IndexOutOfBoundsException when sorting by clusters in ArcChart

[gfbustamante]

In the drawChart method of ArcChart, if sortByCluster is true, a TreeSet is used to order clusters:

https://github.com/HanSolo/charts/blob/jdk17/src/main/java/eu/hansolo/fx/charts/ArcChart.java#L586

        // Sort items by max sum of outgoing reversed
        List<PlotItem> sortedItems;
        if (getSortByCluster()) {
            sortedItems = new LinkedList<>();
            TreeSet<Cluster> clusters = new TreeSet<>(items.stream().filter(item -> null != item.getCluster()).map(PlotItem::getCluster).collect(Collectors.toSet()));
            clusters.forEach(cluster -> sortedItems.addAll(cluster.getSortedItems()));
            if (sortedItems.isEmpty()) { sortedItems.addAll(items); }
        } else {
            sortedItems = new LinkedList<>(items);
            Collections.sort(sortedItems, Comparator.comparingDouble(PlotItem::getSumOfOutgoing).reversed());
        }

The compareTo method of Cluster just uses getMaxValue():

https://github.com/HanSolo/charts/blob/jdk17/src/main/java/eu/hansolo/fx/charts/Cluster.java#L175

    @Override public int compareTo(final Cluster other) {
        if (getMaxValue() < other.getMaxValue()) {
            return 1;
        } else if (getMaxValue() > other.getMaxValue()) {
            return -1;
        } else {
            return 0;
        }
    }

As a result, multiple clusters with the same max value collide in the TreeSet and only a subset of clusters are stored. This causes a discrepancy between the anticipated total items and what is actually added to the sortedItems collection and results in a IndexOutOfBoundsException with the subsequent sortedItems.get(i) operation:

https://github.com/HanSolo/charts/blob/jdk17/src/main/java/eu/hansolo/fx/charts/ArcChart.java#L598

        for (int i = 0 ; i < noOfItems ; i++) {
            PlotItem item     = sortedItems.get(i);
            //double   itemSize = Helper.clamp(minItemSize, maxItemSize, item.getSumOfOutgoing() * itemSizeFactor);
            double   itemX    = insetX + i * stepSizeX;
            double   itemY    = centerY;
            itemPoints.put(item, new Point(itemX, itemY));
        }

Is there a reason that clusters are sorted via a set vs. a non-unique data structure with sorting? My workaround in my project was to subclass Cluster and override compareTo to also use name when maxValues are the same.

I am referencing code from the jdk17 branch.

[HanSolo]

There was probably no specific reason for this. Feel free to create a PR with your solution. Otherwise I will take a look when I’ll find some time.

[gfbustamante]

Thanks @HanSolo. I will take a swing at a PR. You have done some brilliant work here, so I appreciate you being responsive and continuing to support the community.

[HanSolo]

Thx, well it was just for the fun of it and it’s great to see people using it 😁