Always specify UTF-8 targt charset for html_entity_decode(); default is 8-bit ISO-8859-1 which causes things to break when we later pass them through things that expect to work with UTF-8. For instance, running through preg_replace() with the /u option results in NULL, leading to problems with OStatus and SubMirror generating their plaintext versions and doing length-cropping.

1 files changed, 1 insertions, 1 deletions

diff --git a/scripts/install_cli.php b/scripts/install_cli.php
index 61fbe18ef..dadbcf66f 100755
--- a/scripts/install_cli.php
+++ b/scripts/install_cli.php
@@ -208,7 +208,7 @@ END_HELP;
         $breakout = preg_replace('/<a[^>+]\bhref="(.*)"[^>]*>(.*)<\/a>/',
                                  '\2 &lt;\1&gt;',
                                  $html);
-        return html_entity_decode(strip_tags($breakout));
+        return html_entity_decode(strip_tags($breakout), ENT_QUOTES, 'UTF-8');
     }
 }