net_nemein_rss_fetch::raw_fetch() - Code Metrics - Inspection of "Make sure publish date gets set correctly in creat..." - flack/openpsa - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 33e64a...a4094b )

by Andreas

created 2023-04-17 09:20 UTC

net_nemein_rss_fetch::raw_fetch() A

↳ Parent: net_nemein_rss_fetch

Complexity

Conditions	1
Paths	1

Size

Total Lines	6
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	0
CRAP Score	2

Importance

Changes

Metric	Value
cc	1
eloc	4
nc	1
nop	1
dl	0
loc	6
ccs	0
cts	5
cp	0
crap	2
rs	10
c	0
b	0
f	0

<?php
/**
 * @package net.nemein.rss
 * @author The Midgard Project, http://www.midgard-project.org
 * @copyright The Midgard Project, http://www.midgard-project.org
 * @license http://www.gnu.org/licenses/lgpl.html GNU Lesser General Public License
 */

use Symfony\Component\DomCrawler\Crawler;
use midcom\dba\softdelete;
use SimplePie\SimplePie;

/**
 * RSS and Atom feed fetching class. Caches the fetched items as articles
 * in net.nehmer.blog
 *
 * @package net.nemein.rss
 */
class net_nemein_rss_fetch
{
    /**
     * The last error reported by SimplePie, if any
     */
    public $lasterror;

    private net_nemein_rss_feed_dba $_feed;

    /**
     * Property of midcom_db_article we're using for storing the feed item GUIDs
     */
    private string $_guid_property = 'extra2';

    /**
     * Current node we're importing to
     */
    private midcom_db_topic $_node;

    /**
     * Initializes the class with a given feed
     */
    public function __construct(net_nemein_rss_feed_dba $feed)
    {
        $this->_feed = $feed;
        $this->_node = new midcom_db_topic($feed->node);
    }

    public static function get_parser() : SimplePie
    {
        $parser = new SimplePie;
        $parser->get_registry()->register('Item', net_nemein_rss_parser_item::class);
        $parser->set_output_encoding(midcom::get()->i18n->get_current_charset());
        $parser->set_cache_location(midcom::get()->config->get('midcom_tempdir'));
        return $parser;
    }

    /**
     * Actually fetch a feed
     */
    public static function raw_fetch(string $url) : SimplePie
    {
        $parser = self::get_parser();
        $parser->set_feed_url($url);
        $parser->init();
        return $parser;
    }

    /**
     * Fetch given RSS or Atom feed
     *
     * @return net_nemein_rss_parser_item[] Array of normalized feed items
     */
    function fetch() : array
    {
        $parser = self::raw_fetch($this->_feed->url);
        if ($parser->error()) {
            $this->lasterror = $parser->error();
            return [];
        }
        if (!empty($parser->data['headers']['etag'])) {
            // Etag checking
            $etag = trim($parser->data['headers']['etag']);

            $feed_etag = $this->_feed->get_parameter('net.nemein.rss', 'etag');
            if (   !empty($feed_etag)
                && $feed_etag == $etag) {
                // Feed hasn't changed, skip updating
                debug_add("Feed {$this->_feed->url} has not changed since " . date('c', $this->_feed->latestfetch), MIDCOM_LOG_WARN);
                return [];
            }

            $this->_feed->set_parameter('net.nemein.rss', 'etag', $etag);
        }

        $this->_feed->latestfetch = time();
        $this->_feed->_use_rcs = false;
        $this->_feed->update();

        return $parser->get_items();
    }

    /**
     * Fetches and imports items in the feed
     */
    public function import() : array
    {
        if (!$this->_node->component) {
            return [];
        }

        $items = $this->fetch();

        if (empty($items)) {
            // This feed didn't return any items, skip
            return [];
        }

        // Reverse items so that creation times remain in correct order even for feeds without timestamps
        $items = array_reverse($items);

        foreach ($items as $item) {
            if ($guid = $this->import_item($item)) {
                $item->set_local_guid($guid);
                debug_add("Imported item " . $item->get_id() . ' as ' . $guid, MIDCOM_LOG_INFO);
            } else {
                debug_add("Failed to import item " . $item->get_id() . ': ' . midcom_connection::get_error_string(), MIDCOM_LOG_ERROR);
            }
        }

        $this->clean($items);

        return array_reverse($items);
    }

    /**
     * Imports a feed item into the database
     */
    public function import_item(net_nemein_rss_parser_item $item) : ?string
    {
        if ($this->_node->component !== 'net.nehmer.blog') {
            throw new midcom_error("RSS fetching for component {$this->_node->component} is unsupported");
        }
        return $this->import_article($item);
    }

    /**
     * Imports an item as a news article
     */
    private function import_article(net_nemein_rss_parser_item $item) : ?string
    {
        $guid = $item->get_id();
        $title = $item->get_title();

        if (   (   empty($title)
                || trim($title) == '...')
            && empty($guid)) {
            // Something wrong with this entry, skip it
            return null;
        }

        $article = $this->find_article($item, $guid);
        if (!$article) {
            return null;
        }

        $article->allow_name_catenate = true;
        $article->set_rcs_message(sprintf(midcom::get()->i18n->get_string('%s was imported from %s', 'net.nemein.rss'), $title, $this->_feed->title));

        $values = [
            'title' => $title,
            $this->_guid_property => $guid, // FIXME: This breaks with URLs longer than 255 chars
            'content' => $item->get_content(),
            'url' => $item->get_link(),
            'extra1' => '|feed:' . md5($this->_feed->url) . '|',
        ];
        $meta_values = [];

        // Safety, make sure we have sane name (the allow_catenate was set earlier, so this will not clash
        if (empty($article->name)) {
            $values['name'] = midcom_helper_misc::urlize($title);
        }

        $categories = $item->get_categories();
        if (is_array($categories)) {
            // Handle categories provided in the feed
            foreach ($categories as $category) {
                // Clean up the categories and save
                $category = str_replace('|', '_', trim($category->get_term()));
                $values['extra1'] .= "{$category}|";
            }
        }

        $article_author = $this->find_author($item);
        if (!empty($article_author->guid)) {
            $meta_values['authors'] = "|{$article_author->guid}|";
        }

        // Try to figure out item publication date
        if ($article_date = $item->get_date('U')) {
            $meta_values['published'] = $article_date;
        } else {
            $article_date = time();
            if (!$article->id) {
                $meta_values['published'] = $article_date;
            }
        }

        if ($article_date > $this->_feed->latestupdate) {
            // Cache "latest updated" time to feed
            $this->_feed->latestupdate = $article_date;
            $this->_feed->_use_rcs = false;
            $this->_feed->update();
        }

        if ($article->id) {
            if (   $this->apply_values($article, $values, $meta_values)
                && !$article->update()) {
                return null;
            }
        } else {
            $this->apply_values($article, $values, $meta_values);
            if (!$article->create()) {
                return null;
            }
        }

        if ($this->_feed->autoapprove) {
            $article->metadata->approve();
        }

        $this->_parse_tags($article);
        $this->_parse_parameters($article, $item);

        // store <link rel="replies"> url in parameter
        if ($item->get_link(0, 'replies')) {
            $article->set_parameter('net.nemein.rss', 'replies_url', $item->get_link(0, 'replies'));
        }

        return $article->guid;
    }

    private function find_author(net_nemein_rss_parser_item $item) : midcom_db_person
    {
        // Try to figure out item author
        if (   $this->_feed->forceauthor
            && $this->_feed->defaultauthor) {
            // Feed has a "default author" set, use it
            return new midcom_db_person($this->_feed->defaultauthor);
        }
        $author = $this->match_item_author($item);
        $fallback_person_id = 1;
        if (   !$author
            || $author->id == $fallback_person_id) {
            if ($this->_feed->defaultauthor) {
                // Feed has a "default author" set, use it
                $author = new midcom_db_person($this->_feed->defaultauthor);
            } else {
                // Fall back to "Midgard Admin" just in case
                $author = new midcom_db_person($fallback_person_id);
            }
        }
        return $author;
    }

    private function find_article(net_nemein_rss_parser_item $item, string $guid) : ?midcom_db_article
    {
        $qb = midcom_db_article::new_query_builder();
        $qb->add_constraint('topic', '=', $this->_feed->node);
        $qb->add_constraint($this->_guid_property, '=', substr($guid, 0, 255));
        $articles = $qb->execute();
        if (!empty($articles)) {
            // This item has been imported already earlier. Update
            return $articles[0];
        }

        // Check against duplicate hits that may come from different feeds
        if ($link = $item->get_link()) {
            $qb = midcom_db_article::new_query_builder();
            $qb->add_constraint('topic', '=', $this->_feed->node);
            $qb->add_constraint('url', '=', $link);
            if ($qb->count() > 0) {
                // Dupe, skip
                return null;
            }
        }

        // This is a new item
        $article = new midcom_db_article();
        $article->topic = $this->_feed->node;
        return $article;
    }

    private function apply_values(midcom_db_article $article, array $values, array $meta_values) : bool
    {
        $updated = false;

        foreach ($values as $fieldname => $value) {
            if ($article->$fieldname !== $value) {
                $article->$fieldname = $value;
                $updated = true;
            }
        }

        foreach ($meta_values as $fieldname => $value) {
            if ($article->metadata->$fieldname !== $value) {
                $article->metadata->$fieldname = $value;
                $updated = true;
            }
        }

        return $updated;
    }

    /**
     * Cleans up old, removed items from feeds
     *
     * @param net_nemein_rss_parser_item[] $items
     */
    private function clean(array $items)
    {
        if ($this->_feed->keepremoved) {
            // This feed is set up so that we retain items removed from array
            return;
        }

        // Create array of item GUIDs
        $item_guids = [];
        foreach ($items as $item) {
            $item_guids[] = $item->get_id();
        }

        // Find articles resulting from this feed
        $qb = midcom_db_article::new_query_builder();
        $feed_category = md5($this->_feed->url);
        $qb->add_constraint('extra1', 'LIKE', "%|feed:{$feed_category}|%");
        $qb->add_constraint($this->_guid_property, 'NOT IN', $item_guids);
        $local_items = $qb->execute_unchecked();
        $purge_guids = [];
        foreach ($local_items as $item) {
            $purge_guids[] = $item->guid;
            $item->delete();
        }

        softdelete::purge($purge_guids, 'midgard_article');
    }

    /**
     * Parses author formats used by different feed standards and
     * and returns the information
     */
    public static function parse_item_author(net_nemein_rss_parser_item $item) : array
    {
        $author_info = [];

        // First try dig up any information about the author possible
        if ($author = $item->get_author()) {
            $name = $author->get_name();
            $email = $author->get_email();
            if (!empty($name)) {
                $name = html_entity_decode($name, ENT_QUOTES, midcom::get()->i18n->get_current_charset());
                // Atom feed, the value can be either full name or username
                $author_info['user_or_full'] = $name;
            } else {
                $name = html_entity_decode($email, ENT_QUOTES, midcom::get()->i18n->get_current_charset());
            }

            if (!preg_match('/[<\(]/', $name)) {
                $author_info['user_or_full'] = $name;
            } else {
                if (str_contains($name, '<')) {
                    // The classic "Full Name <email>" format
                    $regex = '/(?<fullname>.+) <?(?<email>[a-zA-Z0-9_.-]+?@[a-zA-Z0-9_.-]+)>?[ ,]?/';
                } else {
                    // The classic "email (Full Name)" format
                    $regex = '/^(?<email>[a-zA-Z0-9_.-]+?@[a-zA-Z0-9_.-]+) \((?<fullname>.+)\)$/';
                }
                if (preg_match($regex, $name, $matches)) {
                    $author_info['email'] = $matches['email'];
                    $author_info['user_or_full'] = $matches['fullname'];
                }
            }
        }

        if (isset($author_info['user_or_full'])) {
            $author_info['user_or_full'] = trim($author_info['user_or_full']);
            if (str_contains($author_info['user_or_full'], ' ')) {
                // This value has a space in it, assuming full name
                $author_info['full_name'] = $author_info['user_or_full'];
            } else {
                $author_info['username'] = $author_info['user_or_full'];
            }
            unset($author_info['user_or_full']);
        }

        return $author_info;
    }

    /**
     * Parses author formats used by different feed standards and
     * tries to match to persons in database.
     */
    public function match_item_author(net_nemein_rss_parser_item $item) : ?midcom_db_person
    {
        // Parse the item for author information
        $author_info = self::parse_item_author($item);

        if (!empty($author_info['email'])) {
            // Email is a pretty good identifier, start with it
            $person_qb = midcom_db_person::new_query_builder();
            $person_qb->add_constraint('email', '=', $author_info['email']);
            $persons = $person_qb->execute();
            if (!empty($persons)) {
                return $persons[0];
            }
        }

        if (   !empty($author_info['username'])
            && $person = midcom::get()->auth->get_user_by_name($author_info['username'])) {
            return $person->get_storage();
        }

        if (!empty($author_info['full_name'])) {
            $name_parts = explode(' ', $author_info['full_name']);
            if (count($name_parts) > 1) {
                // We assume the western format Firstname Lastname
                $firstname = $name_parts[0];
                $lastname = $name_parts[1];

                $person_qb = midcom_db_person::new_query_builder();
                $person_qb->add_constraint('firstname', '=', $firstname);
                $person_qb->add_constraint('lastname', '=', $lastname);
                $persons = $person_qb->execute();
                if (!empty($persons)) {
                    return $persons[0];
                }
            }
        }

        return null;
    }

    /**
     * Parses additional metadata in RSS item and sets parameters accordingly
     */
    private function _parse_parameters(midcom_db_article $article, net_nemein_rss_parser_item $item)
    {
        foreach ($item->get_enclosures() as $enclosure) {
            $article->set_parameter('net.nemein.rss:enclosure', 'url', $enclosure->get_link());
            $article->set_parameter('net.nemein.rss:enclosure', 'duration', $enclosure->get_duration());
            $article->set_parameter('net.nemein.rss:enclosure', 'mimetype', $enclosure->get_type());
        }
    }

    /**
     * Parses rel-tag links in article content and tags the object based on them
     */
    private function _parse_tags(midcom_db_article $article)
    {
        $crawler = new Crawler($article->content);
        $nodes = $crawler->filter('a[rel="tag"]');

        $html_tags = $nodes->each(function(Crawler $node) {
            return [
                'href' => $node->attr('href') ?? false,
                'value' => $node->text() ?? false,
            ];
        });

        $tags = [];

        foreach ($html_tags as $html_tag) {
            if (!$html_tag['value']) {
                // No actual tag specified, skip
                continue;
            }

            $tag = strtolower(strip_tags($html_tag['value']));
            $tags[$tag] = $html_tag['href'];
        }
        if (!empty($tags)) {
            net_nemein_tag_handler::tag_object($article, $tags, $this->_node->component);
        }
    }
}


1		<?php
2		/**
3		* @package net.nemein.rss
4		* @author The Midgard Project, http://www.midgard-project.org
5		* @copyright The Midgard Project, http://www.midgard-project.org
6		* @license http://www.gnu.org/licenses/lgpl.html GNU Lesser General Public License
7		*/
8
9		use Symfony\Component\DomCrawler\Crawler;
10		use midcom\dba\softdelete;
11		use SimplePie\SimplePie;
12
13		/**
14		* RSS and Atom feed fetching class. Caches the fetched items as articles
15		* in net.nehmer.blog
16		*
17		* @package net.nemein.rss
18		*/
19		class net_nemein_rss_fetch
20		{
21		/**
22		* The last error reported by SimplePie, if any
23		*/
24		public $lasterror;
25
26		private net_nemein_rss_feed_dba $_feed;
27
28		/**
29		* Property of midcom_db_article we're using for storing the feed item GUIDs
30		*/
31		private string $_guid_property = 'extra2';
32
33		/**
34		* Current node we're importing to
35		*/
36		private midcom_db_topic $_node;
37
38		/**
39		* Initializes the class with a given feed
40		*/
41	2	public function __construct(net_nemein_rss_feed_dba $feed)
42		{
43	2	$this->_feed = $feed;
44	2	$this->_node = new midcom_db_topic($feed->node);
45		}
46
47	2	public static function get_parser() : SimplePie
48		{
49	2	$parser = new SimplePie;
50	2	$parser->get_registry()->register('Item', net_nemein_rss_parser_item::class);
51	2	$parser->set_output_encoding(midcom::get()->i18n->get_current_charset());
52	2	$parser->set_cache_location(midcom::get()->config->get('midcom_tempdir'));
53	2	return $parser;
54		}
55
56		/**
57		* Actually fetch a feed
58		*/
59		public static function raw_fetch(string $url) : SimplePie
60		{
61		$parser = self::get_parser();
62		$parser->set_feed_url($url);
63		$parser->init();
64		return $parser;
65		}
66
67		/**
68		* Fetch given RSS or Atom feed
69		*
70		* @return net_nemein_rss_parser_item[] Array of normalized feed items
71		*/
72		function fetch() : array
73		{
74		$parser = self::raw_fetch($this->_feed->url);
75		if ($parser->error()) {
76		$this->lasterror = $parser->error();
77		return [];
78		}
79		if (!empty($parser->data['headers']['etag'])) {
80		// Etag checking
81		$etag = trim($parser->data['headers']['etag']);
82
83		$feed_etag = $this->_feed->get_parameter('net.nemein.rss', 'etag');
84		if ( !empty($feed_etag)
85		&& $feed_etag == $etag) {
86		// Feed hasn't changed, skip updating
87		debug_add("Feed {$this->_feed->url} has not changed since " . date('c', $this->_feed->latestfetch), MIDCOM_LOG_WARN);
88		return [];
89		}
90
91		$this->_feed->set_parameter('net.nemein.rss', 'etag', $etag);
92		}
93
94		$this->_feed->latestfetch = time();
95		$this->_feed->_use_rcs = false;
96		$this->_feed->update();
97
98		return $parser->get_items();
99		}
100
101		/**
102		* Fetches and imports items in the feed
103		*/
104		public function import() : array
105		{
106		if (!$this->_node->component) {
107		return [];
108		}
109
110		$items = $this->fetch();
111
112		if (empty($items)) {
113		// This feed didn't return any items, skip
114		return [];
115		}
116
117		// Reverse items so that creation times remain in correct order even for feeds without timestamps
118		$items = array_reverse($items);
119
120		foreach ($items as $item) {
121		if ($guid = $this->import_item($item)) {
122		$item->set_local_guid($guid);
123		debug_add("Imported item " . $item->get_id() . ' as ' . $guid, MIDCOM_LOG_INFO);
124		} else {
125		debug_add("Failed to import item " . $item->get_id() . ': ' . midcom_connection::get_error_string(), MIDCOM_LOG_ERROR);
126		}
127		}
128
129		$this->clean($items);
130
131		return array_reverse($items);
132		}
133
134		/**
135		* Imports a feed item into the database
136		*/
137	1	public function import_item(net_nemein_rss_parser_item $item) : ?string
138		{
139	1	if ($this->_node->component !== 'net.nehmer.blog') {
140		throw new midcom_error("RSS fetching for component {$this->_node->component} is unsupported");
141		}
142	1	return $this->import_article($item);
143		}
144
145		/**
146		* Imports an item as a news article
147		*/
148	1	private function import_article(net_nemein_rss_parser_item $item) : ?string
149		{
150	1	$guid = $item->get_id();
151	1	$title = $item->get_title();
152
153	1	if ( ( empty($title)
154	1	\|\| trim($title) == '...')
155	1	&& empty($guid)) {
156		// Something wrong with this entry, skip it
157		return null;
158		}
159
160	1	$article = $this->find_article($item, $guid);
161	1	if (!$article) {
162		return null;
163		}
164
165	1	$article->allow_name_catenate = true;
166	1	$article->set_rcs_message(sprintf(midcom::get()->i18n->get_string('%s was imported from %s', 'net.nemein.rss'), $title, $this->_feed->title));
167
168	1	$values = [
169	1	'title' => $title,
170	1	$this->_guid_property => $guid, // FIXME: This breaks with URLs longer than 255 chars
171	1	'content' => $item->get_content(),
172	1	'url' => $item->get_link(),
173	1	'extra1' => '\|feed:' . md5($this->_feed->url) . '\|',
174	1	];
175	1	$meta_values = [];
176
177		// Safety, make sure we have sane name (the allow_catenate was set earlier, so this will not clash
178	1	if (empty($article->name)) {
179	1	$values['name'] = midcom_helper_misc::urlize($title);
180		}
181
182	1	$categories = $item->get_categories();
183	1	if (is_array($categories)) {
184		// Handle categories provided in the feed
185	1	foreach ($categories as $category) {
186		// Clean up the categories and save
187	1	$category = str_replace('\|', '_', trim($category->get_term()));
188	1	$values['extra1'] .= "{$category}\|";
189		}
190		}
191
192	1	$article_author = $this->find_author($item);
193	1	if (!empty($article_author->guid)) {
194	1	$meta_values['authors'] = "\|{$article_author->guid}\|";
195		}
196
197		// Try to figure out item publication date
198	1	if ($article_date = $item->get_date('U')) {
199	1	$meta_values['published'] = $article_date;
200		} else {
201		$article_date = time();
202		if (!$article->id) {
203		$meta_values['published'] = $article_date;
204		}
205		}
206
207	1	if ($article_date > $this->_feed->latestupdate) {
208		// Cache "latest updated" time to feed
209	1	$this->_feed->latestupdate = $article_date;
210	1	$this->_feed->_use_rcs = false;
211	1	$this->_feed->update();
212		}
213
214	1	if ($article->id) {
215	1	if ( $this->apply_values($article, $values, $meta_values)
216	1	&& !$article->update()) {
217	1	return null;
218		}
219		} else {
220	1	$this->apply_values($article, $values, $meta_values);
221	1	if (!$article->create()) {
222		return null;
223		}
224		}
225
226	1	if ($this->_feed->autoapprove) {
227		$article->metadata->approve();
228		}
229
230	1	$this->_parse_tags($article);
231	1	$this->_parse_parameters($article, $item);
232
233		// store <link rel="replies"> url in parameter
234	1	if ($item->get_link(0, 'replies')) {
235		$article->set_parameter('net.nemein.rss', 'replies_url', $item->get_link(0, 'replies'));
236		}
237
238	1	return $article->guid;
239		}
240
241	1	private function find_author(net_nemein_rss_parser_item $item) : midcom_db_person
242		{
243		// Try to figure out item author
244	1	if ( $this->_feed->forceauthor
245	1	&& $this->_feed->defaultauthor) {
246		// Feed has a "default author" set, use it
247		return new midcom_db_person($this->_feed->defaultauthor);
248		}
249	1	$author = $this->match_item_author($item);
250	1	$fallback_person_id = 1;
251	1	if ( !$author
252	1	\|\| $author->id == $fallback_person_id) {
253	1	if ($this->_feed->defaultauthor) {
254		// Feed has a "default author" set, use it
255		$author = new midcom_db_person($this->_feed->defaultauthor);
256		} else {
257		// Fall back to "Midgard Admin" just in case
258	1	$author = new midcom_db_person($fallback_person_id);
259		}
260		}
261	1	return $author;
262		}
263
264	1	private function find_article(net_nemein_rss_parser_item $item, string $guid) : ?midcom_db_article
265		{
266	1	$qb = midcom_db_article::new_query_builder();
267	1	$qb->add_constraint('topic', '=', $this->_feed->node);
268	1	$qb->add_constraint($this->_guid_property, '=', substr($guid, 0, 255));
269	1	$articles = $qb->execute();
270	1	if (!empty($articles)) {
271		// This item has been imported already earlier. Update
272	1	return $articles[0];
273		}
274
275		// Check against duplicate hits that may come from different feeds
276	1	if ($link = $item->get_link()) {
277	1	$qb = midcom_db_article::new_query_builder();
278	1	$qb->add_constraint('topic', '=', $this->_feed->node);
279	1	$qb->add_constraint('url', '=', $link);
280	1	if ($qb->count() > 0) {
281		// Dupe, skip
282		return null;
283		}
284		}
285
286		// This is a new item
287	1	$article = new midcom_db_article();
288	1	$article->topic = $this->_feed->node;
289	1	return $article;
290		}
291
292	1	private function apply_values(midcom_db_article $article, array $values, array $meta_values) : bool
293		{
294	1	$updated = false;
295
296	1	foreach ($values as $fieldname => $value) {
297	1	if ($article->$fieldname !== $value) {
298	1	$article->$fieldname = $value;
299	1	$updated = true;
300		}
301		}
302
303	1	foreach ($meta_values as $fieldname => $value) {
304	1	if ($article->metadata->$fieldname !== $value) {
305	1	$article->metadata->$fieldname = $value;
306	1	$updated = true;
307		}
308		}
309
310	1	return $updated;
311		}
312
313		/**
314		* Cleans up old, removed items from feeds
315		*
316		* @param net_nemein_rss_parser_item[] $items
317		*/
318		private function clean(array $items)
319		{
320		if ($this->_feed->keepremoved) {
321		// This feed is set up so that we retain items removed from array
322		return;
323		}
324
325		// Create array of item GUIDs
326		$item_guids = [];
327		foreach ($items as $item) {
328		$item_guids[] = $item->get_id();
329		}
330
331		// Find articles resulting from this feed
332		$qb = midcom_db_article::new_query_builder();
333		$feed_category = md5($this->_feed->url);
334		$qb->add_constraint('extra1', 'LIKE', "%\|feed:{$feed_category}\|%");
335		$qb->add_constraint($this->_guid_property, 'NOT IN', $item_guids);
336		$local_items = $qb->execute_unchecked();
337		$purge_guids = [];
338		foreach ($local_items as $item) {
339		$purge_guids[] = $item->guid;
340		$item->delete();
341		}
342
343		softdelete::purge($purge_guids, 'midgard_article');
344		}
345
346		/**
347		* Parses author formats used by different feed standards and
348		* and returns the information
349		*/
350	2	public static function parse_item_author(net_nemein_rss_parser_item $item) : array
351		{
352	2	$author_info = [];
353
354		// First try dig up any information about the author possible
355	2	if ($author = $item->get_author()) {
356	2	$name = $author->get_name();
357	2	$email = $author->get_email();
358	2	if (!empty($name)) {
359		$name = html_entity_decode($name, ENT_QUOTES, midcom::get()->i18n->get_current_charset());
360		// Atom feed, the value can be either full name or username
361		$author_info['user_or_full'] = $name;
362		} else {
363	2	$name = html_entity_decode($email, ENT_QUOTES, midcom::get()->i18n->get_current_charset());
364		}
365
366	2	if (!preg_match('/[<\(]/', $name)) {
367	1	$author_info['user_or_full'] = $name;
368		} else {
369	2	if (str_contains($name, '<')) {
370		// The classic "Full Name <email>" format
371	1	$regex = '/(?<fullname>.+) <?(?<email>[a-zA-Z0-9_.-]+?@[a-zA-Z0-9_.-]+)>?[ ,]?/';
372		} else {
373		// The classic "email (Full Name)" format
374	1	$regex = '/^(?<email>[a-zA-Z0-9_.-]+?@[a-zA-Z0-9_.-]+) \((?<fullname>.+)\)$/';
375		}
376	2	if (preg_match($regex, $name, $matches)) {
377	2	$author_info['email'] = $matches['email'];
378	2	$author_info['user_or_full'] = $matches['fullname'];
379		}
380		}
381		}
382
383	2	if (isset($author_info['user_or_full'])) {
384	2	$author_info['user_or_full'] = trim($author_info['user_or_full']);
385	2	if (str_contains($author_info['user_or_full'], ' ')) {
386		// This value has a space in it, assuming full name
387	2	$author_info['full_name'] = $author_info['user_or_full'];
388		} else {
389	1	$author_info['username'] = $author_info['user_or_full'];
390		}
391	2	unset($author_info['user_or_full']);
392		}
393
394	2	return $author_info;
395		}
396
397		/**
398		* Parses author formats used by different feed standards and
399		* tries to match to persons in database.
400		*/
401	2	public function match_item_author(net_nemein_rss_parser_item $item) : ?midcom_db_person
402		{
403		// Parse the item for author information
404	2	$author_info = self::parse_item_author($item);
405
406	2	if (!empty($author_info['email'])) {
407		// Email is a pretty good identifier, start with it
408	2	$person_qb = midcom_db_person::new_query_builder();
409	2	$person_qb->add_constraint('email', '=', $author_info['email']);
410	2	$persons = $person_qb->execute();
411	2	if (!empty($persons)) {
412	1	return $persons[0];
413		}
414		}
415
416	2	if ( !empty($author_info['username'])
417	2	&& $person = midcom::get()->auth->get_user_by_name($author_info['username'])) {
418	1	return $person->get_storage();
419		}
420
421	2	if (!empty($author_info['full_name'])) {
422	2	$name_parts = explode(' ', $author_info['full_name']);
423	2	if (count($name_parts) > 1) {
424		// We assume the western format Firstname Lastname
425	2	$firstname = $name_parts[0];
426	2	$lastname = $name_parts[1];
427
428	2	$person_qb = midcom_db_person::new_query_builder();
429	2	$person_qb->add_constraint('firstname', '=', $firstname);
430	2	$person_qb->add_constraint('lastname', '=', $lastname);
431	2	$persons = $person_qb->execute();
432	2	if (!empty($persons)) {
433	1	return $persons[0];
434		}
435		}
436		}
437
438	1	return null;
439		}
440
441		/**
442		* Parses additional metadata in RSS item and sets parameters accordingly
443		*/
444	1	private function _parse_parameters(midcom_db_article $article, net_nemein_rss_parser_item $item)
445		{
446	1	foreach ($item->get_enclosures() as $enclosure) {
447	1	$article->set_parameter('net.nemein.rss:enclosure', 'url', $enclosure->get_link());
448	1	$article->set_parameter('net.nemein.rss:enclosure', 'duration', $enclosure->get_duration());
449	1	$article->set_parameter('net.nemein.rss:enclosure', 'mimetype', $enclosure->get_type());
450		}
451		}
452
453		/**
454		* Parses rel-tag links in article content and tags the object based on them
455		*/
456	1	private function _parse_tags(midcom_db_article $article)
457		{
458	1	$crawler = new Crawler($article->content);
459	1	$nodes = $crawler->filter('a[rel="tag"]');
460
461	1	$html_tags = $nodes->each(function(Crawler $node) {
462	1	return [
463	1	'href' => $node->attr('href') ?? false,
464	1	'value' => $node->text() ?? false,
465	1	];
466	1	});
467
468	1	$tags = [];
469
470	1	foreach ($html_tags as $html_tag) {
471	1	if (!$html_tag['value']) {
472		// No actual tag specified, skip
473		continue;
474		}
475
476	1	$tag = strtolower(strip_tags($html_tag['value']));
477	1	$tags[$tag] = $html_tag['href'];
478		}
479	1	if (!empty($tags)) {
480	1	net_nemein_tag_handler::tag_object($article, $tags, $this->_node->component);
481		}
482		}
483		}
484

flack / openpsa

Push — master ( 33e64a...a4094b )

net_nemein_rss_fetch::raw_fetch() A

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like