Showing 2 changed files with 26 additions and 14 deletions
+1 -1
sources/news-lemonde.fr.php
... ...
@@ -25,7 +25,7 @@ if ($xml === false) {
25 25
     $article['timestamp'] = strtotime($item->pubDate);
26 26
     $article['description'] = $item->description;
27 27
     $article['image'] = $item->enclosure['url'];
28
-    $articles[$article['timestamp']] = $article;  
28
+    $articles[$cpt] = $article;
29 29
     echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
30 30
     echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
31 31
     echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;';
+25 -13
sources/news-lexpress.fr.php
... ...
@@ -25,7 +25,6 @@ if ($xml === false) {
25 25
     $article['timestamp'] = strtotime($item->pubDate);
26 26
     $article['description'] = $item->description;
27 27
     $article['image'] = $item->enclosure['url'];
28
-    //$articles[$article['timestamp']] = $article;  
29 28
     $articles[$cpt] = $article;  
30 29
     echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">';
31 30
     echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
... ...
@@ -55,7 +54,7 @@ foreach ($articles as $article ) {
55 54
   echo "<hr>";
56 55
   echo "<a name=\"article-$cpt\">";
57 56
   $article_content = http_get_contents($article['link']);
58
-
57
+/*
59 58
   $SEARCH_SUB1='<h1 class="art-titre list-view">';
60 59
   $pos_start = strpos($article_content, $SEARCH_SUB1);
61 60
   $pos_stop = false;
... ...
@@ -84,15 +83,28 @@ foreach ($articles as $article ) {
84 83
   
85 84
   //echo "pos_start : $pos_start<br>";
86 85
   //echo "pos_stop : $pos_stop<br>";
87
-
86
+ */
87
+  $doc = new DOMDocument();
88
+  $doc->preserveWhiteSpace = false;
89
+  $doc->formatOutput       = true;
90
+  $doc->loadHTML($article_content);
91
+  //$articles = $doc->getElementsByTagName('article_container');
92
+  $articles = $doc->getElementsByTagName('div');
93
+  $article_only="";
94
+  foreach ( $articles as $node) {
95
+    if(0==strcmp("article_container",$node->getAttribute('class'))) {
96
+      $article_only = DOMinnerHTML($node);
97
+    }
98
+  }
99
+  
88 100
   $orgStrings = array( ' href="/',
89 101
     ' src="/sites/',
90 102
     '<img src=',
91 103
     ' data-src="/');
92
-  $newStrings = array( ' href="https://www.lepoint.fr/',
93
-    ' src="https://www.lepoint.fr/sites/',
104
+  $newStrings = array( ' href="https://www.lexpress.fr/',
105
+    ' src="https://www.lexpress.fr/sites/',
94 106
     '<img width="100%" src=',
95
-    ' data-src="https://www.lepoint.fr/' );
107
+    ' data-src="https://www.lexpress.fr/' );
96 108
   $article_only  = str_replace($orgStrings, $newStrings, $article_only);
97 109
   
98 110
   $orgStrings = array( '<div class="row">',
... ...
@@ -118,7 +130,7 @@ foreach ($articles as $article ) {
118 130
     '<div>',
119 131
     '<div>');
120 132
   $article_only = str_replace($orgStrings, $newStrings, $article_only);
121
-  
133
+  /*
122 134
   $temp = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
123 135
   $article_only = $temp;
124 136
   $temp = preg_replace('/\s\s+/', ' ', $article_only);
... ...
@@ -129,11 +141,11 @@ foreach ($articles as $article ) {
129 141
   $re = '/<noscript>(.+?)<\/noscript>/ms';
130 142
   $temp = preg_replace($re, '', $article_only);
131 143
   $article_only = $temp;
132
-  
133
-  $orgStrings = array( 'http://www.lepoint.fr/' );
134
-  $newStrings = array( 'https://www.lepoint.fr/' );
144
+*/ 
145
+  $orgStrings = array( 'http://www.lexpress.fr/' );
146
+  $newStrings = array( 'https://www.lexpress.fr/' );
135 147
   $article_only  = str_replace($orgStrings, $newStrings, $article_only);
136
-
148
+/*
137 149
   $re = '/<img class="w100" alt="(.+?)" data-lazyload="loading" onclick="(.+?)" src="(.+?)" title="(.+?)" data-src="(.+?)" \/>/';
138 150
   $temp = preg_replace($re, '<img width="100%" src="\\5">', $article_only);
139 151
   $article_only = $temp;
... ...
@@ -157,12 +169,12 @@ foreach ($articles as $article ) {
157 169
   $re = '/<a onclick="(.+?)" data-shadowbox="#authform" href="(.+?)" id="connect_viafreemium" class="w100 block txtcenter br-4-all bg-grey-medium-blue pas font-black bold br-4-all upper">Identifiez-vous<\/a>/';
158 170
   $temp = preg_replace($re, '', $article_only);
159 171
   $article_only = $temp;
160
-  
172
+*/ 
161 173
   echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
162 174
   echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
163 175
   echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
164 176
   echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
165
-  echo '<div class="extract-content" id="'.$cpt.'">'.$article_only.'</div>';
177
+  echo '<div class="extract-content" id="'.$cpt.'"><h3>'.$article['title'].'</h3><img width="100%" src="'.$article['image'].'">'.$article_only.'</div>';
166 178
   echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
167 179
   echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-up fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
168 180
   echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';