19885920020

Committed 03 Dec 2025 07:29AM UTC coverage: 94.505% (+0.8%) from 93.75%

Build # 19885920020

Build Type

push

gha

Run Details

344 of 364 relevant lines covered (94.51%)

130.11 hits per line

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

93.83

/src/Parser/DOMParser.php

<?php

declare(strict_types=1);

namespace Isocontent\Parser;

use Isocontent\AST\Builder;
use Isocontent\Exception\UnsupportedFormatException;

/**
 * A simple HTML parser using DOMDocument / LibXML.
 */
final class DOMParser implements Parser
{
    /**
     * @psalm-suppress MixedAssignment
     */
    #[\Override]
    public function parse(Builder $builder, mixed $input): void
    {
        if (!is_string($input)) {
            throw new UnsupportedFormatException();
        }

        $document = new \DOMDocument('1.0', 'UTF-8');
        if (!$input) {
            return;
        }

        $oldUseInternalErrors = libxml_use_internal_errors();
        libxml_use_internal_errors(true);

        /** @var non-empty-string $html */
        $html = '<?xml encoding="UTF-8">'.$input;
        $document->loadHTML($html);

        libxml_clear_errors();
        libxml_use_internal_errors($oldUseInternalErrors);

        foreach ($document->getElementsByTagName('body') as $root) {
            assert($root instanceof \DOMElement);

            foreach ($root->childNodes as $childNode) {
                assert($childNode instanceof \DOMNode);
                $this->parseNode($builder, $childNode);
            }
        }
    }

    #[\Override]
    public function supportsFormat(string $format): bool
    {
        return 'html' === $format;
    }

    private function parseNode(Builder $builder, \DOMNode $node): void
    {
        switch ($node->nodeType) {
            case XML_TEXT_NODE:
                assert($node instanceof \DOMText);
                $builder->addTextNode(preg_replace('#\s{2,}#', ' ', $node->textContent) ?? '');

                return;

            case XML_ELEMENT_NODE:
                assert($node instanceof \DOMElement);
                $blockType = $this->parseBlockType($node);
                $childBuilder = $builder->addBlockNode($blockType[0], $blockType[1] ?? []);

                break;

            default:
                return;
        }

        if (0 === $node->childNodes->length) {
            return;
        }

        foreach ($node->childNodes as $subNode) {
            assert($subNode instanceof \DOMNode);
            $this->parseNode($childBuilder, $subNode);
        }
    }

    /**
     * @return array{0: string, 1?: array<string, scalar>}
     */
    private function parseBlockType(\DOMElement $node): array
    {
        switch ($node->nodeName) {
            case 'h1':
                return ['title', ['level' => 1]];

            case 'h2':
                return ['title', ['level' => 2]];

            case 'h3':
                return ['title', ['level' => 3]];

            case 'h4':
                return ['title', ['level' => 4]];

            case 'h5':
                return ['title', ['level' => 5]];

            case 'h6':
                return ['title', ['level' => 6]];

            case 'p':
                return ['paragraph'];

            case 'em':
                return ['emphasis'];

            case 'strong':
                return ['strong'];

            case 'span':
                return ['inline_text'];

            case 'ul':
                return ['list', ['ordered' => false]];

            case 'ol':
                return ['list', ['ordered' => true]];

            case 'li':
                return ['list_item'];

            case 'blockquote':
                return ['quote'];

            case 'br':
                return ['new_line'];

            case 'a':
                $nodeAttributes = $node->attributes;
                assert($nodeAttributes instanceof \DOMNamedNodeMap);
                $attributes = array_filter(['href' => $nodeAttributes->getNamedItem('href')?->nodeValue]);

                return ['link', $attributes];
            case 'del':
                return ['stripped'];

            case 'hr':
                return ['separator'];

            case 'sub':
                return ['subscript'];

            case 'sup':
                return ['superscript'];

            case 'code':
                return ['code'];

            default:
                return ['generic'];
        }
    }
}

1	<?php
2
3	declare(strict_types=1);
4
5	namespace Isocontent\Parser;
6
7	use Isocontent\AST\Builder;
8	use Isocontent\Exception\UnsupportedFormatException;
9
10	/**
11	* A simple HTML parser using DOMDocument / LibXML.
12	*/
13	final class DOMParser implements Parser
14	{	200✔
15	/**
16	* @psalm-suppress MixedAssignment
17	*/	200✔
18	#[\Override]	200✔
19	public function parse(Builder $builder, mixed $input): void	8✔
20	{
21	if (!is_string($input)) {
22	throw new UnsupportedFormatException();	192✔
23	}	192✔
24
25	$document = new \DOMDocument('1.0', 'UTF-8');	192✔
26	if (!$input) {
27	return;	192✔
28	}	192✔
29
30	$oldUseInternalErrors = libxml_use_internal_errors();	192✔
31	libxml_use_internal_errors(true);	192✔
32		192✔
33	/** @var non-empty-string $html */
34	$html = '<?xml encoding="UTF-8">'.$input;
35	$document->loadHTML($html);
36
37	libxml_clear_errors();	120✔
38	libxml_use_internal_errors($oldUseInternalErrors);
39
40	foreach ($document->getElementsByTagName('body') as $root) {	120✔
41	assert($root instanceof \DOMElement);
42
43	foreach ($root->childNodes as $childNode) {	192✔
44	assert($childNode instanceof \DOMNode);
45	$this->parseNode($builder, $childNode);	192✔
46	}	192✔
47	}	192✔
48	}
49		192✔
50	#[\Override]
51	public function supportsFormat(string $format): bool	192✔
52	{	192✔
53	return 'html' === $format;	192✔
54	}
55		192✔
56	private function parseNode(Builder $builder, \DOMNode $node): void
57	{
58	switch ($node->nodeType) {	8✔
59	case XML_TEXT_NODE:
60	assert($node instanceof \DOMText);
61	$builder->addTextNode(preg_replace('#\s{2,}#', ' ', $node->textContent) ?? '');	192✔
62		24✔
63	return;
64
65	case XML_ELEMENT_NODE:	192✔
66	assert($node instanceof \DOMElement);	192✔
67	$blockType = $this->parseBlockType($node);
68	$childBuilder = $builder->addBlockNode($blockType[0], $blockType[1] ?? []);
69
70	break;
71
72	default:
73	return;	192✔
74	}
75		192✔
76	if (0 === $node->childNodes->length) {	192✔
77	return;	16✔
78	}
79		192✔
80	foreach ($node->childNodes as $subNode) {	16✔
81	assert($subNode instanceof \DOMNode);
82	$this->parseNode($childBuilder, $subNode);	192✔
83	}	16✔
84	}
85		192✔
86	/**	24✔
87	* @return array{0: string, 1?: array<string, scalar>}
88	*/	192✔
89	private function parseBlockType(\DOMElement $node): array	16✔
90	{
91	switch ($node->nodeName) {	192✔
92	case 'h1':	16✔
93	return ['title', ['level' => 1]];
94		192✔
95	case 'h2':	96✔
96	return ['title', ['level' => 2]];
97		176✔
98	case 'h3':	32✔
99	return ['title', ['level' => 3]];
100		176✔
101	case 'h4':	72✔
102	return ['title', ['level' => 4]];
103		144✔
104	case 'h5':	96✔
105	return ['title', ['level' => 5]];
106		56✔
107	case 'h6':	24✔
108	return ['title', ['level' => 6]];
109		56✔
110	case 'p':	24✔
111	return ['paragraph'];
112		56✔
113	case 'em':	24✔
114	return ['emphasis'];
115		48✔
116	case 'strong':	16✔
117	return ['strong'];
118		48✔
119	case 'span':	16✔
120	return ['inline_text'];
121		32✔
122	case 'ul':	16✔
123	return ['list', ['ordered' => false]];	16✔
124		16✔
125	case 'ol':	16✔
126	return ['list', ['ordered' => true]];	16✔
127		8✔
128	case 'li':
129	return ['list_item'];	16✔
130		16✔
131	case 'blockquote':	16✔
132	return ['quote'];
133		16✔
134	case 'br':	×
135	return ['new_line'];
136		16✔
137	case 'a':	×
138	$nodeAttributes = $node->attributes;
139	assert($nodeAttributes instanceof \DOMNamedNodeMap);	16✔
140	$attributes = array_filter(['href' => $nodeAttributes->getNamedItem('href')?->nodeValue]);	×
141
142	return ['link', $attributes];	16✔
143	case 'del':	×
144	return ['stripped'];
145		16✔
146	case 'hr':	×
147	return ['separator'];
148
149	case 'sub':	16✔
150	return ['subscript'];
151
152	case 'sup':
153	return ['superscript'];
154
155	case 'code':
156	return ['code'];
157
158	default:
159	return ['generic'];
160	}
161	}
162	}

un-zero-un / Isocontent / 19885920020

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