Add str::{Split,RSplit,SplitN,RSplitN,SplitTerminator,RSplitTerminator,SplitInclusive}::as_str methods

[WaffleLapkin]

tl;dr this allows viewing unyelded part of str-split-iterators, like so:

let mut split = "Mary had a little lamb".split(' '); 
assert_eq!(split.as_str(), "Mary had a little lamb");
split.next();
assert_eq!(split.as_str(), "had a little lamb");
split.by_ref().for_each(drop);
assert_eq!(split.as_str(), "");

---

This PR adds semi-identical as_str methods to most str-split-iterators with signatures like &'_ Split<'a, P: Pattern<'a>> -> &'a str (Note: output &str lifetime is bound to the 'a, not the '_). The methods are similar to Chars::as_str

SplitInclusive::as_str is under "str_split_inclusive_as_str" feature gate, all other methods are under "str_split_as_str" feature gate.

Before this PR you had to sum lens of all yielded parts or collect into String to emulate as_str.

[rust-highfive]

r? @kennytm

(rust_highfive has picked a reviewer for you, use r? to override)

[LeSeulArtichaut]

Ping from triage: @kennytm do you want to review this? I'd expect someone from T-libs to be assigned here -- maybe cc @sfackler?

[crlf0710]

This will cause surprising result for those who want to calculate ptr offsets of this &str from the whole &str

[WaffleLapkin]

Is there a use case when you need to do those kinds of tricks?

[WaffleLapkin]

I mean the fix is trivial, but for me, it's not clear if we should provide such guarantees.

[loriopatrick]

I would like the return to be a slice of the provided input.

When parsing a http header, I find myself wanting to see how much is remaining in the buffer to determine if I have all the contents (specified by Content-Length).

Example

let http_response_buffer: &str;
let mut header_lines = http_response_buffer.split("\r\n");

let mut content_length = None;
/* read through header and set content_length, header ends on empty header_line */
/* ... */

let unread_buffer_size = unsafe {
   header_lines.as_str().as_ptr().offset_from(response_buffer.as_ptr())
};

if content_length.unwrap() < unread_buffer_size {
   panic!("http_response_buffer does not contain response body");
}

This would break if http_response_buffer was

Content-Length: 100

granted I am using unsafe but this pattern doesn't seem too unreasonable (I might be biased as it's what I want to do).

Edit

Just realized I could just do header_lines.as_str().len(). Late night for me, my justification isn't really holding water anymore.

[WaffleLapkin]

I was thinking about this today and thought that there should be a function (probably in a crate) like

fn substr_bounds(string: &str, sub: &str) -> Option<Range<usize>> { ... }

let string = "aabb";
let sub = &r[1..3];
assert_eq!(substr_bound(string, sub), Some(1..3));
assert_eq!(substr_bound(string, "ab"), None);

It feels natural, but are there any cases when it's useful? Anyway, as said before, for these kinds of things returning an empty substr may be a good idea.

If there is code that may benefit from this, I could make a PR with a fix.

[HyeonuPark]

It should be related to the semantics of raw pointer comparison.
rust-lang/unsafe-code-guidelines#239

[Dylan-DPC-zz]

r? @dtolnay

[bors]

☔ The latest upstream changes (presumably #76325) made this pull request unmergeable. Please resolve the merge conflicts.

Note that reviewers usually do not review pull requests until merge conflicts are resolved! Once you resolve the conflicts, you should change the labels applied by bors to indicate that your PR is ready for review. Post this as a comment to change the labels:

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

[WaffleLapkin]

Hm. I've fixed the conflict, but is it normal that bors didn't change the labels (S-waiting-on-review->S-waiting-on-author)? It said to run a command after resolving conflicts, but labels are already as they should be after resolving conflicts.

[Dylan-DPC-zz]

@WaffleLapkin not sure why, the change has been reported last week as well, will check it accordingly :)

[dtolnay]

Thanks, this looks good!

Please make a tracking issue and update the stability attributes in the PR.

[WaffleLapkin]

@dtolnay done :)

[leonardo-m]

What's an use case of this as_str method?

[Dylan-DPC-zz]

@bors r=dtolnay

[bors]

📌 Commit 7bd6403 has been approved by dtolnay

[mikevoronov]

@leonardo-m I needed this feature while implementing very simple argument parser.

Let's imagine that I want to parse the following command:
call <module_name> <func_name> [args in json]

So in call after <func_name> extracting, I want to get a remainder of string as &str to parse it like a json with serde. In particular, here I had to use join or collect that implies some overhead.

[crumblingstatue]

Would it be possible to also implement this for SplitWhitespace?

[WaffleLapkin]

@crumblingstatue this should be possible, but it's not trivial. SplitWhitespace uses Filter iterator adaptor, making it impossible to reach inner Split and get the remainder. The only solution I can think of - making Filter's iter field pub(crate).