Auto merge of #87062 - poliorcetics:fix-85462, r=dtolnay

Make StrSearcher behave correctly on empty needle

Fix #85462.

This will not affect ABI since the other variant of the enum is bigger.
It may break some code, but that would be very strange: usually people
don't continue after the first `Done` (or `None` for a normal iterator).

`@rustbot` label T-libs A-str A-patterns
This commit is contained in:
bors 2021-07-27 00:31:20 +00:00
commit c51607e031
2 changed files with 58 additions and 2 deletions

View file

@ -1873,6 +1873,47 @@ mod pattern {
"* \t",
[Reject(0, 1), Reject(1, 2), Reject(2, 3),]
);
// See #85462
#[test]
fn str_searcher_empty_needle_after_done() {
// Empty needle and haystack
{
let mut searcher = "".into_searcher("");
assert_eq!(searcher.next(), SearchStep::Match(0, 0));
assert_eq!(searcher.next(), SearchStep::Done);
assert_eq!(searcher.next(), SearchStep::Done);
assert_eq!(searcher.next(), SearchStep::Done);
let mut searcher = "".into_searcher("");
assert_eq!(searcher.next_back(), SearchStep::Match(0, 0));
assert_eq!(searcher.next_back(), SearchStep::Done);
assert_eq!(searcher.next_back(), SearchStep::Done);
assert_eq!(searcher.next_back(), SearchStep::Done);
}
// Empty needle and non-empty haystack
{
let mut searcher = "".into_searcher("a");
assert_eq!(searcher.next(), SearchStep::Match(0, 0));
assert_eq!(searcher.next(), SearchStep::Reject(0, 1));
assert_eq!(searcher.next(), SearchStep::Match(1, 1));
assert_eq!(searcher.next(), SearchStep::Done);
assert_eq!(searcher.next(), SearchStep::Done);
assert_eq!(searcher.next(), SearchStep::Done);
let mut searcher = "".into_searcher("a");
assert_eq!(searcher.next_back(), SearchStep::Match(1, 1));
assert_eq!(searcher.next_back(), SearchStep::Reject(0, 1));
assert_eq!(searcher.next_back(), SearchStep::Match(0, 0));
assert_eq!(searcher.next_back(), SearchStep::Done);
assert_eq!(searcher.next_back(), SearchStep::Done);
assert_eq!(searcher.next_back(), SearchStep::Done);
}
}
}
macro_rules! generate_iterator_test {

View file

@ -928,6 +928,8 @@ struct EmptyNeedle {
end: usize,
is_match_fw: bool,
is_match_bw: bool,
// Needed in case of an empty haystack, see #85462
is_finished: bool,
}
impl<'a, 'b> StrSearcher<'a, 'b> {
@ -941,6 +943,7 @@ impl<'a, 'b> StrSearcher<'a, 'b> {
end: haystack.len(),
is_match_fw: true,
is_match_bw: true,
is_finished: false,
}),
}
} else {
@ -966,13 +969,19 @@ unsafe impl<'a, 'b> Searcher<'a> for StrSearcher<'a, 'b> {
fn next(&mut self) -> SearchStep {
match self.searcher {
StrSearcherImpl::Empty(ref mut searcher) => {
if searcher.is_finished {
return SearchStep::Done;
}
// empty needle rejects every char and matches every empty string between them
let is_match = searcher.is_match_fw;
searcher.is_match_fw = !searcher.is_match_fw;
let pos = searcher.position;
match self.haystack[pos..].chars().next() {
_ if is_match => SearchStep::Match(pos, pos),
None => SearchStep::Done,
None => {
searcher.is_finished = true;
SearchStep::Done
}
Some(ch) => {
searcher.position += ch.len_utf8();
SearchStep::Reject(pos, searcher.position)
@ -1045,12 +1054,18 @@ unsafe impl<'a, 'b> ReverseSearcher<'a> for StrSearcher<'a, 'b> {
fn next_back(&mut self) -> SearchStep {
match self.searcher {
StrSearcherImpl::Empty(ref mut searcher) => {
if searcher.is_finished {
return SearchStep::Done;
}
let is_match = searcher.is_match_bw;
searcher.is_match_bw = !searcher.is_match_bw;
let end = searcher.end;
match self.haystack[..end].chars().next_back() {
_ if is_match => SearchStep::Match(end, end),
None => SearchStep::Done,
None => {
searcher.is_finished = true;
SearchStep::Done
}
Some(ch) => {
searcher.end -= ch.len_utf8();
SearchStep::Reject(searcher.end, end)