1.23.5/php/HtmlFormatter_8php_source.html

<?php

class HtmlFormatter {

    private $doc;


    private $html;

    private $itemsToRemove = array();

    private $elementsToFlatten = array();

    protected $removeMedia = false;


    public function __construct( $html ) {

        $this->html = $html;

    }


    public static function wrapHTML( $html ) {

        return '<!doctype html><html><head></head><body>' . $html . '</body></html>';

    }


    protected function onHtmlReady( $html ) {

        return $html;

    }


    public function getDoc() {

        if ( !$this->doc ) {

            $html = mb_convert_encoding( $this->html, 'HTML-ENTITIES', 'UTF-8' );


            // Workaround for bug that caused spaces before references

            // to disappear during processing:

            // https://bugzilla.wikimedia.org/show_bug.cgi?id=53086

            //

            // Please replace with a better fix if one can be found.

            $html = str_replace( ' <', '&#32;<', $html );


            libxml_use_internal_errors( true );

            $loader = libxml_disable_entity_loader();

            $this->doc = new DOMDocument();

            $this->doc->strictErrorChecking = false;

            $this->doc->loadHTML( $html );

            libxml_disable_entity_loader( $loader );

            libxml_use_internal_errors( false );

            $this->doc->encoding = 'UTF-8';

        }

        return $this->doc;

    }


    public function setRemoveMedia( $flag = true ) {

        $this->removeMedia = $flag;

    }


    public function remove( $selectors ) {

        $this->itemsToRemove = array_merge( $this->itemsToRemove, (array)$selectors );

    }


    public function flatten( $elements ) {

        $this->elementsToFlatten = array_merge( $this->elementsToFlatten, (array)$elements );

    }


    public function flattenAllTags() {

        $this->flatten( '[?!]?[a-z0-9]+' );

    }


    public function filterContent() {

        wfProfileIn( __METHOD__ );

        $removals = $this->parseItemsToRemove();


        if ( !$removals ) {

            wfProfileOut( __METHOD__ );

            return;

        }


        $doc = $this->getDoc();


        // Remove tags


        // You can't remove DOMNodes from a DOMNodeList as you're iterating

        // over them in a foreach loop. It will seemingly leave the internal

        // iterator on the foreach out of wack and results will be quite

        // strange. Though, making a queue of items to remove seems to work.

        $domElemsToRemove = array();

        foreach ( $removals['TAG'] as $tagToRemove ) {

            $tagToRemoveNodes = $doc->getElementsByTagName( $tagToRemove );

            foreach ( $tagToRemoveNodes as $tagToRemoveNode ) {

                if ( $tagToRemoveNode ) {

                    $domElemsToRemove[] = $tagToRemoveNode;

                }

            }

        }


        $this->removeElements( $domElemsToRemove );


        // Elements with named IDs

        $domElemsToRemove = array();

        foreach ( $removals['ID'] as $itemToRemove ) {

            $itemToRemoveNode = $doc->getElementById( $itemToRemove );

            if ( $itemToRemoveNode ) {

                $domElemsToRemove[] = $itemToRemoveNode;

            }

        }

        $this->removeElements( $domElemsToRemove );


        // CSS Classes

        $domElemsToRemove = array();

        $xpath = new DOMXpath( $doc );

        foreach ( $removals['CLASS'] as $classToRemove ) {

            $elements = $xpath->query( '//*[contains(@class, "' . $classToRemove . '")]' );


            foreach ( $elements as $element ) {

                $classes = $element->getAttribute( 'class' );

                if ( preg_match( "/\b$classToRemove\b/", $classes ) && $element->parentNode ) {

                    $domElemsToRemove[] = $element;

                }

            }

        }

        $this->removeElements( $domElemsToRemove );


        // Tags with CSS Classes

        foreach ( $removals['TAG_CLASS'] as $classToRemove ) {

            $parts = explode( '.', $classToRemove );


            $elements = $xpath->query(

                '//' . $parts[0] . '[@class="' . $parts[1] . '"]'

            );


            $this->removeElements( $elements );

        }


        wfProfileOut( __METHOD__ );

    }


    private function removeElements( $elements ) {

        $list = $elements;

        if ( $elements instanceof DOMNodeList ) {

            $list = array();

            foreach ( $elements as $element ) {

                $list[] = $element;

            }

        }

        foreach ( $list as $element ) {

            if ( $element->parentNode ) {

                $element->parentNode->removeChild( $element );

            }

        }

    }


    private function fixLibXML( $html ) {

        wfProfileIn( __METHOD__ );

        static $replacements;

        if ( ! $replacements ) {

            // We don't include rules like '&#34;' => '&amp;quot;' because entities had already been

            // normalized by libxml. Using this function with input not sanitized by libxml is UNSAFE!

            $replacements = new ReplacementArray( array(

                '&quot;' => '&amp;quot;',

                '&amp;' => '&amp;amp;',

                '&lt;' => '&amp;lt;',

                '&gt;' => '&amp;gt;',

            ) );

        }

        $html = $replacements->replace( $html );

        $html = mb_convert_encoding( $html, 'UTF-8', 'HTML-ENTITIES' );

        wfProfileOut( __METHOD__ );

        return $html;

    }


    public function getText( $element = null ) {

        wfProfileIn( __METHOD__ );


        if ( $this->doc ) {

            if ( $element !== null && !( $element instanceof DOMElement ) ) {

                $element = $this->doc->getElementById( $element );

            }

            if ( $element ) {

                $body = $this->doc->getElementsByTagName( 'body' )->item( 0 );

                $nodesArray = array();

                foreach ( $body->childNodes as $node ) {

                    $nodesArray[] = $node;

                }

                foreach ( $nodesArray as $nodeArray ) {

                    $body->removeChild( $nodeArray );

                }

                $body->appendChild( $element );

            }

            $html = $this->doc->saveHTML();

            $html = $this->fixLibXml( $html );

        } else {

            $html = $this->html;

        }

        if ( wfIsWindows() ) {

            // Appears to be cleanup for CRLF misprocessing of unknown origin

            // when running server on Windows platform.

            //

            // If this error continues in the future, please track it down in the

            // XML code paths if possible and fix there.

            $html = str_replace( '&#13;', '', $html );

        }

        $html = preg_replace( '/<!--.*?-->|^.*?<body>|<\/body>.*$/s', '', $html );

        $html = $this->onHtmlReady( $html );


        if ( $this->elementsToFlatten ) {

            $elements = implode( '|', $this->elementsToFlatten );

            $html = preg_replace( "#</?($elements)\\b[^>]*>#is", '', $html );

        }


        wfProfileOut( __METHOD__ );

        return $html;

    }


    protected function parseSelector( $selector, &$type, &$rawName ) {

        if ( strpos( $selector, '.' ) === 0 ) {

            $type = 'CLASS';

            $rawName = substr( $selector, 1 );

        } elseif ( strpos( $selector, '#' ) === 0 ) {

            $type = 'ID';

            $rawName = substr( $selector, 1 );

        } elseif ( strpos( $selector, '.' ) !== 0 && strpos( $selector, '.' ) !== false ) {

            $type = 'TAG_CLASS';

            $rawName = $selector;

        } elseif ( strpos( $selector, '[' ) === false && strpos( $selector, ']' ) === false ) {

            $type = 'TAG';

            $rawName = $selector;

        } else {

            throw new MWException( __METHOD__ . "(): unrecognized selector '$selector'" );

        }


        return true;

    }


    protected function parseItemsToRemove() {

        wfProfileIn( __METHOD__ );

        $removals = array(

            'ID' => array(),

            'TAG' => array(),

            'CLASS' => array(),

            'TAG_CLASS' => array(),

        );


        foreach ( $this->itemsToRemove as $itemToRemove ) {

            $type = '';

            $rawName = '';

            if ( $this->parseSelector( $itemToRemove, $type, $rawName ) ) {

                $removals[$type][] = $rawName;

            }

        }


        if ( $this->removeMedia ) {

            $removals['TAG'][] = 'img';

            $removals['TAG'][] = 'audio';

            $removals['TAG'][] = 'video';

        }


        wfProfileOut( __METHOD__ );

        return $removals;

    }

}