chore: Review fix

2026-04-29 23:01:41 +05:30
parent c8abbadafc
commit 5245cc4245
2 changed files with 81 additions and 7 deletions
@@ -91,6 +91,17 @@ const cutBlockAtMarker = (block, marker) => {
  if (!block.childNodes.length) block.remove();
 };

+// Walk up to the nearest enclosing `<blockquote>` (including `block` itself).
+// Returns null when there is none below `root`.
+const findEnclosingBlockquote = (block, root) => {
+  let cur = block;
+  while (cur && cur !== root) {
+    if (cur.tagName === 'BLOCKQUOTE') return cur;
+    cur = cur.parentElement;
+  }
+  return null;
+};
+
 // Walk up while `block` is the first substantive child of its parent.
 // Promotes the cut to the wrapper, so a divider `<div>` plus the body
 // siblings AFTER it strip together.
@@ -151,13 +162,28 @@ const apply = root => {
  // 3. Trailing <blockquote> as the last top-level child.
  if (root.lastElementChild?.matches?.('blockquote'))
    root.lastElementChild.remove();
-  // 4. Soft headers — same hard-cut, but walk up to the wrapper first.
-  findBlocks(root, isSoftHeader).forEach(b =>
-    cutBlockAtMarker(
-      expandToWrapper(b, root),
-      t => HEADER_LINE.test(t) || ATTRIBUTION.test(t)
-    )
-  );
+  // 4. Soft headers. Three sub-cases:
+  //   (a) match sits inside a <blockquote> — remove the whole blockquote
+  //       (it wraps the entire quote: attribution + body). Apple Mail.
+  //   (b) match has >= 2 header lines (real Outlook attribution shape) —
+  //       hard-cut at the wrapper level so the body siblings go too.
+  //   (c) just an "On … wrote:" attribution → develop-style remove the
+  //       block, so a user reply sitting at root level isn't eaten.
+  findBlocks(root, isSoftHeader).forEach(block => {
+    const enclosingBq = findEnclosingBlockquote(block, root);
+    if (enclosingBq) {
+      enclosingBq.remove();
+      return;
+    }
+    if (countHeaderLines(blockText(block)) >= 2) {
+      cutBlockAtMarker(
+        expandToWrapper(block, root),
+        t => HEADER_LINE.test(t) || ATTRIBUTION.test(t)
+      );
+      return;
+    }
+    block.remove();
+  });
  // 5. Top-level RFC `>` / header tail.
  const start = findTopLevelTailStart(root);
  if (start !== -1) [...root.childNodes].slice(start).forEach(n => n.remove());
@@ -310,6 +310,54 @@ describe('EmailQuoteExtractor', () => {
      expect(cleaned).not.toContain('From: Sam');
    });

+    it('strips Apple-Mail blockquote (attribution + body inside one <blockquote type="cite">)', () => {
+      const html =
+        '<div>Sounds good, see you Friday.</div>' +
+        '<div><br><blockquote type="cite">' +
+        '<div>On Apr 6, 2026, at 11:26 AM, Sam &lt;sam@example.test&gt; wrote:</div>' +
+        '<br><div><div>Hi Pat,</div><div>Locking the Friday slot.</div><div>Sam</div></div>' +
+        '</blockquote></div>';
+      const c = document.createElement('div');
+      c.innerHTML = EmailQuoteExtractor.extractQuotes(html);
+      expect(c.textContent).toContain('Sounds good');
+      expect(c.textContent).not.toContain('On Apr 6, 2026');
+      expect(c.textContent).not.toContain('Hi Pat');
+      expect(c.textContent).not.toContain('Locking the Friday slot');
+    });
+
+    it('strips flat Outlook attribution + body (multi-line From/Sent/To/Subject)', () => {
+      const html =
+        '<p>Confirming I received this — will review tomorrow.</p>' +
+        '<p>Thanks,<br>Pat</p>' +
+        '<p>From: Sam &lt;sam@example.test&gt;<br>Sent: Wednesday, December 4, 2024 5:15 PM<br>To: Pat &lt;pat@example.test&gt;<br>Subject: Quotation</p>' +
+        '<p>Hi Pat,<br>Quotation attached. Let me know if you need changes.<br>Sam</p>';
+      const c = document.createElement('div');
+      c.innerHTML = EmailQuoteExtractor.extractQuotes(html);
+      expect(c.textContent).toContain('Confirming I received this');
+      expect(c.textContent).toContain('Thanks,');
+      expect(c.textContent).not.toContain('From: Sam');
+      expect(c.textContent).not.toContain('Quotation attached');
+    });
+
+    // Inline reply where a soft-header `<blockquote>` is followed by the
+    // user's actual reply at the SAME level. The hard-cut for soft headers
+    // would otherwise eat the reply.
+    it('preserves user reply that follows a soft-header <blockquote>', () => {
+      const html =
+        '<blockquote>On Mon, Sep 22, Sam wrote:<br>Original quoted line.</blockquote><p>My actual reply.</p>';
+      const c = document.createElement('div');
+      c.innerHTML = EmailQuoteExtractor.extractQuotes(html);
+      expect(c.textContent).toContain('My actual reply');
+    });
+
+    it('preserves user reply that follows a wrapper div containing the soft-header block', () => {
+      const html =
+        '<div><blockquote>On Mon, Sam wrote:</blockquote></div><p>My reply outside the wrapper.</p>';
+      const c = document.createElement('div');
+      c.innerHTML = EmailQuoteExtractor.extractQuotes(html);
+      expect(c.textContent).toContain('My reply outside the wrapper');
+    });
+
    it('preserves reply when the From-header sits inside a deep wrapper (Outlook WordSection1)', () => {
      const html = `
        <div class="WordSection1">