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

97.08

/src/Texy/Modules/LinkModule.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\HandlerInvocation;
use Texy\LineParser;
use Texy\Link;
use Texy\Patterns;
use Texy\Regexp;


/**
 * Links module.
 */
final class LinkModule extends Texy\Module
{
        /** root of relative links */
        public ?string $root = null;

        /** linked image class */
        public ?string $imageClass = null;

        /** always use rel="nofollow" for absolute links? */
        public bool $forceNoFollow = false;

        /** shorten URLs to more readable form? */
        public bool $shorten = true;

        /** @var array<string, Link> link references */
        private array $references = [];

        /** @var array<string, bool> */
        private static array $livelock;

        private static string $EMAIL;


        public function __construct(Texy\Texy $texy)
        {
                $this->texy = $texy;

                $texy->allowed['link/definition'] = true;
                $texy->addHandler('newReference', $this->solveNewReference(...));
                $texy->addHandler('linkReference', $this->solve(...));
                $texy->addHandler('linkEmail', $this->solveUrlEmail(...));
                $texy->addHandler('linkURL', $this->solveUrlEmail(...));
                $texy->addHandler('beforeParse', $this->beforeParse(...));

                // [reference]
                $texy->registerLinePattern(
                        $this->patternReference(...),
                        '~(
                                \[
                                [^\[\]\*\n' . Patterns::MARK . ']++  # reference
                                \]
                        )~U',
                        'link/reference',
                );

                // direct url; characters not allowed in URL <>[\]^`{|}
                $texy->registerLinePattern(
                        $this->patternUrlEmail(...),
                        '~
                                (?<= ^ | [\s([<:\x17] )            # must be preceded by these chars
                                (?: https?:// | www\. | ftp:// )   # protocol or www
                                [0-9.' . Patterns::CHAR . '-]      # first char
                                [/\d' . Patterns::CHAR . '+\.\~%&?@=_:;#$!,*()\x{ad}-]{1,1000}  # URL body
                                [/\d' . Patterns::CHAR . '+\~?@=_#$*]  # last char
                        ~',
                        'link/url',
                        '~(?: https?:// | www\. | ftp://)~',
                );

                // direct email
                self::$EMAIL = '
                        [' . Patterns::CHAR . ']                 # first char
                        [0-9.+_' . Patterns::CHAR . '-]{0,63}    # local part
                        @
                        [0-9.+_' . Patterns::CHAR . '\x{ad}-]{1,252} # domain
                        \.
                        [' . Patterns::CHAR . '\x{ad}]{2,19}     # TLD
                ';
                $texy->registerLinePattern(
                        $this->patternUrlEmail(...),
                        '~
                                (?<= ^ | [\s([<\x17] )             # must be preceded by these chars
                                ' . self::$EMAIL . '
                        ~',
                        'link/email',
                        '~' . self::$EMAIL . '~',
                );
        }


        /**
         * Text pre-processing.
         */
        private function beforeParse(Texy\Texy $texy, &$text): void
        {
                self::$livelock = [];

                // [la trine]: http://www.latrine.cz/ text odkazu .(title)[class]{style}
                if (!empty($texy->allowed['link/definition'])) {
                        $text = Texy\Regexp::replace(
                                $text,
                                '~^
                                        \[
                                        ( [^\[\]#\?\*\n]{1,100} )         # reference (1)
                                        \] : \ ++
                                        ( \S{1,1000} )                    # URL (2)
                                        ( [\ \t] .{1,1000} )?             # optional description (3)
                                        ' . Patterns::MODIFIER . '?       # modifier (4)
                                        \s*
                                $~mU',
                                $this->patternReferenceDef(...),
                        );
                }
        }


        /**
         * Callback for: [la trine]: http://www.latrine.cz/ text odkazu .(title)[class]{style}.
         */
        private function patternReferenceDef(array $matches): string
        {
                [, $mRef, $mLink, $mLabel, $mMod] = $matches;
                // [1] => [ (reference) ]
                // [2] => link
                // [3] => ...
                // [4] => .(title)[class]{style}

                $link = new Link($mLink);
                $link->label = trim($mLabel ?? '');
                $link->modifier->setProperties($mMod);
                $this->checkLink($link);
                $this->addReference($mRef, $link);
                return '';
        }


        /**
         * Callback for: [ref].
         */
        public function patternReference(LineParser $parser, array $matches): Texy\HtmlElement|string|null
        {
                [, $mRef] = $matches;
                // [1] => [ref]

                $texy = $this->texy;
                $name = substr($mRef, 1, -1);
                $link = $this->getReference($name);

                if (!$link) {
                        return $texy->invokeAroundHandlers('newReference', $parser, [$name]);
                }

                $link->type = $link::BRACKET;

                if ($link->label != '') { // null or ''
                        // prevent circular references
                        if (isset(self::$livelock[$link->name])) {
                                $content = $link->label;
                        } else {
                                self::$livelock[$link->name] = true;
                                $el = new Texy\HtmlElement;
                                $lineParser = new LineParser($texy, $el);
                                $lineParser->parse($link->label);
                                $content = $el->toString($texy);
                                unset(self::$livelock[$link->name]);
                        }
                } else {
                        $content = $this->textualUrl($link);
                        $content = $this->texy->protect($content, $texy::CONTENT_TEXTUAL);
                }

                return $texy->invokeAroundHandlers('linkReference', $parser, [$link, $content]);
        }


        /**
         * Callback for: http://davidgrudl.com david@grudl.com.
         */
        public function patternUrlEmail(LineParser $parser, array $matches, string $name): Texy\HtmlElement|string|null
        {
                [$mURL] = $matches;
                // [0] => URL

                $link = new Link($mURL);
                $this->checkLink($link);

                return $this->texy->invokeAroundHandlers(
                        $name === 'link/email' ? 'linkEmail' : 'linkURL',
                        $parser,
                        [$link],
                );
        }


        /**
         * Adds new named reference.
         */
        public function addReference(string $name, Link $link): void
        {
                $link->name = Texy\Helpers::toLower($name);
                $this->references[$link->name] = $link;
        }


        /**
         * Returns named reference.
         */
        public function getReference(string $name): ?Link
        {
                $name = Texy\Helpers::toLower($name);
                if (isset($this->references[$name])) {
                        return clone $this->references[$name];

                } else {
                        $pos = strpos($name, '?');
                        if ($pos === false) {
                                $pos = strpos($name, '#');
                        }

                        if ($pos !== false) { // try to extract ?... #... part
                                $name2 = substr($name, 0, $pos);
                                if (isset($this->references[$name2])) {
                                        $link = clone $this->references[$name2];
                                        $link->URL .= substr($name, $pos);
                                        return $link;
                                }
                        }
                }

                return null;
        }


        public function factoryLink(string $dest, ?string $mMod, ?string $label): Link
        {
                $texy = $this->texy;
                $type = Link::COMMON;

                // [ref]
                if (strlen($dest) > 1 && $dest[0] === '[' && $dest[1] !== '*') {
                        $type = Link::BRACKET;
                        $dest = substr($dest, 1, -1);
                        $link = $this->getReference($dest);

                // [* image *]
                } elseif (strlen($dest) > 1 && $dest[0] === '[' && $dest[1] === '*') {
                        $type = Link::IMAGE;
                        $dest = trim(substr($dest, 2, -2));
                        $image = $texy->imageModule->getReference($dest);
                        if ($image) {
                                $link = new Link($image->linkedURL ?? $image->URL);
                                $link->modifier = $image->modifier;
                        }
                }

                if (empty($link)) {
                        $link = new Link(trim($dest));
                        $this->checkLink($link);
                }

                if (str_contains((string) $link->URL, '%s')) {
                        $link->URL = str_replace('%s', urlencode($texy->stringToText($label)), $link->URL);
                }

                $link->modifier->setProperties($mMod);
                $link->type = $type;
                return $link;
        }


        /**
         * Finish invocation.
         */
        public function solve(
                ?HandlerInvocation $invocation,
                Link $link,
                Texy\HtmlElement|string|null $content = null,
        ): Texy\HtmlElement|string
        {
                if ($link->URL == null) {
                        return $content;
                }

                $texy = $this->texy;

                $el = new Texy\HtmlElement('a');

                if (empty($link->modifier)) {
                        $nofollow = false;
                } else {
                        $nofollow = isset($link->modifier->classes['nofollow']);
                        unset($link->modifier->classes['nofollow']);
                        $el->attrs['href'] = null; // trick - move to front
                        $link->modifier->decorate($texy, $el);
                }

                if ($link->type === Link::IMAGE) {
                        // image
                        $el->attrs['href'] = Texy\Helpers::prependRoot($link->URL, $texy->imageModule->linkedRoot);
                        if ($this->imageClass) {
                                $el->attrs['class'][] = $this->imageClass;
                        }
                } else {
                        $el->attrs['href'] = Texy\Helpers::prependRoot($link->URL, $this->root);

                        // rel="nofollow"
                        if ($nofollow || ($this->forceNoFollow && str_contains($el->attrs['href'], '//'))) {
                                $el->attrs['rel'] = 'nofollow';
                        }
                }

                if ($content !== null) {
                        $el->add($content);
                }

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

                return $el;
        }


        /**
         * Finish invocation.
         */
        private function solveUrlEmail(HandlerInvocation $invocation, Link $link): Texy\HtmlElement|string
        {
                $content = $this->textualUrl($link);
                $content = $this->texy->protect($content, Texy\Texy::CONTENT_TEXTUAL);
                return $this->solve(null, $link, $content);
        }


        /**
         * Finish invocation.
         */
        private function solveNewReference(HandlerInvocation $invocation, string $name)
        {
                // no change
        }


        /**
         * Checks and corrects $URL.
         */
        private function checkLink(Link $link): void
        {
                // remove soft hyphens; if not removed by Texy\Texy::process()
                $link->URL = str_replace("\u{AD}", '', $link->URL);

                if (strncasecmp($link->URL, 'www.', 4) === 0) {
                        // special supported case
                        $link->URL = 'http://' . $link->URL;

                } elseif (Regexp::match($link->URL, '~' . self::$EMAIL . '$~A')) {
                        // email
                        $link->URL = 'mailto:' . $link->URL;

                } elseif (!$this->texy->checkURL($link->URL, Texy\Texy::FILTER_ANCHOR)) {
                        $link->URL = null;

                } else {
                        $link->URL = str_replace('&amp;', '&', $link->URL); // replace unwanted &amp;
                }
        }


        /**
         * Returns textual representation of URL.
         */
        private function textualUrl(Link $link): string
        {
                if ($this->texy->obfuscateEmail && Regexp::match($link->raw, '~^' . self::$EMAIL . '$~')) { // email
                        return str_replace('@', '&#64;<!-- -->', $link->raw);
                }

                if ($this->shorten && Regexp::match($link->raw, '~^(https?://|ftp://|www\.|/)~i')) {
                        $raw = strncasecmp($link->raw, 'www.', 4) === 0
                                ? 'none://' . $link->raw
                                : $link->raw;

                        // parse_url() in PHP damages UTF-8 - use regular expression
                        if (!($parts = Regexp::match($raw, '~^
                                (?: (?P<scheme> [a-z]+ ) : )?
                                (?: // (?P<host> [^/?#]+ ) )?
                                (?P<path> (?: / | ^ ) (?! / ) [^?#]* )?
                                (?: \? (?P<query> [^#]* ) )?
                                (?: \# (?P<fragment> .* ) )?
                                $
                        ~'))) {
                                return $link->raw;
                        }

                        $res = '';
                        if ($parts['scheme'] !== null && $parts['scheme'] !== 'none') {
                                $res .= $parts['scheme'] . '://';
                        }

                        if ($parts['host'] !== null) {
                                $res .= $parts['host'];
                        }

                        if ($parts['path'] !== null) {
                                $res .= (iconv_strlen($parts['path'], 'UTF-8') > 16 ? ("/\u{2026}" . iconv_substr($parts['path'], -12, 12, 'UTF-8')) : $parts['path']);
                        }

                        if ($parts['query'] > '') {
                                $res .= iconv_strlen($parts['query'], 'UTF-8') > 4
                                        ? "?\u{2026}"
                                        : ('?' . $parts['query']);
                        } elseif ($parts['fragment'] > '') {
                                $res .= iconv_strlen($parts['fragment'], 'UTF-8') > 4
                                        ? "#\u{2026}"
                                        : ('#' . $parts['fragment']);
                        }

                        return $res;
                }

                return $link->raw;
        }
}

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\HandlerInvocation;
14	use Texy\LineParser;
15	use Texy\Link;
16	use Texy\Patterns;
17	use Texy\Regexp;
18
19
20	/**
21	* Links module.
22	*/
23	final class LinkModule extends Texy\Module
24	{
25	/** root of relative links */
26	public ?string $root = null;
27
28	/** linked image class */
29	public ?string $imageClass = null;
30
31	/** always use rel="nofollow" for absolute links? */
32	public bool $forceNoFollow = false;
33
34	/** shorten URLs to more readable form? */
35	public bool $shorten = true;
36
37	/** @var array<string, Link> link references */
38	private array $references = [];
39
40	/** @var array<string, bool> */
41	private static array $livelock;
42
43	private static string $EMAIL;
44
45
46	public function __construct(Texy\Texy $texy)	1✔
47	{
48	$this->texy = $texy;	1✔
49
50	$texy->allowed['link/definition'] = true;	1✔
51	$texy->addHandler('newReference', $this->solveNewReference(...));	1✔
52	$texy->addHandler('linkReference', $this->solve(...));	1✔
53	$texy->addHandler('linkEmail', $this->solveUrlEmail(...));	1✔
54	$texy->addHandler('linkURL', $this->solveUrlEmail(...));	1✔
55	$texy->addHandler('beforeParse', $this->beforeParse(...));	1✔
56
57	// [reference]
58	$texy->registerLinePattern(	1✔
59	$this->patternReference(...),	1✔
60	'~(
61	\[
62	[^\[\]\*\n' . Patterns::MARK . ']++ # reference	1✔
63	\]
64	)~U',
65	'link/reference',	1✔
66	);
67
68	// direct url; characters not allowed in URL <>[\]^`{\|}
69	$texy->registerLinePattern(	1✔
70	$this->patternUrlEmail(...),	1✔
71	'~
72	(?<= ^ \| [\s([<:\x17] ) # must be preceded by these chars
73	(?: https?:// \| www\. \| ftp:// ) # protocol or www
74	[0-9.' . Patterns::CHAR . '-] # first char
75	[/\d' . Patterns::CHAR . '+\.\~%&?@=_:;#$!,*()\x{ad}-]{1,1000} # URL body
76	[/\d' . Patterns::CHAR . '+\~?@=_#$*] # last char	1✔
77	~',
78	'link/url',	1✔
79	'~(?: https?:// \| www\. \| ftp://)~',	1✔
80	);
81
82	// direct email
83	self::$EMAIL = '	1✔
84	[' . Patterns::CHAR . '] # first char
85	[0-9.+_' . Patterns::CHAR . '-]{0,63} # local part
86	@
87	[0-9.+_' . Patterns::CHAR . '\x{ad}-]{1,252} # domain
88	\.
89	[' . Patterns::CHAR . '\x{ad}]{2,19} # TLD
90	';
91	$texy->registerLinePattern(	1✔
92	$this->patternUrlEmail(...),	1✔
93	'~
94	(?<= ^ \| [\s([<\x17] ) # must be preceded by these chars
95	' . self::$EMAIL . '	1✔
96	~',
97	'link/email',	1✔
98	'~' . self::$EMAIL . '~',	1✔
99	);
100	}	1✔
101
102
103	/**
104	* Text pre-processing.
105	*/
106	private function beforeParse(Texy\Texy $texy, &$text): void	1✔
107	{
108	self::$livelock = [];	1✔
109
110	// [la trine]: http://www.latrine.cz/ text odkazu .(title)[class]{style}
111	if (!empty($texy->allowed['link/definition'])) {	1✔
112	$text = Texy\Regexp::replace(	1✔
113	$text,	1✔
114	'~^
115	\[
116	( [^\[\]#\?\*\n]{1,100} ) # reference (1)
117	\] : \ ++
118	( \S{1,1000} ) # URL (2)
119	( [\ \t] .{1,1000} )? # optional description (3)
120	' . Patterns::MODIFIER . '? # modifier (4)	1✔
121	\s*
122	$~mU',
123	$this->patternReferenceDef(...),	1✔
124	);
125	}
126	}	1✔
127
128
129	/**
130	* Callback for: [la trine]: http://www.latrine.cz/ text odkazu .(title)[class]{style}.
131	*/
132	private function patternReferenceDef(array $matches): string	1✔
133	{
134	[, $mRef, $mLink, $mLabel, $mMod] = $matches;	1✔
135	// [1] => [ (reference) ]
136	// [2] => link
137	// [3] => ...
138	// [4] => .(title)[class]{style}
139
140	$link = new Link($mLink);	1✔
141	$link->label = trim($mLabel ?? '');	1✔
142	$link->modifier->setProperties($mMod);	1✔
143	$this->checkLink($link);	1✔
144	$this->addReference($mRef, $link);	1✔
145	return '';	1✔
146	}
147
148
149	/**
150	* Callback for: [ref].
151	*/
152	public function patternReference(LineParser $parser, array $matches): Texy\HtmlElement\|string\|null	1✔
153	{
154	[, $mRef] = $matches;	1✔
155	// [1] => [ref]
156
157	$texy = $this->texy;	1✔
158	$name = substr($mRef, 1, -1);	1✔
159	$link = $this->getReference($name);	1✔
160
161	if (!$link) {	1✔
162	return $texy->invokeAroundHandlers('newReference', $parser, [$name]);	1✔
163	}
164
165	$link->type = $link::BRACKET;	1✔
166
167	if ($link->label != '') { // null or ''	1✔
168	// prevent circular references
169	if (isset(self::$livelock[$link->name])) {	1✔
UNCOV 170	$content = $link->label;	×
171	} else {
172	self::$livelock[$link->name] = true;	1✔
173	$el = new Texy\HtmlElement;	1✔
174	$lineParser = new LineParser($texy, $el);	1✔
175	$lineParser->parse($link->label);	1✔
176	$content = $el->toString($texy);	1✔
177	unset(self::$livelock[$link->name]);	1✔
178	}
179	} else {
180	$content = $this->textualUrl($link);	1✔
181	$content = $this->texy->protect($content, $texy::CONTENT_TEXTUAL);	1✔
182	}
183
184	return $texy->invokeAroundHandlers('linkReference', $parser, [$link, $content]);	1✔
185	}
186
187
188	/**
189	* Callback for: http://davidgrudl.com david@grudl.com.
190	*/
191	public function patternUrlEmail(LineParser $parser, array $matches, string $name): Texy\HtmlElement\|string\|null	1✔
192	{
193	[$mURL] = $matches;	1✔
194	// [0] => URL
195
196	$link = new Link($mURL);	1✔
197	$this->checkLink($link);	1✔
198
199	return $this->texy->invokeAroundHandlers(	1✔
200	$name === 'link/email' ? 'linkEmail' : 'linkURL',	1✔
201	$parser,
202	[$link],	1✔
203	);
204	}
205
206
207	/**
208	* Adds new named reference.
209	*/
210	public function addReference(string $name, Link $link): void	1✔
211	{
212	$link->name = Texy\Helpers::toLower($name);	1✔
213	$this->references[$link->name] = $link;	1✔
214	}	1✔
215
216
217	/**
218	* Returns named reference.
219	*/
220	public function getReference(string $name): ?Link	1✔
221	{
222	$name = Texy\Helpers::toLower($name);	1✔
223	if (isset($this->references[$name])) {	1✔
224	return clone $this->references[$name];	1✔
225
226	} else {
227	$pos = strpos($name, '?');	1✔
228	if ($pos === false) {	1✔
229	$pos = strpos($name, '#');	1✔
230	}
231
232	if ($pos !== false) { // try to extract ?... #... part	1✔
233	$name2 = substr($name, 0, $pos);	1✔
234	if (isset($this->references[$name2])) {	1✔
235	$link = clone $this->references[$name2];	1✔
236	$link->URL .= substr($name, $pos);	1✔
237	return $link;	1✔
238	}
239	}
240	}
241
242	return null;	1✔
243	}
244
245
246	public function factoryLink(string $dest, ?string $mMod, ?string $label): Link	1✔
247	{
248	$texy = $this->texy;	1✔
249	$type = Link::COMMON;	1✔
250
251	// [ref]
252	if (strlen($dest) > 1 && $dest[0] === '[' && $dest[1] !== '*') {	1✔
253	$type = Link::BRACKET;	1✔
254	$dest = substr($dest, 1, -1);	1✔
255	$link = $this->getReference($dest);	1✔
256
257	// [* image *]
258	} elseif (strlen($dest) > 1 && $dest[0] === '[' && $dest[1] === '*') {	1✔
259	$type = Link::IMAGE;	1✔
260	$dest = trim(substr($dest, 2, -2));	1✔
261	$image = $texy->imageModule->getReference($dest);	1✔
262	if ($image) {	1✔
263	$link = new Link($image->linkedURL ?? $image->URL);	1✔
264	$link->modifier = $image->modifier;	1✔
265	}
266	}
267
268	if (empty($link)) {	1✔
269	$link = new Link(trim($dest));	1✔
270	$this->checkLink($link);	1✔
271	}
272
273	if (str_contains((string) $link->URL, '%s')) {	1✔
UNCOV 274	$link->URL = str_replace('%s', urlencode($texy->stringToText($label)), $link->URL);	×
275	}
276
277	$link->modifier->setProperties($mMod);	1✔
278	$link->type = $type;	1✔
279	return $link;	1✔
280	}
281
282
283	/**
284	* Finish invocation.
285	*/
286	public function solve(	1✔
287	?HandlerInvocation $invocation,
288	Link $link,
289	Texy\HtmlElement\|string\|null $content = null,
290	): Texy\HtmlElement\|string
291	{
292	if ($link->URL == null) {	1✔
293	return $content;	1✔
294	}
295
296	$texy = $this->texy;	1✔
297
298	$el = new Texy\HtmlElement('a');	1✔
299
300	if (empty($link->modifier)) {	1✔
UNCOV 301	$nofollow = false;	×
302	} else {
303	$nofollow = isset($link->modifier->classes['nofollow']);	1✔
304	unset($link->modifier->classes['nofollow']);	1✔
305	$el->attrs['href'] = null; // trick - move to front	1✔
306	$link->modifier->decorate($texy, $el);	1✔
307	}
308
309	if ($link->type === Link::IMAGE) {	1✔
310	// image
311	$el->attrs['href'] = Texy\Helpers::prependRoot($link->URL, $texy->imageModule->linkedRoot);	1✔
312	if ($this->imageClass) {	1✔
313	$el->attrs['class'][] = $this->imageClass;	1✔
314	}
315	} else {
316	$el->attrs['href'] = Texy\Helpers::prependRoot($link->URL, $this->root);	1✔
317
318	// rel="nofollow"
319	if ($nofollow \|\| ($this->forceNoFollow && str_contains($el->attrs['href'], '//'))) {	1✔
320	$el->attrs['rel'] = 'nofollow';	1✔
321	}
322	}
323
324	if ($content !== null) {	1✔
325	$el->add($content);	1✔
326	}
327
328	$texy->summary['links'][] = $el->attrs['href'];	1✔
329
330	return $el;	1✔
331	}
332
333
334	/**
335	* Finish invocation.
336	*/
337	private function solveUrlEmail(HandlerInvocation $invocation, Link $link): Texy\HtmlElement\|string	1✔
338	{
339	$content = $this->textualUrl($link);	1✔
340	$content = $this->texy->protect($content, Texy\Texy::CONTENT_TEXTUAL);	1✔
341	return $this->solve(null, $link, $content);	1✔
342	}
343
344
345	/**
346	* Finish invocation.
347	*/
348	private function solveNewReference(HandlerInvocation $invocation, string $name)	1✔
349	{
350	// no change
351	}	1✔
352
353
354	/**
355	* Checks and corrects $URL.
356	*/
357	private function checkLink(Link $link): void	1✔
358	{
359	// remove soft hyphens; if not removed by Texy\Texy::process()
360	$link->URL = str_replace("\u{AD}", '', $link->URL);	1✔
361
362	if (strncasecmp($link->URL, 'www.', 4) === 0) {	1✔
363	// special supported case
364	$link->URL = 'http://' . $link->URL;	1✔
365
366	} elseif (Regexp::match($link->URL, '~' . self::$EMAIL . '$~A')) {	1✔
367	// email
368	$link->URL = 'mailto:' . $link->URL;	1✔
369
370	} elseif (!$this->texy->checkURL($link->URL, Texy\Texy::FILTER_ANCHOR)) {	1✔
371	$link->URL = null;	1✔
372
373	} else {
374	$link->URL = str_replace('&', '&', $link->URL); // replace unwanted &	1✔
375	}
376	}	1✔
377
378
379	/**
380	* Returns textual representation of URL.
381	*/
382	private function textualUrl(Link $link): string	1✔
383	{
384	if ($this->texy->obfuscateEmail && Regexp::match($link->raw, '~^' . self::$EMAIL . '$~')) { // email	1✔
385	return str_replace('@', '@<!-- -->', $link->raw);	1✔
386	}
387
388	if ($this->shorten && Regexp::match($link->raw, '~^(https?://\|ftp://\|www\.\|/)~i')) {	1✔
389	$raw = strncasecmp($link->raw, 'www.', 4) === 0	1✔
390	? 'none://' . $link->raw	1✔
391	: $link->raw;	1✔
392
393	// parse_url() in PHP damages UTF-8 - use regular expression
394	if (!($parts = Regexp::match($raw, '~^	1✔
395	(?: (?P<scheme> [a-z]+ ) : )?
396	(?: // (?P<host> [^/?#]+ ) )?
397	(?P<path> (?: / \| ^ ) (?! / ) [^?#]* )?
398	(?: \? (?P<query> [^#]* ) )?
399	(?: \# (?P<fragment> .* ) )?
400	$
401	~'))) {
UNCOV 402	return $link->raw;	×
403	}
404
405	$res = '';	1✔
406	if ($parts['scheme'] !== null && $parts['scheme'] !== 'none') {	1✔
407	$res .= $parts['scheme'] . '://';	1✔
408	}
409
410	if ($parts['host'] !== null) {	1✔
411	$res .= $parts['host'];	1✔
412	}
413
414	if ($parts['path'] !== null) {	1✔
415	$res .= (iconv_strlen($parts['path'], 'UTF-8') > 16 ? ("/\u{2026}" . iconv_substr($parts['path'], -12, 12, 'UTF-8')) : $parts['path']);	1✔
416	}
417
418	if ($parts['query'] > '') {	1✔
419	$res .= iconv_strlen($parts['query'], 'UTF-8') > 4	1✔
UNCOV 420	? "?\u{2026}"	×
421	: ('?' . $parts['query']);	1✔
422	} elseif ($parts['fragment'] > '') {	1✔
423	$res .= iconv_strlen($parts['fragment'], 'UTF-8') > 4	1✔
424	? "#\u{2026}"	1✔
425	: ('#' . $parts['fragment']);	1✔
426	}
427
428	return $res;	1✔
429	}
430
431	return $link->raw;	1✔
432	}
433	}

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