Merge pull request #218 from Simounet/feat/jsonld-multiple-authors

JSON LD multiple authors
j0k3r · Nov 12, 2019 · c27bcc8 · c27bcc8
2 parents 590b71d + 16a82d8
commit c27bcc8
Show file tree

Hide file tree

Showing 6 changed files with 45 additions and 24 deletions.
diff --git a/src/Extractor/ContentExtractor.php b/src/Extractor/ContentExtractor.php
@@ -37,9 +37,7 @@ class ContentExtractor
     private $configBuilder;
 
     /**
-     * @param array                $config
-     * @param LoggerInterface|null $logger
-     * @param ConfigBuilder|null   $configBuilder
+     * @param array $config
      */
     public function __construct($config = [], LoggerInterface $logger = null, ConfigBuilder $configBuilder = null)
     {
@@ -730,8 +728,6 @@ protected function addAuthor($authorDirty)
     /**
      * Check if given node list exists and has length more than 0.
      *
-     * @param \DOMNodeList $elems
-     *
      * @return bool
      */
     private function hasElements(\DOMNodeList $elems)
@@ -742,8 +738,7 @@ private function hasElements(\DOMNodeList $elems)
     /**
      * Remove elements.
      *
-     * @param \DOMNodeList $elems
-     * @param string       $logMessage
+     * @param string $logMessage
      */
     private function removeElements(\DOMNodeList $elems, $logMessage = null)
     {
@@ -1250,6 +1245,24 @@ private function extractOpenGraph(\DOMXPath $xpath)
         }
     }
 
+    /**
+     * Clean extract of JSON-LD authors.
+     */
+    private function extractAuthorsFromJsonLdArray(array $authors)
+    {
+        if (isset($authors['name'])) {
+            return $authors['name'];
+        }
+
+        return array_map(function ($author) {
+            if (isset($author['name']) && \is_string($author['name'])) {
+                return $author['name'];
+            }
+
+            return false;
+        }, $authors);
+    }
+
     /**
      * Extract data from JSON-LD information.
      *
@@ -1308,8 +1321,8 @@ private function extractJsonLdInformation(\DOMXPath $xpath)
                 $candidateNames[] = $data['name'];
             }
 
-            if (!empty($data['author']['name'])) {
-                $authors = $data['author']['name'];
+            if (!empty($data['author'])) {
+                $authors = \is_array($data['author']) ? $this->extractAuthorsFromJsonLdArray($data['author']) : $data['author'];
 
                 if (false === \is_array($authors)) {
                     $authors = [$authors];

diff --git a/src/Extractor/HttpClient.php b/src/Extractor/HttpClient.php
@@ -43,9 +43,8 @@ class HttpClient
     private $responseHistory;
 
     /**
-     * @param Client               $client Http client
-     * @param array                $config
-     * @param LoggerInterface|null $logger
+     * @param Client $client Http client
+     * @param array  $config
      */
     public function __construct(Client $client, $config = [], LoggerInterface $logger = null)
     {
@@ -567,8 +566,6 @@ private function getUglyURL($url, $html)
      * Format all headers to avoid unecessary array level.
      * Also lower the header name.
      *
-     * @param ResponseInterface $response
-     *
      * @return array
      */
     private function formatHeaders(ResponseInterface $response)

diff --git a/src/Graby.php b/src/Graby.php
@@ -44,9 +44,8 @@ class Graby
     private $imgNoReferrer = false;
 
     /**
-     * @param array              $config
-     * @param Client|null        $client        Http client
-     * @param ConfigBuilder|null $configBuilder
+     * @param array       $config
+     * @param Client|null $client Http client
      */
     public function __construct($config = [], Client $client = null, ConfigBuilder $configBuilder = null)
     {
@@ -128,8 +127,6 @@ public function __construct($config = [], Client $client = null, ConfigBuilder $
 
     /**
      * Redefine all loggers.
-     *
-     * @param LoggerInterface $logger
      */
     public function setLogger(LoggerInterface $logger)
     {

diff --git a/src/SiteConfig/ConfigBuilder.php b/src/SiteConfig/ConfigBuilder.php
@@ -16,8 +16,7 @@ class ConfigBuilder
     private $cache = [];
 
     /**
-     * @param array                $config
-     * @param LoggerInterface|null $logger
+     * @param array $config
      */
     public function __construct($config = [], LoggerInterface $logger = null)
     {
@@ -336,8 +335,6 @@ public function mergeConfig(SiteConfig $currentConfig, SiteConfig $newConfig)
     /**
      * Parse line from the config file to build the config.
      *
-     * @param array $lines
-     *
      * @return SiteConfig
      */
     public function parseLines(array $lines)

diff --git a/tests/Extractor/ContentExtractorTest.php b/tests/Extractor/ContentExtractorTest.php
@@ -946,6 +946,23 @@ public function testJsonLd()
         $this->assertContains('<p>hihi</p>', $content_block->ownerDocument->saveXML($content_block));
     }
 
+    public function testJsonLdWithMultipleAuthors()
+    {
+        $contentExtractor = new ContentExtractor(self::$contentExtractorConfig);
+
+        $res = $contentExtractor->process(
+            '<script type="application/ld+json">{"@context":"https://schema.org","@type":"NewsArticle","author":[{"@type":"Person","name":"Elisa Thevenet"},{"@type":"Person","name":"Humphrey Bogart"}]}</script>',
+            'https://nativead.io/jsonld'
+        );
+
+        $content_block = $contentExtractor->getContent();
+
+        $this->assertSame([
+            'Elisa Thevenet',
+            'Humphrey Bogart',
+        ], $contentExtractor->getAuthors());
+    }
+
     public function testNoDefinedHtml()
     {
         $contentExtractor = new ContentExtractor(self::$contentExtractorConfig);

diff --git a/tests/GrabyTest.php b/tests/GrabyTest.php
@@ -1192,7 +1192,7 @@ public function dataAuthors()
             [
                 'https://www.liberation.fr/planete/2017/04/05/donald-trump-et-xi-jinping-tentative-de-flirt-en-floride_1560768',
                 'liberation-authors.html',
-                ['Raphaël Balenieri, correspondant à Pékin', 'Frédéric Autran, correspondant à New York'],
+                ['Raphaël Balenieri', 'Frédéric Autran'],
             ],
         ];
     }