Implement Span::sub_span and Span::lines_span.
#456
Conversation
|
This looks reasonable to me at a glance. @dragostis ? |
|
Some concerns I have (that may not actually be real issues):
|
|
I am also interested in |
There are similar requests for additional flexibility ( such as #469 ). There are other types that provide flexibility such as ErrorVariant::CustomError. Subspans could be similarly useful.
I prefer
@tomtau is this PR blocked waiting for better documentation? |
|
@Alextopher unless @tangmi plans to revive this PR soon, yes, I think two main items to fix will be to adjust the docs a little bit (to provide a full example usage) and to resolve a minor conflict with the latest target branch. |
| let start = match range.start_bound() { | ||
| std::ops::Bound::Included(&offset) => offset, | ||
| std::ops::Bound::Excluded(&offset) => offset + 1, | ||
| std::ops::Bound::Unbounded => 0, | ||
| }; | ||
| let end = match range.end_bound() { | ||
| std::ops::Bound::Included(&offset) => Some(offset + 1), | ||
| std::ops::Bound::Excluded(&offset) => Some(offset), | ||
| std::ops::Bound::Unbounded => None, | ||
| }; | ||
| let s = self.as_str(); | ||
| if s.get(start..end.unwrap_or(s.len())).is_some() { |
There was a problem hiding this comment.
@Alextopher one more item to fix: the merge conflict seems to be about core vs std. pest now supports no_std, so these ops may either need to be adjusted for no_std or sub_span should be feature-guarded for std
| Some(unsafe { | ||
| Span::new_unchecked(self.span.input, line_start, self.pos.min(self.span.end)) | ||
| }) |
There was a problem hiding this comment.
@Alextopher could this be done without unsafe? if not, it should have a Safety comment explaining why it is safe or when things could go wrong
|
Sounds good. I started by creating a new PR with just |
|
closing in favour of #681 |
Prototype implementation for #455