Add functions for attribute value normalization

closes tafia#371

Author: dralley

Changelog.md

@@ -14,6 +14,10 @@
 
 - [#541]: Deserialize specially named `$text` enum variant in [externally tagged]
   enums from textual content
+- [#379]: Improved compliance with the XML attribute value normalization process by
+  adding `Attribute::normalized_value()` and `Attribute::normalized_value_with()`,
+  which ought to be used in place of `Attribute::unescape_value()` and
+  `Attribute::unescape_value_with()`
 
 ### Bug Fixes
 
@@ -22,10 +26,15 @@
 
 ### Misc Changes
 
+### New Tests
+
+- [#379]: Added tests for attribute value normalization
+
 [externally tagged]: https://serde.rs/enum-representations.html#externally-tagged
 [#490]: https://github.com/tafia/quick-xml/pull/490
 [#537]: https://github.com/tafia/quick-xml/issues/537
 [#541]: https://github.com/tafia/quick-xml/pull/541
+[#379]: https://github.com/tafia/quick-xml/pull/379
 
 ## 0.27.1 -- 2022-12-28
 

benches/macrobenches.rs

@@ -43,14 +43,13 @@ static INPUTS: &[(&str, &str)] = &[
     ("players.xml", PLAYERS),
 ];
 
-// TODO: use fully normalized attribute values
 fn parse_document_from_str(doc: &str) -> XmlResult<()> {
     let mut r = Reader::from_str(doc);
     loop {
         match criterion::black_box(r.read_event()?) {
             Event::Start(e) | Event::Empty(e) => {
                 for attr in e.attributes() {
-                    criterion::black_box(attr?.decode_and_unescape_value(&r)?);
+                    criterion::black_box(attr?.decode_and_normalize_value(&r)?);
                 }
             }
             Event::Text(e) => {

benches/microbenches.rs

@@ -242,6 +242,50 @@ fn attributes(c: &mut Criterion) {
             assert_eq!(count, 150);
         })
     });
+
+    group.finish();
+}
+
+/// Benchmarks normalizing attribute values
+fn attribute_value_normalization(c: &mut Criterion) {
+    let mut group = c.benchmark_group("attribute_value_normalization");
+
+    group.bench_function("noop_short", |b| {
+        b.iter(|| {
+            criterion::black_box(unescape("foobar")).unwrap();
+        })
+    });
+
+    group.bench_function("noop_long", |b| {
+        b.iter(|| {
+            criterion::black_box(unescape("just a bit of text without any entities")).unwrap();
+        })
+    });
+
+    group.bench_function("replacement_chars", |b| {
+        b.iter(|| {
+            criterion::black_box(unescape("just a bit\n of text without\tany entities")).unwrap();
+        })
+    });
+
+    group.bench_function("char_reference", |b| {
+        b.iter(|| {
+            let text = "prefix "some stuff","more stuff"";
+            criterion::black_box(unescape(text)).unwrap();
+            let text = "&<";
+            criterion::black_box(unescape(text)).unwrap();
+        })
+    });
+
+    group.bench_function("entity_reference", |b| {
+        b.iter(|| {
+            let text = "age > 72 && age < 21";
+            criterion::black_box(unescape(text)).unwrap();
+            let text = ""what's that?"";
+            criterion::black_box(unescape(text)).unwrap();
+        })
+    });
+
     group.finish();
 }
 
@@ -354,6 +398,7 @@ criterion_group!(
     read_resolved_event_into,
     one_event,
     attributes,
+    attribute_value_normalization,
     escaping,
     unescaping,
 );

src/errors.rs

@@ -77,6 +77,7 @@ impl From<EscapeError> for Error {
 }
 
 impl From<AttrError> for Error {
+    /// Creates a new `Error::InvalidAttr` from the given error
     #[inline]
     fn from(error: AttrError) -> Self {
         Error::InvalidAttr(error)