The issue was that smart quotes, as well as any other non-ascii punctuation characters, were not handled like ascii punctuation in the parsing of emphasis/strong tokens.
Solved by including all unicode punctuation in the set of punctuation characters.
The issue was that smart quotes, as well as any other non-ascii punctuation characters, were not handled like ascii punctuation in the parsing of emphasis/strong tokens. Solved by including all unicode punctuation in the set of punctuation characters.