WPDB: When sanity checking read queries, there are some collations we can skip, for improved performance.

Props pento, nacin.

See #21212.



git-svn-id: https://develop.svn.wordpress.org/trunk@32162 602fd350-edb4-49c9-b593-d223f7449a82

src/wp-includes/wp-db.php

@@ -170,6 +170,16 @@ class wpdb {
 	 */
 	protected $check_current_query = true;
 
+	/**
+	 * Flag to ensure we don't run into recursion problems when checking the collation.
+	 *
+	 * @since 4.2.0
+	 * @access private
+	 * @see wpdb::check_safe_collation()
+	 * @var boolean
+	 */
+	private $checking_collation = false;
+
 	/**
 	 * Saved info on the table column
 	 *
@@ -2038,6 +2048,10 @@ class wpdb {
 	public function get_var( $query = null, $x = 0, $y = 0 ) {
 		$this->func_call = "\$db->get_var(\"$query\", $x, $y)";
 
+		if ( $this->check_safe_collation( $query ) ) {
+			$this->check_current_query = false;
+		}
+
 		if ( $query ) {
 			$this->query( $query );
 		}
@@ -2066,6 +2080,11 @@ class wpdb {
 	 */
 	public function get_row( $query = null, $output = OBJECT, $y = 0 ) {
 		$this->func_call = "\$db->get_row(\"$query\",$output,$y)";
+
+		if ( $this->check_safe_collation( $query ) ) {
+			$this->check_current_query = false;
+		}
+
 		if ( $query ) {
 			$this->query( $query );
 		} else {
@@ -2103,6 +2122,10 @@ class wpdb {
 	 * @return array Database query result. Array indexed from 0 by SQL result row number.
 	 */
 	public function get_col( $query = null , $x = 0 ) {
+		if ( $this->check_safe_collation( $query ) ) {
+			$this->check_current_query = false;
+		}
+
 		if ( $query ) {
 			$this->query( $query );
 		}
@@ -2131,6 +2154,10 @@ class wpdb {
 	public function get_results( $query = null, $output = OBJECT ) {
 		$this->func_call = "\$db->get_results(\"$query\", $output)";
 
+		if ( $this->check_safe_collation( $query ) ) {
+			$this->check_current_query = false;
+		}
+
 		if ( $query ) {
 			$this->query( $query );
 		} else {
@@ -2358,6 +2385,47 @@ class wpdb {
 		return false;
 	}
 
+	/**
+	 * Check if the query is accessing a collation considered safe on the current version of MySQL.
+	 *
+	 * @since 4.2.0
+	 * @access protected
+	 *
+	 * @param string $query The query to check.
+	 * @return bool True if the collation is safe, false if it isn't.
+	 */
+	protected function check_safe_collation( $query ) {
+		if ( $this->checking_collation ) {
+			return true;
+		}
+
+		$table = $this->get_table_from_query( $query );
+		if ( ! $table ) {
+			return false;
+		}
+
+		$this->checking_collation = true;
+		$this->get_table_charset( $table );
+		$this->checking_collation = false;
+
+		$table = strtolower( $table );
+		if ( empty( $this->col_meta[ $table ] ) ) {
+			return false;
+		}
+
+		foreach( $this->col_meta[ $table ] as $col ) {
+			if ( empty( $col->Collation ) ) {
+				continue;
+			}
+
+			if ( ! in_array( $col->Collation, array( 'utf8_general_ci', 'utf8_bin', 'utf8mb4_general_ci', 'utf8mb4_bin' ), true ) ) {
+				return false;
+			}
+		}
+
+		return true;
+	}
+
 	/**
 	 * Strips any invalid characters based on value/charset pairs.
 	 *

tests/phpunit/tests/db/charset.php

@@ -463,4 +463,67 @@ class Tests_DB_Charset extends WP_UnitTestCase {
 
 		$this->assertFalse( $wpdb->query( "INSERT INTO {$wpdb->posts} (post_content) VALUES ('foo\xf0\xff\xff\xffbar')" ) );
 	}
+
+	/**
+	 * @ticket 21212
+	 */
+	function data_table_collation_check() {
+		$table_name = 'table_collation_check';
+		$data = array(
+			array(
+				// utf8_bin tables don't need extra sanity checking.
+				"( a VARCHAR(50) COLLATE utf8_bin )", // create
+				true                                  // expected result
+			),
+			array(
+				// Neither do utf8_general_ci tables.
+				"( a VARCHAR(50) COLLATE utf8_general_ci )",
+				true
+			),
+			array(
+				// utf8_unicode_ci tables do.
+				"( a VARCHAR(50) COLLATE utf8_unicode_ci )",
+				false
+			),
+			array(
+				// utf8_bin tables don't need extra sanity checking,
+				// except for when they're not just utf8_bin.
+				"( a VARCHAR(50) COLLATE utf8_bin, b VARCHAR(50) COLLATE big5_chinese_ci )",
+				false
+			),
+			array(
+				// utf8_bin tables don't need extra sanity checking
+				// when the other columns aren't strings.
+				"( a VARCHAR(50) COLLATE utf8_bin, b INT )",
+				true
+			),
+		);
+
+		foreach( $data as &$value ) {
+			$this_table_name = $table_name . '_' . rand_str( 5 );
+
+			$value[0] = "CREATE TABLE $this_table_name {$value[0]}";
+			$value[2] = "SELECT * FROM $this_table_name";
+			$value[3] = "DROP TABLE IF EXISTS $this_table_name";
+		}
+		unset( $value );
+
+		return $data;
+	}
+
+
+	/**
+	 * @dataProvider data_table_collation_check
+	 * @ticket 21212
+	 */
+	function test_table_collation_check( $create, $expected, $query, $drop ) {
+		self::$_wpdb->query( $drop );
+
+		self::$_wpdb->query( $create );
+
+		$return = self::$_wpdb->check_safe_collation( $query );
+		$this->assertEquals( $expected, $return );
+
+		self::$_wpdb->query( $drop );
+	}
 }