Replace regex with DOMParser in GPT document.write rewriting

Author: ChristianPavilonis

crates/js/lib/src/integrations/gpt/script_guard.ts

@@ -190,43 +190,56 @@ function rewriteLinkHref(element: HTMLLinkElement): void {
 // Layer 1: document.write / document.writeln interception
 // ---------------------------------------------------------------------------
 
-/**
- * Regex that matches `src="..."` or `src='...'` attributes inside a
- * `<script>` tag where the URL text mentions the GPT domain token. We capture:
- *   1. Everything before the URL (the `src=` prefix with quote)
- *   2. The URL itself
- *   3. Everything after the URL (the closing quote)
- *
- * This handles the HTML that GPT's `Xd` function produces, e.g.:
- *   `<script src="https://securepubads.g.doubleclick.net/pagead/…/pubads_impl.js" …></script>`
- *
- * Hostname verification still happens in [`maybeRewrite`], so URLs that merely
- * contain the token in query text are left unchanged.
- */
-const SCRIPT_SRC_RE =
-  /(<script\b[^>]*?\bsrc\s*=\s*["'])([^"']*securepubads\.g\.doubleclick\.net[^"']*)(["'])/gi;
-
 /**
  * Rewrite GPT domain URLs inside raw HTML strings passed to
  * `document.write` / `document.writeln`.
+ *
+ * Uses `DOMParser` for robust HTML parsing instead of regex so that
+ * edge-cases (unquoted attributes, unusual spacing, mixed quote styles)
+ * are handled by the browser's native parser. The raw `getAttribute`
+ * value is swapped in the original HTML string so the surrounding markup
+ * is preserved verbatim.
+ *
+ * If the GPT domain is present in the HTML but `DOMParser` is
+ * unavailable or throws, the function **fails closed** (returns an
+ * empty string) rather than passing the unproxied URL through.
  */
 function rewriteHtmlString(html: string): string {
-  SCRIPT_SRC_RE.lastIndex = 0;
-  if (!SCRIPT_SRC_RE.test(html)) return html;
-  SCRIPT_SRC_RE.lastIndex = 0;
-
-  return html.replace(SCRIPT_SRC_RE, (_match, prefix: string, url: string, suffix: string) => {
-    const { url: rewrittenUrl, didRewrite } = maybeRewrite(url);
-    if (!didRewrite) {
-      return `${prefix}${url}${suffix}`;
+  // Fast path: nothing to do when the GPT domain isn't mentioned at all.
+  if (!html.includes(GPT_DOMAIN)) return html;
+
+  if (typeof DOMParser === 'undefined') {
+    log.warn(
+      `${LOG_PREFIX}: DOMParser unavailable, blocking document.write HTML that references GPT domain`
+    );
+    return '';
+  }
+
+  try {
+    const doc = new DOMParser().parseFromString(html, 'text/html');
+    const scripts = doc.querySelectorAll('script[src]');
+    let result = html;
+
+    for (const script of scripts) {
+      const rawSrc = script.getAttribute('src') ?? '';
+      const { url: rewrittenUrl, didRewrite } = maybeRewrite(rawSrc);
+      if (!didRewrite) continue;
+
+      log.info(`${LOG_PREFIX}: rewriting document.write script src`, {
+        original: rawSrc,
+        rewritten: rewrittenUrl,
+      });
+      result = result.replaceAll(rawSrc, rewrittenUrl);
     }
 
-    log.info(`${LOG_PREFIX}: rewriting document.write script src`, {
-      original: url,
-      rewritten: rewrittenUrl,
-    });
-    return `${prefix}${rewrittenUrl}${suffix}`;
-  });
+    return result;
+  } catch (err) {
+    log.warn(
+      `${LOG_PREFIX}: failed to parse document.write HTML containing GPT domain, blocking`,
+      err
+    );
+    return '';
+  }
 }
 
 function installDocumentWritePatch(): void {

crates/js/lib/test/integrations/gpt/script_guard.test.ts

@@ -155,4 +155,154 @@ describe('GPT script guard', () => {
       '/integrations/gpt/pagead/managed/js/gpt/current/pubads_impl.js?foo=bar'
     );
   });
+
+  // -----------------------------------------------------------------------
+  // document.write edge-cases (DOMParser-based rewriting)
+  // -----------------------------------------------------------------------
+
+  it('rewrites document.write script src with single-quoted attribute', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    installGptGuard();
+
+    document.write(
+      "<script src='https://securepubads.g.doubleclick.net/pagead/managed/js/gpt/current/pubads_impl.js'></script>"
+    );
+
+    expect(nativeWriteSpy).toHaveBeenCalledTimes(1);
+    const [writtenHtml] = nativeWriteSpy.mock.calls[0] ?? [];
+    expect(writtenHtml).toContain(window.location.host);
+    expect(writtenHtml).toContain('/integrations/gpt/pagead/managed/js/gpt/current/pubads_impl.js');
+    expect(writtenHtml).not.toContain('securepubads.g.doubleclick.net');
+  });
+
+  it('rewrites document.write script src with extra whitespace around =', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    installGptGuard();
+
+    document.write(
+      '<script  src  =  "https://securepubads.g.doubleclick.net/pagead/managed/js/gpt/current/pubads_impl.js" ></script>'
+    );
+
+    expect(nativeWriteSpy).toHaveBeenCalledTimes(1);
+    const [writtenHtml] = nativeWriteSpy.mock.calls[0] ?? [];
+    expect(writtenHtml).toContain(window.location.host);
+    expect(writtenHtml).toContain('/integrations/gpt/pagead/managed/js/gpt/current/pubads_impl.js');
+    expect(writtenHtml).not.toContain('securepubads.g.doubleclick.net');
+  });
+
+  it('rewrites multiple script tags in a single document.write call', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    installGptGuard();
+
+    document.write(
+      '<script src="https://securepubads.g.doubleclick.net/pagead/a.js"></script>' +
+        '<script src="https://securepubads.g.doubleclick.net/pagead/b.js"></script>'
+    );
+
+    expect(nativeWriteSpy).toHaveBeenCalledTimes(1);
+    const [writtenHtml] = nativeWriteSpy.mock.calls[0] ?? [];
+    expect(writtenHtml).toContain('/integrations/gpt/pagead/a.js');
+    expect(writtenHtml).toContain('/integrations/gpt/pagead/b.js');
+    expect(writtenHtml).not.toContain('securepubads.g.doubleclick.net');
+  });
+
+  it('rewrites document.writeln the same as document.write', () => {
+    const nativeWritelnSpy = vi.fn<(...args: string[]) => void>();
+    document.writeln = nativeWritelnSpy as unknown as typeof document.writeln;
+
+    installGptGuard();
+
+    document.writeln(
+      '<script src="https://securepubads.g.doubleclick.net/pagead/managed/js/gpt/current/pubads_impl.js"></script>'
+    );
+
+    expect(nativeWritelnSpy).toHaveBeenCalledTimes(1);
+    const [writtenHtml] = nativeWritelnSpy.mock.calls[0] ?? [];
+    expect(writtenHtml).toContain(window.location.host);
+    expect(writtenHtml).toContain('/integrations/gpt/pagead/managed/js/gpt/current/pubads_impl.js');
+    expect(writtenHtml).not.toContain('securepubads.g.doubleclick.net');
+  });
+
+  it('passes through HTML with no GPT domain reference unchanged', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    installGptGuard();
+
+    const html = '<script src="https://example.com/tracker.js"></script>';
+    document.write(html);
+
+    expect(nativeWriteSpy).toHaveBeenCalledWith(html);
+  });
+
+  it('rewrites protocol-relative GPT URLs in document.write', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    installGptGuard();
+
+    document.write(
+      '<script src="//securepubads.g.doubleclick.net/pagead/managed/js/gpt/current/pubads_impl.js"></script>'
+    );
+
+    expect(nativeWriteSpy).toHaveBeenCalledTimes(1);
+    const [writtenHtml] = nativeWriteSpy.mock.calls[0] ?? [];
+    expect(writtenHtml).toContain(window.location.host);
+    expect(writtenHtml).toContain('/integrations/gpt/pagead/managed/js/gpt/current/pubads_impl.js');
+    expect(writtenHtml).not.toContain('securepubads.g.doubleclick.net');
+  });
+
+  // -----------------------------------------------------------------------
+  // Fail-closed behaviour
+  // -----------------------------------------------------------------------
+
+  it('fails closed when DOMParser is unavailable', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    const originalDOMParser = globalThis.DOMParser;
+    // @ts-expect-error — simulating an environment without DOMParser
+    delete globalThis.DOMParser;
+
+    try {
+      installGptGuard();
+
+      document.write('<script src="https://securepubads.g.doubleclick.net/pagead/a.js"></script>');
+
+      expect(nativeWriteSpy).toHaveBeenCalledTimes(1);
+      expect(nativeWriteSpy).toHaveBeenCalledWith('');
+    } finally {
+      globalThis.DOMParser = originalDOMParser;
+    }
+  });
+
+  it('fails closed when DOMParser throws', () => {
+    const nativeWriteSpy = vi.fn<(...args: string[]) => void>();
+    document.write = nativeWriteSpy as unknown as typeof document.write;
+
+    const originalDOMParser = globalThis.DOMParser;
+    // @ts-expect-error — injecting a broken DOMParser
+    globalThis.DOMParser = class {
+      parseFromString() {
+        throw new Error('boom');
+      }
+    };
+
+    try {
+      installGptGuard();
+
+      document.write('<script src="https://securepubads.g.doubleclick.net/pagead/a.js"></script>');
+
+      expect(nativeWriteSpy).toHaveBeenCalledTimes(1);
+      expect(nativeWriteSpy).toHaveBeenCalledWith('');
+    } finally {
+      globalThis.DOMParser = originalDOMParser;
+    }
+  });
 });