#!/usr/bin/env python3

import networkx as nx
import argparse
import sys

from deconvolution.d2graph import d2_graph as d2
from deconvolution.d2graph import d2_algorithms as d2a


def parse_arguments():
    parser = argparse.ArgumentParser(description='Reduce the number of edges of a d2 graph. The reduction is performed on edges that are redundant.')
    parser.add_argument('lcp_graph', help='lcp graph to reduce. Must be a gefx formated file.')
    parser.add_argument('--output_lcpg_name', '-o', default="", help="Output file prefix.")
    parser.add_argument('--component_min_size', '-c', type=int, default=10, help="Minimum size of a component to keep it in the lcp graph, after the edge simplifications.")

    args = parser.parse_args()

    if args.output_lcpg_name == "":
        args.output_lcpg_name = args.lcp_graph[:args.lcp_graph.rfind('.')] + "_reduced.gexf"

    return args


def main():
    # Parsing the arguments and validate them
    args = parse_arguments()

    lcpg_name = args.lcp_graph
    if not lcpg_name.endswith(".gexf"):
        print("Inputs file must be gexf formatted", file=sys.stderr)
        exit(1)
    
    # Loading
    print("--- lcp graph loading ---")
    lcpg = d2.D2Graph()
    lcpg.load(lcpg_name)

    # Algorithms for reduction
    print("--- 3-reduction processing ---")
    d2a.transitive_reduction(lcpg)

    # Take the principal component
    print("--- Extract and write the graph ---")
    components = [c for c in nx.connected_components(lcpg) if len(c) > args.component_min_size]
    nodes = set()
    for comp in components:
        nodes.update(comp)
    subgraph = lcpg.subgraph(nodes)

    # Write the simplified graph
    nx.write_gexf(subgraph, args.output_lcpg_name)


if __name__ == "__main__":
    main()