includes/media/IPTC.php

   1 <?php
   2 /**
   3  * Class for some IPTC functions.
   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  * @file
  21  * @ingroup Media
  22  */
  23
  24 /**
  25  * Class for some IPTC functions.
  26  *
  27  * @ingroup Media
  28  */
  29 class IPTC {
  30         /**
  31          * This takes the results of iptcparse() and puts it into a
  32          * form that can be handled by mediawiki. Generally called from
  33          * BitmapMetadataHandler::doApp13.
  34          *
  35          * @see http://www.iptc.org/std/IIM/4.1/specification/IIMV4.1.pdf
  36          *
  37          * @param string $rawData The app13 block from jpeg containing iptc/iim data
  38          * @return array IPTC metadata array
  39          * @suppress PhanTypeArraySuspicious
  40          */
  41         static function parse( $rawData ) {
  42                 $parsed = iptcparse( $rawData );
  43                 $data = [];
  44                 if ( !is_array( $parsed ) ) {
  45                         return $data;
  46                 }
  47
  48                 $c = '';
  49                 // charset info contained in tag 1:90.
  50                 if ( isset( $parsed['1#090'] ) && isset( $parsed['1#090'][0] ) ) {
  51                         $c = self::getCharset( $parsed['1#090'][0] );
  52                         if ( $c === false ) {
  53                                 // Unknown charset. refuse to parse.
  54                                 // note: There is a different between
  55                                 // unknown and no charset specified.
  56                                 return [];
  57                         }
  58                         unset( $parsed['1#090'] );
  59                 }
  60
  61                 foreach ( $parsed as $tag => $val ) {
  62                         if ( isset( $val[0] ) && trim( $val[0] ) == '' ) {
  63                                 wfDebugLog( 'iptc', "IPTC tag $tag had only whitespace as its value." );
  64                                 continue;
  65                         }
  66                         switch ( $tag ) {
  67                                 case '2#120': /*IPTC caption. mapped with exif ImageDescription*/
  68                                         $data['ImageDescription'] = self::convIPTC( $val, $c );
  69                                         break;
  70                                 case '2#116': /* copyright. Mapped with exif copyright */
  71                                         $data['Copyright'] = self::convIPTC( $val, $c );
  72                                         break;
  73                                 case '2#080': /* byline. Mapped with exif Artist */
  74                                         /* merge with byline title (2:85)
  75                                          * like how exif does it with
  76                                          * Title, person. Not sure if this is best
  77                                          * approach since we no longer have the two fields
  78                                          * separate. each byline title entry corresponds to a
  79                                          * specific byline. */
  80
  81                                         $bylines = self::convIPTC( $val, $c );
  82                                         if ( isset( $parsed['2#085'] ) ) {
  83                                                 $titles = self::convIPTC( $parsed['2#085'], $c );
  84                                         } else {
  85                                                 $titles = [];
  86                                         }
  87
  88                                         $titleCount = count( $titles );
  89                                         for ( $i = 0; $i < $titleCount; $i++ ) {
  90                                                 if ( isset( $bylines[$i] ) ) {
  91                                                         // theoretically this should always be set
  92                                                         // but doesn't hurt to be careful.
  93                                                         $bylines[$i] = $titles[$i] . ', ' . $bylines[$i];
  94                                                 }
  95                                         }
  96                                         $data['Artist'] = $bylines;
  97                                         break;
  98                                 case '2#025': /* keywords */
  99                                         $data['Keywords'] = self::convIPTC( $val, $c );
 100                                         break;
 101                                 case '2#101': /* Country (shown) */
 102                                         $data['CountryDest'] = self::convIPTC( $val, $c );
 103                                         break;
 104                                 case '2#095': /* state/province (shown) */
 105                                         $data['ProvinceOrStateDest'] = self::convIPTC( $val, $c );
 106                                         break;
 107                                 case '2#090': /* city (Shown) */
 108                                         $data['CityDest'] = self::convIPTC( $val, $c );
 109                                         break;
 110                                 case '2#092': /* sublocation (shown) */
 111                                         $data['SublocationDest'] = self::convIPTC( $val, $c );
 112                                         break;
 113                                 case '2#005': /* object name/title */
 114                                         $data['ObjectName'] = self::convIPTC( $val, $c );
 115                                         break;
 116                                 case '2#040': /* special instructions */
 117                                         $data['SpecialInstructions'] = self::convIPTC( $val, $c );
 118                                         break;
 119                                 case '2#105': /* headline */
 120                                         $data['Headline'] = self::convIPTC( $val, $c );
 121                                         break;
 122                                 case '2#110': /* credit */
 123                                         /*"Identifies the provider of the objectdata,
 124                                          * not necessarily the owner/creator". */
 125                                         $data['Credit'] = self::convIPTC( $val, $c );
 126                                         break;
 127                                 case '2#115': /* source */
 128                                         /* "Identifies the original owner of the intellectual content of the
 129                                          *objectdata. This could be an agency, a member of an agency or
 130                                          *an individual." */
 131                                         $data['Source'] = self::convIPTC( $val, $c );
 132                                         break;
 133
 134                                 case '2#007': /* edit status (lead, correction, etc) */
 135                                         $data['EditStatus'] = self::convIPTC( $val, $c );
 136                                         break;
 137                                 case '2#015': /* category. deprecated. max 3 letters in theory, often more */
 138                                         $data['iimCategory'] = self::convIPTC( $val, $c );
 139                                         break;
 140                                 case '2#020': /* category. deprecated. */
 141                                         $data['iimSupplementalCategory'] = self::convIPTC( $val, $c );
 142                                         break;
 143                                 case '2#010': /*urgency (1-8. 1 most, 5 normal, 8 low priority)*/
 144                                         $data['Urgency'] = self::convIPTC( $val, $c );
 145                                         break;
 146                                 case '2#022':
 147                                         /* "Identifies objectdata that recurs often and predictably...
 148                                          * Example: Euroweather" */
 149                                         $data['FixtureIdentifier'] = self::convIPTC( $val, $c );
 150                                         break;
 151                                 case '2#026':
 152                                         /* Content location code (iso 3166 + some custom things)
 153                                          * ex: TUR (for turkey), XUN (for UN), XSP (outer space)
 154                                          * See wikipedia article on iso 3166 and appendix D of iim std. */
 155                                         $data['LocationDestCode'] = self::convIPTC( $val, $c );
 156                                         break;
 157                                 case '2#027':
 158                                         /* Content location name. Full printable name
 159                                          * of location of photo. */
 160                                         $data['LocationDest'] = self::convIPTC( $val, $c );
 161                                         break;
 162                                 case '2#065':
 163                                         /* Originating Program.
 164                                          * Combine with Program version (2:70) if present.
 165                                          */
 166                                         $software = self::convIPTC( $val, $c );
 167
 168                                         if ( count( $software ) !== 1 ) {
 169                                                 // according to iim standard this cannot have multiple values
 170                                                 // so if there is more than one, something weird is happening,
 171                                                 // and we skip it.
 172                                                 wfDebugLog( 'iptc', 'IPTC: Wrong count on 2:65 Software field' );
 173                                                 break;
 174                                         }
 175
 176                                         if ( isset( $parsed['2#070'] ) ) {
 177                                                 // if a version is set for the software.
 178                                                 $softwareVersion = self::convIPTC( $parsed['2#070'], $c );
 179                                                 unset( $parsed['2#070'] );
 180                                                 $data['Software'] = [ [ $software[0], $softwareVersion[0] ] ];
 181                                         } else {
 182                                                 $data['Software'] = $software;
 183                                         }
 184                                         break;
 185                                 case '2#075':
 186                                         /* Object cycle.
 187                                          * a for morning (am), p for evening, b for both */
 188                                         $data['ObjectCycle'] = self::convIPTC( $val, $c );
 189                                         break;
 190                                 case '2#100':
 191                                         /* Country/Primary location code.
 192                                          * "Indicates the code of the country/primary location where the
 193                                          * intellectual property of the objectdata was created"
 194                                          * unclear how this differs from 2#026
 195                                          */
 196                                         $data['CountryCodeDest'] = self::convIPTC( $val, $c );
 197                                         break;
 198                                 case '2#103':
 199                                         /* original transmission ref.
 200                                          * "A code representing the location of original transmission ac-
 201                                          * cording to practises of the provider."
 202                                          */
 203                                         $data['OriginalTransmissionRef'] = self::convIPTC( $val, $c );
 204                                         break;
 205                                 case '2#118': /*contact*/
 206                                         $data['Contact'] = self::convIPTC( $val, $c );
 207                                         break;
 208                                 case '2#122':
 209                                         /* Writer/Editor
 210                                          * "Identification of the name of the person involved in the writing,
 211                                          * editing or correcting the objectdata or caption/abstract."
 212                                          */
 213                                         $data['Writer'] = self::convIPTC( $val, $c );
 214                                         break;
 215                                 case '2#135': /* lang code */
 216                                         $data['LanguageCode'] = self::convIPTC( $val, $c );
 217                                         break;
 218
 219                                 // Start date stuff.
 220                                 // It doesn't accept incomplete dates even though they are valid
 221                                 // according to spec.
 222                                 // Should potentially store timezone as well.
 223                                 case '2#055':
 224                                         // Date created (not date digitized).
 225                                         // Maps to exif DateTimeOriginal
 226                                         $time = $parsed['2#060'] ?? [];
 227                                         $timestamp = self::timeHelper( $val, $time, $c );
 228                                         if ( $timestamp ) {
 229                                                 $data['DateTimeOriginal'] = $timestamp;
 230                                         }
 231                                         break;
 232
 233                                 case '2#062':
 234                                         // Date converted to digital representation.
 235                                         // Maps to exif DateTimeDigitized
 236                                         $time = $parsed['2#063'] ?? [];
 237                                         $timestamp = self::timeHelper( $val, $time, $c );
 238                                         if ( $timestamp ) {
 239                                                 $data['DateTimeDigitized'] = $timestamp;
 240                                         }
 241                                         break;
 242
 243                                 case '2#030':
 244                                         // Date released.
 245                                         $time = $parsed['2#035'] ?? [];
 246                                         $timestamp = self::timeHelper( $val, $time, $c );
 247                                         if ( $timestamp ) {
 248                                                 $data['DateTimeReleased'] = $timestamp;
 249                                         }
 250                                         break;
 251
 252                                 case '2#037':
 253                                         // Date expires.
 254                                         $time = $parsed['2#038'] ?? [];
 255                                         $timestamp = self::timeHelper( $val, $time, $c );
 256                                         if ( $timestamp ) {
 257                                                 $data['DateTimeExpires'] = $timestamp;
 258                                         }
 259                                         break;
 260
 261                                 case '2#000': /* iim version */
 262                                         // unlike other tags, this is a 2-byte binary number.
 263                                         // technically this is required if there is iptc data
 264                                         // but in practise it isn't always there.
 265                                         if ( strlen( $val[0] ) == 2 ) {
 266                                                 // if is just to be paranoid.
 267                                                 $versionValue = ord( substr( $val[0], 0, 1 ) ) * 256;
 268                                                 $versionValue += ord( substr( $val[0], 1, 1 ) );
 269                                                 $data['iimVersion'] = $versionValue;
 270                                         }
 271                                         break;
 272
 273                                 case '2#004':
 274                                         // IntellectualGenere.
 275                                         // first 4 characters are an id code
 276                                         // That we're not really interested in.
 277
 278                                         // This prop is weird, since it's
 279                                         // allowed to have multiple values
 280                                         // in iim 4.1, but not in the XMP
 281                                         // stuff. We're going to just
 282                                         // extract the first value.
 283                                         $con = self::convIPTC( $val, $c );
 284                                         if ( strlen( $con[0] ) < 5 ) {
 285                                                 wfDebugLog( 'iptc', 'IPTC: '
 286                                                         . '2:04 too short. '
 287                                                         . 'Ignoring.' );
 288                                                 break;
 289                                         }
 290                                         $extracted = substr( $con[0], 4 );
 291                                         $data['IntellectualGenre'] = $extracted;
 292                                         break;
 293
 294                                 case '2#012':
 295                                         // Subject News code - this is a compound field
 296                                         // at the moment we only extract the subject news
 297                                         // code, which is an 8 digit (ascii) number
 298                                         // describing the subject matter of the content.
 299                                         $codes = self::convIPTC( $val, $c );
 300                                         foreach ( $codes as $ic ) {
 301                                                 $fields = explode( ':', $ic, 3 );
 302
 303                                                 if ( count( $fields ) < 2 || $fields[0] !== 'IPTC' ) {
 304                                                         wfDebugLog( 'IPTC', 'IPTC: '
 305                                                                 . 'Invalid 2:12 - ' . $ic );
 306                                                         break;
 307                                                 }
 308                                                 $data['SubjectNewsCode'] = $fields[1];
 309                                         }
 310                                         break;
 311
 312                                 // purposely does not do 2:125, 2:130, 2:131,
 313                                 // 2:47, 2:50, 2:45, 2:42, 2:8, 2:3
 314                                 // 2:200, 2:201, 2:202
 315                                 // or the audio stuff (2:150 to 2:154)
 316
 317                                 case '2#070':
 318                                 case '2#060':
 319                                 case '2#063':
 320                                 case '2#085':
 321                                 case '2#038':
 322                                 case '2#035':
 323                                         // ignore. Handled elsewhere.
 324                                         break;
 325
 326                                 default:
 327                                         wfDebugLog( 'iptc', "Unsupported iptc tag: $tag. Value: " . implode( ',', $val ) );
 328                                         break;
 329                         }
 330                 }
 331
 332                 return $data;
 333         }
 334
 335         /**
 336          * Convert an iptc date and time tags into the exif format
 337          *
 338          * @todo Potentially this should also capture the timezone offset.
 339          * @param array $date The date tag
 340          * @param array $time The time tag
 341          * @param string $charset
 342          * @return string Date in EXIF format.
 343          */
 344         private static function timeHelper( $date, $time, $charset ) {
 345                 if ( count( $date ) === 1 ) {
 346                         // the standard says this should always be 1
 347                         // just double checking.
 348                         list( $date ) = self::convIPTC( $date, $charset );
 349                 } else {
 350                         return null;
 351                 }
 352
 353                 if ( count( $time ) === 1 ) {
 354                         list( $time ) = self::convIPTC( $time, $charset );
 355                         $dateOnly = false;
 356                 } else {
 357                         $time = '000000+0000'; // placeholder
 358                         $dateOnly = true;
 359                 }
 360
 361                 if ( !( preg_match( '/\d\d\d\d\d\d[-+]\d\d\d\d/', $time )
 362                         && preg_match( '/\d\d\d\d\d\d\d\d/', $date )
 363                         && substr( $date, 0, 4 ) !== '0000'
 364                         && substr( $date, 4, 2 ) !== '00'
 365                         && substr( $date, 6, 2 ) !== '00'
 366                 ) ) {
 367                         // something wrong.
 368                         // Note, this rejects some valid dates according to iptc spec
 369                         // for example: the date 00000400 means the photo was taken in
 370                         // April, but the year and day is unknown. We don't process these
 371                         // types of incomplete dates atm.
 372                         wfDebugLog( 'iptc', "IPTC: invalid time ( $time ) or date ( $date )" );
 373
 374                         return null;
 375                 }
 376
 377                 $unixTS = wfTimestamp( TS_UNIX, $date . substr( $time, 0, 6 ) );
 378                 if ( $unixTS === false ) {
 379                         wfDebugLog( 'iptc', "IPTC: can't convert date to TS_UNIX: $date $time." );
 380
 381                         return null;
 382                 }
 383
 384                 $tz = ( intval( substr( $time, 7, 2 ) ) * 60 * 60 )
 385                         + ( intval( substr( $time, 9, 2 ) ) * 60 );
 386
 387                 if ( substr( $time, 6, 1 ) === '-' ) {
 388                         $tz = -$tz;
 389                 }
 390
 391                 $finalTimestamp = wfTimestamp( TS_EXIF, $unixTS + $tz );
 392                 if ( $finalTimestamp === false ) {
 393                         wfDebugLog( 'iptc', "IPTC: can't make final timestamp. Date: " . ( $unixTS + $tz ) );
 394
 395                         return null;
 396                 }
 397                 if ( $dateOnly ) {
 398                         // return the date only
 399                         return substr( $finalTimestamp, 0, 10 );
 400                 } else {
 401                         return $finalTimestamp;
 402                 }
 403         }
 404
 405         /**
 406          * Helper function to convert charset for iptc values.
 407          * @param string|array $data The iptc string
 408          * @param string $charset
 409          *
 410          * @return string|array
 411          */
 412         private static function convIPTC( $data, $charset ) {
 413                 if ( is_array( $data ) ) {
 414                         foreach ( $data as &$val ) {
 415                                 $val = self::convIPTCHelper( $val, $charset );
 416                         }
 417                 } else {
 418                         $data = self::convIPTCHelper( $data, $charset );
 419                 }
 420
 421                 return $data;
 422         }
 423
 424         /**
 425          * Helper function of a helper function to convert charset for iptc values.
 426          * @param string|array $data The IPTC string
 427          * @param string $charset
 428          *
 429          * @return string
 430          */
 431         private static function convIPTCHelper( $data, $charset ) {
 432                 if ( $charset ) {
 433                         Wikimedia\suppressWarnings();
 434                         $data = iconv( $charset, "UTF-8//IGNORE", $data );
 435                         Wikimedia\restoreWarnings();
 436                         if ( $data === false ) {
 437                                 $data = "";
 438                                 wfDebugLog( 'iptc', __METHOD__ . " Error converting iptc data charset $charset to utf-8" );
 439                         }
 440                 } else {
 441                         // treat as utf-8 if is valid utf-8. otherwise pretend its windows-1252
 442                         // most of the time if there is no 1:90 tag, it is either ascii, latin1, or utf-8
 443                         $oldData = $data;
 444                         UtfNormal\Validator::quickIsNFCVerify( $data ); // make $data valid utf-8
 445                         if ( $data === $oldData ) {
 446                                 return $data; // if validation didn't change $data
 447                         } else {
 448                                 return self::convIPTCHelper( $oldData, 'Windows-1252' );
 449                         }
 450                 }
 451
 452                 return trim( $data );
 453         }
 454
 455         /**
 456          * take the value of 1:90 tag and returns a charset
 457          * @param string $tag 1:90 tag.
 458          * @return string Charset name or "?"
 459          * Warning, this function does not (and is not intended to) detect
 460          * all iso 2022 escape codes. In practise, the code for utf-8 is the
 461          * only code that seems to have wide use. It does detect that code.
 462          */
 463         static function getCharset( $tag ) {
 464                 // According to iim standard, charset is defined by the tag 1:90.
 465                 // in which there are iso 2022 escape sequences to specify the character set.
 466                 // the iim standard seems to encourage that all necessary escape sequences are
 467                 // in the 1:90 tag, but says it doesn't have to be.
 468
 469                 // This is in need of more testing probably. This is definitely not complete.
 470                 // however reading the docs of some other iptc software, it appears that most iptc software
 471                 // only recognizes utf-8. If 1:90 tag is not present content is
 472                 // usually ascii or iso-8859-1 (and sometimes utf-8), but no guarantee.
 473
 474                 // This also won't work if there are more than one escape sequence in the 1:90 tag
 475                 // or if something is put in the G2, or G3 charsets, etc. It will only reliably recognize utf-8.
 476
 477                 // This is just going through the charsets mentioned in appendix C of the iim standard.
 478
 479                 //  \x1b = ESC.
 480                 switch ( $tag ) {
 481                         case "\x1b%G": // utf-8
 482                         // Also call things that are compatible with utf-8, utf-8 (e.g. ascii)
 483                         case "\x1b(B": // ascii
 484                         case "\x1b(@": // iso-646-IRV (ascii in latest version, $ different in older version)
 485                                 $c = 'UTF-8';
 486                                 break;
 487                         case "\x1b(A": // like ascii, but british.
 488                                 $c = 'ISO646-GB';
 489                                 break;
 490                         case "\x1b(C": // some obscure sweedish/finland encoding
 491                                 $c = 'ISO-IR-8-1';
 492                                 break;
 493                         case "\x1b(D":
 494                                 $c = 'ISO-IR-8-2';
 495                                 break;
 496                         case "\x1b(E": // some obscure danish/norway encoding
 497                                 $c = 'ISO-IR-9-1';
 498                                 break;
 499                         case "\x1b(F":
 500                                 $c = 'ISO-IR-9-2';
 501                                 break;
 502                         case "\x1b(G":
 503                                 $c = 'SEN_850200_B'; // aka iso 646-SE; ascii-like
 504                                 break;
 505                         case "\x1b(I":
 506                                 $c = "ISO646-IT";
 507                                 break;
 508                         case "\x1b(L":
 509                                 $c = "ISO646-PT";
 510                                 break;
 511                         case "\x1b(Z":
 512                                 $c = "ISO646-ES";
 513                                 break;
 514                         case "\x1b([":
 515                                 $c = "GREEK7-OLD";
 516                                 break;
 517                         case "\x1b(K":
 518                                 $c = "ISO646-DE";
 519                                 break;
 520                         case "\x1b(N": // crylic
 521                                 $c = "ISO_5427";
 522                                 break;
 523                         case "\x1b(`": // iso646-NO
 524                                 $c = "NS_4551-1";
 525                                 break;
 526                         case "\x1b(f": // iso646-FR
 527                                 $c = "NF_Z_62-010";
 528                                 break;
 529                         case "\x1b(g":
 530                                 $c = "PT2"; // iso646-PT2
 531                                 break;
 532                         case "\x1b(h":
 533                                 $c = "ES2";
 534                                 break;
 535                         case "\x1b(i": // iso646-HU
 536                                 $c = "MSZ_7795.3";
 537                                 break;
 538                         case "\x1b(w":
 539                                 $c = "CSA_Z243.4-1985-1";
 540                                 break;
 541                         case "\x1b(x":
 542                                 $c = "CSA_Z243.4-1985-2";
 543                                 break;
 544                         case "\x1b\$(B":
 545                         case "\x1b\$B":
 546                         case "\x1b&@\x1b\$B":
 547                         case "\x1b&@\x1b\$(B":
 548                                 $c = "JIS_C6226-1983";
 549                                 break;
 550                         case "\x1b-A": // iso-8859-1. at least for the high code characters.
 551                         case "\x1b(@\x1b-A":
 552                         case "\x1b(B\x1b-A":
 553                                 $c = 'ISO-8859-1';
 554                                 break;
 555                         case "\x1b-B": // iso-8859-2. at least for the high code characters.
 556                                 $c = 'ISO-8859-2';
 557                                 break;
 558                         case "\x1b-C": // iso-8859-3. at least for the high code characters.
 559                                 $c = 'ISO-8859-3';
 560                                 break;
 561                         case "\x1b-D": // iso-8859-4. at least for the high code characters.
 562                                 $c = 'ISO-8859-4';
 563                                 break;
 564                         case "\x1b-E": // iso-8859-5. at least for the high code characters.
 565                                 $c = 'ISO-8859-5';
 566                                 break;
 567                         case "\x1b-F": // iso-8859-6. at least for the high code characters.
 568                                 $c = 'ISO-8859-6';
 569                                 break;
 570                         case "\x1b-G": // iso-8859-7. at least for the high code characters.
 571                                 $c = 'ISO-8859-7';
 572                                 break;
 573                         case "\x1b-H": // iso-8859-8. at least for the high code characters.
 574                                 $c = 'ISO-8859-8';
 575                                 break;
 576                         case "\x1b-I": // CSN_369103. at least for the high code characters.
 577                                 $c = 'CSN_369103';
 578                                 break;
 579                         default:
 580                                 wfDebugLog( 'iptc', __METHOD__ . 'Unknown charset in iptc 1:90: ' . bin2hex( $tag ) );
 581                                 // at this point just give up and refuse to parse iptc?
 582                                 $c = false;
 583                 }
 584                 return $c;
 585         }
 586 }