Isn't this more robust though? I feel like using lines to detect next rows is very flimsy. I usually deal with CSV containing full press articles, I am quite sure the CSV.Chunks method would fail without the correct hyperparameter. This method seems more, I dunno, "adaptative".