Use HHVM+EZC internal tidy

EZC doesn't currently support direct access to object properties via the
obj->std.properties hashtable, but tidy uses this extensively. But it
turns out that for production use cases, tidy_repair_string() should be
sufficient. $wgDebugTidy and $wgValidateAllHtml are not used, and
no deployed extension calls MWTidy::checkErrors().

The only difference I know of is that errors from tidy (status==2) lead
to the tidy output being used, rather than discarded. But
TY_(ReportFatal) has very few callers in tidylib -- probably none that
are reachable from stripped parser output.

So, throw an exception if MWTidy::checkErrors() is requested on an HHVM
instance with the tidy extension. For MWTidy::tidy(), use
tidy_repair_string(). Refactor some relevant code.

Bug: T758
Change-Id: I8d5b1c2c9f9ddce46d8ad099a671a2e297d256e0

diff --git a/includes/parser/MWTidy.php b/includes/parser/MWTidy.php
index b310862..ff62e9b 100644 (file)
--- a/includes/parser/MWTidy.php
+++ b/includes/parser/MWTidy.php
@@ -127,17 +127,11 @@ class MWTidy {
         * @return string Corrected HTML output
         */
        public static function tidy( $text ) {
-               global $wgTidyInternal;
-
                $wrapper = new MWTidyWrapper;
                $wrappedtext = $wrapper->getWrapped( $text );
 
                $retVal = null;
-               if ( $wgTidyInternal ) {
-                       $correctedtext = self::execInternalTidy( $wrappedtext, false, $retVal );
-               } else {
-                       $correctedtext = self::execExternalTidy( $wrappedtext, false, $retVal );
-               }
+               $correctedtext = self::clean( $wrappedtext, false, $retVal );
 
                if ( $retVal < 0 ) {
                        wfDebug( "Possible tidy configuration error!\n" );
@@ -160,16 +154,33 @@ class MWTidy {
         * @return bool Whether the HTML is valid
         */
        public static function checkErrors( $text, &$errorStr = null ) {
+               $retval = 0;
+               $errorStr = self::clean( $text, true, $retval );
+               return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
+       }
+
+       /**
+        * Perform a clean/repair operation
+        * @param string $text HTML to check
+        * @param bool $stderr Whether to read result from STDERR rather than STDOUT
+        * @param int &$retval Exit code (-1 on internal error)
+        * @return string|null
+        */
+       private static function clean( $text, $stderr = false, &$retval = null ) {
                global $wgTidyInternal;
 
-               $retval = 0;
                if ( $wgTidyInternal ) {
-                       $errorStr = self::execInternalTidy( $text, true, $retval );
+                       if ( wfIsHHVM() ) {
+                               if ( $stderr ) {
+                                       throw new MWException( __METHOD__.": error text return from HHVM tidy is not supported" );
+                               }
+                               return self::hhvmClean( $text, $retval );
+                       } else {
+                               return self::phpClean( $text, $stderr, $retval );
+                       }
                } else {
-                       $errorStr = self::execExternalTidy( $text, true, $retval );
+                       return self::externalClean( $text, $stderr, $retval );
                }
-
-               return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
        }
 
        /**
@@ -181,7 +192,7 @@ class MWTidy {
         * @param int &$retval Exit code (-1 on internal error)
         * @return string|null
         */
-       private static function execExternalTidy( $text, $stderr = false, &$retval = null ) {
+       private static function externalClean( $text, $stderr = false, &$retval = null ) {
                global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
                wfProfileIn( __METHOD__ );
 
@@ -248,7 +259,7 @@ class MWTidy {
         * @param int &$retval Exit code (-1 on internal error)
         * @return string|null
         */
-       private static function execInternalTidy( $text, $stderr = false, &$retval = null ) {
+       private static function phpClean( $text, $stderr = false, &$retval = null ) {
                global $wgTidyConf, $wgDebugTidy;
                wfProfileIn( __METHOD__ );
 
@@ -288,4 +299,29 @@ class MWTidy {
                wfProfileOut( __METHOD__ );
                return $cleansource;
        }
+
+       /**
+        * Use the tidy extension for HHVM from
+        * https://github.com/wikimedia/mediawiki-php-tidy
+        *
+        * This currently does not support the object-oriented interface, but
+        * tidy_repair_string() can be used for the most common tasks.
+        *
+        * @param string $text HTML to check
+        * @param int &$retval Exit code (-1 on internal error)
+        * @return string|null
+        */
+       private static function hhvmClean( $text, &$retval ) {
+               global $wgTidyConf;
+               wfProfileIn( __METHOD__ );
+               $cleansource = tidy_repair_string( $text, $wgTidyConf, 'utf8' );
+               if ( $cleansource === false ) {
+                       $cleansource = null;
+                       $retval = -1;
+               } else {
+                       $retval = 0;
+               }
+               wfProfileOut( __METHOD__ );
+               return $cleansource;
+       }
 }