From 8f9777ce7aecfe1a163a915e3245466b9dd9ac2e Mon Sep 17 00:00:00 2001 From: Ellie Huxtable Date: Mon, 16 Mar 2026 15:22:49 -0700 Subject: Squashed 'crates/atuin-nucleo/' content from commit 4253de9f git-subtree-dir: crates/atuin-nucleo git-subtree-split: 4253de9faabb4e5c6d81d946a5e35a90f87347ee --- matcher/src/utf32_str/tests.rs | 44 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 44 insertions(+) create mode 100644 matcher/src/utf32_str/tests.rs (limited to 'matcher/src/utf32_str/tests.rs') diff --git a/matcher/src/utf32_str/tests.rs b/matcher/src/utf32_str/tests.rs new file mode 100644 index 00000000..a38c8875 --- /dev/null +++ b/matcher/src/utf32_str/tests.rs @@ -0,0 +1,44 @@ +use crate::{Utf32Str, Utf32String}; + +#[test] +fn test_utf32str_ascii() { + /// Helper function for testing + fn expect_ascii(src: &str, is_ascii: bool) { + let mut buffer = Vec::new(); + assert!(Utf32Str::new(src, &mut buffer).is_ascii() == is_ascii); + assert!(Utf32String::from(src).slice(..).is_ascii() == is_ascii); + assert!(Utf32String::from(src.to_owned()).slice(..).is_ascii() == is_ascii); + } + + // ascii + expect_ascii("", true); + expect_ascii("a", true); + expect_ascii("a\nb", true); + expect_ascii("\n\r", true); + + // not ascii + expect_ascii("aü", false); + expect_ascii("au\u{0308}", false); + + // windows-style newline + expect_ascii("a\r\nb", false); + expect_ascii("ü\r\n", false); + expect_ascii("\r\n", false); +} + +#[test] +fn test_grapheme_truncation() { + // ascii is preserved + let s = Utf32String::from("ab"); + assert_eq!(s.slice(..).get(0), 'a'); + assert_eq!(s.slice(..).get(1), 'b'); + + // windows-style newline is truncated to '\n' + let s = Utf32String::from("\r\n"); + assert_eq!(s.slice(..).get(0), '\n'); + + // normal graphemes are truncated to the first character + let s = Utf32String::from("u\u{0308}\r\n"); + assert_eq!(s.slice(..).get(0), 'u'); + assert_eq!(s.slice(..).get(1), '\n'); +} -- cgit v1.3.1