Replace decode() function with an explicit state machine

instant-xml/src/de.rs

@@ -341,7 +341,7 @@ pub fn borrow_cow_str<'xml>(
     }
 
     if let Some(value) = deserializer.take_str()? {
-        *into = Some(decode(value));
+        *into = Some(decode(value)?);
     };
 
     deserializer.ignore()?;
@@ -357,7 +357,7 @@ pub fn borrow_cow_slice_u8<'xml>(
     }
 
     if let Some(value) = deserializer.take_str()? {
-        *into = Some(match decode(value) {
+        *into = Some(match decode(value)? {
             Cow::Borrowed(v) => Cow::Borrowed(v.as_bytes()),
             Cow::Owned(v) => Cow::Owned(v.into_bytes()),
         });

instant-xml/src/impls.rs

@@ -260,7 +260,7 @@ impl<'xml> FromXml<'xml> for String {
         }
 
         match deserializer.take_str()? {
-            Some(value) => *into = Some(decode(value).into_owned()),
+            Some(value) => *into = Some(decode(value)?.into_owned()),
             None => return Ok(()),
         }
 
@@ -292,7 +292,7 @@ impl<'xml> FromXml<'xml> for &'xml str {
             None => return Ok(()),
         };
 
-        match decode(value) {
+        match decode(value)? {
             Cow::Borrowed(str) => *into = Some(str),
             Cow::Owned(_) => {
                 return Err(Error::UnexpectedValue(format!(
@@ -484,62 +484,68 @@ fn encode(input: &str) -> Result<Cow<'_, str>, Error> {
     Ok(Cow::Owned(result))
 }
 
-pub(crate) fn decode(input: &str) -> Cow<'_, str> {
+pub(crate) fn decode(input: &str) -> Result<Cow<'_, str>, Error> {
     let mut result = String::with_capacity(input.len());
-    let input_len = input.len();
+    let (mut state, mut last_end) = (DecodeState::Normal, 0);
-
+    for (i, &b) in input.as_bytes().iter().enumerate() {
-    let mut last_end = 0;
+        // use a state machine to find entities
-    while input_len - last_end >= 4 {
+        state = match (state, b) {
-        if input.is_char_boundary(last_end + 4) {
+            (DecodeState::Normal, b'&') => DecodeState::Entity([0; 4], 0),
-            match &input[last_end..(last_end + 4)] {
+            (DecodeState::Normal, _) => DecodeState::Normal,
-                "&lt;" => {
+            (DecodeState::Entity(chars, len), b';') => {
-                    result.push('<');
+                let decoded = match chars[..len] {
-                    last_end += 4;
+                    [b'a', b'm', b'p'] => '&',
-                    continue;
+                    [b'a', b'p', b'o', b's'] => '\'',
-                }
+                    [b'g', b't'] => '>',
-                "&gt;" => {
+                    [b'l', b't'] => '<',
-                    result.push('>');
+                    [b'q', b'u', b'o', b't'] => '"',
-                    last_end += 4;
+                    _ => {
-                    continue;
+                        return Err(Error::InvalidEntity(
-                }
+                            String::from_utf8_lossy(&chars[..len]).into_owned(),
-                _ => (),
+                        ))
-            };
-        }
-
-        if input_len - last_end >= 5 && input.is_char_boundary(last_end + 5) {
-            if &input[last_end..(last_end + 5)] == "&amp;" {
-                result.push('&');
-                last_end += 5;
-                continue;
-            }
-
-            if input_len - last_end >= 6 && input.is_char_boundary(last_end + 6) {
-                match &input[last_end..(last_end + 6)] {
-                    "&apos;" => {
-                        result.push('\'');
-                        last_end += 6;
-                        continue;
                     }
-                    "&quot;" => {
-                        result.push('"');
-                        last_end += 6;
-                        continue;
-                    }
-                    _ => (),
                 };
+
+                let start = i - (len + 1); // current position - (length of entity characters + 1 for '&')
+                if last_end < start {
+                    // Unwrap should be safe: `last_end` and `start` must be at character boundaries.
+                    result.push_str(input.get(last_end..start).unwrap());
+                }
+
+                last_end = i + 1;
+                result.push(decoded);
+                DecodeState::Normal
             }
+            (DecodeState::Entity(mut chars, len), b) => {
+                if len >= 4 {
+                    let mut bytes = Vec::with_capacity(5);
+                    bytes.extend(&chars[..len]);
+                    bytes.push(b);
+                    return Err(Error::InvalidEntity(
+                        String::from_utf8_lossy(&bytes).into_owned(),
+                    ));
+                }
+
+                chars[len] = b;
+                DecodeState::Entity(chars, len + 1)
+            }
+        };
+    }
+
+    Ok(match result.is_empty() {
+        true => Cow::Borrowed(input),
+        false => {
+            // Unwrap should be safe: `last_end` and `input.len()` must be at character boundaries.
+            result.push_str(input.get(last_end..input.len()).unwrap());
+            Cow::Owned(result)
         }
+    })
+}
 
-        result.push_str(input.get(last_end..last_end + 1).unwrap());
+#[derive(Debug)]
-        last_end += 1;
+enum DecodeState {
-    }
+    Normal,
-
+    Entity([u8; 4], usize),
-    result.push_str(input.get(last_end..).unwrap());
-    if result.len() == input.len() {
-        return Cow::Borrowed(input);
-    }
-
-    Cow::Owned(result)
 }
 
 impl<'xml, T: FromXml<'xml>> FromXml<'xml> for Vec<T> {
@@ -771,3 +777,25 @@ impl<'xml> FromXml<'xml> for IpAddr {
 
     const KIND: Kind = Kind::Scalar;
 }
+
+#[cfg(test)]
+mod tests {
+    use super::decode;
+
+    #[test]
+    fn test_decode() {
+        assert_eq!(decode("foo").unwrap(), "foo");
+        assert_eq!(decode("foo &amp; bar").unwrap(), "foo & bar");
+        assert_eq!(decode("foo &lt; bar").unwrap(), "foo < bar");
+        assert_eq!(decode("foo &gt; bar").unwrap(), "foo > bar");
+        assert_eq!(decode("foo &quot; bar").unwrap(), "foo \" bar");
+        assert_eq!(decode("foo &apos; bar").unwrap(), "foo ' bar");
+        assert_eq!(decode("foo &amp;lt; bar").unwrap(), "foo &lt; bar");
+        assert_eq!(decode("&amp; foo").unwrap(), "& foo");
+        assert_eq!(decode("foo &amp;").unwrap(), "foo &");
+        assert_eq!(decode("cbdtéda&amp;sü").unwrap(), "cbdtéda&sü");
+        assert!(decode("&foo;").is_err());
+        assert!(decode("&foobar;").is_err());
+        assert!(decode("cbdtéd&ampü").is_err());
+    }
+}

instant-xml/src/lib.rs

@@ -96,6 +96,8 @@ impl<T> FromXmlOwned for T where T: for<'xml> FromXml<'xml> {}
 pub enum Error {
     #[error("format: {0}")]
     Format(#[from] fmt::Error),
+    #[error("invalid entity: {0}")]
+    InvalidEntity(String),
     #[error("parse: {0}")]
     Parse(#[from] xmlparser::Error),
     #[error("other: {0}")]