I feel like it might be possible with dataflow analysis. Stepping through the regex maintaining a liveness set or something like that. Sort of like computing ex

I feel like it might be possible with dataflow analysis. Stepping through the regex maintaining a liveness set or something like that. Sort of like computing exemplar inputs, but with repetition as permitted exemplars. Honestly probably end up re-encoding the regex in some other format, perhaps with 'optimizations applied.'