Fix parsing for invalid control characters in class atoms #85
Conversation
raskad
force-pushed
the
fix-class-atom-invalid-control-escapes
branch
from
9e28b12
to
699f7a6
Compare
| fn try_consume_bracket_class_atom(&mut self) -> Result<Option<ClassAtom>, Error> { | ||
| fn try_consume_bracket_class_atom( | ||
| &mut self, | ||
| ) -> Result<Option<(ClassAtom, Option<ClassAtom>)>, Error> { |
There was a problem hiding this comment.
This is a complicated return type. Here's two alternatives for simplifying it:
- When encountering "\c" in a bracket, instead of returning both, unread the "c" and only return the backslash. The next iteration through the loop will then pick up the "c".
- Instead return a
Result<Vec<ClassAtom>, Error>
I think the first will end up quite clean assuming it works.
There was a problem hiding this comment.
That sounds good, I will try to change it to the first if it works.
There was a problem hiding this comment.
Works well. I implemented it in #88.
| // ClassAtomNoDash :: \ [lookahead = c] | ||
| _ => Ok(Some(( | ||
| ClassAtom::CodePoint(u32::from('\\')), | ||
| Some(ClassAtom::CodePoint(u32::from('c'))), |
There was a problem hiding this comment.
I wasn't able to follow this - from what I can tell in the spec it says:
Return the numeric value of U+005C (REVERSE SOLIDUS).
but the actual behavior in Chrome agrees with your code - do you know where this is documented?
There was a problem hiding this comment.
I think you can read the spec like you have suggested to write to code in your other comment. It just says that if we encounter a \ and the next char is a c we return U+005C. Then the spec "returns" and we read the c as a single char next.
In non unicode mode
ClassAtomNoDash :: \ [lookahead = c]was not being parsed correctly.