ByteSubstring: add an option to used an owned needle (not borrowed)

Author: tsnoam

src/fallback.rs

@@ -1,6 +1,8 @@
 // TODO: Try boxing the closure to see if we can hide the type
 // TODO: Or maybe use a function pointer?
 
+use std::borrow::Cow;
+
 pub struct Bytes<F>
 where
     F: Fn(u8) -> bool,
@@ -22,11 +24,11 @@ where
 }
 
 pub struct ByteSubstring<'a> {
-    needle: &'a [u8],
+    needle: Cow<'a, [u8]>,
 }
 
 impl<'a> ByteSubstring<'a> {
-    pub /* const */ fn new(needle: &'a[u8]) -> Self {
+    pub /* const */ fn from_cow(needle: Cow<'a, [u8]>) -> Self {
         ByteSubstring { needle }
     }
 
@@ -38,6 +40,6 @@ impl<'a> ByteSubstring<'a> {
     pub fn find(&self, haystack: &[u8]) -> Option<usize> {
         haystack
             .windows(self.needle.len())
-            .position(|window| window == self.needle)
+            .position(|window| window == self.needle.as_ref())
     }
 }

src/lib.rs

@@ -152,6 +152,7 @@ extern crate proptest;
 extern crate region;
 
 use std::marker::PhantomData;
+use std::borrow::Cow;
 
 include!(concat!(env!("OUT_DIR"), "/src/macros.rs"));
 
@@ -301,12 +302,17 @@ pub struct ByteSubstring<'a> {
 
 impl<'a> ByteSubstring<'a> {
     pub /* const */ fn new(needle: &'a [u8]) -> Self {
+        Self::from_cow(Cow::Borrowed(needle))
+    }
+
+    pub /* const */ fn from_cow(needle: Cow<'a, [u8]>) -> Self {
+        let needle_clone = needle.clone();
         ByteSubstring {
             #[cfg(target_arch = "x86_64")]
-            simd: simd::ByteSubstring::new(needle),
+            simd: simd::ByteSubstring::from_cow(needle),
 
             #[cfg(not(target_feature = "sse4.2"))]
-            fallback: fallback::ByteSubstring::new(needle),
+            fallback: fallback::ByteSubstring::from_cow(needle_clone),
         }
     }
 

src/simd.rs

@@ -6,6 +6,7 @@ use std::{
     arch::x86_64::{
         __m128i, _mm_cmpestri, _mm_cmpestrm, _mm_extract_epi16, _mm_loadu_si128, _SIDD_CMP_EQUAL_ORDERED,
     },
+    borrow::Cow,
     cmp::min,
     slice,
 };
@@ -245,13 +246,18 @@ impl<'b> PackedCompareControl for &'b Bytes {
 }
 
 pub struct ByteSubstring<'a> {
-    complete_needle: &'a [u8],
+    complete_needle: Cow<'a, [u8]>,
     needle: __m128i,
     needle_len: i32,
 }
 
 impl<'a> ByteSubstring<'a> {
-    pub /* const */ fn new(needle: &'a[u8]) -> Self {
+    #[cfg(test)]
+    pub /* const */ fn new(needle: &'a [u8]) -> Self {
+        Self::from_cow(Cow::Borrowed(needle))
+    }
+
+    pub /* const */ fn from_cow(needle: Cow<'a, [u8]>) -> Self {
         use std::cmp;
 
         let mut simd_needle = [0; 16];
@@ -277,7 +283,7 @@ impl<'a> ByteSubstring<'a> {
         while let Some(idx) = find(PackedCompare::<_, _SIDD_CMP_EQUAL_ORDERED>(self), &haystack[offset..]) {
             let abs_offset = offset + idx;
             // Found a match, but is it really?
-            if haystack[abs_offset..].starts_with(self.complete_needle) {
+            if haystack[abs_offset..].starts_with(&self.complete_needle) {
                 return Some(abs_offset);
             }