fix(rewriting-stream): Don't escape text in special tags (inikulin#434)

jmbpwtw · Feb 16, 2023 · f11ea57 · f11ea57
1 parent 0630256
commit f11ea57
Show file tree

Hide file tree

Showing 4 changed files with 29 additions and 4 deletions.
diff --git a/packages/parse5-html-rewriting-stream/lib/index.ts b/packages/parse5-html-rewriting-stream/lib/index.ts
@@ -1,6 +1,6 @@
 import type { Location } from 'parse5/dist/common/token.js';
 import { SAXParser, EndTag, StartTag, Doctype, Text, Comment, SaxToken } from 'parse5-sax-parser';
-import { escapeString } from 'parse5/dist/serializer/index.js';
+import { hasUnescapedText, escapeString } from 'parse5/dist/serializer/index.js';
 
 /**
  * Streaming [SAX](https://en.wikipedia.org/wiki/Simple_API_for_XML)-style HTML rewriter.
@@ -125,7 +125,11 @@ export class RewritingStream extends SAXParser {
 
     /** Emits serialized text token into the output stream. */
     public emitText({ text }: Text): void {
-        this.push(escapeString(text, false));
+        this.push(
+            !this.parserFeedbackSimulator.inForeignContent && hasUnescapedText(this.tokenizer.lastStartTagName, true)
+                ? text
+                : escapeString(text, false)
+        );
     }
 
     /** Emits serialized comment token into the output stream. */

diff --git a/packages/parse5-html-rewriting-stream/test/rewriting-stream.test.ts b/packages/parse5-html-rewriting-stream/test/rewriting-stream.test.ts
@@ -329,4 +329,21 @@ describe('RewritingStream', () => {
             expected: LONG_TEXT_WITH_COMMENT,
         })
     );
+
+    it(
+        'Should emit text in script without escaping (GH-339)',
+        createRewriterTest({
+            src: '<script></script>',
+            expected: '<script>foo && bar</script>',
+            assignTokenHandlers: (rewriter) => {
+                // On a script tag, emit the text without escaping
+                rewriter.on('startTag', (token) => {
+                    rewriter.emitStartTag(token);
+                    if (token.tagName === 'script') {
+                        rewriter.emitText({ text: 'foo && bar' });
+                    }
+                });
+            },
+        })
+    );
 });
diff --git a/packages/parse5-sax-parser/lib/parser-feedback-simulator.ts b/packages/parse5-sax-parser/lib/parser-feedback-simulator.ts
@@ -8,7 +8,7 @@ import { TAG_ID as $, TAG_NAMES as TN, NAMESPACES as NS, getTagID } from 'parse5
 //Simulates adjustment of the Tokenizer which performed by standard parser during tree construction.
 export class ParserFeedbackSimulator implements TokenHandler {
     private namespaceStack: NS[] = [];
-    private inForeignContent = false;
+    public inForeignContent = false;
     public skipNextNewLine = false;
     public tokenizer: Tokenizer;
 

diff --git a/packages/parse5/lib/serializer/index.ts b/packages/parse5/lib/serializer/index.ts
@@ -32,6 +32,10 @@ const VOID_ELEMENTS = new Set<string>([
 ]);
 const UNESCAPED_TEXT = new Set<string>([$.STYLE, $.SCRIPT, $.XMP, $.IFRAME, $.NOEMBED, $.NOFRAMES, $.PLAINTEXT]);
 
+export function hasUnescapedText(tn: string, scriptingEnabled: boolean): boolean {
+    return UNESCAPED_TEXT.has(tn) || (scriptingEnabled && tn === $.NOSCRIPT);
+}
+
 export interface SerializerOptions<T extends TreeAdapterTypeMap> {
     /**
      * Specifies input tree format.
@@ -197,7 +201,7 @@ function serializeTextNode<T extends TreeAdapterTypeMap>(node: T['textNode'], op
 
     return parentTn &&
         treeAdapter.getNamespaceURI(parent) === NS.HTML &&
-        (UNESCAPED_TEXT.has(parentTn) || (options.scriptingEnabled && parentTn === $.NOSCRIPT))
+        hasUnescapedText(parentTn, options.scriptingEnabled)
         ? content
         : escapeString(content, false);
 }