Remove punctuation

The punctuation people forget is the typographic kind. A word processor silently replaces " with “ and ”, ' with ’, and -- with —, so text copied out of one contains characters that are not on a keyboard and are not in the ASCII punctuation range. A remover built from a list of typed characters leaves every one of them behind, which is why the result still looks wrong.

The apostrophe is the case that needs a decision rather than a rule. Removing it turns isn’t into isnt and it’s into its — which is what you want for text analysis, where the words should match regardless of contraction, and wrong for anything a person will read. There is no correct answer that suits both, so it is worth checking the output rather than assuming.

A common reason for wanting this is preparing text for comparison or search: punctuation makes two otherwise identical strings differ, and stripping it is the standard normalisation step alongside lowercasing and removing accents. For that purpose removing the apostrophe is right.