# Utility functions
# Purpose: check what new edges have been created today.
# Created by Hui on 20 July 2021

def get_edge_set(fname):
    result = set()
    with open(fname) as f:
        for line in f:
            line = line.strip()
            lst = line.split('\t')
            if len(lst) == 10:
                target_id = lst[0].split()[0].upper()
                tf_id = lst[1].split()[0].upper()
                result.add(target_id + '_' + tf_id)
    return result


def make_new_edges_file(Sold, Snew, fname_new, output_file):
    Sdiff = Snew.difference(Sold)
    result = []
    with open(fname_new) as f:
        for line in f:
            line = line.strip()
            lst = line.split('\t')
            if len(lst) == 10:
                target_id = lst[0].split()[0].upper()
                tf_id = lst[1].split()[0].upper()
                if target_id + '_' + tf_id in Sdiff: # this is a new edge line, keep it
                    result.append(line)
    with open(output_file, 'w') as f:
        header = '\t'.join(['target_id target_name', 'source_id source_name', 'score', 'type of score', 'RNA-seqs', 'ChIP-seq', 'Log likelihood', 'Date', 'Strength', 'Inference method']) + '\n'
        f.write(header)
        for line in result:
            f.write(line + '\n')


if __name__ == '__main__':
    S2 = get_edge_set('/home/lanhui/brain/Data/temp/edges.txt')
    S1 = get_edge_set('/home/lanhui/brain/Data/temp/edges.txt.old')
    make_new_edges_file(S1, S2, '/home/lanhui/brain/Data/temp/edges.txt', 'mynewedges.txt')