21344532034

Committed 26 Jan 2026 02:43AM UTC coverage: 91.98% (-0.4%) from 92.376%

Build # 21344532034

Build Type

push

github

Committed by

Commit Message

added CLAUDE.md

Run Details

2397 of 2606 relevant lines covered (91.98%)

0.92 hits per line

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

79.08

/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 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->solveComment(...));
                $texy->addHandler('htmlTag', $this->solveTag(...));

                $texy->registerLinePattern(
                        $this->patternTag(...),
                        '#<(/?)([a-z][a-z0-9_:-]{0,50})((?:\s++[a-z0-9\_:-]++|=\s*+"[^"' . Patterns::MARK . ']*+"|=\s*+\'[^\'' . Patterns::MARK . ']*+\'|=[^\s>' . Patterns::MARK . ']++)*)\s*+(/?)>#isu',
                        'html/tag',
                );

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


        /**
         * Callback for: <!-- comment -->.
         * @param  string[]  $matches
         */
        public function patternComment(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 patternTag(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;
        }


        /**
         * 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
                }

                // 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;
        }


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

                // sanitize comment
                $content = Texy\Regexp::replace($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 (preg_match('#^h[1-6]#i', $name)) {
                        $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
        {
                $matches = $res = [];
                preg_match_all(
                        '#([a-z0-9\_:-]+)\s*(?:=\s*(\'[^\']*\'|"[^"]*"|[^\'"\s]+))?()#isu',
                        $attrs,
                        $matches,
                        PREG_SET_ORDER,
                );

                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 function array_flip, explode, is_array, is_string, preg_match, preg_match_all, str_contains, str_ends_with, strtolower, strtoupper, strtr, substr, trim;
16	use const PREG_SET_ORDER;
17
18
19	/**
20	* Processes HTML tags and comments in input text.
21	*/
22	final class HtmlModule extends Texy\Module
23	{
24	/** pass HTML comments to output? */
25	public bool $passComment = true;
26
27
28	public function __construct(Texy\Texy $texy)	1✔
29	{
30	$this->texy = $texy;	1✔
31
32	$texy->addHandler('htmlComment', $this->solveComment(...));	1✔
33	$texy->addHandler('htmlTag', $this->solveTag(...));	1✔
34
35	$texy->registerLinePattern(	1✔
36	$this->patternTag(...),	1✔
37	'#<(/?)([a-z][a-z0-9_:-]{0,50})((?:\s++[a-z0-9\_:-]++\|=\s+"[^"' . Patterns::MARK . ']+"\|=\s+\'[^\'' . Patterns::MARK . ']+\'\|=[^\s>' . Patterns::MARK . ']++))\s+(/?)>#isu',	1✔
38	'html/tag',	1✔
39	);
40
41	$texy->registerLinePattern(	1✔
42	$this->patternComment(...),	1✔
43	'#<!--([^' . Patterns::MARK . ']*?)-->#is',	1✔
44	'html/comment',	1✔
45	);
46	}	1✔
47
48
49	/**
50	* Callback for: <!-- comment -->.
51	* @param string[] $matches
52	*/
53	public function patternComment(Texy\LineParser $parser, array $matches): HtmlElement\|string\|null	1✔
54	{
55	[, $mComment] = $matches;	1✔
56	return $this->texy->invokeAroundHandlers('htmlComment', $parser, [$mComment]);	1✔
57	}
58
59
60	/**
61	* Callback for: <tag attr="...">.
62	* @param string[] $matches
63	*/
64	public function patternTag(Texy\LineParser $parser, array $matches): ?string	1✔
65	{
66	[, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;	1✔
67	// [1] => /
68	// [2] => tag
69	// [3] => attributes
70	// [4] => /
71
72	$isStart = $mEnd !== '/';	1✔
73	$isEmpty = $mEmpty === '/';	1✔
74	if (!$isEmpty && str_ends_with($mAttr, '/')) { // uvizlo v $mAttr?	1✔
75	$mAttr = substr($mAttr, 0, -1);	×
76	$isEmpty = true;	×
77	}
78
79	// error - can't close empty element
80	if ($isEmpty && !$isStart) {	1✔
81	return null;	×
82	}
83
84	// error - end element with atttrs
85	$mAttr = trim(strtr($mAttr, "\n", ' '));	1✔
86	if ($mAttr && !$isStart) {	1✔
87	return null;	1✔
88	}
89
90	$el = new HtmlElement($mTag);	1✔
91	if ($isStart) {	1✔
92	$el->attrs = $this->parseAttributes($mAttr);	1✔
93	}
94
95	$res = $this->texy->invokeAroundHandlers('htmlTag', $parser, [$el, $isStart, $isEmpty]);	1✔
96
97	if ($res instanceof HtmlElement) {	1✔
98	return $this->texy->protect($isStart ? $res->startTag() : $res->endTag(), $res->getContentType());	1✔
99	}
100
101	return $res;	1✔
102	}
103
104
105	/**
106	* Finish invocation.
107	*/
108	private function solveTag(	1✔
109	Texy\HandlerInvocation $invocation,
110	HtmlElement $el,
111	bool $isStart,
112	?bool $forceEmpty = null,
113	): ?HtmlElement
114	{
115	$texy = $this->texy;	1✔
116
117	// tag & attibutes
118	$allowedTags = $texy->allowedTags; // speed-up	1✔
119	if (!$allowedTags) {	1✔
120	return null; // all tags are disabled	1✔
121	}
122
123	// convert case
124	$name = $el->getName();	1✔
125	$lower = strtolower($name);	1✔
126	if (isset($texy->getDTD()[$lower]) \|\| $name === strtoupper($name)) {	1✔
127	// complete UPPER convert to lower
128	$name = $lower;	1✔
129	$el->setName($name);	1✔
130	}
131
132	if (is_array($allowedTags)) {	1✔
133	if (!isset($allowedTags[$name])) {	1✔
134	return null;	1✔
135	}
136	} else { // allowedTags === Texy\Texy::ALL
137	if ($forceEmpty) {	×
138	$el->setName($name, empty: true);	×
139	}
140	}
141
142	// end tag? we are finished
143	if (!$isStart) {	1✔
144	return $el;	1✔
145	}
146
147	$this->applyAttrs($el->attrs, is_array($allowedTags) ? $allowedTags[$name] : $texy::ALL);	1✔
148	$this->applyClasses($el->attrs, $texy->getAllowedProps()[0]);	1✔
149	$this->applyStyles($el->attrs, $texy->getAllowedProps()[1]);	1✔
150	if (!$this->validateAttrs($el, $texy)) {	1✔
151	return null;	1✔
152	}
153
154	$el->validateAttrs($texy->getDTD());	1✔
155
156	return $el;	1✔
157	}
158
159
160	/**
161	* Finish invocation.
162	*/
163	private function solveComment(Texy\HandlerInvocation $invocation, string $content): string	1✔
164	{
165	if (!$this->passComment) {	1✔
166	return '';	×
167	}
168
169	// sanitize comment
170	$content = Texy\Regexp::replace($content, '#-{2,}#', ' - ');	1✔
171	$content = trim($content, '-');	1✔
172
173	return $this->texy->protect('<!--' . $content . '-->', Texy\Texy::CONTENT_MARKUP);	1✔
174	}
175
176
177	/**
178	* @param array<string, array<string\|int\|bool>\|string\|int\|bool\|null> $attrs
179	* @param bool\|string[] $allowedAttrs
180	*/
181	private function applyAttrs(array &$attrs, bool\|array $allowedAttrs): void	1✔
182	{
183	if (!$allowedAttrs) {	1✔
184	$attrs = [];	1✔
185
186	} elseif (is_array($allowedAttrs)) {	1✔
187	// skip disabled
188	$allowedAttrs = array_flip($allowedAttrs);	1✔
189	foreach ($attrs as $key => $foo) {	1✔
190	if (!isset($allowedAttrs[$key])) {	1✔
191	unset($attrs[$key]);	×
192	}
193	}
194	}
195	}	1✔
196
197
198	/**
199	* @param array<string, string\|int\|bool\|array<string\|int\|bool>\|null> $attrs
200	* @param array<string, int>\|bool $allowedClasses
201	*/
202	private function applyClasses(array &$attrs, bool\|array $allowedClasses): void	1✔
203	{
204	if (!isset($attrs['class'])) {	1✔
205	} elseif (is_array($allowedClasses)) {	×
206	$attrs['class'] = is_string($attrs['class']) ? explode(' ', $attrs['class']) : (array) $attrs['class'];	×
207	foreach ($attrs['class'] as $key => $value) {	×
208	if (!isset($allowedClasses[$value])) {	×
209	unset($attrs['class'][$key]); // id & class are case-sensitive	×
210	}
211	}
212	} elseif ($allowedClasses !== Texy\Texy::ALL) {	×
213	$attrs['class'] = null;	×
214	}
215
216	if (!isset($attrs['id'])) {	1✔
217	} elseif (is_array($allowedClasses)) {	1✔
218	if (!is_string($attrs['id']) \|\| !isset($allowedClasses['#' . $attrs['id']])) {	×
219	$attrs['id'] = null;	×
220	}
221
222	} elseif ($allowedClasses !== Texy\Texy::ALL) {	1✔
223	$attrs['id'] = null;	×
224	}
225	}	1✔
226
227
228	/**
229	* @param array<string, string\|int\|bool\|array<string\|int\|bool>\|null> $attrs
230	* @param array<string, int>\|bool $allowedStyles
231	*/
232	private function applyStyles(array &$attrs, bool\|array $allowedStyles): void	1✔
233	{
234	if (!isset($attrs['style'])) {	1✔
235	} elseif (is_array($allowedStyles)) {	1✔
236	if (is_string($attrs['style'])) {	×
237	$parts = explode(';', $attrs['style']);	×
238	$attrs['style'] = [];	×
239	foreach ($parts as $value) {	×
240	if (count($pair = explode(':', $value, 2)) === 2) {	×
241	$attrs['style'][trim($pair[0])] = trim($pair[1]);	×
242	}
243	}
244	} else {
245	$attrs['style'] = (array) $attrs['style'];	×
246	}
247
248	foreach ($attrs['style'] as $key => $value) {	×
249	if (!isset($allowedStyles[strtolower((string) $key)])) { // CSS is case-insensitive	×
250	unset($attrs['style'][$key]);	×
251	}
252	}
253	} elseif ($allowedStyles !== Texy\Texy::ALL) {	1✔
254	$attrs['style'] = null;	×
255	}
256	}	1✔
257
258
259	private function validateAttrs(HtmlElement $el, Texy\Texy $texy): bool	1✔
260	{
261	foreach (['src', 'href', 'name', 'id'] as $attr) {	1✔
262	if (isset($el->attrs[$attr])) {	1✔
263	$el->attrs[$attr] = is_string($el->attrs[$attr])	1✔
264	? trim($el->attrs[$attr])	1✔
265	: '';	×
266	if ($el->attrs[$attr] === '') {	1✔
267	unset($el->attrs[$attr]);	×
268	}
269	}
270	}
271
272	$name = $el->getName();	1✔
273	if ($name === 'img') {	1✔
274	if (!isset($el->attrs['src'])) {	1✔
275	return false;	×
276	}
277
278	assert(is_string($el->attrs['src']));
279	if (!$texy->checkURL($el->attrs['src'], $texy::FILTER_IMAGE)) {	1✔
280	return false;	×
281	}
282
283	$texy->summary['images'][] = $el->attrs['src'];	1✔
284
285	} elseif ($name === 'a') {	1✔
286	if (!isset($el->attrs['href']) && !isset($el->attrs['name']) && !isset($el->attrs['id'])) {	1✔
287	return false;	1✔
288	}
289
290	if (isset($el->attrs['href'])) {	1✔
291	assert(is_string($el->attrs['href']));
292	if ($texy->linkModule->forceNoFollow && str_contains($el->attrs['href'], '//')) {	1✔
293	$el->attrs['rel'] = (array) ($el->attrs['rel'] ?? []);	1✔
294	$el->attrs['rel'][] = 'nofollow';	1✔
295	}
296
297	if (!$texy->checkURL($el->attrs['href'], $texy::FILTER_ANCHOR)) {	1✔
298	return false;	1✔
299	}
300
301	$texy->summary['links'][] = $el->attrs['href'];	1✔
302	}
303
304	} elseif (preg_match('#^h[1-6]#i', $name)) {	1✔
305	$texy->headingModule->TOC[] = [	1✔
306	'el' => $el,	1✔
307	'level' => (int) substr($name, 1),	1✔
308	'type' => 'html',	1✔
309	];
310	}
311
312	return true;	1✔
313	}
314
315
316	/** @return array<string, string\|bool> */
317	private function parseAttributes(string $attrs): array	1✔
318	{
319	$matches = $res = [];	1✔
320	preg_match_all(	1✔
321	'#([a-z0-9\_:-]+)\s(?:=\s(\'[^\']\'\|"[^"]"\|[^\'"\s]+))?()#isu',	1✔
322	$attrs,	1✔
323	$matches,	1✔
324	PREG_SET_ORDER,	1✔
325	);
326
327	foreach ($matches as $m) {	1✔
328	$key = strtolower($m[1]);	1✔
329	$value = $m[2];	1✔
330	if ($value == null) {	1✔
331	$res[$key] = true;	1✔
332	} elseif ($value[0] === '\'' \|\| $value[0] === '"') {	1✔
333	$res[$key] = Texy\Helpers::unescapeHtml(substr($value, 1, -1));	1✔
334	} else {
335	$res[$key] = Texy\Helpers::unescapeHtml($value);	1✔
336	}
337	}
338
339	return $res;	1✔
340	}
341	}

dg / texy / 21344532034

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