Silence loadHTML error ・ 3825201 ・ Gitprep

- Silence loadHTML error;
- Browse files
- ycawidro commited on 2018-12-07
- 1 parent 7aaabf0
  
  commit 3825201d8b6eb1b50d91a1939675d0fd59c1392b

Showing 9 changed files with 33 additions and 5 deletions

+8 -4

sources/news-conspiracy.php

@@ -28,11 +28,15 @@ if ($xml === false) {
     //Extract image
     $re = '/<img class="size-medium wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
     preg_match($re, $article['content'], $matches);
-    $article['image'] = $matches[2];
-    if ( 0 == strlen($article['image']) ) {
-      $re = '/<img class="aligncenter wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
-      preg_match($re, $article['content'], $matches);
+    if( isset($matches[2]) ) {
       $article['image'] = $matches[2];
+      if ( 0 == strlen($article['image']) ) {
+        $re = '/<img class="aligncenter wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
+        preg_match($re, $article['content'], $matches);
+        $article['image'] = $matches[2];
+      }
+    } else {
+      $article['image'] = "";
     }
 
     $articles[$cpt]=$article;

sources/news-franceinfo.php

View

@@ -57,7 +57,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   $articles = $doc->getElementsByTagName('article');
   $article_only="";
   foreach ( $articles as $node) {

sources/news-huffingtonpost.fr.php

View

@@ -72,7 +72,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   $articles = $doc->getElementsByTagName('article');
   foreach ( $articles as $node) {
     $article_only = DOMinnerHTML($node);

sources/news-lemonde.fr.php

View

@@ -57,7 +57,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   $articles = $doc->getElementsByTagName('article');
   $article_only="";
   if( isset($articles[0]) ) {
@@ -82,7 +85,10 @@ foreach ($articles as $article ) {
     $doc = new DOMDocument();
     $doc->preserveWhiteSpace = false;
     $doc->formatOutput       = true;
+    $libxml_previous_state = libxml_use_internal_errors(true);
     $doc->loadHTML($article_content);
+    libxml_clear_errors();
+    libxml_use_internal_errors($libxml_previous_state);
     $articles = $doc->getElementsByTagName('article');
     $article_only="";
     if( isset($articles[0]) ) {

sources/news-lepoint.fr.php

View

@@ -5,6 +5,7 @@ $rss_content = http_get_contents(NEWS_RSS_LEPOINTFR);
 $xml = simplexml_load_string($rss_content);
 if ($xml === false) {
   echo 'Failed to read RSS';
+  exit(0);
 } else {
   $channel = array();
   $channel['title'] = $xml->channel->title;

+3 -1

sources/news-lesinrocks.fr.php

View

@@ -58,8 +58,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
-  //$doc->loadHTML($article_content);
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content_utf8);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   $articles = $doc->getElementsByTagName('article');
   $article_only="";
   foreach ( $articles as $node) {

sources/news-lexpress.fr.php

View

@@ -90,7 +90,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   //$articles = $doc->getElementsByTagName('article_container');
   $articles = $doc->getElementsByTagName('div');
   $article_only="";

sources/news-liberation.fr.php

View

@@ -69,7 +69,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content_utf8);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   $articles = $doc->getElementsByTagName('article');
   $article_only="";
   foreach ( $articles as $node) {

sources/news-slate.fr.php

View

@@ -58,7 +58,10 @@ foreach ($articles as $article ) {
   $doc = new DOMDocument();
   $doc->preserveWhiteSpace = false;
   $doc->formatOutput       = true;
+  $libxml_previous_state = libxml_use_internal_errors(true);
   $doc->loadHTML($article_content);
+  libxml_clear_errors();
+  libxml_use_internal_errors($libxml_previous_state);
   $articles = $doc->getElementsByTagName('article');
   $article_only="";
   foreach ( $articles as $node) {

...	...	@@ -28,11 +28,15 @@ if ($xml === false) {
28	28	//Extract image
29	29	$re = '/<img class="size-medium wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
30	30	preg_match($re, $article['content'], $matches);
31		- $article['image'] = $matches[2];
32		- if ( 0 == strlen($article['image']) ) {
33		- $re = '/<img class="aligncenter wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
34		- preg_match($re, $article['content'], $matches);
	31	+ if( isset($matches[2]) ) {
35	32	$article['image'] = $matches[2];
	33	+ if ( 0 == strlen($article['image']) ) {
	34	+ $re = '/<img class="aligncenter wp-image-(.+?)" src="(.+?)" alt=(.+?)>/';
	35	+ preg_match($re, $article['content'], $matches);
	36	+ $article['image'] = $matches[2];
	37	+ }
	38	+ } else {
	39	+ $article['image'] = "";
36	40	}
37	41
38	42	$articles[$cpt]=$article;

...	...	@@ -57,7 +57,10 @@ foreach ($articles as $article ) {
57	57	$doc = new DOMDocument();
58	58	$doc->preserveWhiteSpace = false;
59	59	$doc->formatOutput = true;
	60	+ $libxml_previous_state = libxml_use_internal_errors(true);
60	61	$doc->loadHTML($article_content);
	62	+ libxml_clear_errors();
	63	+ libxml_use_internal_errors($libxml_previous_state);
61	64	$articles = $doc->getElementsByTagName('article');
62	65	$article_only="";
63	66	foreach ( $articles as $node) {

...	...	@@ -72,7 +72,10 @@ foreach ($articles as $article ) {
72	72	$doc = new DOMDocument();
73	73	$doc->preserveWhiteSpace = false;
74	74	$doc->formatOutput = true;
	75	+ $libxml_previous_state = libxml_use_internal_errors(true);
75	76	$doc->loadHTML($article_content);
	77	+ libxml_clear_errors();
	78	+ libxml_use_internal_errors($libxml_previous_state);
76	79	$articles = $doc->getElementsByTagName('article');
77	80	foreach ( $articles as $node) {
78	81	$article_only = DOMinnerHTML($node);

...	...	@@ -5,6 +5,7 @@ $rss_content = http_get_contents(NEWS_RSS_LEPOINTFR);
5	5	$xml = simplexml_load_string($rss_content);
6	6	if ($xml === false) {
7	7	echo 'Failed to read RSS';
	8	+ exit(0);
8	9	} else {
9	10	$channel = array();
10	11	$channel['title'] = $xml->channel->title;

...	...	@@ -58,8 +58,10 @@ foreach ($articles as $article ) {
58	58	$doc = new DOMDocument();
59	59	$doc->preserveWhiteSpace = false;
60	60	$doc->formatOutput = true;
61		- //$doc->loadHTML($article_content);
	61	+ $libxml_previous_state = libxml_use_internal_errors(true);
62	62	$doc->loadHTML($article_content_utf8);
	63	+ libxml_clear_errors();
	64	+ libxml_use_internal_errors($libxml_previous_state);
63	65	$articles = $doc->getElementsByTagName('article');
64	66	$article_only="";
65	67	foreach ( $articles as $node) {

...	...	@@ -90,7 +90,10 @@ foreach ($articles as $article ) {
90	90	$doc = new DOMDocument();
91	91	$doc->preserveWhiteSpace = false;
92	92	$doc->formatOutput = true;
	93	+ $libxml_previous_state = libxml_use_internal_errors(true);
93	94	$doc->loadHTML($article_content);
	95	+ libxml_clear_errors();
	96	+ libxml_use_internal_errors($libxml_previous_state);
94	97	//$articles = $doc->getElementsByTagName('article_container');
95	98	$articles = $doc->getElementsByTagName('div');
96	99	$article_only="";

...	...	@@ -69,7 +69,10 @@ foreach ($articles as $article ) {
69	69	$doc = new DOMDocument();
70	70	$doc->preserveWhiteSpace = false;
71	71	$doc->formatOutput = true;
	72	+ $libxml_previous_state = libxml_use_internal_errors(true);
72	73	$doc->loadHTML($article_content_utf8);
	74	+ libxml_clear_errors();
	75	+ libxml_use_internal_errors($libxml_previous_state);
73	76	$articles = $doc->getElementsByTagName('article');
74	77	$article_only="";
75	78	foreach ( $articles as $node) {

...	...	@@ -58,7 +58,10 @@ foreach ($articles as $article ) {
58	58	$doc = new DOMDocument();
59	59	$doc->preserveWhiteSpace = false;
60	60	$doc->formatOutput = true;
	61	+ $libxml_previous_state = libxml_use_internal_errors(true);
61	62	$doc->loadHTML($article_content);
	63	+ libxml_clear_errors();
	64	+ libxml_use_internal_errors($libxml_previous_state);
62	65	$articles = $doc->getElementsByTagName('article');
63	66	$article_only="";
64	67	foreach ( $articles as $node) {