123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318 |
- <?php
- namespace PicoFeed\Parser;
- use SimpleXMLElement;
- use PicoFeed\Filter\Filter;
- /**
- * RSS 1.0 parser.
- *
- * @package PicoFeed\Parser
- * @author Frederic Guillot
- */
- class Rss10 extends Parser
- {
- /**
- * Supported namespaces.
- */
- protected $namespaces = array(
- 'rss' => 'http://purl.org/rss/1.0/',
- 'dc' => 'http://purl.org/dc/elements/1.1/',
- 'content' => 'http://purl.org/rss/1.0/modules/content/',
- 'feedburner' => 'http://rssnamespace.org/feedburner/ext/1.0',
- );
- /**
- * Get the path to the items XML tree.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @return SimpleXMLElement[]
- */
- public function getItemsTree(SimpleXMLElement $xml)
- {
- return XmlParser::getXPathResult($xml, 'rss:item', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'item')
- ?: $xml->item;
- }
- /**
- * Find the feed url.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedUrl(SimpleXMLElement $xml, Feed $feed)
- {
- $feed->setFeedUrl('');
- }
- /**
- * Find the site url.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findSiteUrl(SimpleXMLElement $xml, Feed $feed)
- {
- $value = XmlParser::getXPathResult($xml, 'rss:channel/rss:link', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'channel/link')
- ?: $xml->channel->link;
- $feed->setSiteUrl(XmlParser::getValue($value));
- }
- /**
- * Find the feed description.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedDescription(SimpleXMLElement $xml, Feed $feed)
- {
- $description = XmlParser::getXPathResult($xml, 'rss:channel/rss:description', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'channel/description')
- ?: $xml->channel->description;
- $feed->setDescription(XmlParser::getValue($description));
- }
- /**
- * Find the feed logo url.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedLogo(SimpleXMLElement $xml, Feed $feed)
- {
- $logo = XmlParser::getXPathResult($xml, 'rss:image/rss:url', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'image/url');
- $feed->setLogo(XmlParser::getValue($logo));
- }
- /**
- * Find the feed icon.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedIcon(SimpleXMLElement $xml, Feed $feed)
- {
- $feed->setIcon('');
- }
- /**
- * Find the feed title.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedTitle(SimpleXMLElement $xml, Feed $feed)
- {
- $title = XmlParser::getXPathResult($xml, 'rss:channel/rss:title', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'channel/title')
- ?: $xml->channel->title;
- $feed->setTitle(Filter::stripWhiteSpace(XmlParser::getValue($title)) ?: $feed->getSiteUrl());
- }
- /**
- * Find the feed language.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedLanguage(SimpleXMLElement $xml, Feed $feed)
- {
- $language = XmlParser::getXPathResult($xml, 'rss:channel/dc:language', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'channel/dc:language', $this->namespaces);
- $feed->setLanguage(XmlParser::getValue($language));
- }
- /**
- * Find the feed id.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedId(SimpleXMLElement $xml, Feed $feed)
- {
- $feed->setId($feed->getFeedUrl() ?: $feed->getSiteUrl());
- }
- /**
- * Find the feed date.
- *
- * @param SimpleXMLElement $xml Feed xml
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findFeedDate(SimpleXMLElement $xml, Feed $feed)
- {
- $date = XmlParser::getXPathResult($xml, 'rss:channel/dc:date', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'channel/dc:date', $this->namespaces);
- $feed->setDate($this->getDateParser()->getDateTime(XmlParser::getValue($date)));
- }
- /**
- * Find the item published date.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param Item $item Item object
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findItemPublishedDate(SimpleXMLElement $entry, Item $item, Feed $feed)
- {
- $date = XmlParser::getXPathResult($entry, 'dc:date', $this->namespaces);
- $item->setPublishedDate(!empty($date) ? $this->getDateParser()->getDateTime(XmlParser::getValue($date)) : null);
- }
- /**
- * Find the item updated date.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param Item $item Item object
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findItemUpdatedDate(SimpleXMLElement $entry, Item $item, Feed $feed)
- {
- if ($item->publishedDate === null) {
- $this->findItemPublishedDate($entry, $item, $feed);
- }
- $item->setUpdatedDate($item->getPublishedDate()); // No updated date in RSS 1.0 specifications
- }
- /**
- * Find the item title.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- */
- public function findItemTitle(SimpleXMLElement $entry, Item $item)
- {
- $title = XmlParser::getXPathResult($entry, 'rss:title', $this->namespaces)
- ?: XmlParser::getXPathResult($entry, 'title')
- ?: $entry->title;
- $item->setTitle(Filter::stripWhiteSpace(XmlParser::getValue($title)) ?: $item->getUrl());
- }
- /**
- * Find the item author.
- *
- * @param SimpleXMLElement $xml Feed
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- */
- public function findItemAuthor(SimpleXMLElement $xml, SimpleXMLElement $entry, Item $item)
- {
- $author = XmlParser::getXPathResult($entry, 'dc:creator', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'rss:channel/dc:creator', $this->namespaces)
- ?: XmlParser::getXPathResult($xml, 'channel/dc:creator', $this->namespaces);
- $item->setAuthor(XmlParser::getValue($author));
- }
- /**
- * Find the item author URL.
- *
- * @param SimpleXMLElement $xml Feed
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- */
- public function findItemAuthorUrl(SimpleXMLElement $xml, SimpleXMLElement $entry, Item $item)
- {
- // There appears to be no support for author URL in the dc: terms
- $item->setAuthorUrl('');
- }
- /**
- * Find the item content.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- */
- public function findItemContent(SimpleXMLElement $entry, Item $item)
- {
- $content = XmlParser::getXPathResult($entry, 'content:encoded', $this->namespaces);
- if (XmlParser::getValue($content) === '') {
- $content = XmlParser::getXPathResult($entry, 'rss:description', $this->namespaces)
- ?: XmlParser::getXPathResult($entry, 'description')
- ?: $entry->description;
- }
- $item->setContent(XmlParser::getValue($content));
- }
- /**
- * Find the item URL.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- */
- public function findItemUrl(SimpleXMLElement $entry, Item $item)
- {
- $link = XmlParser::getXPathResult($entry, 'feedburner:origLink', $this->namespaces)
- ?: XmlParser::getXPathResult($entry, 'rss:link', $this->namespaces)
- ?: XmlParser::getXPathResult($entry, 'link')
- ?: $entry->link;
- $item->setUrl(XmlParser::getValue($link));
- }
- /**
- * Genereate the item id.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findItemId(SimpleXMLElement $entry, Item $item, Feed $feed)
- {
- $item->setId($this->generateId(
- $item->getTitle(), $item->getUrl(), $item->getContent()
- ));
- }
- /**
- * Find the item enclosure.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findItemEnclosure(SimpleXMLElement $entry, Item $item, Feed $feed)
- {
- }
- /**
- * Find the item language.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param \PicoFeed\Parser\Item $item Item object
- * @param \PicoFeed\Parser\Feed $feed Feed object
- */
- public function findItemLanguage(SimpleXMLElement $entry, Item $item, Feed $feed)
- {
- $language = XmlParser::getXPathResult($entry, 'dc:language', $this->namespaces);
- $item->setLanguage(XmlParser::getValue($language) ?: $feed->getLanguage());
- }
- /**
- * Find the item categories.
- *
- * @param SimpleXMLElement $entry Feed item
- * @param Item $item Item object
- * @param Feed $feed Feed object
- */
- public function findItemCategories(SimpleXMLElement $entry, Item $item, Feed $feed)
- {
- $categories = XmlParser::getXPathResult($entry, 'dc:subject', $this->namespaces);
- $item->setCategoriesFromXml($categories);
- }
- }
|