Lookahead for closing curly quote when encountering opening curly quote

[estebank]

When encountering curly quotes, the compiler now points at the opening one suggesting changing it for a regular straight quote. It should also look ahead looking for a closing curly quote char and suggest changing that one too, because when fixing only the first one, the parser will interpret this as an unclosed string.

fn main() {
    println!(“”);
}

error: unknown start of token: \u{201c}
 --> src/main.rs:2:14
  |
2 |     println!(“”);
  |              ^
help: Unicode character '“' (Left Double Quotation Mark) looks like '"' (Quotation Mark), but it is not
  |
2 |     println!("”);
  |              ^

After applying suggestion:

error: unterminated double quote string
 --> src/main.rs:2:14
  |
2 |       println!("”);
  |  ______________^
3 | | }
  | |_^

Ideally

error: unknown start of token: \u{201c}
 --> src/main.rs:2:14
  |
2 |     println!(“”);
  |              ^-
help: Unicode character '“' (Left Double Quotation Mark) and '”' (Right Double Quotation Mark) look like '"' (Quotation Mark), but it are not
  |
2 |     println!("");
  |              ^^

[pmccarter]

I could maybe do this with some mentoring, if not too difficult or trivial...

[estebank]

@pmccarter this method is the one that supplies the error:

rust/src/libsyntax/parse/lexer/unicode_chars.rs

Lines 326 to 353 in e544947

	crate fn check_for_substitution<'a>(reader: &StringReader<'a>,
	ch: char,
	err: &mut DiagnosticBuilder<'a>) -> bool {
	UNICODE_ARRAY
	.iter()
	.find(|&&(c, _, _)| c == ch)
	.map(|&(_, u_name, ascii_char)| {
	let span = Span::new(reader.pos, reader.next_pos, NO_EXPANSION);
	match ASCII_ARRAY.iter().find(|&&(c, _)| c == ascii_char) {
	Some(&(ascii_char, ascii_name)) => {
	let msg =
	format!("Unicode character '{}' ({}) looks like '{}' ({}), but it is not",
	ch, u_name, ascii_char, ascii_name);
	err.span_suggestion(
	span,
	&msg,
	ascii_char.to_string(),
	Applicability::MaybeIncorrect);
	true
	},
	None => {
	let msg = format!("substitution character not found for '{}'", ch);
	reader.sess.span_diagnostic.span_bug_no_panic(span, &msg);
	false
	}
	}
	}).unwrap_or(false)
	}

It gets called from the lexer:

rust/src/libsyntax/parse/lexer/mod.rs

Line 1645 in e544947

unicode_chars::check_for_substitution(self, c, &mut err);

It should be possible to special case the “ char to continue advancing, using self.nextch() I think, until finding a ”, and if found, add that to the diagnostic.

The Forge has info on setting up the repo and building your own rustc. The workflow is usually running ./x.py test src/test/ui --stage 1 --bless and modifying/creating test files in src/test/ui.