mirror of
https://github.com/RSS-Bridge/rss-bridge.git
synced 2025-01-29 11:58:23 +01:00
[ReutersBridge] Migrate to new API (#2348)
* [ReutersBridge] Migrate to new API - Add new API, feeds. - Old feed name are perserved for backward compatibility. - Remove 'Special Report' feed. - Some feed continue to use old Wire API due to not available in new one. - Add some new type of content, replace iframe with blockquote for twitter.
This commit is contained in:
csisoap
GitHub
parent
b395fe2641
commit
b9d92150e1
@ -3,11 +3,12 @@ class ReutersBridge extends BridgeAbstract
|
||||
{
|
||||
const MAINTAINER = 'hollowleviathan, spraynard, csisoap';
|
||||
const NAME = 'Reuters Bridge';
|
||||
const URI = 'https://reuters.com/';
|
||||
const URI = 'https://www.reuters.com';
|
||||
const CACHE_TIMEOUT = 1800; // 30min
|
||||
const DESCRIPTION = 'Returns news from Reuters';
|
||||
|
||||
private $feedName = self::NAME;
|
||||
private $useWireAPI = false;
|
||||
|
||||
/**
|
||||
* Wireitem types allowed in the final story output
|
||||
@ -32,40 +33,122 @@ class ReutersBridge extends BridgeAbstract
|
||||
'type' => 'list',
|
||||
'title' => 'Feeds from Reuters U.S/International edition',
|
||||
'values' => array(
|
||||
'Aerospace and Defense' => 'aerospace',
|
||||
'Business' => 'business',
|
||||
'China' => 'china',
|
||||
'Energy' => 'energy',
|
||||
'Entertainment' => 'chan:8ym8q8dl',
|
||||
'Environment' => 'chan:6u4f0jgs',
|
||||
'Fact Check' => 'chan:abtpk0vm',
|
||||
'Health' => 'chan:8hw7807a',
|
||||
'Lifestyle' => 'life',
|
||||
'Markets' => 'markets',
|
||||
'Politics' => 'politics',
|
||||
'Science' => 'science',
|
||||
'Special Reports' => 'special-reports',
|
||||
'Sports' => 'sports',
|
||||
'Tech' => 'tech',
|
||||
'Top News' => 'home/topnews',
|
||||
'UK' => 'chan:61leiu7j',
|
||||
'USA News' => 'us',
|
||||
'Fact Check' => 'chan:abtpk0vm',
|
||||
'Entertainment' => 'chan:8ym8q8dl',
|
||||
'Politics' => 'politics',
|
||||
'Wire' => 'wire',
|
||||
'World' => 'world',
|
||||
'Breakingviews' => '/breakingviews',
|
||||
'World' => array(
|
||||
'World' => 'world',
|
||||
'Africa' => '/world/africa',
|
||||
'Americas' => '/world/americas',
|
||||
'Asia-Pacific' => '/world/asia-pacific',
|
||||
'China' => 'china',
|
||||
'europe' => '/world/europe',
|
||||
'India' => '/world/india',
|
||||
'Middle East' => '/world/middle-east',
|
||||
'UK' => 'chan:61leiu7j',
|
||||
'USA News' => 'us',
|
||||
'The Great Reboot' => '/world/the-great-reboot',
|
||||
'Reuters Next' => '/world/reuters-next'
|
||||
),
|
||||
'Business' => array(
|
||||
'Business' => 'business',
|
||||
'Aerospace and Defense' => 'aerospace',
|
||||
'Autos Transportation' => '/business/autos-transportation',
|
||||
'Energy' => 'energy',
|
||||
'Finance' => '/business/finance',
|
||||
'Health' => 'chan:8hw7807a',
|
||||
'Media Telecom' => '/business/media-telecom',
|
||||
'Retail Consumer' => '/business/retail-consumer',
|
||||
'Sustainable Business' => '/business/sustainable-business',
|
||||
'Change Suite' => '/business/change-suite',
|
||||
'Future of Health' => '/business/future-of-health',
|
||||
'Future of Money' => '/business/future-of-money',
|
||||
'Take Five' => '/business/take-five',
|
||||
'Reuters Impact' => '/business/reuters-impact',
|
||||
),
|
||||
'Legal' => array(
|
||||
'Legal' => '/legal',
|
||||
'Government' => '/legal/government',
|
||||
'Legal Industry' => '/legal/legalindustry',
|
||||
'Litigation' => '/legal/litigation',
|
||||
'Transactional' => '/legal/transactional',
|
||||
),
|
||||
'Markets' => array(
|
||||
'Markets' => 'markets',
|
||||
'Asian Markets' => '/markets/asia',
|
||||
'Commodities' => '/markets/commodities',
|
||||
'Currencies' => '/markets/currencies',
|
||||
'Deals' => '/markets/deals',
|
||||
'European Markets' => '/markets/europe',
|
||||
'Funds' => '/markets/fund',
|
||||
'Global Market Data' => '/markets/global-market-data',
|
||||
'Rates & Bonds' => '/markets/rates-bonds',
|
||||
'Stocks' => '/markets/stocks',
|
||||
'U.S Markets' => '/markets/us',
|
||||
'Wealth' => '/markets/wealth',
|
||||
'Macro Matters' => '/markets/macromatters',
|
||||
),
|
||||
'Technology' => array(
|
||||
'Technology' => 'tech',
|
||||
'Disrupted' => '/technology/disrupted',
|
||||
'Reuters Momentum' => '/technology/reuters-momentum',
|
||||
),
|
||||
'Sports' => array(
|
||||
'Sports' => 'sports',
|
||||
'Athletics' => '/lifestyle/sports/athletics',
|
||||
'Cricket' => '/lifestyle/sports/cricket',
|
||||
'Cycling' => '/lifestyle/sports/cycling',
|
||||
'Golf' => '/lifestyle/sports/golf',
|
||||
'Motor Sports' => '/lifestyle/sports/motor-sports',
|
||||
'Soccer' => '/lifestyle/sports/soccer',
|
||||
'Tennis' => '/lifestyle/sports/tennis',
|
||||
),
|
||||
'Lifestyle' => array(
|
||||
'Lifestyle' => 'life',
|
||||
'Oddly Enough' => '/lifestyle/oddly-enough',
|
||||
'Science' => 'science',
|
||||
)
|
||||
)
|
||||
)
|
||||
)
|
||||
);
|
||||
|
||||
const BACKWARD_COMPATIBILITY = array(
|
||||
'world' => '/world',
|
||||
'china' => '/world/china',
|
||||
'chan:61leiu7j' => '/world/uk',
|
||||
'us' => '/world/us',
|
||||
'business' => '/business',
|
||||
'aerospace' => '/business/aerospace-defense',
|
||||
'energy' => '/business/energy',
|
||||
'environment' => '/business/environment',
|
||||
'chan:8hw7807a' => '/business/healthcare-pharmaceuticals',
|
||||
'markets' => '/markets',
|
||||
'tech' => '/technology',
|
||||
'sports' => '/lifestyle/sports',
|
||||
'life' => '/lifestyle',
|
||||
'science' => '/lifestyle/science',
|
||||
);
|
||||
|
||||
const OLD_WIRE_SECTION = array(
|
||||
'home/topnews',
|
||||
'chan:abtpk0vm',
|
||||
'chan:8ym8q8dl',
|
||||
'politics',
|
||||
'wire'
|
||||
);
|
||||
|
||||
/**
|
||||
* Performs an HTTP request to the Reuters API and returns decoded JSON
|
||||
* in the form of an associative array
|
||||
* @param string $feed_uri Parameter string to the Reuters API
|
||||
* @param string $feed_uri Full API URL to fetch data
|
||||
* @return array
|
||||
*/
|
||||
private function getJson($feed_uri)
|
||||
private function getJson($uri)
|
||||
{
|
||||
$uri = "https://wireapi.reuters.com/v8$feed_uri";
|
||||
$returned_data = getContents($uri);
|
||||
return json_decode($returned_data, true);
|
||||
}
|
||||
@ -117,19 +200,94 @@ class ReutersBridge extends BridgeAbstract
|
||||
return $reuters_wireitem_templates;
|
||||
}
|
||||
|
||||
private function getSectionEndpoint() {
|
||||
$endpoint = $this->getInput('feed');
|
||||
if(isset(self::BACKWARD_COMPATIBILITY[$endpoint])) {
|
||||
$endpoint = self::BACKWARD_COMPATIBILITY[$endpoint];
|
||||
} elseif (in_array($endpoint, self::OLD_WIRE_SECTION)) {
|
||||
$this->useWireAPI = true;
|
||||
}
|
||||
return $endpoint;
|
||||
}
|
||||
|
||||
/**
|
||||
* @param string $endpoint - Provide section's endpoint to Reuters API.
|
||||
* @param string $fetch_type - Provide what kind of fetch do you want? Article or Section.
|
||||
* @return string A completed API URL to fetch data
|
||||
*/
|
||||
private function getAPIURL($endpoint, $fetch_type) {
|
||||
$base_url = self::URI . '/pf/api/v3/content/fetch/';
|
||||
$wire_url = 'https://wireapi.reuters.com/v8';
|
||||
switch($fetch_type) {
|
||||
case 'article':
|
||||
if($this->useWireAPI) {
|
||||
return $wire_url . $endpoint;
|
||||
}
|
||||
$query = array(
|
||||
'website_url' => $endpoint,
|
||||
'website' => 'reuters'
|
||||
);
|
||||
$json_query = json_encode($query);
|
||||
return $base_url . 'article-by-id-or-url-v1?query=' . $json_query;
|
||||
break;
|
||||
case 'section':
|
||||
if($this->useWireAPI) {
|
||||
if(strpos($endpoint, 'chan:') !== false) {
|
||||
// Now checking whether that feed has unique ID or not.
|
||||
$feed_uri = "/feed/rapp/us/wirefeed/$endpoint";
|
||||
} else {
|
||||
$feed_uri = "/feed/rapp/us/tabbar/feeds/$endpoint";
|
||||
}
|
||||
return $wire_url . $feed_uri;
|
||||
}
|
||||
$query = array(
|
||||
'fetch_type' => 'section',
|
||||
'section_id' => $endpoint,
|
||||
'size' => 30,
|
||||
'website' => 'reuters'
|
||||
);
|
||||
$json_query = json_encode($query);
|
||||
return $base_url . 'articles-by-section-alias-or-id-v1?query=' . $json_query;
|
||||
break;
|
||||
}
|
||||
returnServerError('unsupported endpoint');
|
||||
}
|
||||
|
||||
private function getArticle($feed_uri)
|
||||
{
|
||||
// This will make another request to API to get full detail of article and author's name.
|
||||
$rawData = $this->getJson($feed_uri);
|
||||
$reuters_wireitems = $rawData['wireitems'];
|
||||
$processedData = $this->processData($reuters_wireitems);
|
||||
$url = $this->getAPIURL($feed_uri, 'article');
|
||||
$rawData = $this->getJson($url);
|
||||
$article_content = '';
|
||||
$authorlist = '';
|
||||
$category = array();
|
||||
$image_list = array();
|
||||
$published_at = '';
|
||||
if($this->useWireAPI) {
|
||||
$reuters_wireitems = $rawData['wireitems'];
|
||||
$processedData = $this->processData($reuters_wireitems);
|
||||
|
||||
$first = reset($processedData);
|
||||
$article_content = $first['story']['body_items'];
|
||||
$authorlist = $first['story']['authors'];
|
||||
$category = $first['story']['channel']['name'];
|
||||
$image_list = $first['story']['images'];
|
||||
$published_at = $first['story']['published_at'];
|
||||
$first = reset($processedData);
|
||||
$article_content = $first['story']['body_items'];
|
||||
$authorlist = $first['story']['authors'];
|
||||
$category = array($first['story']['channel']['name']);
|
||||
$image_list = $first['story']['images'];
|
||||
$published_at = $first['story']['published_at'];
|
||||
} else {
|
||||
$article_content = $rawData['result']['content_elements'];
|
||||
$authorlist = $rawData['result']['authors'];
|
||||
$category = array($rawData['result']['taxonomy']['ads_primary_section']['name']);
|
||||
$image_list = array();
|
||||
if(!empty($rawData['result']['related_content']['galleries'])) {
|
||||
$galleries = $rawData['result']['related_content']['galleries'];
|
||||
foreach($galleries as $gallery) {
|
||||
$image_list = array_merge($image_list, $gallery['content_elements']);
|
||||
}
|
||||
} else if(!empty($rawData['result']['related_content']['images'])) {
|
||||
$image_list = $rawData['result']['related_content']['images'];
|
||||
}
|
||||
$published_at = $rawData['result']['published_time'];
|
||||
}
|
||||
|
||||
$content_detail = array(
|
||||
'content' => $this->handleArticleContent($article_content),
|
||||
@ -161,15 +319,14 @@ class ReutersBridge extends BridgeAbstract
|
||||
$counter = 0;
|
||||
foreach ($authors as $data) {
|
||||
//Formatting author's name.
|
||||
$counter++;
|
||||
$name = $data['name'];
|
||||
if ($counter == count($authors)) {
|
||||
$author = $author . $name;
|
||||
$counter++;
|
||||
if($counter == count($authors)) {
|
||||
$author .= $name;
|
||||
} else {
|
||||
$author = $author . "$name, ";
|
||||
$author .= $name . ', ';
|
||||
}
|
||||
}
|
||||
|
||||
return $author;
|
||||
}
|
||||
|
||||
@ -211,19 +368,16 @@ class ReutersBridge extends BridgeAbstract
|
||||
$embed = '';
|
||||
switch ($media_type) {
|
||||
case 'tweet':
|
||||
$url = "https://platform.twitter.com/embed/Tweet.html?id=$cid";
|
||||
$embed .= <<<EOD
|
||||
<iframe
|
||||
src="{$url}"
|
||||
title="Twitter Tweet"
|
||||
scrolling="no"
|
||||
frameborder="0"
|
||||
allowtransparency="true"
|
||||
allowfullscreen="true"
|
||||
style="width: 550px;height: 225px;"
|
||||
>
|
||||
</iframe>
|
||||
EOD;
|
||||
try {
|
||||
$tweet_url = "https://twitter.com/dummyname/statuses/$cid";
|
||||
$get_embed_url = 'https://publish.twitter.com/oembed?url='
|
||||
. urlencode($tweet_url) .
|
||||
'&partner=&hide_thread=false';
|
||||
$oembed_json = json_decode(getContents($get_embed_url), true);
|
||||
$embed .= $oembed_json['html'];
|
||||
} catch (Exception $e) { // In case not found any tweet.
|
||||
$embed .= '';
|
||||
}
|
||||
break;
|
||||
case 'instagram':
|
||||
$url = "https://instagram.com/p/$cid/media/?size=l";
|
||||
@ -250,6 +404,34 @@ EOD;
|
||||
}
|
||||
$description .= $embed;
|
||||
break;
|
||||
case 'social_media':
|
||||
if ($content['sub_type'] == 'twitter') {
|
||||
$description .= $content['html'];
|
||||
}
|
||||
break;
|
||||
case 'table':
|
||||
$table = '<table>';
|
||||
$theaders = $content['header'];
|
||||
$tr = '<tr>';
|
||||
foreach($theaders as $header) {
|
||||
$tr .= '<th>' . $header . '</th>';
|
||||
}
|
||||
$tr .= '</tr>';
|
||||
$table .= $tr;
|
||||
$rows = $content['rows'];
|
||||
foreach($rows as $row) {
|
||||
$tr = '<tr>';
|
||||
foreach($row as $data) {
|
||||
$tr .= '<td>' . $data . '</td>';
|
||||
}
|
||||
$tr .= '</tr>';
|
||||
$table .= $tr;
|
||||
}
|
||||
$table .= '</table>';
|
||||
$description .= $table;
|
||||
break;
|
||||
case 'image':
|
||||
$description .= $this->handleImage(array($content));
|
||||
}
|
||||
}
|
||||
|
||||
@ -262,47 +444,72 @@ EOD;
|
||||
|
||||
public function collectData()
|
||||
{
|
||||
$reuters_feed_name = $this->getInput('feed');
|
||||
$endpoint = $this->getSectionEndpoint();
|
||||
$url = $this->getAPIURL($endpoint, 'section');
|
||||
$data = $this->getJson($url);
|
||||
|
||||
if(strpos($reuters_feed_name, 'chan:') !== false) {
|
||||
// Now checking whether that feed has unique ID or not.
|
||||
$feed_uri = "/feed/rapp/us/wirefeed/$reuters_feed_name";
|
||||
$stories = array();
|
||||
$section_name = '';
|
||||
if($this->useWireAPI) {
|
||||
$reuters_wireitems = $data['wireitems'];
|
||||
$section_name = $data['wire_name'];
|
||||
$processedData = $this->processData($reuters_wireitems);
|
||||
|
||||
// Merge all articles from Editor's Highlight section into existing array of templates.
|
||||
$top_section = reset($processedData);
|
||||
if ($top_section['type'] == 'headlines') {
|
||||
$top_section = array_shift($processedData);
|
||||
$articles = $top_section['headlines'];
|
||||
$processedData = array_merge($articles, $processedData);
|
||||
}
|
||||
$stories = $processedData;
|
||||
} else {
|
||||
$feed_uri = "/feed/rapp/us/tabbar/feeds/$reuters_feed_name";
|
||||
}
|
||||
|
||||
$data = $this->getJson($feed_uri);
|
||||
|
||||
$reuters_wireitems = $data['wireitems'];
|
||||
$this->feedName = $data['wire_name'] . ' | Reuters';
|
||||
$processedData = $this->processData($reuters_wireitems);
|
||||
|
||||
// Merge all articles from Editor's Highlight section into existing array of templates.
|
||||
$top_section = reset($processedData);
|
||||
if ($top_section['type'] == 'headlines') {
|
||||
$top_section = array_shift($processedData);
|
||||
$articles = $top_section['headlines'];
|
||||
$processedData = array_merge($articles, $processedData);
|
||||
}
|
||||
|
||||
foreach ($processedData as $story) {
|
||||
$item['uid'] = $story['story']['usn'];
|
||||
$article_uri = $story['template_action']['api_path'];
|
||||
$content_detail = $this->getArticle($article_uri);
|
||||
$description = $content_detail['content'];
|
||||
$author = $content_detail['author'];
|
||||
$images = $content_detail['images'];
|
||||
$item['categories'] = array($content_detail['category']);
|
||||
$item['author'] = $author;
|
||||
if (!(bool) $description) {
|
||||
$description = $story['story']['lede']; // Just in case the content doesn't have anything.
|
||||
$section_name = $data['result']['section']['name'];
|
||||
if(isset($data['arcResult']['articles'])) {
|
||||
$stories = $data['arcResult']['articles'];
|
||||
} else {
|
||||
$item['content'] = "$description $images";
|
||||
$stories = $data['result']['articles'];
|
||||
}
|
||||
}
|
||||
$this->feedName = $section_name . ' | Reuters';
|
||||
|
||||
foreach ($stories as $story) {
|
||||
$uid = '';
|
||||
$author = '';
|
||||
$category = array();
|
||||
$content = '';
|
||||
$title = '';
|
||||
$timestamp = '';
|
||||
$url = '';
|
||||
$article_uri = '';
|
||||
if($this->useWireAPI) {
|
||||
$uid = $story['story']['usn'];
|
||||
$article_uri = $story['template_action']['api_path'];
|
||||
$title = $story['story']['hed'];
|
||||
$url = $story['template_action']['url'];
|
||||
} else {
|
||||
$uid = $story['id'];
|
||||
$url = self::URI . $story['canonical_url'];
|
||||
$title = $story['title'];
|
||||
$article_uri = $story['canonical_url'];
|
||||
}
|
||||
|
||||
$item['title'] = $story['story']['hed'];
|
||||
$item['timestamp'] = $content_detail['published_at'];
|
||||
$item['uri'] = $story['template_action']['url'];
|
||||
$content_detail = $this->getArticle($article_uri);
|
||||
$description = $content_detail['content'];
|
||||
$description = defaultLinkTo($description, $this->getURI());
|
||||
$author = $content_detail['author'];
|
||||
$images = $content_detail['images'];
|
||||
$category = $content_detail['category'];
|
||||
$content = "$description $images";
|
||||
$timestamp = $content_detail['published_at'];
|
||||
|
||||
$item['uid'] = $uid;
|
||||
$item['categories'] = $category;
|
||||
$item['author'] = $author;
|
||||
$item['content'] = $content;
|
||||
$item['title'] = $title;
|
||||
$item['timestamp'] = $timestamp;
|
||||
$item['uri'] = $url;
|
||||
$this->items[] = $item;
|
||||
}
|
||||
}
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user