21345344909

Committed 26 Jan 2026 03:32AM UTC coverage: 92.382% (-0.4%) from 92.744%

Build Type

push

github

Committed by

Commit Message

HtmlElement: removed toHtml() & toText()

Run Details

18 of 19 new or added lines in 5 files covered. (94.74%)

149 existing lines in 21 files now uncovered.

2401 of 2599 relevant lines covered (92.38%)

0.92 hits per line

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

78.81

/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\HtmlElement;
use Texy\Patterns;
use Texy\Regexp;
use function array_flip, explode, is_array, is_string, preg_match, preg_match_all, str_contains, str_ends_with, strtolower, strtoupper, strtr, substr, trim;
use const PREG_SET_ORDER;


/**
 * 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(Texy\Texy $texy)
        {
                $this->texy = $texy;

                $texy->addHandler('htmlComment', $this->commentToElement(...));
                $texy->addHandler('htmlTag', $this->tagToElement(...));

                $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',
                );

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


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


        /**
         * Callback for: <tag attr="...">.
         * @param  string[]  $matches
         */
        public function parseTag(Texy\LineParser $parser, array $matches): ?string
        {
                [, $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;
        }


        public function tagToElement(
                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
                }

                // convert case
                $name = $el->getName();
                $lower = strtolower($name);
                if (isset($texy->getDTD()[$lower]) || $name === strtoupper($name)) {
                        // complete UPPER convert to lower
                        $name = $lower;
                        $el->setName($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;
                }

                $el->validateAttrs($texy->getDTD());

                return $el;
        }


        public function commentToElement(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;
                        }

                        $texy->summary['images'][] = $el->attrs['src'];

                } 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;
                                }

                                $texy->summary['links'][] = $el->attrs['href'];
                        }

                } 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,
                );

                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
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\HtmlElement;
14	use Texy\Patterns;
15	use Texy\Regexp;
16	use function array_flip, explode, is_array, is_string, preg_match, preg_match_all, str_contains, str_ends_with, strtolower, strtoupper, strtr, substr, trim;
17	use const PREG_SET_ORDER;
18
19
20	/**
21	* Processes HTML tags and comments in input text.
22	*/
23	final class HtmlModule extends Texy\Module
24	{
25	/** pass HTML comments to output? */
26	public bool $passComment = true;
27
28
29	public function __construct(Texy\Texy $texy)	1✔
30	{
31	$this->texy = $texy;	1✔
32
33	$texy->addHandler('htmlComment', $this->commentToElement(...));	1✔
34	$texy->addHandler('htmlTag', $this->tagToElement(...));	1✔
35
36	$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
45	= \s+ \' [^\'' . Patterns::MARK . ']+ \' \| # attribute value in single quotes
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	$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	* Callback for: <!-- comment -->.
70	* @param string[] $matches
71	*/
72	public function parseComment(Texy\LineParser $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	* Callback for: <tag attr="...">.
81	* @param string[] $matches
82	*/
83	public function parseTag(Texy\LineParser $parser, array $matches): ?string	1✔
84	{
85	[, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;	1✔
86	// [1] => /
87	// [2] => tag
88	// [3] => attributes
89	// [4] => /
90
91	$isStart = $mEnd !== '/';	1✔
92	$isEmpty = $mEmpty === '/';	1✔
93	if (!$isEmpty && str_ends_with($mAttr, '/')) { // uvizlo v $mAttr?	1✔
UNCOV 94	$mAttr = substr($mAttr, 0, -1);	×
UNCOV 95	$isEmpty = true;	×
96	}
97
98	// error - can't close empty element
99	if ($isEmpty && !$isStart) {	1✔
UNCOV 100	return null;	×
101	}
102
103	// error - end element with atttrs
104	$mAttr = trim(strtr($mAttr, "\n", ' '));	1✔
105	if ($mAttr && !$isStart) {	1✔
106	return null;	1✔
107	}
108
109	$el = new HtmlElement($mTag);	1✔
110	if ($isStart) {	1✔
111	$el->attrs = $this->parseAttributes($mAttr);	1✔
112	}
113
114	$res = $this->texy->invokeAroundHandlers('htmlTag', $parser, [$el, $isStart, $isEmpty]);	1✔
115
116	if ($res instanceof HtmlElement) {	1✔
117	return $this->texy->protect($isStart ? $res->startTag() : $res->endTag(), $res->getContentType());	1✔
118	}
119
120	return $res;	1✔
121	}
122
123
124	public function tagToElement(	1✔
125	Texy\HandlerInvocation $invocation,
126	HtmlElement $el,
127	bool $isStart,
128	?bool $forceEmpty = null,
129	): ?HtmlElement
130	{
131	$texy = $this->texy;	1✔
132
133	// tag & attibutes
134	$allowedTags = $texy->allowedTags; // speed-up	1✔
135	if (!$allowedTags) {	1✔
136	return null; // all tags are disabled	1✔
137	}
138
139	// convert case
140	$name = $el->getName();	1✔
141	$lower = strtolower($name);	1✔
142	if (isset($texy->getDTD()[$lower]) \|\| $name === strtoupper($name)) {	1✔
143	// complete UPPER convert to lower
144	$name = $lower;	1✔
145	$el->setName($name);	1✔
146	}
147
148	if (is_array($allowedTags)) {	1✔
149	if (!isset($allowedTags[$name])) {	1✔
150	return null;	1✔
151	}
152	} else { // allowedTags === Texy\Texy::ALL
UNCOV 153	if ($forceEmpty) {	×
UNCOV 154	$el->setName($name, empty: true);	×
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	$el->validateAttrs($texy->getDTD());	1✔
171
172	return $el;	1✔
173	}
174
175
176	public function commentToElement(Texy\HandlerInvocation $invocation, string $content): string	1✔
177	{
178	if (!$this->passComment) {	1✔
UNCOV 179	return '';	×
180	}
181
182	// sanitize comment
183	$content = Regexp::replace($content, '~-{2,}~', ' - ');	1✔
184	$content = trim($content, '-');	1✔
185
186	return $this->texy->protect('<!--' . $content . '-->', Texy\Texy::CONTENT_MARKUP);	1✔
187	}
188
189
190	/**
191	* @param array<string, array<string\|int\|bool>\|string\|int\|bool\|null> $attrs
192	* @param bool\|string[] $allowedAttrs
193	*/
194	private function applyAttrs(array &$attrs, bool\|array $allowedAttrs): void	1✔
195	{
196	if (!$allowedAttrs) {	1✔
197	$attrs = [];	1✔
198
199	} elseif (is_array($allowedAttrs)) {	1✔
200	// skip disabled
201	$allowedAttrs = array_flip($allowedAttrs);	1✔
202	foreach ($attrs as $key => $foo) {	1✔
203	if (!isset($allowedAttrs[$key])) {	1✔
UNCOV 204	unset($attrs[$key]);	×
205	}
206	}
207	}
208	}	1✔
209
210
211	/**
212	* @param array<string, string\|int\|bool\|array<string\|int\|bool>\|null> $attrs
213	* @param array<string, int>\|bool $allowedClasses
214	*/
215	private function applyClasses(array &$attrs, bool\|array $allowedClasses): void	1✔
216	{
217	if (!isset($attrs['class'])) {	1✔
UNCOV 218	} elseif (is_array($allowedClasses)) {	×
UNCOV 219	$attrs['class'] = is_string($attrs['class']) ? explode(' ', $attrs['class']) : (array) $attrs['class'];	×
UNCOV 220	foreach ($attrs['class'] as $key => $value) {	×
221	if (!isset($allowedClasses[$value])) {	×
222	unset($attrs['class'][$key]); // id & class are case-sensitive	×
223	}
224	}
225	} elseif ($allowedClasses !== Texy\Texy::ALL) {	×
UNCOV 226	$attrs['class'] = null;	×
227	}
228
229	if (!isset($attrs['id'])) {	1✔
230	} elseif (is_array($allowedClasses)) {	1✔
UNCOV 231	if (!is_string($attrs['id']) \|\| !isset($allowedClasses['#' . $attrs['id']])) {	×
UNCOV 232	$attrs['id'] = null;	×
233	}
234
235	} elseif ($allowedClasses !== Texy\Texy::ALL) {	1✔
236	$attrs['id'] = null;	×
237	}
238	}	1✔
239
240
241	/**
242	* @param array<string, string\|int\|bool\|array<string\|int\|bool>\|null> $attrs
243	* @param array<string, int>\|bool $allowedStyles
244	*/
245	private function applyStyles(array &$attrs, bool\|array $allowedStyles): void	1✔
246	{
247	if (!isset($attrs['style'])) {	1✔
248	} elseif (is_array($allowedStyles)) {	1✔
UNCOV 249	if (is_string($attrs['style'])) {	×
UNCOV 250	$parts = explode(';', $attrs['style']);	×
UNCOV 251	$attrs['style'] = [];	×
UNCOV 252	foreach ($parts as $value) {	×
UNCOV 253	if (count($pair = explode(':', $value, 2)) === 2) {	×
UNCOV 254	$attrs['style'][trim($pair[0])] = trim($pair[1]);	×
255	}
256	}
257	} else {
UNCOV 258	$attrs['style'] = (array) $attrs['style'];	×
259	}
260
UNCOV 261	foreach ($attrs['style'] as $key => $value) {	×
UNCOV 262	if (!isset($allowedStyles[strtolower((string) $key)])) { // CSS is case-insensitive	×
UNCOV 263	unset($attrs['style'][$key]);	×
264	}
265	}
266	} elseif ($allowedStyles !== Texy\Texy::ALL) {	1✔
UNCOV 267	$attrs['style'] = null;	×
268	}
269	}	1✔
270
271
272	private function validateAttrs(HtmlElement $el, Texy\Texy $texy): bool	1✔
273	{
274	foreach (['src', 'href', 'name', 'id'] as $attr) {	1✔
275	if (isset($el->attrs[$attr])) {	1✔
276	$el->attrs[$attr] = is_string($el->attrs[$attr])	1✔
277	? trim($el->attrs[$attr])	1✔
UNCOV 278	: '';	×
279	if ($el->attrs[$attr] === '') {	1✔
UNCOV 280	unset($el->attrs[$attr]);	×
281	}
282	}
283	}
284
285	$name = $el->getName();	1✔
286	if ($name === 'img') {	1✔
287	if (!isset($el->attrs['src'])) {	1✔
UNCOV 288	return false;	×
289	}
290
291	assert(is_string($el->attrs['src']));
292	if (!$texy->checkURL($el->attrs['src'], $texy::FILTER_IMAGE)) {	1✔
UNCOV 293	return false;	×
294	}
295
296	$texy->summary['images'][] = $el->attrs['src'];	1✔
297
298	} elseif ($name === 'a') {	1✔
299	if (!isset($el->attrs['href']) && !isset($el->attrs['name']) && !isset($el->attrs['id'])) {	1✔
300	return false;	1✔
301	}
302
303	if (isset($el->attrs['href'])) {	1✔
304	assert(is_string($el->attrs['href']));
305	if ($texy->linkModule->forceNoFollow && str_contains($el->attrs['href'], '//')) {	1✔
306	$el->attrs['rel'] = (array) ($el->attrs['rel'] ?? []);	1✔
307	$el->attrs['rel'][] = 'nofollow';	1✔
308	}
309
310	if (!$texy->checkURL($el->attrs['href'], $texy::FILTER_ANCHOR)) {	1✔
311	return false;	1✔
312	}
313
314	$texy->summary['links'][] = $el->attrs['href'];	1✔
315	}
316
317	} elseif (Regexp::match($name, '~^h[1-6]~i')) {	1✔
318	$texy->headingModule->TOC[] = [	1✔
319	'el' => $el,	1✔
320	'level' => (int) substr($name, 1),	1✔
321	'type' => 'html',	1✔
322	];
323	}
324
325	return true;	1✔
326	}
327
328
329	/** @return array<string, string\|bool> */
330	private function parseAttributes(string $attrs): array	1✔
331	{
332	$res = [];	1✔
333	$matches = Regexp::matchAll(	1✔
334	$attrs,	1✔
335	<<<'X'
336	~	1✔
337	([a-z0-9_:-]+) # attribute name
338	\s*
339	(?:
340	= \s* # equals sign
341	(
342	' [^']* ' \| # single quoted value
343	" [^"]* " \| # double quoted value
344	[^'"\s]+ # unquoted value
345	)
346	)?
347	~is
348	X,
349	);
350
351	foreach ($matches as $m) {	1✔
352	$key = strtolower($m[1]);	1✔
353	$value = $m[2];	1✔
354	if ($value == null) {	1✔
355	$res[$key] = true;	1✔
356	} elseif ($value[0] === '\'' \|\| $value[0] === '"') {	1✔
357	$res[$key] = Texy\Helpers::unescapeHtml(substr($value, 1, -1));	1✔
358	} else {
359	$res[$key] = Texy\Helpers::unescapeHtml($value);	1✔
360	}
361	}
362
363	return $res;	1✔
364	}
365	}

dg / texy / 21345344909

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