Fix JSX comment duplication by properly handling comments in JSX text scanning

Co-authored-by: RyanCavanaugh <[email protected]>

Author: Copilot

src/compiler/emitter.ts

@@ -3896,11 +3896,8 @@ export function createPrinter(printerOptions: PrinterOptions = {}, handlers: Pri
         writePunctuation(">");
     }
 
-    function emitJsxText(node: JsxText) {
-        // Strip comments from JSX text content to prevent duplication
-        // Comments like /* comment */ should not be treated as text content in JSX
-        const textWithoutComments = node.text.replace(/\/\*[\s\S]*?\*\//g, '');
-        writer.writeLiteral(textWithoutComments);
+    function emitJsxText(node: JsxText) {
+        writer.writeLiteral(node.text);
     }
 
     function emitJsxClosingElementOrFragment(node: JsxClosingElement | JsxClosingFragment) {

src/compiler/scanner.ts

@@ -3720,6 +3720,9 @@ export function createScanner(
         // These initial values are special because the first line is:
         // firstNonWhitespace = 0 to indicate that we want leading whitespace,
 
+        // Build the text content excluding comments
+        let textContent = "";
+
         while (pos < end) {
             char = charCodeUnchecked(pos);
             if (char === CharacterCodes.openBrace) {
@@ -3739,6 +3742,21 @@ export function createScanner(
                 error(Diagnostics.Unexpected_token_Did_you_mean_or_rbrace, pos, 1);
             }
 
+            // Handle comments - skip /* ... */ style comments
+            if (char === CharacterCodes.slash && charCodeUnchecked(pos + 1) === CharacterCodes.asterisk) {
+                pos += 2; // Skip /*
+                // Find the end of the comment
+                while (pos < end) {
+                    if (charCodeUnchecked(pos) === CharacterCodes.asterisk && charCodeUnchecked(pos + 1) === CharacterCodes.slash) {
+                        pos += 2; // Skip */
+                        break;
+                    }
+                    pos++;
+                }
+                // Continue without adding comment content to textContent
+                continue;
+            }
+
             // FirstNonWhitespace is 0, then we only see whitespaces so far. If we see a linebreak, we want to ignore that whitespaces.
             // i.e (- : whitespace)
             //      <div>----
@@ -3754,13 +3772,16 @@ export function createScanner(
                 break;
             }
             else if (!isWhiteSpaceLike(char)) {
-                firstNonWhitespace = pos;
+                if (firstNonWhitespace === 0) {
+                    firstNonWhitespace = textContent.length;
+                }
             }
 
+            textContent += String.fromCharCode(char);
             pos++;
         }
 
-        tokenValue = text.substring(fullStartPos, pos);
+        tokenValue = textContent;
 
         return firstNonWhitespace === -1 ? SyntaxKind.JsxTextAllWhiteSpaces : SyntaxKind.JsxText;
     }

src/compiler/transformers/jsx.ts

@@ -554,37 +554,22 @@ export function transformJsx(context: TransformationContext): (x: SourceFile | B
         return fixed === undefined ? undefined : factory.createStringLiteral(fixed);
     }
 
-    /**
-     * Remove comments from JSX text content.
-     * Comments like slash-star comment star-slash should not be treated as text content in JSX.
-     */
-    function stripCommentsFromJsxText(text: string): string {
-        // Only strip comments when not in preserve mode
-        if (compilerOptions.jsx === JsxEmit.Preserve) {
-            return text;
-        }
-        // Remove /* ... */ style comments from JSX text
-        return text.replace(/\/\*[\s\S]*?\*\//g, '');
-    }
-
-    /**
-     * JSX trims whitespace at the end and beginning of lines, except that the
-     * start/end of a tag is considered a start/end of a line only if that line is
-     * on the same line as the closing tag. See examples in
-     * tests/cases/conformance/jsx/tsxReactEmitWhitespace.tsx
-     * See also https://www.w3.org/TR/html4/struct/text.html#h-9.1 and https://www.w3.org/TR/CSS2/text.html#white-space-model
-     *
-     * An equivalent algorithm would be:
-     * - If there is only one line, return it.
-     * - If there is only whitespace (but multiple lines), return `undefined`.
-     * - Split the text into lines.
-     * - 'trimRight' the first line, 'trimLeft' the last line, 'trim' middle lines.
-     * - Decode entities on each line (individually).
-     * - Remove empty lines and join the rest with " ".
-     */
-    function fixupWhitespaceAndDecodeEntities(text: string): string | undefined {
-        // First, strip comments from the text
-        text = stripCommentsFromJsxText(text);
+    /**
+     * JSX trims whitespace at the end and beginning of lines, except that the
+     * start/end of a tag is considered a start/end of a line only if that line is
+     * on the same line as the closing tag. See examples in
+     * tests/cases/conformance/jsx/tsxReactEmitWhitespace.tsx
+     * See also https://www.w3.org/TR/html4/struct/text.html#h-9.1 and https://www.w3.org/TR/CSS2/text.html#white-space-model
+     *
+     * An equivalent algorithm would be:
+     * - If there is only one line, return it.
+     * - If there is only whitespace (but multiple lines), return `undefined`.
+     * - Split the text into lines.
+     * - 'trimRight' the first line, 'trimLeft' the last line, 'trim' middle lines.
+     * - Decode entities on each line (individually).
+     * - Remove empty lines and join the rest with " ".
+     */
+    function fixupWhitespaceAndDecodeEntities(text: string): string | undefined {
         let acc: string | undefined;
         // First non-whitespace character on this line.
         let firstNonWhitespace = 0;