From c7d551f185ab30c9c7a4e0df2e37260c204fc8ba Mon Sep 17 00:00:00 2001
From: Linus Rath <139418639+rathlinus@users.noreply.github.com>
Date: Tue, 7 Jul 2026 23:54:50 +0200
Subject: [PATCH] fix: attachment reminder ignores quoted text on reply/forward
 #570

---
 components/email/email-composer.tsx        |  9 ++-
 lib/__tests__/email-composer-utils.test.ts | 73 ++++++++++++++++++++++
 lib/email-composer-utils.ts                | 55 ++++++++++++++++
 3 files changed, 136 insertions(+), 1 deletion(-)

diff --git a/components/email/email-composer.tsx b/components/email/email-composer.tsx
index ef5bb310..c2b231f1 100644
--- a/components/email/email-composer.tsx
+++ b/components/email/email-composer.tsx
@@ -46,6 +46,7 @@ import {
   formatRecipientList,
   splitPastedRecipients,
   waitForPendingUploads,
+  extractUserAuthoredText,
   type Recipient,
 } from "@/lib/email-composer-utils";
 import { isValidEmail } from "@/lib/validation";
@@ -1599,7 +1600,13 @@ export function EmailComposer({
     if (!skipAttachmentCheck && attachmentReminderEnabled) {
       const hasAttachments = attachmentsRef.current.some(att => att.blobId && !att.uploading && !att.error);
       if (!hasAttachments) {
-        const bodyText = htmlToPlainText(body);
+        // Scan only the user-authored text: the quoted original of a
+        // reply/forward often mentions an attachment itself, which used to fire
+        // the reminder even when the user typed no keyword and added nothing (#570).
+        const bodyText = extractUserAuthoredText(body, {
+          plainTextMode,
+          forwardedSeparator: tQuote('forwarded_separator'),
+        });
         const searchText = `${subject} ${bodyText}`.toLowerCase();
         const matched = attachmentReminderKeywords.find(kw => searchText.includes(kw.toLowerCase()));
         if (matched) {
diff --git a/lib/__tests__/email-composer-utils.test.ts b/lib/__tests__/email-composer-utils.test.ts
index f2d742a8..36956d94 100644
--- a/lib/__tests__/email-composer-utils.test.ts
+++ b/lib/__tests__/email-composer-utils.test.ts
@@ -11,8 +11,81 @@ import {
   formatRecipientList,
   splitPastedRecipients,
   waitForPendingUploads,
+  extractUserAuthoredText,
 } from "../email-composer-utils";
 
+const FORWARDED_SEPARATOR = "---------- Forwarded message ----------";
+
+describe("extractUserAuthoredText", () => {
+  const scan = (body: string, plainTextMode: boolean) =>
+    extractUserAuthoredText(body, {
+      plainTextMode,
+      forwardedSeparator: FORWARDED_SEPARATOR,
+    }).toLowerCase();
+
+  it("keeps user text and drops the quoted island on an HTML reply (#570)", () => {
+    const body =
+      "<p>Here is my reply.</p>" +
+      '<div>On Mon, Someone wrote:</div>' +
+      '<div data-quoted-html><p>Please find attached the invoice (anexo).</p></div>';
+    const result = scan(body, false);
+    expect(result).toContain("here is my reply");
+    expect(result).not.toContain("anexo");
+    expect(result).not.toContain("attached");
+  });
+
+  it("drops a <blockquote> quote when the original had no HTML part", () => {
+    const body =
+      "<p>Thanks!</p>" +
+      '<blockquote>segue em anexo o documento</blockquote>';
+    const result = scan(body, false);
+    expect(result).toContain("thanks");
+    expect(result).not.toContain("anexo");
+  });
+
+  it("drops the forwarded header and original on an HTML forward", () => {
+    const body =
+      "<p>FYI</p><br><br>" +
+      FORWARDED_SEPARATOR +
+      "<br>From: a@b.com<br>Subject: Invoice attached<br><br>" +
+      '<div data-quoted-html><p>em anexo</p></div>';
+    const result = scan(body, false);
+    expect(result).toContain("fyi");
+    expect(result).not.toContain("anexo");
+    expect(result).not.toContain("attached");
+    expect(result).not.toContain("forwarded message");
+  });
+
+  it("drops '>' quoted lines on a plain-text reply", () => {
+    const body = "My reply here.\n\nOn Mon, X wrote:\n> please find attached\n> anexo";
+    const result = scan(body, true);
+    expect(result).toContain("my reply here");
+    expect(result).not.toContain("attached");
+    expect(result).not.toContain("anexo");
+  });
+
+  it("drops the bare forwarded original on a plain-text forward", () => {
+    const body =
+      "See below.\n\n" +
+      FORWARDED_SEPARATOR +
+      "\nFrom: a@b.com\nSubject: hi\n\nem anexo o contrato";
+    const result = scan(body, true);
+    expect(result).toContain("see below");
+    expect(result).not.toContain("anexo");
+  });
+
+  it("still surfaces a keyword the user actually typed", () => {
+    const body =
+      "<p>See the attached file.</p>" +
+      '<div data-quoted-html><p>nothing here</p></div>';
+    expect(scan(body, false)).toContain("attached");
+  });
+
+  it("tolerates a missing forwarded separator", () => {
+    expect(scan("<p>plain reply</p>", false)).toContain("plain reply");
+  });
+});
+
 describe("plainTextToComposerBody", () => {
   it("returns an empty string for empty input", () => {
     expect(plainTextToComposerBody("")).toBe("");
diff --git a/lib/email-composer-utils.ts b/lib/email-composer-utils.ts
index f927ddd2..d3f0555c 100644
--- a/lib/email-composer-utils.ts
+++ b/lib/email-composer-utils.ts
@@ -1,4 +1,5 @@
 import { isValidEmail } from "@/lib/validation";
+import { htmlToPlainText } from "@/lib/html-to-text";
 
 const HTML_ESCAPE_MAP = {
   "&": "&amp;",
@@ -54,6 +55,60 @@ export function rewriteCidImagesForEditor(html: string): string {
   return touched ? doc.body.innerHTML : html;
 }
 
+/**
+ * Reduce a composer body to just the user-authored text for the attachment
+ * reminder's keyword scan, dropping the quoted original of a reply/forward.
+ *
+ * Scanning the whole body triggered false positives whenever the quoted message
+ * mentioned an attachment - common, since the original often did carry one, and
+ * the default keyword list is broad and multilingual (#570). We strip:
+ *   - HTML mode: the QuotedHtml island ([data-quoted-html]) and any <blockquote>
+ *     (the wrapper used when the original had no HTML part), then convert to text.
+ *   - Plain-text mode: lines prefixed with ">" (the reply quote).
+ *   - Both modes: everything from the "Forwarded message" separator onward, which
+ *     also removes the forwarded From/Date/Subject header lines and the bare
+ *     forwarded original (which carries no blockquote/island wrapper).
+ *
+ * `forwardedSeparator` is the localized quote_header.forwarded_separator string;
+ * pass it so the forward cut works in the active locale.
+ */
+export function extractUserAuthoredText(
+  body: string,
+  options: { plainTextMode: boolean; forwardedSeparator?: string }
+): string {
+  const { plainTextMode, forwardedSeparator } = options;
+
+  let text: string;
+  if (plainTextMode) {
+    text = body
+      .split("\n")
+      .filter((line) => !/^\s*>/.test(line))
+      .join("\n");
+  } else {
+    const doc = new DOMParser().parseFromString(`<body>${body}</body>`, "text/html");
+    doc
+      .querySelectorAll("[data-quoted-html], blockquote")
+      .forEach((el) => el.remove());
+    text = htmlToPlainText(doc.body.innerHTML, { paragraphSpacing: true });
+  }
+
+  // Cut everything from the forwarded-message separator onward. htmlToPlainText
+  // collapses the separator's internal whitespace, so match with a
+  // whitespace-flexible, regex-escaped pattern rather than an exact string.
+  const trimmedSeparator = forwardedSeparator?.trim();
+  if (trimmedSeparator) {
+    const pattern = trimmedSeparator
+      .replace(/[.*+?^${}()|[\]\\]/g, "\\$&")
+      .replace(/\s+/g, "\\s+");
+    const match = text.match(new RegExp(pattern));
+    if (match && match.index !== undefined) {
+      text = text.slice(0, match.index);
+    }
+  }
+
+  return text;
+}
+
 /** A composer recipient. Display name is optional; email is required. */
 export type Recipient = { name?: string; email: string };