ExtractFormatter.php

Go to the documentation of this file.

<?php
 
namespace TextExtracts;
 
use DOMElement;
use HtmlFormatter\HtmlFormatter;
 
class ExtractFormatter extends HtmlFormatter {
    const SECTION_MARKER_START = "\1\2";
    const SECTION_MARKER_END = "\2\1";
 
    private $plainText;
 
    public function __construct( $text, $plainText ) {
        parent::__construct( HtmlFormatter::wrapHTML( $text ) );
        $this->plainText = $plainText;
 
        $this->setRemoveMedia( true );
 
        if ( $plainText ) {
            $this->flattenAllTags();
        } else {
            $this->flatten( [ 'a' ] );
        }
    }
 
    public function getText( $element = null ) {
        $this->filterContent();
        $text = parent::getText();
        if ( $this->plainText ) {
            $text = html_entity_decode( $text );
            // replace nbsp with space
            $text = str_replace( "\u{00A0}", ' ', $text );
            // for Windows
            $text = str_replace( "\r", "\n", $text );
            // normalise newlines
            $text = preg_replace( "/\n{3,}/", "\n\n", $text );
        }
        return trim( $text );
    }
 
    public function onHtmlReady( $html ) {
        if ( $this->plainText ) {
            $html = preg_replace( '/\s*(<h([1-6])\b)/i',
                "\n\n" . self::SECTION_MARKER_START . '$2' . self::SECTION_MARKER_END . '$1',
                $html
            );
        }
        return $html;
    }
 
    public function filterContent() {
        $removed = parent::filterContent();
 
        $doc = $this->getDoc();
        $spans = $doc->getElementsByTagName( 'span' );
 
        foreach ( $spans as $span ) {
            $span->removeAttribute( 'class' );
            $span->removeAttribute( 'style' );
        }
 
        return $removed;
    }
}