Changeset 33478

Timestamp:

07/29/2015 06:35:00 AM (10 years ago)

Author:

pento

Message:

WPDB: Remove some of the complexities in ::strip_invalid_text() associated with switching character sets between queries. Instead of trying to dynamically change connection character sets, we now rely on the value of ::charset. This also fixes the case where queries were being blocked when DB_CHARSET was utf8, but the column character set was non-utf8.

Merge of [33308] to the 4.1 branch.

See #32165.

Location:

branches/4.1

Files:

• src/wp-includes/wp-db.php (modified) (2 diffs)
• tests/phpunit/tests/db/charset.php (modified) (4 diffs)

branches/4.1/src/wp-includes/wp-db.php

@@ -2652 +2652 @@
             foreach ( $data as $col => $value ) {
                 if ( ! empty( $value['db'] ) ) {
-                    if ( ! isset( $queries[ $value['charset'] ] ) ) {
-                        $queries[ $value['charset'] ] = array();
-                    }
-
                     // We're going to need to truncate by characters or bytes, depending on the length value we have.
                     if ( 'byte' === $value['length']['type'] ) {
-                        // Split the CONVERT() calls by charset, so we can make sure the connection is right
-                        $queries[ $value['charset'] ][ $col ] = $this->prepare( "CONVERT( LEFT( CONVERT( %s USING binary ), %.0f ) USING {$value['charset']} )", $value['value'], $value['length']['length'] );
+                        // Using binary causes LEFT() to truncate by bytes.
+                        $charset = 'binary';
                     } else {
-                        $queries[ $value['charset'] ][ $col ] = $this->prepare( "LEFT( CONVERT( %s USING {$value['charset']} ), %.0f )", $value['value'], $value['length']['length'] );
+                        $charset = $value['charset'];
                     }
+
+                    $queries[ $col ] = $this->prepare( "CONVERT( LEFT( CONVERT( %s USING $charset ), %.0f ) USING {$this->charset} )", $value['value'], $value['length']['length'] );
 
                     unset( $data[ $col ]['db'] );
@@ -2668 +2666 @@
             }
 
-            $connection_charset = $this->charset;
-            foreach ( $queries as $charset => $query ) {
+            $sql = array();
+            foreach ( $queries as $column => $query ) {
                 if ( ! $query ) {
                     continue;
                 }
 
-                // Change the charset to match the string(s) we're converting
-                if ( $charset !== $connection_charset ) {
-                    $connection_charset = $charset;
-                    $this->set_charset( $this->dbh, $charset );
-                }
-
-                $this->check_current_query = false;
-
-                $sql = array();
-                foreach ( $query as $column => $column_query ) {
-                    $sql[] = $column_query . " AS x_$column";
-                }
-
-                $row = $this->get_row( "SELECT " . implode( ', ', $sql ), ARRAY_A );
-                if ( ! $row ) {
-                    $this->set_charset( $this->dbh, $connection_charset );
-                    return new WP_Error( 'wpdb_strip_invalid_text_failure' );
-                }
-
-                foreach ( array_keys( $query ) as $column ) {
-                    $data[ $column ]['value'] = $row["x_$column"];
-                }
-            }
-
-            // Don't forget to change the charset back!
-            if ( $connection_charset !== $this->charset ) {
-                $this->set_charset( $this->dbh );
+                $sql[] = $query . " AS x_$column";
+            }
+
+            $this->check_current_query = false;
+            $row = $this->get_row( "SELECT " . implode( ', ', $sql ), ARRAY_A );
+            if ( ! $row ) {
+                return new WP_Error( 'wpdb_strip_invalid_text_failure' );
+            }
+
+            foreach ( array_keys( $data ) as $column ) {
+                $data[ $column ]['value'] = $row["x_$column"];
             }
         }

branches/4.1/tests/phpunit/tests/db/charset.php

@@ -244 +244 @@
                 'expected' => str_repeat( "\xcc\xe3", 5 ),
                 'length'   => array( 'type' => 'byte', 'length' => 10 ),
+            ),
+            'ujis_with_utf8_connection' => array(
+                'charset'            => 'ujis',
+                'connection_charset' => 'utf8',
+                'value'              => '自動下書き',
+                'expected'           => '自動下書き',
+                'length'             => array( 'type' => 'byte', 'length' => 100 ),
+            ),
+            'ujis_with_utf8_connection_char_length' => array(
+                'charset'            => 'ujis',
+                'connection_charset' => 'utf8',
+                'value'              => '自動下書き',
+                'expected'           => '自動下書',
+                'length'             => array( 'type' => 'char', 'length' => 4 ),
+            ),
+            'ujis_with_utf8_connection_byte_length' => array(
+                'charset'            => 'ujis',
+                'connection_charset' => 'utf8',
+                'value'              => '自動下書き',
+                'expected'           => '自動',
+                'length'             => array( 'type' => 'byte', 'length' => 6 ),
             ),
             'false' => array(
@@ -290 +311 @@
             $expected = $field;
             $expected['value'] = $expected['expected'];
-            unset( $expected['expected'], $field['expected'] );
+            unset( $expected['expected'], $field['expected'], $expected['connection_charset'] );
 
             // We're keeping track of these for our multiple-field test.
@@ -304 +325 @@
         }
 
-        // Time for our test of multiple fields at once.
-        $data_provider[] = array( $multiple, $multiple_expected, 'multiple fields/charsets' );
-
         return $data_provider;
     }
@@ -315 +333 @@
      */
     function test_strip_invalid_text( $data, $expected, $message ) {
+        $charset = self::$_wpdb->charset;
+        if ( isset( $data[0]['connection_charset'] ) ) {
+            $new_charset = $data[0]['connection_charset'];
+            unset( $data[0]['connection_charset'] );
+        } else {
+            $new_charset = $data[0]['charset'];
+        }
+
+        self::$_wpdb->charset = $new_charset;
+        self::$_wpdb->set_charset( self::$_wpdb->dbh, $new_charset );
+
         $actual = self::$_wpdb->strip_invalid_text( $data );
+
+        self::$_wpdb->charset = $charset;
+        self::$_wpdb->set_charset( self::$_wpdb->dbh, $charset );
+
         $this->assertSame( $expected, $actual, $message );
     }