Added Straits Times ・ 0fb35ce ・ Gitprep

- Added Straits Times;
- Browse files
- Yanik Cawidrone commited on 2019-03-22
- 1 parent ab25467
  
  commit 0fb35ce0c3aed81d32d7a09d8a44ba9fbf5d6b3f

Showing 3 changed files with 232 additions and 2 deletions

+112

sources/news-stasia.php

...	...	@@ -0,0 +1,112 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_STASIA);
	5	+$orgStrings = array( 'media:thumbnail');
	6	+$newStrings = array( 'mediacontent');
	7	+$rss_content = str_replace($orgStrings, $newStrings, $rss_content);
	8	+$xml = simplexml_load_string($rss_content);
	9	+if ($xml === false) {
	10	+ echo 'Failed to read RSS';
	11	+} else {
	12	+ $channel = array();
	13	+ $channel['title'] = $xml->channel->title;
	14	+ $channel['link'] = $xml->channel->link;
	15	+ $channel['description'] = $xml->channel->description;
	16	+ $channel['pubDate'] = $xml->channel->pubDate;
	17	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	18	+ echo '<h4>' . $channel['title'] . '</h4>';
	19	+ $cpt=0;
	20	+ foreach ($xml->channel->item as $item) {
	21	+ $article = array();
	22	+ $article['title'] = $item->title;
	23	+ $article['link'] = $item->link;
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ //$article['image'] = $item->mediacontent['url'];
	28	+ $articles[$cpt] = $article;
	29	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	30	+// echo '<img id="img-list" src="'.$article['image'].'" style="display:block;"><br>';
	31	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	33	+ echo $article['title'].'  ';
	34	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	35	+ $cpt++;
	36	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+
	42	+echo '</div><!-- ./col-4 -->'.PHP_EOL;
	43	+echo '<div id="article-display" class="col-6">'.PHP_EOL;
	44	+echo '<a name="article-top"></a><div id="article-current"></div>'.PHP_EOL;
	45	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	46	+
	47	+$cpt=0;
	48	+foreach ($articles as $article ) {
	49	+ $cpt_prev=$cpt-1;
	50	+ $cpt_next=$cpt+1;
	51	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	52	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	53	+ echo "<hr>";
	54	+ echo "<a name=\"article-$cpt\">";
	55	+ $article_content = file_get_contents($article['link']);
	56	+ $doc = new DOMDocument();
	57	+ $doc->preserveWhiteSpace = false;
	58	+ $doc->formatOutput = true;
	59	+ $libxml_previous_state = libxml_use_internal_errors(true);
	60	+ //Hack
	61	+ $orgStrings = array('itemprop="articleBody"');
	62	+ $newStrings = array('id="articleBody"');
	63	+ $article_content = str_replace($orgStrings, $newStrings, $article_content);
	64	+ $doc->loadHTML($article_content);
	65	+ libxml_clear_errors();
	66	+ libxml_use_internal_errors($libxml_previous_state);
	67	+ $articleBody = $doc->getElementByID('articleBody');
	68	+ if(NULL!= $article) {
	69	+ $article_only = DOMinnerHTML($articleBody);
	70	+ }
	71	+
	72	+ $SEARCH_SUB='<div class="c-save-for-later--default is-article-bottom-position"';
	73	+ $pos_start=strpos($article_only, $SEARCH_SUB);
	74	+ if($pos_start) {
	75	+ $article_only = substr($article_only,0,$pos_start);
	76	+ }
	77	+ //Extract image
	78	+ //$re = '/<link rel="image_src" href="(.+?)"\/>/';
	79	+ $re = '/<div class="media-entity" rel="" resource="(.+?)">/';
	80	+ $imageurl="";
	81	+ preg_match($re, $article_content, $matches);
	82	+ if(isset($matches[1]))
	83	+ $imageurl = $matches[1];
	84	+
	85	+
	86	+ //Clean some sections
	87	+ //$article_only = re_remove($article_only, '//');
	88	+ $article_only = re_remove($article_only, '/googletag\.display$"(.+?)"$;/');
	89	+
	90	+ //Finally remove empty lines
	91	+ $article_only = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	92	+
	93	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	94	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	95	+ echo '<button id="html-btn" onclick="printHTML()" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button>'.PHP_EOL;
	96	+ echo '<button id="pdf-btn" onclick="printPDF()" style="display:inline;"><img src="img/pdf.png" width="24px" height="24px"></button>'.PHP_EOL;
	97	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	98	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
	99	+ echo '<div class="extract-content" id="'.$cpt.'"><img src="'.$imageurl.'">'.$article_only.'</div>'.PHP_EOL;
	100	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	101	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     '.PHP_EOL;
	102	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	103	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	104	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
	105	+ $cpt++;
	106	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	107	+ break;
	108	+ }
	109	+}
	110	+?>
	111	+
	112	+

+112

sources/news-stsg.php

View

...	...	@@ -0,0 +1,112 @@
	1	+<?php
	2	+include_once( 'news-constants.php' );
	3	+
	4	+$rss_content = file_get_contents(NEWS_RSS_STSG);
	5	+$orgStrings = array( 'media:thumbnail');
	6	+$newStrings = array( 'mediacontent');
	7	+$rss_content = str_replace($orgStrings, $newStrings, $rss_content);
	8	+$xml = simplexml_load_string($rss_content);
	9	+if ($xml === false) {
	10	+ echo 'Failed to read RSS';
	11	+} else {
	12	+ $channel = array();
	13	+ $channel['title'] = $xml->channel->title;
	14	+ $channel['link'] = $xml->channel->link;
	15	+ $channel['description'] = $xml->channel->description;
	16	+ $channel['pubDate'] = $xml->channel->pubDate;
	17	+ $channel['timestamp'] = strtotime($xml->channel->pubDate);
	18	+ echo '<h4>' . $channel['title'] . '</h4>';
	19	+ $cpt=0;
	20	+ foreach ($xml->channel->item as $item) {
	21	+ $article = array();
	22	+ $article['title'] = $item->title;
	23	+ $article['link'] = $item->link;
	24	+ $article['pubDate'] = $item->pubDate;
	25	+ $article['timestamp'] = strtotime($item->pubDate);
	26	+ $article['description'] = $item->description;
	27	+ //$article['image'] = $item->mediacontent['url'];
	28	+ $articles[$cpt] = $article;
	29	+ echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
	30	+// echo '<img id="img-list" src="'.$article['image'].'" style="display:block;"><br>';
	31	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>  ';
	32	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div> ';
	33	+ echo $article['title'].'  ';
	34	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
	35	+ $cpt++;
	36	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	37	+ break;
	38	+ }
	39	+ }
	40	+}
	41	+
	42	+echo '</div><!-- ./col-4 -->'.PHP_EOL;
	43	+echo '<div id="article-display" class="col-6">'.PHP_EOL;
	44	+echo '<a name="article-top"></a><div id="article-current"></div>'.PHP_EOL;
	45	+echo '</div><!-- ./col-6 -->'.PHP_EOL;
	46	+
	47	+$cpt=0;
	48	+foreach ($articles as $article ) {
	49	+ $cpt_prev=$cpt-1;
	50	+ $cpt_next=$cpt+1;
	51	+ echo '<!-- ==================== article '.$cpt.'============== -->';
	52	+ echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
	53	+ echo "<hr>";
	54	+ echo "<a name=\"article-$cpt\">";
	55	+ $article_content = file_get_contents($article['link']);
	56	+ $doc = new DOMDocument();
	57	+ $doc->preserveWhiteSpace = false;
	58	+ $doc->formatOutput = true;
	59	+ $libxml_previous_state = libxml_use_internal_errors(true);
	60	+ //Hack
	61	+ $orgStrings = array('itemprop="articleBody"');
	62	+ $newStrings = array('id="articleBody"');
	63	+ $article_content = str_replace($orgStrings, $newStrings, $article_content);
	64	+ $doc->loadHTML($article_content);
	65	+ libxml_clear_errors();
	66	+ libxml_use_internal_errors($libxml_previous_state);
	67	+ $articleBody = $doc->getElementByID('articleBody');
	68	+ if(NULL!= $article) {
	69	+ $article_only = DOMinnerHTML($articleBody);
	70	+ }
	71	+
	72	+ $SEARCH_SUB='<div class="c-save-for-later--default is-article-bottom-position"';
	73	+ $pos_start=strpos($article_only, $SEARCH_SUB);
	74	+ if($pos_start) {
	75	+ $article_only = substr($article_only,0,$pos_start);
	76	+ }
	77	+ //Extract image
	78	+ //$re = '/<link rel="image_src" href="(.+?)"\/>/';
	79	+ $re = '/<div class="media-entity" rel="" resource="(.+?)">/';
	80	+ $imageurl="";
	81	+ preg_match($re, $article_content, $matches);
	82	+ if(isset($matches[1]))
	83	+ $imageurl = $matches[1];
	84	+
	85	+
	86	+ //Clean some sections
	87	+ //$article_only = re_remove($article_only, '//');
	88	+ $article_only = re_remove($article_only, '/googletag\.display$"(.+?)"$;/');
	89	+
	90	+ //Finally remove empty lines
	91	+ $article_only = preg_replace("/(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+/", "\n", $article_only);
	92	+
	93	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	94	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	95	+ echo '<button id="html-btn" onclick="printHTML()" style="display:inline;"><img src="img/html5.png" width="24px" height="24px"></button>'.PHP_EOL;
	96	+ echo '<button id="pdf-btn" onclick="printPDF()" style="display:inline;"><img src="img/pdf.png" width="24px" height="24px"></button>'.PHP_EOL;
	97	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	98	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
	99	+ echo '<div class="extract-content" id="'.$cpt.'"><img src="'.$imageurl.'">'.$article_only.'</div>'.PHP_EOL;
	100	+ echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>     '.PHP_EOL;
	101	+ echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>     '.PHP_EOL;
	102	+ echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>     '.PHP_EOL;
	103	+ echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>     '.PHP_EOL;
	104	+ echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
	105	+ $cpt++;
	106	+ if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
	107	+ break;
	108	+ }
	109	+}
	110	+?>
	111	+
	112	+

+8 -2

sources/rss.php

View

@@ -7,6 +7,8 @@ define('NEWS_RSS_BBC', 'http://feeds.bbci.co.uk/news/rss.xml');
 define('NEWS_RSS_AJE', 'https://www.aljazeera.com/xml/rss/all.xml');
 define('NEWS_RSS_SMH', 'https://www.smh.com.au/rss/world.xml');
 define('NEWS_RSS_SCMP', 'https://www.scmp.com/rss/91/feed');
+define('NEWS_RSS_STSG', 'https://www.straitstimes.com/news/singapore/rss.xml');
+define('NEWS_RSS_STASIA', 'https://www.straitstimes.com/news/asia/rss.xml');
 //define('NEWS_RSS_STAR', '');
 
 $array_title=array(
@@ -16,7 +18,9 @@ $array_title=array(
   "bbc" => "BBC Top Stories",
   "aje" => "AlJazeera English",
   "smh" => "Sydney Morning Herald",
-  "scmp" => "South China Morning Post"
+  "scmp" => "South China Morning Post",
+  "stsg" => "ST Singapore",
+  "stasia" => "ST Asia"
 );
 $array_url=array(
   "nyt" => NEWS_RSS_NYT,
@@ -25,7 +29,9 @@ $array_url=array(
   "bbc" => NEWS_RSS_BBC,
   "aje" => NEWS_RSS_AJE,
   "smh" => NEWS_RSS_SMH,
-  "scmp" => NEWS_RSS_SCMP
+  "scmp" => NEWS_RSS_SCMP,
+  "stsg" => NEWS_RSS_STSG,
+  "stasia" => NEWS_RSS_STASIA
 );
 
 function DEBUG($Message) {

...	...	@@ -7,6 +7,8 @@ define('NEWS_RSS_BBC', 'http://feeds.bbci.co.uk/news/rss.xml');
7	7	define('NEWS_RSS_AJE', 'https://www.aljazeera.com/xml/rss/all.xml');
8	8	define('NEWS_RSS_SMH', 'https://www.smh.com.au/rss/world.xml');
9	9	define('NEWS_RSS_SCMP', 'https://www.scmp.com/rss/91/feed');
	10	+define('NEWS_RSS_STSG', 'https://www.straitstimes.com/news/singapore/rss.xml');
	11	+define('NEWS_RSS_STASIA', 'https://www.straitstimes.com/news/asia/rss.xml');
10	12	//define('NEWS_RSS_STAR', '');
11	13
12	14	$array_title=array(
...	...	@@ -16,7 +18,9 @@ $array_title=array(
16	18	"bbc" => "BBC Top Stories",
17	19	"aje" => "AlJazeera English",
18	20	"smh" => "Sydney Morning Herald",
19		- "scmp" => "South China Morning Post"
	21	+ "scmp" => "South China Morning Post",
	22	+ "stsg" => "ST Singapore",
	23	+ "stasia" => "ST Asia"
20	24	);
21	25	$array_url=array(
22	26	"nyt" => NEWS_RSS_NYT,
...	...	@@ -25,7 +29,9 @@ $array_url=array(
25	29	"bbc" => NEWS_RSS_BBC,
26	30	"aje" => NEWS_RSS_AJE,
27	31	"smh" => NEWS_RSS_SMH,
28		- "scmp" => NEWS_RSS_SCMP
	32	+ "scmp" => NEWS_RSS_SCMP,
	33	+ "stsg" => NEWS_RSS_STSG,
	34	+ "stasia" => NEWS_RSS_STASIA
29	35	);
30	36
31	37	function DEBUG($Message) {