Encode core::str::CharSearcher::utf8_size as enum

Author: GnomedDev

library/core/src/str/pattern.rs

@@ -348,6 +348,45 @@ pub trait DoubleEndedSearcher<'a>: ReverseSearcher<'a> {}
 // Impl for char
 /////////////////////////////////////////////////////////////////////////////
 
+#[derive(Clone, Copy, Debug)]
+enum Utf8Size {
+    // Values are indexes, so `- 1`
+    One = 0,
+    Two = 1,
+    Three = 2,
+    Four = 3,
+}
+
+impl Utf8Size {
+    fn new(size: usize) -> Option<Self> {
+        match size {
+            1 => Some(Self::One),
+            2 => Some(Self::Two),
+            3 => Some(Self::Three),
+            4 => Some(Self::Four),
+            _ => None,
+        }
+    }
+
+    // # Safety
+    //
+    // `size` must be more than `0` and less than `5`
+    unsafe fn new_unchecked(size: usize) -> Self {
+        // SAFETY: Invariant held by caller
+        unsafe { Self::new(size).unwrap_unchecked() }
+    }
+
+    /// Returns the size value
+    fn get_raw(self) -> usize {
+        (self as usize) + 1
+    }
+
+    fn index(self, arr: &[u8; 4]) -> &u8 {
+        // SAFETY: max value is 3, which indexes to the 4th element.
+        unsafe { arr.get_unchecked(self as usize) }
+    }
+}
+
 /// Associated type for `<char as Pattern<'a>>::Searcher`.
 #[derive(Clone, Debug)]
 pub struct CharSearcher<'a> {
@@ -368,9 +407,8 @@ pub struct CharSearcher<'a> {
     /// The character being searched for
     needle: char,
 
-    // safety invariant: `utf8_size` must be less than 5
     /// The number of bytes `needle` takes up when encoded in utf8.
-    utf8_size: usize,
+    utf8_size: Utf8Size,
     /// A utf8 encoded copy of the `needle`
     utf8_encoded: [u8; 4],
 }
@@ -413,8 +451,7 @@ unsafe impl<'a> Searcher<'a> for CharSearcher<'a> {
             // get the haystack after the last character found
             let bytes = self.haystack.as_bytes().get(self.finger..self.finger_back)?;
             // the last byte of the utf8 encoded needle
-            // SAFETY: we have an invariant that `utf8_size < 5`
-            let last_byte = unsafe { *self.utf8_encoded.get_unchecked(self.utf8_size - 1) };
+            let last_byte = *self.utf8_size.index(&self.utf8_encoded);
             if let Some(index) = memchr::memchr(last_byte, bytes) {
                 // The new finger is the index of the byte we found,
                 // plus one, since we memchr'd for the last byte of the character.
@@ -434,10 +471,12 @@ unsafe impl<'a> Searcher<'a> for CharSearcher<'a> {
                 // find something. When we find something the `finger` will be set
                 // to a UTF8 boundary.
                 self.finger += index + 1;
-                if self.finger >= self.utf8_size {
-                    let found_char = self.finger - self.utf8_size;
+
+                let utf8_size = self.utf8_size.get_raw();
+                if self.finger >= utf8_size {
+                    let found_char = self.finger - utf8_size;
                     if let Some(slice) = self.haystack.as_bytes().get(found_char..self.finger) {
-                        if slice == &self.utf8_encoded[0..self.utf8_size] {
+                        if slice == &self.utf8_encoded[0..utf8_size] {
                             return Some((found_char, self.finger));
                         }
                     }
@@ -481,8 +520,7 @@ unsafe impl<'a> ReverseSearcher<'a> for CharSearcher<'a> {
             // get the haystack up to but not including the last character searched
             let bytes = haystack.get(self.finger..self.finger_back)?;
             // the last byte of the utf8 encoded needle
-            // SAFETY: we have an invariant that `utf8_size < 5`
-            let last_byte = unsafe { *self.utf8_encoded.get_unchecked(self.utf8_size - 1) };
+            let last_byte = *self.utf8_size.index(&self.utf8_encoded);
             if let Some(index) = memchr::memrchr(last_byte, bytes) {
                 // we searched a slice that was offset by self.finger,
                 // add self.finger to recoup the original index
@@ -493,14 +531,15 @@ unsafe impl<'a> ReverseSearcher<'a> for CharSearcher<'a> {
                 // char in the paradigm of reverse iteration). For
                 // multibyte chars we need to skip down by the number of more
                 // bytes they have than ASCII
-                let shift = self.utf8_size - 1;
+                let utf8_size = self.utf8_size.get_raw();
+                let shift = utf8_size - 1;
                 if index >= shift {
                     let found_char = index - shift;
-                    if let Some(slice) = haystack.get(found_char..(found_char + self.utf8_size)) {
-                        if slice == &self.utf8_encoded[0..self.utf8_size] {
+                    if let Some(slice) = haystack.get(found_char..(found_char + utf8_size)) {
+                        if slice == &self.utf8_encoded[0..utf8_size] {
                             // move finger to before the character found (i.e., at its start index)
                             self.finger_back = found_char;
-                            return Some((self.finger_back, self.finger_back + self.utf8_size));
+                            return Some((self.finger_back, self.finger_back + utf8_size));
                         }
                     }
                 }
@@ -543,6 +582,9 @@ impl<'a> Pattern<'a> for char {
     fn into_searcher(self, haystack: &'a str) -> Self::Searcher {
         let mut utf8_encoded = [0; 4];
         let utf8_size = self.encode_utf8(&mut utf8_encoded).len();
+
+        // SAFETY: utf8_size is above 0 and below 5
+        let utf8_size = unsafe { Utf8Size::new_unchecked(utf8_size) };
         CharSearcher {
             haystack,
             finger: 0,