def resolve_first_ref()

in dataset-construction/src/ndb_data/generation/map_kelm.py [0:0]

• 50 lines of code
• 13 McCabe index (conditional complexity)

def resolve_first_ref(ref):
    ref = ref.replace("$COMMA$", ",")
    toks = ref.split()

    parsed = []
    next_id = 0
    is_relation = False

    iteration = 0
    prev_end = 0
    while next_id < len(toks) and iteration <= 2:
        iteration += 1

        text, resolved_id, startptr, nextptr = get_longest(
            toks, next_id, restrict_relation=is_relation
        )

        if toks[next_id] == ",":
            next_id += 1
            continue

        if text is not None and resolved_id is not None:
            next_id = nextptr + 1
            is_relation = not is_relation
            parsed.append((text, resolved_id))

            # Fix previous resolution
            if prev_end != startptr:
                recovered = try_recovery(" ".join(toks[:startptr]))
                if recovered is not None:
                    parsed[-2] = recovered
                else:

                    print(f"Failed to fix {ref}")
                    print(parsed)
                    del parsed[-2]

            elif len(parsed) >= 3:
                # Fix this resolution
                aa = clean(" ".join(toks[startptr:])).split()
                if len(parsed[-1][0].split()) < len(aa):
                    recovery = try_recovery(" ".join(toks[startptr:]))
                    if recovery is not None:
                        parsed[-1] = recovery
                    else:
                        print(f"Failed to fix2 {ref}")
                        print(parsed)
                        del parsed[-1]

            prev_end = nextptr + 1
            if len(parsed) >= 3:
                break

        else:
            if len(parsed) > 1:
                print(ref)
                print("Early stop")
                print(toks[next_id:])
                print(parsed)
                print()
                break

    return parsed