?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
SpecialLinkSearch: clean up munged query variable handling
[mediawiki.git] / includes / media / FormatMetadata.php
blob7246072ded2b6466ed1d8e59960c0e1967b2a289
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 http://www.gnu.org/copyleft/gpl.html GNU General Public License
24 * @see http://exif.org/Exif2-2.PDF The Exif 2.2 specification
25 * @file
26 */
27
28 /**
29 * Format Image metadata values into a human readable form.
30 *
31 * Note lots of these messages use the prefix 'exif' even though
32 * they may not be exif properties. For example 'exif-ImageDescription'
33 * can be the Exif ImageDescription, or it could be the iptc-iim caption
34 * property, or it could be the xmp dc:description property. This
35 * is because these messages should be independent of how the data is
36 * stored, sine the user doesn't care if the description is stored in xmp,
37 * exif, etc only that its a description. (Additionally many of these properties
38 * are merged together following the MWG standard, such that for example,
39 * exif properties override XMP properties that mean the same thing if
40 * there is a conflict).
41 *
42 * It should perhaps use a prefix like 'metadata' instead, but there
43 * is already a large number of messages using the 'exif' prefix.
44 *
45 * @ingroup Media
46 * @since 1.23 the class extends ContextSource and various formerly-public
47 * internal methods are private
48 */
49 class FormatMetadata extends ContextSource {
50 /**
51 * Only output a single language for multi-language fields
52 * @var bool
53 * @since 1.23
54 */
55 protected $singleLang = false;
56
57 /**
58 * Trigger only outputting single language for multilanguage fields
59 *
60 * @param bool $val
61 * @since 1.23
62 */
63 public function setSingleLanguage( $val ) {
64 $this->singleLang = $val;
65 }
66
67 /**
68 * Numbers given by Exif user agents are often magical, that is they
69 * should be replaced by a detailed explanation depending on their
70 * value which most of the time are plain integers. This function
71 * formats Exif (and other metadata) values into human readable form.
72 *
73 * This is the usual entry point for this class.
74 *
75 * @param array $tags The Exif data to format ( as returned by
76 * Exif::getFilteredData() or BitmapMetadataHandler )
77 * @param bool|IContextSource $context Context to use (optional)
78 * @return array
79 */
80 public static function getFormattedData( $tags, $context = false ) {
81 $obj = new FormatMetadata;
82 if ( $context ) {
83 $obj->setContext( $context );
84 }
85
86 return $obj->makeFormattedData( $tags );
87 }
88
89 /**
90 * Numbers given by Exif user agents are often magical, that is they
91 * should be replaced by a detailed explanation depending on their
92 * value which most of the time are plain integers. This function
93 * formats Exif (and other metadata) values into human readable form.
94 *
95 * @param array $tags The Exif data to format ( as returned by
96 * Exif::getFilteredData() or BitmapMetadataHandler )
97 * @return array
98 * @since 1.23
99 */
100 public function makeFormattedData( $tags ) {
101 $resolutionunit = !isset( $tags['ResolutionUnit'] ) || $tags['ResolutionUnit'] == 2 ? 2 : 3;
102 unset( $tags['ResolutionUnit'] );
103
104 foreach ( $tags as $tag => &$vals ) {
105
106 // This seems ugly to wrap non-array's in an array just to unwrap again,
107 // especially when most of the time it is not an array
108 if ( !is_array( $tags[$tag] ) ) {
109 $vals = array( $vals );
110 }
111
112 // _type is a special value to say what array type
113 if ( isset( $tags[$tag]['_type'] ) ) {
114 $type = $tags[$tag]['_type'];
115 unset( $vals['_type'] );
116 } else {
117 $type = 'ul'; // default unordered list.
118 }
119
120 //This is done differently as the tag is an array.
121 if ( $tag == 'GPSTimeStamp' && count( $vals ) === 3 ) {
122 //hour min sec array
123
124 $h = explode( '/', $vals[0] );
125 $m = explode( '/', $vals[1] );
126 $s = explode( '/', $vals[2] );
127
128 // this should already be validated
129 // when loaded from file, but it could
130 // come from a foreign repo, so be
131 // paranoid.
132 if ( !isset( $h[1] )
133 || !isset( $m[1] )
134 || !isset( $s[1] )
135 || $h[1] == 0
136 || $m[1] == 0
137 || $s[1] == 0
138 ) {
139 continue;
140 }
141 $tags[$tag] = str_pad( intval( $h[0] / $h[1] ), 2, '0', STR_PAD_LEFT )
142 . ':' . str_pad( intval( $m[0] / $m[1] ), 2, '0', STR_PAD_LEFT )
143 . ':' . str_pad( intval( $s[0] / $s[1] ), 2, '0', STR_PAD_LEFT );
144
145 try {
146 $time = wfTimestamp( TS_MW, '1971:01:01 ' . $tags[$tag] );
147 // the 1971:01:01 is just a placeholder, and not shown to user.
148 if ( $time && intval( $time ) > 0 ) {
149 $tags[$tag] = $this->getLanguage()->time( $time );
150 }
151 } catch ( TimestampException $e ) {
152 // This shouldn't happen, but we've seen bad formats
153 // such as 4-digit seconds in the wild.
154 // leave $tags[$tag] as-is
155 }
156 continue;
157 }
158
159 // The contact info is a multi-valued field
160 // instead of the other props which are single
161 // valued (mostly) so handle as a special case.
162 if ( $tag === 'Contact' ) {
163 $vals = $this->collapseContactInfo( $vals );
164 continue;
165 }
166
167 foreach ( $vals as &$val ) {
168
169 switch ( $tag ) {
170 case 'Compression':
171 switch ( $val ) {
172 case 1:
173 case 2:
174 case 3:
175 case 4:
176 case 5:
177 case 6:
178 case 7:
179 case 8:
180 case 32773:
181 case 32946:
182 case 34712:
183 $val = $this->exifMsg( $tag, $val );
184 break;
185 default:
186 /* If not recognized, display as is. */
187 break;
188 }
189 break;
190
191 case 'PhotometricInterpretation':
192 switch ( $val ) {
193 case 2:
194 case 6:
195 $val = $this->exifMsg( $tag, $val );
196 break;
197 default:
198 /* If not recognized, display as is. */
199 break;
200 }
201 break;
202
203 case 'Orientation':
204 switch ( $val ) {
205 case 1:
206 case 2:
207 case 3:
208 case 4:
209 case 5:
210 case 6:
211 case 7:
212 case 8:
213 $val = $this->exifMsg( $tag, $val );
214 break;
215 default:
216 /* If not recognized, display as is. */
217 break;
218 }
219 break;
220
221 case 'PlanarConfiguration':
222 switch ( $val ) {
223 case 1:
224 case 2:
225 $val = $this->exifMsg( $tag, $val );
226 break;
227 default:
228 /* If not recognized, display as is. */
229 break;
230 }
231 break;
232
233 // TODO: YCbCrSubSampling
234 case 'YCbCrPositioning':
235 switch ( $val ) {
236 case 1:
237 case 2:
238 $val = $this->exifMsg( $tag, $val );
239 break;
240 default:
241 /* If not recognized, display as is. */
242 break;
243 }
244 break;
245
246 case 'XResolution':
247 case 'YResolution':
248 switch ( $resolutionunit ) {
249 case 2:
250 $val = $this->exifMsg( 'XYResolution', 'i', $this->formatNum( $val ) );
251 break;
252 case 3:
253 $val = $this->exifMsg( 'XYResolution', 'c', $this->formatNum( $val ) );
254 break;
255 default:
256 /* If not recognized, display as is. */
257 break;
258 }
259 break;
260
261 // TODO: YCbCrCoefficients #p27 (see annex E)
262 case 'ExifVersion':
263 case 'FlashpixVersion':
264 $val = "$val" / 100;
265 break;
266
267 case 'ColorSpace':
268 switch ( $val ) {
269 case 1:
270 case 65535:
271 $val = $this->exifMsg( $tag, $val );
272 break;
273 default:
274 /* If not recognized, display as is. */
275 break;
276 }
277 break;
278
279 case 'ComponentsConfiguration':
280 switch ( $val ) {
281 case 0:
282 case 1:
283 case 2:
284 case 3:
285 case 4:
286 case 5:
287 case 6:
288 $val = $this->exifMsg( $tag, $val );
289 break;
290 default:
291 /* If not recognized, display as is. */
292 break;
293 }
294 break;
295
296 case 'DateTime':
297 case 'DateTimeOriginal':
298 case 'DateTimeDigitized':
299 case 'DateTimeReleased':
300 case 'DateTimeExpires':
301 case 'GPSDateStamp':
302 case 'dc-date':
303 case 'DateTimeMetadata':
304 if ( $val == '0000:00:00 00:00:00' || $val == ' : : : : ' ) {
305 $val = $this->msg( 'exif-unknowndate' )->text();
306 } elseif ( preg_match(
307 '/^(?:\d{4}):(?:\d\d):(?:\d\d) (?:\d\d):(?:\d\d):(?:\d\d)$/D',
308 $val
309 ) ) {
310 // Full date.
311 $time = wfTimestamp( TS_MW, $val );
312 if ( $time && intval( $time ) > 0 ) {
313 $val = $this->getLanguage()->timeanddate( $time );
314 }
315 } elseif ( preg_match( '/^(?:\d{4}):(?:\d\d):(?:\d\d) (?:\d\d):(?:\d\d)$/D', $val ) ) {
316 // No second field. Still format the same
317 // since timeanddate doesn't include seconds anyways,
318 // but second still available in api
319 $time = wfTimestamp( TS_MW, $val . ':00' );
320 if ( $time && intval( $time ) > 0 ) {
321 $val = $this->getLanguage()->timeanddate( $time );
322 }
323 } elseif ( preg_match( '/^(?:\d{4}):(?:\d\d):(?:\d\d)$/D', $val ) ) {
324 // If only the date but not the time is filled in.
325 $time = wfTimestamp( TS_MW, substr( $val, 0, 4 )
326 . substr( $val, 5, 2 )
327 . substr( $val, 8, 2 )
328 . '000000' );
329 if ( $time && intval( $time ) > 0 ) {
330 $val = $this->getLanguage()->date( $time );
331 }
332 }
333 // else it will just output $val without formatting it.
334 break;
335
336 case 'ExposureProgram':
337 switch ( $val ) {
338 case 0:
339 case 1:
340 case 2:
341 case 3:
342 case 4:
343 case 5:
344 case 6:
345 case 7:
346 case 8:
347 $val = $this->exifMsg( $tag, $val );
348 break;
349 default:
350 /* If not recognized, display as is. */
351 break;
352 }
353 break;
354
355 case 'SubjectDistance':
356 $val = $this->exifMsg( $tag, '', $this->formatNum( $val ) );
357 break;
358
359 case 'MeteringMode':
360 switch ( $val ) {
361 case 0:
362 case 1:
363 case 2:
364 case 3:
365 case 4:
366 case 5:
367 case 6:
368 case 7:
369 case 255:
370 $val = $this->exifMsg( $tag, $val );
371 break;
372 default:
373 /* If not recognized, display as is. */
374 break;
375 }
376 break;
377
378 case 'LightSource':
379 switch ( $val ) {
380 case 0:
381 case 1:
382 case 2:
383 case 3:
384 case 4:
385 case 9:
386 case 10:
387 case 11:
388 case 12:
389 case 13:
390 case 14:
391 case 15:
392 case 17:
393 case 18:
394 case 19:
395 case 20:
396 case 21:
397 case 22:
398 case 23:
399 case 24:
400 case 255:
401 $val = $this->exifMsg( $tag, $val );
402 break;
403 default:
404 /* If not recognized, display as is. */
405 break;
406 }
407 break;
408
409 case 'Flash':
410 $flashDecode = array(
411 'fired' => $val & bindec( '00000001' ),
412 'return' => ( $val & bindec( '00000110' ) ) >> 1,
413 'mode' => ( $val & bindec( '00011000' ) ) >> 3,
414 'function' => ( $val & bindec( '00100000' ) ) >> 5,
415 'redeye' => ( $val & bindec( '01000000' ) ) >> 6,
416 // 'reserved' => ($val & bindec( '10000000' )) >> 7,
417 );
418 $flashMsgs = array();
419 # We do not need to handle unknown values since all are used.
420 foreach ( $flashDecode as $subTag => $subValue ) {
421 # We do not need any message for zeroed values.
422 if ( $subTag != 'fired' && $subValue == 0 ) {
423 continue;
424 }
425 $fullTag = $tag . '-' . $subTag;
426 $flashMsgs[] = $this->exifMsg( $fullTag, $subValue );
427 }
428 $val = $this->getLanguage()->commaList( $flashMsgs );
429 break;
430
431 case 'FocalPlaneResolutionUnit':
432 switch ( $val ) {
433 case 2:
434 $val = $this->exifMsg( $tag, $val );
435 break;
436 default:
437 /* If not recognized, display as is. */
438 break;
439 }
440 break;
441
442 case 'SensingMethod':
443 switch ( $val ) {
444 case 1:
445 case 2:
446 case 3:
447 case 4:
448 case 5:
449 case 7:
450 case 8:
451 $val = $this->exifMsg( $tag, $val );
452 break;
453 default:
454 /* If not recognized, display as is. */
455 break;
456 }
457 break;
458
459 case 'FileSource':
460 switch ( $val ) {
461 case 3:
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 'SceneType':
471 switch ( $val ) {
472 case 1:
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 'CustomRendered':
482 switch ( $val ) {
483 case 0:
484 case 1:
485 $val = $this->exifMsg( $tag, $val );
486 break;
487 default:
488 /* If not recognized, display as is. */
489 break;
490 }
491 break;
492
493 case 'ExposureMode':
494 switch ( $val ) {
495 case 0:
496 case 1:
497 case 2:
498 $val = $this->exifMsg( $tag, $val );
499 break;
500 default:
501 /* If not recognized, display as is. */
502 break;
503 }
504 break;
505
506 case 'WhiteBalance':
507 switch ( $val ) {
508 case 0:
509 case 1:
510 $val = $this->exifMsg( $tag, $val );
511 break;
512 default:
513 /* If not recognized, display as is. */
514 break;
515 }
516 break;
517
518 case 'SceneCaptureType':
519 switch ( $val ) {
520 case 0:
521 case 1:
522 case 2:
523 case 3:
524 $val = $this->exifMsg( $tag, $val );
525 break;
526 default:
527 /* If not recognized, display as is. */
528 break;
529 }
530 break;
531
532 case 'GainControl':
533 switch ( $val ) {
534 case 0:
535 case 1:
536 case 2:
537 case 3:
538 case 4:
539 $val = $this->exifMsg( $tag, $val );
540 break;
541 default:
542 /* If not recognized, display as is. */
543 break;
544 }
545 break;
546
547 case 'Contrast':
548 switch ( $val ) {
549 case 0:
550 case 1:
551 case 2:
552 $val = $this->exifMsg( $tag, $val );
553 break;
554 default:
555 /* If not recognized, display as is. */
556 break;
557 }
558 break;
559
560 case 'Saturation':
561 switch ( $val ) {
562 case 0:
563 case 1:
564 case 2:
565 $val = $this->exifMsg( $tag, $val );
566 break;
567 default:
568 /* If not recognized, display as is. */
569 break;
570 }
571 break;
572
573 case 'Sharpness':
574 switch ( $val ) {
575 case 0:
576 case 1:
577 case 2:
578 $val = $this->exifMsg( $tag, $val );
579 break;
580 default:
581 /* If not recognized, display as is. */
582 break;
583 }
584 break;
585
586 case 'SubjectDistanceRange':
587 switch ( $val ) {
588 case 0:
589 case 1:
590 case 2:
591 case 3:
592 $val = $this->exifMsg( $tag, $val );
593 break;
594 default:
595 /* If not recognized, display as is. */
596 break;
597 }
598 break;
599
600 //The GPS...Ref values are kept for compatibility, probably won't be reached.
601 case 'GPSLatitudeRef':
602 case 'GPSDestLatitudeRef':
603 switch ( $val ) {
604 case 'N':
605 case 'S':
606 $val = $this->exifMsg( 'GPSLatitude', $val );
607 break;
608 default:
609 /* If not recognized, display as is. */
610 break;
611 }
612 break;
613
614 case 'GPSLongitudeRef':
615 case 'GPSDestLongitudeRef':
616 switch ( $val ) {
617 case 'E':
618 case 'W':
619 $val = $this->exifMsg( 'GPSLongitude', $val );
620 break;
621 default:
622 /* If not recognized, display as is. */
623 break;
624 }
625 break;
626
627 case 'GPSAltitude':
628 if ( $val < 0 ) {
629 $val = $this->exifMsg( 'GPSAltitude', 'below-sealevel', $this->formatNum( -$val, 3 ) );
630 } else {
631 $val = $this->exifMsg( 'GPSAltitude', 'above-sealevel', $this->formatNum( $val, 3 ) );
632 }
633 break;
634
635 case 'GPSStatus':
636 switch ( $val ) {
637 case 'A':
638 case 'V':
639 $val = $this->exifMsg( $tag, $val );
640 break;
641 default:
642 /* If not recognized, display as is. */
643 break;
644 }
645 break;
646
647 case 'GPSMeasureMode':
648 switch ( $val ) {
649 case 2:
650 case 3:
651 $val = $this->exifMsg( $tag, $val );
652 break;
653 default:
654 /* If not recognized, display as is. */
655 break;
656 }
657 break;
658
659 case 'GPSTrackRef':
660 case 'GPSImgDirectionRef':
661 case 'GPSDestBearingRef':
662 switch ( $val ) {
663 case 'T':
664 case 'M':
665 $val = $this->exifMsg( 'GPSDirection', $val );
666 break;
667 default:
668 /* If not recognized, display as is. */
669 break;
670 }
671 break;
672
673 case 'GPSLatitude':
674 case 'GPSDestLatitude':
675 $val = $this->formatCoords( $val, 'latitude' );
676 break;
677 case 'GPSLongitude':
678 case 'GPSDestLongitude':
679 $val = $this->formatCoords( $val, 'longitude' );
680 break;
681
682 case 'GPSSpeedRef':
683 switch ( $val ) {
684 case 'K':
685 case 'M':
686 case 'N':
687 $val = $this->exifMsg( 'GPSSpeed', $val );
688 break;
689 default:
690 /* If not recognized, display as is. */
691 break;
692 }
693 break;
694
695 case 'GPSDestDistanceRef':
696 switch ( $val ) {
697 case 'K':
698 case 'M':
699 case 'N':
700 $val = $this->exifMsg( 'GPSDestDistance', $val );
701 break;
702 default:
703 /* If not recognized, display as is. */
704 break;
705 }
706 break;
707
708 case 'GPSDOP':
709 // See http://en.wikipedia.org/wiki/Dilution_of_precision_(GPS)
710 if ( $val <= 2 ) {
711 $val = $this->exifMsg( $tag, 'excellent', $this->formatNum( $val ) );
712 } elseif ( $val <= 5 ) {
713 $val = $this->exifMsg( $tag, 'good', $this->formatNum( $val ) );
714 } elseif ( $val <= 10 ) {
715 $val = $this->exifMsg( $tag, 'moderate', $this->formatNum( $val ) );
716 } elseif ( $val <= 20 ) {
717 $val = $this->exifMsg( $tag, 'fair', $this->formatNum( $val ) );
718 } else {
719 $val = $this->exifMsg( $tag, 'poor', $this->formatNum( $val ) );
720 }
721 break;
722
723 // This is not in the Exif standard, just a special
724 // case for our purposes which enables wikis to wikify
725 // the make, model and software name to link to their articles.
726 case 'Make':
727 case 'Model':
728 $val = $this->exifMsg( $tag, '', $val );
729 break;
730
731 case 'Software':
732 if ( is_array( $val ) ) {
733 //if its a software, version array.
734 $val = $this->msg( 'exif-software-version-value', $val[0], $val[1] )->text();
735 } else {
736 $val = $this->exifMsg( $tag, '', $val );
737 }
738 break;
739
740 case 'ExposureTime':
741 // Show the pretty fraction as well as decimal version
742 $val = $this->msg( 'exif-exposuretime-format',
743 $this->formatFraction( $val ), $this->formatNum( $val ) )->text();
744 break;
745 case 'ISOSpeedRatings':
746 // If its = 65535 that means its at the
747 // limit of the size of Exif::short and
748 // is really higher.
749 if ( $val == '65535' ) {
750 $val = $this->exifMsg( $tag, 'overflow' );
751 } else {
752 $val = $this->formatNum( $val );
753 }
754 break;
755 case 'FNumber':
756 $val = $this->msg( 'exif-fnumber-format',
757 $this->formatNum( $val ) )->text();
758 break;
759
760 case 'FocalLength':
761 case 'FocalLengthIn35mmFilm':
762 $val = $this->msg( 'exif-focallength-format',
763 $this->formatNum( $val ) )->text();
764 break;
765
766 case 'MaxApertureValue':
767 if ( strpos( $val, '/' ) !== false ) {
768 // need to expand this earlier to calculate fNumber
769 list( $n, $d ) = explode( '/', $val );
770 if ( is_numeric( $n ) && is_numeric( $d ) ) {
771 $val = $n / $d;
772 }
773 }
774 if ( is_numeric( $val ) ) {
775 $fNumber = pow( 2, $val / 2 );
776 if ( $fNumber !== false ) {
777 $val = $this->msg( 'exif-maxaperturevalue-value',
778 $this->formatNum( $val ),
779 $this->formatNum( $fNumber, 2 )
780 )->text();
781 }
782 }
783 break;
784
785 case 'iimCategory':
786 switch ( strtolower( $val ) ) {
787 // See pg 29 of IPTC photo
788 // metadata standard.
789 case 'ace':
790 case 'clj':
791 case 'dis':
792 case 'fin':
793 case 'edu':
794 case 'evn':
795 case 'hth':
796 case 'hum':
797 case 'lab':
798 case 'lif':
799 case 'pol':
800 case 'rel':
801 case 'sci':
802 case 'soi':
803 case 'spo':
804 case 'war':
805 case 'wea':
806 $val = $this->exifMsg(
807 'iimcategory',
808 $val
809 );
810 }
811 break;
812 case 'SubjectNewsCode':
813 // Essentially like iimCategory.
814 // 8 (numeric) digit hierarchical
815 // classification. We decode the
816 // first 2 digits, which provide
817 // a broad category.
818 $val = $this->convertNewsCode( $val );
819 break;
820 case 'Urgency':
821 // 1-8 with 1 being highest, 5 normal
822 // 0 is reserved, and 9 is 'user-defined'.
823 $urgency = '';
824 if ( $val == 0 || $val == 9 ) {
825 $urgency = 'other';
826 } elseif ( $val < 5 && $val > 1 ) {
827 $urgency = 'high';
828 } elseif ( $val == 5 ) {
829 $urgency = 'normal';
830 } elseif ( $val <= 8 && $val > 5 ) {
831 $urgency = 'low';
832 }
833
834 if ( $urgency !== '' ) {
835 $val = $this->exifMsg( 'urgency',
836 $urgency, $val
837 );
838 }
839 break;
840
841 // Things that have a unit of pixels.
842 case 'OriginalImageHeight':
843 case 'OriginalImageWidth':
844 case 'PixelXDimension':
845 case 'PixelYDimension':
846 case 'ImageWidth':
847 case 'ImageLength':
848 $val = $this->formatNum( $val ) . ' ' . $this->msg( 'unit-pixel' )->text();
849 break;
850
851 // Do not transform fields with pure text.
852 // For some languages the formatNum()
853 // conversion results to wrong output like
854 // foo,bar@example,com or foo٫bar@example٫com.
855 // Also some 'numeric' things like Scene codes
856 // are included here as we really don't want
857 // commas inserted.
858 case 'ImageDescription':
859 case 'Artist':
860 case 'Copyright':
861 case 'RelatedSoundFile':
862 case 'ImageUniqueID':
863 case 'SpectralSensitivity':
864 case 'GPSSatellites':
865 case 'GPSVersionID':
866 case 'GPSMapDatum':
867 case 'Keywords':
868 case 'WorldRegionDest':
869 case 'CountryDest':
870 case 'CountryCodeDest':
871 case 'ProvinceOrStateDest':
872 case 'CityDest':
873 case 'SublocationDest':
874 case 'WorldRegionCreated':
875 case 'CountryCreated':
876 case 'CountryCodeCreated':
877 case 'ProvinceOrStateCreated':
878 case 'CityCreated':
879 case 'SublocationCreated':
880 case 'ObjectName':
881 case 'SpecialInstructions':
882 case 'Headline':
883 case 'Credit':
884 case 'Source':
885 case 'EditStatus':
886 case 'FixtureIdentifier':
887 case 'LocationDest':
888 case 'LocationDestCode':
889 case 'Writer':
890 case 'JPEGFileComment':
891 case 'iimSupplementalCategory':
892 case 'OriginalTransmissionRef':
893 case 'Identifier':
894 case 'dc-contributor':
895 case 'dc-coverage':
896 case 'dc-publisher':
897 case 'dc-relation':
898 case 'dc-rights':
899 case 'dc-source':
900 case 'dc-type':
901 case 'Lens':
902 case 'SerialNumber':
903 case 'CameraOwnerName':
904 case 'Label':
905 case 'Nickname':
906 case 'RightsCertificate':
907 case 'CopyrightOwner':
908 case 'UsageTerms':
909 case 'WebStatement':
910 case 'OriginalDocumentID':
911 case 'LicenseUrl':
912 case 'MorePermissionsUrl':
913 case 'AttributionUrl':
914 case 'PreferredAttributionName':
915 case 'PNGFileComment':
916 case 'Disclaimer':
917 case 'ContentWarning':
918 case 'GIFFileComment':
919 case 'SceneCode':
920 case 'IntellectualGenre':
921 case 'Event':
922 case 'OrginisationInImage':
923 case 'PersonInImage':
924
925 $val = htmlspecialchars( $val );
926 break;
927
928 case 'ObjectCycle':
929 switch ( $val ) {
930 case 'a':
931 case 'p':
932 case 'b':
933 $val = $this->exifMsg( $tag, $val );
934 break;
935 default:
936 $val = htmlspecialchars( $val );
937 break;
938 }
939 break;
940 case 'Copyrighted':
941 switch ( $val ) {
942 case 'True':
943 case 'False':
944 $val = $this->exifMsg( $tag, $val );
945 break;
946 }
947 break;
948 case 'Rating':
949 if ( $val == '-1' ) {
950 $val = $this->exifMsg( $tag, 'rejected' );
951 } else {
952 $val = $this->formatNum( $val );
953 }
954 break;
955
956 case 'LanguageCode':
957 $lang = Language::fetchLanguageName( strtolower( $val ), $this->getLanguage()->getCode() );
958 if ( $lang ) {
959 $val = htmlspecialchars( $lang );
960 } else {
961 $val = htmlspecialchars( $val );
962 }
963 break;
964
965 default:
966 $val = $this->formatNum( $val );
967 break;
968 }
969 }
970 // End formatting values, start flattening arrays.
971 $vals = $this->flattenArrayReal( $vals, $type );
972 }
973
974 return $tags;
975 }
976
977 /**
978 * Flatten an array, using the content language for any messages.
979 *
980 * @param array $vals Array of values
981 * @param string $type Type of array (either lang, ul, ol).
982 * lang = language assoc array with keys being the lang code
983 * ul = unordered list, ol = ordered list
984 * type can also come from the '_type' member of $vals.
985 * @param bool $noHtml If to avoid returning anything resembling HTML.
986 * (Ugly hack for backwards compatibility with old MediaWiki).
987 * @param bool|IContextSource $context
988 * @return string Single value (in wiki-syntax).
989 * @since 1.23
990 */
991 public static function flattenArrayContentLang( $vals, $type = 'ul',
992 $noHtml = false, $context = false
993 ) {
994 global $wgContLang;
995 $obj = new FormatMetadata;
996 if ( $context ) {
997 $obj->setContext( $context );
998 }
999 $context = new DerivativeContext( $obj->getContext() );
1000 $context->setLanguage( $wgContLang );
1001 $obj->setContext( $context );
1002
1003 return $obj->flattenArrayReal( $vals, $type, $noHtml );
1004 }
1005
1006 /**
1007 * Flatten an array, using the user language for any messages.
1008 *
1009 * @param array $vals Array of values
1010 * @param string $type Type of array (either lang, ul, ol).
1011 * lang = language assoc array with keys being the lang code
1012 * ul = unordered list, ol = ordered list
1013 * type can also come from the '_type' member of $vals.
1014 * @param bool $noHtml If to avoid returning anything resembling HTML.
1015 * (Ugly hack for backwards compatibility with old MediaWiki).
1016 * @param bool|IContextSource $context
1017 * @return string Single value (in wiki-syntax).
1018 */
1019 public static function flattenArray( $vals, $type = 'ul', $noHtml = false, $context = false ) {
1020 $obj = new FormatMetadata;
1021 if ( $context ) {
1022 $obj->setContext( $context );
1023 }
1024
1025 return $obj->flattenArrayReal( $vals, $type, $noHtml );
1026 }
1027
1028 /**
1029 * A function to collapse multivalued tags into a single value.
1030 * This turns an array of (for example) authors into a bulleted list.
1031 *
1032 * This is public on the basis it might be useful outside of this class.
1033 *
1034 * @param array $vals Array of values
1035 * @param string $type Type of array (either lang, ul, ol).
1036 * lang = language assoc array with keys being the lang code
1037 * ul = unordered list, ol = ordered list
1038 * type can also come from the '_type' member of $vals.
1039 * @param bool $noHtml If to avoid returning anything resembling HTML.
1040 * (Ugly hack for backwards compatibility with old mediawiki).
1041 * @return string Single value (in wiki-syntax).
1042 * @since 1.23
1043 */
1044 public function flattenArrayReal( $vals, $type = 'ul', $noHtml = false ) {
1045 if ( !is_array( $vals ) ) {
1046 return $vals; // do nothing if not an array;
1047 }
1048
1049 if ( isset( $vals['_type'] ) ) {
1050 $type = $vals['_type'];
1051 unset( $vals['_type'] );
1052 }
1053
1054 if ( !is_array( $vals ) ) {
1055 return $vals; // do nothing if not an array;
1056 } elseif ( count( $vals ) === 1 && $type !== 'lang' ) {
1057 return $vals[0];
1058 } elseif ( count( $vals ) === 0 ) {
1059 wfDebug( __METHOD__ . " metadata array with 0 elements!\n" );
1060
1061 return ""; // paranoia. This should never happen
1062 } else {
1063 /* @todo FIXME: This should hide some of the list entries if there are
1064 * say more than four. Especially if a field is translated into 20
1065 * languages, we don't want to show them all by default
1066 */
1067 switch ( $type ) {
1068 case 'lang':
1069 // Display default, followed by ContLang,
1070 // followed by the rest in no particular
1071 // order.
1072
1073 // Todo: hide some items if really long list.
1074
1075 $content = '';
1076
1077 $priorityLanguages = $this->getPriorityLanguages();
1078 $defaultItem = false;
1079 $defaultLang = false;
1080
1081 // If default is set, save it for later,
1082 // as we don't know if it's equal to
1083 // one of the lang codes. (In xmp
1084 // you specify the language for a
1085 // default property by having both
1086 // a default prop, and one in the language
1087 // that are identical)
1088 if ( isset( $vals['x-default'] ) ) {
1089 $defaultItem = $vals['x-default'];
1090 unset( $vals['x-default'] );
1091 }
1092 foreach ( $priorityLanguages as $pLang ) {
1093 if ( isset( $vals[$pLang] ) ) {
1094 $isDefault = false;
1095 if ( $vals[$pLang] === $defaultItem ) {
1096 $defaultItem = false;
1097 $isDefault = true;
1098 }
1099 $content .= $this->langItem(
1100 $vals[$pLang], $pLang,
1101 $isDefault, $noHtml );
1102
1103 unset( $vals[$pLang] );
1104
1105 if ( $this->singleLang ) {
1106 return Html::rawElement( 'span',
1107 array( 'lang' => $pLang ), $vals[$pLang] );
1108 }
1109 }
1110 }
1111
1112 // Now do the rest.
1113 foreach ( $vals as $lang => $item ) {
1114 if ( $item === $defaultItem ) {
1115 $defaultLang = $lang;
1116 continue;
1117 }
1118 $content .= $this->langItem( $item,
1119 $lang, false, $noHtml );
1120 if ( $this->singleLang ) {
1121 return Html::rawElement( 'span',
1122 array( 'lang' => $lang ), $item );
1123 }
1124 }
1125 if ( $defaultItem !== false ) {
1126 $content = $this->langItem( $defaultItem,
1127 $defaultLang, true, $noHtml ) .
1128 $content;
1129 if ( $this->singleLang ) {
1130 return $defaultItem;
1131 }
1132 }
1133 if ( $noHtml ) {
1134 return $content;
1135 }
1136
1137 return '<ul class="metadata-langlist">' .
1138 $content .
1139 '</ul>';
1140 case 'ol':
1141 if ( $noHtml ) {
1142 return "\n#" . implode( "\n#", $vals );
1143 }
1144
1145 return "<ol><li>" . implode( "</li>\n<li>", $vals ) . '</li></ol>';
1146 case 'ul':
1147 default:
1148 if ( $noHtml ) {
1149 return "\n*" . implode( "\n*", $vals );
1150 }
1151
1152 return "<ul><li>" . implode( "</li>\n<li>", $vals ) . '</li></ul>';
1153 }
1154 }
1155 }
1156
1157 /** Helper function for creating lists of translations.
1158 *
1159 * @param string $value Value (this is not escaped)
1160 * @param string $lang Lang code of item or false
1161 * @param bool $default If it is default value.
1162 * @param bool $noHtml If to avoid html (for back-compat)
1163 * @throws MWException
1164 * @return string Language item (Note: despite how this looks, this is
1165 * treated as wikitext, not as HTML).
1166 */
1167 private function langItem( $value, $lang, $default = false, $noHtml = false ) {
1168 if ( $lang === false && $default === false ) {
1169 throw new MWException( '$lang and $default cannot both '
1170 . 'be false.' );
1171 }
1172
1173 if ( $noHtml ) {
1174 $wrappedValue = $value;
1175 } else {
1176 $wrappedValue = '<span class="mw-metadata-lang-value">'
1177 . $value . '</span>';
1178 }
1179
1180 if ( $lang === false ) {
1181 $msg = $this->msg( 'metadata-langitem-default', $wrappedValue );
1182 if ( $noHtml ) {
1183 return $msg->text() . "\n\n";
1184 } /* else */
1185
1186 return '<li class="mw-metadata-lang-default">'
1187 . $msg->text()
1188 . "</li>\n";
1189 }
1190
1191 $lowLang = strtolower( $lang );
1192 $langName = Language::fetchLanguageName( $lowLang );
1193 if ( $langName === '' ) {
1194 //try just the base language name. (aka en-US -> en ).
1195 list( $langPrefix ) = explode( '-', $lowLang, 2 );
1196 $langName = Language::fetchLanguageName( $langPrefix );
1197 if ( $langName === '' ) {
1198 // give up.
1199 $langName = $lang;
1200 }
1201 }
1202 // else we have a language specified
1203
1204 $msg = $this->msg( 'metadata-langitem', $wrappedValue, $langName, $lang );
1205 if ( $noHtml ) {
1206 return '*' . $msg->text();
1207 } /* else: */
1208
1209 $item = '<li class="mw-metadata-lang-code-'
1210 . $lang;
1211 if ( $default ) {
1212 $item .= ' mw-metadata-lang-default';
1213 }
1214 $item .= '" lang="' . $lang . '">';
1215 $item .= $msg->text();
1216 $item .= "</li>\n";
1217
1218 return $item;
1219 }
1220
1221 /**
1222 * Convenience function for getFormattedData()
1223 *
1224 * @param string $tag The tag name to pass on
1225 * @param string $val The value of the tag
1226 * @param string $arg An argument to pass ($1)
1227 * @param string $arg2 A 2nd argument to pass ($2)
1228 * @return string The text content of "exif-$tag-$val" message in lower case
1229 */
1230 private function exifMsg( $tag, $val, $arg = null, $arg2 = null ) {
1231 global $wgContLang;
1232
1233 if ( $val === '' ) {
1234 $val = 'value';
1235 }
1236
1237 return $this->msg( $wgContLang->lc( "exif-$tag-$val" ), $arg, $arg2 )->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 = array();
1250 if ( is_array( $num ) ) {
1251 $out = array();
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 $m = array();
1286 if ( preg_match( '/^(-?\d+)\/(\d+)$/', $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 // http://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 = array();
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 = array();
1576 $lines = explode( "\n", wfMessage( 'metadata-fields' )->inContentLanguage()->text() );
1577 foreach ( $lines as $line ) {
1578 $matches = array();
1579 if ( preg_match( '/^\\*\s*(.*?)\s*$/', $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 global $wgMemc;
1597
1598 // If revision deleted, exit immediately
1599 if ( $file->isDeleted( File::DELETED_FILE ) ) {
1600
1601 return array();
1602 }
1603
1604 $cacheKey = wfMemcKey(
1605 'getExtendedMetadata',
1606 $this->getLanguage()->getCode(),
1607 (int)$this->singleLang,
1608 $file->getSha1()
1609 );
1610
1611 $cachedValue = $wgMemc->get( $cacheKey );
1612 if (
1613 $cachedValue
1614 && Hooks::run( 'ValidateExtendedMetadataCache', array( $cachedValue['timestamp'], $file ) )
1615 ) {
1616 $extendedMetadata = $cachedValue['data'];
1617 } else {
1618 $maxCacheTime = ( $file instanceof ForeignAPIFile ) ? 60 * 60 * 12 : 60 * 60 * 24 * 30;
1619 $fileMetadata = $this->getExtendedMetadataFromFile( $file );
1620 $extendedMetadata = $this->getExtendedMetadataFromHook( $file, $fileMetadata, $maxCacheTime );
1621 if ( $this->singleLang ) {
1622 $this->resolveMultilangMetadata( $extendedMetadata );
1623 }
1624 $this->discardMultipleValues( $extendedMetadata );
1625 // Make sure the metadata won't break the API when an XML format is used.
1626 // This is an API-specific function so it would be cleaner to call it from
1627 // outside fetchExtendedMetadata, but this way we don't need to redo the
1628 // computation on a cache hit.
1629 $this->sanitizeArrayForXml( $extendedMetadata );
1630 $valueToCache = array( 'data' => $extendedMetadata, 'timestamp' => wfTimestampNow() );
1631 $wgMemc->set( $cacheKey, $valueToCache, $maxCacheTime );
1632 }
1633
1634 return $extendedMetadata;
1635 }
1636
1637 /**
1638 * Get file-based metadata in standardized format.
1639 *
1640 * Note that for a remote file, this might return metadata supplied by extensions.
1641 *
1642 * @param File $file File to use
1643 * @return array [<property name> => ['value' => <value>]], or [] on error
1644 * @since 1.23
1645 */
1646 protected function getExtendedMetadataFromFile( File $file ) {
1647 // If this is a remote file accessed via an API request, we already
1648 // have remote metadata so we just ignore any local one
1649 if ( $file instanceof ForeignAPIFile ) {
1650 // In case of error we pretend no metadata - this will get cached.
1651 // Might or might not be a good idea.
1652 return $file->getExtendedMetadata() ?: array();
1653 }
1654
1655 $uploadDate = wfTimestamp( TS_ISO_8601, $file->getTimestamp() );
1656
1657 $fileMetadata = array(
1658 // This is modification time, which is close to "upload" time.
1659 'DateTime' => array(
1660 'value' => $uploadDate,
1661 'source' => 'mediawiki-metadata',
1662 ),
1663 );
1664
1665 $title = $file->getTitle();
1666 if ( $title ) {
1667 $text = $title->getText();
1668 $pos = strrpos( $text, '.' );
1669
1670 if ( $pos ) {
1671 $name = substr( $text, 0, $pos );
1672 } else {
1673 $name = $text;
1674 }
1675
1676 $fileMetadata['ObjectName'] = array(
1677 'value' => $name,
1678 'source' => 'mediawiki-metadata',
1679 );
1680 }
1681
1682 return $fileMetadata;
1683 }
1684
1685 /**
1686 * Get additional metadata from hooks in standardized format.
1687 *
1688 * @param File $file File to use
1689 * @param array $extendedMetadata
1690 * @param int $maxCacheTime Hook handlers might use this parameter to override cache time
1691 *
1692 * @return array [<property name> => ['value' => <value>]], or [] on error
1693 * @since 1.23
1694 */
1695 protected function getExtendedMetadataFromHook( File $file, array $extendedMetadata,
1696 &$maxCacheTime
1697 ) {
1698
1699 Hooks::run( 'GetExtendedMetadata', array(
1700 &$extendedMetadata,
1701 $file,
1702 $this->getContext(),
1703 $this->singleLang,
1704 &$maxCacheTime
1705 ) );
1706
1707 $visible = array_flip( self::getVisibleFields() );
1708 foreach ( $extendedMetadata as $key => $value ) {
1709 if ( !isset( $visible[strtolower( $key )] ) ) {
1710 $extendedMetadata[$key]['hidden'] = '';
1711 }
1712 }
1713
1714 return $extendedMetadata;
1715 }
1716
1717 /**
1718 * Turns an XMP-style multilang array into a single value.
1719 * If the value is not a multilang array, it is returned unchanged.
1720 * See mediawiki.org/wiki/Manual:File_metadata_handling#Multi-language_array_format
1721 * @param mixed $value
1722 * @return mixed Value in best language, null if there were no languages at all
1723 * @since 1.23
1724 */
1725 protected function resolveMultilangValue( $value ) {
1726 if (
1727 !is_array( $value )
1728 || !isset( $value['_type'] )
1729 || $value['_type'] != 'lang'
1730 ) {
1731 return $value; // do nothing if not a multilang array
1732 }
1733
1734 // choose the language best matching user or site settings
1735 $priorityLanguages = $this->getPriorityLanguages();
1736 foreach ( $priorityLanguages as $lang ) {
1737 if ( isset( $value[$lang] ) ) {
1738 return $value[$lang];
1739 }
1740 }
1741
1742 // otherwise go with the default language, if set
1743 if ( isset( $value['x-default'] ) ) {
1744 return $value['x-default'];
1745 }
1746
1747 // otherwise just return any one language
1748 unset( $value['_type'] );
1749 if ( !empty( $value ) ) {
1750 return reset( $value );
1751 }
1752
1753 // this should not happen; signal error
1754 return null;
1755 }
1756
1757 /**
1758 * Turns an XMP-style multivalue array into a single value by dropping all but the first value.
1759 * If the value is not a multivalue array (or a multivalue array inside a multilang array), it is returned unchanged.
1760 * See mediawiki.org/wiki/Manual:File_metadata_handling#Multi-language_array_format
1761 * @param mixed $value
1762 * @return mixed The value, or the first value if there were multiple ones
1763 * @since 1.25
1764 */
1765 protected function resolveMultivalueValue( $value ) {
1766 if ( !is_array( $value ) ) {
1767 return $value;
1768 } elseif ( isset( $value['_type'] ) && $value['_type'] === 'lang' ) { // if this is a multilang array, process fields separately
1769 $newValue = array();
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 list( $k, $v ) = each( $value );
1776 if ( $k === '_type' ) {
1777 $v = current( $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 /**
1824 * Makes sure the given array is a valid API response fragment
1825 * (can be transformed into XML)
1826 * @param array $arr
1827 */
1828 protected function sanitizeArrayForXml( &$arr ) {
1829 if ( !is_array( $arr ) ) {
1830 return;
1831 }
1832
1833 $counter = 1;
1834 foreach ( $arr as $key => &$value ) {
1835 $sanitizedKey = $this->sanitizeKeyForXml( $key );
1836 if ( $sanitizedKey !== $key ) {
1837 if ( isset( $arr[$sanitizedKey] ) ) {
1838 // Make the sanitized keys hopefully unique.
1839 // To make it definitely unique would be too much effort, given that
1840 // sanitizing is only needed for misformatted metadata anyway, but
1841 // this at least covers the case when $arr is numeric.
1842 $sanitizedKey .= $counter;
1843 ++$counter;
1844 }
1845 $arr[$sanitizedKey] = $arr[$key];
1846 unset( $arr[$key] );
1847 }
1848 if ( is_array( $value ) ) {
1849 $this->sanitizeArrayForXml( $value );
1850 }
1851 }
1852 }
1853
1854 /**
1855 * Turns a string into a valid XML identifier.
1856 * Used to ensure that keys of an associative array in the
1857 * API response do not break the XML formatter.
1858 * @param string $key
1859 * @return string
1860 * @since 1.23
1861 */
1862 protected function sanitizeKeyForXml( $key ) {
1863 // drop all characters which are not valid in an XML tag name
1864 // a bunch of non-ASCII letters would be valid but probably won't
1865 // be used so we take the easy way
1866 $key = preg_replace( '/[^a-zA-z0-9_:.-]/', '', $key );
1867 // drop characters which are invalid at the first position
1868 $key = preg_replace( '/^[\d-.]+/', '', $key );
1869
1870 if ( $key == '' ) {
1871 $key = '_';
1872 }
1873
1874 // special case for an internal keyword
1875 if ( $key == '_element' ) {
1876 $key = 'element';
1877 }
1878
1879 return $key;
1880 }
1881
1882 /**
1883 * Returns a list of languages (first is best) to use when formatting multilang fields,
1884 * based on user and site preferences.
1885 * @return array
1886 * @since 1.23
1887 */
1888 protected function getPriorityLanguages() {
1889 $priorityLanguages =
1890 Language::getFallbacksIncludingSiteLanguage( $this->getLanguage()->getCode() );
1891 $priorityLanguages = array_merge(
1892 (array)$this->getLanguage()->getCode(),
1893 $priorityLanguages[0],
1894 $priorityLanguages[1]
1895 );
1896
1897 return $priorityLanguages;
1898 }
1899 }
MediaWiki Core