OcrService::process() - Code Metrics - Inspection of "Semaphore features" - janis91/ocr - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#17)

by Janis

created 2016-09-21 18:42 UTC

OcrService::process() C

↳ Parent: OcrService

Complexity

Conditions	7
Paths	16

Size

Total Lines	38
Code Lines	19

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	6
CRAP Score	22.6944

Importance

Changes

Metric	Value
dl	0
loc	38
ccs	6
cts	19
cp	0.3158
rs	6.7272
c	0
b	0
f	0
cc	7
eloc	19
nc	16
nop	2
crap	22.6944

<?php
/**
 * nextCloud - ocr
 *
 * This file is licensed under the Affero General Public License version 3 or
 * later. See the COPYING file.
 *
 * @author Janis Koehr <[email protected]>
 * @copyright Janis Koehr 2016
 */

namespace OCA\Ocr\Service;

use Exception;
use OC\Files\View;
use OCA\Ocr\Db\OcrStatus;
use OCA\Ocr\Db\OcrStatusMapper;
use OCP\Files;
use OCP\Files\FileInfo;
use OCP\IConfig;
use OCP\IL10N;
use OCP\ILogger;
use OCP\ITempManager;


/**
 * Class OcrService
 * @package OCA\Ocr\Service
 */
class OcrService {

	/**
	 * @var ILogger
	 */
	private $logger;

	/**
	 * @var ITempManager
	 */
	private $tempM;

	/**
	 * @var IConfig
	 */
	private $config;

	/**
	 * @var QueueService
	 */
	private $queueService;

	/**
	 * @var OcrStatusMapper
	 */
	private $statusMapper;

	/**
	 * @var View
	 */
	private $view;

	/**
	 * @var
	 */
	private $userId;

	/**
	 * @var IL10N
	 */
	private $l10n;

	/**
	 * Array of allowed mimetypes for ocr processing
	 */
	const ALLOWED_MIMETYPES = ['application/pdf', 'image/png', 'image/jpeg', 'image/tiff'];

	/**
	 * the correct mimetype for a pdf file
	 */
	const MIMETYPE_PDF = 'application/pdf';

	/**
	 * the only allowed image mimetypes by tesseract
	 */
	const MIMETYPES_IMAGE = ['image/png', 'image/jpeg', 'image/tiff'];

	/**
	 * OcrService constructor.
	 *
	 * @param ITempManager $tempManager
	 * @param IConfig $config
	 * @param QueueService $queueService
	 * @param OcrStatusMapper $mapper
	 * @param View $view
	 * @param $userId
	 * @param IL10N $l10n
	 * @param ILogger $logger
	 */
	public function __construct(ITempManager $tempManager, IConfig $config, QueueService $queueService, OcrStatusMapper $mapper, View $view, $userId, IL10N $l10n, ILogger $logger) {
		$this->logger = $logger;
		$this->tempM = $tempManager;
		$this->config = $config;
		$this->queueService = $queueService;
		$this->statusMapper = $mapper;
		$this->view = $view;
		$this->userId = $userId;
		$this->l10n = $l10n;
	}

	/**
	 * Gets the list of all available tesseract-ocr languages.
	 *
	 * @return array Languages
	 */
	public function listLanguages() {
		try {
			$success = -1;
			$this->logger->debug('Fetching languages. ', ['app' => 'ocr']);
			exec('tesseract --list-langs 2>&1', $result, $success);
			if ($success === 0 && count($result) > 0) {
				if (is_array($result)) {
					$traineddata = $result;
				} else {
					$traineddata = explode(' ', $result);
				}
				$languages = array();
				foreach ($traineddata as $td) {
					$tdname = trim($td);
					if (strlen($tdname) === 3) {
						array_push($languages, $tdname);
					}
				}
				$this->logger->debug('Fetched languages: ' . json_encode($languages), ['app' => 'ocr']);
				return $languages;
			} else {
				throw new NotFoundException($this->l10n->t('No languages found.'));
			}
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * Processes and prepares the files for ocr.
	 * Sends the stuff to the client in order to ocr async.
	 *
	 * @param string $language
	 * @param array $files
	 * @return string
	 */
	public function process($language, $files) {
		try {
			$this->logger->debug('Will now process files: ' . json_encode($files) . ' with language: ' . json_encode($language), ['app' => 'ocr']);
			// Check if files and language not empty
			if (!empty($files) && !empty($language) && in_array($language, $this->listLanguages())) {
				// get the array with full fileinfo
				$fileInfo = $this->buildFileInfo($files);
				foreach ($fileInfo as $fInfo) {
					// Check if filelock existing
					// TODO: FileLock maybe \OC\Files\View::lockFile()
					// get new name for saving purpose
					$newName = $this->buildNewName($fInfo);

					// create a temp file for ocr processing purposes
					$tempFile = $this->tempM->getTemporaryFile();

					// set the running type
					if ($fInfo->getMimetype() === $this::MIMETYPE_PDF) {
						$ftype = 'mypdf';
					} else {
						$ftype = 'tess';
					}

					// Create status object
					$status = new OcrStatus('PENDING', $fInfo->getId(), $newName, $tempFile, $ftype, $this->userId);

					// Init client and send task / job
					// Feed the worker
					$this->queueService->clientSend($status, $this->config->getSystemValue('datadirectory'), $fInfo->getPath(), $language, \OC::$SERVERROOT);
				}
				return 'PROCESSING';
			} else {
				throw new NotFoundException($this->l10n->t('Empty passed parameters.'));
			}
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * A function which returns the JSONResponse for all required status checks and tasks.
	 * It will check for already processed, pending and failed ocr tasks and return them as needed.
	 *
	 * @return string
	 */
	public function status() {
		try {
			// TODO: release lock
			$processed = $this->handleProcessed();

			$failed = count($this->handleFailed());

			$pending = count($this->statusMapper->findAllPending($this->userId));

			return ['processed' => $processed, 'failed' => $failed, 'pending' => $pending];
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * The command ocr:complete for occ will call this function in order to set the status.
	 * the worker should call it automatically after each processing step.
	 *
	 * @param $statusId
	 * @param boolean $failed
	 */
	public function complete($statusId, $failed) {
		try {
			$status = $this->statusMapper->find($statusId);
			if (!$failed) {
				$status->setStatus('PROCESSED');
				$this->statusMapper->update($status);
			} else {
				$status->setStatus('FAILED');
				$this->statusMapper->update($status);
			}
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * Finishes all Processed files by copying them to the right path and deleteing the temp files.
	 * Returns the number of processed files.
	 *
	 * @return int
	 */
	private function handleProcessed() {
		try {
			$this->logger->debug('Find processed ocr files and put them to the right dirs.', ['app' => 'ocr']);
			$processed = $this->statusMapper->findAllProcessed($this->userId);
			foreach ($processed as $status) {
				if ($status->getType() === 'tess' && file_exists($status->getTempFile() . '.txt')) {
					//Save the tmp file with newname
					$this->view->file_put_contents($status->getNewName(), file_get_contents($status->getTempFile() . '.txt')); // need .txt because tesseract saves it like this
					// Cleaning temp files
					$this->statusMapper->delete($status);
					exec('rm ' . $status->getTempFile() . '.txt');
				} elseif ($status->getType() === 'mypdf' && file_exists($status->getTempFile())) {
					//Save the tmp file with newname
					$this->view->file_put_contents($status->getNewName(), file_get_contents($status->getTempFile())); // don't need to extend with .pdf / it uses the tmp file to save
					$this->statusMapper->delete($status);
					exec('rm ' . $status->getTempFile());
				} else {
					throw new NotFoundException($this->l10n->t('Temp file does not exist.'));
				}
			}
			return count($processed);
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * Handles all failed orders of ocr processing queue and returns the status objects.
	 *
	 * @return array
	 */
	private function handleFailed() {
		try {
			$failed = $this->statusMapper->findAllFailed($this->userId);
			foreach ($failed as $status) {
				// clean the tempfile
				exec('rm ' . $status->getTempFile());
				// clean from db
				$this->statusMapper->delete($status);
			}
			$this->logger->debug('Following status objects failed: ' . json_encode($failed), ['app' => 'ocr']);
			return $failed;
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}


	/**
	 * Returns a not existing file name for pdf or image processing
	 * protected as of testing issues with static methods. (Actually
	 * it will be mocked partially) FIXME: Change this behaviour as soon as the buidlNotExistingFileName function is not static anymore
	 *
	 * @param FileInfo $fileInfo
	 * @return string
	 */
	protected function buildNewName(FileInfo $fileInfo) {
		// get rid of the .png or .pdf and so on
		$fileName = substr($fileInfo->getName(), 0, -4);
		// eliminate the file name from the path
		$filePath = str_replace($fileInfo->getName(), '', $fileInfo->getPath());
		// and get the path on top of the user/files/ dir
		$filePath = str_replace('/' . $this->userId . '/files', '', $filePath);
		if ($fileInfo->getMimetype() === $this::MIMETYPE_PDF) {
			// PDFs:
			return Files::buildNotExistingFileName($filePath, $fileName . '_OCR.pdf');
		} else {
			// IMAGES:
			return Files::buildNotExistingFileName($filePath, $fileName . '_OCR.txt');
		}
	}

	/**
	 * Returns the fileInfo for each file in files and checks
	 * if it has a allowed mimetype and some other conditions.
	 *
	 * @param array $files
	 * @return array of Files\FileInfo
	 * @throws NotFoundException
	 */
	private function buildFileInfo(array $files) {
		try {
			$fileArray = array();
			foreach ($files as $file) {
				// Check if anything is missing and file type is correct
				if ((!empty($file['path']) || !empty($file['directory'])) && $file['type'] === 'file') {
					// get correct path
					$path = $this->getCorrectPath($file);
					$fileInfo = $this->view->getFileInfo($path);
					$this->checkMimeType($fileInfo);
					array_push($fileArray, $fileInfo);
				} else {
					throw new NotFoundException($this->l10n->t('Wrong path parameter.'));
				}
			}
			return $fileArray;
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * Checks a Mimetype for a specific given FileInfo.
	 * @param Files\FileInfo $fileInfo
	 */
	private function checkMimeType(FileInfo $fileInfo) {
		try {
			if (!$fileInfo || !in_array($fileInfo->getMimetype(), $this::ALLOWED_MIMETYPES)) {
				$this->logger->debug('Getting FileInfo did not work or not included in the ALLOWED_MIMETYPES array.', ['app' => 'ocr']);
				throw new NotFoundException($this->l10n->t('Wrong parameters or wrong mimetype.'));
			}
		} catch (Exception $e) {
			$this->handleException($e);
		}
	}

	/**
	 * Returns the correct path based on delivered file variable
	 * @param $file
	 * @return string
	 */
	private function getCorrectPath($file) {
		if (empty($file['path'])) {
			//Because new updated files have the property directory instead of path
			$file['path'] = $file['directory'];
		}
		if ($file['path'] === '/') {
			$path = $file['path'] . $file['name'];
		} else {
			$path = $file['path'] . '/' . $file['name'];
		}
		return $path;
	}

	/**
	 * Handle the possible thrown Exceptions from all methods of this class.
	 *
	 * @param Exception $e
	 * @throws Exception
	 * @throws NotFoundException
	 */
	private function handleException($e) {
		$this->logger->logException($e, ['app' => 'ocr', 'message' => 'Exception during ocr service function processing']);
		if ($e instanceof NotFoundException) {
			throw new NotFoundException($e->getMessage());
		} else {
			throw $e;
		}
	}
}

1			<?php
2			/**
3			* nextCloud - ocr
4			*
5			* This file is licensed under the Affero General Public License version 3 or
6			* later. See the COPYING file.
7			*
8			* @author Janis Koehr <[email protected]>
9			* @copyright Janis Koehr 2016
10			*/
11
12			namespace OCA\Ocr\Service;
13
14			use Exception;
15			use OC\Files\View;
16			use OCA\Ocr\Db\OcrStatus;
17			use OCA\Ocr\Db\OcrStatusMapper;
18			use OCP\Files;
19			use OCP\Files\FileInfo;
20			use OCP\IConfig;
21			use OCP\IL10N;
22			use OCP\ILogger;
23			use OCP\ITempManager;
24
25
26			/**
27			* Class OcrService
28			* @package OCA\Ocr\Service
29			*/
30			class OcrService {
31
32			/**
33			* @var ILogger
34			*/
35			private $logger;
36
37			/**
38			* @var ITempManager
39			*/
40			private $tempM;
41
42			/**
43			* @var IConfig
44			*/
45			private $config;
46
47			/**
48			* @var QueueService
49			*/
50			private $queueService;
51
52			/**
53			* @var OcrStatusMapper
54			*/
55			private $statusMapper;
56
57			/**
58			* @var View
59			*/
60			private $view;
61
62			/**
63			* @var
64			*/
65			private $userId;
66
67			/**
68			* @var IL10N
69			*/
70			private $l10n;
71
72			/**
73			* Array of allowed mimetypes for ocr processing
74			*/
75			const ALLOWED_MIMETYPES = ['application/pdf', 'image/png', 'image/jpeg', 'image/tiff'];
76
77			/**
78			* the correct mimetype for a pdf file
79			*/
80			const MIMETYPE_PDF = 'application/pdf';
81
82			/**
83			* the only allowed image mimetypes by tesseract
84			*/
85			const MIMETYPES_IMAGE = ['image/png', 'image/jpeg', 'image/tiff'];
86
87			/**
88			* OcrService constructor.
89			*
90			* @param ITempManager $tempManager
91			* @param IConfig $config
92			* @param QueueService $queueService
93			* @param OcrStatusMapper $mapper
94			* @param View $view
95			* @param $userId
96			* @param IL10N $l10n
97			* @param ILogger $logger
98			*/
99	10		public function __construct(ITempManager $tempManager, IConfig $config, QueueService $queueService, OcrStatusMapper $mapper, View $view, $userId, IL10N $l10n, ILogger $logger) {
100	10		$this->logger = $logger;
101	10		$this->tempM = $tempManager;
102	10		$this->config = $config;
103	10		$this->queueService = $queueService;
104	10		$this->statusMapper = $mapper;
105	10		$this->view = $view;
106	10		$this->userId = $userId;
107	10		$this->l10n = $l10n;
108	10		}
109
110			/**
111			* Gets the list of all available tesseract-ocr languages.
112			*
113			* @return array Languages
114			*/
115	3		public function listLanguages() {
116			try {
117	3		$success = -1;
118	3		$this->logger->debug('Fetching languages. ', ['app' => 'ocr']);
119	3		exec('tesseract --list-langs 2>&1', $result, $success);
120	3		if ($success === 0 && count($result) > 0) {
121			if (is_array($result)) {
122			$traineddata = $result;
123			} else {
124			$traineddata = explode(' ', $result);
125			}
126			$languages = array();
127			foreach ($traineddata as $td) {
128			$tdname = trim($td);
129			if (strlen($tdname) === 3) {
130			array_push($languages, $tdname);
131			}
132			}
133			$this->logger->debug('Fetched languages: ' . json_encode($languages), ['app' => 'ocr']);
134			return $languages;
135			} else {
136	3		throw new NotFoundException($this->l10n->t('No languages found.'));
137			}
138	3		} catch (Exception $e) {
139	3		$this->handleException($e);
140			}
141			}
142
143			/**
144			* Processes and prepares the files for ocr.
145			* Sends the stuff to the client in order to ocr async.
146			*
147			* @param string $language
148			* @param array $files
149			* @return string
150			*/
151	4		public function process($language, $files) {
152			try {
153	4		$this->logger->debug('Will now process files: ' . json_encode($files) . ' with language: ' . json_encode($language), ['app' => 'ocr']);
154			// Check if files and language not empty
155	4		if (!empty($files) && !empty($language) && in_array($language, $this->listLanguages())) {
156			// get the array with full fileinfo
157			$fileInfo = $this->buildFileInfo($files);
158			foreach ($fileInfo as $fInfo) {
159			// Check if filelock existing
160			// TODO: FileLock maybe \OC\Files\View::lockFile()
161			// get new name for saving purpose
162			$newName = $this->buildNewName($fInfo);
163
164			// create a temp file for ocr processing purposes
165			$tempFile = $this->tempM->getTemporaryFile();
166
167			// set the running type
168			if ($fInfo->getMimetype() === $this::MIMETYPE_PDF) {
169			$ftype = 'mypdf';
170			} else {
171			$ftype = 'tess';
172			}
173
174			// Create status object
175			$status = new OcrStatus('PENDING', $fInfo->getId(), $newName, $tempFile, $ftype, $this->userId);
176
177			// Init client and send task / job
178			// Feed the worker
179			$this->queueService->clientSend($status, $this->config->getSystemValue('datadirectory'), $fInfo->getPath(), $language, \OC::$SERVERROOT);
180			}
181			return 'PROCESSING';
182			} else {
183	2		throw new NotFoundException($this->l10n->t('Empty passed parameters.'));
184			}
185	4		} catch (Exception $e) {
186	4		$this->handleException($e);
187			}
188			}
189
190			/**
191			* A function which returns the JSONResponse for all required status checks and tasks.
192			* It will check for already processed, pending and failed ocr tasks and return them as needed.
193			*
194			* @return string
195			*/
196			public function status() {
197			try {
198			// TODO: release lock
199			$processed = $this->handleProcessed();
200
201			$failed = count($this->handleFailed());
202
203			$pending = count($this->statusMapper->findAllPending($this->userId));
204
205			return ['processed' => $processed, 'failed' => $failed, 'pending' => $pending];
206			} catch (Exception $e) {
207			$this->handleException($e);
208			}
209			}
210
211			/**
212			* The command ocr:complete for occ will call this function in order to set the status.
213			* the worker should call it automatically after each processing step.
214			*
215			* @param $statusId
216			* @param boolean $failed
217			*/
218	3		public function complete($statusId, $failed) {
219			try {
220	3		$status = $this->statusMapper->find($statusId);
221	2		if (!$failed) {
222	1		$status->setStatus('PROCESSED');
223	1		$this->statusMapper->update($status);
224	1		} else {
225	1		$status->setStatus('FAILED');
226	1		$this->statusMapper->update($status);
227			}
228	3		} catch (Exception $e) {
229	1		$this->handleException($e);
230			}
231	2		}
232
233			/**
234			* Finishes all Processed files by copying them to the right path and deleteing the temp files.
235			* Returns the number of processed files.
236			*
237			* @return int
238			*/
239			private function handleProcessed() {
240			try {
241			$this->logger->debug('Find processed ocr files and put them to the right dirs.', ['app' => 'ocr']);
242			$processed = $this->statusMapper->findAllProcessed($this->userId);
243			foreach ($processed as $status) {
244			if ($status->getType() === 'tess' && file_exists($status->getTempFile() . '.txt')) {
245			//Save the tmp file with newname
246			$this->view->file_put_contents($status->getNewName(), file_get_contents($status->getTempFile() . '.txt')); // need .txt because tesseract saves it like this
247			// Cleaning temp files
248			$this->statusMapper->delete($status);
249			exec('rm ' . $status->getTempFile() . '.txt');
250			} elseif ($status->getType() === 'mypdf' && file_exists($status->getTempFile())) {
251			//Save the tmp file with newname
252			$this->view->file_put_contents($status->getNewName(), file_get_contents($status->getTempFile())); // don't need to extend with .pdf / it uses the tmp file to save
253			$this->statusMapper->delete($status);
254			exec('rm ' . $status->getTempFile());
255			} else {
256			throw new NotFoundException($this->l10n->t('Temp file does not exist.'));
257			}
258			}
259			return count($processed);
260			} catch (Exception $e) {
261			$this->handleException($e);
262			}
263			}
264
265			/**
266			* Handles all failed orders of ocr processing queue and returns the status objects.
267			*
268			* @return array
269			*/
270			private function handleFailed() {
271			try {
272			$failed = $this->statusMapper->findAllFailed($this->userId);
273			foreach ($failed as $status) {
274			// clean the tempfile
275			exec('rm ' . $status->getTempFile());
276			// clean from db
277			$this->statusMapper->delete($status);
278			}
279			$this->logger->debug('Following status objects failed: ' . json_encode($failed), ['app' => 'ocr']);
280			return $failed;
281			} catch (Exception $e) {
282			$this->handleException($e);
283			}
284			}
285
286
287			/**
288			* Returns a not existing file name for pdf or image processing
289			* protected as of testing issues with static methods. (Actually
290			* it will be mocked partially) FIXME: Change this behaviour as soon as the buidlNotExistingFileName function is not static anymore
291			*
292			* @param FileInfo $fileInfo
293			* @return string
294			*/
295			protected function buildNewName(FileInfo $fileInfo) {
296			// get rid of the .png or .pdf and so on
297			$fileName = substr($fileInfo->getName(), 0, -4);
298			// eliminate the file name from the path
299			$filePath = str_replace($fileInfo->getName(), '', $fileInfo->getPath());
300			// and get the path on top of the user/files/ dir
301			$filePath = str_replace('/' . $this->userId . '/files', '', $filePath);
302			if ($fileInfo->getMimetype() === $this::MIMETYPE_PDF) {
303			// PDFs:
304			return Files::buildNotExistingFileName($filePath, $fileName . '_OCR.pdf');
305			} else {
306			// IMAGES:
307			return Files::buildNotExistingFileName($filePath, $fileName . '_OCR.txt');
308			}
309			}
310
311			/**
312			* Returns the fileInfo for each file in files and checks
313			* if it has a allowed mimetype and some other conditions.
314			*
315			* @param array $files
316			* @return array of Files\FileInfo
317			* @throws NotFoundException
318			*/
319			private function buildFileInfo(array $files) {
320			try {
321			$fileArray = array();
322			foreach ($files as $file) {
323			// Check if anything is missing and file type is correct
324			if ((!empty($file['path']) \|\| !empty($file['directory'])) && $file['type'] === 'file') {
325			// get correct path
326			$path = $this->getCorrectPath($file);
327			$fileInfo = $this->view->getFileInfo($path);
328			$this->checkMimeType($fileInfo);
329			array_push($fileArray, $fileInfo);
330			} else {
331			throw new NotFoundException($this->l10n->t('Wrong path parameter.'));
332			}
333			}
334			return $fileArray;
335			} catch (Exception $e) {
336			$this->handleException($e);
337			}
338			}
339
340			/**
341			* Checks a Mimetype for a specific given FileInfo.
342			* @param Files\FileInfo $fileInfo
343			*/
344			private function checkMimeType(FileInfo $fileInfo) {
345			try {
346			if (!$fileInfo \|\| !in_array($fileInfo->getMimetype(), $this::ALLOWED_MIMETYPES)) {
347			$this->logger->debug('Getting FileInfo did not work or not included in the ALLOWED_MIMETYPES array.', ['app' => 'ocr']);
348			throw new NotFoundException($this->l10n->t('Wrong parameters or wrong mimetype.'));
349			}
350			} catch (Exception $e) {
351			$this->handleException($e);
352			}
353			}
354
355			/**
356			* Returns the correct path based on delivered file variable
357			* @param $file
358			* @return string
359			*/
360			private function getCorrectPath($file) {
361			if (empty($file['path'])) {
362			//Because new updated files have the property directory instead of path
363			$file['path'] = $file['directory'];
364			}
365			if ($file['path'] === '/') {
366			$path = $file['path'] . $file['name'];
367			} else {
368			$path = $file['path'] . '/' . $file['name'];
369			}
370			return $path;
371			}
372
373			/**
374			* Handle the possible thrown Exceptions from all methods of this class.
375			*
376			* @param Exception $e
377			* @throws Exception
378			* @throws NotFoundException
379			*/
380	6	View Code Duplication	private function handleException($e) {
381	6		$this->logger->logException($e, ['app' => 'ocr', 'message' => 'Exception during ocr service function processing']);
382	6		if ($e instanceof NotFoundException) {
383	6		throw new NotFoundException($e->getMessage());
384			} else {
385			throw $e;
386			}
387			}
388			}

janis91 / ocr

Pull Request — master (#17)

OcrService::process() C

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like