forked from blallo/rss-bridge
Merge pull request #293 from ORelio/master
Add Nextgov & TheHackerNews, ZDNet insert img in article.
This commit is contained in:
Mitsu
GitHub
commit
8831ad068e
3 changed files with 216 additions and 7 deletions
110
bridges/NextgovBridge.php
Normal file
110
bridges/NextgovBridge.php
Normal file
|
|
@ -0,0 +1,110 @@
|
|||
<?php
|
||||
class NextgovBridge extends BridgeAbstract {
|
||||
|
||||
public function loadMetadatas() {
|
||||
|
||||
$this->maintainer = 'ORelio';
|
||||
$this->name = $this->getName();
|
||||
$this->uri = $this->getURI();
|
||||
$this->description = 'USA Federal technology news, best practices, and web 2.0 tools.';
|
||||
$this->update = '2016-07-20';
|
||||
|
||||
$this->parameters[] =
|
||||
'[
|
||||
{
|
||||
"name" : "Category",
|
||||
"type" : "list",
|
||||
"identifier" : "category",
|
||||
"values" :
|
||||
[
|
||||
{ "name" : "All", "value" : "all" },
|
||||
{ "name" : "Technology News", "value" : "technology-news" },
|
||||
{ "name" : "CIO Briefing", "value" : "cio-briefing" },
|
||||
{ "name" : "Emerging Tech", "value" : "emerging-tech" },
|
||||
{ "name" : "Cloud", "value" : "cloud-computing" },
|
||||
{ "name" : "Cybersecurity", "value" : "cybersecurity" },
|
||||
{ "name" : "Mobile", "value" : "mobile" },
|
||||
{ "name" : "Health", "value" : "health" },
|
||||
{ "name" : "Defense", "value" : "defense" },
|
||||
{ "name" : "Big Data", "value" : "big-data" }
|
||||
]
|
||||
}
|
||||
]';
|
||||
|
||||
}
|
||||
|
||||
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;
|
||||
}
|
||||
|
||||
$category = $param['category'];
|
||||
if (empty($category))
|
||||
$category = 'all';
|
||||
if ($category !== preg_replace('/[^a-z-]+/', '', $category) || strlen($category > 32))
|
||||
$this->returnError('Invalid "category" parameter.', 400);
|
||||
$url = $this->getURI().'rss/'.$category.'/';
|
||||
$html = $this->file_get_html($url) or $this->returnError('Could not request Nextgov: '.$url, 500);
|
||||
$limit = 0;
|
||||
|
||||
foreach ($html->find('item') as $element) {
|
||||
if ($limit < 10) {
|
||||
|
||||
$article_url = ExtractFromDelimiters($element->innertext, '<link>', '</link>');
|
||||
$article_author = ExtractFromDelimiters($element->innertext, 'dc/elements/1.1/">', '</dc:creator>');
|
||||
$article_title = $element->find('title', 0)->plaintext;
|
||||
$article_subtitle = $element->find('description', 0)->plaintext;
|
||||
$article_timestamp = strtotime($element->find('pubDate', 0)->plaintext);
|
||||
$article_thumbnail = ExtractFromDelimiters($element->innertext, '<media:content url="', '"');
|
||||
$article = $this->file_get_html($article_url) or $this->returnError('Could not request Nextgov: '.$article_url, 500);
|
||||
|
||||
$contents = $article->find('div.wysiwyg', 0)->innertext;
|
||||
$contents = StripWithDelimiters($contents, '<div class="ad-container">', '</div>');
|
||||
$contents = StripWithDelimiters($contents, '<div', '</div>'); //ad outer div
|
||||
$contents = StripWithDelimiters($contents, '<script', '</script>');
|
||||
$contents = ($article_thumbnail == '' ? '' : '<p><img src="'.$article_thumbnail.'" /></p>')
|
||||
.'<p><b>'.$article_subtitle.'</b></p>'
|
||||
.trim($contents);
|
||||
|
||||
if ($article_thumbnail == '')
|
||||
$article_thumbnail = 'http://cdn.nextgov.com/nextgov/images/logo.png';
|
||||
|
||||
$item = new \Item();
|
||||
$item->uri = $article_url;
|
||||
$item->title = $article_title;
|
||||
$item->author = $article_author;
|
||||
$item->thumbnailUri = $article_thumbnail;
|
||||
$item->timestamp = $article_timestamp;
|
||||
$item->content = $contents;
|
||||
$this->items[] = $item;
|
||||
$limit++;
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
public function getName() {
|
||||
return 'Nextgov Bridge';
|
||||
}
|
||||
|
||||
public function getURI() {
|
||||
return 'https://www.nextgov.com/';
|
||||
}
|
||||
|
||||
public function getCacheDuration() {
|
||||
return 3600; //1 hour
|
||||
}
|
||||
}
|
||||
90
bridges/TheHackerNewsBridge.php
Normal file
90
bridges/TheHackerNewsBridge.php
Normal file
|
|
@ -0,0 +1,90 @@
|
|||
<?php
|
||||
class TheHackerNewsBridge extends BridgeAbstract {
|
||||
|
||||
public function loadMetadatas() {
|
||||
|
||||
$this->maintainer = 'ORelio';
|
||||
$this->name = $this->getName();
|
||||
$this->uri = $this->getURI();
|
||||
$this->description = 'Cyber Security, Hacking, Technology News.';
|
||||
$this->update = '2016-07-22';
|
||||
|
||||
}
|
||||
|
||||
public function collectData(array $param) {
|
||||
|
||||
function StripWithDelimiters($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 $string;
|
||||
}
|
||||
|
||||
function StripRecursiveHTMLSection($string, $tag_name, $tag_start) {
|
||||
$open_tag = '<'.$tag_name;
|
||||
$close_tag = '</'.$tag_name.'>';
|
||||
$close_tag_length = strlen($close_tag);
|
||||
if (strpos($tag_start, $open_tag) === 0) {
|
||||
while (strpos($string, $tag_start) !== false) {
|
||||
$max_recursion = 100;
|
||||
$section_to_remove = null;
|
||||
$section_start = strpos($string, $tag_start);
|
||||
$search_offset = $section_start;
|
||||
do {
|
||||
$max_recursion--;
|
||||
$section_end = strpos($string, $close_tag, $search_offset);
|
||||
$search_offset = $section_end + $close_tag_length;
|
||||
$section_to_remove = substr($string, $section_start, $section_end - $section_start + $close_tag_length);
|
||||
$open_tag_count = substr_count($section_to_remove, $open_tag);
|
||||
$close_tag_count = substr_count($section_to_remove, $close_tag);
|
||||
} while ($open_tag_count > $close_tag_count && $max_recursion > 0);
|
||||
$string = str_replace($section_to_remove, '', $string);
|
||||
}
|
||||
}
|
||||
return $string;
|
||||
}
|
||||
|
||||
$html = $this->file_get_html($this->getURI()) or $this->returnError('Could not request TheHackerNews: '.$this->getURI(), 500);
|
||||
$limit = 0;
|
||||
|
||||
foreach ($html->find('article') as $element) {
|
||||
if ($limit < 5) {
|
||||
|
||||
$article_url = $element->find('a.entry-title', 0)->href;
|
||||
$article_author = trim($element->find('span.vcard', 0)->plaintext);
|
||||
$article_title = $element->find('a.entry-title', 0)->plaintext;
|
||||
$article_timestamp = strtotime($element->find('span.updated', 0)->plaintext);
|
||||
$article_thumbnail = $element->find('img', 0)->src;
|
||||
$article = $this->file_get_html($article_url) or $this->returnError('Could not request TheHackerNews: '.$article_url, 500);
|
||||
|
||||
$contents = $article->find('div.articlebodyonly', 0)->innertext;
|
||||
$contents = StripRecursiveHTMLSection($contents, 'div', '<div class=\'clear\'');
|
||||
$contents = StripWithDelimiters($contents, '<script', '</script>');
|
||||
|
||||
$item = new \Item();
|
||||
$item->uri = $article_url;
|
||||
$item->title = $article_title;
|
||||
$item->author = $article_author;
|
||||
$item->thumbnailUri = $article_thumbnail;
|
||||
$item->timestamp = $article_timestamp;
|
||||
$item->content = trim($contents);
|
||||
$this->items[] = $item;
|
||||
$limit++;
|
||||
}
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
public function getName() {
|
||||
return 'The Hacker News Bridge';
|
||||
}
|
||||
|
||||
public function getURI() {
|
||||
return 'https://thehackernews.com/';
|
||||
}
|
||||
|
||||
public function getCacheDuration() {
|
||||
return 3600; //1 hour
|
||||
}
|
||||
}
|
||||
|
|
@ -6,8 +6,8 @@ class ZDNetBridge extends BridgeAbstract {
|
|||
$this->maintainer = 'ORelio';
|
||||
$this->name = $this->getName();
|
||||
$this->uri = $this->getURI();
|
||||
$this->description = 'Returns the newest articles.';
|
||||
$this->update = '2016-07-18';
|
||||
$this->description = 'Technology News, Analysis, Comments and Product Reviews for IT Professionals.';
|
||||
$this->update = '2016-07-20';
|
||||
|
||||
$this->parameters[] =
|
||||
// http://www.zdnet.com/zdnet.opml
|
||||
|
|
@ -261,7 +261,7 @@ class ZDNetBridge extends BridgeAbstract {
|
|||
$thumbnail = $article->find('meta[itemprop=image]', 0);
|
||||
if (is_object($thumbnail))
|
||||
$thumbnail = $thumbnail->content;
|
||||
else $thumbnail = 'http://zdnet1.cbsistatic.com/fly/bundles/zdnetcss/images/logos/logo-192x192.png';
|
||||
else $thumbnail = '';
|
||||
|
||||
$contents = $article->find('article', 0)->innertext;
|
||||
foreach (array(
|
||||
|
|
@ -277,8 +277,17 @@ class ZDNetBridge extends BridgeAbstract {
|
|||
}
|
||||
$contents = StripWithDelimiters($contents, '<script', '</script>');
|
||||
$contents = StripWithDelimiters($contents, '<meta itemprop="image"', '>');
|
||||
$contents = StripWithDelimiters($contents, '<section class="sharethrough-top', '</section>');
|
||||
$contents = '<p><b>'.$article_subtitle.'</b></p>'.trim($contents);
|
||||
$contents = trim(StripWithDelimiters($contents, '<section class="sharethrough-top', '</section>'));
|
||||
$content_img = strpos($contents, '<img'); //Look for first image
|
||||
if (($content_img !== false && $content_img < 512) || $thumbnail == '')
|
||||
$content_img = ''; //Image already present on article beginning or no thumbnail
|
||||
else $content_img = '<p><img src="'.$thumbnail.'" /></p>'; //Include thumbnail
|
||||
$contents = $content_img
|
||||
.'<p><b>'.$article_subtitle.'</b></p>'
|
||||
.$contents;
|
||||
|
||||
if ($thumbnail == '')
|
||||
$thumbnail = 'http://zdnet1.cbsistatic.com/fly/bundles/zdnetcss/images/logos/logo-192x192.png';
|
||||
|
||||
$item = new \Item();
|
||||
$item->author = $author;
|
||||
|
|
@ -303,6 +312,6 @@ class ZDNetBridge extends BridgeAbstract {
|
|||
}
|
||||
|
||||
public function getCacheDuration() {
|
||||
return 3600;
|
||||
return 3600; //1 hour
|
||||
}
|
||||
}
|
||||
Loading…
Reference in a new issue