25553882355

Committed 08 May 2026 11:47AM UTC coverage: 11.696% (+1.2%) from 10.457%

Build # 25553882355

Build Type

push

github

Committed by

web-flow

Commit Message

Merge pull request #651 from Yoast/task/optimize-bodyparser

[TASK] Optimized BodyParser to support nested TYPO3SEARCH markers and allow h6 headings in the content

Coverage Stats

33 of 35 new or added lines in 1 file covered. (94.29%)

320 of 2736 relevant lines covered (11.7%)

0.57 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

94.74

/Classes/PageParser/BodyParser.php

<?php

/**
 * This file is part of the "yoast_seo" extension for TYPO3 CMS.
 *
 * For the full copyright and license information, please read the
 * LICENSE.txt file that was distributed with this source code.
 */

declare(strict_types=1);

namespace YoastSeoForTypo3\YoastSeo\PageParser;

use MaxServ\FrontendRequest\Dto\RequestContext;

class BodyParser extends \MaxServ\FrontendRequest\PageParser\Parser\BodyParser
{
    protected const INDEXABLE_PATTERN = '/<!--\s*TYPO3SEARCH_(begin|end)\s*-->/i';
    protected const ALLOWED_TAGS = '<h1><h2><h3><h4><h5><h6><p><a><img>';
    protected const STRIP_WITH_CONTENT_TAGS = '<script><noscript>';

    public function parse(string $html, RequestContext $context): string
    {
        $body = parent::parse($html, $context);

        $sections = $this->extractIndexableSections($body);
        if ($sections !== []) {
            $body = implode(' ', array_map(trim(...), $sections));
        }

        return $this->prepareBody($body);
    }

    protected function prepareBody(string $body): string
    {
        $body = $this->stripTagsContent($body, self::STRIP_WITH_CONTENT_TAGS);
        $body = (string)preg_replace('/\s+/', ' ', $body);
        $body = strip_tags($body, self::ALLOWED_TAGS);

        return trim($body);
    }

    protected function stripTagsContent(string $text, string $tags = ''): string
    {
        if ($tags === '') {
            return $text;
        }

        preg_match_all('/<([a-z][a-z0-9]*)\b/i', $tags, $found);
        $names = array_unique($found[1]);
        if ($names === []) {
            return $text;
        }

        $pattern = '@<(' . implode('|', array_map('preg_quote', $names)) . ')\b[^>]*>.*?</\1\s*>@is';
        return (string)preg_replace($pattern, '', $text);
    }

    /**
     * @return list<string>
     */
    protected function extractIndexableSections(string $body): array
    {
        if (!preg_match_all(self::INDEXABLE_PATTERN, $body, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) {
            return [];
        }

        $sections = [];
        $depth = 0;
        $sectionStart = null;

        foreach ($matches as $match) {
            $markerText = $match[0][0];
            $markerOffset = (int)$match[0][1];
            $markerType = strtolower($match[1][0]);

            if ($markerType === 'begin') {
                if ($depth === 0) {
                    $sectionStart = $markerOffset + strlen($markerText);
                }
                $depth++;
                continue;
            }

            if ($depth === 0) {
                continue;
            }
            $depth--;

            if ($depth === 0 && $sectionStart !== null) {
                $sections[] = substr($body, $sectionStart, $markerOffset - $sectionStart);
                $sectionStart = null;
            }
        }

        return $sections;
    }
}

1	<?php
2
3	/**
4	* This file is part of the "yoast_seo" extension for TYPO3 CMS.
5	*
6	* For the full copyright and license information, please read the
7	* LICENSE.txt file that was distributed with this source code.
8	*/
9
10	declare(strict_types=1);
11
12	namespace YoastSeoForTypo3\YoastSeo\PageParser;
13
14	use MaxServ\FrontendRequest\Dto\RequestContext;
15
16	class BodyParser extends \MaxServ\FrontendRequest\PageParser\Parser\BodyParser
17	{
18	protected const INDEXABLE_PATTERN = '/<!--\sTYPO3SEARCH_(begin\|end)\s-->/i';
19	protected const ALLOWED_TAGS = '<h1><h2><h3><h4><h5><h6><p><a><img>';
20	protected const STRIP_WITH_CONTENT_TAGS = '<script><noscript>';
21
22	public function parse(string $html, RequestContext $context): string
23	{
24	$body = parent::parse($html, $context);	33✔
25
26	$sections = $this->extractIndexableSections($body);	33✔
27	if ($sections !== []) {	33✔
28	$body = implode(' ', array_map(trim(...), $sections));	12✔
29	}
30
31	return $this->prepareBody($body);	33✔
32	}
33
34	protected function prepareBody(string $body): string
35	{
36	$body = $this->stripTagsContent($body, self::STRIP_WITH_CONTENT_TAGS);	33✔
37	$body = (string)preg_replace('/\s+/', ' ', $body);	33✔
38	$body = strip_tags($body, self::ALLOWED_TAGS);	33✔
39
40	return trim($body);	33✔
41	}
42
43	protected function stripTagsContent(string $text, string $tags = ''): string
44	{
45	if ($tags === '') {	33✔
NEW 46	return $text;	×
47	}
48
49	preg_match_all('/<([a-z][a-z0-9]*)\b/i', $tags, $found);	33✔
50	$names = array_unique($found[1]);	33✔
51	if ($names === []) {	33✔
NEW 52	return $text;	×
53	}
54
55	$pattern = '@<(' . implode('\|', array_map('preg_quote', $names)) . ')\b[^>]>.?</\1\s*>@is';	33✔
56	return (string)preg_replace($pattern, '', $text);	33✔
57	}
58
59	/**
60	* @return list<string>
61	*/
62	protected function extractIndexableSections(string $body): array
63	{
64	if (!preg_match_all(self::INDEXABLE_PATTERN, $body, $matches, PREG_SET_ORDER \| PREG_OFFSET_CAPTURE)) {	33✔
65	return [];	20✔
66	}
67
68	$sections = [];	13✔
69	$depth = 0;	13✔
70	$sectionStart = null;	13✔
71
72	foreach ($matches as $match) {	13✔
73	$markerText = $match[0][0];	13✔
74	$markerOffset = (int)$match[0][1];	13✔
75	$markerType = strtolower($match[1][0]);	13✔
76
77	if ($markerType === 'begin') {	13✔
78	if ($depth === 0) {	13✔
79	$sectionStart = $markerOffset + strlen($markerText);	13✔
80	}
81	$depth++;	13✔
82	continue;	13✔
83	}
84
85	if ($depth === 0) {	12✔
86	continue;	1✔
87	}
88	$depth--;	12✔
89
90	if ($depth === 0 && $sectionStart !== null) {	12✔
91	$sections[] = substr($body, $sectionStart, $markerOffset - $sectionStart);	12✔
92	$sectionStart = null;	12✔
93	}
94	}
95
96	return $sections;	13✔
97	}
98	}

Yoast / Yoast-SEO-for-TYPO3 / 25553882355

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous