Upgrade task to move utf8 mysql databases to utf8mb4 (#394)

include/admin/upgrader.inc.php

@@ -275,6 +275,11 @@ $tasks = array(array('version'   => '0.5.1',
                      'function'  => 'serendipity_upgrader_spamblock_moveForce',
                      'title'     => 'Swap value forcemoderation',
                      'desc'      => 'If force moderation of comments is activated, activate the new option moderation_auto'),
+                     
+               array('version'   => '2.4-alpha2',
+                     'function'  => 'serendipity_upgradeUTF8_UTF8mb4',
+                     'title'     => 'Enable full Unicode support for MySQL',
+                     'desc'      => 'Move MySQL databases from MyIsam and UTF8 to InnoDB and utf8mb4, enabling full unicode input.'),
 );
 
 /* Fetch SQL files which needs to be run */

include/functions_upgrader.inc.php

@@ -458,3 +458,50 @@ function serendipity_upgrader_move_syndication_config() {
         }
     }
 }
+
+# When mysql is used in a version new enough (at least 5.6.4/10.0.5), upgrade all UTF8 tables to utf8mb4. For that to work well
+# we also need to set the storage engine to InnoDB. Earlier version set it to MyISAM.
+function serendipity_upgradeUTF8_UTF8mb4() {
+    global $serendipity;
+
+    # are we even using mysql?
+    if ($serendipity['dbType'] != 'mysqli') {
+        return false;
+    }
+
+    if (serendipity_utf8mb4_ready()) {
+        # get all core tables
+        //$tables = serendipity_db_query("SHOW TABLES LIKE '" . str_replace('_', '\_', serendipity_db_escape_string($prefix)) . "%'");
+        $tables = serendipity_db_query("SHOW TABLES LIKE 'serendipity_entries'");
+        if (!is_array($tables)) {
+            echo 'Could not analyze existing tables via SHOW TABLES, please check permissions.' . $tables;
+            return false;
+        }
+        # now collect all their columns charsets
+        # they all need to be utf8 for us to be able to securely set them to utf8mb4
+        foreach ($tables as $table) {
+            $table = $table[0];
+            $columns = serendipity_db_query('SHOW FULL COLUMNS FROM ' . $table);
+            foreach($columns as $column) {
+                if ($column['Collation'] && ( ! ($column['Collation'] == 'utf8_unicode_ci' || $column['Collation'] == 'utf8_general_ci')) ) {
+                    return false;
+                }
+            }
+        }
+
+        # but before we can set them to utf8mb4, the storage engine needs to be set to innodb and the format to dynamic
+        # Without InnoDB the index size is just 1000 bytes, which would not work with our db schema
+        foreach ($tables AS $table) {
+            $table = $table[0];
+            serendipity_db_query('ALTER TABLE `' . $table . '` ENGINE=InnoDB');
+            serendipity_db_query('ALTER TABLE `' . $table . '` ROW_FORMAT=DYNAMIC');
+        }
+
+        # Good, now we can set it it utf8mb4. utf8mb4 is fully compatible with utf8, so there won't be broken chars
+        foreach ($tables AS $table) {
+            $table = $table[0];
+            serendipity_db_query('ALTER TABLE `' . $table . '` CONVERT TO CHARACTER SET utf8mb4 COLLATE utf8mb4_general_ci');
+        }
+        return true;
+    }
+}

serendipity_config.inc.php

@@ -47,7 +47,7 @@ if (defined('USE_MEMSNAP')) {
 }
 
 // The version string
-$serendipity['version'] = '2.4-alpha1';
+$serendipity['version'] = '2.4-alpha2';
 
 
 // Setting this to 'false' will enable debugging output. All alpha/beta/cvs snapshot versions will emit debug information by default. To increase the debug level (to enable Smarty debugging), set this flag to 'debug'.