Added a fix for #122 and changed output format of image processing to…

… pdf
janis91 · Jan 28, 2018 · 24f5a5f · 24f5a5f
1 parent 9eb18dc
commit 24f5a5f
Show file tree

Hide file tree

Showing 14 changed files with 131 additions and 106 deletions.
diff --git a/AUTHORS.md b/AUTHORS.md
@@ -2,4 +2,5 @@
 
 * Janis Koehr: <janiskoehr@icloud.com> __Main Author__
 * stweil
-* comradekingu
+* comradekingu
+* mathiasconradt
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,4 +1,7 @@
 #Changelog
+nextcloud-ocr (3.2.0)
+* **Feature**: Now the result of a processed image will always be a pdf file, not a txt file anymore.
+* **Bugfix**: The ENOENT error in the worker should be fixed now.
 nextcloud-ocr (3.1.0-beta.1)
 * **Feature**: Now files can be replaced by selecting a checkbox.
 nextcloud-ocr (3.0.0-beta.8)

diff --git a/README.md b/README.md
@@ -8,7 +8,7 @@ The app uses a docker container with tesseract-ocr, OCRmyPDF and communicates ov
 ## Prerequisites, Requirements and Dependencies
 The OCR app has some prerequisites:
  - **[Nextcloud 12](https://nextcloud.com/)**. For older versions take an older major version of this app.
- - **Linux** server as environment. (tested with Debian 8, Raspbian and Ubuntu 14.04 (Trusty))
+ - **Linux** server as environment. (tested with Debian 8 and Ubuntu 14.04 (Trusty)) **currently not compatible to ARM processors like raspberry**
  - **Docker** is used for processing files. tesseract-ocr and OCRmyPDF reside in a docker container.
  - **php-redis** is used for the communication and has to be a part of your php.
 

diff --git a/appinfo/info.xml b/appinfo/info.xml
@@ -36,7 +36,7 @@ Please read the related topics in the [wiki](https://github.com/janis91/ocr/wiki
 The software is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
 CONDITIONS OF ANY KIND, either express or implied.
 ]]></description>
-	<version>3.1.0-beta.1</version>
+	<version>3.2.0</version>
 	<licence>agpl</licence>
 	<author>Janis Koehr</author>
 	<namespace>Ocr</namespace>

diff --git a/lib/AppInfo/Application.php b/lib/AppInfo/Application.php
@@ -154,7 +154,7 @@ function (IAppContainer $c) {
                     /** @var \OC\Server $server */
                     $server = $c->query('ServerContainer');
                     return new RedisService($c->query('OcrJobMapper'), $c->query('FileUtil'), $c->query('RedisUtil'), 
-                            $server->getL10N('ocr'), $server->getLogger());
+                            $server->getL10N('ocr'), $server->getLogger(), $server->getTempManager());
                 });
         /**
          * Register the Job Service

diff --git a/lib/Service/FileService.php b/lib/Service/FileService.php
@@ -198,7 +198,7 @@ private function buildTargetForShared(File $fileInfo, $replace) {
             }
         } else {
             // IMAGES:
-            return $this->fileUtil->buildNotExistingFilename($filePath, $fileName . '.txt');
+            return $this->fileUtil->buildNotExistingFilename($filePath, $fileName . '.pdf');
         }
     }
 
@@ -240,7 +240,7 @@ private function buildTargetNotForShared(File $fileInfo, $replace) {
             }
         } else {
             // IMAGES:
-            return $this->fileUtil->buildNotExistingFilename($filePath, $fileName . '.txt');
+            return $this->fileUtil->buildNotExistingFilename($filePath, $fileName . '.pdf');
         }
     }
 

diff --git a/lib/Service/JobService.php b/lib/Service/JobService.php
@@ -164,7 +164,7 @@ public function process($languages, $files, $replace) {
                     // set the running type
                     $fType = $this->fileService->getCorrectType($fInfo);
                     // create a temp file for ocr processing purposes
-                    $tempFile = $this->getTempFile($fType);
+                    $tempFile = $this->getTempFile();
                     // Create job object
                     $job = new OcrJob(OcrConstants::STATUS_PENDING, $source, $target, $tempFile, $fType, $this->userId, 
                             false, $fInfo->getName(), null, $replace);
@@ -274,11 +274,11 @@ public function handleProcessed() {
             $this->logger->debug('Check if files were processed by ocr and if so, put them to the right dirs.');
             $processed = $this->jobMapper->findAllProcessed($this->userId);
             foreach ($processed as $job) {
-                if ($this->fileUtil->fileExists($job->getTempFile())) {
+                if ($this->fileUtil->fileExists($this->tempM->getTempBaseDir().'/'.$job->getTempFile().'.pdf')) {
                     // Save the tmp file with newname
                     $this->pullResult($job);
                     $this->jobMapper->delete($job);
-                    $this->fileUtil->execRemove($job->getTempFile());
+                    $this->fileUtil->execRemove($this->tempM->getTempBaseDir().'/'.$job->getTempFile().'.pdf');
                 } else {
                     $job->setStatus(OcrConstants::STATUS_FAILED);
                     $job->setErrorLog('Temp file does not exist.');
@@ -302,7 +302,7 @@ public function handleFailed() {
             $failed = $this->jobMapper->findAllFailed($this->userId);
             foreach ($failed as $job) {
                 // clean the tempfile
-                $this->fileUtil->execRemove($job->getTempFile());
+                $this->fileUtil->execRemove($this->tempM->getTempBaseDir().'/'.$job->getTempFile().'.pdf');
                 // set error displayed
                 $job->setErrorDisplayed(true);
                 $this->jobMapper->update($job);
@@ -326,7 +326,7 @@ private function pullResult($job) {
         if ($job->getReplace()) {
             $this->view->unlink(str_replace($this->userId . '/files', '', $job->getSource()));
         }
-        $result = $this->view->file_put_contents($job->getTarget(), $this->fileUtil->getFileContents($job->getTempFile()));
+        $result = $this->view->file_put_contents($job->getTarget(), $this->fileUtil->getFileContents($this->tempM->getTempBaseDir().'/'.$job->getTempFile().'.pdf'));
         if (!$result) {
             throw new NotFoundException($this->l10n->t('OCR could not put processed file to the right target folder. If you selected the replace option, you can restore the file by using the trash bin.'));
         }
@@ -360,21 +360,17 @@ private function jobFinished($jobId, $error, $log) {
     /**
      * Gives a temp file name back depending on the type of the OCR.
      * Later in the worker this file is used as an output.
-     * 
-     * @param integer $type            
+     *         
      * @return string
      */
-    private function getTempFile($type) {
-        if ($type === OcrConstants::TESSERACT) {
+    private function getTempFile() {
             $fileName = $this->phpUtil->tempnamWrapper($this->tempM->getTempBaseDir(), OcrConstants::TEMPFILE_PREFIX);
             $this->phpUtil->unlinkWrapper($fileName);
-            $fileNameWithPostfix = $fileName . '.txt';
+            $fileNameWithPostfix = $fileName . '.pdf';
             $this->phpUtil->touchWrapper($fileNameWithPostfix);
             $this->phpUtil->chmodWrapper($fileNameWithPostfix, 0600);
-            return $fileNameWithPostfix;
-        } else {
-            return $this->phpUtil->tempnamWrapper($this->tempM->getTempBaseDir(), OcrConstants::TEMPFILE_PREFIX);
-        }
+            $fileName = basename($fileName);
+            return $fileName;
     }
 
     /**

diff --git a/lib/Service/RedisService.php b/lib/Service/RedisService.php
@@ -15,6 +15,7 @@
 use OCA\Ocr\Db\OcrJob;
 use OCA\Ocr\Db\OcrJobMapper;
 use OCP\IL10N;
+use OCP\ITempManager;
 use OCP\ILogger;
 use OCA\Ocr\Constants\OcrConstants;
 use OCA\Ocr\Util\FileUtil;
@@ -57,6 +58,12 @@ class RedisService {
      * @var IL10N
      */
     private $l10n;
+
+    /**
+     *
+     * @var ITempManager
+     */
+    private $tempM;
 
     /**
      * QueueService constructor.
@@ -68,12 +75,13 @@ class RedisService {
      * @param ILogger $logger            
      */
     public function __construct(OcrJobMapper $mapper, FileUtil $fileUtil, RedisUtil $redisUtil, IL10N $l10n, 
-            ILogger $logger) {
+            ILogger $logger, ITempManager $tempManager) {
         $this->mapper = $mapper;
         $this->fileUtil = $fileUtil;
         $this->logger = $logger;
         $this->l10n = $l10n;
         $this->redisUtil = $redisUtil;
+        $this->tempM = $tempManager;
     }
 
     /**
@@ -99,7 +107,7 @@ public function sendJob($job, $languages) {
                 throw new NotFoundException($this->l10n->t('Could not add files to the Redis OCR processing queue.'));
             }
         } catch (Exception $e) {
-            $this->fileUtil->execRemove($job->getTempFile());
+            $this->fileUtil->execRemove($this->tempM->getTempBaseDir().'/'.$job->getTempFile().'.pdf');
             $job->setStatus(OcrConstants::STATUS_FAILED);
             $job->setErrorLog($e->getMessage());
             $this->mapper->update($job);

diff --git a/tests/Unit/Service/FileServiceTest.php b/tests/Unit/Service/FileServiceTest.php
@@ -87,10 +87,10 @@ public function testBuildTargetForSharedForImage() {
             ->will($this->returnValue($share));
         $this->fileUtilMock->expects($this->once())
             ->method('buildNotExistingFilename')
-            ->with('test/path/to', 'file.txt')
-            ->will($this->returnValue('/test/path/to/file.txt'));
+            ->with('test/path/to', 'file.pdf')
+            ->will($this->returnValue('/test/path/to/file.pdf'));
         $result = $this->cut->buildTarget($this->fileInfoSharedPng, true, false);
-        $this->assertEquals('/test/path/to/file.txt', $result);
+        $this->assertEquals('/test/path/to/file.pdf', $result);
     }
 
     public function testBuildTargetForSharedForPdf() {
@@ -130,10 +130,10 @@ public function testBuildTargetForSharedForPdfAndReplaceOnRoot() {
     public function testBuildTargetNotForSharedForImage() {
         $this->fileUtilMock->expects($this->once())
             ->method('buildNotExistingFilename')
-            ->with('test/path/to', 'file.txt')
-            ->will($this->returnValue('/test/path/to/file.txt'));
+            ->with('test/path/to', 'file.pdf')
+            ->will($this->returnValue('/test/path/to/file.pdf'));
             $result = $this->cut->buildTarget($this->fileInfoNotSharedPng, false, false);
-        $this->assertEquals('/test/path/to/file.txt', $result);
+        $this->assertEquals('/test/path/to/file.pdf', $result);
     }
 
     public function testBuildTargetNotForSharedForPdf() {