Merge pull request #384 from jaschaio/tiptap-aware-html-detection

feat: make detectCustomHtmlPatterns aware of TipTap's actual capabilities

apps/api/src/services/EmailService.ts

@@ -659,12 +659,16 @@ export class EmailService {
   /**
    * Detects if HTML contains custom patterns that indicate it was written in the HTML editor
    * rather than the visual editor. Mirrors the same logic in apps/web/src/lib/emailStyles.ts.
+   *
+   * The TipTap editor loads StarterKit + TextAlign + Color + TextStyle + Link +
+   * ResizableImage + VariableMention. TextStyle/Color/Link round-trip <span style="..."> and
+   * <a style="..."> markup. This detection therefore PERMITS span + inline styles and only
+   * REJECTS markup TipTap cannot represent (tables, divs, forms, embeds, custom attrs,
+   * <style> blocks, etc).
    */
   private static detectCustomHtmlPatterns(html: string): boolean {
     if (!html || html.trim() === '') return false;
 
-    const hasInlineStyles = /<[^>]+style\s*=\s*["'][^"']*["']/i.test(html);
-
     const classMatches = html.matchAll(/class\s*=\s*["']([^"']*)["']/gi);
     let hasCustomClasses = false;
     for (const match of classMatches) {
@@ -679,21 +683,21 @@ export class EmailService {
       }
     }
 
-    const hasCustomAttributes = /<[^>]+(?:data-|aria-|role=|id=)/i.test(html);
-    const hasComplexTables = /<table[^>]*>[\s\S]*?<table/i.test(html);
-    const hasCustomElements = /<(?:div|span|section|article|header|footer|nav|aside)[^>]*>/i.test(html);
+    // Element-attribute-scoped regex; the leading [\s"'] guard prevents `id=` inside
+    // href URLs (e.g. `?id=...`) from false-matching as an HTML id attribute.
+    const hasCustomAttributes = /<[a-z][^>]*?[\s"'](?:data-|aria-|role=|id=)/i.test(html);
+
+    // Elements TipTap cannot round-trip with the currently-loaded extension set.
+    // <span> is intentionally excluded -- TipTap's TextStyle extension handles it.
+    const hasCustomElements =
+      /<(?:div|section|article|header|footer|nav|aside|main|table|tr|td|th|tbody|thead|tfoot|colgroup|col|form|input|button|select|textarea|iframe|video|audio|svg|object|embed|details|summary|dialog)\b/i.test(
+        html,
+      );
+
     const hasMediaQueries = /@media/i.test(html);
     const hasStyleTags = /<style[^>]*>/i.test(html);
 
-    return (
-      hasInlineStyles ||
-      hasCustomClasses ||
-      hasCustomAttributes ||
-      hasComplexTables ||
-      hasCustomElements ||
-      hasMediaQueries ||
-      hasStyleTags
-    );
+    return hasCustomClasses || hasCustomAttributes || hasCustomElements || hasMediaQueries || hasStyleTags;
   }
 
   /**

apps/web/src/lib/__tests__/emailStyles.test.ts

@@ -0,0 +1,126 @@
+import {describe, expect, it} from 'vitest';
+import {detectCustomHtmlPatterns} from '../emailStyles';
+
+describe('detectCustomHtmlPatterns', () => {
+  describe('empty / whitespace input', () => {
+    it('returns false for empty string', () => {
+      expect(detectCustomHtmlPatterns('')).toBe(false);
+    });
+
+    it('returns false for whitespace-only string', () => {
+      expect(detectCustomHtmlPatterns('   \n\t  ')).toBe(false);
+    });
+  });
+
+  describe('content TipTap can round-trip (should NOT be flagged as custom)', () => {
+    it('returns false for a basic paragraph', () => {
+      expect(detectCustomHtmlPatterns('<p>Hello</p>')).toBe(false);
+    });
+
+    it('returns false for headings, lists, blockquote, bold, italic', () => {
+      expect(
+        detectCustomHtmlPatterns(
+          '<h1>Title</h1><p><strong>bold</strong> <em>italic</em></p><ul><li>one</li></ul><blockquote>quote</blockquote>',
+        ),
+      ).toBe(false);
+    });
+
+    it('returns false for <span> with inline color style (TipTap TextStyle output)', () => {
+      expect(detectCustomHtmlPatterns('<span style="color: rgb(220, 38, 38)">red</span>')).toBe(false);
+    });
+
+    it('returns false for <span> with background-color: initial (TipTap export artifact)', () => {
+      expect(detectCustomHtmlPatterns('<span style="background-color: initial">stuff</span>')).toBe(false);
+    });
+
+    it('returns false for <span> with background-color: transparent (TipTap export artifact)', () => {
+      expect(detectCustomHtmlPatterns('<span style="background-color: transparent">stuff</span>')).toBe(false);
+    });
+
+    it('returns false for <a> with inline style (TipTap Link output)', () => {
+      expect(detectCustomHtmlPatterns('<a href="https://example.com" style="color: red">link</a>')).toBe(false);
+    });
+
+    it('returns false for paragraph with inline text-align style', () => {
+      expect(detectCustomHtmlPatterns('<p style="text-align: center">centered</p>')).toBe(false);
+    });
+
+    it('returns false when an href URL contains "id=" or "contactId=" (must not match custom-attr regex)', () => {
+      expect(
+        detectCustomHtmlPatterns('<a href="https://example.com/u?contactId=abc&id=123">unsub</a>'),
+      ).toBe(false);
+    });
+
+    it('returns false for allowed class prefixes', () => {
+      expect(detectCustomHtmlPatterns('<p class="prose">x</p>')).toBe(false);
+      expect(detectCustomHtmlPatterns('<span class="variable-mention">x</span>')).toBe(false);
+      expect(detectCustomHtmlPatterns('<img class="email-image" src="x" />')).toBe(false);
+    });
+
+    it('returns false for a TipTap-style colored span wrapped in a paragraph', () => {
+      expect(
+        detectCustomHtmlPatterns('<p>Hello <span style="color: rgb(220, 38, 38);">world</span>!</p>'),
+      ).toBe(false);
+    });
+  });
+
+  describe('content TipTap can NOT round-trip (should be flagged as custom)', () => {
+    it('returns true for <div>', () => {
+      expect(detectCustomHtmlPatterns('<div>stuff</div>')).toBe(true);
+    });
+
+    it('returns true for <table> markup (no TipTap Table extension loaded)', () => {
+      expect(detectCustomHtmlPatterns('<table><tr><td>x</td></tr></table>')).toBe(true);
+    });
+
+    it('returns true for a single <table> tag', () => {
+      expect(detectCustomHtmlPatterns('<table>x</table>')).toBe(true);
+    });
+
+    it('returns true for <style> tag', () => {
+      expect(detectCustomHtmlPatterns('<style>p { color: red; }</style>')).toBe(true);
+    });
+
+    it('returns true for @media query inside a style block', () => {
+      expect(detectCustomHtmlPatterns('@media (max-width: 600px) { ... }')).toBe(true);
+    });
+
+    it('returns true for custom data-* attribute', () => {
+      expect(detectCustomHtmlPatterns('<p data-foo="bar">x</p>')).toBe(true);
+    });
+
+    it('returns true for aria-* attribute', () => {
+      expect(detectCustomHtmlPatterns('<p aria-label="x">y</p>')).toBe(true);
+    });
+
+    it('returns true for role= attribute', () => {
+      expect(detectCustomHtmlPatterns('<p role="presentation">x</p>')).toBe(true);
+    });
+
+    it('returns true for id= attribute on an element', () => {
+      expect(detectCustomHtmlPatterns('<p id="main">x</p>')).toBe(true);
+    });
+
+    it('returns true for a disallowed CSS class', () => {
+      expect(detectCustomHtmlPatterns('<p class="custom">x</p>')).toBe(true);
+    });
+
+    it('returns true for <section>, <article>, <header>, <footer>, <nav>, <aside>, <main>', () => {
+      expect(detectCustomHtmlPatterns('<section>x</section>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<article>x</article>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<header>x</header>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<footer>x</footer>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<nav>x</nav>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<aside>x</aside>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<main>x</main>')).toBe(true);
+    });
+
+    it('returns true for form/input/button/iframe/svg', () => {
+      expect(detectCustomHtmlPatterns('<form>x</form>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<input type="text" />')).toBe(true);
+      expect(detectCustomHtmlPatterns('<button>x</button>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<iframe src="x"></iframe>')).toBe(true);
+      expect(detectCustomHtmlPatterns('<svg><circle /></svg>')).toBe(true);
+    });
+  });
+});

apps/web/src/lib/emailStyles.ts

@@ -1,10 +1,15 @@
 // Detects if HTML contains custom patterns that indicate it was written in the HTML editor
 // rather than the visual editor. Custom HTML should render as-is without prose wrapper.
+//
+// The TipTap editor in EmailEditor.tsx loads: StarterKit (paragraphs, headings, lists,
+// blockquote, code, hr, bold, italic, strike, etc.), TextAlign, Color, TextStyle, Link,
+// ResizableImage, and VariableMention. Of these, TextStyle + Color + Link natively
+// round-trip <span style="color: ..."> / <a style="color: ..."> markup that TipTap itself
+// generates when you change text color or style a link. We must therefore PERMIT what
+// TipTap can represent and REJECT only what it can't.
 export const detectCustomHtmlPatterns = (html: string): boolean => {
   if (!html || html.trim() === '') return false;
 
-  const hasInlineStyles = /<[^>]+style\s*=\s*["'][^"']*["']/i.test(html);
-
   const classMatches = html.matchAll(/class\s*=\s*["']([^"']*)["']/gi);
   let hasCustomClasses = false;
   for (const match of classMatches) {
@@ -28,21 +33,26 @@ export const detectCustomHtmlPatterns = (html: string): boolean => {
     }
   }
 
-  const hasCustomAttributes = /<[^>]+(?:data-|aria-|role=|id=)/i.test(html);
-  const hasComplexTables = /<table[^>]*>[\s\S]*?<table/i.test(html);
-  const hasCustomElements = /<(?:div|span|section|article|header|footer|nav|aside)[^>]*>/i.test(html);
+  // Custom attributes that carry semantics TipTap doesn't preserve. We require an
+  // attribute-boundary (whitespace, `=`, or quote) before the prefix so that query
+  // strings like `?id=...` inside an `href="..."` value don't false-match.
+  const hasCustomAttributes = /<[a-z][^>]*?[\s"'](?:data-|aria-|role=|id=)/i.test(html);
+
+  // Elements TipTap cannot round-trip with the currently-loaded extension set.
+  // - No Table/TableRow/TableCell extensions are loaded -> all table markup is custom.
+  // - No Div/Section/etc. block-layout extensions -> reject layout containers.
+  // - Form/embed/media/interactive elements have no TipTap representation here.
+  // <span> is intentionally NOT in this list: TipTap's TextStyle extension emits and
+  // accepts <span style="..."> for things like text color.
+  const hasCustomElements =
+    /<(?:div|section|article|header|footer|nav|aside|main|table|tr|td|th|tbody|thead|tfoot|colgroup|col|form|input|button|select|textarea|iframe|video|audio|svg|object|embed|details|summary|dialog)\b/i.test(
+      html,
+    );
+
   const hasMediaQueries = /@media/i.test(html);
   const hasStyleTags = /<style[^>]*>/i.test(html);
 
-  return (
-    hasInlineStyles ||
-    hasCustomClasses ||
-    hasCustomAttributes ||
-    hasComplexTables ||
-    hasCustomElements ||
-    hasMediaQueries ||
-    hasStyleTags
-  );
+  return hasCustomClasses || hasCustomAttributes || hasCustomElements || hasMediaQueries || hasStyleTags;
 };
 
 export const wrapEmailWithStyles = (htmlBody: string): string => {