|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* Nextcloud - OCR |
|
5
|
|
|
* |
|
6
|
|
|
* This file is licensed under the Affero General Public License version 3 or |
|
7
|
|
|
* later. 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\OcrJob; |
|
15
|
|
|
use OCA\Ocr\Db\FileMapper; |
|
16
|
|
|
use OCA\Ocr\Db\File; |
|
17
|
|
|
use OCP\ILogger; |
|
18
|
|
|
use OCA\Ocr\Db\ShareMapper; |
|
19
|
|
|
use OCP\IL10N; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Class FileService |
|
23
|
|
|
* |
|
24
|
|
|
* @package OCA\Ocr\Service |
|
25
|
|
|
*/ |
|
26
|
|
|
class FileService { |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* |
|
30
|
|
|
* @var ILogger |
|
31
|
|
|
*/ |
|
32
|
|
|
private $logger; |
|
33
|
|
|
|
|
34
|
|
|
/** |
|
35
|
|
|
* |
|
36
|
|
|
* @var FileMapper |
|
37
|
|
|
*/ |
|
38
|
|
|
private $fileMapper; |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* |
|
42
|
|
|
* @var ShareMapper |
|
43
|
|
|
*/ |
|
44
|
|
|
private $shareMapper; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* |
|
48
|
|
|
* @var string |
|
49
|
|
|
*/ |
|
50
|
|
|
private $userId; |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* |
|
54
|
|
|
* @var IL10N |
|
55
|
|
|
*/ |
|
56
|
|
|
private $l10n; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Supported MIME types for OCR processing |
|
60
|
|
|
*/ |
|
61
|
|
|
private static $ALLOWED_MIMETYPES = [ |
|
|
|
|
|
|
62
|
|
|
'application/pdf', |
|
63
|
|
|
'image/png', |
|
64
|
|
|
'image/jpeg', |
|
65
|
|
|
'image/tiff', |
|
66
|
|
|
'image/jp2', |
|
67
|
|
|
'image/jpm', |
|
68
|
|
|
'image/jpx', |
|
69
|
|
|
'image/webp', |
|
70
|
|
|
'image/gif' |
|
71
|
|
|
]; |
|
72
|
|
|
|
|
73
|
|
|
/** |
|
74
|
|
|
* The correct MIME type for a PDF file |
|
75
|
|
|
*/ |
|
76
|
|
|
private static $MIMETYPE_PDF = 'application/pdf'; |
|
|
|
|
|
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* The supported image MIME types by tesseract |
|
80
|
|
|
*/ |
|
81
|
|
|
private static $MIMETYPES_IMAGE = [ |
|
|
|
|
|
|
82
|
|
|
'image/png', |
|
83
|
|
|
'image/jpeg', |
|
84
|
|
|
'image/tiff', |
|
85
|
|
|
'image/jp2', |
|
86
|
|
|
'image/jpm', |
|
87
|
|
|
'image/jpx', |
|
88
|
|
|
'image/webp', |
|
89
|
|
|
'image/gif' |
|
90
|
|
|
]; |
|
91
|
|
|
public function __construct(IL10N $l10n, ILogger $logger, $userId, FileMapper $fileMapper, ShareMapper $shareMapper) { |
|
92
|
|
|
$this->l10n = $l10n; |
|
93
|
|
|
$this->logger = $logger; |
|
94
|
|
|
$this->userId = $userId; |
|
95
|
|
|
$this->fileMapper = $fileMapper; |
|
96
|
|
|
$this->shareMapper = $shareMapper; |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
/** |
|
100
|
|
|
* Checks if shared with the process initiator |
|
101
|
|
|
* |
|
102
|
|
|
* @param File $fileInfo |
|
103
|
|
|
* @return boolean|null |
|
104
|
|
|
*/ |
|
105
|
|
|
public function checkSharedWithInitiator($fileInfo) { |
|
106
|
|
|
try { |
|
107
|
|
|
$owner = str_replace ( 'home::', '', $fileInfo->getStoragename () ); |
|
108
|
|
|
if ($this->userId === $owner) { |
|
109
|
|
|
// user is owner (no shared file) |
|
110
|
|
|
return false; |
|
111
|
|
|
} else { |
|
112
|
|
|
// user is not owner (shared file) |
|
113
|
|
|
return true; |
|
114
|
|
|
} |
|
115
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
|
|
116
|
|
|
$this->handleException ( $e ); |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* Builds the target name. |
|
122
|
|
|
* |
|
123
|
|
|
* @param File $fileInfo |
|
124
|
|
|
* @param boolean $shared |
|
125
|
|
|
* @return string |
|
126
|
|
|
*/ |
|
127
|
|
|
public function buildTarget($fileInfo, $shared) { |
|
128
|
|
|
if ($shared) { |
|
129
|
|
|
$target = $this->buildTargetForShared ( $fileInfo ); |
|
130
|
|
|
} else { |
|
131
|
|
|
$target = $this->buildTargetNotForShared ( $fileInfo ); |
|
132
|
|
|
} |
|
133
|
|
|
return $target; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
/** |
|
137
|
|
|
* Builds the source name. |
|
138
|
|
|
* |
|
139
|
|
|
* @param File $fileInfo |
|
140
|
|
|
* @param boolean $shared |
|
141
|
|
|
* @return string |
|
142
|
|
|
*/ |
|
143
|
|
|
public function buildSource($fileInfo, $shared) { |
|
144
|
|
|
$source = $fileInfo->getPath (); |
|
145
|
|
|
if ($shared) { |
|
146
|
|
|
$source = str_replace ( 'home::', '', $fileInfo->getStoragename () ) . '/' . $source; |
|
147
|
|
|
} else { |
|
148
|
|
|
$source = $this->userId . '/' . $source; |
|
149
|
|
|
} |
|
150
|
|
|
return $source; |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
/** |
|
154
|
|
|
* Returns the fileInfo for each file in files and checks |
|
155
|
|
|
* if it has a allowed MIME type and some other conditions. |
|
156
|
|
|
* |
|
157
|
|
|
* @param array $files |
|
158
|
|
|
* @return File[] |
|
159
|
|
|
* @throws NotFoundException |
|
160
|
|
|
*/ |
|
161
|
|
|
public function buildFileInfo($files) { |
|
162
|
|
|
try { |
|
163
|
|
|
$fileArray = array (); |
|
164
|
|
|
foreach ( $files as $file ) { |
|
165
|
|
|
// Check if anything is missing and file type is correct |
|
166
|
|
|
if (! empty ( $file ['id'] )) { |
|
167
|
|
|
|
|
168
|
|
|
$fileInfo = $this->fileMapper->find ( $file ['id'] ); |
|
169
|
|
|
$this->checkMimeType ( $fileInfo ); |
|
170
|
|
|
|
|
171
|
|
|
array_push ( $fileArray, $fileInfo ); |
|
172
|
|
|
} else { |
|
173
|
|
|
throw new NotFoundException ( $this->l10n->t ( 'Wrong parameter.' ) ); |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
return $fileArray; |
|
177
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
|
|
178
|
|
|
$this->handleException ( $e ); |
|
179
|
|
|
} |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
|
|
/** |
|
183
|
|
|
* Removes ".txt" from the newName of a OCR job |
|
184
|
|
|
* |
|
185
|
|
|
* @param $job OcrJob |
|
186
|
|
|
* @return string |
|
187
|
|
|
*/ |
|
188
|
|
|
public function removeFileExtension($job) { |
|
189
|
|
|
return substr ( $job->getTarget (), 0, strrpos ( $job->getTarget (), '_OCR' ) ); |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
/** |
|
193
|
|
|
* Determines the correct type for the ocr process worker. |
|
194
|
|
|
* TODO: adjust according to expected |
|
195
|
|
|
* |
|
196
|
|
|
* @param File $fileInfo |
|
197
|
|
|
* @return string |
|
198
|
|
|
*/ |
|
199
|
|
|
public function getCorrectType($fileInfo) { |
|
200
|
|
|
if ($fileInfo->getMimetype () === FileService::$MIMETYPE_PDF) { |
|
201
|
|
|
return 'mypdf'; |
|
202
|
|
|
} else { |
|
203
|
|
|
return 'tess'; |
|
204
|
|
|
} |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
/** |
|
208
|
|
|
* Executes the exec function with a remove statement for a given file path. |
|
209
|
|
|
* |
|
210
|
|
|
* @codeCoverageIgnore |
|
211
|
|
|
* |
|
212
|
|
|
* @param string $pathToFile |
|
213
|
|
|
*/ |
|
214
|
|
|
public function execRemove($pathToFile) { |
|
215
|
|
|
exec ( 'rm ' . $pathToFile ); |
|
216
|
|
|
} |
|
217
|
|
|
|
|
218
|
|
|
/** |
|
219
|
|
|
* Wraps the static file_get_contents method of php. |
|
220
|
|
|
* |
|
221
|
|
|
* @codeCoverageIgnore |
|
222
|
|
|
* |
|
223
|
|
|
* @param string $pathToFile |
|
224
|
|
|
* @return string |
|
225
|
|
|
*/ |
|
226
|
|
|
public function getFileContents($pathToFile) { |
|
227
|
|
|
return file_get_contents ( $pathToFile ); |
|
228
|
|
|
} |
|
229
|
|
|
|
|
230
|
|
|
/** |
|
231
|
|
|
* Wraps the static file_exists method of php. |
|
232
|
|
|
* |
|
233
|
|
|
* @codeCoverageIgnore |
|
234
|
|
|
* |
|
235
|
|
|
* @param string $pathToFile |
|
236
|
|
|
* @return boolean |
|
237
|
|
|
*/ |
|
238
|
|
|
public function fileExists($pathToFile) { |
|
239
|
|
|
return file_exists ( $pathToFile ); |
|
240
|
|
|
} |
|
241
|
|
|
|
|
242
|
|
|
/** |
|
243
|
|
|
* Returns a not existing file name for pdf or image processing |
|
244
|
|
|
* protected as of testing issues with static methods. |
|
245
|
|
|
* (Actually |
|
246
|
|
|
* it will be mocked partially) FIXME: Change this behaviour as soon as the buidlNotExistingFileName function is not static anymore |
|
247
|
|
|
* @codeCoverageIgnore |
|
248
|
|
|
* |
|
249
|
|
|
* @param File $fileInfo |
|
250
|
|
|
* @return string |
|
251
|
|
|
*/ |
|
252
|
|
|
protected function buildTargetForShared(File $fileInfo) { |
|
253
|
|
|
try { |
|
254
|
|
|
$share = $this->shareMapper->find ( $fileInfo->getFileid (), $this->userId, str_replace ( 'home::', '', $fileInfo->getStoragename () ) ); |
|
255
|
|
|
|
|
256
|
|
|
// get rid of the .png or .pdf and so on |
|
257
|
|
|
$fileName = substr ( $share->getFileTarget (), 0, (strrpos ( $share->getFileTarget (), '.' )) ); // '/thedom.png' => '/thedom' || '/Test/thedom.png' => '/Test/thedom' |
|
|
|
|
|
|
258
|
|
|
|
|
259
|
|
|
// remove everything in front of and including of the first appearance of a slash from behind |
|
260
|
|
|
$fileName = substr ( strrchr ( $fileName, "/" ), 1 ); // '/thedom' => 'thedom' || '/Test/thedom' => 'thedom' |
|
|
|
|
|
|
261
|
|
|
|
|
262
|
|
|
// eliminate the file name from the path |
|
263
|
|
|
$filePath = dirname ( $share->getFileTarget () ); // '/thedom.png' => '/' || '/Test/thedom.png' => '/Test' |
|
|
|
|
|
|
264
|
|
|
|
|
265
|
|
|
// replace the first slash |
|
266
|
|
|
$pos = strpos ( $filePath, '/' ); |
|
267
|
|
|
if ($pos !== false) { |
|
268
|
|
|
$filePath = substr_replace ( $filePath, '', $pos, strlen ( '/' ) ); // '/' => '' || '/Test/' => 'Test' |
|
|
|
|
|
|
269
|
|
|
} |
|
270
|
|
|
|
|
271
|
|
View Code Duplication |
if ($fileInfo->getMimetype () === FileService::$MIMETYPE_PDF) { |
|
|
|
|
|
|
272
|
|
|
// PDFs: |
|
273
|
|
|
return \OCP\Files::buildNotExistingFileName ( $filePath, $fileName . '_OCR.pdf' ); |
|
274
|
|
|
} else { |
|
275
|
|
|
// IMAGES: |
|
276
|
|
|
return \OCP\Files::buildNotExistingFileName ( $filePath, $fileName . '_OCR.txt' ); |
|
277
|
|
|
} |
|
278
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
|
|
279
|
|
|
$this->handleException ( $e ); |
|
280
|
|
|
} |
|
281
|
|
|
} |
|
282
|
|
|
|
|
283
|
|
|
/** |
|
284
|
|
|
* Returns a not existing file name for PDF or image processing |
|
285
|
|
|
* protected as of testing issues with static methods. |
|
286
|
|
|
* (Actually |
|
287
|
|
|
* it will be mocked partially) FIXME: Change this behaviour as soon as the buidlNotExistingFileName function is not static anymore |
|
288
|
|
|
* @codeCoverageIgnore |
|
289
|
|
|
* |
|
290
|
|
|
* @param File $fileInfo |
|
291
|
|
|
* @return string |
|
292
|
|
|
*/ |
|
293
|
|
|
protected function buildTargetNotForShared(File $fileInfo) { |
|
294
|
|
|
try { |
|
295
|
|
|
// get rid of the .png or .pdf and so on |
|
296
|
|
|
$fileName = substr ( $fileInfo->getName (), 0, (strrpos ( $fileInfo->getName (), '.' )) ); // 'thedom.png' => 'thedom' |
|
|
|
|
|
|
297
|
|
|
|
|
298
|
|
|
// eliminate the file name from the path |
|
299
|
|
|
$filePath = str_replace ( $fileInfo->getName (), '', $fileInfo->getPath () ); // 'files/Test/thedom.png' => 'files/Test/' || 'files/thedom.png' => 'files/' |
|
|
|
|
|
|
300
|
|
|
|
|
301
|
|
|
// and get the path on top of the files/ dir |
|
302
|
|
|
$filePath = str_replace ( 'files', '', $filePath ); // 'files/Test/' => '/Test/' || 'files/' => '/' |
|
|
|
|
|
|
303
|
|
|
|
|
304
|
|
|
// remove the last slash |
|
305
|
|
|
$filePath = substr_replace ( $filePath, '', strrpos ( $filePath, '/' ), strlen ( '/' ) ); // '/Test/' => '/Test' || '/' => '' |
|
|
|
|
|
|
306
|
|
|
|
|
307
|
|
|
// replace the first slash |
|
308
|
|
|
$pos = strpos ( $filePath, '/' ); |
|
309
|
|
|
if ($pos !== false) { |
|
310
|
|
|
$filePath = substr_replace ( $filePath, '', $pos, strlen ( '/' ) ); // '/Test' => '// 'Test' || '/' => '' |
|
311
|
|
|
} |
|
312
|
|
|
|
|
313
|
|
View Code Duplication |
if ($fileInfo->getMimetype () === FileService::$MIMETYPE_PDF) { |
|
|
|
|
|
|
314
|
|
|
// PDFs: |
|
315
|
|
|
return \OCP\Files::buildNotExistingFileName ( $filePath, $fileName . '_OCR.pdf' ); |
|
316
|
|
|
} else { |
|
317
|
|
|
// IMAGES: |
|
318
|
|
|
return \OCP\Files::buildNotExistingFileName ( $filePath, $fileName . '_OCR.txt' ); |
|
319
|
|
|
} |
|
320
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
|
|
321
|
|
|
$this->handleException ( $e ); |
|
322
|
|
|
} |
|
323
|
|
|
} |
|
324
|
|
|
|
|
325
|
|
|
/** |
|
326
|
|
|
* Checks a MIME type for a specifically given FileInfo. |
|
327
|
|
|
* |
|
328
|
|
|
* @param File $fileInfo |
|
329
|
|
|
*/ |
|
330
|
|
|
private function checkMimeType(File $fileInfo) { |
|
331
|
|
|
try { |
|
332
|
|
|
if (! $fileInfo || ! in_array ( $fileInfo->getMimetype (), FileService::$ALLOWED_MIMETYPES )) { |
|
333
|
|
|
$this->logger->debug ( 'Getting FileInfo did not work or not included in the ALLOWED_MIMETYPES array.', [ |
|
334
|
|
|
'app' => 'ocr' |
|
335
|
|
|
] ); |
|
336
|
|
|
throw new NotFoundException ( $this->l10n->t ( 'Wrong MIME type.' ) ); |
|
337
|
|
|
} |
|
338
|
|
|
} catch ( Exception $e ) { |
|
|
|
|
|
|
339
|
|
|
$this->handleException ( $e ); |
|
340
|
|
|
} |
|
341
|
|
|
} |
|
342
|
|
|
|
|
343
|
|
|
/** |
|
344
|
|
|
* Handle the possible thrown Exceptions from all methods of this class. |
|
345
|
|
|
* |
|
346
|
|
|
* @param Exception $e |
|
347
|
|
|
* @throws Exception |
|
348
|
|
|
* @throws NotFoundException |
|
349
|
|
|
*/ |
|
350
|
|
View Code Duplication |
private function handleException($e) { |
|
|
|
|
|
|
351
|
|
|
$this->logger->logException ( $e, [ |
|
352
|
|
|
'app' => 'ocr', |
|
353
|
|
|
'message' => 'Exception during file service function processing' |
|
354
|
|
|
] ); |
|
355
|
|
|
if ($e instanceof NotFoundException) { |
|
356
|
|
|
throw new NotFoundException ( $e->getMessage () ); |
|
357
|
|
|
} else { |
|
358
|
|
|
throw $e; |
|
359
|
|
|
} |
|
360
|
|
|
} |
|
361
|
|
|
} |
This check marks private properties in classes that are never used. Those properties can be removed.