1<?php
2
3// SPDX-FileCopyrightText: 2004-2023 Ryan Parman, Sam Sneddon, Ryan McCue
4// SPDX-License-Identifier: BSD-3-Clause
5
6declare(strict_types=1);
7
8namespace SimplePie;
9
10use DomDocument;
11use Psr\Http\Client\ClientInterface;
12use Psr\Http\Message\RequestFactoryInterface;
13use Psr\Http\Message\UriFactoryInterface;
14use SimplePie\HTTP\Client;
15use SimplePie\HTTP\ClientException;
16use SimplePie\HTTP\FileClient;
17use SimplePie\HTTP\Psr18Client;
18use SimplePie\HTTP\Response;
19
20/**
21 * Used for feed auto-discovery
22 *
23 *
24 * This class can be overloaded with {@see \SimplePie\SimplePie::set_locator_class()}
25 */
26class Locator implements RegistryAware
27{
28 /** @var ?string */
29 public $useragent = null;
30 /** @var int */
31 public $timeout = 10;
32 /** @var File */
33 public $file;
34 /** @var string[] */
35 public $local = [];
36 /** @var string[] */
37 public $elsewhere = [];
38 /** @var array<mixed> */
39 public $cached_entities = [];
40 /** @var string */
41 public $http_base;
42 /** @var string */
43 public $base;
44 /** @var int */
45 public $base_location = 0;
46 /** @var int */
47 public $checked_feeds = 0;
48 /** @var int */
49 public $max_checked_feeds = 10;
50 /** @var bool */
51 public $force_fsockopen = false;
52 /** @var array<int, mixed> */
53 public $curl_options = [];
54 /** @var ?\DomDocument */
55 public $dom;
56 /** @var ?Registry */
57 protected $registry;
58
59 /**
60 * @var Client|null
61 */
62 private $http_client = null;
63
64 /**
65 * @param array<int, mixed> $curl_options
66 */
67 public function __construct(File $file, int $timeout = 10, ?string $useragent = null, int $max_checked_feeds = 10, bool $force_fsockopen = false, array $curl_options = [])
68 {
69 $this->file = $file;
70 $this->useragent = $useragent;
71 $this->timeout = $timeout;
72 $this->max_checked_feeds = $max_checked_feeds;
73 $this->force_fsockopen = $force_fsockopen;
74 $this->curl_options = $curl_options;
75
76 $body = $this->file->get_body_content();
77
78 if (class_exists('DOMDocument') && $body != '') {
79 $this->dom = new \DOMDocument();
80
81 set_error_handler([Misc::class, 'silence_errors']);
82 try {
83 $this->dom->loadHTML($body);
84 } catch (\Throwable $ex) {
85 $this->dom = null;
86 }
87 restore_error_handler();
88 } else {
89 $this->dom = null;
90 }
91 }
92
93 /**
94 * Set a PSR-18 client and PSR-17 factories
95 *
96 * Allows you to use your own HTTP client implementations.
97 */
98 final public function set_http_client(
99 ClientInterface $http_client,
100 RequestFactoryInterface $request_factory,
101 UriFactoryInterface $uri_factory
102 ): void {
103 $this->http_client = new Psr18Client($http_client, $request_factory, $uri_factory);
104 }
105
106 /**
107 * @return void
108 */
109 public function set_registry(\SimplePie\Registry $registry)
110 {
111 $this->registry = $registry;
112 }
113
114 /**
115 * @param SimplePie::LOCATOR_* $type
116 * @param array<Response>|null $working
117 * @return Response|null
118 */
119 public function find(int $type = \SimplePie\SimplePie::LOCATOR_ALL, ?array &$working = null)
120 {
121 assert($this->registry !== null);
122
123 if ($this->is_feed($this->file)) {
124 return $this->file;
125 }
126
127 if (Misc::is_remote_uri($this->file->get_final_requested_uri())) {
128 $sniffer = $this->registry->create(Content\Type\Sniffer::class, [$this->file]);
129 if ($sniffer->get_type() !== 'text/html') {
130 return null;
131 }
132 }
133
134 if ($type & ~\SimplePie\SimplePie::LOCATOR_NONE) {
135 $this->get_base();
136 }
137
138 if ($type & \SimplePie\SimplePie::LOCATOR_AUTODISCOVERY && $working = $this->autodiscovery()) {
139 return $working[0];
140 }
141
142 if ($type & (\SimplePie\SimplePie::LOCATOR_LOCAL_EXTENSION | \SimplePie\SimplePie::LOCATOR_LOCAL_BODY | \SimplePie\SimplePie::LOCATOR_REMOTE_EXTENSION | \SimplePie\SimplePie::LOCATOR_REMOTE_BODY) && $this->get_links()) {
143 if ($type & \SimplePie\SimplePie::LOCATOR_LOCAL_EXTENSION && $working = $this->extension($this->local)) {
144 return $working[0];
145 }
146
147 if ($type & \SimplePie\SimplePie::LOCATOR_LOCAL_BODY && $working = $this->body($this->local)) {
148 return $working[0];
149 }
150
151 if ($type & \SimplePie\SimplePie::LOCATOR_REMOTE_EXTENSION && $working = $this->extension($this->elsewhere)) {
152 return $working[0];
153 }
154
155 if ($type & \SimplePie\SimplePie::LOCATOR_REMOTE_BODY && $working = $this->body($this->elsewhere)) {
156 return $working[0];
157 }
158 }
159 return null;
160 }
161
162 /**
163 * @return bool
164 */
165 public function is_feed(Response $file, bool $check_html = false)
166 {
167 assert($this->registry !== null);
168
169 if (Misc::is_remote_uri($file->get_final_requested_uri())) {
170 $sniffer = $this->registry->create(Content\Type\Sniffer::class, [$file]);
171 $sniffed = $sniffer->get_type();
172 $mime_types = ['application/rss+xml', 'application/rdf+xml',
173 'text/rdf', 'application/atom+xml', 'text/xml',
174 'application/xml', 'application/x-rss+xml'];
175 if ($check_html) {
176 $mime_types[] = 'text/html';
177 }
178
179 return in_array($sniffed, $mime_types);
180 } elseif (is_file($file->get_final_requested_uri())) {
181 return true;
182 } else {
183 return false;
184 }
185 }
186
187 /**
188 * @return void
189 */
190 public function get_base()
191 {
192 assert($this->registry !== null);
193
194 if ($this->dom === null) {
195 throw new \SimplePie\Exception('DOMDocument not found, unable to use locator');
196 }
197 $this->http_base = $this->file->get_final_requested_uri();
198 $this->base = $this->http_base;
199 $elements = $this->dom->getElementsByTagName('base');
200 foreach ($elements as $element) {
201 if ($element->hasAttribute('href')) {
202 $base = $this->registry->call(Misc::class, 'absolutize_url', [trim($element->getAttribute('href')), $this->http_base]);
203 if ($base === false) {
204 continue;
205 }
206 $this->base = $base;
207 $this->base_location = method_exists($element, 'getLineNo') ? $element->getLineNo() : 0;
208 break;
209 }
210 }
211 }
212
213 /**
214 * @return array<Response>|null
215 */
216 public function autodiscovery()
217 {
218 $done = [];
219 $feeds = [];
220 $feeds = array_merge($feeds, $this->search_elements_by_tag('link', $done, $feeds));
221 $feeds = array_merge($feeds, $this->search_elements_by_tag('a', $done, $feeds));
222 $feeds = array_merge($feeds, $this->search_elements_by_tag('area', $done, $feeds));
223
224 if (!empty($feeds)) {
225 return array_values($feeds);
226 }
227
228 return null;
229 }
230
231 /**
232 * @param string[] $done
233 * @param array<string, Response> $feeds
234 * @return array<string, Response>
235 */
236 protected function search_elements_by_tag(string $name, array &$done, array $feeds)
237 {
238 assert($this->registry !== null);
239
240 if ($this->dom === null) {
241 throw new \SimplePie\Exception('DOMDocument not found, unable to use locator');
242 }
243
244 $links = $this->dom->getElementsByTagName($name);
245 foreach ($links as $link) {
246 if ($this->checked_feeds === $this->max_checked_feeds) {
247 break;
248 }
249 if ($link->hasAttribute('href') && $link->hasAttribute('rel')) {
250 $rel = array_unique($this->registry->call(Misc::class, 'space_separated_tokens', [strtolower($link->getAttribute('rel'))]));
251 $line = method_exists($link, 'getLineNo') ? $link->getLineNo() : 1;
252
253 if ($this->base_location < $line) {
254 $href = $this->registry->call(Misc::class, 'absolutize_url', [trim($link->getAttribute('href')), $this->base]);
255 } else {
256 $href = $this->registry->call(Misc::class, 'absolutize_url', [trim($link->getAttribute('href')), $this->http_base]);
257 }
258 if ($href === false) {
259 continue;
260 }
261
262 if (!in_array($href, $done) && in_array('feed', $rel) || (in_array('alternate', $rel) && !in_array('stylesheet', $rel) && $link->hasAttribute('type') && in_array(strtolower($this->registry->call(Misc::class, 'parse_mime', [$link->getAttribute('type')])), ['text/html', 'application/rss+xml', 'application/atom+xml'])) && !isset($feeds[$href])) {
263 $this->checked_feeds++;
264 $headers = [
265 'Accept' => SimplePie::DEFAULT_HTTP_ACCEPT_HEADER,
266 ];
267
268 try {
269 $feed = $this->get_http_client()->request(Client::METHOD_GET, $href, $headers);
270
271 if ((!Misc::is_remote_uri($feed->get_final_requested_uri()) || ($feed->get_status_code() === 200 || $feed->get_status_code() > 206 && $feed->get_status_code() < 300)) && $this->is_feed($feed, true)) {
272 $feeds[$href] = $feed;
273 }
274 } catch (ClientException $th) {
275 // Just mark it as done and continue.
276 }
277 }
278 $done[] = $href;
279 }
280 }
281
282 return $feeds;
283 }
284
285 /**
286 * @return true|null
287 */
288 public function get_links()
289 {
290 assert($this->registry !== null);
291
292 if ($this->dom === null) {
293 throw new \SimplePie\Exception('DOMDocument not found, unable to use locator');
294 }
295
296 $links = $this->dom->getElementsByTagName('a');
297 foreach ($links as $link) {
298 if ($link->hasAttribute('href')) {
299 $href = trim($link->getAttribute('href'));
300 $parsed = $this->registry->call(Misc::class, 'parse_url', [$href]);
301 if ($parsed['scheme'] === '' || preg_match('/^(https?|feed)?$/i', $parsed['scheme'])) {
302 if (method_exists($link, 'getLineNo') && $this->base_location < $link->getLineNo()) {
303 $href = $this->registry->call(Misc::class, 'absolutize_url', [trim($link->getAttribute('href')), $this->base]);
304 } else {
305 $href = $this->registry->call(Misc::class, 'absolutize_url', [trim($link->getAttribute('href')), $this->http_base]);
306 }
307 if ($href === false) {
308 continue;
309 }
310
311 $current = $this->registry->call(Misc::class, 'parse_url', [$this->file->get_final_requested_uri()]);
312
313 if ($parsed['authority'] === '' || $parsed['authority'] === $current['authority']) {
314 $this->local[] = $href;
315 } else {
316 $this->elsewhere[] = $href;
317 }
318 }
319 }
320 }
321 $this->local = array_unique($this->local);
322 $this->elsewhere = array_unique($this->elsewhere);
323 if (!empty($this->local) || !empty($this->elsewhere)) {
324 return true;
325 }
326 return null;
327 }
328
329 /**
330 * Extracts first `link` element with given `rel` attribute inside the `head` element.
331 *
332 * @return string|null
333 */
334 public function get_rel_link(string $rel)
335 {
336 assert($this->registry !== null);
337
338 if ($this->dom === null) {
339 throw new \SimplePie\Exception('DOMDocument not found, unable to use '.
340 'locator');
341 }
342 if (!class_exists('DOMXpath')) {
343 throw new \SimplePie\Exception('DOMXpath not found, unable to use '.
344 'get_rel_link');
345 }
346
347 $xpath = new \DOMXpath($this->dom);
348 $query = '(//head)[1]/link[@rel and @href]';
349 /** @var \DOMNodeList<\DOMElement> */
350 $queryResult = $xpath->query($query);
351 foreach ($queryResult as $link) {
352 $href = trim($link->getAttribute('href'));
353 $parsed = $this->registry->call(Misc::class, 'parse_url', [$href]);
354 if ($parsed['scheme'] === '' ||
355 preg_match('/^https?$/i', $parsed['scheme'])) {
356 if (method_exists($link, 'getLineNo') &&
357 $this->base_location < $link->getLineNo()) {
358 $href = $this->registry->call(
359 Misc::class,
360 'absolutize_url',
361 [trim($link->getAttribute('href')), $this->base]
362 );
363 } else {
364 $href = $this->registry->call(
365 Misc::class,
366 'absolutize_url',
367 [trim($link->getAttribute('href')), $this->http_base]
368 );
369 }
370 if ($href === false) {
371 return null;
372 }
373 $rel_values = explode(' ', strtolower($link->getAttribute('rel')));
374 if (in_array($rel, $rel_values)) {
375 return $href;
376 }
377 }
378 }
379
380 return null;
381 }
382
383 /**
384 * @param string[] $array
385 * @return array<Response>|null
386 */
387 public function extension(array &$array)
388 {
389 foreach ($array as $key => $value) {
390 if ($this->checked_feeds === $this->max_checked_feeds) {
391 break;
392 }
393 $extension = strrchr($value, '.');
394 if ($extension !== false && in_array(strtolower($extension), ['.rss', '.rdf', '.atom', '.xml'])) {
395 $this->checked_feeds++;
396
397 $headers = [
398 'Accept' => SimplePie::DEFAULT_HTTP_ACCEPT_HEADER,
399 ];
400
401 try {
402 $feed = $this->get_http_client()->request(Client::METHOD_GET, $value, $headers);
403
404 if ((!Misc::is_remote_uri($feed->get_final_requested_uri()) || ($feed->get_status_code() === 200 || $feed->get_status_code() > 206 && $feed->get_status_code() < 300)) && $this->is_feed($feed)) {
405 return [$feed];
406 }
407 } catch (ClientException $th) {
408 // Just unset and continue.
409 }
410
411 unset($array[$key]);
412 }
413 }
414 return null;
415 }
416
417 /**
418 * @param string[] $array
419 * @return array<Response>|null
420 */
421 public function body(array &$array)
422 {
423 foreach ($array as $key => $value) {
424 if ($this->checked_feeds === $this->max_checked_feeds) {
425 break;
426 }
427 if (preg_match('/(feed|rss|rdf|atom|xml)/i', $value)) {
428 $this->checked_feeds++;
429 $headers = [
430 'Accept' => SimplePie::DEFAULT_HTTP_ACCEPT_HEADER,
431 ];
432
433 try {
434 $feed = $this->get_http_client()->request(Client::METHOD_GET, $value, $headers);
435
436 if ((!Misc::is_remote_uri($feed->get_final_requested_uri()) || ($feed->get_status_code() === 200 || $feed->get_status_code() > 206 && $feed->get_status_code() < 300)) && $this->is_feed($feed)) {
437 return [$feed];
438 }
439 } catch (ClientException $th) {
440 // Just unset and continue.
441 }
442
443 unset($array[$key]);
444 }
445 }
446 return null;
447 }
448
449 /**
450 * Get a HTTP client
451 */
452 private function get_http_client(): Client
453 {
454 assert($this->registry !== null);
455
456 if ($this->http_client === null) {
457 $options = [
458 'timeout' => $this->timeout,
459 'redirects' => 5,
460 'force_fsockopen' => $this->force_fsockopen,
461 'curl_options' => $this->curl_options,
462 ];
463
464 if ($this->useragent !== null) {
465 $options['useragent'] = $this->useragent;
466 }
467
468 return new FileClient(
469 $this->registry,
470 $options
471 );
472 }
473
474 return $this->http_client;
475 }
476}
477
478class_alias('SimplePie\Locator', 'SimplePie_Locator', false);
479