Adding pdftotext for PDF support in Hypercube (#77)

Islandora · Sep 18, 2019 · 4173b13 · 4173b13
1 parent ff58f34
commit 4173b13
Show file tree

Hide file tree

Showing 4 changed files with 84 additions and 14 deletions.
diff --git a/Hypercube/cfg/config.example.yaml b/Hypercube/cfg/config.example.yaml
@@ -2,7 +2,8 @@
 
 hypercube:
   # path to the convert executable
-  executable: tesseract
+  tesseract_executable: tesseract
+  pdftotext_executable: pdftotext
 
 fedora_resource:
   base_url: http://localhost:8080/fcrepo/rest
@@ -20,4 +21,4 @@ syn:
   # Path to the syn config file for authentication.
   # example can be found here:
   # https://github.com/Islandora-CLAW/Syn/blob/master/conf/syn-settings.example.xml
-  config: ../syn-settings.xml
+  config: ../syn-settings.xml
diff --git a/Hypercube/src/Controller/HypercubeController.php b/Hypercube/src/Controller/HypercubeController.php
@@ -4,6 +4,7 @@
 
 use GuzzleHttp\Psr7\StreamWrapper;
 use Islandora\Crayfish\Commons\CmdExecuteService;
+use Monolog\Logger;
 use Psr\Http\Message\ResponseInterface;
 use Symfony\Component\HttpFoundation\BinaryFileResponse;
 use Symfony\Component\HttpFoundation\Request;
@@ -25,17 +26,35 @@ class HypercubeController
     /**
      * @var string
      */
-    protected $executable;
+    protected $tesseract_executable;
+
+    /**
+     * @var string
+     */
+    protected $pdftotext_executable;
+
+    /**
+     * @var \Monolog\Logger
+     */
+    protected $log;
 
     /**
      * HypercubeController constructor.
      * @param \Islandora\Crayfish\Commons\CmdExecuteService $cmd
-     * @param string $executable
+     * @param string $tesseract_executable
+     * @param string $pdftotext_executable
+     * @param $log
      */
-    public function __construct(CmdExecuteService $cmd, $executable)
-    {
+    public function __construct(
+        CmdExecuteService $cmd,
+        $tesseract_executable,
+        $pdftotext_executable,
+        Logger $log
+    ) {
         $this->cmd = $cmd;
-        $this->executable = $executable;
+        $this->tesseract_executable = $tesseract_executable;
+        $this->pdftotext_executable = $pdftotext_executable;
+        $this->log = $log;
     }
 
     /**
@@ -50,10 +69,21 @@ public function get(Request $request)
         // Get tiff as a resource.
         $body = StreamWrapper::getResource($fedora_resource->getBody());
 
-        // Arguments to OCR command are sent as a custom header
+        // Arguments to command line are sent as a custom header
         $args = $request->headers->get('X-Islandora-Args');
 
-        $cmd_string = $this->executable . ' stdin stdout ' . $args;
+        // Check content type and use the appropriate command line tool.
+        $content_type = $fedora_resource->getHeader('Content-Type')[0];
+
+        $this->log->debug("Got Content-Type:", ['type' => $content_type]);
+
+        if ($content_type == 'application/pdf') {
+            $cmd_string = $this->pdftotext_executable . " $args - -";
+        } else {
+            $cmd_string = $this->tesseract_executable . " stdin stdout $args";
+        }
+
+        $this->log->debug("Executing command:", ['cmd' => $cmd_string]);
 
         // Return response.
         try {

diff --git a/Hypercube/src/app.php b/Hypercube/src/app.php
@@ -15,7 +15,9 @@
 $app['hypercube.controller'] = function ($app) {
     return new HypercubeController(
         $app['crayfish.cmd_execute_service'],
-        $app['crayfish.hypercube.executable']
+        $app['crayfish.hypercube.tesseract_executable'],
+        $app['crayfish.hypercube.pdftotext_executable'],
+        $app['monolog']
     );
 };
 

diff --git a/Hypercube/tests/Islandora/Hypercube/Tests/HypercubeControllerTest.php b/Hypercube/tests/Islandora/Hypercube/Tests/HypercubeControllerTest.php
@@ -4,6 +4,7 @@
 
 use Islandora\Crayfish\Commons\CmdExecuteService;
 use Islandora\Hypercube\Controller\HypercubeController;
+use Monolog\Logger;
 use Prophecy\Argument;
 use Psr\Http\Message\ResponseInterface;
 use Psr\Http\Message\StreamInterface;
@@ -21,9 +22,12 @@ class HypercubeControllerTest extends \PHPUnit_Framework_TestCase
     public function testOptions()
     {
         $mock_service = $this->prophesize(CmdExecuteService::class)->reveal();
+        $mock_logger = $this->prophesize(Logger::class)->reveal();
         $controller = new HypercubeController(
             $mock_service,
-            ''
+            'tesseract',
+            'pdftotext',
+            $mock_logger
         );
 
         $response = $controller->options();
@@ -39,12 +43,27 @@ public function testOptions()
      * @covers ::get
      */
     public function testTesseractErrorReturns500()
+    {
+        $this->errorReturns500('image/tiff');
+    }
+
+    /**
+     * @covers ::__construct
+     * @covers ::get
+     */
+    public function testPdfToTextErrorReturns500()
+    {
+        $this->errorReturns500('application/pdf');
+    }
+
+    protected function errorReturns500($mimetype)
     {
         // Mock a TesseractService to create a controller.
         $prophecy = $this->prophesize(CmdExecuteService::class);
         $prophecy->execute(Argument::any(), Argument::any())->willThrow(new \RuntimeException("ERROR", 500));
         $mock_service = $prophecy->reveal();
-        $controller = new HypercubeController($mock_service, '');
+        $mock_logger = $this->prophesize(Logger::class)->reveal();
+        $controller = new HypercubeController($mock_service, 'tesseract', 'pdftotext', $mock_logger);
 
         // Mock a stream body for a Fedora response.
         $prophecy = $this->prophesize(StreamInterface::class);
@@ -54,6 +73,7 @@ public function testTesseractErrorReturns500()
 
         // Mock a Fedora response.
         $prophecy = $this->prophesize(ResponseInterface::class);
+        $prophecy->getHeader('Content-Type')->willReturn(['image/tiff']);
         $prophecy->getStatusCode()->willReturn(200);
         $prophecy->getBody()->willReturn($mock_stream);
         $mock_fedora_response = $prophecy->reveal();
@@ -78,10 +98,25 @@ public function testTesseractErrorReturns500()
      */
     public function testTesseractSuccessReturns200()
     {
-        // Mock a TesseractService to create a controller.
+        $this->successReturns200('image/tiff');
+    }
+
+    /**
+     * @covers ::__construct
+     * @covers ::get
+     */
+    public function testPdfToTextSuccessReturns200()
+    {
+        $this->successReturns200('application/pdf');
+    }
+
+    protected function successReturns200($mimetype)
+    {
+        // Mock a controller.
         $prophecy = $this->prophesize(CmdExecuteService::class);
         $mock_service = $prophecy->reveal();
-        $controller = new HypercubeController($mock_service, '');
+        $mock_logger = $this->prophesize(Logger::class)->reveal();
+        $controller = new HypercubeController($mock_service, 'tesseract', 'pdftotext', $mock_logger);
 
         // Mock a stream body for a Fedora response.
         $prophecy = $this->prophesize(StreamInterface::class);
@@ -91,6 +126,7 @@ public function testTesseractSuccessReturns200()
 
         // Mock a Fedora response.
         $prophecy = $this->prophesize(ResponseInterface::class);
+        $prophecy->getHeader('Content-Type')->willReturn([$mimetype]);
         $prophecy->getStatusCode()->willReturn(200);
         $prophecy->getBody()->willReturn($mock_stream);
         $mock_fedora_response = $prophecy->reveal();
@@ -104,6 +140,7 @@ public function testTesseractSuccessReturns200()
         $request->headers->set('ApixLdpResource', 'http://localhost:8080/fcrepo/rest/foo');
         $request->attributes->set('fedora_resource', $mock_fedora_response);
 
+        // Check success.
         $response = $controller->get($request);
         $this->assertTrue($response->getStatusCode() == 200, "Response must return 200");
     }