The visitor optionally lets us ignore a UTF-8 BOM on the first header field if present, however, if that field starts with a double-quote, then the tokenizer will fail to treat it as quoted.
The good news is that with the new model I'm doing for #12, all the CsvInput implementations could be the ones that optionally ignore a UTF-8 BOM if present.
At the time of this comment, the top result for a Google search for the term "FEFF", on its own, is a blog post of someone having this exact same problem with Ruby's CSV parser. Cursively is in good company.
The visitor optionally lets us ignore a UTF-8 BOM on the first header field if present, however, if that field starts with a double-quote, then the tokenizer will fail to treat it as quoted.
The good news is that with the new model I'm doing for #12, all the
CsvInput
implementations could be the ones that optionally ignore a UTF-8 BOM if present.