (var j = i - backwardWindowSize; j <= i + forwardWindowSize; j++) {
                if (j >= 0 && j < values.length) {
                    sum += values[j];
                    count++;
                }
            }
            averages[i] = sum / count;
        }
        return averages;
    }

    this.cumulativeMovingAverage = function (values) {
        var averages = new Array(values.length);
        var sum = 0;
        for (var i = 0; i < values.length; i++) {
            sum += values[i];
            averages[i] = sum / (i + 1);
        }
        return averages;
    }

    this.exponentialMovingAverage = function (values, smoothingFactor) {
        var averages = new Array(values.length);
        var movingAverage = values[0];
        averages[0] = movingAverage;
        for (var i = 1; i < values.length; i++) {
            movingAverage = smoothingFactor * values[i] + (1 - smoothingFactor) * movingAverage;
            averages[i] = movingAverage;
        }
        return averages;
    }

    // The return value is the starting indices of each segment.
    this.segmentTimeSeriesGreedyWithStudentsTTest = function (values, minLength) {
        if (values.length < 2)
            return [0];
        var segments = new Array;
        recursivelySplitIntoTwoSegmentsAtMaxTIfSignificantlyDifferent(values, 0, values.length, minLength, segments);
        segments.push(values.length);
        return segments;
    }

    this.debuggingSegmentation = false;
    this.segmentTimeSeriesByMaximizingSchwarzCriterion = function (values, segmentCountWeight, gridSize) {
        // Split the time series into grids since splitIntoSegmentsUntilGoodEnough is O(n^2).
        var gridLength = gridSize || 500;
        var totalSegmentation = [0];
        for (var gridCount = 0; gridCount < Math.ceil(values.length / gridLength); gridCount++) {
            var gridValues = values.slice(gridCount * gridLength, (gridCount + 1) * gridLength);
            var segmentation = splitIntoSegmentsUntilGoodEnough(gridValues, segmentCountWeight);

            if (Statistics.debuggingSegmentation)
                console.log('grid=' + gridCount, segmentation);

            for (var i = 1; i < segmentation.length - 1; i++)
                totalSegmentation.push(gridCount * gridLength + segmentation[i]);
        }

        if (Statistics.debuggingSegmentation)
            console.log('Final Segmentation', totalSegmentation);

        totalSegmentation.push(values.length);

        return totalSegmentation;
    }

    function recursivelySplitIntoTwoSegmentsAtMaxTIfSignificantlyDifferent(values, startIndex, length, minLength, segments) {
        var tMax = 0;
        var argTMax = null;
        for (var i = 1; i < length - 1; i++) {
            var firstLength = i;
            var secondLength = length - i;
            if (firstLength < minLength || secondLength < minLength)
                continue;
            var result = Statistics.computeWelchsT(values, startIndex, firstLength, values, startIndex + i, secondLength, 0.9);
            if (result.significantlyDifferent && result.t > tMax) {
                tMax = result.t;
                argTMax = i;
            }
        }
        if (!tMax) {
            segments.push(startIndex);
            return;
        }
        recursivelySplitIntoTwoSegmentsAtMaxTIfSignificantlyDifferent(values, startIndex, argTMax, minLength, segments);
        recursivelySplitIntoTwoSegmentsAtMaxTIfSignificantlyDifferent(values, startIndex + argTMax, length - argTMax, minLength, segments);
    }

    var tDistributionByOneSidedProbability = {
        0.9: [
            3.077684, 1.885618, 1.637744, 1.533206, 1.475884, 1.439756, 1.414924, 1.396815, 1.383029, 1.372184,
            1.363430, 1.356217, 1.350171, 1.345030, 1.340606, 1.336757, 1.333379, 1.330391, 1.327728, 1.325341,
            1.323188, 1.321237, 1.319460, 1.317836, 1.316345, 1.314972, 1.313703, 1.312527, 1.311434, 1.310415,
            1.309464, 1.308573, 1.307737, 1.306952, 1.306212, 1.305514, 1.304854, 1.304230, 1.303639, 1.303077,
            1.302543, 1.302035, 1.301552, 1.301090, 1.300649, 1.300228, 1.299825, 1.299439, 1.299069, 1.298714,

            1.298373, 1.298045, 1.297730, 1.297426, 1.297134, 1.296853, 1.296581, 1.296319, 1.296066, 1.295821,
            1.295585, 1.295356, 1.295134, 1.294920, 1.294712, 1.294511, 1.294315, 1.294126, 1.293942, 1.293763,
            1.293589, 1.293421, 1.293256, 1.293097, 1.292941, 1.292790, 1.292643, 1.292500, 1.292360, 1.292224,
            1.292091, 1.291961, 1.291835, 1.291711, 1.291591, 1.291473, 1.291358, 1.291246, 1.291136, 1.291029,
            1.290924, 1.290821, 1.290721, 1.290623, 1.290527, 1.290432, 1.290340, 1.290250, 1.290161, 1.290075],
        0.95: [
            6.313752, 2.919986, 2.353363, 2.131847, 2.015048, 1.943180, 1.894579, 1.859548, 1.833113, 1.812461,
            1.795885, 1.782288, 1.770933, 1.761310, 1.753050, 1.745884, 1.739607, 1.734064, 1.729133, 1.724718,
            1.720743, 1.717144, 1.713872, 1.710882, 1.708141, 1.705618, 1.703288, 1.701131, 1.699127, 1.697261,
            1.695519, 1.693889, 1.692360, 1.690924, 1.689572, 1.688298, 1.687094, 1.685954, 1.684875, 1.683851,
            1.682878, 1.681952, 1.681071, 1.680230, 1.679427, 1.678660, 1.677927, 1.677224, 1.676551, 1.675905,

            1.675285, 1.674689, 1.674116, 1.673565, 1.673034, 1.672522, 1.672029, 1.671553, 1.671093, 1.670649,
            1.670219, 1.669804, 1.669402, 1.669013, 1.668636, 1.668271, 1.667916, 1.667572, 1.667239, 1.666914,
            1.666600, 1.666294, 1.665996, 1.665707, 1.665425, 1.665151, 1.664885, 1.664625, 1.664371, 1.664125,
            1.663884, 1.663649, 1.663420, 1.663197, 1.662978, 1.662765, 1.662557, 1.662354, 1.662155, 1.661961,
            1.661771, 1.661585, 1.661404, 1.661226, 1.661052, 1.660881, 1.660715, 1.660551, 1.660391, 1.660234],
        0.975: [
            12.706205, 4.302653, 3.182446, 2.776445, 2.570582, 2.446912, 2.364624, 2.306004, 2.262157, 2.228139,
            2.200985, 2.178813, 2.160369, 2.144787, 2.131450, 2.119905, 2.109816, 2.100922, 2.093024, 2.085963,
            2.079614, 2.073873, 2.068658, 2.063899, 2.059539, 2.055529, 2.051831, 2.048407, 2.045230, 2.042272,
            2.039513, 2.036933, 2.034515, 2.032245, 2.030108, 2.028094, 2.026192, 2.024394, 2.022691, 2.021075,
            2.019541, 2.018082, 2.016692, 2.015368, 2.014103, 2.012896, 2.011741, 2.010635, 2.009575, 2.008559,

            2.007584, 2.006647, 2.005746, 2.004879, 2.004045, 2.003241, 2.002465, 2.001717, 2.000995, 2.000298,
            1.999624, 1.998972, 1.998341, 1.997730, 1.997138, 1.996564, 1.996008, 1.995469, 1.994945, 1.994437,
            1.993943, 1.993464, 1.992997, 1.992543, 1.992102, 1.991673, 1.991254, 1.990847, 1.990450, 1.990063,
            1.989686, 1.989319, 1.988960, 1.988610, 1.988268, 1.987934, 1.987608, 1.987290, 1.986979, 1.986675,
            1.986377, 1.986086, 1.985802, 1.985523, 1.985251, 1.984984, 1.984723, 1.984467, 1.984217, 1.983972],
        0.99: [
            31.820516, 6.964557, 4.540703, 3.746947, 3.364930, 3.142668, 2.997952, 2.896459, 2.821438, 2.763769,
            2.718079, 2.680998, 2.650309, 2.624494, 2.602480, 2.583487, 2.566934, 2.552380, 2.539483, 2.527977,
            2.517648, 2.508325, 2.499867, 2.492159, 2.485107, 2.478630, 2.472660, 2.467140, 2.462021, 2.457262,
            2.452824, 2.448678, 2.444794, 2.441150, 2.437723, 2.434494, 2.431447, 2.428568, 2.425841, 2.423257,
            2.420803, 2.418470, 2.416250, 2.414134, 2.412116, 2.410188, 2.408345, 2.406581, 2.404892, 2.403272,

            2.401718, 2.400225, 2.398790, 2.397410, 2.396081, 2.394801, 2.393568, 2.392377, 2.391229, 2.390119,
            2.389047, 2.388011, 2.387008, 2.386037, 2.385097, 2.384186, 2.383302, 2.382446, 2.381615, 2.380807,
            2.380024, 2.379262, 2.378522, 2.377802, 2.377102, 2.376420, 2.375757, 2.375111, 2.374482, 2.373868,
            2.373270, 2.372687, 2.372119, 2.371564, 2.371022, 2.370493, 2.369977, 2.369472, 2.368979, 2.368497,
            2.368026, 2.367566, 2.367115, 2.366674, 2.366243, 2.365821, 2.365407, 2.365002, 2.364606, 2.364217]
    };
    function oneSidedToTwoSidedProbability(probability) { return 2 * probability - 1; }
    function twoSidedToOneSidedProbability(probability) { return (1 - (1 - probability) / 2); }

    function splitIntoSegmentsUntilGoodEnough(values, BirgeAndMassartC) {
        if (values.length < 2)
            return [0, values.length];

        var matrix = new SampleVarianceUpperTriangularMatrix(values);

        var SchwarzCriterionBeta = Math.log1p(values.length - 1) / values.length;

        BirgeAndMassartC = BirgeAndMassartC || 2.5; // Suggested by the authors.
        var BirgeAndMassartPenalization = function (segmentCount) {
            return segmentCount * (1 + BirgeAndMassartC * Math.log1p(values.length / segmentCount - 1));
        }

        var segmentation;
        var minTotalCost = Infinity;
        var maxK = Math.min(50, values.length);

        for (var k = 1; k < maxK; k++) {
            var start = Date.now();
            var result = findOptimalSegmentation(values, matrix, k);
            var cost = result.cost / values.length;
            var penalty = SchwarzCriterionBeta * BirgeAndMassartPenalization(k);
            if (cost + penalty < minTotalCost) {
                minTotalCost = cost + penalty;
                segmentation = result.segmentation;
            } else
                maxK = Math.min(maxK, k + 3);
            if (Statistics.debuggingSegmentation)
                console.log('splitIntoSegmentsUntilGoodEnough', k, Date.now() - start, cost + penalty);
        }

        return segmentation;
    }

    function allocateCostUpperTriangularForSegmentation(values, segmentCount)
    {
        // Dynamic programming. cost[i][k] = The cost to segmenting values up to i into k segments.
        var cost = new Array(values.length);
        for (var segmentEnd = 0; segmentEnd < values.length; segmentEnd++)
            cost[segmentEnd] = new Float32Array(segmentCount + 1);
        return cost;
    }

    function allocatePreviousNodeForSegmentation(values, segmentCount)
    {
        // previousNode[i][k] = The start of the last segment in an optimal segmentation that ends at i with k segments.
        var previousNode = new Array(values.length);
        for (var i = 0; i < values.length; i++)
            previousNode[i] = new Array(segmentCount + 1);
        return previousNode;
    }

    function findOptimalSegmentationInternal(cost, previousNode, values, costMatrix, segmentCount)
    {
        cost[0] = new Float32Array([0]); // The cost of segmenting single value is always 0.
        previousNode[0] = [-1];
        for (var segmentStart = 0; segmentStart < values.length; segmentStart++) {
            var costOfOptimalSegmentationThatEndAtCurrentStart = cost[segmentStart];
            for (var k = 0; k < segmentCount; k++) {
                var noSegmentationOfLenghtKEndsAtCurrentStart = previousNode[segmentStart][k] === undefined;
                if (noSegmentationOfLenghtKEndsAtCurrentStart)
                    continue;
                for (var segmentEnd = segmentStart + 1; segmentEnd < values.length; segmentEnd++) {
                    var costOfOptimalSegmentationOfLengthK = costOfOptimalSegmentationThatEndAtCurrentStart[k];
                    var costOfCurrentSegment = costMatrix.costBetween(segmentStart, segmentEnd);
                    var totalCost = costOfOptimalSegmentationOfLengthK + costOfCurrentSegment;
                    if (previousNode[segmentEnd][k + 1] === undefined || totalCost < cost[segmentEnd][k + 1]) {
                        cost[segmentEnd][k + 1] = totalCost;
                        previousNode[segmentEnd][k + 1] = segmentStart;
                    }
                }
            }
        }
    }

    function findOptimalSegmentation(values, costMatrix, segmentCount) {
        var cost = allocateCostUpperTriangularForSegmentation(values, segmentCount);
        var previousNode = allocatePreviousNodeForSegmentation(values, segmentCount);

        findOptimalSegmentationInternal(cost, previousNode, values, costMatrix, segmentCount);

        if (Statistics.debuggingSegmentation) {
            console.log('findOptimalSegmentation with', segmentCount, 'segments');
            for (var end = 0; end < values.length; end++) {
                for (var k = 0; k <= segmentCount; k++) {
                    var start = previousNode[end][k];
                    if (start === undefined)
                        continue;
                    console.log(`C(segment=[${start}, ${end + 1}], segmentCount=${k})=${cost[end][k]}`);
                }
            }
        }

        var segmentEnd = values.length - 1;
        var segmentation = new Array(segmentCount + 1);
        segmentation[segmentCount] = values.length;
        for (var k = segmentCount; k > 0; k--) {
            segmentEnd = previousNode[segmentEnd][k];
            segmentation[k - 1] = segmentEnd;
        }
        var costOfOptimalSegmentation = cost[values.length - 1][segmentCount];

        if (Statistics.debuggingSegmentation)
            console.log('Optimal segmentation:', segmentation, 'with cost =', costOfOptimalSegmentation);

        return {segmentation: segmentation, cost: costOfOptimalSegmentation};
    }

    function SampleVarianceUpperTriangularMatrix(values) {
        // The cost of segment (i, j].
        var costMatrix = new Array(values.length - 1);
        for (var i = 0; i < values.length - 1; i++) {
            var remainingValueCount = values.length - i - 1;
            costMatrix[i] = new Float32Array(remainingValueCount);
            var sum = values[i];
            var squareSum = sum * sum;
            costMatrix[i][0] = 0;
            for (var j = i + 1; j < values.length; j++) {
                var currentValue = values[j];
                sum += currentValue;
                squareSum += currentValue * currentValue;
                var sampleSize = j - i + 1;
                var stdev = Statistics.sampleStandardDeviation(sampleSize, sum, squareSum);
                costMatrix[i][j - i - 1] = stdev > 0 ? sampleSize * Math.log(stdev * stdev) : 0;
            }
        }
        this.costMatrix = costMatrix;
    }

    SampleVarianceUpperTriangularMatrix.prototype.costBetween = function(from, to) {
        if (from >= this.costMatrix.length || from == to)
            return 0; // The cost of the segment that starts at the last data point is 0.
        return this.costMatrix[from][to - from - 1];
    }

})();

if (typeof module != 'undefined')
    module.exports = Statistics;

class AsyncTask {

    constructor(method, args)
    {
        this._method = method;
        this._args = args;
    }

    execute()
    {
        if (!(this._method in Statistics))
            throw `${this._method} is not a valid method of Statistics`;

        AsyncTask._asyncMessageId++;

        var startTime = Date.now();
        var method = this._method;
        var args = this._args;
        return new Promise(function (resolve, reject) {
            AsyncTaskWorker.waitForAvailableWorker(function (worker) {
                worker.sendTask({id: AsyncTask._asyncMessageId, method: method, args: args}).then(function (data) {
                    var startLatency = data.workerStartTime - startTime;
                    var totalTime = Date.now() - startTime;
                    var callback = data.status == 'resolve' ? resolve : reject;
                    callback({result: data.result, workerId: worker.id(), startLatency: startLatency, totalTime: totalTime, workerTime: data.workerTime});
                });
            });
        });
    }

    static isAvailable()
    {
        return typeof Worker !== 'undefined';
    }
}

AsyncTask._asyncMessageId = 0;

class AsyncTaskWorker {

    // Takes a callback instead of returning a promise because a worker can become unavailable before the end of the current microtask.
    static waitForAvailableWorker(callback)
    {
        var worker = this._makeWorkerEventuallyAvailable();
        if (worker)
            callback(worker);
        else
            this._queue.push(callback);
    }

    static _makeWorkerEventuallyAvailable()
    {
        var worker = this._findAvailableWorker();
        if (worker)
            return worker;

        var canStartMoreWorker = this._workerSet.size < this._maxWorkerCount;
        if (!canStartMoreWorker)
            return null;

        if (this._latestStartTime > Date.now() - 50) {
            setTimeout(function () {
                if (!AsyncTaskWorker._queue.length)
                    return;
                var worker = AsyncTaskWorker._findAvailableWorker();
                if (worker)
                    AsyncTaskWorker._queue.pop()(worker);
            }, 50);
            return null;
        }
        return new AsyncTaskWorker;
    }

    static _findAvailableWorker()
    {
        for (var worker of this._workerSet) {
            if (!worker._currentTaskId)
                return worker;
        }
        return null;
    }

    constructor()
    {
        this._webWorker = new Worker('async-task.js');
        this._webWorker.onmessage = this._didRecieveMessage.bind(this);
        this._id = AsyncTaskWorker._workerId;
        this._startTime = Date.now();
        this._currentTaskId = null;
        this._callback = null;

        AsyncTaskWorker._latestStartTime = this._startTime;
        AsyncTaskWorker._workerId++;
        AsyncTaskWorker._workerSet.add(this);
    }

    id() { return this._id; }

    sendTask(task)
    {
        console.assert(!this._currentTaskId);
        console.assert(task.id);
        var self = this;
        this._currentTaskId = task.id;
        return new Promise(function (resolve) {
            self._webWorker.postMessage(task);
            self._callback = resolve;
        });
    }

    _didRecieveMessage(event)
    {
        var callback = this._callback;

        console.assert(this._currentTaskId);
        this._currentTaskId = null;
        this._callback = null;

        if (AsyncTaskWorker._queue.length)
            AsyncTaskWorker._queue.pop()(this);
        else {
            var self = this;
            setTimeout(function () {
                if (self._currentTaskId == null)
                    AsyncTaskWorker._workerSet.delete(self);
            }, 1000);
        }

        callback(event.data);
    }

    static workerDidRecieveMessage(event)
    {
        var data = event.data;
        var id = data.id;
        var method = Statistics[data.method];
        var startTime = Date.now();
        try {
            var returnValue = method.apply(Statistics, data.args);
            postMessage({'id': id, 'status': 'resolve', 'result': returnValue, 'workerStartTime': startTime, 'workerTime': Date.now() - startTime});
        } catch (error) {
            postMessage({'id': id, 'status': 'reject', 'result': error.toString(), 'workerStartTime': startTime, 'workerTime': Date.now() - startTime});
            throw error;
        }
    }
}

AsyncTaskWorker._maxWorkerCount = 4;
AsyncTaskWorker._workerSet = new Set;
AsyncTaskWorker._queue = [];
AsyncTaskWorker._workerId = 1;
AsyncTaskWorker._latestStartTime = 0;

if (typeof module == 'undefined' && typeof window == 'undefined' && typeof importScripts != 'undefined') { // Inside a worker
    //importScripts('/shared/statistics.js');
    onmessage = AsyncTaskWorker.workerDidRecieveMessage.bind(AsyncTaskWorker);
}

if (typeof module != 'undefined')
    module.exports.AsyncTask = AsyncTask;