From 00ad5f864eab80a4b62f3c9c11ed81fcf9bab1d0 Mon Sep 17 00:00:00 2001
From: "Peter J. Holzer" <hjp@hjp.at>
Date: Mon, 20 May 2019 23:25:49 +0200
Subject: [PATCH] Fix charset handling for text/html parts

---
 mbox2web | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/mbox2web b/mbox2web
index ac973a1..6f7d273 100755
--- a/mbox2web
+++ b/mbox2web
@@ -207,7 +207,10 @@ def render_body(msg, extra=None):
         return bodytmpl.render(context)
     def render_text_html(msg, extra=None):
         htmlpart = HTMLPart(extra)
-        htmlpart.feed(msg.get_payload(decode=True).decode(msg.get_charset() or "iso-8859-1"))
+        ct_params = dict(msg.get_params())
+        charset = ct_params.get("charset", "iso-8859-1")
+        raw_text = msg.get_payload(decode=True).decode(charset, errors="replace")
+        htmlpart.feed(raw_text)
         bodytmpl = jenv.get_template("body_text_html.html")
         context = {
             "body": jinja2.Markup(htmlpart.as_string())