12879605443

Committed 21 Jan 2025 03:31AM UTC coverage: 92.224% (+0.03%) from 92.197%

Build Type

push

github

Committed by

Commit Message

regexp: uses unmatched as null (BC break)

Run Details

14 of 14 new or added lines in 6 files covered. (100.0%)

101 existing lines in 14 files now uncovered.

2372 of 2572 relevant lines covered (92.22%)

0.92 hits per line

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

80.56

/src/Texy/Modules/HtmlModule.php

<?php

/**
 * This file is part of the Texy! (https://texy.info)
 * 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;


/**
 * Html tags module.
 */
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(...),
                        '~
                                < (/?)                          # 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->patternComment(...),
                        '~
                                <!--
                                ( [^' . Patterns::MARK . ']*? )
                                -->
                        ~is',
                        'html/comment',
                );
        }


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


        /**
         * Callback for: <tag attr="...">.
         */
        public function patternTag(Texy\LineParser $parser, array $matches): HtmlElement|string|null
        {
                [, $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|string|null
        {
                $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, 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
                $content = Regexp::replace($content, '~-{2,}~', ' - ');
                $content = trim($content, '-');

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


        private function applyAttrs(&$attrs, $allowedAttrs): void
        {
                if (!$allowedAttrs) {
                        $attrs = [];

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


        private function applyClasses(&$attrs, $allowedClasses): void
        {
                if (!isset($attrs['class'])) {
                } elseif (is_array($allowedClasses)) {
                        $attrs['class'] = explode(' ', $attrs['class']);
                        foreach ($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 (!isset($allowedClasses['#' . $attrs['id']])) {
                                $attrs['id'] = null;
                        }
                } elseif ($allowedClasses !== Texy\Texy::ALL) {
                        $attrs['id'] = null;
                }
        }


        private function applyStyles(&$attrs, $allowedStyles): void
        {
                if (!isset($attrs['style'])) {
                } elseif (is_array($allowedStyles)) {
                        $tmp = explode(';', $attrs['style']);
                        $attrs['style'] = null;
                        foreach ($tmp as $value) {
                                $pair = explode(':', $value, 2);
                                $prop = trim($pair[0]);
                                if (isset($pair[1], $allowedStyles[strtolower($prop)])) { // CSS is case-insensitive
                                        $attrs['style'][$prop] = $pair[1];
                                }
                        }
                } elseif ($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($el->attrs['src']) || !$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($el->attrs['href'])) {
                                if ($texy->linkModule->forceNoFollow && str_contains($el->attrs['href'], '//')) {
                                        if (isset($el->attrs['rel'])) {
                                                $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;
        }


        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.info)
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
17
18	/**
19	* Html tags module.
20	*/
21	final class HtmlModule extends Texy\Module
22	{
23	/** pass HTML comments to output? */
24	public bool $passComment = true;
25
26
27	public function __construct(Texy\Texy $texy)	1✔
28	{
29	$this->texy = $texy;	1✔
30
31	$texy->addHandler('htmlComment', $this->solveComment(...));	1✔
32	$texy->addHandler('htmlTag', $this->solveTag(...));	1✔
33
34	$texy->registerLinePattern(	1✔
35	$this->patternTag(...),	1✔
36	'~
37	< (/?) # tag begin
38	([a-z][a-z0-9_:-]{0,50}) # tag name
39	(
40	(?:
41	\s++ [a-z0-9\_:-]++ \| # attribute name
42	= \s+ " [^"' . Patterns::MARK . ']+ " \| # attribute value in double quotes
43	= \s+ \' [^\'' . Patterns::MARK . ']+ \' \| # attribute value in single quotes
44	= [^\s>' . Patterns::MARK . ']++ # attribute value without quotes	1✔
45	)*
46	)
47	\s*+
48	(/?) # self-closing slash
49	>
50	~is',
51	'html/tag',	1✔
52	);
53
54	$texy->registerLinePattern(	1✔
55	$this->patternComment(...),	1✔
56	'~
57	<!--
58	( [^' . Patterns::MARK . ']*? )	1✔
59	-->
60	~is',
61	'html/comment',	1✔
62	);
63	}	1✔
64
65
66	/**
67	* Callback for: <!-- comment -->.
68	*/
69	public function patternComment(Texy\LineParser $parser, array $matches): HtmlElement\|string\|null	1✔
70	{
71	[, $mComment] = $matches;	1✔
72	return $this->texy->invokeAroundHandlers('htmlComment', $parser, [$mComment]);	1✔
73	}
74
75
76	/**
77	* Callback for: <tag attr="...">.
78	*/
79	public function patternTag(Texy\LineParser $parser, array $matches): HtmlElement\|string\|null	1✔
80	{
81	[, $mEnd, $mTag, $mAttr, $mEmpty] = $matches;	1✔
82	// [1] => /
83	// [2] => tag
84	// [3] => attributes
85	// [4] => /
86
87	$isStart = $mEnd !== '/';	1✔
88	$isEmpty = $mEmpty === '/';	1✔
89	if (!$isEmpty && str_ends_with($mAttr, '/')) { // uvizlo v $mAttr?	1✔
UNCOV 90	$mAttr = substr($mAttr, 0, -1);	×
UNCOV 91	$isEmpty = true;	×
92	}
93
94	// error - can't close empty element
95	if ($isEmpty && !$isStart) {	1✔
UNCOV 96	return null;	×
97	}
98
99	// error - end element with atttrs
100	$mAttr = trim(strtr($mAttr, "\n", ' '));	1✔
101	if ($mAttr && !$isStart) {	1✔
102	return null;	1✔
103	}
104
105	$el = new HtmlElement($mTag);	1✔
106	if ($isStart) {	1✔
107	$el->attrs = $this->parseAttributes($mAttr);	1✔
108	}
109
110	$res = $this->texy->invokeAroundHandlers('htmlTag', $parser, [$el, $isStart, $isEmpty]);	1✔
111
112	if ($res instanceof HtmlElement) {	1✔
113	return $this->texy->protect($isStart ? $res->startTag() : $res->endTag(), $res->getContentType());	1✔
114	}
115
116	return $res;	1✔
117	}
118
119
120	/**
121	* Finish invocation.
122	*/
123	private function solveTag(	1✔
124	Texy\HandlerInvocation $invocation,
125	HtmlElement $el,
126	bool $isStart,
127	?bool $forceEmpty = null,
128	): HtmlElement\|string\|null
129	{
130	$texy = $this->texy;	1✔
131
132	// tag & attibutes
133	$allowedTags = $texy->allowedTags; // speed-up	1✔
134	if (!$allowedTags) {	1✔
135	return null; // all tags are disabled	1✔
136	}
137
138	// convert case
139	$name = $el->getName();	1✔
140	$lower = strtolower($name);	1✔
141	if (isset($texy->getDTD()[$lower]) \|\| $name === strtoupper($name)) {	1✔
142	// complete UPPER convert to lower
143	$name = $lower;	1✔
144	$el->setName($name);	1✔
145	}
146
147	if (is_array($allowedTags)) {	1✔
148	if (!isset($allowedTags[$name])) {	1✔
149	return null;	1✔
150	}
151	} else { // allowedTags === Texy\Texy::ALL
UNCOV 152	if ($forceEmpty) {	×
UNCOV 153	$el->setName($name, true);	×
154	}
155	}
156
157	// end tag? we are finished
158	if (!$isStart) {	1✔
159	return $el;	1✔
160	}
161
162	$this->applyAttrs($el->attrs, is_array($allowedTags) ? $allowedTags[$name] : $texy::ALL);	1✔
163	$this->applyClasses($el->attrs, $texy->getAllowedProps()[0]);	1✔
164	$this->applyStyles($el->attrs, $texy->getAllowedProps()[1]);	1✔
165	if (!$this->validateAttrs($el, $texy)) {	1✔
166	return null;	1✔
167	}
168
169	$el->validateAttrs($texy->getDTD());	1✔
170
171	return $el;	1✔
172	}
173
174
175	/**
176	* Finish invocation.
177	*/
178	private function solveComment(Texy\HandlerInvocation $invocation, string $content): string	1✔
179	{
180	if (!$this->passComment) {	1✔
UNCOV 181	return '';	×
182	}
183
184	// sanitize comment
185	$content = Regexp::replace($content, '~-{2,}~', ' - ');	1✔
186	$content = trim($content, '-');	1✔
187
188	return $this->texy->protect('<!--' . $content . '-->', Texy\Texy::CONTENT_MARKUP);	1✔
189	}
190
191
192	private function applyAttrs(&$attrs, $allowedAttrs): void
193	{
194	if (!$allowedAttrs) {	1✔
195	$attrs = [];	1✔
196
197	} elseif (is_array($allowedAttrs)) {	1✔
198	// skip disabled
199	$allowedAttrs = array_flip($allowedAttrs);	1✔
200	foreach ($attrs as $key => $foo) {	1✔
201	if (!isset($allowedAttrs[$key])) {	1✔
202	unset($attrs[$key]);	×
203	}
204	}
205	}
206	}	1✔
207
208
209	private function applyClasses(&$attrs, $allowedClasses): void
210	{
211	if (!isset($attrs['class'])) {	1✔
UNCOV 212	} elseif (is_array($allowedClasses)) {	×
UNCOV 213	$attrs['class'] = explode(' ', $attrs['class']);	×
UNCOV 214	foreach ($attrs['class'] as $key => $value) {	×
UNCOV 215	if (!isset($allowedClasses[$value])) {	×
UNCOV 216	unset($attrs['class'][$key]); // id & class are case-sensitive	×
217	}
218	}
UNCOV 219	} elseif ($allowedClasses !== Texy\Texy::ALL) {	×
220	$attrs['class'] = null;	×
221	}
222
223	if (!isset($attrs['id'])) {	1✔
224	} elseif (is_array($allowedClasses)) {	1✔
225	if (!isset($allowedClasses['#' . $attrs['id']])) {	×
226	$attrs['id'] = null;	×
227	}
228	} elseif ($allowedClasses !== Texy\Texy::ALL) {	1✔
UNCOV 229	$attrs['id'] = null;	×
230	}
231	}	1✔
232
233
234	private function applyStyles(&$attrs, $allowedStyles): void
235	{
236	if (!isset($attrs['style'])) {	1✔
237	} elseif (is_array($allowedStyles)) {	1✔
UNCOV 238	$tmp = explode(';', $attrs['style']);	×
UNCOV 239	$attrs['style'] = null;	×
UNCOV 240	foreach ($tmp as $value) {	×
241	$pair = explode(':', $value, 2);	×
UNCOV 242	$prop = trim($pair[0]);	×
243	if (isset($pair[1], $allowedStyles[strtolower($prop)])) { // CSS is case-insensitive	×
UNCOV 244	$attrs['style'][$prop] = $pair[1];	×
245	}
246	}
247	} elseif ($allowedStyles !== Texy\Texy::ALL) {	1✔
UNCOV 248	$attrs['style'] = null;	×
249	}
250	}	1✔
251
252
253	private function validateAttrs(HtmlElement $el, Texy\Texy $texy): bool	1✔
254	{
255	foreach (['src', 'href', 'name', 'id'] as $attr) {	1✔
256	if (isset($el->attrs[$attr])) {	1✔
257	$el->attrs[$attr] = is_string($el->attrs[$attr])	1✔
258	? trim($el->attrs[$attr])	1✔
UNCOV 259	: '';	×
260	if ($el->attrs[$attr] === '') {	1✔
UNCOV 261	unset($el->attrs[$attr]);	×
262	}
263	}
264	}
265
266	$name = $el->getName();	1✔
267	if ($name === 'img') {	1✔
268	if (!isset($el->attrs['src']) \|\| !$texy->checkURL($el->attrs['src'], $texy::FILTER_IMAGE)) {	1✔
UNCOV 269	return false;	×
270	}
271
272	$texy->summary['images'][] = $el->attrs['src'];	1✔
273
274	} elseif ($name === 'a') {	1✔
275	if (!isset($el->attrs['href']) && !isset($el->attrs['name']) && !isset($el->attrs['id'])) {	1✔
276	return false;	1✔
277	}
278
279	if (isset($el->attrs['href'])) {	1✔
280	if ($texy->linkModule->forceNoFollow && str_contains($el->attrs['href'], '//')) {	1✔
281	if (isset($el->attrs['rel'])) {	1✔
282	$el->attrs['rel'] = (array) $el->attrs['rel'];	1✔
283	}
284
285	$el->attrs['rel'][] = 'nofollow';	1✔
286	}
287
288	if (!$texy->checkURL($el->attrs['href'], $texy::FILTER_ANCHOR)) {	1✔
289	return false;	1✔
290	}
291
292	$texy->summary['links'][] = $el->attrs['href'];	1✔
293	}
294	} elseif (Regexp::match($name, '~^h[1-6]~i')) {	1✔
295	$texy->headingModule->TOC[] = [	1✔
296	'el' => $el,	1✔
297	'level' => (int) substr($name, 1),	1✔
298	'type' => 'html',	1✔
299	];
300	}
301
302	return true;	1✔
303	}
304
305
306	private function parseAttributes(string $attrs): array	1✔
307	{
308	$res = [];	1✔
309	$matches = Regexp::matchAll(	1✔
310	$attrs,	1✔
311	<<<'X'
312	~	1✔
313	([a-z0-9\_:-]+) # attribute name
314	\s*
315	(?:
316	= \s* # equals sign
317	(
318	' [^']* ' \| # single quoted value
319	" [^"]* " \| # double quoted value
320	[^'"\s]+ # unquoted value
321	)
322	)?
323	~is
324	X,
325	);
326
327	foreach ($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 / 12879605443

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