Added support for main content extraction (fixes #5)

Author: vaites

src/Client.php

@@ -120,6 +120,20 @@ public function getText($file)
         return $this->request('text', $file);
     }
 
+    /**
+     * Extracts main text.
+     *
+     * @param string $file
+     *
+     * @return string
+     *
+     * @throws \Exception
+     */
+    public function getMainText($file)
+    {
+        return $this->request('text-main', $file);
+    }
+
     /**
      * Returns current Tika version.
      *

src/Clients/CLIClient.php

@@ -187,6 +187,10 @@ protected function getArguments($type, $file = null)
                 $arguments[] = '--text';
                 break;
 
+            case 'text-main':
+                $arguments[] = '--text-main';
+                break;
+
             case 'version':
                 $arguments[] = '--version';
                 break;

src/Clients/WebClient.php

@@ -353,6 +353,11 @@ protected function getParameters($type, $file = null)
                 $headers[] = 'Accept: text/plain';
                 break;
 
+            case 'text-main':
+                $resource = 'tika/main';
+                $headers[] = 'Accept: text/plain';
+                break;
+
             case 'version':
                 $resource = 'version';
                 break;

tests/BaseTest.php

@@ -186,6 +186,27 @@ public function testDocumentText($file)
         $this->assertContains('Zenonis est, inquam, hoc Stoici', self::$client->getText($file));
     }
 
+    /**
+     * Main text test
+     *
+     * @dataProvider    fileProvider
+     *
+     * @param   string $file
+     */
+    public function testDocumentMainText($file)
+    {
+        $client =& self::$client;
+
+        if($client::MODE == 'web' && version_compare(self::$version, '1.15') < 0)
+        {
+            $this->markTestSkipped('Apache Tika ' . self::$version . 'lacks main content extraction');
+        }
+        else
+        {
+            $this->assertContains('Sed quia studebat laudi et dignitati', self::$client->getMainText($file));
+        }
+    }
+
     /**
      * Metadata test
      *