|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* Nextcloud - OCR |
|
5
|
|
|
* This file is licensed under the Affero General Public License version 3 or |
|
6
|
|
|
* later. |
|
7
|
|
|
* See the COPYING file. |
|
8
|
|
|
* |
|
9
|
|
|
* @author Janis Koehr <[email protected]> |
|
10
|
|
|
* @copyright Janis Koehr 2017 |
|
11
|
|
|
*/ |
|
12
|
|
|
namespace OCA\Ocr\Service; |
|
13
|
|
|
|
|
14
|
|
|
use OCA\Ocr\Db\OcrJobMapper; |
|
15
|
|
|
use OC\Files\View; |
|
16
|
|
|
use OCP\ILogger; |
|
17
|
|
|
use OCP\IL10N; |
|
18
|
|
|
use OCP\ITempManager; |
|
19
|
|
|
use OCA\Ocr\Db\OcrJob; |
|
20
|
|
|
use OCA\Ocr\Constants\OcrConstants; |
|
21
|
|
|
use OCA\Ocr\Util\PHPUtil; |
|
22
|
|
|
use Exception; |
|
23
|
|
|
use OCP\AppFramework\Db\DoesNotExistException; |
|
24
|
|
|
use OCA\Ocr\Util\FileUtil; |
|
25
|
|
|
|
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* Class JobService |
|
29
|
|
|
* |
|
30
|
|
|
* @package OCA\Ocr\Service |
|
31
|
|
|
*/ |
|
32
|
|
|
class JobService { |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* |
|
36
|
|
|
* @var ILogger |
|
37
|
|
|
*/ |
|
38
|
|
|
private $logger; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* |
|
42
|
|
|
* @var RedisService |
|
43
|
|
|
*/ |
|
44
|
|
|
private $redisService; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* |
|
48
|
|
|
* @var OcrJobMapper |
|
49
|
|
|
*/ |
|
50
|
|
|
private $jobMapper; |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* |
|
54
|
|
|
* @var View |
|
55
|
|
|
*/ |
|
56
|
|
|
private $view; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* |
|
60
|
|
|
* @var String |
|
61
|
|
|
*/ |
|
62
|
|
|
private $userId; |
|
63
|
|
|
|
|
64
|
|
|
/** |
|
65
|
|
|
* |
|
66
|
|
|
* @var IL10N |
|
67
|
|
|
*/ |
|
68
|
|
|
private $l10n; |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* |
|
72
|
|
|
* @var FileService |
|
73
|
|
|
*/ |
|
74
|
|
|
private $fileService; |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* |
|
78
|
|
|
* @var ITempManager |
|
79
|
|
|
*/ |
|
80
|
|
|
private $tempM; |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* |
|
84
|
|
|
* @var AppConfigService |
|
85
|
|
|
*/ |
|
86
|
|
|
private $appConfigService; |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* |
|
90
|
|
|
* @var PHPUtil |
|
91
|
|
|
*/ |
|
92
|
|
|
private $phpUtil; |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* |
|
96
|
|
|
* @var FileUtil |
|
97
|
|
|
*/ |
|
98
|
|
|
private $fileUtil; |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* JobService constructor. |
|
102
|
|
|
* |
|
103
|
|
|
* @param IL10N $l10n |
|
104
|
|
|
* @param ILogger $logger |
|
105
|
|
|
* @param string $userId |
|
106
|
|
|
* @param View $view |
|
107
|
|
|
* @param RedisService $queueService |
|
108
|
|
|
* @param OcrJobMapper $mapper |
|
109
|
|
|
* @param FileService $fileService |
|
110
|
|
|
* @param AppConfigService $appConfigService |
|
111
|
|
|
* @param PHPUtil $phpUtil |
|
112
|
|
|
* @param FileUtil $fileUtil |
|
113
|
|
|
*/ |
|
114
|
22 |
|
public function __construct(IL10N $l10n, ILogger $logger, $userId, View $view, ITempManager $tempManager, |
|
115
|
|
|
RedisService $queueService, OcrJobMapper $mapper, FileService $fileService, |
|
116
|
|
|
AppConfigService $appConfigService, PHPUtil $phpUtil, FileUtil $fileUtil) { |
|
117
|
22 |
|
$this->logger = $logger; |
|
118
|
22 |
|
$this->redisService = $queueService; |
|
119
|
22 |
|
$this->jobMapper = $mapper; |
|
120
|
22 |
|
$this->view = $view; |
|
121
|
22 |
|
$this->userId = $userId; |
|
122
|
22 |
|
$this->l10n = $l10n; |
|
123
|
22 |
|
$this->fileService = $fileService; |
|
124
|
22 |
|
$this->tempM = $tempManager; |
|
125
|
22 |
|
$this->appConfigService = $appConfigService; |
|
126
|
22 |
|
$this->phpUtil = $phpUtil; |
|
127
|
22 |
|
$this->fileUtil = $fileUtil; |
|
128
|
22 |
|
} |
|
129
|
|
|
|
|
130
|
|
|
/** |
|
131
|
|
|
* Processes and prepares the files for OCR. |
|
132
|
|
|
* Sends the stuff to the client in order to OCR async. |
|
133
|
|
|
* |
|
134
|
|
|
* @param string[] $languages |
|
135
|
|
|
* @param array $files |
|
136
|
|
|
* @return string |
|
137
|
|
|
*/ |
|
138
|
9 |
|
public function process($languages, $files) { |
|
139
|
|
|
try { |
|
140
|
9 |
|
$this->logger->debug('Will now process files: {files} with languages: {languages}', |
|
141
|
|
|
[ |
|
142
|
9 |
|
'files' => json_encode($files), |
|
143
|
9 |
|
'languages' => json_encode($languages) |
|
144
|
|
|
]); |
|
145
|
|
|
// Check if files and language not empty |
|
146
|
9 |
|
$noLang = $this->noLanguage($languages); |
|
147
|
9 |
|
if (!empty($files) && ($this->checkForAcceptedLanguages($languages) || $noLang)) { |
|
148
|
|
|
// language part: |
|
149
|
6 |
|
if ($noLang) { |
|
150
|
5 |
|
$languages = []; |
|
151
|
|
|
} |
|
152
|
|
|
// file part: |
|
153
|
6 |
|
$fileInfo = $this->fileService->buildFileInfo($files); |
|
154
|
6 |
|
foreach ($fileInfo as $fInfo) { |
|
155
|
|
|
// Check Shared |
|
156
|
6 |
|
$shared = $this->fileService->checkSharedWithInitiator($fInfo); |
|
157
|
6 |
|
$target = $this->fileService->buildTarget($fInfo, $shared); |
|
158
|
6 |
|
$source = $this->fileService->buildSource($fInfo, $shared); |
|
159
|
|
|
// set the running type |
|
160
|
6 |
|
$fType = $this->fileService->getCorrectType($fInfo); |
|
161
|
|
|
// create a temp file for ocr processing purposes |
|
162
|
6 |
|
$tempFile = $this->getTempFile($fType); |
|
163
|
|
|
// Create job object |
|
164
|
2 |
|
$job = new OcrJob(OcrConstants::STATUS_PENDING, $source, $target, $tempFile, $fType, $this->userId, |
|
165
|
2 |
|
false, $fInfo->getName(), null); |
|
|
|
|
|
|
166
|
|
|
// Init client and send task / job |
|
167
|
|
|
// Feed the worker |
|
168
|
2 |
|
$this->redisService->sendJob($job, $languages); |
|
169
|
|
|
} |
|
170
|
2 |
|
return 'PROCESSING'; |
|
171
|
|
|
} else { |
|
172
|
3 |
|
throw new NotFoundException($this->l10n->t('Empty parameters passed.')); |
|
173
|
|
|
} |
|
174
|
7 |
|
} catch (Exception $e) { |
|
175
|
7 |
|
$this->handleException($e); |
|
176
|
|
|
} |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
|
|
/** |
|
180
|
|
|
* Delete an ocr job for a given id and userId. |
|
181
|
|
|
* |
|
182
|
|
|
* @param |
|
183
|
|
|
* $jobId |
|
184
|
|
|
* @param string $userId |
|
185
|
|
|
* @return OcrJob |
|
186
|
|
|
*/ |
|
187
|
3 |
|
public function deleteJob($jobId, $userId) { |
|
188
|
|
|
try { |
|
189
|
3 |
|
$job = $this->jobMapper->find($jobId); |
|
190
|
2 |
|
if ($job->getUserId() !== $userId) { |
|
191
|
1 |
|
throw new NotFoundException($this->l10n->t('Cannot delete because of wrong owner.')); |
|
192
|
|
|
} else { |
|
193
|
1 |
|
$job = $this->jobMapper->delete($job); |
|
194
|
|
|
} |
|
195
|
1 |
|
$job->setTarget(null); |
|
196
|
1 |
|
$job->setSource(null); |
|
197
|
1 |
|
$job->setStatus(null); |
|
198
|
1 |
|
$job->setTempFile(null); |
|
199
|
1 |
|
$job->setType(null); |
|
200
|
1 |
|
$job->setUserId(null); |
|
201
|
1 |
|
$job->setErrorDisplayed(null); |
|
202
|
1 |
|
$job->setErrorLog(null); |
|
203
|
1 |
|
return $job; |
|
204
|
2 |
|
} catch (Exception $e) { |
|
205
|
2 |
|
if ($e instanceof DoesNotExistException) { |
|
|
|
|
|
|
206
|
1 |
|
$ex = new NotFoundException($this->l10n->t('Cannot delete because of wrong ID.')); |
|
207
|
1 |
|
$this->handleException($ex); |
|
208
|
|
|
} else { |
|
209
|
1 |
|
$this->handleException($e); |
|
210
|
|
|
} |
|
211
|
|
|
} |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
/** |
|
215
|
|
|
* Gets all job objects for a specific user. |
|
216
|
|
|
* |
|
217
|
|
|
* @param string $userId |
|
218
|
|
|
* @return OcrJob[] |
|
219
|
|
|
*/ |
|
220
|
1 |
|
public function getAllJobsForUser($userId) { |
|
221
|
|
|
try { |
|
222
|
1 |
|
$jobs = $this->jobMapper->findAll($userId); |
|
223
|
1 |
|
$jobsNew = array(); |
|
224
|
1 |
|
foreach ($jobs as $job) { |
|
225
|
1 |
|
$job->setTarget(null); |
|
226
|
1 |
|
$job->setSource(null); |
|
227
|
1 |
|
$job->setTempFile(null); |
|
228
|
1 |
|
$job->setType(null); |
|
229
|
1 |
|
$job->setUserId(null); |
|
230
|
1 |
|
$job->setErrorDisplayed(null); |
|
231
|
1 |
|
array_push($jobsNew, $job); |
|
232
|
|
|
} |
|
233
|
1 |
|
return $jobsNew; |
|
234
|
|
|
} catch (Exception $e) { |
|
235
|
|
|
$this->handleException($e); |
|
236
|
|
|
} |
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
|
|
/** |
|
240
|
|
|
* The function checks if there are finished jobs to process finally. |
|
241
|
|
|
* |
|
242
|
|
|
* @throws NotFoundException |
|
243
|
|
|
*/ |
|
244
|
2 |
|
public function checkForFinishedJobs() { |
|
245
|
|
|
try { |
|
246
|
2 |
|
$finishedJobs = $this->redisService->readingFinishedJobs(); |
|
247
|
2 |
|
foreach ($finishedJobs as $finishedJob) { |
|
248
|
1 |
|
$fJob = $this->transformJob($finishedJob); |
|
249
|
1 |
|
$this->logger->debug('The following job finished: {job}', |
|
250
|
|
|
[ |
|
251
|
1 |
|
'job' => $fJob |
|
252
|
|
|
]); |
|
253
|
1 |
|
$this->jobFinished($fJob->id, $fJob->error, $fJob->log); |
|
254
|
|
|
} |
|
255
|
1 |
|
} catch (Exception $e) { |
|
256
|
1 |
|
throw new NotFoundException($this->l10n->t('Reading the finished jobs from redis went wrong.')); |
|
257
|
|
|
} |
|
258
|
1 |
|
} |
|
259
|
|
|
|
|
260
|
|
|
/** |
|
261
|
|
|
* Finishes all Processed files by copying them to the right path and deleteing the temp files. |
|
262
|
|
|
* Returns the number of processed files. |
|
263
|
|
|
* |
|
264
|
|
|
* @return array |
|
265
|
|
|
*/ |
|
266
|
2 |
|
public function handleProcessed() { |
|
267
|
|
|
try { |
|
268
|
2 |
|
$this->logger->debug('Check if files were processed by ocr and if so, put them to the right dirs.'); |
|
269
|
2 |
|
$processed = $this->jobMapper->findAllProcessed($this->userId); |
|
270
|
2 |
|
foreach ($processed as $job) { |
|
271
|
2 |
|
if ($this->fileUtil->fileExists($job->getTempFile())) { |
|
272
|
|
|
// Save the tmp file with newname |
|
273
|
1 |
|
$this->view->file_put_contents($job->getTarget(), |
|
274
|
1 |
|
$this->fileUtil->getFileContents($job->getTempFile())); |
|
275
|
1 |
|
$this->jobMapper->delete($job); |
|
276
|
1 |
|
$this->fileUtil->execRemove($job->getTempFile()); |
|
277
|
|
|
} else { |
|
278
|
1 |
|
$job->setStatus(OcrConstants::STATUS_FAILED); |
|
279
|
1 |
|
$job->setErrorLog('Temp file does not exist.'); |
|
280
|
1 |
|
$this->jobMapper->update($job); |
|
281
|
2 |
|
throw new NotFoundException($this->l10n->t('Temp file does not exist.')); |
|
282
|
|
|
} |
|
283
|
|
|
} |
|
284
|
1 |
|
return $processed; |
|
285
|
1 |
|
} catch (Exception $e) { |
|
286
|
1 |
|
$this->handleException($e); |
|
287
|
|
|
} |
|
288
|
|
|
} |
|
289
|
|
|
|
|
290
|
|
|
/** |
|
291
|
|
|
* Handles all failed orders of ocr processing queue and returns the job objects. |
|
292
|
|
|
* |
|
293
|
|
|
* @return array |
|
294
|
|
|
*/ |
|
295
|
1 |
|
public function handleFailed() { |
|
296
|
|
|
try { |
|
297
|
1 |
|
$failed = $this->jobMapper->findAllFailed($this->userId); |
|
298
|
1 |
|
foreach ($failed as $job) { |
|
299
|
|
|
// clean the tempfile |
|
300
|
1 |
|
$this->fileUtil->execRemove($job->getTempFile()); |
|
301
|
|
|
// set error displayed |
|
302
|
1 |
|
$job->setErrorDisplayed(true); |
|
303
|
1 |
|
$this->jobMapper->update($job); |
|
304
|
|
|
} |
|
305
|
1 |
|
$this->logger->debug('Following jobs failed: {failed}', |
|
306
|
|
|
[ |
|
307
|
1 |
|
'failed' => json_encode($failed) |
|
308
|
|
|
]); |
|
309
|
1 |
|
return $failed; |
|
310
|
|
|
} catch (Exception $e) { |
|
311
|
|
|
$this->handleException($e); |
|
312
|
|
|
} |
|
313
|
|
|
} |
|
314
|
|
|
|
|
315
|
|
|
/** |
|
316
|
|
|
* The function the worker will call in order to set the jobs status. |
|
317
|
|
|
* The worker should call it automatically after each processing step. |
|
318
|
|
|
* |
|
319
|
|
|
* @param integer $jobId |
|
320
|
|
|
* @param boolean $error |
|
321
|
|
|
* @param string $log |
|
322
|
|
|
*/ |
|
323
|
1 |
|
private function jobFinished($jobId, $error, $log) { |
|
324
|
|
|
try { |
|
325
|
1 |
|
$job = $this->jobMapper->find($jobId); |
|
326
|
1 |
|
if (!$error) { |
|
327
|
1 |
|
$job->setStatus(OcrConstants::STATUS_PROCESSED); |
|
328
|
1 |
|
$this->jobMapper->update($job); |
|
329
|
|
|
} else { |
|
330
|
1 |
|
$job->setStatus(OcrConstants::STATUS_FAILED); |
|
331
|
1 |
|
$job->setErrorLog($log); |
|
332
|
1 |
|
$this->jobMapper->update($job); |
|
333
|
1 |
|
$this->logger->error($log); |
|
334
|
|
|
} |
|
335
|
|
|
} catch (Exception $e) { |
|
336
|
|
|
$this->handleException($e); |
|
337
|
|
|
} |
|
338
|
1 |
|
} |
|
339
|
|
|
|
|
340
|
|
|
/** |
|
341
|
|
|
* Gives a temp file name back depending on the type of the OCR. |
|
342
|
|
|
* Later in the worker this file is used as an output. |
|
343
|
|
|
* |
|
344
|
|
|
* @param integer $type |
|
345
|
|
|
* @return string |
|
346
|
|
|
*/ |
|
347
|
6 |
|
private function getTempFile($type) { |
|
348
|
6 |
|
if ($type === OcrConstants::TESSERACT) { |
|
349
|
5 |
|
$fileName = $this->phpUtil->tempnamWrapper($this->tempM->getTempBaseDir(), OcrConstants::TEMPFILE_PREFIX); |
|
350
|
5 |
|
$this->phpUtil->unlinkWrapper($fileName); |
|
351
|
4 |
|
$fileNameWithPostfix = $fileName . '.txt'; |
|
352
|
4 |
|
$this->phpUtil->touchWrapper($fileNameWithPostfix); |
|
353
|
3 |
|
$this->phpUtil->chmodWrapper($fileNameWithPostfix, 0600); |
|
354
|
2 |
|
return $fileNameWithPostfix; |
|
355
|
|
|
} else { |
|
356
|
3 |
|
return $this->phpUtil->tempnamWrapper($this->tempM->getTempBaseDir(), OcrConstants::TEMPFILE_PREFIX); |
|
357
|
|
|
} |
|
358
|
|
|
} |
|
359
|
|
|
|
|
360
|
|
|
/** |
|
361
|
|
|
* Takes care of transforming an incoming finished job into a php readable object. |
|
362
|
|
|
* |
|
363
|
|
|
* @param string $job |
|
364
|
|
|
* @throws NotFoundException |
|
365
|
|
|
* @return mixed |
|
366
|
|
|
*/ |
|
367
|
1 |
|
private function transformJob($job) { |
|
368
|
1 |
|
$decoded = json_decode($job); |
|
369
|
1 |
|
if ($decoded !== null && isset($decoded->id)) { |
|
370
|
1 |
|
return $decoded; |
|
371
|
|
|
} else { |
|
372
|
|
|
$this->logger->debug('The finished job retrieved by Redis was corrupt.'); |
|
373
|
|
|
throw new NotFoundException('The finished job retrieved by Redis was corrupt.'); |
|
374
|
|
|
} |
|
375
|
|
|
} |
|
376
|
|
|
|
|
377
|
|
|
/** |
|
378
|
|
|
* Checks if the given languages are supported or not. |
|
379
|
|
|
* |
|
380
|
|
|
* @param string[] $languages |
|
381
|
|
|
* @return boolean |
|
382
|
|
|
*/ |
|
383
|
8 |
|
private function checkForAcceptedLanguages($languages) { |
|
384
|
8 |
|
$installedLanguages = explode(';', $this->appConfigService->getAppValue('languages')); |
|
385
|
8 |
|
if (count(array_diff($languages, $installedLanguages)) === 0) { |
|
386
|
1 |
|
return true; |
|
387
|
|
|
} else { |
|
388
|
7 |
|
return false; |
|
389
|
|
|
} |
|
390
|
|
|
} |
|
391
|
|
|
|
|
392
|
|
|
/** |
|
393
|
|
|
* Checks if the process should be initiated without any language specified. |
|
394
|
|
|
* |
|
395
|
|
|
* @param string[] $languages |
|
396
|
|
|
* @return boolean |
|
397
|
|
|
*/ |
|
398
|
9 |
|
private function noLanguage($languages) { |
|
399
|
9 |
|
if (in_array('any', $languages)) { |
|
400
|
6 |
|
return true; |
|
401
|
|
|
} else { |
|
402
|
3 |
|
return false; |
|
403
|
|
|
} |
|
404
|
|
|
} |
|
405
|
|
|
|
|
406
|
|
|
/** |
|
407
|
|
|
* Handle the possible thrown Exceptions from all methods of this class. |
|
408
|
|
|
* |
|
409
|
|
|
* @param Exception $e |
|
410
|
|
|
* @throws Exception |
|
411
|
|
|
* @throws NotFoundException |
|
412
|
|
|
*/ |
|
413
|
10 |
|
private function handleException($e) { |
|
414
|
10 |
|
$this->logger->logException($e, |
|
415
|
|
|
[ |
|
416
|
|
|
'message' => 'Exception during job service function processing' |
|
417
|
10 |
|
]); |
|
418
|
10 |
|
throw $e; |
|
419
|
|
|
} |
|
420
|
|
|
} |