From c7580d5532828e80dc36bc9575f285a334f19838 Mon Sep 17 00:00:00 2001
From: pdontthink <pdontthink@7612ce4b-ef26-0410-bec9-ea0150e637f0>
Date: Mon, 23 May 2022 00:25:07 +0000
Subject: [PATCH] Greatly fix the plaintext display of messages that do not
 have a text part. Patch from Alexey Shpakovsky (#496)

git-svn-id: https://svn.code.sf.net/p/squirrelmail/code/trunk/squirrelmail@14961 7612ce4b-ef26-0410-bec9-ea0150e637f0
---
 functions/mime.php | 36 ++++++++++++++++++++++++++++--------
 1 file changed, 28 insertions(+), 8 deletions(-)

diff --git a/functions/mime.php b/functions/mime.php
index 7e67f00d..1f02367b 100644
--- a/functions/mime.php
+++ b/functions/mime.php
@@ -396,18 +396,38 @@ function formatBody($imap_stream, $message, $color, $wrap_at, $ent_num, $id, $ma
          */
 
         if ($body_message->header->type1 == 'html') {
-            if ($show_html_default <> 1) {
+            // Do we need to make an HTML part viewable as non-HTML plain text?
+            if ($show_html_default != 1) {
                 $entity_conv = array('&nbsp;' => ' ',
-                                     '<p>'    => "\n",
-                                     '<P>'    => "\n",
-                                     '<br>'   => "\n",
-                                     '<BR>'   => "\n",
-                                     '<br />' => "\n",
-                                     '<BR />' => "\n",
+                                     // These are better done by regex (below)
+                                     // '<p>'    => "\n",
+                                     // '<P>'    => "\n",
+                                     // '<br>'   => "\n",
+                                     // '<BR>'   => "\n",
+                                     // '<br />' => "\n",
+                                     // '<BR />' => "\n",
+                                     // '<tr>'   => "\n",
+                                     // '<div>'  => "\n",
                                      '&gt;'   => '>',
-                                     '&lt;'   => '<');
+                                     '&lt;'   => '<',
+                                     '&amp;'   => '&',
+                                     '&copy;'   => '©');
+                // first, completely remove <style> tags as they aren't useful in this context
+                $body = preg_replace('/<style.*>.*<\/style.*>/isU', '', $body);
+                // emulate how newlines are treated as spaces in HTML
+                $body = preg_replace('/(\r|\n)+/', ' ', $body);
+                // now replace the tags listed just above
                 $body = strtr($body, $entity_conv);
+                // <p>, <br>, <tr> and <div> are best replaced by a newline
+                $body = preg_replace('/<(p|br|tr|div).*>/isU', "\n", $body);
+                // remove the rest of the HTML tags
                 $body = strip_tags($body);
+                // condense multiple spaces into one
+                $body = preg_replace('/[ \t]+/', ' ', $body);
+                // trim each line
+                $body = preg_replace('/ *\n */', "\n", $body);
+                // allow maximum two newlines
+                $body = preg_replace('/\n\n\n+/', "\n\n", $body);
                 $body = trim($body);
                 translateText($body, $wrap_at,
                         $body_message->header->getParameter('charset'));
-- 
2.25.1