Showing 1 changed files with 70 additions and 26 deletions
+70 -26
sources/news-huffingtonpost.fr.php
... ...
@@ -39,35 +39,33 @@ if ($xml === false) {
39 39
     $pos_stop = strpos($item->description, $SEARCH_SUB2, $pos_start);
40 40
     $article['image']=$item->mediacontent['url'];
41 41
     $articles[$cpt] = $article;  
42
-    echo '               <div onclick="onArticle('.$cpt.')" style="display:inline;">';
43
-    echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>';
44
-    echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;';
42
+    echo '<div onclick="onArticle('.$cpt.')" style="display:inline;">'.PHP_EOL;
43
+    echo '<img src="'.$article['image'].'" style="display:inline;" width="100%"><br>'.PHP_EOL;
44
+    echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;'.PHP_EOL;
45 45
     echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-down fa-2x"></i></a></div>&nbsp;';
46
-    echo $article['title'].'&nbsp;&nbsp;';
47
-    echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>';
46
+    echo $article['title'].'&nbsp;&nbsp;'.PHP_EOL;
47
+    echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a><br></div></div>'.PHP_EOL;
48 48
     $cpt++;
49 49
     if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
50 50
       break;
51 51
     }
52 52
   }
53 53
 }
54
-echo '            </div><!-- ./panel-body -->';
55
-echo '            </div><!-- ./panel panel-default -->';
56
-echo '            </div><!-- ./col-md-6 -->';
57
-echo '<div class="col-md-6">';
58
-echo '<div class="panel panel-default">';
59
-echo '<div class="panel-body">';
60
-echo '<a name="article-top"></a><div id="article-current"></div>';
54
+echo '</div><!-- ./panel-body -->'.PHP_EOL;
55
+echo '</div><!-- ./panel panel-default -->'.PHP_EOL;
56
+echo '</div><!-- ./col-md-6 -->'.PHP_EOL;
57
+echo '<div class="col-md-6">'.PHP_EOL;
58
+echo '<div class="panel panel-default">'.PHP_EOL;
59
+echo '<div class="panel-body">'.PHP_EOL;
60
+echo '<a name="article-top"></a><div id="article-current"></div>'.PHP_EOL;
61 61
 $cpt=0;
62 62
 foreach ($articles as $article ) {
63 63
   $cpt_prev=$cpt-1;
64 64
   $cpt_next=$cpt+1;
65
-  echo "\n";
66
-  echo '<!-- ==================== article '.$cpt.'============== -->';
67
-  echo "\n";
65
+  echo PHP_EOL.PHP_EOL."<!-- ==================== article '.$cpt.'============== -->".PHP_EOL;
68 66
   echo "<div class=\"article\" id=\"article-$cpt\" style=\"display: none;\">\n";
69 67
   echo "<hr>";
70
-  echo "<a name=\"article-$cpt\">";
68
+  echo "<a name=\"article-$cpt\">\n";
71 69
   $article_content = file_get_contents($article['link']);
72 70
   $doc = new DOMDocument();
73 71
   $doc->preserveWhiteSpace = false;
... ...
@@ -96,6 +94,25 @@ foreach ($articles as $article ) {
96 94
   if($pos_stop) {
97 95
     $temp = substr($article_only, 0, $pos_stop);
98 96
     $article_only = $temp."</blockquote>";
97
+  } else {
98
+    //End of article cleanup
99
+    $SEARCHSUB='<h3 class="related-entries__title">LIRE AUSSI</h3>';
100
+    $pos_stop=false;
101
+    $pos_stop = strpos($article_only, $SEARCHSUB);
102
+    if($pos_stop) {
103
+      $temp = substr($article_only, 0, $pos_stop);
104
+      $article_only = $temp;
105
+    } else {
106
+      $SEARCHSUB='<h3 class="recirc__title trending">';
107
+      $pos_stop=false;
108
+      $pos_stop = strpos($article_only, $SEARCHSUB);
109
+      if($pos_stop) {
110
+        $temp = substr($article_only, 0, $pos_stop);
111
+        $article_only = $temp;
112
+      } else {
113
+        DEBUG("Article($cpt) : No Match");
114
+      }
115
+    }
99 116
   }
100 117
 
101 118
   $re = '/alt="(.+?)" width="(.+?)" height="(.+?)"/';
... ...
@@ -138,19 +155,46 @@ foreach ($articles as $article ) {
138 155
     '',
139 156
     '');
140 157
   $article_only  = str_replace($orgStrings, $newStrings, $article_only);
158
+
159
+  // Cleanup "Social"
160
+  $re='/<script id="share-bar__social-mob" (.+)<\/script>/';
161
+  $article_only = preg_replace($re, '', $article_only);
162
+  $re='/<li class="(facebook|twitter|gplus|pinterest|linkedin|email) share-bar__item">(.+)<\/a><\/li>/';
163
+  $article_only = preg_replace($re, '', $article_only);
164
+  $re='/<li class="(facebook|twitter|snapchat|instagram) follow-us-networks-item " (.+?)<\/span> <\/a> <\/li>/';
165
+  $article_only = preg_replace($re, '', $article_only);
166
+  $re='/<div class="social-overlay__qr">(.+)<\/a><\/div>/';
167
+  $article_only = preg_replace($re, '', $article_only);
168
+  $re='/<script type="text\/javascript"> \(function\(\)\{ var dfp_placement_slug(.+)<\/script>/';
169
+  $article_only = preg_replace($re, '', $article_only);
170
+  $re='/<form class="newsletter-signup" method="POST">(.+)<\/form>/';
171
+  $article_only = preg_replace($re, '', $article_only);
172
+  $re='/<div class="follow-us__header"> <h3>SUIVEZ-NOUS<\/h3> <\/div>/';
173
+  $article_only = preg_replace($re, '', $article_only);
174
+  $re='/<div class="follow-us__description"> <p>Avec la newsletter quotidienne du HuffPost(.+) target="_blank">En savoir plus<\/a> <\/p> <\/div>/';
175
+  $article_only = preg_replace($re, '', $article_only);
176
+  $re='/<li class="(facebook|instagram) follow-us-networks-item even" (.+) <\/a> <\/li>/';
177
+  $article_only = preg_replace($re, '', $article_only);
178
+  //Clean Empty lines
179
+  $article_only = preg_replace("/(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+/", "\n", $article_only);
180
+  $article_only = preg_replace('/\s\s+/', ' ', $article_only);
181
+
182
+
183
+
141 184
   
142
-  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;';
143
-  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;';
144
-  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
145
-  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>';
185
+  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;'.PHP_EOL;
186
+  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;'.PHP_EOL;
187
+  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
188
+  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div>'.PHP_EOL;
146 189
   echo '<div class="extract-content" id="'.$cpt.'">';
147 190
   echo '<h4>'.$article['title'].'</h4>';
148
-  echo $article_only.'</div>';
149
-  echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
150
-  echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
151
-  echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
152
-  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;';
153
-  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>';
191
+  echo $article_only.'</div>'.PHP_EOL;
192
+  echo '<div id="nav-up" style="display:inline;"><a href="#top"><i class="fa fa-home fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
193
+  echo '<div id="nav-up" style="display:inline;"><a href="#article-top"><i class="fa fa-chevron-upi fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
194
+  echo '<div id="nav-source" style="display:inline;"><a href="'.$article['link'].'" target="new-'.$cpt.'"><i class="fa fa-link fa-2x"></i></a></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
195
+  echo '<div id="nav-prev" onclick="onArticle('.$cpt_prev.')" style="display:inline;"><i class="fa fa-chevron-left fa-2x"></i></div>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;'.PHP_EOL;
196
+  echo '<div id="nav-next" onclick="onArticle('.$cpt_next.')" style="display:inline;"><i class="fa fa-chevron-right fa-2x"></i></div></div>'.PHP_EOL;
197
+  echo '<!-- ENDOF ==================== article '.$cpt.'============== -->'.PHP_EOL;
154 198
   $cpt++;
155 199
   if( $cpt > $NEWS_RSS_MAX_ITEMS ) {
156 200
       break;