Added recursive metadata support (fixes #16)

Author: vaites

src/Client.php

@@ -231,13 +231,26 @@ public function setDownloadRemote($download)
     /**
      * Gets file metadata
      *
+     * @link    https://wiki.apache.org/tika/TikaJAXRS#Recursive_Metadata_and_Content
      * @param   string  $file
+     * @param   string  $recursive
      * @return  \Vaites\ApacheTika\Metadata\Metadata
      * @throws  \Exception
      */
-    public function getMetadata($file)
+    public function getMetadata($file, $recursive = null)
     {
-        return $this->request('meta', $file);
+        if(is_null($recursive))
+        {
+            return $this->request('meta', $file);
+        }
+        elseif(in_array($recursive, ['text', 'html', 'ignore']))
+        {
+            return $this->request("rmeta/$recursive", $file);
+        }
+        else
+        {
+            throw new Exception("Unknown recursive type (must be text, html, ignore or null)");
+        }
     }
 
     /**

src/Clients/CLIClient.php

@@ -271,6 +271,12 @@ protected function getArguments($type, $file = null)
                 $arguments[] = '--metadata --json';
                 break;
 
+            case 'rmeta/ignore':
+            case 'rmeta/html':
+            case 'rmeta/text':
+                throw new Exception('Recursive metadata is not supported in command line mode');
+                break;
+
             case 'text':
                 $arguments[] = '--text';
                 break;

src/Clients/WebClient.php

@@ -353,7 +353,7 @@ public function request($type, $file = null)
         // request completed successfully
         if($status == 200)
         {
-            if($type == 'meta')
+            if(preg_match('/^(meta|rmeta)/', $type))
             {
                 $response = Metadata::make($response, $file);
             }
@@ -497,7 +497,10 @@ protected function getParameters($type, $file = null)
                 break;
 
             case 'meta':
-                $resource = 'meta';
+            case 'rmeta/html':
+            case 'rmeta/ignore':
+            case 'rmeta/text':
+                $resource = $type;
                 $headers[] = 'Accept: application/json';
                 break;
 

src/Metadata/DocumentMetadata.php

@@ -164,6 +164,10 @@ protected function setAttribute($key, $value)
                 $this->encoding = $value;
                 break;
 
+            case 'x-tika:content':
+                $this->content = $value;
+                break;
+
             default:
                 return false;
         }

src/Metadata/ImageMetadata.php

@@ -60,6 +60,10 @@ protected function setAttribute($key, $value)
                 $this->width = (int) $value;
                 break;
 
+            case 'x-tika:content':
+                $this->content = $value;
+                break;
+
             default:
                 return false;
         }

src/Metadata/Metadata.php

@@ -11,6 +11,13 @@
  */
 abstract class Metadata
 {
+    /**
+     * Content
+     *
+     * @var string
+     */
+    public $content = null;
+
     /**
      * MIME type
      *
@@ -86,7 +93,10 @@ public static function make($response, $file)
         }
 
         // decode the JSON response
-        $meta = json_decode($response);
+        $json = json_decode($response);
+
+        // get the meta info
+        $meta = is_array($json) ? current($json) : $json;
 
         // exceptions if metadata is not valid
         if(json_last_error())

tests/ErrorTest.php

@@ -150,6 +150,24 @@ public function testRequestRestrictedOptions()
         }
     }
 
+    /**
+     * Test wrong recursive metadata type
+     */
+    public function testRequestMetadataType()
+    {
+        try
+        {
+            $client = Client::prepare('localhost', 9998, [CURLOPT_PROXY => 'localhost']);
+            $client->getMetadata(dirname(__DIR__) . '/samples/sample3.png', 'bad');
+
+            $this->fail();
+        }
+        catch(Exception $exception)
+        {
+            $this->assertContains('Unknown recursive type', $exception->getMessage());
+        }
+    }
+
     /**
      * Test unsupported media type
      */

tests/WebTest.php

@@ -115,4 +115,49 @@ public function testSetRetries()
 
         $this->assertEquals(5, $client->getRetries());
     }
+
+    /**
+     * Recursive text metadata test
+     *
+     * @dataProvider    ocrProvider
+     *
+     * @param   string $file
+     * @throws  \Exception
+     */
+    public function testTextRecursiveMetadata($file)
+    {
+        $metadata = self::$client->getMetadata($file, 'text');
+
+        $this->assertContains('Ut enim ad minim veniam', $metadata->content);
+    }
+
+    /**
+     * Recursive HTML metadata test
+     *
+     * @dataProvider    ocrProvider
+     *
+     * @param   string $file
+     * @throws  \Exception
+     */
+    public function testHtmlRecursiveMetadata($file)
+    {
+        $metadata = self::$client->getMetadata($file, 'html');
+
+        $this->assertContains('Ut enim ad minim veniam', $metadata->content);
+    }
+
+    /**
+     * Recursive ignore metadata test
+     *
+     * @dataProvider    ocrProvider
+     *
+     * @param   string $file
+     * @throws  \Exception
+     */
+    public function testIgnoreRecursiveMetadata($file)
+    {
+        $metadata = self::$client->getMetadata($file, 'ignore');
+
+        $this->assertNull($metadata->content);
+    }
 }