Fix extra leading newlines in plain text converted from HTML (#8973)

2 years ago · e69cd1b5a6
3 changed files with 26 additions and 3 deletions
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@ -32,6 +32,7 @@
 - Fix bug where it wasn't possible to scroll lists by clicking middle mouse button (#8942)
 - Fix bug where label text in a single-input dialog could be partially invisible in some locales (#8905)
 - Fix bug where LDAP (fulltext) search didn't work without 'search_fields' in config (#8874)
+- Fix extra leading newlines in plain text converted from HTML (#8973)

 ## Release 1.6.1

--- a/program/lib/Roundcube/rcube_html2text.php
+++ b/program/lib/Roundcube/rcube_html2text.php
@ -143,9 +143,10 @@ class rcube_html2text
     */
    protected $search = [
        '/\r/',                                  // Non-legal carriage return
-        '/<head[^>]*>.*?<\/head>/is',            // <head>
-        '/<script[^>]*>.*?<\/script>/is',        // <script>
-        '/<style[^>]*>.*?<\/style>/is',          // <style>
+        '/\n*<\/?html>\n*/is',                   // <html>
+        '/\n*<head[^>]*>.*?<\/head>\n*/is',      // <head>
+        '/\n*<script[^>]*>.*?<\/script>\n*/is',  // <script>
+        '/\n*<style[^>]*>.*?<\/style>\n*/is',    // <style>
        '/[\n\t]+/',                             // Newlines and tabs
        '/<p[^>]*>/i',                           // <p>
        '/<\/p>[\s\n\t]*<div[^>]*>/i',           // </p> before <div>
@ -172,6 +173,7 @@ class rcube_html2text
     */
    protected $replace = [
        '',                                     // Non-legal carriage return
+        '',                                     // <html>|</html>
        '',                                     // <head>
        '',                                     // <script>
        '',                                     // <style>
@ -502,6 +504,7 @@ class rcube_html2text
        if (($pos = stripos($text, '<body')) !== false) {
            $pos = strpos($text, '>', $pos);
            $text = substr($text, $pos + 1);
+            $text = ltrim($text);
        }

        // Run our defined tags search-and-replace
--- a/tests/Framework/Html2text.php
+++ b/tests/Framework/Html2text.php
@ -71,6 +71,25 @@ class rc_html2text extends PHPUnit\Framework\TestCase
                'in'    => '<b><strong>&#347;</strong></b>',
                'out'   => 'ś',
            ],
+            12 => [
+                'title' => 'Full HTML handling (html tag only)',
+                'in'    => "<html>\n<p>test</p></html>",
+                'out'   => 'test',
+            ],
+            13 => [
+                'title' => 'Full HTML handling (html+head tags)',
+                'in'    => '<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /></head>'
+                    . "\n<p>test</p></html>\n",
+                'out'   => 'test',
+            ],
+            14 => [
+                'title' => 'Full HTML handling (html+head+body tags)',
+                'in'    => '<html><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /></head>' . "\n"
+                    . '<body style="font-size: 10pt; font-family: Verdana,Geneva,sans-serif">' . "\n"
+                    . '<p>test</p>'
+                    . '</body></html>',
+                'out'   => 'test',
+            ],
        ];
    }