MediaWiki  master
DjVuHandler.php
Go to the documentation of this file.
1 <?php
27 
33 class DjVuHandler extends ImageHandler {
34  private const EXPENSIVE_SIZE_LIMIT = 10485760; // 10MiB
35 
36  // Constants for getHandlerState
37  private const STATE_DJVU_IMAGE = 'djvuImage';
38  private const STATE_TEXT_TREE = 'djvuTextTree';
39  private const STATE_META_TREE = 'djvuMetaTree';
40  private const CACHE_VERSION = 'v2';
41 
45  public function isEnabled() {
46  $djvuRenderer = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::DjvuRenderer );
47  $djvuDump = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::DjvuDump );
48  if ( !$djvuRenderer || !$djvuDump ) {
49  wfDebug( "DjVu is disabled, please set \$wgDjvuRenderer and \$wgDjvuDump" );
50 
51  return false;
52  } else {
53  return true;
54  }
55  }
56 
61  public function mustRender( $file ) {
62  return true;
63  }
64 
70  public function isExpensiveToThumbnail( $file ) {
71  return $file->getSize() > static::EXPENSIVE_SIZE_LIMIT;
72  }
73 
78  public function isMultiPage( $file ) {
79  return true;
80  }
81 
85  public function getParamMap() {
86  return [
87  'img_width' => 'width',
88  'img_page' => 'page',
89  ];
90  }
91 
97  public function validateParam( $name, $value ) {
98  if ( $name === 'page' && trim( $value ) !== (string)intval( $value ) ) {
99  // Extra junk on the end of page, probably actually a caption
100  // e.g. [[File:Foo.djvu|thumb|Page 3 of the document shows foo]]
101  return false;
102  }
103  return in_array( $name, [ 'width', 'height', 'page' ] ) && $value > 0;
104  }
105 
110  public function makeParamString( $params ) {
111  $page = $params['page'] ?? 1;
112  if ( !isset( $params['width'] ) ) {
113  return false;
114  }
115 
116  return "page{$page}-{$params['width']}px";
117  }
118 
123  public function parseParamString( $str ) {
124  $m = false;
125  if ( preg_match( '/^page(\d+)-(\d+)px$/', $str, $m ) ) {
126  return [ 'width' => $m[2], 'page' => $m[1] ];
127  } else {
128  return false;
129  }
130  }
131 
136  protected function getScriptParams( $params ) {
137  return [
138  'width' => $params['width'],
139  'page' => $params['page'],
140  ];
141  }
142 
151  public function doTransform( $image, $dstPath, $dstUrl, $params, $flags = 0 ) {
152  $djvuRenderer = MediaWikiServices::getInstance()->getMainConfig()->get( MainConfigNames::DjvuRenderer );
153  $djvuPostProcessor = MediaWikiServices::getInstance()->getMainConfig()
154  ->get( MainConfigNames::DjvuPostProcessor );
155  if ( !$this->normaliseParams( $image, $params ) ) {
156  return new TransformParameterError( $params );
157  }
158  $width = $params['width'];
159  $height = $params['height'];
160  $page = $params['page'];
161 
162  if ( $flags & self::TRANSFORM_LATER ) {
163  $params = [
164  'width' => $width,
165  'height' => $height,
166  'page' => $page
167  ];
168 
169  return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
170  }
171 
172  if ( !wfMkdirParents( dirname( $dstPath ), null, __METHOD__ ) ) {
173  return new MediaTransformError(
174  'thumbnail_error',
175  $width,
176  $height,
177  wfMessage( 'thumbnail_dest_directory' )
178  );
179  }
180 
181  // Get local copy source for shell scripts
182  // Thumbnail extraction is very inefficient for large files.
183  // Provide a way to pool count limit the number of downloaders.
184  if ( $image->getSize() >= 1e7 ) { // 10 MB
185  $work = new PoolCounterWorkViaCallback( 'GetLocalFileCopy', sha1( $image->getName() ),
186  [
187  'doWork' => static function () use ( $image ) {
188  return $image->getLocalRefPath();
189  }
190  ]
191  );
192  $srcPath = $work->execute();
193  } else {
194  $srcPath = $image->getLocalRefPath();
195  }
196 
197  if ( $srcPath === false ) { // Failed to get local copy
198  wfDebugLog( 'thumbnail',
199  sprintf( 'Thumbnail failed on %s: could not get local copy of "%s"',
200  wfHostname(), $image->getName() ) );
201 
202  return new MediaTransformError( 'thumbnail_error',
203  $params['width'], $params['height'],
204  wfMessage( 'filemissing' )
205  );
206  }
207 
208  # Use a subshell (brackets) to aggregate stderr from both pipeline commands
209  # before redirecting it to the overall stdout. This works in both Linux and Windows XP.
210  $cmd = '(' . Shell::escape(
211  $djvuRenderer,
212  "-format=ppm",
213  "-page={$page}",
214  "-size={$params['physicalWidth']}x{$params['physicalHeight']}",
215  $srcPath );
216  if ( $djvuPostProcessor ) {
217  $cmd .= " | {$djvuPostProcessor}";
218  }
219  $cmd .= ' > ' . Shell::escape( $dstPath ) . ') 2>&1';
220  wfDebug( __METHOD__ . ": $cmd" );
221  $retval = '';
222  $err = wfShellExec( $cmd, $retval );
223 
224  $removed = $this->removeBadFile( $dstPath, $retval );
225  if ( $retval != 0 || $removed ) {
226  $this->logErrorForExternalProcess( $retval, $err, $cmd );
227  return new MediaTransformError( 'thumbnail_error', $width, $height, $err );
228  } else {
229  $params = [
230  'width' => $width,
231  'height' => $height,
232  'page' => $page
233  ];
234 
235  return new ThumbnailImage( $image, $dstUrl, $dstPath, $params );
236  }
237  }
238 
247  private function getDjVuImage( $state, $path ) {
248  $deja = $state->getHandlerState( self::STATE_DJVU_IMAGE );
249  if ( !$deja ) {
250  $deja = new DjVuImage( $path );
251  $state->setHandlerState( self::STATE_DJVU_IMAGE, $deja );
252  }
253  return $deja;
254  }
255 
264  private function getMetadataInternal( File $file, $gettext ) {
265  $itemNames = [ 'error', '_error', 'data' ];
266  if ( $gettext ) {
267  $itemNames[] = 'text';
268  }
269  $unser = $file->getMetadataItems( $itemNames );
270 
271  if ( isset( $unser['error'] ) ) {
272  return false;
273  } elseif ( isset( $unser['_error'] ) ) {
274  return false;
275  } else {
276  return $unser;
277  }
278  }
279 
286  public function getMetaTree( $image, $gettext = false ) {
287  if ( $gettext && $image->getHandlerState( self::STATE_TEXT_TREE ) ) {
288  return $image->getHandlerState( self::STATE_TEXT_TREE );
289  }
290  if ( !$gettext && $image->getHandlerState( self::STATE_META_TREE ) ) {
291  return $image->getHandlerState( self::STATE_META_TREE );
292  }
293 
294  $metadata = $this->getMetadataInternal( $image, $gettext );
295  if ( !$metadata ) {
296  return false;
297  }
298 
299  if ( $gettext ) {
300  return $metadata;
301  } else {
302  unset( $metadata['text'] );
303  return $metadata;
304  }
305  }
306 
307  public function getThumbType( $ext, $mime, $params = null ) {
308  $djvuOutputExtension = MediaWikiServices::getInstance()->getMainConfig()
309  ->get( MainConfigNames::DjvuOutputExtension );
310  static $mime;
311  if ( !isset( $mime ) ) {
312  $magic = MediaWikiServices::getInstance()->getMimeAnalyzer();
313  $mime = $magic->getMimeTypeFromExtensionOrNull( $djvuOutputExtension );
314  }
315 
316  return [ $djvuOutputExtension, $mime ];
317  }
318 
319  public function getSizeAndMetadata( $state, $path ) {
320  wfDebug( "Getting DjVu metadata for $path" );
321 
322  $djvuImage = $this->getDjVuImage( $state, $path );
323  $metadata = $djvuImage->retrieveMetaData();
324  if ( $metadata === false ) {
325  // Special value so that we don't repetitively try and decode a broken file.
326  $metadata = [ 'error' => 'Error extracting metadata' ];
327  }
328  return [ 'metadata' => $metadata ] + $djvuImage->getImageSize();
329  }
330 
331  public function getMetadataType( $image ) {
332  // historical reasons
333  return 'djvuxml';
334  }
335 
336  public function isFileMetadataValid( $image ) {
337  return $image->getMetadataArray() ? self::METADATA_GOOD : self::METADATA_BAD;
338  }
339 
340  public function pageCount( File $image ) {
341  $info = $this->getDimensionInfo( $image );
342 
343  return $info ? $info['pageCount'] : false;
344  }
345 
346  public function getPageDimensions( File $image, $page ) {
347  $index = $page - 1; // MW starts pages at 1
348 
349  $info = $this->getDimensionInfo( $image );
350  if ( $info && isset( $info['dimensionsByPage'][$index] ) ) {
351  return $info['dimensionsByPage'][$index];
352  }
353 
354  return false;
355  }
356 
357  protected function getDimensionInfo( File $file ) {
358  $cache = MediaWikiServices::getInstance()->getMainWANObjectCache();
359  return $cache->getWithSetCallback(
360  $cache->makeKey( 'file-djvu', 'dimensions', self::CACHE_VERSION, $file->getSha1() ),
361  $cache::TTL_INDEFINITE,
362  function () use ( $file ) {
363  $tree = $this->getMetaTree( $file );
364  return $this->getDimensionInfoFromMetaTree( $tree );
365  },
366  [ 'pcTTL' => $cache::TTL_INDEFINITE ]
367  );
368  }
369 
375  protected function getDimensionInfoFromMetaTree( $metatree ) {
376  if ( !$metatree ) {
377  return false;
378  }
379  $dimsByPage = [];
380 
381  if ( !isset( $metatree['data'] ) || !$metatree['data'] ) {
382  return false;
383  }
384  foreach ( $metatree['data']['pages'] as $page ) {
385  if ( !$page ) {
386  $dimsByPage[] = false;
387  } else {
388  $dimsByPage[] = [
389  'width' => (int)$page['width'],
390  'height' => (int)$page['height'],
391  ];
392  }
393  }
394  return [
395  'pageCount' => count( $metatree['data']['pages'] ),
396  'dimensionsByPage' => $dimsByPage
397  ];
398  }
399 
405  public function getPageText( File $image, $page ) {
406  $tree = $this->getMetaTree( $image, true );
407  if ( !$tree ) {
408  return false;
409  }
410  if ( isset( $tree['text'] ) && isset( $tree['text'][$page - 1] ) ) {
411  return $tree['text'][$page - 1];
412  }
413  return false;
414  }
415 
416  public function useSplitMetadata() {
417  return true;
418  }
419 }
wfDebug( $text, $dest='all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
wfShellExec( $cmd, &$retval=null, $environ=[], $limits=[], $options=[])
Execute a shell command, with time and memory limits mirrored from the PHP configuration if supported...
wfHostname()
Get host name of the current machine, for use in error reporting.
wfDebugLog( $logGroup, $text, $dest='all', array $context=[])
Send a line to a supplementary debug log file, if configured, or main debug log if not.
wfMkdirParents( $dir, $mode=null, $caller=null)
Make directory, and make all parent directories if they don't exist.
wfMessage( $key,... $params)
This is the function for getting translated interface messages.
Handler for DjVu images.
Definition: DjVuHandler.php:33
getDimensionInfoFromMetaTree( $metatree)
Given the metadata, returns dimension information about the document.
makeParamString( $params)
isExpensiveToThumbnail( $file)
True if creating thumbnails from the file is large or otherwise resource-intensive.
Definition: DjVuHandler.php:70
validateParam( $name, $value)
Definition: DjVuHandler.php:97
isMultiPage( $file)
Definition: DjVuHandler.php:78
getDimensionInfo(File $file)
getScriptParams( $params)
doTransform( $image, $dstPath, $dstUrl, $params, $flags=0)
pageCount(File $image)
Page count for a multi-page document, false if unsupported or unknown.
getPageText(File $image, $page)
isFileMetadataValid( $image)
Check if the metadata is valid for this handler.
getSizeAndMetadata( $state, $path)
Get image size information and metadata array.
getPageDimensions(File $image, $page)
Get an associative array of page dimensions Currently "width" and "height" are understood,...
getMetadataType( $image)
Get a string describing the type of metadata, for display purposes.
getMetaTree( $image, $gettext=false)
Cache a document tree for the DjVu metadata.
getThumbType( $ext, $mime, $params=null)
Get the thumbnail extension and MIME type for a given source MIME type.
useSplitMetadata()
If this returns true, LocalFile may split metadata up and store its constituent items separately.
mustRender( $file)
Definition: DjVuHandler.php:61
parseParamString( $str)
Support for detecting/validating DjVu image files and getting some basic file metadata (resolution et...
Definition: DjVuImage.php:41
Implements some public methods and some protected utility functions which are required by multiple ch...
Definition: File.php:68
Media handler abstract base class for images.
normaliseParams( $image, &$params)
Changes the parameter array as necessary, ready for transformation.Should be idempotent....
const METADATA_BAD
logErrorForExternalProcess( $retval, $err, $cmd)
Log an error that occurred in an external process.
const METADATA_GOOD
removeBadFile( $dstPath, $retval=0)
Check for zero-sized thumbnails.
Basic media transform error class.
A class containing constants representing the names of configuration variables.
Service locator for MediaWiki core services.
Executes shell commands.
Definition: Shell.php:46
Convenience class for dealing with PoolCounters using callbacks.
execute( $skipcache=false)
Get the result of the work (whatever it is), or the result of the error() function.
Media transform output for images.
Shortcut class for parameter validation errors.
$cache
Definition: mcc.php:33
$mime
Definition: router.php:60
if(PHP_SAPI !='cli-server') if(!isset( $_SERVER['SCRIPT_FILENAME'])) $file
Item class for a filearchive table row.
Definition: router.php:42
if(!is_readable( $file)) $ext
Definition: router.php:48