public function collectData(array $param)
{
function StripCDATA($string)
{
$string = str_replace('<![CDATA[', '', $string);
$string = str_replace(']]>', '', $string);
return $string;
}
function StripWithDelimiters($string, $start, $end)
{
while (strpos($string, $start) !== false) {
$section_to_remove = substr($string, strpos($string, $start));
$section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
$string = str_replace($section_to_remove, '', $string);
}
return $string;
}
function CleanArticle($article_html)
{
$article_html = StripWithDelimiters($article_html, '<script', '</script>');
$article_html = StripWithDelimiters($article_html, '<h1 class="cleanprint-title"', '</h1>');
return $article_html;
}
$feedUrl = 'http://www.lemondeinformatique.fr/rss/rss.xml';
$html = $this->file_get_html($feedUrl) or $this->returnError('Could not request LeMondeInformatique: ' . $feedUrl, 500);
$limit = 0;
foreach ($html->find('item') as $element) {
if ($limit < 5) {
//Retrieve article details
$article_uri = $element->innertext;
$article_uri = substr($article_uri, strpos($article_uri, '<link>') + 6);
$article_uri = substr($article_uri, 0, strpos($article_uri, '</link>'));
$article_html = $this->file_get_html($article_uri) or $this->returnError('Could not request LeMondeInformatique: ' . $article_uri, 500);
$thumbnailUri = $article_html->find('div#article', 0)->find('img#illustration', 0)->src;
$article_content = CleanArticle($article_html->find('div#article', 0)->innertext);
$article_title = $article_html->find('h1.cleanprint-title', 0)->plaintext;
//Build and add final item
$item = new \Item();
$item->uri = $article_uri;
$item->thumbnailUri = $thumbnailUri;
$item->title = $article_title;
$item->author = StripCDATA($element->find('dc:creator', 0)->innertext);
$item->timestamp = strtotime($element->find('dc:date', 0)->plaintext);
$item->content = $article_content;
$this->items[] = $item;
$limit++;
}
}
}
public function collectData(array $param)
{
function ExtractFromDelimiters($string, $start, $end)
{
if (strpos($string, $start) !== false) {
$section_retrieved = substr($string, strpos($string, $start) + strlen($start));
$section_retrieved = substr($section_retrieved, 0, strpos($section_retrieved, $end));
return $section_retrieved;
}
return false;
}
function StripWithDelimiters($string, $start, $end)
{
while (strpos($string, $start) !== false) {
$section_to_remove = substr($string, strpos($string, $start));
$section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
$string = str_replace($section_to_remove, '', $string);
}
return $string;
}
function CleanArticle($article_html)
{
$article_html = '<p>' . substr($article_html, strpos($article_html, '<p>') + 3);
$article_html = StripWithDelimiters($article_html, '<span class="credit">', '</span>');
$article_html = StripWithDelimiters($article_html, '<script', '</script>');
$article_html = StripWithDelimiters($article_html, '<div class="shortcode related-links', '</div>');
$article_html = StripWithDelimiters($article_html, '<a class="clickToEnlarge">', '</a>');
return $article_html;
}
if (!empty($param['topic'])) {
$this->topicName = $param['topic'];
}
$pageUrl = 'http://www.cnet.com/' . (empty($this->topicName) ? '' : 'topics/' . $this->topicName . '/');
$html = $this->file_get_html($pageUrl) or $this->returnError('Could not request CNET: ' . $pageUrl, 500);
$limit = 0;
foreach ($html->find('div.assetBody') as $element) {
if ($limit < 8) {
$article_title = trim($element->find('h2', 0)->plaintext);
$article_uri = 'http://www.cnet.com' . $element->find('a', 0)->href;
$article_thumbnail = $element->parent()->find('img', 0)->src;
$article_timestamp = strtotime($element->find('time.assetTime', 0)->plaintext);
$article_author = trim($element->find('a[rel=author]', 0)->plaintext);
if (!empty($article_title) && !empty($article_uri) && strpos($article_uri, '/news/') !== false) {
$article_html = $this->file_get_html($article_uri) or $this->returnError('Could not request CNET: ' . $article_uri, 500);
if (is_null($article_thumbnail)) {
$article_thumbnail = $article_html->find('div.originalImage', 0);
}
if (is_null($article_thumbnail)) {
$article_thumbnail = $article_html->find('span.imageContainer', 0);
}
if (is_object($article_thumbnail)) {
$article_thumbnail = $article_thumbnail->find('img', 0)->src;
}
$article_content = trim(CleanArticle(ExtractFromDelimiters($article_html, '<div class="articleContent', '<footer>')));
$item = new \Item();
$item->uri = $article_uri;
$item->thumbnailUri = $article_thumbnail;
$item->title = $article_title;
$item->author = $article_author;
$item->timestamp = $article_timestamp;
$item->content = $article_content;
$this->items[] = $item;
$limit++;
}
}
}
}
public function collectData(array $param)
{
function ExtractFromDelimiters($string, $start, $end)
{
if (strpos($string, $start) !== false) {
$section_retrieved = substr($string, strpos($string, $start) + strlen($start));
$section_retrieved = substr($section_retrieved, 0, strpos($section_retrieved, $end));
return $section_retrieved;
}
return false;
}
function StripWithDelimiters($string, $start, $end)
{
while (strpos($string, $start) !== false) {
$section_to_remove = substr($string, strpos($string, $start));
$section_to_remove = substr($section_to_remove, 0, strpos($section_to_remove, $end) + strlen($end));
$string = str_replace($section_to_remove, '', $string);
}
return $string;
}
function CleanArticle($article_html)
{
$article_html = '<p>' . substr($article_html, strpos($article_html, '</script></div><p>') + 18);
$article_html = StripWithDelimiters($article_html, '<script>', '</script>');
$article_html = StripWithDelimiters($article_html, '<div class="shortcode related-links', '</div>');
$article_html = StripWithDelimiters($article_html, '<a class="clickToEnlarge">', '</a>');
return $article_html;
}
if (!empty($param['topic'])) {
$this->topicName = $param['topic'];
}
$pageUrl = 'http://www.cnet.com/' . (empty($this->topicName) ? '' : 'topics/' . $this->topicName . '/');
$html = file_get_html($pageUrl) or $this->returnError('Could not request CNET: ' . $pageUrl, 500);
$limit = 0;
foreach ($html->find('div.socialSharingSmall') as $element) {
if ($limit < 4) {
$article_meta = json_decode(ExtractFromDelimiters($element->outertext, 'data-social-counts-options=\'', '\'>'));
$article_title = $article_meta->title;
$article_uri = $article_meta->url;
if (!empty($article_title) && !empty($article_uri) && strpos($article_uri, '/news/') !== false) {
$article_html = file_get_html($article_uri) or $this->returnError('Could not request CNET: ' . $article_uri, 500);
$article_timestamp = strtotime(ExtractFromDelimiters($article_html->innertext, '<time itemprop="datePublished" class="', '">'));
$article_thumbnail = $article_html->find('div.originalImage', 0);
if (is_null($article_thumbnail)) {
$article_thumbnail = $article_html->find('span.imageContainer', 0);
}
if (!is_null($article_thumbnail)) {
$article_thumbnail = $article_thumbnail->find('img', 0)->src;
}
$article_content = trim(CleanArticle(ExtractFromDelimiters($article_html, '</div></div></div><div class="col-8">', '<footer>')));
$article_author = trim($article_html->find('a.author', 0)->plaintext);
$item = new \Item();
$item->uri = $article_uri;
$item->thumbnailUri = $article_thumbnail;
$item->title = $article_title;
$item->author = $article_author;
$item->timestamp = $article_timestamp;
$item->content = $article_content;
$this->items[] = $item;
$limit++;
}
}
}
}
