22283286087

Committed 22 Feb 2026 06:58PM UTC coverage: 93.01% (+0.02%) from 92.991%

Build Type

push

github

Committed by

Commit Message

LinkModule: deprecated label and modifiers in link definitions

Run Details

3 of 3 new or added lines in 1 file covered. (100.0%)

72 existing lines in 16 files now uncovered.

2089 of 2246 relevant lines covered (93.01%)

0.93 hits per line

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

81.21

/src/Texy/Modules/HtmlModule.php

<?php declare(strict_types=1);

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

namespace Texy\Modules;

use Texy;
use Texy\HtmlElement;
use Texy\Patterns;
use Texy\Regexp;
use function array_flip, count, explode, is_array, is_string, str_contains, str_ends_with, strtolower, strtr, substr, trim;


/**
 * Processes HTML tags and comments in input text.
 */
final class HtmlModule extends Texy\Module
{
        /** pass HTML comments to output? */
        public bool $passComment = true;


        public function __construct(
                private Texy\Texy $texy,
        ) {
                $texy->addHandler('htmlComment', $this->solveComment(...));
                $texy->addHandler('htmlTag', $this->solveTag(...));
        }


        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',
                        'html/tag',
                );

                $this->texy->registerLinePattern(
                        $this->parseComment(...),
                        '~
                                <!--
                                ( [^' . Patterns::MARK . ']*? )
                                -->
                        ~is',
                        'html/comment',
                );
        }


        /**
         * Parses <!-- comment -->
         * @param  array<?string>  $matches
         */
        public function parseComment(Texy\InlineParser $parser, array $matches): HtmlElement|string|null
        {
                [, $mComment] = $matches;
                return $this->texy->invokeAroundHandlers('htmlComment', $parser, [$mComment]);
        }


        /**
         * Parses <tag attr="...">
         * @param  array<?string>  $matches
         */
        public function parseTag(Texy\InlineParser $parser, array $matches): ?string
        {
                /** @var array{string, string, string, string, string} $matches */
                [, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;
                // [1] => /
                // [2] => tag
                // [3] => attributes
                // [4] => /

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

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

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

                $el = new HtmlElement($mTag);
                if ($isStart) {
                        $el->attrs = $this->parseAttributes($mAttr);
                }

                $res = $this->texy->invokeAroundHandlers('htmlTag', $parser, [$el, $isStart, $isEmpty]);

                if ($res instanceof HtmlElement) {
                        return $this->texy->protect($isStart ? $res->startTag() : $res->endTag(), $res->getContentType());
                }

                return $res;
        }


        /**
         * Finish invocation.
         */
        private function solveTag(
                Texy\HandlerInvocation $invocation,
                HtmlElement $el,
                bool $isStart,
                ?bool $forceEmpty = null,
        ): ?HtmlElement
        {
                $texy = $this->texy;

                // tag & attibutes
                $allowedTags = $texy->allowedTags; // speed-up
                if (!$allowedTags) {
                        return null; // all tags are disabled
                }

                $name = $el->getName();
                assert($name !== null);
                $name = strtolower($name);
                $el->setName($name);

                if (is_array($allowedTags)) {
                        if (!isset($allowedTags[$name])) {
                                return null;
                        }
                } else { // allowedTags === Texy\Texy::ALL
                        if ($forceEmpty) {
                                $el->setName($name, empty: true);
                        }
                }

                // end tag? we are finished
                if (!$isStart) {
                        return $el;
                }

                $this->applyAttrs($el->attrs, is_array($allowedTags) ? $allowedTags[$name] : $texy::ALL);
                $this->applyClasses($el->attrs, $texy->getAllowedProps()[0]);
                $this->applyStyles($el->attrs, $texy->getAllowedProps()[1]);
                if (!$this->validateAttrs($el, $texy)) {
                        return null;
                }

                return $el;
        }


        /**
         * Finish invocation.
         */
        private function solveComment(Texy\HandlerInvocation $invocation, string $content): string
        {
                if (!$this->passComment) {
                        return '';
                }

                // sanitize comment
                $content = Regexp::replace($content, '~-{2,}~', ' - ');
                $content = trim($content, '-');

                return $this->texy->protect('<!--' . $content . '-->', Texy\Texy::CONTENT_MARKUP);
        }


        /**
         * @param  array<string, array<string|int|bool>|string|int|bool|null>  $attrs
         * @param  bool|string[]  $allowedAttrs
         */
        private function applyAttrs(array &$attrs, bool|array $allowedAttrs): void
        {
                if (!$allowedAttrs) {
                        $attrs = [];

                } elseif (is_array($allowedAttrs)) {
                        // skip disabled
                        $allowedAttrs = array_flip($allowedAttrs);
                        foreach ($attrs as $key => $foo) {
                                if (!isset($allowedAttrs[$key])) {
                                        unset($attrs[$key]);
                                }
                        }
                }
        }


        /**
         * @param  array<string, string|int|bool|array<string|int|bool>|null>  $attrs
         * @param  array<string, int>|bool  $allowedClasses
         */
        private function applyClasses(array &$attrs, bool|array $allowedClasses): void
        {
                if (!isset($attrs['class'])) {
                } elseif (is_array($allowedClasses)) {
                        $attrs['class'] = is_string($attrs['class']) ? explode(' ', $attrs['class']) : (array) $attrs['class'];
                        foreach ($attrs['class'] as $key => $value) {
                                if (!isset($allowedClasses[$value])) {
                                        unset($attrs['class'][$key]); // id & class are case-sensitive
                                }
                        }
                } elseif ($allowedClasses !== Texy\Texy::ALL) {
                        $attrs['class'] = null;
                }

                if (!isset($attrs['id'])) {
                } elseif (is_array($allowedClasses)) {
                        if (!is_string($attrs['id']) || !isset($allowedClasses['#' . $attrs['id']])) {
                                $attrs['id'] = null;
                        }

                } elseif ($allowedClasses !== Texy\Texy::ALL) {
                        $attrs['id'] = null;
                }
        }


        /**
         * @param  array<string, string|int|bool|array<string|int|bool>|null>  $attrs
         * @param  array<string, int>|bool  $allowedStyles
         */
        private function applyStyles(array &$attrs, bool|array $allowedStyles): void
        {
                if (!isset($attrs['style'])) {
                } elseif (is_array($allowedStyles)) {
                        if (is_string($attrs['style'])) {
                                $parts = explode(';', $attrs['style']);
                                $attrs['style'] = [];
                                foreach ($parts as $value) {
                                        if (count($pair = explode(':', $value, 2)) === 2) {
                                                $attrs['style'][trim($pair[0])] = trim($pair[1]);
                                        }
                                }
                        } else {
                                $attrs['style'] = (array) $attrs['style'];
                        }

                        foreach ($attrs['style'] as $key => $value) {
                                if (!isset($allowedStyles[strtolower((string) $key)])) { // CSS is case-insensitive
                                        unset($attrs['style'][$key]);
                                }
                        }
                } elseif ($allowedStyles !== Texy\Texy::ALL) {
                        $attrs['style'] = null;
                }
        }


        private function validateAttrs(HtmlElement $el, Texy\Texy $texy): bool
        {
                foreach (['src', 'href', 'name', 'id'] as $attr) {
                        if (isset($el->attrs[$attr])) {
                                $el->attrs[$attr] = is_string($el->attrs[$attr])
                                        ? trim($el->attrs[$attr])
                                        : '';
                                if ($el->attrs[$attr] === '') {
                                        unset($el->attrs[$attr]);
                                }
                        }
                }

                $name = $el->getName();
                if ($name === 'img') {
                        if (!isset($el->attrs['src'])) {
                                return false;
                        }

                        assert(is_string($el->attrs['src']));
                        if (!$texy->checkURL($el->attrs['src'], $texy::FILTER_IMAGE)) {
                                return false;
                        }

                } elseif ($name === 'a') {
                        if (!isset($el->attrs['href']) && !isset($el->attrs['name']) && !isset($el->attrs['id'])) {
                                return false;
                        }

                        if (isset($el->attrs['href'])) {
                                assert(is_string($el->attrs['href']));
                                if ($texy->linkModule->forceNoFollow && str_contains($el->attrs['href'], '//')) {
                                        $el->attrs['rel'] = (array) ($el->attrs['rel'] ?? []);
                                        $el->attrs['rel'][] = 'nofollow';
                                }

                                if (!$texy->checkURL($el->attrs['href'], $texy::FILTER_ANCHOR)) {
                                        return false;
                                }
                        }

                } elseif (Regexp::match($name ?? '', '~^h[1-6]~i')) {
                        $texy->headingModule->TOC[] = [
                                'el' => $el,
                                'level' => (int) substr($name, 1),
                                'type' => 'html',
                        ];
                }

                return true;
        }


        /** @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,
                );

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

                return $res;
        }
}

1	<?php declare(strict_types=1);
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	namespace Texy\Modules;
9
10	use Texy;
11	use Texy\HtmlElement;
12	use Texy\Patterns;
13	use Texy\Regexp;
14	use function array_flip, count, explode, is_array, is_string, str_contains, str_ends_with, strtolower, strtr, substr, trim;
15
16
17	/**
18	* Processes HTML tags and comments in input text.
19	*/
20	final class HtmlModule extends Texy\Module
21	{
22	/** pass HTML comments to output? */
23	public bool $passComment = true;
24
25
26	public function __construct(	1✔
27	private Texy\Texy $texy,
28	) {
29	$texy->addHandler('htmlComment', $this->solveComment(...));	1✔
30	$texy->addHandler('htmlTag', $this->solveTag(...));	1✔
31	}	1✔
32
33
34	public function beforeParse(string &$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	'html/tag',	1✔
54	);
55
56	$this->texy->registerLinePattern(	1✔
57	$this->parseComment(...),	1✔
58	'~
59	<!--
60	( [^' . Patterns::MARK . ']*? )	1✔
61	-->
62	~is',
63	'html/comment',	1✔
64	);
65	}	1✔
66
67
68	/**
69	* Parses <!-- comment -->
70	* @param array<?string> $matches
71	*/
72	public function parseComment(Texy\InlineParser $parser, array $matches): HtmlElement\|string\|null	1✔
73	{
74	[, $mComment] = $matches;	1✔
75	return $this->texy->invokeAroundHandlers('htmlComment', $parser, [$mComment]);	1✔
76	}
77
78
79	/**
80	* Parses <tag attr="...">
81	* @param array<?string> $matches
82	*/
83	public function parseTag(Texy\InlineParser $parser, array $matches): ?string	1✔
84	{
85	/** @var array{string, string, string, string, string} $matches */
86	[, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;	1✔
87	// [1] => /
88	// [2] => tag
89	// [3] => attributes
90	// [4] => /
91
92	$isStart = $mEnd !== '/';	1✔
93	$isEmpty = $mEmpty === '/';	1✔
94	if (!$isEmpty && str_ends_with($mAttr, '/')) { // uvizlo v $mAttr?	1✔
95	$mAttr = substr($mAttr, 0, -1);	×
UNCOV 96	$isEmpty = true;	×
97	}
98
99	// error - can't close empty element
100	if ($isEmpty && !$isStart) {	1✔
UNCOV 101	return null;	×
102	}
103
104	// error - end element with atttrs
105	$mAttr = trim(strtr($mAttr, "\n", ' '));	1✔
106	if ($mAttr && !$isStart) {	1✔
107	return null;	1✔
108	}
109
110	$el = new HtmlElement($mTag);	1✔
111	if ($isStart) {	1✔
112	$el->attrs = $this->parseAttributes($mAttr);	1✔
113	}
114
115	$res = $this->texy->invokeAroundHandlers('htmlTag', $parser, [$el, $isStart, $isEmpty]);	1✔
116
117	if ($res instanceof HtmlElement) {	1✔
118	return $this->texy->protect($isStart ? $res->startTag() : $res->endTag(), $res->getContentType());	1✔
119	}
120
121	return $res;	1✔
122	}
123
124
125	/**
126	* Finish invocation.
127	*/
128	private function solveTag(	1✔
129	Texy\HandlerInvocation $invocation,
130	HtmlElement $el,
131	bool $isStart,
132	?bool $forceEmpty = null,
133	): ?HtmlElement
134	{
135	$texy = $this->texy;	1✔
136
137	// tag & attibutes
138	$allowedTags = $texy->allowedTags; // speed-up	1✔
139	if (!$allowedTags) {	1✔
140	return null; // all tags are disabled	1✔
141	}
142
143	$name = $el->getName();	1✔
144	assert($name !== null);
145	$name = strtolower($name);	1✔
146	$el->setName($name);	1✔
147
148	if (is_array($allowedTags)) {	1✔
149	if (!isset($allowedTags[$name])) {	1✔
150	return null;	1✔
151	}
152	} else { // allowedTags === Texy\Texy::ALL
153	if ($forceEmpty) {	1✔
154	$el->setName($name, empty: true);	1✔
155	}
156	}
157
158	// end tag? we are finished
159	if (!$isStart) {	1✔
160	return $el;	1✔
161	}
162
163	$this->applyAttrs($el->attrs, is_array($allowedTags) ? $allowedTags[$name] : $texy::ALL);	1✔
164	$this->applyClasses($el->attrs, $texy->getAllowedProps()[0]);	1✔
165	$this->applyStyles($el->attrs, $texy->getAllowedProps()[1]);	1✔
166	if (!$this->validateAttrs($el, $texy)) {	1✔
167	return null;	1✔
168	}
169
170	return $el;	1✔
171	}
172
173
174	/**
175	* Finish invocation.
176	*/
177	private function solveComment(Texy\HandlerInvocation $invocation, string $content): string	1✔
178	{
179	if (!$this->passComment) {	1✔
UNCOV 180	return '';	×
181	}
182
183	// sanitize comment
184	$content = Regexp::replace($content, '~-{2,}~', ' - ');	1✔
185	$content = trim($content, '-');	1✔
186
187	return $this->texy->protect('<!--' . $content . '-->', Texy\Texy::CONTENT_MARKUP);	1✔
188	}
189
190
191	/**
192	* @param array<string, array<string\|int\|bool>\|string\|int\|bool\|null> $attrs
193	* @param bool\|string[] $allowedAttrs
194	*/
195	private function applyAttrs(array &$attrs, bool\|array $allowedAttrs): void	1✔
196	{
197	if (!$allowedAttrs) {	1✔
198	$attrs = [];	1✔
199
200	} elseif (is_array($allowedAttrs)) {	1✔
201	// skip disabled
202	$allowedAttrs = array_flip($allowedAttrs);	1✔
203	foreach ($attrs as $key => $foo) {	1✔
204	if (!isset($allowedAttrs[$key])) {	1✔
UNCOV 205	unset($attrs[$key]);	×
206	}
207	}
208	}
209	}	1✔
210
211
212	/**
213	* @param array<string, string\|int\|bool\|array<string\|int\|bool>\|null> $attrs
214	* @param array<string, int>\|bool $allowedClasses
215	*/
216	private function applyClasses(array &$attrs, bool\|array $allowedClasses): void	1✔
217	{
218	if (!isset($attrs['class'])) {	1✔
219	} elseif (is_array($allowedClasses)) {	1✔
220	$attrs['class'] = is_string($attrs['class']) ? explode(' ', $attrs['class']) : (array) $attrs['class'];	×
UNCOV 221	foreach ($attrs['class'] as $key => $value) {	×
UNCOV 222	if (!isset($allowedClasses[$value])) {	×
UNCOV 223	unset($attrs['class'][$key]); // id & class are case-sensitive	×
224	}
225	}
226	} elseif ($allowedClasses !== Texy\Texy::ALL) {	1✔
UNCOV 227	$attrs['class'] = null;	×
228	}
229
230	if (!isset($attrs['id'])) {	1✔
231	} elseif (is_array($allowedClasses)) {	1✔
UNCOV 232	if (!is_string($attrs['id']) \|\| !isset($allowedClasses['#' . $attrs['id']])) {	×
UNCOV 233	$attrs['id'] = null;	×
234	}
235
236	} elseif ($allowedClasses !== Texy\Texy::ALL) {	1✔
UNCOV 237	$attrs['id'] = null;	×
238	}
239	}	1✔
240
241
242	/**
243	* @param array<string, string\|int\|bool\|array<string\|int\|bool>\|null> $attrs
244	* @param array<string, int>\|bool $allowedStyles
245	*/
246	private function applyStyles(array &$attrs, bool\|array $allowedStyles): void	1✔
247	{
248	if (!isset($attrs['style'])) {	1✔
249	} elseif (is_array($allowedStyles)) {	1✔
250	if (is_string($attrs['style'])) {	×
251	$parts = explode(';', $attrs['style']);	×
252	$attrs['style'] = [];	×
UNCOV 253	foreach ($parts as $value) {	×
UNCOV 254	if (count($pair = explode(':', $value, 2)) === 2) {	×
UNCOV 255	$attrs['style'][trim($pair[0])] = trim($pair[1]);	×
256	}
257	}
258	} else {
259	$attrs['style'] = (array) $attrs['style'];	×
260	}
261
UNCOV 262	foreach ($attrs['style'] as $key => $value) {	×
UNCOV 263	if (!isset($allowedStyles[strtolower((string) $key)])) { // CSS is case-insensitive	×
UNCOV 264	unset($attrs['style'][$key]);	×
265	}
266	}
267	} elseif ($allowedStyles !== Texy\Texy::ALL) {	1✔
UNCOV 268	$attrs['style'] = null;	×
269	}
270	}	1✔
271
272
273	private function validateAttrs(HtmlElement $el, Texy\Texy $texy): bool	1✔
274	{
275	foreach (['src', 'href', 'name', 'id'] as $attr) {	1✔
276	if (isset($el->attrs[$attr])) {	1✔
277	$el->attrs[$attr] = is_string($el->attrs[$attr])	1✔
278	? trim($el->attrs[$attr])	1✔
UNCOV 279	: '';	×
280	if ($el->attrs[$attr] === '') {	1✔
UNCOV 281	unset($el->attrs[$attr]);	×
282	}
283	}
284	}
285
286	$name = $el->getName();	1✔
287	if ($name === 'img') {	1✔
288	if (!isset($el->attrs['src'])) {	1✔
UNCOV 289	return false;	×
290	}
291
292	assert(is_string($el->attrs['src']));
293	if (!$texy->checkURL($el->attrs['src'], $texy::FILTER_IMAGE)) {	1✔
UNCOV 294	return false;	×
295	}
296
297	} elseif ($name === 'a') {	1✔
298	if (!isset($el->attrs['href']) && !isset($el->attrs['name']) && !isset($el->attrs['id'])) {	1✔
299	return false;	1✔
300	}
301
302	if (isset($el->attrs['href'])) {	1✔
303	assert(is_string($el->attrs['href']));
304	if ($texy->linkModule->forceNoFollow && str_contains($el->attrs['href'], '//')) {	1✔
305	$el->attrs['rel'] = (array) ($el->attrs['rel'] ?? []);	1✔
306	$el->attrs['rel'][] = 'nofollow';	1✔
307	}
308
309	if (!$texy->checkURL($el->attrs['href'], $texy::FILTER_ANCHOR)) {	1✔
310	return false;	1✔
311	}
312	}
313
314	} elseif (Regexp::match($name ?? '', '~^h[1-6]~i')) {	1✔
315	$texy->headingModule->TOC[] = [	1✔
316	'el' => $el,	1✔
317	'level' => (int) substr($name, 1),	1✔
318	'type' => 'html',	1✔
319	];
320	}
321
322	return true;	1✔
323	}
324
325
326	/** @return array<string, string\|bool> */
327	private function parseAttributes(string $attrs): array	1✔
328	{
329	$res = [];	1✔
330	$matches = Regexp::matchAll(	1✔
331	$attrs,	1✔
332	<<<'X'
333	~	1✔
334	([a-z0-9_:-]+) # attribute name
335	\s*
336	(?:
337	= \s* # equals sign
338	(
339	' [^']* ' \| # single quoted value
340	" [^"]* " \| # double quoted value
341	[^'"\s]+ # unquoted value
342	)
343	)?
344	~is
345	X,
346	);
347
348	/** @var array{string, string, ?string} $m */
349	foreach ($matches as $m) {	1✔
350	$key = strtolower($m[1]);	1✔
351	$value = $m[2];	1✔
352	if ($value == null) {	1✔
353	$res[$key] = true;	1✔
354	} elseif ($value[0] === '\'' \|\| $value[0] === '"') {	1✔
355	$res[$key] = Texy\Helpers::unescapeHtml(substr($value, 1, -1));	1✔
356	} else {
357	$res[$key] = Texy\Helpers::unescapeHtml($value);	1✔
358	}
359	}
360
361	return $res;	1✔
362	}
363	}

dg / texy / 22283286087

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