21501721037

Committed 30 Jan 2026 02:00AM UTC coverage: 91.159% (-1.3%) from 92.426%

Build Type

push

github

Committed by

Commit Message

wip

Run Details

2681 of 2941 relevant lines covered (91.16%)

0.91 hits per line

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

88.89

/src/Texy/Modules/HtmlModule.php

<?php

/**
 * This file is part of the Texy! (https://texy.nette.org)
 * Copyright (c) 2004 David Grudl (https://davidgrudl.com)
 */

declare(strict_types=1);

namespace Texy\Modules;

use Texy;
use Texy\Nodes\HtmlCommentNode;
use Texy\Nodes\HtmlTagNode;
use Texy\ParseContext;
use Texy\Patterns;
use Texy\Position;
use Texy\Regexp;
use Texy\Syntax;
use function str_ends_with, strlen, strtolower, strtr, substr, trim;


/**
 * Processes HTML tags and comments in input text.
 */
final class HtmlModule extends Texy\Module
{
        public function __construct(
                private Texy\Texy $texy,
        ) {
        }


        public function beforeParse(string &$text): void
        {
                $this->texy->registerLinePattern(
                        $this->parseTag(...),
                        '~
                                < (/?)                          # tag begin
                                ([a-z][a-z0-9_:-]{0,50})        # tag name
                                (
                                        (?:
                                                \s++ [a-z0-9_:-]++ |   # attribute name
                                                = \s*+ " [^"' . Patterns::MARK . ']*+ " |     # attribute value in double quotes
                                                = \s*+ \' [^\'' . Patterns::MARK . ']*+ \' |  # attribute value in single quotes
                                                = [^\s>' . Patterns::MARK . ']++              # attribute value without quotes
                                        )*
                                )
                                \s*+
                                (/?)                             # self-closing slash
                                >
                        ~is',
                        Syntax::HtmlTag,
                );

                $this->texy->registerLinePattern(
                        fn(?ParseContext $context, array $matches, array $offsets) => new HtmlCommentNode($matches[1], new Position($offsets[0], strlen($matches[0]))),
                        '~
                                <!--
                                ( [^' . Patterns::MARK . ']*? )
                                -->
                        ~is',
                        Syntax::HtmlComment,
                );
        }


        /**
         * Parses <tag attr="...">
         * @param  array<?string>  $matches
         * @param  array<?int>  $offsets
         */
        public function parseTag(?ParseContext $context, array $matches, array $offsets): ?HtmlTagNode
        {
                [, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;

                $isStart = $mEnd !== '/';
                $isEmpty = $mEmpty === '/';
                if (!$isEmpty && str_ends_with($mAttr, '/')) {
                        $mAttr = substr($mAttr, 0, -1);
                        $isEmpty = true;
                }

                // error - can't close empty element
                if ($isEmpty && !$isStart) {
                        return null;
                }

                // error - end element with attrs
                $mAttr = trim(strtr($mAttr, "\n", ' '));
                if ($mAttr && !$isStart) {
                        return null;
                }

                return new HtmlTagNode(
                        $mTag,
                        $isStart ? $this->parseAttributes($mAttr) : [],
                        closing: !$isStart,
                        selfClosing: $isEmpty,
                        position: new Position($offsets[0], strlen($matches[0])),
                );
        }


        /** @return array<string, string|bool> */
        private function parseAttributes(string $attrs): array
        {
                $res = [];
                $matches = Regexp::matchAll(
                        $attrs,
                        <<<'X'
                                ~
                                ([a-z0-9_:-]+)                 # attribute name
                                \s*
                                (?:
                                        = \s*                      # equals sign
                                        (
                                                ' [^']* ' |            # single quoted value
                                                " [^"]* " |            # double quoted value
                                                [^'"\s]+               # unquoted value
                                        )
                                )?
                                ~is
                                X,
                );

                foreach ($matches as $m) {
                        $key = strtolower($m[1]);
                        $value = $m[2];
                        if ($value == null) {
                                $res[$key] = true;
                        } elseif ($value[0] === '\'' || $value[0] === '"') {
                                $res[$key] = trim(Texy\Helpers::unescapeHtml(substr($value, 1, -1)));
                        } else {
                                $res[$key] = trim(Texy\Helpers::unescapeHtml($value));
                        }
                }

                return $res;
        }
}

1	<?php
2
3	/**
4	* This file is part of the Texy! (https://texy.nette.org)
5	* Copyright (c) 2004 David Grudl (https://davidgrudl.com)
6	*/
7
8	declare(strict_types=1);
9
10	namespace Texy\Modules;
11
12	use Texy;
13	use Texy\Nodes\HtmlCommentNode;
14	use Texy\Nodes\HtmlTagNode;
15	use Texy\ParseContext;
16	use Texy\Patterns;
17	use Texy\Position;
18	use Texy\Regexp;
19	use Texy\Syntax;
20	use function str_ends_with, strlen, strtolower, strtr, substr, trim;
21
22
23	/**
24	* Processes HTML tags and comments in input text.
25	*/
26	final class HtmlModule extends Texy\Module
27	{
28	public function __construct(	1✔
29	private Texy\Texy $texy,
30	) {
31	}	1✔
32
33
34	public function beforeParse(string &$text): void	1✔
35	{
36	$this->texy->registerLinePattern(	1✔
37	$this->parseTag(...),	1✔
38	'~
39	< (/?) # tag begin
40	([a-z][a-z0-9_:-]{0,50}) # tag name
41	(
42	(?:
43	\s++ [a-z0-9_:-]++ \| # attribute name
44	= \s+ " [^"' . Patterns::MARK . ']+ " \| # attribute value in double quotes	1✔
45	= \s+ \' [^\'' . Patterns::MARK . ']+ \' \| # attribute value in single quotes	1✔
46	= [^\s>' . Patterns::MARK . ']++ # attribute value without quotes	1✔
47	)*
48	)
49	\s*+
50	(/?) # self-closing slash
51	>
52	~is',
53	Syntax::HtmlTag,	1✔
54	);
55
56	$this->texy->registerLinePattern(	1✔
57	fn(?ParseContext $context, array $matches, array $offsets) => new HtmlCommentNode($matches[1], new Position($offsets[0], strlen($matches[0]))),	1✔
58	'~
59	<!--
60	( [^' . Patterns::MARK . ']*? )	1✔
61	-->
62	~is',
63	Syntax::HtmlComment,	1✔
64	);
65	}	1✔
66
67
68	/**
69	* Parses <tag attr="...">
70	* @param array<?string> $matches
71	* @param array<?int> $offsets
72	*/
73	public function parseTag(?ParseContext $context, array $matches, array $offsets): ?HtmlTagNode	1✔
74	{
75	[, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;	1✔
76
77	$isStart = $mEnd !== '/';	1✔
78	$isEmpty = $mEmpty === '/';	1✔
79	if (!$isEmpty && str_ends_with($mAttr, '/')) {	1✔
80	$mAttr = substr($mAttr, 0, -1);	×
81	$isEmpty = true;	×
82	}
83
84	// error - can't close empty element
85	if ($isEmpty && !$isStart) {	1✔
86	return null;	×
87	}
88
89	// error - end element with attrs
90	$mAttr = trim(strtr($mAttr, "\n", ' '));	1✔
91	if ($mAttr && !$isStart) {	1✔
92	return null;	×
93	}
94
95	return new HtmlTagNode(	1✔
96	$mTag,	1✔
97	$isStart ? $this->parseAttributes($mAttr) : [],	1✔
98	closing: !$isStart,	1✔
99	selfClosing: $isEmpty,
100	position: new Position($offsets[0], strlen($matches[0])),	1✔
101	);
102	}
103
104
105	/** @return array<string, string\|bool> */
106	private function parseAttributes(string $attrs): array	1✔
107	{
108	$res = [];	1✔
109	$matches = Regexp::matchAll(	1✔
110	$attrs,	1✔
111	<<<'X'
112	~	1✔
113	([a-z0-9_:-]+) # attribute name
114	\s*
115	(?:
116	= \s* # equals sign
117	(
118	' [^']* ' \| # single quoted value
119	" [^"]* " \| # double quoted value
120	[^'"\s]+ # unquoted value
121	)
122	)?
123	~is
124	X,
125	);
126
127	foreach ($matches as $m) {	1✔
128	$key = strtolower($m[1]);	1✔
129	$value = $m[2];	1✔
130	if ($value == null) {	1✔
131	$res[$key] = true;	×
132	} elseif ($value[0] === '\'' \|\| $value[0] === '"') {	1✔
133	$res[$key] = trim(Texy\Helpers::unescapeHtml(substr($value, 1, -1)));	1✔
134	} else {
135	$res[$key] = trim(Texy\Helpers::unescapeHtml($value));	1✔
136	}
137	}
138
139	return $res;	1✔
140	}
141	}

dg / texy / 21501721037

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