MediaWiki  master
DjVuImage.php
Go to the documentation of this file.
1 <?php
28 
38 class DjVuImage {
39 
43  const DJVUTXT_MEMORY_LIMIT = 300000;
44 
46  private $mFilename;
47 
51  function __construct( $filename ) {
52  $this->mFilename = $filename;
53  }
54 
59  public function isValid() {
60  $info = $this->getInfo();
61 
62  return $info !== false;
63  }
64 
69  public function getImageSize() {
70  $data = $this->getInfo();
71 
72  if ( $data !== false ) {
73  $width = $data['width'];
74  $height = $data['height'];
75 
76  return [ $width, $height, 'DjVu',
77  "width=\"$width\" height=\"$height\"" ];
78  }
79 
80  return false;
81  }
82 
83  // ---------
84 
88  function dump() {
89  $file = fopen( $this->mFilename, 'rb' );
90  $header = fread( $file, 12 );
91  $arr = unpack( 'a4magic/a4chunk/NchunkLength', $header );
92  $chunk = $arr['chunk'];
93  $chunkLength = $arr['chunkLength'];
94  echo "$chunk $chunkLength\n";
95  $this->dumpForm( $file, $chunkLength, 1 );
96  fclose( $file );
97  }
98 
99  private function dumpForm( $file, $length, $indent ) {
100  $start = ftell( $file );
101  $secondary = fread( $file, 4 );
102  echo str_repeat( ' ', $indent * 4 ) . "($secondary)\n";
103  while ( ftell( $file ) - $start < $length ) {
104  $chunkHeader = fread( $file, 8 );
105  if ( $chunkHeader == '' ) {
106  break;
107  }
108  $arr = unpack( 'a4chunk/NchunkLength', $chunkHeader );
109  $chunk = $arr['chunk'];
110  $chunkLength = $arr['chunkLength'];
111  echo str_repeat( ' ', $indent * 4 ) . "$chunk $chunkLength\n";
112 
113  if ( $chunk == 'FORM' ) {
114  $this->dumpForm( $file, $chunkLength, $indent + 1 );
115  } else {
116  fseek( $file, $chunkLength, SEEK_CUR );
117  if ( $chunkLength & 1 ) {
118  // Padding byte between chunks
119  fseek( $file, 1, SEEK_CUR );
120  }
121  }
122  }
123  }
124 
125  function getInfo() {
126  Wikimedia\suppressWarnings();
127  $file = fopen( $this->mFilename, 'rb' );
128  Wikimedia\restoreWarnings();
129  if ( $file === false ) {
130  wfDebug( __METHOD__ . ": missing or failed file read\n" );
131 
132  return false;
133  }
134 
135  $header = fread( $file, 16 );
136  $info = false;
137 
138  if ( strlen( $header ) < 16 ) {
139  wfDebug( __METHOD__ . ": too short file header\n" );
140  } else {
141  $arr = unpack( 'a4magic/a4form/NformLength/a4subtype', $header );
142 
143  $subtype = $arr['subtype'];
144  if ( $arr['magic'] != 'AT&T' ) {
145  wfDebug( __METHOD__ . ": not a DjVu file\n" );
146  } elseif ( $subtype == 'DJVU' ) {
147  // Single-page document
148  $info = $this->getPageInfo( $file );
149  } elseif ( $subtype == 'DJVM' ) {
150  // Multi-page document
151  $info = $this->getMultiPageInfo( $file, $arr['formLength'] );
152  } else {
153  wfDebug( __METHOD__ . ": unrecognized DJVU file type '{$arr['subtype']}'\n" );
154  }
155  }
156  fclose( $file );
157 
158  return $info;
159  }
160 
161  private function readChunk( $file ) {
162  $header = fread( $file, 8 );
163  if ( strlen( $header ) < 8 ) {
164  return [ false, 0 ];
165  } else {
166  $arr = unpack( 'a4chunk/Nlength', $header );
167 
168  return [ $arr['chunk'], $arr['length'] ];
169  }
170  }
171 
172  private function skipChunk( $file, $chunkLength ) {
173  fseek( $file, $chunkLength, SEEK_CUR );
174 
175  if ( ( $chunkLength & 1 ) && !feof( $file ) ) {
176  // padding byte
177  fseek( $file, 1, SEEK_CUR );
178  }
179  }
180 
181  private function getMultiPageInfo( $file, $formLength ) {
182  // For now, we'll just look for the first page in the file
183  // and report its information, hoping others are the same size.
184  $start = ftell( $file );
185  do {
186  list( $chunk, $length ) = $this->readChunk( $file );
187  if ( !$chunk ) {
188  break;
189  }
190 
191  if ( $chunk == 'FORM' ) {
192  $subtype = fread( $file, 4 );
193  if ( $subtype == 'DJVU' ) {
194  wfDebug( __METHOD__ . ": found first subpage\n" );
195 
196  return $this->getPageInfo( $file );
197  }
198  $this->skipChunk( $file, $length - 4 );
199  } else {
200  wfDebug( __METHOD__ . ": skipping '$chunk' chunk\n" );
201  $this->skipChunk( $file, $length );
202  }
203  } while ( $length != 0 && !feof( $file ) && ftell( $file ) - $start < $formLength );
204 
205  wfDebug( __METHOD__ . ": multi-page DJVU file contained no pages\n" );
206 
207  return false;
208  }
209 
210  private function getPageInfo( $file ) {
211  list( $chunk, $length ) = $this->readChunk( $file );
212  if ( $chunk != 'INFO' ) {
213  wfDebug( __METHOD__ . ": expected INFO chunk, got '$chunk'\n" );
214 
215  return false;
216  }
217 
218  if ( $length < 9 ) {
219  wfDebug( __METHOD__ . ": INFO should be 9 or 10 bytes, found $length\n" );
220 
221  return false;
222  }
223  $data = fread( $file, $length );
224  if ( strlen( $data ) < $length ) {
225  wfDebug( __METHOD__ . ": INFO chunk cut off\n" );
226 
227  return false;
228  }
229 
230  $arr = unpack(
231  'nwidth/' .
232  'nheight/' .
233  'Cminor/' .
234  'Cmajor/' .
235  'vresolution/' .
236  'Cgamma', $data );
237 
238  # Newer files have rotation info in byte 10, but we don't use it yet.
239 
240  return [
241  'width' => $arr['width'],
242  'height' => $arr['height'],
243  'version' => "{$arr['major']}.{$arr['minor']}",
244  'resolution' => $arr['resolution'],
245  'gamma' => $arr['gamma'] / 10.0 ];
246  }
247 
252  function retrieveMetaData() {
254 
255  if ( !$this->isValid() ) {
256  return false;
257  }
258 
259  if ( isset( $wgDjvuDump ) ) {
260  # djvudump is faster as of version 3.5
261  # https://sourceforge.net/p/djvu/bugs/71/
262  $cmd = Shell::escape( $wgDjvuDump ) . ' ' . Shell::escape( $this->mFilename );
263  $dump = wfShellExec( $cmd );
264  $xml = $this->convertDumpToXML( $dump );
265  } elseif ( isset( $wgDjvuToXML ) ) {
266  $cmd = Shell::escape( $wgDjvuToXML ) . ' --without-anno --without-text ' .
267  Shell::escape( $this->mFilename );
268  $xml = wfShellExec( $cmd );
269  } else {
270  $xml = null;
271  }
272  # Text layer
273  if ( isset( $wgDjvuTxt ) ) {
274  $cmd = Shell::escape( $wgDjvuTxt ) . ' --detail=page ' . Shell::escape( $this->mFilename );
275  wfDebug( __METHOD__ . ": $cmd\n" );
276  $retval = '';
277  $txt = wfShellExec( $cmd, $retval, [], [ 'memory' => self::DJVUTXT_MEMORY_LIMIT ] );
278  if ( $retval == 0 ) {
279  # Strip some control characters
280  $txt = preg_replace( "/[\013\035\037]/", "", $txt );
281  $reg = <<<EOR
282  /\(page\s[\d-]*\s[\d-]*\s[\d-]*\s[\d-]*\s*"
283  ((?> # Text to match is composed of atoms of either:
284  \\\\. # - any escaped character
285  | # - any character different from " and \
286  [^"\\\\]+
287  )*?)
288  "\s*\)
289  | # Or page can be empty ; in this case, djvutxt dumps ()
290  \(\s*()\)/sx
291 EOR;
292  $txt = preg_replace_callback( $reg, [ $this, 'pageTextCallback' ], $txt );
293  $txt = "<DjVuTxt>\n<HEAD></HEAD>\n<BODY>\n" . $txt . "</BODY>\n</DjVuTxt>\n";
294  $xml = preg_replace( "/<DjVuXML>/", "<mw-djvu><DjVuXML>", $xml, 1 ) .
295  $txt .
296  '</mw-djvu>';
297  }
298  }
299 
300  return $xml;
301  }
302 
304  # Get rid of invalid UTF-8, strip control characters
305  $val = htmlspecialchars( UtfNormal\Validator::cleanUp( stripcslashes( $matches[1] ) ) );
306  $val = str_replace( [ "\n", '�' ], [ '&#10;', '' ], $val );
307  return '<PAGE value="' . $val . '" />';
308  }
309 
315  function convertDumpToXML( $dump ) {
316  if ( strval( $dump ) == '' ) {
317  return false;
318  }
319 
320  $xml = <<<EOT
321 <?xml version="1.0" ?>
322 <!DOCTYPE DjVuXML PUBLIC "-//W3C//DTD DjVuXML 1.1//EN" "pubtext/DjVuXML-s.dtd">
323 <DjVuXML>
324 <HEAD></HEAD>
325 <BODY>
326 EOT;
327 
328  $dump = str_replace( "\r", '', $dump );
329  $line = strtok( $dump, "\n" );
330  $m = false;
331  $good = false;
332  if ( preg_match( '/^( *)FORM:DJVU/', $line, $m ) ) {
333  # Single-page
334  if ( $this->parseFormDjvu( $line, $xml ) ) {
335  $good = true;
336  } else {
337  return false;
338  }
339  } elseif ( preg_match( '/^( *)FORM:DJVM/', $line, $m ) ) {
340  # Multi-page
341  $parentLevel = strlen( $m[1] );
342  # Find DIRM
343  $line = strtok( "\n" );
344  while ( $line !== false ) {
345  $childLevel = strspn( $line, ' ' );
346  if ( $childLevel <= $parentLevel ) {
347  # End of chunk
348  break;
349  }
350 
351  if ( preg_match( '/^ *DIRM.*indirect/', $line ) ) {
352  wfDebug( "Indirect multi-page DjVu document, bad for server!\n" );
353 
354  return false;
355  }
356  if ( preg_match( '/^ *FORM:DJVU/', $line ) ) {
357  # Found page
358  if ( $this->parseFormDjvu( $line, $xml ) ) {
359  $good = true;
360  } else {
361  return false;
362  }
363  }
364  $line = strtok( "\n" );
365  }
366  }
367  if ( !$good ) {
368  return false;
369  }
370 
371  $xml .= "</BODY>\n</DjVuXML>\n";
372 
373  return $xml;
374  }
375 
376  function parseFormDjvu( $line, &$xml ) {
377  $parentLevel = strspn( $line, ' ' );
378  $line = strtok( "\n" );
379 
380  # Find INFO
381  while ( $line !== false ) {
382  $childLevel = strspn( $line, ' ' );
383  if ( $childLevel <= $parentLevel ) {
384  # End of chunk
385  break;
386  }
387 
388  if ( preg_match(
389  '/^ *INFO *\[\d*\] *DjVu *(\d+)x(\d+), *\w*, *(\d+) *dpi, *gamma=([0-9.-]+)/',
390  $line,
391  $m
392  ) ) {
393  $xml .= Xml::tags(
394  'OBJECT',
395  [
396  # 'data' => '',
397  # 'type' => 'image/x.djvu',
398  'height' => $m[2],
399  'width' => $m[1],
400  # 'usemap' => '',
401  ],
402  "\n" .
403  Xml::element( 'PARAM', [ 'name' => 'DPI', 'value' => $m[3] ] ) . "\n" .
404  Xml::element( 'PARAM', [ 'name' => 'GAMMA', 'value' => $m[4] ] ) . "\n"
405  ) . "\n";
406 
407  return true;
408  }
409  $line = strtok( "\n" );
410  }
411 
412  # Not found
413  return false;
414  }
415 }
if(PHP_SAPI !='cli-server') if(!isset( $_SERVER['SCRIPT_FILENAME'])) $file
Item class for a filearchive table row.
Definition: router.php:42
pageTextCallback( $matches)
Definition: DjVuImage.php:303
parseFormDjvu( $line, &$xml)
Definition: DjVuImage.php:376
$wgDjvuTxt
Path of the djvutxt DJVU text extraction utility Enable this and $wgDjvuDump to enable text layer ext...
dumpForm( $file, $length, $indent)
Definition: DjVuImage.php:99
getMultiPageInfo( $file, $formLength)
Definition: DjVuImage.php:181
__construct( $filename)
Definition: DjVuImage.php:51
getImageSize()
Return data in the style of getimagesize()
Definition: DjVuImage.php:69
wfShellExec( $cmd, &$retval=null, $environ=[], $limits=[], $options=[])
Execute a shell command, with time and memory limits mirrored from the PHP configuration if supported...
Support for detecting/validating DjVu image files and getting some basic file metadata (resolution et...
Definition: DjVuImage.php:38
dump()
For debugging; dump the IFF chunk structure.
Definition: DjVuImage.php:88
wfDebug( $text, $dest='all', array $context=[])
Sends a line to the debug log if enabled or, optionally, to a comment in output.
readChunk( $file)
Definition: DjVuImage.php:161
static tags( $element, $attribs, $contents)
Same as Xml::element(), but does not escape contents.
Definition: Xml.php:130
$header
const DJVUTXT_MEMORY_LIMIT
Memory limit for the DjVu description software.
Definition: DjVuImage.php:43
skipChunk( $file, $chunkLength)
Definition: DjVuImage.php:172
string $mFilename
Definition: DjVuImage.php:46
$wgDjvuDump
Path of the djvudump executable Enable this and $wgDjvuRenderer to enable djvu rendering example: $wg...
isValid()
Check if the given file is indeed a valid DjVu image file.
Definition: DjVuImage.php:59
getPageInfo( $file)
Definition: DjVuImage.php:210
static element( $element, $attribs=null, $contents='', $allowShortTag=true)
Format an XML element with given attributes and, optionally, text content.
Definition: Xml.php:41
$wgDjvuToXML
Path of the djvutoxml executable This works like djvudump except much, much slower as of version 3...
$line
Definition: mcc.php:119
retrieveMetaData()
Return an XML string describing the DjVu image.
Definition: DjVuImage.php:252
convertDumpToXML( $dump)
Hack to temporarily work around djvutoxml bug.
Definition: DjVuImage.php:315
$matches