called_alot.py

import numpy as np

def make_exclusionlist(fragment, capped, max_length, ignore_capping_hydrogens = True):
    # is this int? - else change
    frag_serials = np.asarray(fragment.serials, dtype=int)
    valid_capped_fragments  = np.fromiter((frag for frag in capped if capfrag.ignore == False), int)
    valid_capped_fragment_serials  = [[serial for serial, name in zip(capfrag.serials, capfrag.names) if (ignore_capping_hydrogens and name != " H* ")] for capfrag in valid_capped_fragments]
    # are serials unique? - else change
    matches = (np.intersect1d(fragment.serials, cap_serials, assume_unique = True) for cap_serials in valid_capped_fragment_serials)

    # this can probably be done faster
    # for each element, get all the serials of capped fragment containing element
    excl = [[] for _ in frag_serials]
    for i, (cap_serials, matched) in enumerate(zip(valid_capped_fragment_serials, matches)):
        for j in matched:
            excl[j].extend(cap_serials)

    # all the unique elements with zeros padded
    excl = [np.unique(x).resize(max_length).tolist() for x in excl]

    # serial == -1 should return [-1] + [0]*(max_length-1) right?
    for i, v in enumerate(frag_serials):
        if v == -1:
            excl[i] = - np.ones(1).resize(max_length).tolist()


    return excl