Faster escape routines

Author: dralley

Cargo.toml

@@ -16,6 +16,8 @@ document-features = { version = "0.2", optional = true }
 encoding_rs = { version = "0.8", optional = true }
 serde = { version = "1.0", optional = true }
 memchr = "2.5"
+jetscii = "0.5.2"
+once_cell = "1.12.0"
 
 [dev-dependencies]
 criterion = "0.3"

src/escapei.rs

@@ -1,13 +1,20 @@
 //! Manage xml character escapes
 
-use memchr;
 use std::borrow::Cow;
 use std::collections::HashMap;
 use std::ops::Range;
 
+use jetscii::bytes;
+use memchr;
+use once_cell::sync::Lazy;
+
 #[cfg(test)]
 use pretty_assertions::assert_eq;
 
+static XML_ESCAPE_BYTES: Lazy<jetscii::BytesConst> =
+    Lazy::new(|| bytes!(b'<', b'>', b'&', b'\'', b'"'));
+static XML_PARTIAL_ESCAPE_BYTES: Lazy<jetscii::BytesConst> = Lazy::new(|| bytes!(b'<', b'>', b'&'));
+
 /// Error for XML escape/unescqpe.
 #[derive(Debug)]
 pub enum EscapeError {
@@ -66,31 +73,17 @@ impl std::error::Error for EscapeError {}
 /// Escapes a `&[u8]` and replaces all xml special characters (<, >, &, ', ") with their
 /// corresponding xml escaped value.
 pub fn escape(raw: &[u8]) -> Cow<[u8]> {
-    #[inline]
-    fn to_escape(b: u8) -> bool {
-        match b {
-            b'<' | b'>' | b'\'' | b'&' | b'"' => true,
-            _ => false,
-        }
-    }
-
-    _escape(raw, to_escape)
+    // _escape(raw, |ch| matches!(ch, b'<' | b'>' | b'\'' | b'&' | b'"'))
+    simd_escape(raw, &XML_ESCAPE_BYTES)
 }
 
 /// Should only be used for escaping text content. In xml text content, it is allowed
 /// (though not recommended) to leave the quote special characters " and ' unescaped.
 /// This function escapes a `&[u8]` and replaces xml special characters (<, >, &) with
 /// their corresponding xml escaped value, but does not escape quote characters.
 pub fn partial_escape(raw: &[u8]) -> Cow<[u8]> {
-    #[inline]
-    fn to_escape(b: u8) -> bool {
-        match b {
-            b'<' | b'>' | b'&' => true,
-            _ => false,
-        }
-    }
-
-    _escape(raw, to_escape)
+    // _escape(raw, |ch| matches!(ch, b'<' | b'>' | b'&'))
+    simd_escape(raw, &XML_PARTIAL_ESCAPE_BYTES)
 }
 
 /// Escapes a `&[u8]` and replaces a subset of xml special characters (<, >, &, ', ") with their
@@ -112,7 +105,46 @@ fn _escape<F: Fn(u8) -> bool>(raw: &[u8], escape_chars: F) -> Cow<[u8]> {
             b'\'' => escaped.extend_from_slice(b"&apos;"),
             b'&' => escaped.extend_from_slice(b"&amp;"),
             b'"' => escaped.extend_from_slice(b"&quot;"),
-            _ => unreachable!("Only '<', '>','\', '&' and '\"' are escaped"),
+            c @ _ => unreachable!(
+                "Found {} but only '<', '>', ', '&' and '\"' are escaped",
+                c as char
+            ),
+        }
+        pos = new_pos + 1;
+    }
+
+    if let Some(mut escaped) = escaped {
+        if let Some(raw) = raw.get(pos..) {
+            escaped.extend_from_slice(raw);
+        }
+        Cow::Owned(escaped)
+    } else {
+        Cow::Borrowed(raw)
+    }
+}
+
+/// Escapes a `&[u8]` and replaces all xml special characters (<, >, &, ', ") with their
+/// corresponding xml escaped value.
+pub fn simd_escape<'a>(raw: &'a [u8], escape_matcher: &jetscii::BytesConst) -> Cow<'a, [u8]> {
+    let mut escaped = None;
+    let mut pos = 0;
+    while let Some(i) = escape_matcher.find(&raw[pos..]) {
+        if escaped.is_none() {
+            escaped = Some(Vec::with_capacity(raw.len()));
+        }
+        let escaped = escaped.as_mut().expect("initialized");
+        let new_pos = pos + i;
+        escaped.extend_from_slice(&raw[pos..new_pos]);
+        match raw[new_pos] {
+            b'<' => escaped.extend_from_slice(b"&lt;"),
+            b'>' => escaped.extend_from_slice(b"&gt;"),
+            b'\'' => escaped.extend_from_slice(b"&apos;"),
+            b'&' => escaped.extend_from_slice(b"&amp;"),
+            b'"' => escaped.extend_from_slice(b"&quot;"),
+            c @ _ => unreachable!(
+                "Found {} but only '<', '>', ', '&' and '\"' are escaped",
+                c as char
+            ),
         }
         pos = new_pos + 1;
     }