I have a networkx graph, in which each node has a x and y attributes that form a geogrpahic coordinate. I want to use this information to merge nodes that are within a specific distance of each other, say 10 meters. I have tried this as a iterative with a while function that merges nodes until none are within 10 meters. I currently am working on creating a list of all the nodes that need to be merged together. I am facing two difficulties because the runtime for computing the distances is very long for my almost 2000 nodes and I am not sure if I should make this a iterative approach where I compute distances then merge and then compute again until no nodes are within the 10 meters anymore or if I can somehow immediately combine the nodes into a group of nodes and merge them all at once.

This is the current functions I am using

def compute_distances(graph):
    distances = {}
    for node1, node2 in combinations(graph.nodes(data=True), 2):
        try:
            if node1[1]['type'] == node2[1]['type']:
                distance = geodesic((node1[1]['y'], node1[1]['x']), (node2[1]['y'], node2[1]['x'])).meters
                distances[(node1[0], node2[0])] = distance
                print(f"Distance between {node1[0]} and {node2[0]}: {distance} meters")
        except KeyError as e:
            print(f"KeyError: {e} for nodes {node1[0]} and {node2[0]}")
    return distances

def find_nodes_to_merge(distances, threshold_distance):
    nodes_to_merge = []

    for (node1, node2), distance in distances.items():
        if distance < threshold_distance:
            nodes_to_merge.append((node1, node2))

    return nodes_to_merge

I would like that instead of pairs of nodes for nodes_to_merge I get all the nodes that need to be merged grouped into clusters already in a list of lists potentially.