feat(transformer): implement some of jsx decode entities (#1086)

crates/oxc_codegen/src/gen.rs

@@ -1076,6 +1076,9 @@ fn print_str<const MINIFY: bool>(s: &str, p: &mut Codegen<{ MINIFY }>) {
             // <https://github.com/tc39/proposal-json-superset>
             LS => p.print_str(b"\\u2028"),
             PS => p.print_str(b"\\u2029"),
+            '\u{a0}' => {
+                p.print_str(b"\\xA0");
+            }
             _ => p.print_str(c.escape_default().to_string().as_bytes()),
         }
     }

crates/oxc_transformer/src/react_jsx/mod.rs

@@ -442,19 +442,7 @@ impl<'a> ReactJsx<'a> {
 
     fn transform_jsx_child(&mut self, child: &JSXChild<'a>) -> Option<Expression<'a>> {
         match child {
-            JSXChild::Text(text) => {
-                let text = text.value.trim();
-                (!text.trim().is_empty()).then(|| {
-                    let text = text
-                        .split(char::is_whitespace)
-                        .map(str::trim)
-                        .filter(|c| !c.is_empty())
-                        .collect::<std::vec::Vec<_>>()
-                        .join(" ");
-                    let s = StringLiteral::new(SPAN, text.into());
-                    self.ast.literal_string_expression(s)
-                })
-            }
+            JSXChild::Text(text) => self.transform_jsx_text(text),
             JSXChild::ExpressionContainer(e) => match &e.expression {
                 JSXExpression::Expression(e) => Some(self.ast.copy(e)),
                 JSXExpression::EmptyExpression(_) => None,
@@ -467,4 +455,53 @@ impl<'a> ReactJsx<'a> {
             }
         }
     }
+
+    fn transform_jsx_text(&self, text: &JSXText) -> Option<Expression<'a>> {
+        let text = text.value.trim();
+        (!text.trim().is_empty()).then(|| {
+            let text = text
+                .split(char::is_whitespace)
+                .map(str::trim)
+                .filter(|c| !c.is_empty())
+                .map(Self::decode_jsx_text)
+                .collect::<std::vec::Vec<_>>()
+                .join(" ");
+            let s = StringLiteral::new(SPAN, text.into());
+            self.ast.literal_string_expression(s)
+        })
+    }
+
+    /// * Replace entities like "&nbsp;", "&#123;", and "&#xDEADBEEF;" with the characters they encode.
+    /// * See https://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references
+    /// Code adapted from <https://github.com/microsoft/TypeScript/blob/514f7e639a2a8466c075c766ee9857a30ed4e196/src/compiler/transformers/jsx.ts#L617C1-L635>
+    fn decode_jsx_text(s: &str) -> String {
+        let mut buffer = vec![];
+        let mut chars = s.bytes().enumerate();
+        let mut prev = 0;
+        while let Some((i, c)) = chars.next() {
+            if c == b'&' {
+                let start = i;
+                let mut end = None;
+                for (j, c) in chars.by_ref() {
+                    if c == b';' {
+                        end.replace(j);
+                        break;
+                    }
+                }
+                if let Some(end) = end {
+                    let word = &s[start + 1..end];
+                    buffer.extend_from_slice(s[prev..start].as_bytes());
+                    prev = end + 1;
+                    match word {
+                        "amp" => buffer.extend_from_slice(b"&"),
+                        "nbsp" => buffer.extend_from_slice("\u{a0}".as_bytes()),
+                        _ => {}
+                    }
+                }
+            }
+        }
+        buffer.extend_from_slice(s[prev..].as_bytes());
+        // Safety: The buffer is constructed from valid utf chars.
+        unsafe { String::from_utf8_unchecked(buffer) }
+    }
 }

tasks/transform_conformance/babel.snap.md

@@ -1,4 +1,4 @@
-Passed: 226/1083
+Passed: 229/1083
 
 # All Passed:
 * babel-plugin-transform-numeric-separator
@@ -804,7 +804,7 @@ Passed: 226/1083
 * regression/11061/input.mjs
 * variable-declaration/non-null-in-optional-chain/input.ts
 
-# babel-plugin-transform-react-jsx (77/172)
+# babel-plugin-transform-react-jsx (80/172)
 * autoImport/after-polyfills-compiled-to-cjs/input.mjs
 * autoImport/after-polyfills-script-not-supported/input.js
 * autoImport/auto-import-react-source-type-module/input.js
@@ -853,12 +853,10 @@ Passed: 226/1083
 * react/should-escape-xhtml-jsxtext-babel-7/input.js
 * react/should-handle-attributed-elements/input.js
 * react/should-not-strip-nbsp-even-coupled-with-other-whitespace/input.js
-* react/should-not-strip-tags-with-a-single-child-of-nbsp/input.js
 * react/should-support-xml-namespaces-if-flag/input.js
 * react/should-throw-error-namespaces-if-not-flag/input.js
 * react/should-warn-when-importSource-is-set/input.js
 * react/should-warn-when-importSource-pragma-is-set/input.js
-* react/weird-symbols/input.js
 * react/wraps-props-in-react-spread-for-first-spread-attributes-babel-7/input.js
 * react/wraps-props-in-react-spread-for-last-spread-attributes-babel-7/input.js
 * react/wraps-props-in-react-spread-for-middle-spread-attributes-babel-7/input.js
@@ -880,7 +878,6 @@ Passed: 226/1083
 * react-automatic/should-escape-xhtml-jsxtext-babel-7/input.js
 * react-automatic/should-handle-attributed-elements/input.js
 * react-automatic/should-not-strip-nbsp-even-coupled-with-other-whitespace/input.js
-* react-automatic/should-not-strip-tags-with-a-single-child-of-nbsp/input.js
 * react-automatic/should-properly-handle-comments-between-props/input.js
 * react-automatic/should-throw-error-namespaces-if-not-flag/input.js
 * react-automatic/should-throw-when-filter-is-specified/input.js