IndexError in find_atom_pairs: matches/iso_matches index mismatch (frag.network.decorate)

[alanbchristie]

frag.network.decorate.find_atom_pairs() raises IndexError: list index out of range for some molecules.

Where (frag/network/decorate.py, v1.1.0)

def find_atom_pairs(smiles_input, get_indices, iso_smiles):
    matches = re.findall(XE_PATT, smiles_input)
    if iso_smiles:
        iso_matches = re.findall(XE_PATT, iso_smiles)
    else:
        iso_matches = None
        isotope = None
    ind_list = []
    for i, match in enumerate(matches):
        index = int(match[:-2])
        if iso_matches:
            isotope = int(iso_matches[i][:-2])   # IndexError when len(iso_matches) < len(matches)
        indices = ret_comb_index(index, get_indices, isotope)
        ind_list.append(indices)
    return ind_list

Cause

matches comes from smiles_input; iso_matches comes from a different string, iso_smiles. The loop runs over matches but subscripts iso_matches[i], assuming the two lists are equal length and aligned 1:1. When iso_smiles yields fewer [<n>Xe] tokens than smiles_input, iso_matches[i] goes out of range.

Reproduced via

get_vect_indices_for_mol() -> del_link_coord(..., iso_labels=False) -> get_atom_coords -> find_atom_pairs. Hit in production via the fragalysis-backend /api/vector endpoint (xchem/fragalysis-backend#941):

  File "frag/network/decorate.py", line 225, in find_atom_pairs
    isotope = int(iso_matches[i][:-2])
IndexError: list index out of range

Suggested fix

Don't assume positional correspondence between matches and iso_matches -- bounds-check i < len(iso_matches), or rework how isotope labels are paired to the plain matches.

Version: 1.1.0.