public function build(\SimplePie_Item $feed)
{
$data = new ArrayCollection();
$data->set('title', $feed->get_title());
$data->set('url', $feed->get_permalink());
$data->set('description', $feed->get_description());
return $data;
}
/**
*
* Check if a feed item is suitable for import into database
*
* @param \SimplePie_Item $item
* @param \Datetime $itemUploadedToFeed
* @param \Datetime $lastImportDate
* @return bool
*/
protected function checkSuitableForImport(\SimplePie_Item $item, $itemUploadedToFeed, $lastImportDate)
{
// only check new import items (ie added since last import date)
if ($itemUploadedToFeed < $lastImportDate) {
return false;
}
// check db to see if item was:
// - manually added by a protalk team member
// - somehow re-added to the feed, at a later date, but with the same title
// - previously imported but a protalk team member edited the title (hence permalink check)
$repository = $this->entityManager->getRepository('ProtalkMediaBundle:Media');
$itemExists = $repository->itemExists($item->get_permalink());
if ($itemExists) {
return false;
}
return true;
}
public function getFavIcon(SimplePie_Item $item)
{
$shortname = '';
$categories = $item->get_categories();
foreach ($categories as $category) {
$matches = array();
$matched = preg_match('/short=([a-z-]+)/', $category->get_term(), $matches);
if ($matched === 1) {
$shortname = $matches[1];
break;
}
}
if (array_key_exists($shortname, self::$map)) {
$favIcon = self::$map[$shortname];
} else {
$favIcon = $item->get_feed()->get_favicon();
}
return $favIcon;
}
/**
* Parses a since feed item.
*
* @param \Devour\Table\TableInterface $table
* A table obeject.
* @param \SimplePie_Item $item
* A SimplePie item.
*/
protected function parseItem(TableInterface $table, \SimplePie_Item $item)
{
// @todo Add more fields.
$row = $table->getNewRow();
$row->set('id', $item->get_id())->set('permalink', $item->get_permalink())->set('title', $item->get_title())->set('date', $item->get_gmdate('U'))->set('content', $item->get_content());
if ($author = $item->get_author()) {
$row->set('author_name', $author->get_name())->set('author_email', $author->get_email());
}
}
/**
* The constructor stores feed item attributes as properties.
*
* @param type $feed the feed the new item belongs to
* @param SimplePie_Item $item SimplePie_Item that contains information
* about the item
*/
public function __construct($feed, $item)
{
$this->_feed = $feed;
$this->_title = $item->get_title();
$this->_link = $item->get_permalink();
$this->_description = $item->get_content();
$this->_date = $item->get_date('U');
$author = $item->get_author();
$author_name = $author ? $author->get_name() : NULL;
$this->_author = $author && empty($author_name) ? $author->get_email() : $author_name;
}
function filter(SimplePie_Item $item, array $patterns)
{
foreach ($patterns as $pattern) {
switch ($pattern["field"]) {
case "title":
if (preg_match($pattern["regex"], $item->get_title())) {
return false;
}
break;
case "summary":
if (preg_match($pattern["regex"], $item->get_description())) {
return false;
}
break;
case "content":
if (preg_match($pattern["regex"], $item->get_content())) {
return false;
}
break;
case "url":
if (preg_match($pattern["regex"], $item->get_permalink())) {
return false;
}
break;
case "category":
foreach ($item->get_categories() as $category) {
if (preg_match($pattern["regex"], $category->get_term()) || preg_match($pattern["regex"], $category->get_label())) {
return false;
}
}
break;
default:
die("Filter not implemented for field " . $pattern["field"]);
}
}
return true;
}
/**
* Populate an item from a feed item
*
* @param $item \Planetflow3\Domain\Model\Item
* @param $feedItem \SimplePie_Item
* @return void
*/
protected function populateItemProperties(Item $item, \SimplePie_Item $feedItem)
{
$item->setLink($feedItem->get_link());
$item->setTitle($feedItem->get_title());
$item->setDescription($feedItem->get_description());
$item->setContent($feedItem->get_content(TRUE));
$item->setPublicationDate(new \DateTime($feedItem->get_date()));
$item->setAuthor($feedItem->get_author());
}
/**
* @param \SimplePie_Item $tweet
* @return \Swiftriver\Core\ObjectModel\DuplicationIdentificationFieldCollection
*/
private function ParseTweetToSanitizedTweetDiffCollection($tweet)
{
//Get the original text
$tweetText = $tweet->get_title();
//Break the text down into words
$tweetTextParts = explode(" ", $tweetText);
//Set up the sanitized return string
$sanitizedText = "";
//loop through all the words
foreach ($tweetTextParts as $part) {
//to lowwer the word
$part = strtolower($part);
//If the word contains none standard chars, continue
if (preg_match("/[^\\w\\d\\.\\(\\)\\!']/si", $part)) {
continue;
}
//if the owrd is just rt then continue
if ($part == "rt") {
continue;
}
//Add the word to the sanitized
$sanitizedText .= $part . " ";
}
//Create a new Diff
$dif = new \Swiftriver\Core\ObjectModel\DuplicationIdentificationField("Sanitized Tweet", utf8_encode($sanitizedText));
//Create the new diff collection
$difCollection = new \Swiftriver\Core\ObjectModel\DuplicationIdentificationFieldCollection("Sanitized Tweet", array($dif));
//Return the diff collection
return $difCollection;
}
/**
* Parses the simplepie item to a content item
* @param \SimplePie_Item $tweet
* @param \Swiftriver\Core\ObjectModel\Source
* @return \Swiftriver\Core\ObjectModel\Content
*/
private function ParseTweetFromATOMItem($tweet, $channel)
{
//Extract all the relevant feedItem info
$title = $tweet->get_title();
//$description = $tweet->get_description();
$contentLink = $tweet->get_permalink();
$date = $tweet->get_date();
//Create the source
$author = $tweet->get_author();
$source_name = $author != null ? $author->get_name() : $channel->name;
$source = \Swiftriver\Core\ObjectModel\ObjectFactories\SourceFactory::CreateSourceFromIdentifier($source_name);
$source->name = $source_name;
$source->email = $author != null ? $tweet->get_author()->get_email() : null;
$source->link = $author != null ? $tweet->get_author()->get_link() : null;
$source->parent = $channel->id;
$source->type = $channel->type;
$source->subType = $channel->subType;
//Create a new Content item
$item = \Swiftriver\Core\ObjectModel\ObjectFactories\ContentFactory::CreateContent($source);
//Fill the Content Item
$item->text[] = new \Swiftriver\Core\ObjectModel\LanguageSpecificText(null, $title, array());
$item->link = $contentLink;
$item->date = strtotime($date);
return $item;
}
/**
* Constructor
*/
function WP_SimplePieAtomPub_Item($feed, $data)
{
parent::SimplePie_Item($feed, $data);
}
/**
* @return string twitterified content
*/
function get_content($content_only = false)
{
$content = $this->twitterify(parent::get_content($content_only));
return $content;
}
protected function getLinkFromItem(SimplePie_Item $i)
{
/* @var $e SimplePie_Enclosure */
$e = $i->get_enclosure(0);
if (empty($e)) {
return '';
}
return $e->get_link();
}
/**
* @param SimplePie_Item $item
* @return string
*/
function socialbookmarks_get_tags($item)
{
global $serendipity;
$return = '';
$taglink = $serendipity['baseURL'] . ($serendipity['rewrite'] == 'none' ? $serendipity['indexFile'] . '?/' : '') . 'plugin/tag/';
switch ($this->get_config('socialbookmarksService')) {
case 'del.icio.us':
// quite easy
$return .= '<br/><p style="font-size:.7em;margin:0;padding:0" class="serendipity_socialbookmarks_tags">[Tags:';
$tags = $item->get_categories();
/** @var SimplePie_Category $tag */
foreach ($tags as $tag) {
$return .= ' <a href="' . $taglink . socialbookmarks_freetag_compat(strtolower($tag->get_term())) . '" rel="tag">' . strtolower($tag->get_term()) . '</a>';
}
$return .= ']</p>';
break;
case 'ma.gnolia':
// they've changed this recently
$return .= '<br/><p style="font-size:.7em;margin:0;padding:0" class="serendipity_socialbookmarks_tags">[Tags:';
$tags = $item->get_categories();
//$tags = explode(' ', $tags[0]);
foreach ($tags as $tag) {
$return .= ' <a href="' . $taglink . socialbookmarks_freetag_compat(strtolower($tag)) . '" rel="tag">' . strtolower($tag) . '</a>';
}
$return .= ']</p>';
break;
case 'furl':
$return .= '<br/><p style="font-size:.7em;margin:0;padding:0" class="serendipity_socialbookmarks_tags">[Tags:';
$tags = $item->get_category();
foreach ($tags as $tag) {
$return .= ' <a href="' . $taglink . socialbookmarks_freetag_compat(strtolower($tag)) . '" rel="tag">' . strtolower($tag) . '</a>';
}
$return .= ']</p>';
break;
case 'misterwong':
case 'linkroll':
// services don't provide tags in their RSS feeds (yet)!?
// services don't provide tags in their RSS feeds (yet)!?
default:
break;
}
return $return;
}
function get_categories() {
$cats = parent::get_categories();
$mycats = array();
if (!is_null($cats)) {
foreach ($cats as $cat) {
$mycats[] = $cat->term;
}
}
return $mycats;
}
function get_source()
{
return apply_filters('feedwordpie_item_get_source', parent::get_source(), $this);
}
/**
* Parses the simplepie item to a content item
* @param \SimplePie_Item $tweet
* @param \Swiftriver\Core\ObjectModel\Channel $channel
*/
private function ParseTweetFromATOMItem($tweet, $channel)
{
//Extract all the relevant feedItem info
$title = $tweet->get_title();
//$description = $tweet->get_description();
$contentLink = $tweet->get_permalink();
$date = $tweet->get_date();
//Create the source
$author = $tweet->get_author();
$source_name = $author != null ? $author->get_name() : $channel->name;
$source = \Swiftriver\Core\ObjectModel\ObjectFactories\SourceFactory::CreateSourceFromIdentifier($source_name, $channel->trusted);
$source->name = $source_name;
$source->email = $author != null ? $tweet->get_author()->get_email() : null;
$source->link = $author != null ? $tweet->get_author()->get_link() : null;
$source->parent = $channel->id;
$source->type = $channel->type;
$source->subType = $channel->subType;
//Add location data
//Long and lat
$location = $tweet->get_item_tags("http://www.georss.org/georss", "point");
$long = 0;
$lat = 0;
$name = "";
if (is_array($location)) {
$lat_lon_array = split(" ", $location[0]["data"]);
$long = $lat_lon_array[1];
$lat = $lat_lon_array[0];
//Name
$location_name = $tweet->get_item_tags("http://api.twitter.com", "place");
if (is_array($location_name)) {
if (isset($location_name[0]["child"]["http://api.twitter.com"]["full_name"][0]["data"])) {
$name = $location_name[0]["child"]["http://api.twitter.com"]["full_name"][0]["data"];
}
}
$source->gisData = array(new \Swiftriver\Core\ObjectModel\GisData($long, $lat, $name));
}
//Create a new Content item
$item = \Swiftriver\Core\ObjectModel\ObjectFactories\ContentFactory::CreateContent($source);
//Fill the Content Item
$item->text[] = new \Swiftriver\Core\ObjectModel\LanguageSpecificText(null, $title, array());
$item->link = $contentLink;
$item->date = strtotime($date);
//Sanitize the tweet text into a DIF collection
$sanitizedTweetDiffCollection = $this->ParseTweetToSanitizedTweetDiffCollection($tweet);
//Add the dif collection to the item
$item->difs = array($sanitizedTweetDiffCollection);
//return the item
return $item;
}
/**
* MagpieFromSimplePie::normalize_author_inheritance
*
* @param SimplePie_Item $original
*
* @uses SimplePie_Item::get_authors
* @uses SimplePie_Author::get_name
* @uses SimplePie_Author::get_link
* @uses SimplePie_Author::get_email
* @uses MagpieFromSimplePie::increment_element
*/
function normalize_author_inheritance(&$item, $original)
{
// "If an atom:entry element does not contain
// atom:author elements, then the atom:author elements
// of the contained atom:source element are considered
// to apply. In an Atom Feed Document, the atom:author
// elements of the containing atom:feed element are
// considered to apply to the entry if there are no
// atom:author elements in the locations described
// above." <http://atompub.org/2005/08/17/draft-ietf-atompub-format-11.html#rfc.section.4.2.1>
if (!isset($item["author#"])) {
$authors = $original->get_authors();
foreach ($authors as $author) {
$tag = $this->increment_element($item, 'author', 'atom', array());
$item[$tag] = $item["{$tag}_name"] = $author->get_name();
if ($author->get_link()) {
$item["{$tag}_uri"] = $item["{$tag}_url"] = $author->get_link();
}
if ($author->get_email()) {
$item["{$tag}_email"] = $author->get_email();
}
}
}
}
/**
* Implements CollectionAbstract::buildDocument().
*
* @param IndexDocument $document
* @param \SimplePie_Item $data
*/
public function buildDocument(IndexDocument $document, $data)
{
$document->source = $this->_feed->get_title();
$document->subject = $this->_feed->get_description();
$document->title = $data->get_title();
$document->link = $data->get_link();
$document->description = $data->get_description();
$document->creator = (array) $data->get_author();
$document->date = $data->get_date();
// PHP properties cannot have dashes (-), and the fields below have
// dashes in the field name.
$document->source_link = $this->_feed->get_link();
$document->getField('source_link')->setName('source-link');
$document->item_subject = $this->_feed->get_link();
$document->getField('item_subject')->setName('item-subject');
}
/**
* Pieces together the title attribute without using inline ternary conditional statements
*
* @access public
* @param SimplePie_Item $item The item to generate the title attribute for.
* @param integer $length The number of characters to return in the description.
* @param string $date_format The format to use when displaying dates on items. Uses values from http://php.net/strftime, NOT http://php.net/date.
* @return string The string for the title attribute.
*/
function get_title_attr($item, $length, $date_format)
{
// Get a handle for the item's parent feed object for PHP 4.x.
$parent = $item->get_feed();
$title_attr = '';
$title_attr .= $item->get_title();
// The title of the post
$title_attr .= ' :: ';
// The separator between the title and the description (required by MooTools)
$title_attr .= newsblocks::cleanup($item->get_description(), $length);
// The cleaned-up and shortened version of the description
$title_attr .= '<span>';
// This marks the beginning of the date/domain line (and is CSS styleable)
// Does the item have a timestamp?
if ($item->get_local_date($date_format)) {
$title_attr .= $item->get_local_date($date_format);
// Use the locale-friendly version for non-English languages.
$title_attr .= ' // ';
// Visual separator.
}
$title_attr .= newsblocks::name($parent->subscribe_url());
// The domain name that the item is coming from.
$title_attr .= '</span>';
// Mark the end of the date/domain line.
return $title_attr;
}
