TextTruncator.php

Go to the documentation of this file.

<?php
 
namespace TextExtracts;
 
use MediaWiki\Tidy\TidyDriverBase;
 
class TextTruncator {
 
    private $tidyDriver;
 
    public function __construct( TidyDriverBase $tidy = null ) {
        $this->tidyDriver = $tidy;
    }
 
    public function getFirstSentences( $text, $requestedSentenceCount ) {
        if ( $requestedSentenceCount <= 0 ) {
            return '';
        }
 
        // Based on code from OpenSearchXml by Brion Vibber
        $endchars = [
            // regular ASCII
            '\P{Lu}\.(?=[ \n]|$)',
            '[!?](?=[ \n]|$)',
            // full-width ideographic full-stop
            '。',
            // double-width roman forms
            '．',
            '！',
            '？',
            // half-width ideographic full stop
            '｡',
        ];
 
        $regexp = '/(?:' . implode( '|', $endchars ) . ')+/u';
        $res = preg_match_all( $regexp, $text, $matches, PREG_OFFSET_CAPTURE );
 
        if ( !$res ) {
            // Just return the first line
            $lines = explode( "\n", $text, 2 );
            return trim( $lines[0] );
        }
 
        $index = min( $requestedSentenceCount, $res ) - 1;
        list( $tail, $length ) = $matches[0][$index];
        // PCRE returns raw offsets, so using substr() instead of mb_substr()
        $text = substr( $text, 0, $length ) . $tail;
 
        return $this->tidy( $text );
    }
 
    public function getFirstChars( $text, $requestedLength ) {
        if ( $requestedLength <= 0 ) {
            return '';
        }
 
        $length = mb_strlen( $text );
        if ( $length <= $requestedLength ) {
            return $text;
        }
 
        // This ungreedy pattern always matches, just might return an empty string
        $pattern = '/^[\w\/]*>?/su';
        preg_match( $pattern, mb_substr( $text, $requestedLength ), $m );
        $text = mb_substr( $text, 0, $requestedLength ) . $m[0];
 
        return $this->tidy( $text );
    }
 
    private function tidy( $text ) {
        if ( $this->tidyDriver ) {
            // Fix possibly unclosed HTML tags.
            $text = $this->tidyDriver->tidy( $text );
        }
 
        return trim( $text );
    }
 
}