Rollup merge of #100311 - xfix:lines-fix-handling-of-bare-cr, r=ChrisDenton

Fix handling of trailing bare CR in str::lines

Continuing from #91191.

Fixes #94435.
This commit is contained in:
Dylan DPC 2023-03-23 00:00:30 +05:30 committed by GitHub
commit d694f47baa
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 26 additions and 14 deletions

View File

@ -1499,13 +1499,25 @@ fn test_split_whitespace() {
#[test]
fn test_lines() {
let data = "\nMäry häd ä little lämb\n\r\nLittle lämb\n";
let lines: Vec<&str> = data.lines().collect();
assert_eq!(lines, ["", "Märy häd ä little lämb", "", "Little lämb"]);
let data = "\r\nMäry häd ä little lämb\n\nLittle lämb"; // no trailing \n
let lines: Vec<&str> = data.lines().collect();
assert_eq!(lines, ["", "Märy häd ä little lämb", "", "Little lämb"]);
fn t(data: &str, expected: &[&str]) {
let lines: Vec<&str> = data.lines().collect();
assert_eq!(lines, expected);
}
t("", &[]);
t("\n", &[""]);
t("\n2nd", &["", "2nd"]);
t("\r\n", &[""]);
t("bare\r", &["bare\r"]);
t("bare\rcr", &["bare\rcr"]);
t("Text\n\r", &["Text", "\r"]);
t(
"\nMäry häd ä little lämb\n\r\nLittle lämb\n",
&["", "Märy häd ä little lämb", "", "Little lämb"],
);
t(
"\r\nMäry häd ä little lämb\n\nLittle lämb",
&["", "Märy häd ä little lämb", "", "Little lämb"],
);
}
#[test]

View File

@ -13,7 +13,7 @@
use super::pattern::Pattern;
use super::pattern::{DoubleEndedSearcher, ReverseSearcher, Searcher};
use super::validations::{next_code_point, next_code_point_reverse};
use super::LinesAnyMap;
use super::LinesMap;
use super::{BytesIsNotEmpty, UnsafeBytesToStr};
use super::{CharEscapeDebugContinue, CharEscapeDefault, CharEscapeUnicode};
use super::{IsAsciiWhitespace, IsNotEmpty, IsWhitespace};
@ -1104,7 +1104,7 @@ fn next_back(&mut self) -> Option<&'a str>
#[stable(feature = "rust1", since = "1.0.0")]
#[must_use = "iterators are lazy and do nothing unless consumed"]
#[derive(Clone, Debug)]
pub struct Lines<'a>(pub(super) Map<SplitTerminator<'a, char>, LinesAnyMap>);
pub struct Lines<'a>(pub(super) Map<SplitInclusive<'a, char>, LinesMap>);
#[stable(feature = "rust1", since = "1.0.0")]
impl<'a> Iterator for Lines<'a> {

View File

@ -1011,7 +1011,7 @@ pub fn split_ascii_whitespace(&self) -> SplitAsciiWhitespace<'_> {
#[stable(feature = "rust1", since = "1.0.0")]
#[inline]
pub fn lines(&self) -> Lines<'_> {
Lines(self.split_terminator('\n').map(LinesAnyMap))
Lines(self.split_inclusive('\n').map(LinesMap))
}
/// An iterator over the lines of a string.
@ -2604,10 +2604,10 @@ fn default() -> Self {
impl_fn_for_zst! {
/// A nameable, cloneable fn type
#[derive(Clone)]
struct LinesAnyMap impl<'a> Fn = |line: &'a str| -> &'a str {
let l = line.len();
if l > 0 && line.as_bytes()[l - 1] == b'\r' { &line[0 .. l - 1] }
else { line }
struct LinesMap impl<'a> Fn = |line: &'a str| -> &'a str {
let Some(line) = line.strip_suffix('\n') else { return line };
let Some(line) = line.strip_suffix('\r') else { return line };
line
};
#[derive(Clone)]