/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result, StateInterface $state)
{
// Set time zone to GMT for parsing dates with strtotime().
$tz = date_default_timezone_get();
date_default_timezone_set('GMT');
$raw = trim($fetcher_result->getRaw());
if (!strlen($raw)) {
throw new EmptyFeedException();
}
// Yes, using a DOM parser is a bit inefficient, but will do for now.
// @todo XML error handling.
$this->startXmlErrorHandling();
$xml = new \SimpleXMLElement($raw);
$this->stopXmlErrorHandling();
$result = new ParserResult();
foreach ($xml->url as $url) {
$item = new SitemapItem();
$item->set('url', (string) $url->loc);
if ($url->lastmod) {
$item->set('lastmod', strtotime($url->lastmod));
}
if ($url->changefreq) {
$item->set('changefreq', (string) $url->changefreq);
}
if ($url->priority) {
$item->set('priority', (string) $url->priority);
}
$result->addItem($item);
}
date_default_timezone_set($tz);
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result)
{
$feed_config = $feed->getConfigurationFor($this);
$state = $feed->getState(StateInterface::PARSE);
// Load and configure parser.
$parser = new ParserCSV();
$delimiter = $feed_config['delimiter'] == 'TAB' ? "\t" : $feed_config['delimiter'];
$parser->setDelimiter($delimiter);
$iterator = new ParserCSVIterator($fetcher_result->getFilePath());
if (empty($feed_config['no_headers'])) {
// Get first line and use it for column names, convert them to lower case.
$header = $this->parseHeader($parser, $iterator);
if (!$header) {
return;
}
$parser->setColumnNames($header);
}
// Determine section to parse, parse.
$start = $state->pointer ? $state->pointer : $parser->lastLinePos();
$limit = $this->importer->getLimit();
$rows = $this->parseItems($parser, $iterator, $start, $limit);
// Report progress.
$state->total = filesize($fetcher_result->getFilePath());
$state->pointer = $parser->lastLinePos();
$progress = $parser->lastLinePos() ? $parser->lastLinePos() : $state->total;
$state->progress($state->total, $progress);
// Create a result object and return it.
return new ParserResult($rows, $feed->id());
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result, StateInterface $state)
{
$feed_config = $feed->getConfigurationFor($this);
if (!filesize($fetcher_result->getFilePath())) {
throw new EmptyFeedException();
}
// Load and configure parser.
$parser = CsvFileParser::createFromFilePath($fetcher_result->getFilePath())->setDelimiter($feed_config['delimiter'] === 'TAB' ? "\t" : $feed_config['delimiter'])->setHasHeader(!$feed_config['no_headers'])->setStartByte((int) $state->pointer);
// Wrap parser in a limit iterator.
$parser = new \LimitIterator($parser, 0, $this->configuration['line_limit']);
$header = !$feed_config['no_headers'] ? $parser->getHeader() : [];
$result = new ParserResult();
foreach ($parser as $row) {
$item = new DynamicItem();
foreach ($row as $delta => $cell) {
$key = isset($header[$delta]) ? $header[$delta] : $delta;
$item->set($key, $cell);
}
$result->addItem($item);
}
// Report progress.
$state->total = filesize($fetcher_result->getFilePath());
$state->pointer = $parser->lastLinePos();
$state->progress($state->total, $state->pointer);
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result)
{
$parser = new GenericOPMLParser($fetcher_result->getRaw());
$opml = $parser->parse(TRUE);
$result = new ParserResult();
$result->items = $this->getItems($opml['outlines'], array());
$result->title = $opml['head']['#title'];
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result, StateInterface $state)
{
$raw = $fetcher_result->getRaw();
if (!strlen(trim($raw))) {
throw new EmptyFeedException();
}
$result = new ParserResult();
$parser = new GenericOpmlParser($fetcher_result->getRaw());
$opml = $parser->parse(TRUE);
foreach ($this->getItems($opml['outlines']) as $item) {
$item->set('feed_title', $opml['head']['#title']);
$result->addItem($item);
}
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result, StateInterface $state)
{
$result = new ParserResult();
Reader::setExtensionManager(\Drupal::service('feed.bridge.reader'));
Reader::registerExtension('GeoRSS');
$raw = $fetcher_result->getRaw();
if (!strlen(trim($raw))) {
throw new EmptyFeedException();
}
try {
$channel = Reader::importString($raw);
} catch (ExceptionInterface $e) {
$args = ['%site' => $feed->label(), '%error' => trim($e->getMessage())];
throw new \RuntimeException($this->t('The feed from %site seems to be broken because of error "%error".', $args));
}
foreach ($channel as $delta => $entry) {
$item = new SyndicationItem();
// Move the values to an array as expected by processors.
$item->set('title', $entry->getTitle())->set('guid', $entry->getId())->set('url', $entry->getLink())->set('guid', $entry->getId())->set('url', $entry->getLink())->set('description', $entry->getDescription())->set('tags', $entry->getCategories()->getValues())->set('feed_title', $channel->getTitle())->set('feed_description', $channel->getDescription())->set('feed_url', $channel->getLink());
if ($image = $channel->getImage()) {
$item->set('feed_image_uri', $image['uri']);
}
if ($enclosure = $entry->getEnclosure()) {
$item->set('enclosures', [rawurldecode($enclosure->url)]);
}
if ($author = $entry->getAuthor()) {
$author += ['name' => '', 'email' => ''];
$item->set('author_name', $author['name'])->set('author_email', $author['email']);
}
if ($date = $entry->getDateModified()) {
$item->set('timestamp', $date->getTimestamp());
}
if ($point = $entry->getGeoPoint()) {
$item->set('georss_lat', $point['lat'])->set('georss_lon', $point['lon']);
}
$result->addItem($item);
}
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result, StateInterface $state)
{
// Set time zone to GMT for parsing dates with strtotime().
$tz = date_default_timezone_get();
date_default_timezone_set('GMT');
// Reset item counter.
$this->items_count = 0;
// Get raw data.
$raw = trim($fetcher_result->getRaw());
if (!strlen($raw)) {
throw new EmptyFeedException();
}
$data = Json::decode($raw);
$result = new ParserResult();
if ($data && count($data['items']) > 0) {
$this->processItems($data['items'], $result);
}
if ($data['pageInfo']['totalResults'] && $data['pageInfo']['resultsPerPage'] && $data['pageInfo']['totalResults'] > $data['pageInfo']['resultsPerPage']) {
$number_of_pages = $data['pageInfo']['totalResults'] / $data['pageInfo']['resultsPerPage'];
if ($number_of_pages > 1) {
$feed_type = $feed->getType();
$fetcher_configuration = $feed_type->getFetcher()->getConfiguration();
$yt_state = ['channel_id' => $feed->getSource(), 'api_key' => $fetcher_configuration['api_key'], 'import_limit' => $fetcher_configuration['import_limit'], 'page_limit' => $fetcher_configuration['page_limit'], 'pageToken' => ''];
for ($i = 0; $i <= $number_of_pages; $i++) {
if (!$data) {
throw new EmptyFeedException();
}
if ($data['nextPageToken']) {
$yt_state['pageToken'] = $data['nextPageToken'];
$data = Json::decode($this->fetchInternal($feed, $yt_state));
$this->processItems($data['items'], $result);
}
}
}
}
date_default_timezone_set($tz);
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result)
{
// Set time zone to GMT for parsing dates with strtotime().
$tz = date_default_timezone_get();
date_default_timezone_set('GMT');
// Yes, using a DOM parser is a bit inefficient, but will do for now.
$xml = new \SimpleXMLElement($fetcher_result->getRaw());
$result = new ParserResult();
foreach ($xml->url as $url) {
$item = array('url' => (string) $url->loc);
if ($url->lastmod) {
$item['lastmod'] = strtotime($url->lastmod);
}
if ($url->changefreq) {
$item['changefreq'] = (string) $url->changefreq;
}
if ($url->priority) {
$item['priority'] = (string) $url->priority;
}
$result->items[] = $item;
}
date_default_timezone_set($tz);
return $result;
}
/**
* {@inheritdoc}
*/
public function parse(FeedInterface $feed, FetcherResultInterface $fetcher_result)
{
$result = new ParserResult();
Reader::setExtensionManager(\Drupal::service('feed.bridge.reader'));
try {
$channel = Reader::importString($fetcher_result->getRaw());
} catch (ExceptionInterface $e) {
watchdog_exception('feeds', $e);
drupal_set_message($this->t('The feed from %site seems to be broken because of error "%error".', array('%site' => $feed->label(), '%error' => $e->getMessage())), 'error');
return $result;
}
$result->title = $channel->getTitle();
$result->description = $channel->getDescription();
$result->link = $channel->getLink();
foreach ($channel as $item) {
// Reset the parsed item.
$parsed_item = array();
// Move the values to an array as expected by processors.
$parsed_item['title'] = $item->getTitle();
$parsed_item['guid'] = $item->getId();
$parsed_item['url'] = $item->getLink();
$parsed_item['description'] = $item->getDescription();
if ($enclosure = $item->getEnclosure()) {
$parsed_item['enclosures'][] = urldecode($enclosure->url);
}
if ($author = $item->getAuthor()) {
$parsed_item['author_name'] = $author['name'];
}
if ($date = $item->getDateModified()) {
$parsed_item['timestamp'] = $date->getTimestamp();
}
$parsed_item['tags'] = $item->getCategories()->getValues();
$result->items[] = $parsed_item;
}
return $result;
}
/**
* Finds a hub from a fetcher result.
*
* @param \Drupal\feeds\Result\FetcherResultInterface $fetcher_result
* The fetcher result.
*
* @return string|null
* The hub URL or null if one wasn't found.
*/
protected function findRelation(FetcherResultInterface $fetcher_result, $relation)
{
if ($fetcher_result instanceof HttpFetcherResultInterface) {
if ($rel = HttpHelpers::findLinkHeader($fetcher_result->getHeaders(), $relation)) {
return $rel;
}
}
return HttpHelpers::findRelationFromXml($fetcher_result->getRaw(), $relation);
}
