Have been added:

1) Code strip tags. 2) Settings for necessary tags which will be no cutting off.
deepslam · May 18, 2017 · dc8157b · dc8157b
1 parent 941a720
commit dc8157b
Show file tree

Hide file tree

Showing 5 changed files with 53 additions and 3 deletions.
diff --git a/src/ContentParser.php b/src/ContentParser.php
@@ -53,6 +53,9 @@ public function __construct($url, Array $params = []) {
         $this->url = $url;
         $this->urlParams = parse_url($this->getURL());
         $this->parsingResult = $this->parse();
+        if ($this->needsCodeStrip()) {
+            $this->parsingResult->stripContent();
+        }
         if ($this->needsCodeClean()) {
             $this->parsingResult->cleanContent();
         }
@@ -96,6 +99,15 @@ final protected function needsCodeClean() {
         return ((bool)config('deepslam.parser.clean_code'));
     }
 
+    /**
+     * Checks whether have to strip tags or not.
+     *
+     * @return bool True - need to strip tags, false - needn't to strip tags
+     */
+    final protected function needsCodeStrip() {
+        return ((bool)config('deepslam.parser.strip_tags'));
+    }
+
     /*** STATIC LAYER ***/
 
     /**

diff --git a/src/ContentParserMercury.php b/src/ContentParserMercury.php
@@ -9,7 +9,7 @@ final protected function parse():ParsingResult
     {
         $result = new ParsingResult();
         $response = Curl::to('https://mercury.postlight.com/parser?url='.urlencode($this->getURL()))
-            ->withHeaders(array('x-api-key: '.config('deepslam.mercury.api-key')))
+            ->withHeaders(array('x-api-key: '.config('deepslam.mercury.api_key')))
             ->asJson()
             ->get();
         if (!is_null($response)) {

diff --git a/src/ParsingResult.php b/src/ParsingResult.php
@@ -73,5 +73,14 @@ public function isEmpty() {
     public function cleanContent() {
         $this->params["content"] = preg_replace( '/\s?(style|class|id)=[\'"]{1}.*[\'"]{1}/sUi', '', $this->params["content"], -1 );
     }
+
+    /**
+     * Let's strip tags from unwilling tags
+     *
+     *@return String Stripped content
+     */
+    public function stripContent() {
+        $this->params["content"] = strip_tags($this->params["content"], implode('',config('deepslam.parser.allowed_tags')));
+    }
 }
 ?>
diff --git a/src/config/mercury.php b/src/config/mercury.php
@@ -1,5 +1,5 @@
 <?php
     return [
-        'api-key' => 'YVMpLAlWqWk9DSJHcWMCdsFhqMSrIhoK0YaQwfX6',
+        'api_key' => 'YVMpLAlWqWk9DSJHcWMCdsFhqMSrIhoK0YaQwfX6',
     ];
 ?>
diff --git a/src/config/parser.php b/src/config/parser.php
@@ -1,6 +1,35 @@
 <?php
 return [
-    "clean_code" => true,
+    "clean_code" => true,//Clean code from different styles, classes etc.
+    "strip_tags" => true,//Strip tags?
+    "allowed_tags" => [//List of the allowed tags
+        '<br>',
+        '<p>',
+        '<a>',
+        '<ul>',
+        '<ol>',
+        '<li>',
+        '<i>',
+        '<b>',
+        '<strong>',
+        '<img>',
+        '<span>',
+        '<pre>',
+        '<code>',
+        '<table>',
+        '<tr>',
+        '<th>',
+        '<td>',
+        '<tbody>',
+        '<thead>',
+        '<tfoot>',
+        '<h1>',
+        '<h2>',
+        '<h3>',
+        '<h4>',
+        '<h5>',
+        '<h6>',
+    ],
     "parsers" => [
         'mercury' => \Deepslam\ContentParser\ContentParserMercury::class,
         'graby' => \Deepslam\ContentParser\ContentParserGraby::class,