Make UniStr it’s own thing

1 files changed, 84 insertions, 0 deletions

diff --git a/oryxc/src/unistr.rs b/oryxc/src/unistr.rs
new file mode 100644
index 0000000..5dc8160
--- /dev/null
+++ b/oryxc/src/unistr.rs
@@ -0,0 +1,84 @@
+use std::hash::{Hash, Hasher};
+
+use unicode_normalization::{self, IsNormalized, UnicodeNormalization};
+
+#[repr(transparent)]
+#[derive(Copy, Clone, Debug, Eq)]
+pub struct UniStr<'a>(pub &'a str);
+
+impl Hash for UniStr<'_> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+		/* In the ASCII common case we use .bytes() to avoid decoding
+		 * every codepoint (a no-op in ASCII) */
+		if self.0.is_ascii() {
+			self.0.bytes().for_each(|c| (c as char).hash(state));
+		} else if unicode_normalization::is_nfkd_quick(self.0.chars())
+			== IsNormalized::Yes
+		{
+			self.0.chars().for_each(|c| c.hash(state));
+		} else {
+			self.0.nfkd().for_each(|c| c.hash(state));
+		}
+    }
+}
+
+impl PartialEq for UniStr<'_> {
+	fn eq(&self, other: &Self) -> bool {
+		/* Most code is ASCII, and normalization is obviously a lot
+		 * slower than not normalizing, so we try to only normalize when
+		 * we have to */
+
+		if self.0.is_ascii() && other.0.is_ascii() {
+			return self.0 == other.0;
+		}
+
+		return match (
+			unicode_normalization::is_nfkd_quick(self.0.chars())
+				== IsNormalized::Yes,
+			unicode_normalization::is_nfkd_quick(other.0.chars())
+				== IsNormalized::Yes,
+		) {
+			(true, true) => self.0 == other.0,
+			(true, false) => {
+				self.0.chars().map(|b| b as char).eq(other.0.nfkd())
+			},
+			(false, true) => {
+				self.0.nfkd().eq(other.0.chars().map(|b| b as char))
+			},
+			(false, false) => self.0.nfkd().eq(other.0.nfkd()),
+		};
+	}
+}
+
+#[cfg(test)]
+mod tests {
+    use std::hash::{DefaultHasher, Hash, Hasher};
+
+    use super::UniStr;
+
+    #[test]
+    fn test_unistr_eq() {
+	    assert_eq!(UniStr("fishi"), UniStr("fishᵢ"));
+	    assert_eq!(UniStr("fishi"), UniStr("fishi"));
+	    assert_eq!(UniStr("fishi"), UniStr("fishᵢ"));
+	    assert_eq!(UniStr("fishᵢ"), UniStr("fishᵢ"));
+	    assert_eq!(UniStr("corné"), UniStr("corné"));
+    }
+
+    #[test]
+    fn test_unistr_hash() {
+	    for (lhs, rhs) in &[
+		    (UniStr("fishi"), UniStr("fishᵢ")),
+		    (UniStr("fishi"), UniStr("fishi")),
+		    (UniStr("fishi"), UniStr("fishᵢ")),
+		    (UniStr("fishᵢ"), UniStr("fishᵢ")),
+		    (UniStr("corné"), UniStr("corné")),
+	    ] {
+		    let mut hashl = DefaultHasher::new();
+		    let mut hashr = DefaultHasher::new();
+		    lhs.hash(&mut hashl);
+		    rhs.hash(&mut hashr);
+		    assert_eq!(hashl.finish(), hashr.finish());
+	    }
+    }
+}