Fix HtmlSource meta tags behaviour

Only for "keywords" and "description" meta tags "content" attr should be treated as user visible text.
mekras · Mar 11, 2017 · b58d2c3 · b58d2c3
1 parent 1938cf0
commit b58d2c3
Show file tree

Hide file tree

Showing 3 changed files with 52 additions and 8 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,12 @@
 # Changelog
 
+## 1.5.1 - 2017-03-11
+
+### Fixed
+
+- HtmlSource: only for "keywords" and "description" meta tags "content" attr should be treated as
+  user visible text.
+
 ## 1.5 - 2017-03-11
 
 ### Added

diff --git a/src/Source/HtmlSource.php b/src/Source/HtmlSource.php
@@ -24,7 +24,6 @@ class HtmlSource implements Source
     static private $textAttributes = [
         'abbr',
         'alt',
-        'content',
         'label',
         'placeholder',
         'title'
@@ -61,7 +60,7 @@ public function getAsString()
         $document = new \DOMDocument('1.0');
         $document->loadHTML($this->html);
 
-        return $this->extractText($document->documentElement);
+        return $this->extractFromNode($document->documentElement);
     }
 
     /**
@@ -71,26 +70,51 @@ public function getAsString()
      *
      * @return string
      */
-    private function extractText(\DOMNode $node)
+    private function extractFromNode(\DOMNode $node)
     {
         if ($node instanceof \DOMText) {
             return trim($node->textContent);
         }
 
-        $text = '';
+        $text = [];
 
         if ($node instanceof \DOMElement) {
             foreach ($node->attributes as $attr) {
                 /** @var \DOMAttr $attr */
                 if (in_array($attr->name, self::$textAttributes, true)) {
-                    $text .= ' ' . trim($attr->textContent);
+                    $text[] = trim($attr->textContent);
                 }
             }
+            $text[] = $this->extractFromMeta($node);
             foreach ($node->childNodes as $child) {
-                $text .= ' ' . $this->extractText($child);
+                $text[] = $this->extractFromNode($child);
             }
         }
 
-        return trim($text);
+        return trim(implode(' ', $text));
+    }
+
+    /**
+     * Extract text from meta tag.
+     *
+     * @param \DOMElement $node
+     *
+     * @return string
+     */
+    private function extractFromMeta(\DOMElement $node)
+    {
+        if (strtolower($node->nodeName) !== 'meta') {
+            return '';
+        }
+
+        if (!($node->hasAttribute('name') && $node->hasAttribute('content'))) {
+            return '';
+        }
+
+        if (!in_array(strtolower($node->getAttribute('name')), ['description', 'keywords'], true)) {
+            return '';
+        }
+
+        return trim($node->getAttribute('content'));
     }
 }
diff --git a/tests/Source/HtmlSourceTest.php b/tests/Source/HtmlSourceTest.php
@@ -25,6 +25,19 @@ class HtmlSourceTest extends TestCase
     public function testBasics()
     {
         $source = new HtmlSource('<a href="#" title="Foo">Bar</a> Baz');
-        static::assertEquals('Foo Bar Baz', $source->getAsString());
+        static::assertEquals('Foo  Bar Baz', $source->getAsString());
+    }
+
+    /**
+     * Only for "keywords" and "description" meta tags "content" attr should be treated as string.
+     */
+    public function testMetaContent()
+    {
+        $source = new HtmlSource(
+            '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">' .
+            '<meta name="Keywords" content="Foo">' .
+            '<meta name="description" content="Bar">'
+        );
+        static::assertEquals('Foo Bar', $source->getAsString());
     }
 }