| Index: trunk/extensions/PagedTiffHandler/PagedTiffHandler.image.php |
| — | — | @@ -186,31 +186,6 @@ |
| 187 | 187 | return $this->_meta; |
| 188 | 188 | } |
| 189 | 189 | |
| 190 | | - private function addPageEntry( &$entry, &$metadata, &$prevPage ) { |
| 191 | | - if ( !isset( $entry['page'] ) ) { |
| 192 | | - $entry['page'] = $prevPage +1; |
| 193 | | - } else { |
| 194 | | - if ( $prevPage >= $entry['page'] ) { |
| 195 | | - $metadata['errors'][] = "inconsistent page numbering in TIFF directory"; |
| 196 | | - return false; |
| 197 | | - } |
| 198 | | - } |
| 199 | | - |
| 200 | | - if ( isset( $entry['width'] ) && isset( $entry['height'] ) ) { |
| 201 | | - $prevPage = max($prevPage, $entry['page']); |
| 202 | | - |
| 203 | | - if ( !isset( $entry['alpha'] ) ) { |
| 204 | | - $entry['alpha'] = 'false'; |
| 205 | | - } |
| 206 | | - |
| 207 | | - $entry['pixels'] = $entry['height'] * $entry['width']; |
| 208 | | - $metadata['page_data'][$entry['page']] = $entry; |
| 209 | | - } |
| 210 | | - |
| 211 | | - $entry = array(); |
| 212 | | - return true; |
| 213 | | - } |
| 214 | | - |
| 215 | 190 | /** |
| 216 | 191 | * helper function of retrieveMetaData(). |
| 217 | 192 | * parses shell return from tiffinfo-command into an array. |
| — | — | @@ -221,27 +196,25 @@ |
| 222 | 197 | $dump = preg_replace( '/ Image Length:/', "\n Image Length:", $dump ); #HACK: width and length are given on a single line... |
| 223 | 198 | $rows = preg_split('/[\r\n]+\s*/', $dump); |
| 224 | 199 | |
| 225 | | - $data = array(); |
| 226 | | - $data['page_data'] = array(); |
| | 200 | + $state = new PagedTiffInfoParserState(); |
| 227 | 201 | |
| 228 | 202 | $ignoreIFDs = array(); |
| 229 | | - $entry = array(); |
| 230 | | - |
| 231 | 203 | $ignore = false; |
| 232 | | - $prevPage = 0; |
| 233 | 204 | |
| 234 | 205 | foreach ( $rows as $row ) { |
| 235 | 206 | $row = trim( $row ); |
| 236 | 207 | |
| 237 | | - if ( preg_match('/^<|^$/', $row) ) { |
| | 208 | + # ignore XML rows |
| | 209 | + if ( preg_match('/^<|^$/', $row) ) { |
| 238 | 210 | continue; |
| 239 | 211 | } |
| 240 | 212 | |
| 241 | 213 | $error = false; |
| 242 | 214 | |
| | 215 | + # handle fatal errors |
| 243 | 216 | foreach ( $wgTiffTiffinfoRejectMessages as $pattern ) { |
| 244 | 217 | if ( preg_match( $pattern, trim( $row ) ) ) { |
| 245 | | - $data['errors'][] = $row; |
| | 218 | + $state->addError( $row ); |
| 246 | 219 | $error = true; |
| 247 | 220 | break; |
| 248 | 221 | } |
| — | — | @@ -250,19 +223,25 @@ |
| 251 | 224 | if ( $error ) continue; |
| 252 | 225 | |
| 253 | 226 | if ( preg_match('/^TIFF Directory at offset 0x[a-f0-9]+ \((\d+)\)/', $row, $m) ) { |
| | 227 | + # new IFD starting, flush previous page |
| | 228 | + |
| 254 | 229 | if ( $ignore ) { |
| 255 | | - $entry = array(); |
| 256 | | - } else if ( $entry ) { |
| 257 | | - $ok = $this->addPageEntry($entry, $data, $prevPage); |
| | 230 | + $state->resetPage(); |
| | 231 | + } else { |
| | 232 | + $ok = $state->finishPage(); |
| | 233 | + |
| 258 | 234 | if ( !$ok ) { |
| 259 | 235 | $error = true; |
| 260 | 236 | continue; |
| 261 | 237 | } |
| 262 | 238 | } |
| 263 | 239 | |
| | 240 | + # check if the next IFD is to be ignored |
| 264 | 241 | $offset = (int)$m[1]; |
| 265 | 242 | $ignore = !empty( $ignoreIFDs[ $offset ] ); |
| 266 | 243 | } else if ( preg_match('#^(TIFF.*?Directory): (.*?/.*?): (.*)#i', $row, $m) ) { |
| | 244 | + # handle warnings |
| | 245 | + |
| 267 | 246 | $bypass = false; |
| 268 | 247 | $msg = $m[3]; |
| 269 | 248 | |
| — | — | @@ -274,23 +253,25 @@ |
| 275 | 254 | } |
| 276 | 255 | |
| 277 | 256 | if ( !$bypass ) { |
| 278 | | - $data['warnings'][] = $msg; |
| | 257 | + $state->addWarning( $msg ); |
| 279 | 258 | } |
| 280 | 259 | } else if ( preg_match('/^\s*(.*?)\s*:\s*(.*?)\s*$/', $row, $m) ) { |
| | 260 | + # handle key/value pair |
| | 261 | + |
| 281 | 262 | $key = $m[1]; |
| 282 | 263 | $value = $m[2]; |
| 283 | 264 | |
| 284 | 265 | if ( $key == 'Page Number' && preg_match('/(\d+)-(\d+)/', $value, $m) ) { |
| 285 | | - $data['page_amount'] = (int)$m[2]; |
| 286 | | - $entry['page'] = (int)$m[1] +1; |
| | 266 | + $state->setFileProperty('page_amount', (int)$m[2]); |
| | 267 | + $state->setPageProperty('page', (int)$m[1] +1); |
| 287 | 268 | } else if ( $key == 'Samples/Pixel' ) { |
| 288 | | - if ($value == '4') $entry['alpha'] = 'true'; |
| | 269 | + if ($value == '4') $state->setPageProperty('alpha', 'true'); |
| 289 | 270 | } else if ( $key == 'Extra samples' ) { |
| 290 | | - if (preg_match('.*alpha.*', $value)) $entry['alpha'] = 'true'; |
| | 271 | + if (preg_match('.*alpha.*', $value)) $state->setPageProperty('alpha', 'true'); |
| 291 | 272 | } else if ( $key == 'Image Width' || $key == 'PixelXDimension' ) { |
| 292 | | - $entry['width'] = (int)$value; |
| | 273 | + $state->setPageProperty('width', (int)$value); |
| 293 | 274 | } else if ( $key == 'Image Length' || $key == 'PixelYDimension' ) { |
| 294 | | - $entry['height'] = (int)$value; |
| | 275 | + $state->setPageProperty('height', (int)$value); |
| 295 | 276 | } else if ( preg_match('/.*IFDOffset/', $key) ) { |
| 296 | 277 | # ignore extra IFDs, see <http://www.awaresystems.be/imaging/tiff/tifftags/exififd.html> |
| 297 | 278 | # Note: we assume that we will always see the reference before the actual IFD, so we know which IFDs to ignore |
| — | — | @@ -303,18 +284,9 @@ |
| 304 | 285 | |
| 305 | 286 | } |
| 306 | 287 | |
| 307 | | - if ( $entry && !$ignore ) { |
| 308 | | - $ok = $this->addPageEntry($entry, $data, $prevPage); |
| 309 | | - } |
| | 288 | + $state->finish( !$ignore ); |
| 310 | 289 | |
| 311 | | - if ( !isset( $data['page_amount'] ) ) { |
| 312 | | - $data['page_amount'] = count( $data['page_data'] ); |
| 313 | | - } |
| 314 | | - |
| 315 | | - if ( ! $data['page_data'] ) { |
| 316 | | - $data['errors'][] = 'no page data found in tiff directory!'; |
| 317 | | - } |
| 318 | | - |
| | 290 | + $data = $state->getMetadata(); |
| 319 | 291 | return $data; |
| 320 | 292 | } |
| 321 | 293 | |
| — | — | @@ -342,18 +314,17 @@ |
| 343 | 315 | protected function parseIdentifyOutput( $dump ) { |
| 344 | 316 | global $wgTiffIdentifyRejectMessages, $wgTiffIdentifyBypassMessages; |
| 345 | 317 | |
| 346 | | - $data = array(); |
| | 318 | + $state = new PagedTiffInfoParserState(); |
| | 319 | + |
| 347 | 320 | if ( strval( $dump ) == '' ) { |
| 348 | | - $data['errors'][] = "no metadata"; |
| 349 | | - return $data; |
| | 321 | + $state->addError( "no metadata" ); |
| | 322 | + return $state->getMetadata(); |
| 350 | 323 | } |
| 351 | 324 | |
| 352 | 325 | $infos = null; |
| 353 | 326 | preg_match_all( '/\[BEGIN\](.+?)\[END\]/si', $dump, $infos, PREG_SET_ORDER ); |
| 354 | | - $data['page_amount'] = count( $infos ); |
| 355 | | - $data['page_data'] = array(); |
| 356 | 327 | foreach ( $infos as $info ) { |
| 357 | | - $entry = array(); |
| | 328 | + $state->resetPage(); |
| 358 | 329 | $lines = explode( "\n", $info[1] ); |
| 359 | 330 | foreach ( $lines as $line ) { |
| 360 | 331 | if ( trim( $line ) == '' ) { |
| — | — | @@ -363,22 +334,21 @@ |
| 364 | 335 | if ( trim( $parts[0] ) == 'alpha' && trim( $parts[1] ) == '%A' ) { |
| 365 | 336 | continue; |
| 366 | 337 | } |
| 367 | | - if ( trim( $parts[0] ) == 'alpha2' && !isset( $entry['alpha'] ) ) { |
| | 338 | + if ( trim( $parts[0] ) == 'alpha2' && !$state->hasPageProperty( 'alpha' ) ) { |
| 368 | 339 | switch( trim( $parts[1] ) ) { |
| 369 | 340 | case 'DirectClassRGBMatte': |
| 370 | 341 | case 'DirectClassRGBA': |
| 371 | | - $entry['alpha'] = 'true'; |
| | 342 | + $state->setPageProperty('alpha', 'true'); |
| 372 | 343 | break; |
| 373 | 344 | default: |
| 374 | | - $entry['alpha'] = 'false'; |
| | 345 | + $state->setPageProperty('alpha', 'false'); |
| 375 | 346 | break; |
| 376 | 347 | } |
| 377 | 348 | continue; |
| 378 | 349 | } |
| 379 | | - $entry[trim( $parts[0] )] = trim( $parts[1] ); |
| | 350 | + $state->setPageProperty( trim( $parts[0] ), trim( $parts[1] ) ); |
| 380 | 351 | } |
| 381 | | - $entry['pixels'] = $entry['height'] * $entry['width']; |
| 382 | | - $data['page_data'][$entry['page']] = $entry; |
| | 352 | + $state->finishPage(); |
| 383 | 353 | } |
| 384 | 354 | |
| 385 | 355 | |
| — | — | @@ -393,7 +363,7 @@ |
| 394 | 364 | $knownError = false; |
| 395 | 365 | foreach ( $wgTiffIdentifyRejectMessages as $msg ) { |
| 396 | 366 | if ( preg_match( $msg, trim( $error ) ) ) { |
| 397 | | - $data['errors'][] = $error; |
| | 367 | + $state->addError( $error ); |
| 398 | 368 | $knownError = true; |
| 399 | 369 | break; |
| 400 | 370 | } |
| — | — | @@ -402,17 +372,109 @@ |
| 403 | 373 | // ignore messages that match $wgTiffIdentifyBypassMessages |
| 404 | 374 | foreach ( $wgTiffIdentifyBypassMessages as $msg ) { |
| 405 | 375 | if ( preg_match( $msg, trim( $error ) ) ) { |
| 406 | | - // $data['warnings'][] = $error; |
| 407 | 376 | $knownError = true; |
| 408 | 377 | break; |
| 409 | 378 | } |
| 410 | 379 | } |
| 411 | 380 | } |
| 412 | 381 | if ( !$knownError ) { |
| 413 | | - $data['warnings'][] = $error; |
| | 382 | + $state->addWarning( $error ); |
| 414 | 383 | } |
| 415 | 384 | } |
| 416 | 385 | } |
| | 386 | + |
| | 387 | + $state->finish(); |
| | 388 | + |
| | 389 | + $data = $state->getMetadata(); |
| 417 | 390 | return $data; |
| 418 | 391 | } |
| 419 | 392 | } |
| | 393 | + |
| | 394 | +class PagedTiffInfoParserState { |
| | 395 | + var $metadata; # all data |
| | 396 | + var $page; # current page |
| | 397 | + var $prevPage; |
| | 398 | + |
| | 399 | + function __construct() { |
| | 400 | + $this->metadata = array(); |
| | 401 | + $this->page = array(); |
| | 402 | + $this->prevPage = 0; |
| | 403 | + |
| | 404 | + $this->metadata['page_data'] = array(); |
| | 405 | + } |
| | 406 | + |
| | 407 | + function finish( $finishPage = true ) { |
| | 408 | + if ( $finishPage ) { |
| | 409 | + $this->finishPage( ); |
| | 410 | + } |
| | 411 | + |
| | 412 | + if ( !isset( $this->metadata['page_amount'] ) ) { |
| | 413 | + $this->metadata['page_amount'] = count( $this->metadata['page_data'] ); |
| | 414 | + } |
| | 415 | + |
| | 416 | + if ( ! $this->metadata['page_data'] ) { |
| | 417 | + $this->metadata['errors'][] = 'no page data found in tiff directory!'; |
| | 418 | + } |
| | 419 | + } |
| | 420 | + |
| | 421 | + function resetPage( ) { |
| | 422 | + $this->page = array(); |
| | 423 | + } |
| | 424 | + |
| | 425 | + function finishPage( ) { |
| | 426 | + if ( !isset( $this->page['page'] ) ) { |
| | 427 | + $this->page['page'] = $this->prevPage +1; |
| | 428 | + } else { |
| | 429 | + if ( $this->prevPage >= $this->page['page'] ) { |
| | 430 | + $this->metadata['errors'][] = "inconsistent page numbering in TIFF directory"; |
| | 431 | + return false; |
| | 432 | + } |
| | 433 | + } |
| | 434 | + |
| | 435 | + if ( isset( $this->page['width'] ) && isset( $this->page['height'] ) ) { |
| | 436 | + $this->prevPage = max($this->prevPage, $this->page['page']); |
| | 437 | + |
| | 438 | + if ( !isset( $this->page['alpha'] ) ) { |
| | 439 | + $this->page['alpha'] = 'false'; |
| | 440 | + } |
| | 441 | + |
| | 442 | + $this->page['pixels'] = $this->page['height'] * $this->page['width']; |
| | 443 | + $this->metadata['page_data'][$this->page['page']] = $this->page; |
| | 444 | + } |
| | 445 | + |
| | 446 | + $this->page = array(); |
| | 447 | + return true; |
| | 448 | + } |
| | 449 | + |
| | 450 | + function setPageProperty( $key, $value ) { |
| | 451 | + $this->page[$key] = $value; |
| | 452 | + } |
| | 453 | + |
| | 454 | + function hasPageProperty( $key ) { |
| | 455 | + return isset( $this->page[$key] ) && ! is_null( $this->page[$key] ); |
| | 456 | + } |
| | 457 | + |
| | 458 | + function setFileProperty( $key, $value ) { |
| | 459 | + $this->metadata[$key] = $value; |
| | 460 | + } |
| | 461 | + |
| | 462 | + function hasFileProperty( $key, $value ) { |
| | 463 | + return isset( $this->metadata[$key] ) && ! is_null( $this->metadata[$key] ); |
| | 464 | + } |
| | 465 | + |
| | 466 | + function addError( $message ) { |
| | 467 | + $this->metadata['errors'][] = $message; |
| | 468 | + } |
| | 469 | + |
| | 470 | + function addWarning( $message ) { |
| | 471 | + $this->metadata['warnings'][] = $message; |
| | 472 | + } |
| | 473 | + |
| | 474 | + function getMetadata( ) { |
| | 475 | + return $this->metadata; |
| | 476 | + } |
| | 477 | + |
| | 478 | + function hasErrors() { |
| | 479 | + return !empty( $this->metadata['errors'] ); |
| | 480 | + } |
| | 481 | +} |
| \ No newline at end of file |