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