From b6a1ad7d9d44a12ebedec4b8bf44b1c19a1f9d3b Mon Sep 17 00:00:00 2001
From: TooonyChen <26252739+TooonyChen@users.noreply.github.com>
Date: Thu, 9 Jul 2026 15:43:18 +1200
Subject: [PATCH] fix: keep anchor hrefs when stripping HTML and accept bare
 text/html fragments
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Forwarded mails (e.g. Yandex webmail) arrive as a single text/html part
with no <html>/<body> wrapper and no text/plain alternative. Two bugs made
their codes unextractable:

- stripHtmlTags dropped href attributes, so button-only mails (Netflix
  'Get Code') reached the LLM with no URL at all → codeExist=0.
  Anchors now render as 'label [url]' with &amp; decoded.
- extractMailBodies only recognised single-part HTML wrapped in
  <html>/<body>, so htmlBody was null and the Rendered preview was empty.
  A bare fragment with Content-Type text/html now counts as the html body.

Verified against the real forwarded Netflix mail from production raw_mails:
content now carries the travel/verify link and classifies 5/5 as
login_code.
---
 src/services/mime.ts | 33 ++++++++++++++++++++++++++-------
 test/mime.test.ts    | 29 +++++++++++++++++++++++++++++
 2 files changed, 55 insertions(+), 7 deletions(-)

diff --git a/src/services/mime.ts b/src/services/mime.ts
index 8ef3437..8499903 100644
--- a/src/services/mime.ts
+++ b/src/services/mime.ts
@@ -100,12 +100,25 @@ export function decodeMimeHeader(value: string | null): string | null {
 }
 
 export function stripHtmlTags(text: string): string {
-  return text
-    .replace(/<style[\s\S]*?<\/style>/gi, "")
-    .replace(/<script[\s\S]*?<\/script>/gi, "")
-    .replace(/<[^>]+>/g, " ")
-    .replace(/\s+/g, " ")
-    .trim();
+  return (
+    text
+      .replace(/<style[\s\S]*?<\/style>/gi, "")
+      .replace(/<script[\s\S]*?<\/script>/gi, "")
+      // 保留链接: <a href="url">text</a> → "text [url]"。
+      // 验证码邮件常常只有一个按钮链接 (如 Netflix 的 Get Code), 剥掉 href 会让
+      // LLM 无码可提取。仅 html-only 邮件走到这里, 多部分邮件优先用 text/plain。
+      .replace(
+        /<a\b[^>]*?href\s*=\s*(?:"([^"]*)"|'([^']*)')[^>]*>([\s\S]*?)<\/a>/gi,
+        (_match, doubleQuoted, singleQuoted, inner) => {
+          const url = (doubleQuoted ?? singleQuoted ?? "").replace(/&amp;/gi, "&").trim();
+          const label = inner.replace(/<[^>]+>/g, " ").replace(/\s+/g, " ").trim();
+          return /^https?:\/\//i.test(url) ? ` ${label} [${url}] ` : ` ${label} `;
+        },
+      )
+      .replace(/<[^>]+>/g, " ")
+      .replace(/\s+/g, " ")
+      .trim()
+  );
 }
 
 // 主题看起来像验证码/事务性邮件 (verification, OTP, sign-in, reset, 验证码…)。
@@ -199,7 +212,13 @@ export function extractMailBodies(rawEmail: string): {
   const decodedBody = decodeBodyPart(topBody, encoding, getCharset(topContentType));
   const topMimeType = topContentType?.split(";")[0]?.trim().toLowerCase() ?? "";
   const htmlMatch = decodedBody.match(/<html[\s\S]*<\/html>/i) ?? decodedBody.match(/<body[\s\S]*<\/body>/i);
-  const htmlBody = htmlMatch ? htmlMatch[0].trim() : null;
+  // Content-Type 明确是 text/html 时整个正文就是 HTML, 即使没有 <html>/<body>
+  // 包裹 (Yandex 等客户端转发出的就是裸片段), 否则 Rendered 预览会拿不到内容。
+  const htmlBody = htmlMatch
+    ? htmlMatch[0].trim()
+    : topMimeType.includes("text/html") && decodedBody.trim()
+      ? decodedBody.trim()
+      : null;
 
   const decodedText = topMimeType.includes("text/html") ? stripHtmlTags(decodedBody) : decodedBody.trim();
   const textBody = decodedText ? decodedText : htmlBody ? stripHtmlTags(htmlBody) : null;
diff --git a/test/mime.test.ts b/test/mime.test.ts
index 53d1a91..93214e6 100644
--- a/test/mime.test.ts
+++ b/test/mime.test.ts
@@ -4,6 +4,7 @@ import {
   decodeQuotedPrintable,
   extractMailBodies,
   isPromotionalEmail,
+  stripHtmlTags,
 } from "../src/services/mime";
 
 describe("mime decoding", () => {
@@ -92,3 +93,31 @@ describe("isPromotionalEmail", () => {
     expect(isPromotionalEmail(headers, "Subject: Hello\r\n\r\nbody")).toBe(false);
   });
 });
+
+describe("stripHtmlTags", () => {
+  it("preserves anchor hrefs so button-only mails keep their link", () => {
+    const html = '<p>Hi Thomas</p><a class="h5" href="https://www.netflix.com/verify?a=1&amp;b=2" style="x">Get Code</a>';
+    expect(stripHtmlTags(html)).toContain("Get Code [https://www.netflix.com/verify?a=1&b=2]");
+  });
+
+  it("drops non-http hrefs but keeps the label", () => {
+    expect(stripHtmlTags('<a href="mailto:x@y.com">contact us</a>')).toBe("contact us");
+  });
+});
+
+describe("extractMailBodies single-part html", () => {
+  it("treats a bare text/html fragment (no <html> wrapper) as the html body", () => {
+    const rawEmail = [
+      "Subject: code",
+      "Content-Type: text/html; charset=utf-8",
+      "Content-Transfer-Encoding: base64",
+      "",
+      Buffer.from('<div>Your code: <a href="https://x.com/v?t=1">Get Code</a></div>').toString("base64"),
+      "",
+    ].join("\r\n");
+
+    const bodies = extractMailBodies(rawEmail);
+    expect(bodies.htmlBody).toContain("https://x.com/v?t=1");
+    expect(bodies.textBody).toContain("Get Code [https://x.com/v?t=1]");
+  });
+});