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 SimplePie\XML\Declaration\Parser as DeclarationParser;
 11use XMLParser;
 12
 13/**
 14 * Parses XML into something sane
 15 *
 16 *
 17 * This class can be overloaded with {@see \SimplePie\SimplePie::set_parser_class()}
 18 */
 19class Parser implements RegistryAware
 20{
 21    /** @var int */
 22    public $error_code;
 23    /** @var string */
 24    public $error_string;
 25    /** @var int */
 26    public $current_line;
 27    /** @var int */
 28    public $current_column;
 29    /** @var int */
 30    public $current_byte;
 31    /** @var string */
 32    public $separator = ' ';
 33    /** @var string[] */
 34    public $namespace = [''];
 35    /** @var string[] */
 36    public $element = [''];
 37    /** @var string[] */
 38    public $xml_base = [''];
 39    /** @var bool[] */
 40    public $xml_base_explicit = [false];
 41    /** @var string[] */
 42    public $xml_lang = [''];
 43    /** @var array<string, mixed> */
 44    public $data = [];
 45    /** @var array<array<string, mixed>> */
 46    public $datas = [[]];
 47    /** @var int */
 48    public $current_xhtml_construct = -1;
 49    /** @var string */
 50    public $encoding;
 51    /** @var Registry */
 52    protected $registry;
 53
 54    /**
 55     * @return void
 56     */
 57    public function set_registry(\SimplePie\Registry $registry)
 58    {
 59        $this->registry = $registry;
 60    }
 61
 62    /**
 63     * @return bool
 64     */
 65    public function parse(string &$data, string $encoding, string $url = '')
 66    {
 67        if (class_exists('DOMXpath') && function_exists('Mf2\parse')) {
 68            $doc = new \DOMDocument();
 69            @$doc->loadHTML($data);
 70            $xpath = new \DOMXpath($doc);
 71            // Check for both h-feed and h-entry, as both a feed with no entries
 72            // and a list of entries without an h-feed wrapper are both valid.
 73            $query = '//*[contains(concat(" ", @class, " "), " h-feed ") or '.
 74                'contains(concat(" ", @class, " "), " h-entry ")]';
 75            /** @var \DOMNodeList<\DOMElement> $result */
 76            $result = $xpath->query($query);
 77            if ($result->length !== 0) {
 78                return $this->parse_microformats($data, $url);
 79            }
 80        }
 81
 82        // Use UTF-8 if we get passed US-ASCII, as every US-ASCII character is a UTF-8 character
 83        if (strtoupper($encoding) === 'US-ASCII') {
 84            $this->encoding = 'UTF-8';
 85        } else {
 86            $this->encoding = $encoding;
 87        }
 88
 89        // Strip BOM:
 90        // UTF-32 Big Endian BOM
 91        if (substr($data, 0, 4) === "\x00\x00\xFE\xFF") {
 92            $data = substr($data, 4);
 93        }
 94        // UTF-32 Little Endian BOM
 95        elseif (substr($data, 0, 4) === "\xFF\xFE\x00\x00") {
 96            $data = substr($data, 4);
 97        }
 98        // UTF-16 Big Endian BOM
 99        elseif (substr($data, 0, 2) === "\xFE\xFF") {
100            $data = substr($data, 2);
101        }
102        // UTF-16 Little Endian BOM
103        elseif (substr($data, 0, 2) === "\xFF\xFE") {
104            $data = substr($data, 2);
105        }
106        // UTF-8 BOM
107        elseif (substr($data, 0, 3) === "\xEF\xBB\xBF") {
108            $data = substr($data, 3);
109        }
110
111        if (substr($data, 0, 5) === '<?xml' && strspn(substr($data, 5, 1), "\x09\x0A\x0D\x20") && ($pos = strpos($data, '?>')) !== false) {
112            $declaration = $this->registry->create(DeclarationParser::class, [substr($data, 5, $pos - 5)]);
113            if ($declaration->parse()) {
114                $data = substr($data, $pos + 2);
115                $data = '<?xml version="' . $declaration->version . '" encoding="' . $encoding . '" standalone="' . (($declaration->standalone) ? 'yes' : 'no') . '"?>' . "\n" .
116                    self::set_doctype($data);
117            } else {
118                $this->error_string = 'SimplePie bug! Please report this!';
119                return false;
120            }
121        } else {
122            $data = self::set_doctype($data);
123        }
124
125        $return = true;
126
127        static $xml_is_sane = null;
128        if ($xml_is_sane === null) {
129            $parser_check = xml_parser_create();
130            xml_parse_into_struct($parser_check, '<foo>&amp;</foo>', $values);
131            if (\PHP_VERSION_ID < 80000) {
132                xml_parser_free($parser_check);
133            }
134            $xml_is_sane = isset($values[0]['value']);
135        }
136
137        // Create the parser
138        if ($xml_is_sane) {
139            $xml = xml_parser_create_ns($this->encoding, $this->separator);
140            xml_parser_set_option($xml, XML_OPTION_SKIP_WHITE, 1);
141            xml_parser_set_option($xml, XML_OPTION_CASE_FOLDING, 0);
142            xml_set_character_data_handler($xml, [$this, 'cdata']);
143            xml_set_element_handler($xml, [$this, 'tag_open'], [$this, 'tag_close']);
144
145            // Parse!
146            $wrapper = @is_writable(sys_get_temp_dir()) ? 'php://temp' : 'php://memory';
147            if (($stream = fopen($wrapper, 'r+')) &&
148                fwrite($stream, $data) &&
149                rewind($stream)) {
150                //Parse by chunks not to use too much memory
151                do {
152                    $stream_data = (string) fread($stream, 1048576);
153
154                    if (!xml_parse($xml, $stream_data, feof($stream))) {
155                        $this->error_code = xml_get_error_code($xml);
156                        $this->error_string = xml_error_string($this->error_code) ?: "Unknown";
157                        $return = false;
158                        break;
159                    }
160                } while (!feof($stream));
161                fclose($stream);
162            } else {
163                $return = false;
164            }
165
166            $this->current_line = xml_get_current_line_number($xml);
167            $this->current_column = xml_get_current_column_number($xml);
168            $this->current_byte = xml_get_current_byte_index($xml);
169            if (\PHP_VERSION_ID < 80000) {
170                xml_parser_free($xml);
171            }
172            return $return;
173        }
174
175        libxml_clear_errors();
176        $xml = new \XMLReader();
177        $xml->xml($data);
178        while (@$xml->read()) {
179            switch ($xml->nodeType) {
180                case \XMLReader::END_ELEMENT:
181                    if ($xml->namespaceURI !== '') {
182                        $tagName = $xml->namespaceURI . $this->separator . $xml->localName;
183                    } else {
184                        $tagName = $xml->localName;
185                    }
186                    $this->tag_close(null, $tagName);
187                    break;
188                case \XMLReader::ELEMENT:
189                    $empty = $xml->isEmptyElement;
190                    if ($xml->namespaceURI !== '') {
191                        $tagName = $xml->namespaceURI . $this->separator . $xml->localName;
192                    } else {
193                        $tagName = $xml->localName;
194                    }
195                    $attributes = [];
196                    while ($xml->moveToNextAttribute()) {
197                        if ($xml->namespaceURI !== '') {
198                            $attrName = $xml->namespaceURI . $this->separator . $xml->localName;
199                        } else {
200                            $attrName = $xml->localName;
201                        }
202                        $attributes[$attrName] = $xml->value;
203                    }
204                    $this->tag_open(null, $tagName, $attributes);
205                    if ($empty) {
206                        $this->tag_close(null, $tagName);
207                    }
208                    break;
209                case \XMLReader::TEXT:
210
211                case \XMLReader::CDATA:
212                    $this->cdata(null, $xml->value);
213                    break;
214            }
215        }
216        if ($error = libxml_get_last_error()) {
217            $this->error_code = $error->code;
218            $this->error_string = $error->message;
219            $this->current_line = $error->line;
220            $this->current_column = $error->column;
221            return false;
222        }
223
224        return true;
225    }
226
227    /**
228     * @return int
229     */
230    public function get_error_code()
231    {
232        return $this->error_code;
233    }
234
235    /**
236     * @return string
237     */
238    public function get_error_string()
239    {
240        return $this->error_string;
241    }
242
243    /**
244     * @return int
245     */
246    public function get_current_line()
247    {
248        return $this->current_line;
249    }
250
251    /**
252     * @return int
253     */
254    public function get_current_column()
255    {
256        return $this->current_column;
257    }
258
259    /**
260     * @return int
261     */
262    public function get_current_byte()
263    {
264        return $this->current_byte;
265    }
266
267    /**
268     * @return array<string, mixed>
269     */
270    public function get_data()
271    {
272        return $this->data;
273    }
274
275    /**
276     * @param XMLParser|resource|null $parser
277     * @param array<string, string> $attributes
278     * @return void
279     */
280    public function tag_open($parser, string $tag, array $attributes)
281    {
282        [$this->namespace[], $this->element[]] = $this->split_ns($tag);
283
284        $attribs = [];
285        foreach ($attributes as $name => $value) {
286            [$attrib_namespace, $attribute] = $this->split_ns($name);
287            $attribs[$attrib_namespace][$attribute] = $value;
288        }
289
290        if (isset($attribs[\SimplePie\SimplePie::NAMESPACE_XML]['base'])) {
291            $base = $this->registry->call(Misc::class, 'absolutize_url', [$attribs[\SimplePie\SimplePie::NAMESPACE_XML]['base'], end($this->xml_base)]);
292            if ($base !== false) {
293                $this->xml_base[] = $base;
294                $this->xml_base_explicit[] = true;
295            }
296        } else {
297            $this->xml_base[] = end($this->xml_base) ?: '';
298            $this->xml_base_explicit[] = end($this->xml_base_explicit);
299        }
300
301        if (isset($attribs[\SimplePie\SimplePie::NAMESPACE_XML]['lang'])) {
302            $this->xml_lang[] = $attribs[\SimplePie\SimplePie::NAMESPACE_XML]['lang'];
303        } else {
304            $this->xml_lang[] = end($this->xml_lang) ?: '';
305        }
306
307        if ($this->current_xhtml_construct >= 0) {
308            $this->current_xhtml_construct++;
309            if (end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_XHTML) {
310                $this->data['data'] .= '<' . end($this->element);
311                if (isset($attribs[''])) {
312                    foreach ($attribs[''] as $name => $value) {
313                        $this->data['data'] .= ' ' . $name . '="' . htmlspecialchars($value, ENT_COMPAT, $this->encoding) . '"';
314                    }
315                }
316                $this->data['data'] .= '>';
317            }
318        } else {
319            $this->datas[] = &$this->data;
320            $this->data = &$this->data['child'][end($this->namespace)][end($this->element)][];
321            $this->data = ['data' => '', 'attribs' => $attribs, 'xml_base' => end($this->xml_base), 'xml_base_explicit' => end($this->xml_base_explicit), 'xml_lang' => end($this->xml_lang)];
322            if ((end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_ATOM_03 && in_array(end($this->element), ['title', 'tagline', 'copyright', 'info', 'summary', 'content']) && isset($attribs['']['mode']) && $attribs['']['mode'] === 'xml')
323            || (end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_ATOM_10 && in_array(end($this->element), ['rights', 'subtitle', 'summary', 'info', 'title', 'content']) && isset($attribs['']['type']) && $attribs['']['type'] === 'xhtml')
324            || (end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_RSS_20 && in_array(end($this->element), ['title']))
325            || (end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_RSS_090 && in_array(end($this->element), ['title']))
326            || (end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_RSS_10 && in_array(end($this->element), ['title']))) {
327                $this->current_xhtml_construct = 0;
328            }
329        }
330    }
331
332    /**
333     * @param XMLParser|resource|null $parser
334     * @return void
335     */
336    public function cdata($parser, string $cdata)
337    {
338        if ($this->current_xhtml_construct >= 0) {
339            $this->data['data'] .= htmlspecialchars($cdata, ENT_QUOTES, $this->encoding);
340        } else {
341            $this->data['data'] .= $cdata;
342        }
343    }
344
345    /**
346     * @param XMLParser|resource|null $parser
347     * @return void
348     */
349    public function tag_close($parser, string $tag)
350    {
351        if ($this->current_xhtml_construct >= 0) {
352            $this->current_xhtml_construct--;
353            if (end($this->namespace) === \SimplePie\SimplePie::NAMESPACE_XHTML && !in_array(end($this->element), ['area', 'base', 'basefont', 'br', 'col', 'frame', 'hr', 'img', 'input', 'isindex', 'link', 'meta', 'param'])) {
354                $this->data['data'] .= '</' . end($this->element) . '>';
355            }
356        }
357        if ($this->current_xhtml_construct === -1) {
358            $this->data = &$this->datas[count($this->datas) - 1];
359            array_pop($this->datas);
360        }
361
362        array_pop($this->element);
363        array_pop($this->namespace);
364        array_pop($this->xml_base);
365        array_pop($this->xml_base_explicit);
366        array_pop($this->xml_lang);
367    }
368
369    /**
370     * @return array{string, string}
371     */
372    public function split_ns(string $string)
373    {
374        static $cache = [];
375        if (!isset($cache[$string])) {
376            if ($pos = strpos($string, $this->separator)) {
377                static $separator_length;
378                if (!$separator_length) {
379                    $separator_length = strlen($this->separator);
380                }
381                $namespace = substr($string, 0, $pos);
382                $local_name = substr($string, $pos + $separator_length);
383                if (strtolower($namespace) === \SimplePie\SimplePie::NAMESPACE_ITUNES) {
384                    $namespace = \SimplePie\SimplePie::NAMESPACE_ITUNES;
385                }
386
387                // Normalize the Media RSS namespaces
388                if ($namespace === \SimplePie\SimplePie::NAMESPACE_MEDIARSS_WRONG ||
389                    $namespace === \SimplePie\SimplePie::NAMESPACE_MEDIARSS_WRONG2 ||
390                    $namespace === \SimplePie\SimplePie::NAMESPACE_MEDIARSS_WRONG3 ||
391                    $namespace === \SimplePie\SimplePie::NAMESPACE_MEDIARSS_WRONG4 ||
392                    $namespace === \SimplePie\SimplePie::NAMESPACE_MEDIARSS_WRONG5) {
393                    $namespace = \SimplePie\SimplePie::NAMESPACE_MEDIARSS;
394                }
395                $cache[$string] = [$namespace, $local_name];
396            } else {
397                $cache[$string] = ['', $string];
398            }
399        }
400        return $cache[$string];
401    }
402
403    /**
404     * @param array<string, mixed> $data
405     */
406    private function parse_hcard(array $data, bool $category = false): string
407    {
408        $name = '';
409        $link = '';
410        // Check if h-card is set and pass that information on in the link.
411        if (isset($data['type']) && in_array('h-card', $data['type'])) {
412            if (isset($data['properties']['name'][0])) {
413                $name = $data['properties']['name'][0];
414            }
415            if (isset($data['properties']['url'][0])) {
416                $link = $data['properties']['url'][0];
417                if ($name === '') {
418                    $name = $link;
419                } else {
420                    // can't have commas in categories.
421                    $name = str_replace(',', '', $name);
422                }
423                $person_tag = $category ? '<span class="person-tag"></span>' : '';
424                return '<a class="h-card" href="'.$link.'">'.$person_tag.$name.'</a>';
425            }
426        }
427        return $data['value'] ?? '';
428    }
429
430    /**
431     * @return true
432     */
433    private function parse_microformats(string &$data, string $url): bool
434    {
435        // For PHPStan, we already check that in call site.
436        \assert(function_exists('Mf2\parse'));
437        \assert(function_exists('Mf2\fetch'));
438        $feed_title = '';
439        $feed_author = null;
440        $author_cache = [];
441        $items = [];
442        $entries = [];
443        $mf = \Mf2\parse($data, $url);
444        // First look for an h-feed.
445        $h_feed = [];
446        foreach ($mf['items'] as $mf_item) {
447            if (in_array('h-feed', $mf_item['type'])) {
448                $h_feed = $mf_item;
449                break;
450            }
451            // Also look for h-feed or h-entry in the children of each top level item.
452            if (!isset($mf_item['children'][0]['type'])) {
453                continue;
454            }
455            if (in_array('h-feed', $mf_item['children'][0]['type'])) {
456                $h_feed = $mf_item['children'][0];
457                // In this case the parent of the h-feed may be an h-card, so use it as
458                // the feed_author.
459                if (in_array('h-card', $mf_item['type'])) {
460                    $feed_author = $mf_item;
461                }
462                break;
463            } elseif (in_array('h-entry', $mf_item['children'][0]['type'])) {
464                $entries = $mf_item['children'];
465                // In this case the parent of the h-entry list may be an h-card, so use
466                // it as the feed_author.
467                if (in_array('h-card', $mf_item['type'])) {
468                    $feed_author = $mf_item;
469                }
470                break;
471            }
472        }
473        if (isset($h_feed['children'])) {
474            $entries = $h_feed['children'];
475            // Also set the feed title and store author from the h-feed if available.
476            if (isset($mf['items'][0]['properties']['name'][0])) {
477                $feed_title = $mf['items'][0]['properties']['name'][0];
478            }
479            if (isset($mf['items'][0]['properties']['author'][0])) {
480                $feed_author = $mf['items'][0]['properties']['author'][0];
481            }
482        } elseif (count($entries) === 0) {
483            $entries = $mf['items'];
484        }
485        for ($i = 0; $i < count($entries); $i++) {
486            $entry = $entries[$i];
487            if (in_array('h-entry', $entry['type'])) {
488                $item = [];
489                $title = '';
490                $description = '';
491                if (isset($entry['properties']['url'][0])) {
492                    $link = $entry['properties']['url'][0];
493                    if (isset($link['value'])) {
494                        $link = $link['value'];
495                    }
496                    $item['link'] = [['data' => $link]];
497                }
498                if (isset($entry['properties']['uid'][0])) {
499                    $guid = $entry['properties']['uid'][0];
500                    if (isset($guid['value'])) {
501                        $guid = $guid['value'];
502                    }
503                    $item['guid'] = [['data' => $guid]];
504                }
505                if (isset($entry['properties']['name'][0])) {
506                    $title = $entry['properties']['name'][0];
507                    if (isset($title['value'])) {
508                        $title = $title['value'];
509                    }
510                    $item['title'] = [['data' => $title]];
511                }
512                if (isset($entry['properties']['author'][0]) || isset($feed_author)) {
513                    // author is a special case, it can be plain text or an h-card array.
514                    // If it's plain text it can also be a url that should be followed to
515                    // get the actual h-card.
516                    $author = $entry['properties']['author'][0] ?? $feed_author;
517                    if (!is_string($author)) {
518                        $author = $this->parse_hcard($author);
519                    } elseif (strpos($author, 'http') === 0) {
520                        if (isset($author_cache[$author])) {
521                            $author = $author_cache[$author];
522                        } else {
523                            if ($mf = \Mf2\fetch($author)) {
524                                foreach ($mf['items'] as $hcard) {
525                                    // Only interested in an h-card by itself in this case.
526                                    if (!in_array('h-card', $hcard['type'])) {
527                                        continue;
528                                    }
529                                    // It must have a url property matching what we fetched.
530                                    if (!isset($hcard['properties']['url']) ||
531                                            !(in_array($author, $hcard['properties']['url']))) {
532                                        continue;
533                                    }
534                                    // Save parse_hcard the trouble of finding the correct url.
535                                    $hcard['properties']['url'][0] = $author;
536                                    // Cache this h-card for the next h-entry to check.
537                                    $author_cache[$author] = $this->parse_hcard($hcard);
538                                    $author = $author_cache[$author];
539                                    break;
540                                }
541                            }
542                        }
543                    }
544                    $item['author'] = [['data' => $author]];
545                }
546                if (isset($entry['properties']['photo'][0])) {
547                    // If a photo is also in content, don't need to add it again here.
548                    $content = '';
549                    if (isset($entry['properties']['content'][0]['html'])) {
550                        $content = $entry['properties']['content'][0]['html'];
551                    }
552                    $photo_list = [];
553                    for ($j = 0; $j < count($entry['properties']['photo']); $j++) {
554                        $photo = $entry['properties']['photo'][$j];
555                        if (!empty($photo) && strpos($content, $photo) === false) {
556                            $photo_list[] = $photo;
557                        }
558                    }
559                    // When there's more than one photo show the first and use a lightbox.
560                    // Need a permanent, unique name for the image set, but don't have
561                    // anything unique except for the content itself, so use that.
562                    $count = count($photo_list);
563                    if ($count > 1) {
564                        $image_set_id = preg_replace('/[[:^alnum:]]/', '', $photo_list[0]);
565                        $description = '<p>';
566                        for ($j = 0; $j < $count; $j++) {
567                            $hidden = $j === 0 ? '' : 'class="hidden" ';
568                            $description .= '<a href="'.$photo_list[$j].'" '.$hidden.
569                                'data-lightbox="image-set-'.$image_set_id.'">'.
570                                '<img src="'.$photo_list[$j].'"></a>';
571                        }
572                        $description .= '<br><b>'.$count.' photos</b></p>';
573                    } elseif ($count == 1) {
574                        $description = '<p><img src="'.$photo_list[0].'"></p>';
575                    }
576                }
577                if (isset($entry['properties']['content'][0]['html'])) {
578                    // e-content['value'] is the same as p-name when they are on the same
579                    // element. Use this to replace title with a strip_tags version so
580                    // that alt text from images is not included in the title.
581                    if ($entry['properties']['content'][0]['value'] === $title) {
582                        $title = strip_tags($entry['properties']['content'][0]['html']);
583                        $item['title'] = [['data' => $title]];
584                    }
585                    $description .= $entry['properties']['content'][0]['html'];
586                    if (isset($entry['properties']['in-reply-to'][0])) {
587                        $in_reply_to = '';
588                        if (is_string($entry['properties']['in-reply-to'][0])) {
589                            $in_reply_to = $entry['properties']['in-reply-to'][0];
590                        } elseif (isset($entry['properties']['in-reply-to'][0]['value'])) {
591                            $in_reply_to = $entry['properties']['in-reply-to'][0]['value'];
592                        }
593                        if ($in_reply_to !== '') {
594                            $description .= '<p><span class="in-reply-to"></span> '.
595                                '<a href="'.$in_reply_to.'">'.$in_reply_to.'</a><p>';
596                        }
597                    }
598                    $item['description'] = [['data' => $description]];
599                }
600                if (isset($entry['properties']['category'])) {
601                    $category_csv = '';
602                    // Categories can also contain h-cards.
603                    foreach ($entry['properties']['category'] as $category) {
604                        if ($category_csv !== '') {
605                            $category_csv .= ', ';
606                        }
607                        if (is_string($category)) {
608                            // Can't have commas in categories.
609                            $category_csv .= str_replace(',', '', $category);
610                        } else {
611                            $category_csv .= $this->parse_hcard($category, true);
612                        }
613                    }
614                    $item['category'] = [['data' => $category_csv]];
615                }
616                if (isset($entry['properties']['published'][0])) {
617                    $timestamp = strtotime($entry['properties']['published'][0]);
618                    $pub_date = date('F j Y g:ia', $timestamp).' GMT';
619                    $item['pubDate'] = [['data' => $pub_date]];
620                }
621                // The title and description are set to the empty string to represent
622                // a deleted item (which also makes it an invalid rss item).
623                if (isset($entry['properties']['deleted'][0])) {
624                    $item['title'] = [['data' => '']];
625                    $item['description'] = [['data' => '']];
626                }
627                $items[] = ['child' => ['' => $item]];
628            }
629        }
630        // Mimic RSS data format when storing microformats.
631        $link = [['data' => $url]];
632        $image = '';
633        if (!is_string($feed_author) &&
634                isset($feed_author['properties']['photo'][0])) {
635            $image = [['child' => ['' => ['url' =>
636                [['data' => $feed_author['properties']['photo'][0]]]]]]];
637        }
638        // Use the name given for the h-feed, or get the title from the html.
639        if ($feed_title !== '') {
640            $feed_title = [['data' => htmlspecialchars($feed_title)]];
641        } elseif ($position = strpos($data, '<title>')) {
642            $start = $position < 200 ? 0 : $position - 200;
643            $check = substr($data, $start, 400);
644            $matches = [];
645            if (preg_match('/<title>(.+)<\/title>/', $check, $matches)) {
646                $feed_title = [['data' => htmlspecialchars($matches[1])]];
647            }
648        }
649        $channel = ['channel' => [['child' => ['' =>
650            ['link' => $link, 'image' => $image, 'title' => $feed_title,
651                  'item' => $items]]]]];
652        $rss = [['attribs' => ['' => ['version' => '2.0']],
653                           'child' => ['' => $channel]]];
654        $this->data = ['child' => ['' => ['rss' => $rss]]];
655        return true;
656    }
657
658    private static function set_doctype(string $data): string
659    {
660        // Strip DOCTYPE except if containing an [internal subset]
661        $data = preg_replace('/^\\s*<!DOCTYPE\\s[^>\\[\\]]*>\s*/', '', $data) ?? $data;
662        // Declare HTML entities only if no remaining DOCTYPE
663        $doctype = preg_match('/^\\s*<!DOCTYPE\\s/', $data) ? '' : self::declare_html_entities();
664        return $doctype . $data;
665    }
666
667    private static function declare_html_entities(): string
668    {
669        // This is required because the RSS specification says that entity-encoded
670        // html is allowed, but the xml specification says they must be declared.
671        return '<!DOCTYPE rss [ <!ENTITY nbsp "&#x00A0;"> <!ENTITY iexcl "&#x00A1;"> <!ENTITY cent "&#x00A2;"> <!ENTITY pound "&#x00A3;"> <!ENTITY curren "&#x00A4;"> <!ENTITY yen "&#x00A5;"> <!ENTITY brvbar "&#x00A6;"> <!ENTITY sect "&#x00A7;"> <!ENTITY uml "&#x00A8;"> <!ENTITY copy "&#x00A9;"> <!ENTITY ordf "&#x00AA;"> <!ENTITY laquo "&#x00AB;"> <!ENTITY not "&#x00AC;"> <!ENTITY shy "&#x00AD;"> <!ENTITY reg "&#x00AE;"> <!ENTITY macr "&#x00AF;"> <!ENTITY deg "&#x00B0;"> <!ENTITY plusmn "&#x00B1;"> <!ENTITY sup2 "&#x00B2;"> <!ENTITY sup3 "&#x00B3;"> <!ENTITY acute "&#x00B4;"> <!ENTITY micro "&#x00B5;"> <!ENTITY para "&#x00B6;"> <!ENTITY middot "&#x00B7;"> <!ENTITY cedil "&#x00B8;"> <!ENTITY sup1 "&#x00B9;"> <!ENTITY ordm "&#x00BA;"> <!ENTITY raquo "&#x00BB;"> <!ENTITY frac14 "&#x00BC;"> <!ENTITY frac12 "&#x00BD;"> <!ENTITY frac34 "&#x00BE;"> <!ENTITY iquest "&#x00BF;"> <!ENTITY Agrave "&#x00C0;"> <!ENTITY Aacute "&#x00C1;"> <!ENTITY Acirc "&#x00C2;"> <!ENTITY Atilde "&#x00C3;"> <!ENTITY Auml "&#x00C4;"> <!ENTITY Aring "&#x00C5;"> <!ENTITY AElig "&#x00C6;"> <!ENTITY Ccedil "&#x00C7;"> <!ENTITY Egrave "&#x00C8;"> <!ENTITY Eacute "&#x00C9;"> <!ENTITY Ecirc "&#x00CA;"> <!ENTITY Euml "&#x00CB;"> <!ENTITY Igrave "&#x00CC;"> <!ENTITY Iacute "&#x00CD;"> <!ENTITY Icirc "&#x00CE;"> <!ENTITY Iuml "&#x00CF;"> <!ENTITY ETH "&#x00D0;"> <!ENTITY Ntilde "&#x00D1;"> <!ENTITY Ograve "&#x00D2;"> <!ENTITY Oacute "&#x00D3;"> <!ENTITY Ocirc "&#x00D4;"> <!ENTITY Otilde "&#x00D5;"> <!ENTITY Ouml "&#x00D6;"> <!ENTITY times "&#x00D7;"> <!ENTITY Oslash "&#x00D8;"> <!ENTITY Ugrave "&#x00D9;"> <!ENTITY Uacute "&#x00DA;"> <!ENTITY Ucirc "&#x00DB;"> <!ENTITY Uuml "&#x00DC;"> <!ENTITY Yacute "&#x00DD;"> <!ENTITY THORN "&#x00DE;"> <!ENTITY szlig "&#x00DF;"> <!ENTITY agrave "&#x00E0;"> <!ENTITY aacute "&#x00E1;"> <!ENTITY acirc "&#x00E2;"> <!ENTITY atilde "&#x00E3;"> <!ENTITY auml "&#x00E4;"> <!ENTITY aring "&#x00E5;"> <!ENTITY aelig "&#x00E6;"> <!ENTITY ccedil "&#x00E7;"> <!ENTITY egrave "&#x00E8;"> <!ENTITY eacute "&#x00E9;"> <!ENTITY ecirc "&#x00EA;"> <!ENTITY euml "&#x00EB;"> <!ENTITY igrave "&#x00EC;"> <!ENTITY iacute "&#x00ED;"> <!ENTITY icirc "&#x00EE;"> <!ENTITY iuml "&#x00EF;"> <!ENTITY eth "&#x00F0;"> <!ENTITY ntilde "&#x00F1;"> <!ENTITY ograve "&#x00F2;"> <!ENTITY oacute "&#x00F3;"> <!ENTITY ocirc "&#x00F4;"> <!ENTITY otilde "&#x00F5;"> <!ENTITY ouml "&#x00F6;"> <!ENTITY divide "&#x00F7;"> <!ENTITY oslash "&#x00F8;"> <!ENTITY ugrave "&#x00F9;"> <!ENTITY uacute "&#x00FA;"> <!ENTITY ucirc "&#x00FB;"> <!ENTITY uuml "&#x00FC;"> <!ENTITY yacute "&#x00FD;"> <!ENTITY thorn "&#x00FE;"> <!ENTITY yuml "&#x00FF;"> <!ENTITY OElig "&#x0152;"> <!ENTITY oelig "&#x0153;"> <!ENTITY Scaron "&#x0160;"> <!ENTITY scaron "&#x0161;"> <!ENTITY Yuml "&#x0178;"> <!ENTITY fnof "&#x0192;"> <!ENTITY circ "&#x02C6;"> <!ENTITY tilde "&#x02DC;"> <!ENTITY Alpha "&#x0391;"> <!ENTITY Beta "&#x0392;"> <!ENTITY Gamma "&#x0393;"> <!ENTITY Epsilon "&#x0395;"> <!ENTITY Zeta "&#x0396;"> <!ENTITY Eta "&#x0397;"> <!ENTITY Theta "&#x0398;"> <!ENTITY Iota "&#x0399;"> <!ENTITY Kappa "&#x039A;"> <!ENTITY Lambda "&#x039B;"> <!ENTITY Mu "&#x039C;"> <!ENTITY Nu "&#x039D;"> <!ENTITY Xi "&#x039E;"> <!ENTITY Omicron "&#x039F;"> <!ENTITY Pi "&#x03A0;"> <!ENTITY Rho "&#x03A1;"> <!ENTITY Sigma "&#x03A3;"> <!ENTITY Tau "&#x03A4;"> <!ENTITY Upsilon "&#x03A5;"> <!ENTITY Phi "&#x03A6;"> <!ENTITY Chi "&#x03A7;"> <!ENTITY Psi "&#x03A8;"> <!ENTITY Omega "&#x03A9;"> <!ENTITY alpha "&#x03B1;"> <!ENTITY beta "&#x03B2;"> <!ENTITY gamma "&#x03B3;"> <!ENTITY delta "&#x03B4;"> <!ENTITY epsilon "&#x03B5;"> <!ENTITY zeta "&#x03B6;"> <!ENTITY eta "&#x03B7;"> <!ENTITY theta "&#x03B8;"> <!ENTITY iota "&#x03B9;"> <!ENTITY kappa "&#x03BA;"> <!ENTITY lambda "&#x03BB;"> <!ENTITY mu "&#x03BC;"> <!ENTITY nu "&#x03BD;"> <!ENTITY xi "&#x03BE;"> <!ENTITY omicron "&#x03BF;"> <!ENTITY pi "&#x03C0;"> <!ENTITY rho "&#x03C1;"> <!ENTITY sigmaf "&#x03C2;"> <!ENTITY sigma "&#x03C3;"> <!ENTITY tau "&#x03C4;"> <!ENTITY upsilon "&#x03C5;"> <!ENTITY phi "&#x03C6;"> <!ENTITY chi "&#x03C7;"> <!ENTITY psi "&#x03C8;"> <!ENTITY omega "&#x03C9;"> <!ENTITY thetasym "&#x03D1;"> <!ENTITY upsih "&#x03D2;"> <!ENTITY piv "&#x03D6;"> <!ENTITY ensp "&#x2002;"> <!ENTITY emsp "&#x2003;"> <!ENTITY thinsp "&#x2009;"> <!ENTITY zwnj "&#x200C;"> <!ENTITY zwj "&#x200D;"> <!ENTITY lrm "&#x200E;"> <!ENTITY rlm "&#x200F;"> <!ENTITY ndash "&#x2013;"> <!ENTITY mdash "&#x2014;"> <!ENTITY lsquo "&#x2018;"> <!ENTITY rsquo "&#x2019;"> <!ENTITY sbquo "&#x201A;"> <!ENTITY ldquo "&#x201C;"> <!ENTITY rdquo "&#x201D;"> <!ENTITY bdquo "&#x201E;"> <!ENTITY dagger "&#x2020;"> <!ENTITY Dagger "&#x2021;"> <!ENTITY bull "&#x2022;"> <!ENTITY hellip "&#x2026;"> <!ENTITY permil "&#x2030;"> <!ENTITY prime "&#x2032;"> <!ENTITY Prime "&#x2033;"> <!ENTITY lsaquo "&#x2039;"> <!ENTITY rsaquo "&#x203A;"> <!ENTITY oline "&#x203E;"> <!ENTITY frasl "&#x2044;"> <!ENTITY euro "&#x20AC;"> <!ENTITY image "&#x2111;"> <!ENTITY weierp "&#x2118;"> <!ENTITY real "&#x211C;"> <!ENTITY trade "&#x2122;"> <!ENTITY alefsym "&#x2135;"> <!ENTITY larr "&#x2190;"> <!ENTITY uarr "&#x2191;"> <!ENTITY rarr "&#x2192;"> <!ENTITY darr "&#x2193;"> <!ENTITY harr "&#x2194;"> <!ENTITY crarr "&#x21B5;"> <!ENTITY lArr "&#x21D0;"> <!ENTITY uArr "&#x21D1;"> <!ENTITY rArr "&#x21D2;"> <!ENTITY dArr "&#x21D3;"> <!ENTITY hArr "&#x21D4;"> <!ENTITY forall "&#x2200;"> <!ENTITY part "&#x2202;"> <!ENTITY exist "&#x2203;"> <!ENTITY empty "&#x2205;"> <!ENTITY nabla "&#x2207;"> <!ENTITY isin "&#x2208;"> <!ENTITY notin "&#x2209;"> <!ENTITY ni "&#x220B;"> <!ENTITY prod "&#x220F;"> <!ENTITY sum "&#x2211;"> <!ENTITY minus "&#x2212;"> <!ENTITY lowast "&#x2217;"> <!ENTITY radic "&#x221A;"> <!ENTITY prop "&#x221D;"> <!ENTITY infin "&#x221E;"> <!ENTITY ang "&#x2220;"> <!ENTITY and "&#x2227;"> <!ENTITY or "&#x2228;"> <!ENTITY cap "&#x2229;"> <!ENTITY cup "&#x222A;"> <!ENTITY int "&#x222B;"> <!ENTITY there4 "&#x2234;"> <!ENTITY sim "&#x223C;"> <!ENTITY cong "&#x2245;"> <!ENTITY asymp "&#x2248;"> <!ENTITY ne "&#x2260;"> <!ENTITY equiv "&#x2261;"> <!ENTITY le "&#x2264;"> <!ENTITY ge "&#x2265;"> <!ENTITY sub "&#x2282;"> <!ENTITY sup "&#x2283;"> <!ENTITY nsub "&#x2284;"> <!ENTITY sube "&#x2286;"> <!ENTITY supe "&#x2287;"> <!ENTITY oplus "&#x2295;"> <!ENTITY otimes "&#x2297;"> <!ENTITY perp "&#x22A5;"> <!ENTITY sdot "&#x22C5;"> <!ENTITY lceil "&#x2308;"> <!ENTITY rceil "&#x2309;"> <!ENTITY lfloor "&#x230A;"> <!ENTITY rfloor "&#x230B;"> <!ENTITY lang "&#x2329;"> <!ENTITY rang "&#x232A;"> <!ENTITY loz "&#x25CA;"> <!ENTITY spades "&#x2660;"> <!ENTITY clubs "&#x2663;"> <!ENTITY hearts "&#x2665;"> <!ENTITY diams "&#x2666;"> ]>';
672    }
673}
674
675class_alias('SimplePie\Parser', 'SimplePie_Parser');
676