exception: Avoid preg_replace for literal swap

[lhc/web/wiklou.git] / includes / parser / ParserOutput.php

diff --git a/includes/parser/ParserOutput.php b/includes/parser/ParserOutput.php
index 7f417a2..445981b 100644 (file)
--- a/includes/parser/ParserOutput.php
+++ b/includes/parser/ParserOutput.php
@@ -31,9 +31,9 @@ class ParserOutput extends CacheTime {
        const SUPPORTS_UNWRAP_TRANSFORM = 1;
 
        /**
-        * @var string $mText The output text
+        * @var string|null $mText The output text
         */
-       public $mText;
+       public $mText = null;
 
        /**
         * @var array $mLanguageLinks List of the full text of language links,
@@ -232,6 +232,15 @@ class ParserOutput extends CacheTime {
        const SLOW_AR_TTL = 3600; // adaptive TTL for "slow" pages
        const MIN_AR_TTL = 15; // min adaptive TTL (for sanity, pool counter, and edit stashing)
 
+       /**
+        * @param string|null $text HTML. Use null to indicate that this ParserOutput contains only
+        *        meta-data, and the HTML output is undetermined, as opposed to empty. Passing null
+        *        here causes hasText() to return false.
+        * @param array $languageLinks
+        * @param array $categoryLinks
+        * @param bool $unused
+        * @param string $titletext
+        */
        public function __construct( $text = '', $languageLinks = [], $categoryLinks = [],
                $unused = false, $titletext = ''
        ) {
@@ -241,6 +250,20 @@ class ParserOutput extends CacheTime {
                $this->mTitleText = $titletext;
        }
 
+       /**
+        * Returns true if text was passed to the constructor, or set using setText(). Returns false
+        * if null was passed to the $text parameter of the constructor to indicate that this
+        * ParserOutput only contains meta-data, and the HTML output is undetermined.
+        *
+        * @since 1.32
+        *
+        * @return bool Whether this ParserOutput contains rendered text. If this returns false, the
+        *         ParserOutput contains meta-data only.
+        */
+       public function hasText() {
+               return ( $this->mText !== null );
+       }
+
        /**
         * Get the cacheable text with <mw:editsection> markers still in it. The
         * return value is suitable for writing back via setText() but is not valid
@@ -250,6 +273,10 @@ class ParserOutput extends CacheTime {
         * @since 1.27
         */
        public function getRawText() {
+               if ( $this->mText === null ) {
+                       throw new LogicException( 'This ParserOutput contains no text!' );
+               }
+
                return $this->mText;
        }
 
@@ -276,6 +303,7 @@ class ParserOutput extends CacheTime {
         *    the scheme-specific-part of the href is the (percent-encoded) value
         *    of the `data-mw-deduplicate` attribute.
         * @return string HTML
+        * @return-taint escaped
         */
        public function getText( $options = [] ) {
                $options += [
@@ -285,7 +313,7 @@ class ParserOutput extends CacheTime {
                        'deduplicateStyles' => true,
                        'wrapperDivClass' => $this->getWrapperDivClass(),
                ];
-               $text = $this->mText;
+               $text = $this->getRawText();
 
                Hooks::runWithoutAbort( 'ParserOutputPostCacheTransform', [ $this, &$text, &$options ] );
 
@@ -1262,6 +1290,42 @@ class ParserOutput extends CacheTime {
                );
        }
 
+       // TODO remove this method once old parser cache objects have expired, probably mid-October 2018
+       public function __wakeup() {
+               // T203716 remove wrapper that was added by logic in an older version of this class,
+               // where the wrapper was included in mText. This might sometimes remove a wrapper that's
+               // genuine content (manually added to a system message), but that will work out OK, see below.
+               $text = $this->getRawText();
+               $start = Html::openElement( 'div', [
+                       'class' => 'mw-parser-output'
+               ] );
+               $startLen = strlen( $start );
+               $end = Html::closeElement( 'div' );
+               $endPos = strrpos( $text, $end );
+               $endLen = strlen( $end );
+               if ( substr( $text, 0, $startLen ) === $start && $endPos !== false
+                        // if the closing div is followed by real content, bail out of unwrapping
+                        && preg_match( '/^(?>\s*<!--.*?-->)*\s*$/s', substr( $text, $endPos + $endLen ) )
+               ) {
+                       $text = substr( $text, $startLen );
+                       $text = substr( $text, 0, $endPos - $startLen ) .
+                                       substr( $text, $endPos - $startLen + $endLen );
+                       $this->setText( $text );
+                       // We found a wrapper to remove, so the ParserOutput was probably created by the
+                       // code path that now contains an addWrapperDivClass( 'mw-parser-output' ) call,
+                       // but it did not contain it when this object was cached, so we need to fix the
+                       // wrapper class variable.
+                       // If this was a message with a manually added wrapper, we are technically wrong about
+                       // this but we were wrong about the unwrapping as well so it will work out just right,
+                       // except when this is a normal page view of such a message page, in which case
+                       // it will be single-wrapped instead of double-wrapped (harmless) or something wants
+                       // render the message with unwrap=true (in which case the message won't be wrapped even
+                       // though it should, but the few code paths using unwrap=true only do it for real pages).
+                       $this->clearWrapperDivClass();
+                       $this->addWrapperDivClass( 'mw-parser-output' );
+               }
+       }
+
        /**
         * Merges internal metadata such as flags, accessed options, and profiling info
         * from $source into this ParserOutput. This should be used whenever the state of $source