Merge pull request #22 from helsingborg-stad/fix/do-not-index-body-of…

…-tags

fix: do not include contents of some tags from source.

source/php/Index.php

@@ -316,7 +316,7 @@ private static function getPost($post)
               'ID' => $post->ID,
               'post_title' => apply_filters('the_title', $post->post_title),
               'post_excerpt' => self::getTheExcerpt($post),
-              'content' => strip_tags(apply_filters('the_content', $post->post_content)),
+              'content' => self::stripTags(apply_filters('the_content', $post->post_content)),
               'permalink' => get_permalink($post->ID),
               'post_date' => strtotime($post->post_date),
               'post_date_formatted' => date(get_option('date_format'), strtotime($post->post_date)),
@@ -352,6 +352,21 @@ private static function getPost($post)
         return null;
     }
 
+    public static function stripTags($content) {
+        $removeBodyOfTags = [
+            'script',
+            'style',
+            'noscript'
+        ];
+
+        $content = preg_replace(sprintf(
+            '/<(%s)\b[^>]*>.*?<\/\1>/is', 
+            implode('|', $removeBodyOfTags)
+        ), '', $content);
+
+        return strip_tags($content);
+    }
+
     public static function getTheExcerpt($post, int $numberOfWords = 55) {
 
         $excerpt = get_the_excerpt($post);