Ability to define connection and table creation charset and collation

qa-config-example.php

@@ -49,9 +49,17 @@
 	5. Place all the Question2Answer files on your server.
 	6. Open the appropriate URL, and follow the instructions.
 
-	More detailed installation instructions here: http://www.question2answer.org/
+	More detailed installation instructions here: https://www.question2answer.org
 */
 
+
+/*
+	The QA_MYSQL_CHARSET and QA_MYSQL_COLLATION define the charset and collation used by Q2A to
+	connect to the server, and to create tables and fields.
+*/
+
+	define('QA_MYSQL_CHARSET', 'utf8mb4');
+	define('QA_MYSQL_COLLATION', 'utf8mb4_unicode_ci');
 /*
 	======================================================================
 	 OPTIONAL CONSTANT DEFINITIONS, INCLUDING SUPPORT FOR SINGLE SIGN-ON

qa-include/db/install.php

@@ -714,7 +714,7 @@ function qa_db_create_table_sql($rawname, $definition)
 		if (isset($coldef))
 			$querycols .= (strlen($querycols) ? ', ' : '') . (is_int($colname) ? $coldef : ($colname . ' ' . $coldef));
 
-	return 'CREATE TABLE ^' . $rawname . ' (' . $querycols . ') ENGINE=InnoDB CHARSET=utf8';
+	return 'CREATE TABLE ^' . $rawname . ' (' . $querycols . ') ENGINE=InnoDB ' . qa_get_table_charset_collation();
 }
 
 

qa-include/db/selects.php

@@ -1218,7 +1218,7 @@ function qa_db_tag_recent_qs_selectspec($voteuserid, $tag, $start, $full = false
 	$selectspec = qa_db_posts_basic_selectspec($voteuserid, $full);
 
 	// use two tests here - one which can use the index, and the other which narrows it down exactly - then limit to 1 just in case
-	$selectspec['source'] .= " JOIN (SELECT postid FROM ^posttags WHERE wordid=(SELECT wordid FROM ^words WHERE word=$ AND word=$ COLLATE utf8_bin LIMIT 1) ORDER BY postcreated DESC LIMIT #,#) y ON ^posts.postid=y.postid";
+	$selectspec['source'] .= " JOIN (SELECT postid FROM ^posttags WHERE wordid=(SELECT wordid FROM ^words WHERE word=$ AND word=$ LIMIT 1) ORDER BY postcreated DESC LIMIT #,#) y ON ^posts.postid=y.postid";
 	array_push($selectspec['arguments'], $tag, qa_strtolower($tag), $start, $count);
 	$selectspec['sortdesc'] = 'created';
 
@@ -1235,7 +1235,7 @@ function qa_db_tag_word_selectspec($tag)
 {
 	return array(
 		'columns' => array('wordid', 'word', 'tagcount'),
-		'source' => '^words WHERE word=$ AND word=$ COLLATE utf8_bin',
+		'source' => '^words WHERE word=$ AND word=$',
 		'arguments' => array($tag, qa_strtolower($tag)),
 		'single' => true,
 	);

qa-include/qa-base.php

@@ -313,6 +313,13 @@ function qa_undo_wordpress_quoting($param, $isget)
 		define('QA_FINAL_MYSQL_PORT', QA_MYSQL_PORT);
 	}
 
+	if (!defined('QA_MYSQL_CHARSET')) {
+		define('QA_MYSQL_CHARSET', 'utf8'); // Use hardcoded default until v1.8.8
+	}
+	if (!defined('QA_MYSQL_COLLATION')) {
+		define('QA_MYSQL_COLLATION', null); // Use hardcoded default until v1.8.8
+	}
+
 	// Possible URL schemes for Q2A and the string used for url scheme testing
 
 	define('QA_URL_FORMAT_INDEX', 0);  // http://...../index.php/123/why-is-the-sky-blue

qa-include/qa-db.php

@@ -94,8 +94,15 @@ function qa_db_connect($failhandler = null)
 	// From Q2A 1.5, we explicitly set the character encoding of the MySQL connection, instead of using lots of "SELECT BINARY col"-style queries.
 	// Testing showed that overhead is minimal, so this seems worth trading off against the benefit of more straightforward queries, especially
 	// for plugin developers.
-	if (!$db->set_charset('utf8'))
+	if ($db->set_charset(QA_MYSQL_CHARSET)) {
+		$sql = 'SET NAMES ' . QA_MYSQL_CHARSET;
+		if (QA_MYSQL_COLLATION !== null) {
+			$sql .= ' COLLATE ' . QA_MYSQL_COLLATION;
+		}
+		$db->query($sql);
+	} else {
 		qa_db_fail_error('set_charset', $db->errno, $db->error);
+	}
 
 	qa_report_process_stage('db_connected');
 
@@ -174,6 +181,23 @@ function qa_db_disconnect()
 	}
 }
 
+/**
+ * Return the default charset and collation string to use in CREATE TABLE statements. For fields that
+ * have not been set an implicit charset or collation, these values will be assigned at creation time.
+ * Collation might be null to provide backwards compatibility with Q2A 1.8.6 and before.
+ * @param string $charset
+ * @param string $collation
+ * @return string
+ */
+function qa_get_table_charset_collation($charset = QA_MYSQL_CHARSET, $collation = QA_MYSQL_COLLATION)
+{
+	$sql = 'DEFAULT CHARACTER SET ' . $charset;
+	if ($collation !== null) {
+		$sql .= ' COLLATE ' . $collation;
+	}
+
+	return $sql;
+}
 
 /**
  * Run the raw $query, call the global failure handler if necessary, otherwise return the result resource.

qa-plugin/event-logger/qa-event-logger.php

@@ -44,7 +44,7 @@ public function init_queries($table_list)
 					'KEY ipaddress (ipaddress),' .
 					'KEY userid (userid),' .
 					'KEY event (event)' .
-					') ENGINE=MyISAM DEFAULT CHARSET=utf8';
+					') ENGINE=MyISAM ' . qa_get_table_charset_collation();
 			} else {
 				// table exists: check it has the correct schema
 				$column = qa_db_read_one_assoc(qa_db_query_sub('SHOW COLUMNS FROM ^eventlog WHERE Field="ipaddress"'));