dépôts / lhc / web / wiklou.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Merge "Make DBAccessBase use DBConnRef, rename $wiki, and hide getLoadBalancer()"
[lhc/web/wiklou.git] / includes / media / FormatMetadata.php
1 <?php
2 /**
3 * Formatting of image metadata values into human readable form.
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License along
16 * with this program; if not, write to the Free Software Foundation, Inc.,
17 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
19 *
20 * @ingroup Media
21 * @author Ævar Arnfjörð Bjarmason <avarab@gmail.com>
22 * @copyright Copyright © 2005, Ævar Arnfjörð Bjarmason, 2009 Brent Garber, 2010 Brian Wolff
23 * @license GPL-2.0-or-later
24 * @see http://exif.org/Exif2-2.PDF The Exif 2.2 specification
25 * @file
26 */
27 use MediaWiki\MediaWikiServices;
28 use Wikimedia\Timestamp\TimestampException;
29
30 /**
31 * Format Image metadata values into a human readable form.
32 *
33 * Note lots of these messages use the prefix 'exif' even though
34 * they may not be exif properties. For example 'exif-ImageDescription'
35 * can be the Exif ImageDescription, or it could be the iptc-iim caption
36 * property, or it could be the xmp dc:description property. This
37 * is because these messages should be independent of how the data is
38 * stored, sine the user doesn't care if the description is stored in xmp,
39 * exif, etc only that its a description. (Additionally many of these properties
40 * are merged together following the MWG standard, such that for example,
41 * exif properties override XMP properties that mean the same thing if
42 * there is a conflict).
43 *
44 * It should perhaps use a prefix like 'metadata' instead, but there
45 * is already a large number of messages using the 'exif' prefix.
46 *
47 * @ingroup Media
48 * @since 1.23 the class extends ContextSource and various formerly-public
49 * internal methods are private
50 */
51 class FormatMetadata extends ContextSource {
52 /**
53 * Only output a single language for multi-language fields
54 * @var bool
55 * @since 1.23
56 */
57 protected $singleLang = false;
58
59 /**
60 * Trigger only outputting single language for multilanguage fields
61 *
62 * @param bool $val
63 * @since 1.23
64 */
65 public function setSingleLanguage( $val ) {
66 $this->singleLang = $val;
67 }
68
69 /**
70 * Numbers given by Exif user agents are often magical, that is they
71 * should be replaced by a detailed explanation depending on their
72 * value which most of the time are plain integers. This function
73 * formats Exif (and other metadata) values into human readable form.
74 *
75 * This is the usual entry point for this class.
76 *
77 * @param array $tags The Exif data to format ( as returned by
78 * Exif::getFilteredData() or BitmapMetadataHandler )
79 * @param bool|IContextSource $context Context to use (optional)
80 * @return array
81 */
82 public static function getFormattedData( $tags, $context = false ) {
83 $obj = new FormatMetadata;
84 if ( $context ) {
85 $obj->setContext( $context );
86 }
87
88 return $obj->makeFormattedData( $tags );
89 }
90
91 /**
92 * Numbers given by Exif user agents are often magical, that is they
93 * should be replaced by a detailed explanation depending on their
94 * value which most of the time are plain integers. This function
95 * formats Exif (and other metadata) values into human readable form.
96 *
97 * @param array $tags The Exif data to format ( as returned by
98 * Exif::getFilteredData() or BitmapMetadataHandler )
99 * @return array
100 * @since 1.23
101 * @suppress PhanTypeArraySuspiciousNullable
102 */
103 public function makeFormattedData( $tags ) {
104 $resolutionunit = !isset( $tags['ResolutionUnit'] ) || $tags['ResolutionUnit'] == 2 ? 2 : 3;
105 unset( $tags['ResolutionUnit'] );
106
107 foreach ( $tags as $tag => &$vals ) {
108 // This seems ugly to wrap non-array's in an array just to unwrap again,
109 // especially when most of the time it is not an array
110 if ( !is_array( $tags[$tag] ) ) {
111 $vals = [ $vals ];
112 }
113
114 // _type is a special value to say what array type
115 if ( isset( $tags[$tag]['_type'] ) ) {
116 $type = $tags[$tag]['_type'];
117 unset( $vals['_type'] );
118 } else {
119 $type = 'ul'; // default unordered list.
120 }
121
122 // This is done differently as the tag is an array.
123 if ( $tag == 'GPSTimeStamp' && count( $vals ) === 3 ) {
124 // hour min sec array
125
126 $h = explode( '/', $vals[0] );
127 $m = explode( '/', $vals[1] );
128 $s = explode( '/', $vals[2] );
129
130 // this should already be validated
131 // when loaded from file, but it could
132 // come from a foreign repo, so be
133 // paranoid.
134 if ( !isset( $h[1] )
135 || !isset( $m[1] )
136 || !isset( $s[1] )
137 || $h[1] == 0
138 || $m[1] == 0
139 || $s[1] == 0
140 ) {
141 continue;
142 }
143 $tags[$tag] = str_pad( intval( $h[0] / $h[1] ), 2, '0', STR_PAD_LEFT )
144 . ':' . str_pad( intval( $m[0] / $m[1] ), 2, '0', STR_PAD_LEFT )
145 . ':' . str_pad( intval( $s[0] / $s[1] ), 2, '0', STR_PAD_LEFT );
146
147 try {
148 $time = wfTimestamp( TS_MW, '1971:01:01 ' . $tags[$tag] );
149 // the 1971:01:01 is just a placeholder, and not shown to user.
150 if ( $time && intval( $time ) > 0 ) {
151 $tags[$tag] = $this->getLanguage()->time( $time );
152 }
153 } catch ( TimestampException $e ) {
154 // This shouldn't happen, but we've seen bad formats
155 // such as 4-digit seconds in the wild.
156 // leave $tags[$tag] as-is
157 }
158 continue;
159 }
160
161 // The contact info is a multi-valued field
162 // instead of the other props which are single
163 // valued (mostly) so handle as a special case.
164 if ( $tag === 'Contact' ) {
165 $vals = $this->collapseContactInfo( $vals );
166 continue;
167 }
168
169 foreach ( $vals as &$val ) {
170 switch ( $tag ) {
171 case 'Compression':
172 switch ( $val ) {
173 case 1:
174 case 2:
175 case 3:
176 case 4:
177 case 5:
178 case 6:
179 case 7:
180 case 8:
181 case 32773:
182 case 32946:
183 case 34712:
184 $val = $this->exifMsg( $tag, $val );
185 break;
186 default:
187 /* If not recognized, display as is. */
188 break;
189 }
190 break;
191
192 case 'PhotometricInterpretation':
193 switch ( $val ) {
194 case 0:
195 case 1:
196 case 2:
197 case 3:
198 case 4:
199 case 5:
200 case 6:
201 case 8:
202 case 9:
203 case 10:
204 case 32803:
205 case 34892:
206 $val = $this->exifMsg( $tag, $val );
207 break;
208 default:
209 /* If not recognized, display as is. */
210 break;
211 }
212 break;
213
214 case 'Orientation':
215 switch ( $val ) {
216 case 1:
217 case 2:
218 case 3:
219 case 4:
220 case 5:
221 case 6:
222 case 7:
223 case 8:
224 $val = $this->exifMsg( $tag, $val );
225 break;
226 default:
227 /* If not recognized, display as is. */
228 break;
229 }
230 break;
231
232 case 'PlanarConfiguration':
233 switch ( $val ) {
234 case 1:
235 case 2:
236 $val = $this->exifMsg( $tag, $val );
237 break;
238 default:
239 /* If not recognized, display as is. */
240 break;
241 }
242 break;
243
244 // TODO: YCbCrSubSampling
245 case 'YCbCrPositioning':
246 switch ( $val ) {
247 case 1:
248 case 2:
249 $val = $this->exifMsg( $tag, $val );
250 break;
251 default:
252 /* If not recognized, display as is. */
253 break;
254 }
255 break;
256
257 case 'XResolution':
258 case 'YResolution':
259 switch ( $resolutionunit ) {
260 case 2:
261 $val = $this->exifMsg( 'XYResolution', 'i', $this->formatNum( $val ) );
262 break;
263 case 3:
264 $val = $this->exifMsg( 'XYResolution', 'c', $this->formatNum( $val ) );
265 break;
266 default:
267 /* If not recognized, display as is. */
268 break;
269 }
270 break;
271
272 // TODO: YCbCrCoefficients #p27 (see annex E)
273 case 'ExifVersion':
274 case 'FlashpixVersion':
275 $val = (int)$val / 100;
276 break;
277
278 case 'ColorSpace':
279 switch ( $val ) {
280 case 1:
281 case 65535:
282 $val = $this->exifMsg( $tag, $val );
283 break;
284 default:
285 /* If not recognized, display as is. */
286 break;
287 }
288 break;
289
290 case 'ComponentsConfiguration':
291 switch ( $val ) {
292 case 0:
293 case 1:
294 case 2:
295 case 3:
296 case 4:
297 case 5:
298 case 6:
299 $val = $this->exifMsg( $tag, $val );
300 break;
301 default:
302 /* If not recognized, display as is. */
303 break;
304 }
305 break;
306
307 case 'DateTime':
308 case 'DateTimeOriginal':
309 case 'DateTimeDigitized':
310 case 'DateTimeReleased':
311 case 'DateTimeExpires':
312 case 'GPSDateStamp':
313 case 'dc-date':
314 case 'DateTimeMetadata':
315 if ( $val == '0000:00:00 00:00:00' || $val == ' : : : : ' ) {
316 $val = $this->msg( 'exif-unknowndate' )->text();
317 } elseif ( preg_match(
318 '/^(?:\d{4}):(?:\d\d):(?:\d\d) (?:\d\d):(?:\d\d):(?:\d\d)$/D',
319 $val
320 ) ) {
321 // Full date.
322 $time = wfTimestamp( TS_MW, $val );
323 if ( $time && intval( $time ) > 0 ) {
324 $val = $this->getLanguage()->timeanddate( $time );
325 }
326 } elseif ( preg_match( '/^(?:\d{4}):(?:\d\d):(?:\d\d) (?:\d\d):(?:\d\d)$/D', $val ) ) {
327 // No second field. Still format the same
328 // since timeanddate doesn't include seconds anyways,
329 // but second still available in api
330 $time = wfTimestamp( TS_MW, $val . ':00' );
331 if ( $time && intval( $time ) > 0 ) {
332 $val = $this->getLanguage()->timeanddate( $time );
333 }
334 } elseif ( preg_match( '/^(?:\d{4}):(?:\d\d):(?:\d\d)$/D', $val ) ) {
335 // If only the date but not the time is filled in.
336 $time = wfTimestamp( TS_MW, substr( $val, 0, 4 )
337 . substr( $val, 5, 2 )
338 . substr( $val, 8, 2 )
339 . '000000' );
340 if ( $time && intval( $time ) > 0 ) {
341 $val = $this->getLanguage()->date( $time );
342 }
343 }
344 // else it will just output $val without formatting it.
345 break;
346
347 case 'ExposureProgram':
348 switch ( $val ) {
349 case 0:
350 case 1:
351 case 2:
352 case 3:
353 case 4:
354 case 5:
355 case 6:
356 case 7:
357 case 8:
358 $val = $this->exifMsg( $tag, $val );
359 break;
360 default:
361 /* If not recognized, display as is. */
362 break;
363 }
364 break;
365
366 case 'SubjectDistance':
367 $val = $this->exifMsg( $tag, '', $this->formatNum( $val ) );
368 break;
369
370 case 'MeteringMode':
371 switch ( $val ) {
372 case 0:
373 case 1:
374 case 2:
375 case 3:
376 case 4:
377 case 5:
378 case 6:
379 case 7:
380 case 255:
381 $val = $this->exifMsg( $tag, $val );
382 break;
383 default:
384 /* If not recognized, display as is. */
385 break;
386 }
387 break;
388
389 case 'LightSource':
390 switch ( $val ) {
391 case 0:
392 case 1:
393 case 2:
394 case 3:
395 case 4:
396 case 9:
397 case 10:
398 case 11:
399 case 12:
400 case 13:
401 case 14:
402 case 15:
403 case 17:
404 case 18:
405 case 19:
406 case 20:
407 case 21:
408 case 22:
409 case 23:
410 case 24:
411 case 255:
412 $val = $this->exifMsg( $tag, $val );
413 break;
414 default:
415 /* If not recognized, display as is. */
416 break;
417 }
418 break;
419
420 case 'Flash':
421 $flashDecode = [
422 'fired' => $val & 0b00000001,
423 'return' => ( $val & 0b00000110 ) >> 1,
424 'mode' => ( $val & 0b00011000 ) >> 3,
425 'function' => ( $val & 0b00100000 ) >> 5,
426 'redeye' => ( $val & 0b01000000 ) >> 6,
427 // 'reserved' => ( $val & 0b10000000 ) >> 7,
428 ];
429 $flashMsgs = [];
430 # We do not need to handle unknown values since all are used.
431 foreach ( $flashDecode as $subTag => $subValue ) {
432 # We do not need any message for zeroed values.
433 if ( $subTag != 'fired' && $subValue == 0 ) {
434 continue;
435 }
436 $fullTag = $tag . '-' . $subTag;
437 $flashMsgs[] = $this->exifMsg( $fullTag, $subValue );
438 }
439 $val = $this->getLanguage()->commaList( $flashMsgs );
440 break;
441
442 case 'FocalPlaneResolutionUnit':
443 switch ( $val ) {
444 case 2:
445 $val = $this->exifMsg( $tag, $val );
446 break;
447 default:
448 /* If not recognized, display as is. */
449 break;
450 }
451 break;
452
453 case 'SensingMethod':
454 switch ( $val ) {
455 case 1:
456 case 2:
457 case 3:
458 case 4:
459 case 5:
460 case 7:
461 case 8:
462 $val = $this->exifMsg( $tag, $val );
463 break;
464 default:
465 /* If not recognized, display as is. */
466 break;
467 }
468 break;
469
470 case 'FileSource':
471 switch ( $val ) {
472 case 3:
473 $val = $this->exifMsg( $tag, $val );
474 break;
475 default:
476 /* If not recognized, display as is. */
477 break;
478 }
479 break;
480
481 case 'SceneType':
482 switch ( $val ) {
483 case 1:
484 $val = $this->exifMsg( $tag, $val );
485 break;
486 default:
487 /* If not recognized, display as is. */
488 break;
489 }
490 break;
491
492 case 'CustomRendered':
493 switch ( $val ) {
494 case 0: /* normal */
495 case 1: /* custom */
496 /* The following are unofficial Apple additions */
497 case 2: /* HDR (no original saved) */
498 case 3: /* HDR (original saved) */
499 case 4: /* Original (for HDR) */
500 /* Yes 5 is not present ;) */
501 case 6: /* Panorama */
502 case 7: /* Portrait HDR */
503 case 8: /* Portrait */
504 $val = $this->exifMsg( $tag, $val );
505 break;
506 default:
507 /* If not recognized, display as is. */
508 break;
509 }
510 break;
511
512 case 'ExposureMode':
513 switch ( $val ) {
514 case 0:
515 case 1:
516 case 2:
517 $val = $this->exifMsg( $tag, $val );
518 break;
519 default:
520 /* If not recognized, display as is. */
521 break;
522 }
523 break;
524
525 case 'WhiteBalance':
526 switch ( $val ) {
527 case 0:
528 case 1:
529 $val = $this->exifMsg( $tag, $val );
530 break;
531 default:
532 /* If not recognized, display as is. */
533 break;
534 }
535 break;
536
537 case 'SceneCaptureType':
538 switch ( $val ) {
539 case 0:
540 case 1:
541 case 2:
542 case 3:
543 $val = $this->exifMsg( $tag, $val );
544 break;
545 default:
546 /* If not recognized, display as is. */
547 break;
548 }
549 break;
550
551 case 'GainControl':
552 switch ( $val ) {
553 case 0:
554 case 1:
555 case 2:
556 case 3:
557 case 4:
558 $val = $this->exifMsg( $tag, $val );
559 break;
560 default:
561 /* If not recognized, display as is. */
562 break;
563 }
564 break;
565
566 case 'Contrast':
567 switch ( $val ) {
568 case 0:
569 case 1:
570 case 2:
571 $val = $this->exifMsg( $tag, $val );
572 break;
573 default:
574 /* If not recognized, display as is. */
575 break;
576 }
577 break;
578
579 case 'Saturation':
580 switch ( $val ) {
581 case 0:
582 case 1:
583 case 2:
584 $val = $this->exifMsg( $tag, $val );
585 break;
586 default:
587 /* If not recognized, display as is. */
588 break;
589 }
590 break;
591
592 case 'Sharpness':
593 switch ( $val ) {
594 case 0:
595 case 1:
596 case 2:
597 $val = $this->exifMsg( $tag, $val );
598 break;
599 default:
600 /* If not recognized, display as is. */
601 break;
602 }
603 break;
604
605 case 'SubjectDistanceRange':
606 switch ( $val ) {
607 case 0:
608 case 1:
609 case 2:
610 case 3:
611 $val = $this->exifMsg( $tag, $val );
612 break;
613 default:
614 /* If not recognized, display as is. */
615 break;
616 }
617 break;
618
619 // The GPS...Ref values are kept for compatibility, probably won't be reached.
620 case 'GPSLatitudeRef':
621 case 'GPSDestLatitudeRef':
622 switch ( $val ) {
623 case 'N':
624 case 'S':
625 $val = $this->exifMsg( 'GPSLatitude', $val );
626 break;
627 default:
628 /* If not recognized, display as is. */
629 break;
630 }
631 break;
632
633 case 'GPSLongitudeRef':
634 case 'GPSDestLongitudeRef':
635 switch ( $val ) {
636 case 'E':
637 case 'W':
638 $val = $this->exifMsg( 'GPSLongitude', $val );
639 break;
640 default:
641 /* If not recognized, display as is. */
642 break;
643 }
644 break;
645
646 case 'GPSAltitude':
647 if ( $val < 0 ) {
648 $val = $this->exifMsg( 'GPSAltitude', 'below-sealevel', $this->formatNum( -$val, 3 ) );
649 } else {
650 $val = $this->exifMsg( 'GPSAltitude', 'above-sealevel', $this->formatNum( $val, 3 ) );
651 }
652 break;
653
654 case 'GPSStatus':
655 switch ( $val ) {
656 case 'A':
657 case 'V':
658 $val = $this->exifMsg( $tag, $val );
659 break;
660 default:
661 /* If not recognized, display as is. */
662 break;
663 }
664 break;
665
666 case 'GPSMeasureMode':
667 switch ( $val ) {
668 case 2:
669 case 3:
670 $val = $this->exifMsg( $tag, $val );
671 break;
672 default:
673 /* If not recognized, display as is. */
674 break;
675 }
676 break;
677
678 case 'GPSTrackRef':
679 case 'GPSImgDirectionRef':
680 case 'GPSDestBearingRef':
681 switch ( $val ) {
682 case 'T':
683 case 'M':
684 $val = $this->exifMsg( 'GPSDirection', $val );
685 break;
686 default:
687 /* If not recognized, display as is. */
688 break;
689 }
690 break;
691
692 case 'GPSLatitude':
693 case 'GPSDestLatitude':
694 $val = $this->formatCoords( $val, 'latitude' );
695 break;
696 case 'GPSLongitude':
697 case 'GPSDestLongitude':
698 $val = $this->formatCoords( $val, 'longitude' );
699 break;
700
701 case 'GPSSpeedRef':
702 switch ( $val ) {
703 case 'K':
704 case 'M':
705 case 'N':
706 $val = $this->exifMsg( 'GPSSpeed', $val );
707 break;
708 default:
709 /* If not recognized, display as is. */
710 break;
711 }
712 break;
713
714 case 'GPSDestDistanceRef':
715 switch ( $val ) {
716 case 'K':
717 case 'M':
718 case 'N':
719 $val = $this->exifMsg( 'GPSDestDistance', $val );
720 break;
721 default:
722 /* If not recognized, display as is. */
723 break;
724 }
725 break;
726
727 case 'GPSDOP':
728 // See https://en.wikipedia.org/wiki/Dilution_of_precision_(GPS)
729 if ( $val <= 2 ) {
730 $val = $this->exifMsg( $tag, 'excellent', $this->formatNum( $val ) );
731 } elseif ( $val <= 5 ) {
732 $val = $this->exifMsg( $tag, 'good', $this->formatNum( $val ) );
733 } elseif ( $val <= 10 ) {
734 $val = $this->exifMsg( $tag, 'moderate', $this->formatNum( $val ) );
735 } elseif ( $val <= 20 ) {
736 $val = $this->exifMsg( $tag, 'fair', $this->formatNum( $val ) );
737 } else {
738 $val = $this->exifMsg( $tag, 'poor', $this->formatNum( $val ) );
739 }
740 break;
741
742 // This is not in the Exif standard, just a special
743 // case for our purposes which enables wikis to wikify
744 // the make, model and software name to link to their articles.
745 case 'Make':
746 case 'Model':
747 $val = $this->exifMsg( $tag, '', $val );
748 break;
749
750 case 'Software':
751 if ( is_array( $val ) ) {
752 if ( count( $val ) > 1 ) {
753 // if its a software, version array.
754 $val = $this->msg( 'exif-software-version-value', $val[0], $val[1] )->text();
755 } else {
756 // https://phabricator.wikimedia.org/T178130
757 $val = $this->exifMsg( $tag, '', $val[0] );
758 }
759 } else {
760 $val = $this->exifMsg( $tag, '', $val );
761 }
762 break;
763
764 case 'ExposureTime':
765 // Show the pretty fraction as well as decimal version
766 $val = $this->msg( 'exif-exposuretime-format',
767 $this->formatFraction( $val ), $this->formatNum( $val ) )->text();
768 break;
769 case 'ISOSpeedRatings':
770 // If its = 65535 that means its at the
771 // limit of the size of Exif::short and
772 // is really higher.
773 if ( $val == '65535' ) {
774 $val = $this->exifMsg( $tag, 'overflow' );
775 } else {
776 $val = $this->formatNum( $val );
777 }
778 break;
779 case 'FNumber':
780 $val = $this->msg( 'exif-fnumber-format',
781 $this->formatNum( $val ) )->text();
782 break;
783
784 case 'FocalLength':
785 case 'FocalLengthIn35mmFilm':
786 $val = $this->msg( 'exif-focallength-format',
787 $this->formatNum( $val ) )->text();
788 break;
789
790 case 'MaxApertureValue':
791 if ( strpos( $val, '/' ) !== false ) {
792 // need to expand this earlier to calculate fNumber
793 list( $n, $d ) = explode( '/', $val );
794 if ( is_numeric( $n ) && is_numeric( $d ) ) {
795 $val = $n / $d;
796 }
797 }
798 if ( is_numeric( $val ) ) {
799 $fNumber = 2 ** ( $val / 2 );
800 if ( $fNumber !== false ) {
801 $val = $this->msg( 'exif-maxaperturevalue-value',
802 $this->formatNum( $val ),
803 $this->formatNum( $fNumber, 2 )
804 )->text();
805 }
806 }
807 break;
808
809 case 'iimCategory':
810 switch ( strtolower( $val ) ) {
811 // See pg 29 of IPTC photo
812 // metadata standard.
813 case 'ace':
814 case 'clj':
815 case 'dis':
816 case 'fin':
817 case 'edu':
818 case 'evn':
819 case 'hth':
820 case 'hum':
821 case 'lab':
822 case 'lif':
823 case 'pol':
824 case 'rel':
825 case 'sci':
826 case 'soi':
827 case 'spo':
828 case 'war':
829 case 'wea':
830 $val = $this->exifMsg(
831 'iimcategory',
832 $val
833 );
834 }
835 break;
836 case 'SubjectNewsCode':
837 // Essentially like iimCategory.
838 // 8 (numeric) digit hierarchical
839 // classification. We decode the
840 // first 2 digits, which provide
841 // a broad category.
842 $val = $this->convertNewsCode( $val );
843 break;
844 case 'Urgency':
845 // 1-8 with 1 being highest, 5 normal
846 // 0 is reserved, and 9 is 'user-defined'.
847 $urgency = '';
848 if ( $val == 0 || $val == 9 ) {
849 $urgency = 'other';
850 } elseif ( $val < 5 && $val > 1 ) {
851 $urgency = 'high';
852 } elseif ( $val == 5 ) {
853 $urgency = 'normal';
854 } elseif ( $val <= 8 && $val > 5 ) {
855 $urgency = 'low';
856 }
857
858 if ( $urgency !== '' ) {
859 $val = $this->exifMsg( 'urgency',
860 $urgency, $val
861 );
862 }
863 break;
864
865 // Things that have a unit of pixels.
866 case 'OriginalImageHeight':
867 case 'OriginalImageWidth':
868 case 'PixelXDimension':
869 case 'PixelYDimension':
870 case 'ImageWidth':
871 case 'ImageLength':
872 $val = $this->formatNum( $val ) . ' ' . $this->msg( 'unit-pixel' )->text();
873 break;
874
875 // Do not transform fields with pure text.
876 // For some languages the formatNum()
877 // conversion results to wrong output like
878 // foo,bar@example,com or foo٫bar@example٫com.
879 // Also some 'numeric' things like Scene codes
880 // are included here as we really don't want
881 // commas inserted.
882 case 'ImageDescription':
883 case 'UserComment':
884 case 'Artist':
885 case 'Copyright':
886 case 'RelatedSoundFile':
887 case 'ImageUniqueID':
888 case 'SpectralSensitivity':
889 case 'GPSSatellites':
890 case 'GPSVersionID':
891 case 'GPSMapDatum':
892 case 'Keywords':
893 case 'WorldRegionDest':
894 case 'CountryDest':
895 case 'CountryCodeDest':
896 case 'ProvinceOrStateDest':
897 case 'CityDest':
898 case 'SublocationDest':
899 case 'WorldRegionCreated':
900 case 'CountryCreated':
901 case 'CountryCodeCreated':
902 case 'ProvinceOrStateCreated':
903 case 'CityCreated':
904 case 'SublocationCreated':
905 case 'ObjectName':
906 case 'SpecialInstructions':
907 case 'Headline':
908 case 'Credit':
909 case 'Source':
910 case 'EditStatus':
911 case 'FixtureIdentifier':
912 case 'LocationDest':
913 case 'LocationDestCode':
914 case 'Writer':
915 case 'JPEGFileComment':
916 case 'iimSupplementalCategory':
917 case 'OriginalTransmissionRef':
918 case 'Identifier':
919 case 'dc-contributor':
920 case 'dc-coverage':
921 case 'dc-publisher':
922 case 'dc-relation':
923 case 'dc-rights':
924 case 'dc-source':
925 case 'dc-type':
926 case 'Lens':
927 case 'SerialNumber':
928 case 'CameraOwnerName':
929 case 'Label':
930 case 'Nickname':
931 case 'RightsCertificate':
932 case 'CopyrightOwner':
933 case 'UsageTerms':
934 case 'WebStatement':
935 case 'OriginalDocumentID':
936 case 'LicenseUrl':
937 case 'MorePermissionsUrl':
938 case 'AttributionUrl':
939 case 'PreferredAttributionName':
940 case 'PNGFileComment':
941 case 'Disclaimer':
942 case 'ContentWarning':
943 case 'GIFFileComment':
944 case 'SceneCode':
945 case 'IntellectualGenre':
946 case 'Event':
947 case 'OrginisationInImage':
948 case 'PersonInImage':
949
950 $val = htmlspecialchars( $val );
951 break;
952
953 case 'ObjectCycle':
954 switch ( $val ) {
955 case 'a':
956 case 'p':
957 case 'b':
958 $val = $this->exifMsg( $tag, $val );
959 break;
960 default:
961 $val = htmlspecialchars( $val );
962 break;
963 }
964 break;
965 case 'Copyrighted':
966 switch ( $val ) {
967 case 'True':
968 case 'False':
969 $val = $this->exifMsg( $tag, $val );
970 break;
971 }
972 break;
973 case 'Rating':
974 if ( $val == '-1' ) {
975 $val = $this->exifMsg( $tag, 'rejected' );
976 } else {
977 $val = $this->formatNum( $val );
978 }
979 break;
980
981 case 'LanguageCode':
982 $lang = Language::fetchLanguageName( strtolower( $val ), $this->getLanguage()->getCode() );
983 $val = htmlspecialchars( $lang ?: $val );
984 break;
985
986 default:
987 $val = $this->formatNum( $val );
988 break;
989 }
990 }
991 // End formatting values, start flattening arrays.
992 $vals = $this->flattenArrayReal( $vals, $type );
993 }
994
995 return $tags;
996 }
997
998 /**
999 * Flatten an array, using the content language for any messages.
1000 *
1001 * @param array $vals Array of values
1002 * @param string $type Type of array (either lang, ul, ol).
1003 * lang = language assoc array with keys being the lang code
1004 * ul = unordered list, ol = ordered list
1005 * type can also come from the '_type' member of $vals.
1006 * @param bool $noHtml If to avoid returning anything resembling HTML.
1007 * (Ugly hack for backwards compatibility with old MediaWiki).
1008 * @param bool|IContextSource $context
1009 * @return string Single value (in wiki-syntax).
1010 * @since 1.23
1011 */
1012 public static function flattenArrayContentLang( $vals, $type = 'ul',
1013 $noHtml = false, $context = false
1014 ) {
1015 $obj = new FormatMetadata;
1016 if ( $context ) {
1017 $obj->setContext( $context );
1018 }
1019 $context = new DerivativeContext( $obj->getContext() );
1020 $context->setLanguage( MediaWikiServices::getInstance()->getContentLanguage() );
1021 $obj->setContext( $context );
1022
1023 return $obj->flattenArrayReal( $vals, $type, $noHtml );
1024 }
1025
1026 /**
1027 * A function to collapse multivalued tags into a single value.
1028 * This turns an array of (for example) authors into a bulleted list.
1029 *
1030 * This is public on the basis it might be useful outside of this class.
1031 *
1032 * @param array $vals Array of values
1033 * @param string $type Type of array (either lang, ul, ol).
1034 * lang = language assoc array with keys being the lang code
1035 * ul = unordered list, ol = ordered list
1036 * type can also come from the '_type' member of $vals.
1037 * @param bool $noHtml If to avoid returning anything resembling HTML.
1038 * (Ugly hack for backwards compatibility with old mediawiki).
1039 * @return string Single value (in wiki-syntax).
1040 * @since 1.23
1041 */
1042 public function flattenArrayReal( $vals, $type = 'ul', $noHtml = false ) {
1043 if ( !is_array( $vals ) ) {
1044 return $vals; // do nothing if not an array;
1045 }
1046
1047 if ( isset( $vals['_type'] ) ) {
1048 $type = $vals['_type'];
1049 unset( $vals['_type'] );
1050 }
1051
1052 if ( !is_array( $vals ) ) {
1053 return $vals; // do nothing if not an array;
1054 } elseif ( count( $vals ) === 1 && $type !== 'lang' && isset( $vals[0] ) ) {
1055 return $vals[0];
1056 } elseif ( count( $vals ) === 0 ) {
1057 wfDebug( __METHOD__ . " metadata array with 0 elements!\n" );
1058
1059 return ""; // paranoia. This should never happen
1060 } else {
1061 /* @todo FIXME: This should hide some of the list entries if there are
1062 * say more than four. Especially if a field is translated into 20
1063 * languages, we don't want to show them all by default
1064 */
1065 switch ( $type ) {
1066 case 'lang':
1067 // Display default, followed by ContentLanguage,
1068 // followed by the rest in no particular
1069 // order.
1070
1071 // Todo: hide some items if really long list.
1072
1073 $content = '';
1074
1075 $priorityLanguages = $this->getPriorityLanguages();
1076 $defaultItem = false;
1077 $defaultLang = false;
1078
1079 // If default is set, save it for later,
1080 // as we don't know if it's equal to
1081 // one of the lang codes. (In xmp
1082 // you specify the language for a
1083 // default property by having both
1084 // a default prop, and one in the language
1085 // that are identical)
1086 if ( isset( $vals['x-default'] ) ) {
1087 $defaultItem = $vals['x-default'];
1088 unset( $vals['x-default'] );
1089 }
1090 foreach ( $priorityLanguages as $pLang ) {
1091 if ( isset( $vals[$pLang] ) ) {
1092 $isDefault = false;
1093 if ( $vals[$pLang] === $defaultItem ) {
1094 $defaultItem = false;
1095 $isDefault = true;
1096 }
1097 $content .= $this->langItem(
1098 $vals[$pLang], $pLang,
1099 $isDefault, $noHtml );
1100
1101 unset( $vals[$pLang] );
1102
1103 if ( $this->singleLang ) {
1104 return Html::rawElement( 'span',
1105 [ 'lang' => $pLang ], $vals[$pLang] );
1106 }
1107 }
1108 }
1109
1110 // Now do the rest.
1111 foreach ( $vals as $lang => $item ) {
1112 if ( $item === $defaultItem ) {
1113 $defaultLang = $lang;
1114 continue;
1115 }
1116 $content .= $this->langItem( $item,
1117 $lang, false, $noHtml );
1118 if ( $this->singleLang ) {
1119 return Html::rawElement( 'span',
1120 [ 'lang' => $lang ], $item );
1121 }
1122 }
1123 if ( $defaultItem !== false ) {
1124 $content = $this->langItem( $defaultItem,
1125 $defaultLang, true, $noHtml ) .
1126 $content;
1127 if ( $this->singleLang ) {
1128 return $defaultItem;
1129 }
1130 }
1131 if ( $noHtml ) {
1132 return $content;
1133 }
1134
1135 return '<ul class="metadata-langlist">' .
1136 $content .
1137 '</ul>';
1138 case 'ol':
1139 if ( $noHtml ) {
1140 return "\n#" . implode( "\n#", $vals );
1141 }
1142
1143 return "<ol><li>" . implode( "</li>\n<li>", $vals ) . '</li></ol>';
1144 case 'ul':
1145 default:
1146 if ( $noHtml ) {
1147 return "\n*" . implode( "\n*", $vals );
1148 }
1149
1150 return "<ul><li>" . implode( "</li>\n<li>", $vals ) . '</li></ul>';
1151 }
1152 }
1153 }
1154
1155 /** Helper function for creating lists of translations.
1156 *
1157 * @param string $value Value (this is not escaped)
1158 * @param string $lang Lang code of item or false
1159 * @param bool $default If it is default value.
1160 * @param bool $noHtml If to avoid html (for back-compat)
1161 * @throws MWException
1162 * @return string Language item (Note: despite how this looks, this is
1163 * treated as wikitext, not as HTML).
1164 */
1165 private function langItem( $value, $lang, $default = false, $noHtml = false ) {
1166 if ( $lang === false && $default === false ) {
1167 throw new MWException( '$lang and $default cannot both '
1168 . 'be false.' );
1169 }
1170
1171 if ( $noHtml ) {
1172 $wrappedValue = $value;
1173 } else {
1174 $wrappedValue = '<span class="mw-metadata-lang-value">'
1175 . $value . '</span>';
1176 }
1177
1178 if ( $lang === false ) {
1179 $msg = $this->msg( 'metadata-langitem-default', $wrappedValue );
1180 if ( $noHtml ) {
1181 return $msg->text() . "\n\n";
1182 } /* else */
1183
1184 return '<li class="mw-metadata-lang-default">'
1185 . $msg->text()
1186 . "</li>\n";
1187 }
1188
1189 $lowLang = strtolower( $lang );
1190 $langName = Language::fetchLanguageName( $lowLang );
1191 if ( $langName === '' ) {
1192 // try just the base language name. (aka en-US -> en ).
1193 $langPrefix = explode( '-', $lowLang, 2 )[0];
1194 $langName = Language::fetchLanguageName( $langPrefix );
1195 if ( $langName === '' ) {
1196 // give up.
1197 $langName = $lang;
1198 }
1199 }
1200 // else we have a language specified
1201
1202 $msg = $this->msg( 'metadata-langitem', $wrappedValue, $langName, $lang );
1203 if ( $noHtml ) {
1204 return '*' . $msg->text();
1205 } /* else: */
1206
1207 $item = '<li class="mw-metadata-lang-code-'
1208 . $lang;
1209 if ( $default ) {
1210 $item .= ' mw-metadata-lang-default';
1211 }
1212 $item .= '" lang="' . $lang . '">';
1213 $item .= $msg->text();
1214 $item .= "</li>\n";
1215
1216 return $item;
1217 }
1218
1219 /**
1220 * Convenience function for getFormattedData()
1221 *
1222 * @param string $tag The tag name to pass on
1223 * @param string $val The value of the tag
1224 * @param string $arg An argument to pass ($1)
1225 * @param string $arg2 A 2nd argument to pass ($2)
1226 * @return string The text content of "exif-$tag-$val" message in lower case
1227 */
1228 private function exifMsg( $tag, $val, $arg = null, $arg2 = null ) {
1229 if ( $val === '' ) {
1230 $val = 'value';
1231 }
1232
1233 return $this->msg(
1234 MediaWikiServices::getInstance()->getContentLanguage()->lc( "exif-$tag-$val" ),
1235 $arg,
1236 $arg2
1237 )->text();
1238 }
1239
1240 /**
1241 * Format a number, convert numbers from fractions into floating point
1242 * numbers, joins arrays of numbers with commas.
1243 *
1244 * @param mixed $num The value to format
1245 * @param float|int|bool $round Digits to round to or false.
1246 * @return mixed A floating point number or whatever we were fed
1247 */
1248 private function formatNum( $num, $round = false ) {
1249 $m = [];
1250 if ( is_array( $num ) ) {
1251 $out = [];
1252 foreach ( $num as $number ) {
1253 $out[] = $this->formatNum( $number );
1254 }
1255
1256 return $this->getLanguage()->commaList( $out );
1257 }
1258 if ( preg_match( '/^(-?\d+)\/(\d+)$/', $num, $m ) ) {
1259 if ( $m[2] != 0 ) {
1260 $newNum = $m[1] / $m[2];
1261 if ( $round !== false ) {
1262 $newNum = round( $newNum, $round );
1263 }
1264 } else {
1265 $newNum = $num;
1266 }
1267
1268 return $this->getLanguage()->formatNum( $newNum );
1269 } else {
1270 if ( is_numeric( $num ) && $round !== false ) {
1271 $num = round( $num, $round );
1272 }
1273
1274 return $this->getLanguage()->formatNum( $num );
1275 }
1276 }
1277
1278 /**
1279 * Format a rational number, reducing fractions
1280 *
1281 * @param mixed $num The value to format
1282 * @return mixed A floating point number or whatever we were fed
1283 */
1284 private function formatFraction( $num ) {
1285 $m = [];
1286 if ( preg_match( '/^(-?\d+)\/(\d+)$/', $num, $m ) ) {
1287 $numerator = intval( $m[1] );
1288 $denominator = intval( $m[2] );
1289 $gcd = $this->gcd( abs( $numerator ), $denominator );
1290 if ( $gcd != 0 ) {
1291 // 0 shouldn't happen! ;)
1292 return $this->formatNum( $numerator / $gcd ) . '/' . $this->formatNum( $denominator / $gcd );
1293 }
1294 }
1295
1296 return $this->formatNum( $num );
1297 }
1298
1299 /**
1300 * Calculate the greatest common divisor of two integers.
1301 *
1302 * @param int $a Numerator
1303 * @param int $b Denominator
1304 * @return int
1305 */
1306 private function gcd( $a, $b ) {
1307 /*
1308 // https://en.wikipedia.org/wiki/Euclidean_algorithm
1309 // Recursive form would be:
1310 if( $b == 0 )
1311 return $a;
1312 else
1313 return gcd( $b, $a % $b );
1314 */
1315 while ( $b != 0 ) {
1316 $remainder = $a % $b;
1317
1318 // tail recursion...
1319 $a = $b;
1320 $b = $remainder;
1321 }
1322
1323 return $a;
1324 }
1325
1326 /**
1327 * Fetch the human readable version of a news code.
1328 * A news code is an 8 digit code. The first two
1329 * digits are a general classification, so we just
1330 * translate that.
1331 *
1332 * Note, leading 0's are significant, so this is
1333 * a string, not an int.
1334 *
1335 * @param string $val The 8 digit news code.
1336 * @return string The human readable form
1337 */
1338 private function convertNewsCode( $val ) {
1339 if ( !preg_match( '/^\d{8}$/D', $val ) ) {
1340 // Not a valid news code.
1341 return $val;
1342 }
1343 $cat = '';
1344 switch ( substr( $val, 0, 2 ) ) {
1345 case '01':
1346 $cat = 'ace';
1347 break;
1348 case '02':
1349 $cat = 'clj';
1350 break;
1351 case '03':
1352 $cat = 'dis';
1353 break;
1354 case '04':
1355 $cat = 'fin';
1356 break;
1357 case '05':
1358 $cat = 'edu';
1359 break;
1360 case '06':
1361 $cat = 'evn';
1362 break;
1363 case '07':
1364 $cat = 'hth';
1365 break;
1366 case '08':
1367 $cat = 'hum';
1368 break;
1369 case '09':
1370 $cat = 'lab';
1371 break;
1372 case '10':
1373 $cat = 'lif';
1374 break;
1375 case '11':
1376 $cat = 'pol';
1377 break;
1378 case '12':
1379 $cat = 'rel';
1380 break;
1381 case '13':
1382 $cat = 'sci';
1383 break;
1384 case '14':
1385 $cat = 'soi';
1386 break;
1387 case '15':
1388 $cat = 'spo';
1389 break;
1390 case '16':
1391 $cat = 'war';
1392 break;
1393 case '17':
1394 $cat = 'wea';
1395 break;
1396 }
1397 if ( $cat !== '' ) {
1398 $catMsg = $this->exifMsg( 'iimcategory', $cat );
1399 $val = $this->exifMsg( 'subjectnewscode', '', $val, $catMsg );
1400 }
1401
1402 return $val;
1403 }
1404
1405 /**
1406 * Format a coordinate value, convert numbers from floating point
1407 * into degree minute second representation.
1408 *
1409 * @param int $coord Degrees, minutes and seconds
1410 * @param string $type Latitude or longitude (for if its a NWS or E)
1411 * @return mixed A floating point number or whatever we were fed
1412 */
1413 private function formatCoords( $coord, $type ) {
1414 $ref = '';
1415 if ( $coord < 0 ) {
1416 $nCoord = -$coord;
1417 if ( $type === 'latitude' ) {
1418 $ref = 'S';
1419 } elseif ( $type === 'longitude' ) {
1420 $ref = 'W';
1421 }
1422 } else {
1423 $nCoord = $coord;
1424 if ( $type === 'latitude' ) {
1425 $ref = 'N';
1426 } elseif ( $type === 'longitude' ) {
1427 $ref = 'E';
1428 }
1429 }
1430
1431 $deg = floor( $nCoord );
1432 $min = floor( ( $nCoord - $deg ) * 60.0 );
1433 $sec = round( ( ( $nCoord - $deg ) - $min / 60 ) * 3600, 2 );
1434
1435 $deg = $this->formatNum( $deg );
1436 $min = $this->formatNum( $min );
1437 $sec = $this->formatNum( $sec );
1438
1439 return $this->msg( 'exif-coordinate-format', $deg, $min, $sec, $ref, $coord )->text();
1440 }
1441
1442 /**
1443 * Format the contact info field into a single value.
1444 *
1445 * This function might be called from
1446 * JpegHandler::convertMetadataVersion which is why it is
1447 * public.
1448 *
1449 * @param array $vals Array with fields of the ContactInfo
1450 * struct defined in the IPTC4XMP spec. Or potentially
1451 * an array with one element that is a free form text
1452 * value from the older iptc iim 1:118 prop.
1453 * @return string HTML-ish looking wikitext
1454 * @since 1.23 no longer static
1455 */
1456 public function collapseContactInfo( $vals ) {
1457 if ( !( isset( $vals['CiAdrExtadr'] )
1458 || isset( $vals['CiAdrCity'] )
1459 || isset( $vals['CiAdrCtry'] )
1460 || isset( $vals['CiEmailWork'] )
1461 || isset( $vals['CiTelWork'] )
1462 || isset( $vals['CiAdrPcode'] )
1463 || isset( $vals['CiAdrRegion'] )
1464 || isset( $vals['CiUrlWork'] )
1465 ) ) {
1466 // We don't have any sub-properties
1467 // This could happen if its using old
1468 // iptc that just had this as a free-form
1469 // text value.
1470 // Note: We run this through htmlspecialchars
1471 // partially to be consistent, and partially
1472 // because people often insert >, etc into
1473 // the metadata which should not be interpreted
1474 // but we still want to auto-link urls.
1475 foreach ( $vals as &$val ) {
1476 $val = htmlspecialchars( $val );
1477 }
1478
1479 return $this->flattenArrayReal( $vals );
1480 } else {
1481 // We have a real ContactInfo field.
1482 // Its unclear if all these fields have to be
1483 // set, so assume they do not.
1484 $url = $tel = $street = $city = $country = '';
1485 $email = $postal = $region = '';
1486
1487 // Also note, some of the class names this uses
1488 // are similar to those used by hCard. This is
1489 // mostly because they're sensible names. This
1490 // does not (and does not attempt to) output
1491 // stuff in the hCard microformat. However it
1492 // might output in the adr microformat.
1493
1494 if ( isset( $vals['CiAdrExtadr'] ) ) {
1495 // Todo: This can potentially be multi-line.
1496 // Need to check how that works in XMP.
1497 $street = '<span class="extended-address">'
1498 . htmlspecialchars(
1499 $vals['CiAdrExtadr'] )
1500 . '</span>';
1501 }
1502 if ( isset( $vals['CiAdrCity'] ) ) {
1503 $city = '<span class="locality">'
1504 . htmlspecialchars( $vals['CiAdrCity'] )
1505 . '</span>';
1506 }
1507 if ( isset( $vals['CiAdrCtry'] ) ) {
1508 $country = '<span class="country-name">'
1509 . htmlspecialchars( $vals['CiAdrCtry'] )
1510 . '</span>';
1511 }
1512 if ( isset( $vals['CiEmailWork'] ) ) {
1513 $emails = [];
1514 // Have to split multiple emails at commas/new lines.
1515 $splitEmails = explode( "\n", $vals['CiEmailWork'] );
1516 foreach ( $splitEmails as $e1 ) {
1517 // Also split on comma
1518 foreach ( explode( ',', $e1 ) as $e2 ) {
1519 $finalEmail = trim( $e2 );
1520 if ( $finalEmail == ',' || $finalEmail == '' ) {
1521 continue;
1522 }
1523 if ( strpos( $finalEmail, '<' ) !== false ) {
1524 // Don't do fancy formatting to
1525 // "My name" <foo@bar.com> style stuff
1526 $emails[] = $finalEmail;
1527 } else {
1528 $emails[] = '[mailto:'
1529 . $finalEmail
1530 . ' <span class="email">'
1531 . $finalEmail
1532 . '</span>]';
1533 }
1534 }
1535 }
1536 $email = implode( ', ', $emails );
1537 }
1538 if ( isset( $vals['CiTelWork'] ) ) {
1539 $tel = '<span class="tel">'
1540 . htmlspecialchars( $vals['CiTelWork'] )
1541 . '</span>';
1542 }
1543 if ( isset( $vals['CiAdrPcode'] ) ) {
1544 $postal = '<span class="postal-code">'
1545 . htmlspecialchars(
1546 $vals['CiAdrPcode'] )
1547 . '</span>';
1548 }
1549 if ( isset( $vals['CiAdrRegion'] ) ) {
1550 // Note this is province/state.
1551 $region = '<span class="region">'
1552 . htmlspecialchars(
1553 $vals['CiAdrRegion'] )
1554 . '</span>';
1555 }
1556 if ( isset( $vals['CiUrlWork'] ) ) {
1557 $url = '<span class="url">'
1558 . htmlspecialchars( $vals['CiUrlWork'] )
1559 . '</span>';
1560 }
1561
1562 return $this->msg( 'exif-contact-value', $email, $url,
1563 $street, $city, $region, $postal, $country,
1564 $tel )->text();
1565 }
1566 }
1567
1568 /**
1569 * Get a list of fields that are visible by default.
1570 *
1571 * @return array
1572 * @since 1.23
1573 */
1574 public static function getVisibleFields() {
1575 $fields = [];
1576 $lines = explode( "\n", wfMessage( 'metadata-fields' )->inContentLanguage()->text() );
1577 foreach ( $lines as $line ) {
1578 $matches = [];
1579 if ( preg_match( '/^\\*\s*(.*?)\s*$/', $line, $matches ) ) {
1580 $fields[] = $matches[1];
1581 }
1582 }
1583 $fields = array_map( 'strtolower', $fields );
1584
1585 return $fields;
1586 }
1587
1588 /**
1589 * Get an array of extended metadata. (See the imageinfo API for format.)
1590 *
1591 * @param File $file File to use
1592 * @return array [<property name> => ['value' => <value>]], or [] on error
1593 * @since 1.23
1594 */
1595 public function fetchExtendedMetadata( File $file ) {
1596 $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
1597
1598 // If revision deleted, exit immediately
1599 if ( $file->isDeleted( File::DELETED_FILE ) ) {
1600 return [];
1601 }
1602
1603 $cacheKey = $cache->makeKey(
1604 'getExtendedMetadata',
1605 $this->getLanguage()->getCode(),
1606 (int)$this->singleLang,
1607 $file->getSha1()
1608 );
1609
1610 $cachedValue = $cache->get( $cacheKey );
1611 if (
1612 $cachedValue
1613 && Hooks::run( 'ValidateExtendedMetadataCache', [ $cachedValue['timestamp'], $file ] )
1614 ) {
1615 $extendedMetadata = $cachedValue['data'];
1616 } else {
1617 $maxCacheTime = ( $file instanceof ForeignAPIFile ) ? 60 * 60 * 12 : 60 * 60 * 24 * 30;
1618 $fileMetadata = $this->getExtendedMetadataFromFile( $file );
1619 $extendedMetadata = $this->getExtendedMetadataFromHook( $file, $fileMetadata, $maxCacheTime );
1620 if ( $this->singleLang ) {
1621 $this->resolveMultilangMetadata( $extendedMetadata );
1622 }
1623 $this->discardMultipleValues( $extendedMetadata );
1624 // Make sure the metadata won't break the API when an XML format is used.
1625 // This is an API-specific function so it would be cleaner to call it from
1626 // outside fetchExtendedMetadata, but this way we don't need to redo the
1627 // computation on a cache hit.
1628 $this->sanitizeArrayForAPI( $extendedMetadata );
1629 $valueToCache = [ 'data' => $extendedMetadata, 'timestamp' => wfTimestampNow() ];
1630 $cache->set( $cacheKey, $valueToCache, $maxCacheTime );
1631 }
1632
1633 return $extendedMetadata;
1634 }
1635
1636 /**
1637 * Get file-based metadata in standardized format.
1638 *
1639 * Note that for a remote file, this might return metadata supplied by extensions.
1640 *
1641 * @param File $file File to use
1642 * @return array [<property name> => ['value' => <value>]], or [] on error
1643 * @since 1.23
1644 */
1645 protected function getExtendedMetadataFromFile( File $file ) {
1646 // If this is a remote file accessed via an API request, we already
1647 // have remote metadata so we just ignore any local one
1648 if ( $file instanceof ForeignAPIFile ) {
1649 // In case of error we pretend no metadata - this will get cached.
1650 // Might or might not be a good idea.
1651 return $file->getExtendedMetadata() ?: [];
1652 }
1653
1654 $uploadDate = wfTimestamp( TS_ISO_8601, $file->getTimestamp() );
1655
1656 $fileMetadata = [
1657 // This is modification time, which is close to "upload" time.
1658 'DateTime' => [
1659 'value' => $uploadDate,
1660 'source' => 'mediawiki-metadata',
1661 ],
1662 ];
1663
1664 $title = $file->getTitle();
1665 if ( $title ) {
1666 $text = $title->getText();
1667 $pos = strrpos( $text, '.' );
1668
1669 if ( $pos ) {
1670 $name = substr( $text, 0, $pos );
1671 } else {
1672 $name = $text;
1673 }
1674
1675 $fileMetadata['ObjectName'] = [
1676 'value' => $name,
1677 'source' => 'mediawiki-metadata',
1678 ];
1679 }
1680
1681 return $fileMetadata;
1682 }
1683
1684 /**
1685 * Get additional metadata from hooks in standardized format.
1686 *
1687 * @param File $file File to use
1688 * @param array $extendedMetadata
1689 * @param int &$maxCacheTime Hook handlers might use this parameter to override cache time
1690 *
1691 * @return array [<property name> => ['value' => <value>]], or [] on error
1692 * @since 1.23
1693 */
1694 protected function getExtendedMetadataFromHook( File $file, array $extendedMetadata,
1695 &$maxCacheTime
1696 ) {
1697 Hooks::run( 'GetExtendedMetadata', [
1698 &$extendedMetadata,
1699 $file,
1700 $this->getContext(),
1701 $this->singleLang,
1702 &$maxCacheTime
1703 ] );
1704
1705 $visible = array_flip( self::getVisibleFields() );
1706 foreach ( $extendedMetadata as $key => $value ) {
1707 if ( !isset( $visible[strtolower( $key )] ) ) {
1708 $extendedMetadata[$key]['hidden'] = '';
1709 }
1710 }
1711
1712 return $extendedMetadata;
1713 }
1714
1715 /**
1716 * Turns an XMP-style multilang array into a single value.
1717 * If the value is not a multilang array, it is returned unchanged.
1718 * See mediawiki.org/wiki/Manual:File_metadata_handling#Multi-language_array_format
1719 * @param mixed $value
1720 * @return mixed Value in best language, null if there were no languages at all
1721 * @since 1.23
1722 */
1723 protected function resolveMultilangValue( $value ) {
1724 if (
1725 !is_array( $value )
1726 || !isset( $value['_type'] )
1727 || $value['_type'] != 'lang'
1728 ) {
1729 return $value; // do nothing if not a multilang array
1730 }
1731
1732 // choose the language best matching user or site settings
1733 $priorityLanguages = $this->getPriorityLanguages();
1734 foreach ( $priorityLanguages as $lang ) {
1735 if ( isset( $value[$lang] ) ) {
1736 return $value[$lang];
1737 }
1738 }
1739
1740 // otherwise go with the default language, if set
1741 if ( isset( $value['x-default'] ) ) {
1742 return $value['x-default'];
1743 }
1744
1745 // otherwise just return any one language
1746 unset( $value['_type'] );
1747 if ( !empty( $value ) ) {
1748 return reset( $value );
1749 }
1750
1751 // this should not happen; signal error
1752 return null;
1753 }
1754
1755 /**
1756 * Turns an XMP-style multivalue array into a single value by dropping all but the first
1757 * value. If the value is not a multivalue array (or a multivalue array inside a multilang
1758 * array), it is returned unchanged.
1759 * See mediawiki.org/wiki/Manual:File_metadata_handling#Multi-language_array_format
1760 * @param mixed $value
1761 * @return mixed The value, or the first value if there were multiple ones
1762 * @since 1.25
1763 */
1764 protected function resolveMultivalueValue( $value ) {
1765 if ( !is_array( $value ) ) {
1766 return $value;
1767 } elseif ( isset( $value['_type'] ) && $value['_type'] === 'lang' ) {
1768 // if this is a multilang array, process fields separately
1769 $newValue = [];
1770 foreach ( $value as $k => $v ) {
1771 $newValue[$k] = $this->resolveMultivalueValue( $v );
1772 }
1773 return $newValue;
1774 } else { // _type is 'ul' or 'ol' or missing in which case it defaults to 'ul'
1775 $v = reset( $value );
1776 if ( key( $value ) === '_type' ) {
1777 $v = next( $value );
1778 }
1779 return $v;
1780 }
1781 }
1782
1783 /**
1784 * Takes an array returned by the getExtendedMetadata* functions,
1785 * and resolves multi-language values in it.
1786 * @param array &$metadata
1787 * @since 1.23
1788 */
1789 protected function resolveMultilangMetadata( &$metadata ) {
1790 if ( !is_array( $metadata ) ) {
1791 return;
1792 }
1793 foreach ( $metadata as &$field ) {
1794 if ( isset( $field['value'] ) ) {
1795 $field['value'] = $this->resolveMultilangValue( $field['value'] );
1796 }
1797 }
1798 }
1799
1800 /**
1801 * Takes an array returned by the getExtendedMetadata* functions,
1802 * and turns all fields into single-valued ones by dropping extra values.
1803 * @param array &$metadata
1804 * @since 1.25
1805 */
1806 protected function discardMultipleValues( &$metadata ) {
1807 if ( !is_array( $metadata ) ) {
1808 return;
1809 }
1810 foreach ( $metadata as $key => &$field ) {
1811 if ( $key === 'Software' || $key === 'Contact' ) {
1812 // we skip some fields which have composite values. They are not particularly interesting
1813 // and you can get them via the metadata / commonmetadata APIs anyway.
1814 continue;
1815 }
1816 if ( isset( $field['value'] ) ) {
1817 $field['value'] = $this->resolveMultivalueValue( $field['value'] );
1818 }
1819 }
1820 }
1821
1822 /**
1823 * Makes sure the given array is a valid API response fragment
1824 * @param array &$arr
1825 */
1826 protected function sanitizeArrayForAPI( &$arr ) {
1827 if ( !is_array( $arr ) ) {
1828 return;
1829 }
1830
1831 $counter = 1;
1832 foreach ( $arr as $key => &$value ) {
1833 $sanitizedKey = $this->sanitizeKeyForAPI( $key );
1834 if ( $sanitizedKey !== $key ) {
1835 if ( isset( $arr[$sanitizedKey] ) ) {
1836 // Make the sanitized keys hopefully unique.
1837 // To make it definitely unique would be too much effort, given that
1838 // sanitizing is only needed for misformatted metadata anyway, but
1839 // this at least covers the case when $arr is numeric.
1840 $sanitizedKey .= $counter;
1841 ++$counter;
1842 }
1843 $arr[$sanitizedKey] = $arr[$key];
1844 unset( $arr[$key] );
1845 }
1846 if ( is_array( $value ) ) {
1847 $this->sanitizeArrayForAPI( $value );
1848 }
1849 }
1850
1851 // Handle API metadata keys (particularly "_type")
1852 $keys = array_filter( array_keys( $arr ), 'ApiResult::isMetadataKey' );
1853 if ( $keys ) {
1854 ApiResult::setPreserveKeysList( $arr, $keys );
1855 }
1856 }
1857
1858 /**
1859 * Turns a string into a valid API identifier.
1860 * @param string $key
1861 * @return string
1862 * @since 1.23
1863 */
1864 protected function sanitizeKeyForAPI( $key ) {
1865 // drop all characters which are not valid in an XML tag name
1866 // a bunch of non-ASCII letters would be valid but probably won't
1867 // be used so we take the easy way
1868 $key = preg_replace( '/[^a-zA-z0-9_:.\-]/', '', $key );
1869 // drop characters which are invalid at the first position
1870 $key = preg_replace( '/^[\d\-.]+/', '', $key );
1871
1872 if ( $key == '' ) {
1873 $key = '_';
1874 }
1875
1876 // special case for an internal keyword
1877 if ( $key == '_element' ) {
1878 $key = 'element';
1879 }
1880
1881 return $key;
1882 }
1883
1884 /**
1885 * Returns a list of languages (first is best) to use when formatting multilang fields,
1886 * based on user and site preferences.
1887 * @return array
1888 * @since 1.23
1889 */
1890 protected function getPriorityLanguages() {
1891 $priorityLanguages =
1892 Language::getFallbacksIncludingSiteLanguage( $this->getLanguage()->getCode() );
1893 $priorityLanguages = array_merge(
1894 (array)$this->getLanguage()->getCode(),
1895 $priorityLanguages[0],
1896 $priorityLanguages[1]
1897 );
1898
1899 return $priorityLanguages;
1900 }
1901 }
Wiklou, le Wiki du Wiklou