includes/Math.php

   1 <?php
   2 /**
   3  * Contain everything related to <math> </math> parsing
   4  * @file
   5  * @ingroup Parser
   6  */
   7
   8 /**
   9  * Takes LaTeX fragments, sends them to a helper program (texvc) for rendering
  10  * to rasterized PNG and HTML and MathML approximations. An appropriate
  11  * rendering form is picked and returned.
  12  *
  13  * @author Tomasz Wegrzanowski, with additions by Brion Vibber (2003, 2004)
  14  * @ingroup Parser
  15  */
  16 class MathRenderer {
  17         var $mode = MW_MATH_MODERN;
  18         var $tex = '';
  19         var $inputhash = '';
  20         var $hash = '';
  21         var $html = '';
  22         var $mathml = '';
  23         var $conservativeness = 0;
  24
  25         function __construct( $tex, $params=array() ) {
  26                 $this->tex = $tex;
  27                 $this->params = $params;
  28         }
  29
  30         function setOutputMode( $mode ) {
  31                 $this->mode = $mode;
  32         }
  33
  34         function render() {
  35                 global $wgTmpDirectory, $wgInputEncoding;
  36                 global $wgTexvc, $wgMathCheckFiles, $wgTexvcBackgroundColor;
  37
  38                 if( $this->mode == MW_MATH_SOURCE ) {
  39                         # No need to render or parse anything more!
  40                         # New lines are replaced with spaces, which avoids confusing our parser (bugs 23190, 22818)
  41                         return ('<span class="tex">$ ' . str_replace( "\n", " ", htmlspecialchars( $this->tex ) ) . ' $</span>');
  42                 }
  43                 if( $this->tex == '' ) {
  44                         return; # bug 8372
  45                 }
  46
  47                 if( !$this->_recall() ) {
  48                         if( $wgMathCheckFiles ) {
  49                                 # Ensure that the temp and output directories are available before continuing...
  50                                 if( !file_exists( $wgTmpDirectory ) ) {
  51                                         if( !wfMkdirParents( $wgTmpDirectory ) ) {
  52                                                 return $this->_error( 'math_bad_tmpdir' );
  53                                         }
  54                                 } elseif( !is_dir( $wgTmpDirectory ) || !is_writable( $wgTmpDirectory ) ) {
  55                                         return $this->_error( 'math_bad_tmpdir' );
  56                                 }
  57                         }
  58
  59                         if( !is_executable( $wgTexvc ) ) {
  60                                 return $this->_error( 'math_notexvc' );
  61                         }
  62                         $cmd = $wgTexvc . ' ' .
  63                                         escapeshellarg( $wgTmpDirectory ).' '.
  64                                         escapeshellarg( $wgTmpDirectory ).' '.
  65                                         escapeshellarg( $this->tex ).' '.
  66                                         escapeshellarg( $wgInputEncoding ).' '.
  67                                         escapeshellarg( $wgTexvcBackgroundColor );
  68
  69                         if ( wfIsWindows() ) {
  70                                 # Invoke it within cygwin sh, because texvc expects sh features in its default shell
  71                                 $cmd = 'sh -c ' . wfEscapeShellArg( $cmd );
  72                         }
  73
  74                         wfDebug( "TeX: $cmd\n" );
  75                         $contents = wfShellExec( $cmd );
  76                         wfDebug( "TeX output:\n $contents\n---\n" );
  77
  78                         if (strlen($contents) == 0) {
  79                                 return $this->_error( 'math_unknown_error' );
  80                         }
  81
  82                         $retval = substr ($contents, 0, 1);
  83                         $errmsg = '';
  84                         if (($retval == 'C') || ($retval == 'M') || ($retval == 'L')) {
  85                                 if ($retval == 'C') {
  86                                         $this->conservativeness = 2;
  87                                 } else if ($retval == 'M') {
  88                                         $this->conservativeness = 1;
  89                                 } else {
  90                                         $this->conservativeness = 0;
  91                                 }
  92                                 $outdata = substr ($contents, 33);
  93
  94                                 $i = strpos($outdata, "\000");
  95
  96                                 $this->html = substr($outdata, 0, $i);
  97                                 $this->mathml = substr($outdata, $i+1);
  98                         } else if (($retval == 'c') || ($retval == 'm') || ($retval == 'l'))  {
  99                                 $this->html = substr ($contents, 33);
 100                                 if ($retval == 'c') {
 101                                         $this->conservativeness = 2;
 102                                 } else if ($retval == 'm') {
 103                                         $this->conservativeness = 1;
 104                                 } else {
 105                                         $this->conservativeness = 0;
 106                                 }
 107                                 $this->mathml = null;
 108                         } else if ($retval == 'X') {
 109                                 $this->html = null;
 110                                 $this->mathml = substr ($contents, 33);
 111                                 $this->conservativeness = 0;
 112                         } else if ($retval == '+') {
 113                                 $this->html = null;
 114                                 $this->mathml = null;
 115                                 $this->conservativeness = 0;
 116                         } else {
 117                                 $errbit = htmlspecialchars( substr($contents, 1) );
 118                                 switch( $retval ) {
 119                                         case 'E':
 120                                                 $errmsg = $this->_error( 'math_lexing_error', $errbit );
 121                                                 break;
 122                                         case 'S':
 123                                                 $errmsg = $this->_error( 'math_syntax_error', $errbit );
 124                                                 break;
 125                                         case 'F':
 126                                                 $errmsg = $this->_error( 'math_unknown_function', $errbit );
 127                                                 break;
 128                                         default:
 129                                                 $errmsg = $this->_error( 'math_unknown_error', $errbit );
 130                                 }
 131                         }
 132
 133                         if ( !$errmsg ) {
 134                                  $this->hash = substr ($contents, 1, 32);
 135                         }
 136
 137                         wfRunHooks( 'MathAfterTexvc', array( &$this, &$errmsg ) );
 138
 139                         if ( $errmsg ) {
 140                                  return $errmsg;
 141                         }
 142
 143                         if (!preg_match("/^[a-f0-9]{32}$/", $this->hash)) {
 144                                 return $this->_error( 'math_unknown_error' );
 145                         }
 146
 147                         if( !file_exists( "$wgTmpDirectory/{$this->hash}.png" ) ) {
 148                                 return $this->_error( 'math_image_error' );
 149                         }
 150
 151                         if( filesize( "$wgTmpDirectory/{$this->hash}.png" ) == 0 ) {
 152                                 return $this->_error( 'math_image_error' );
 153                         }
 154
 155                         $hashpath = $this->_getHashPath();
 156                         if( !file_exists( $hashpath ) ) {
 157                                 if( !@wfMkdirParents( $hashpath, 0755 ) ) {
 158                                         return $this->_error( 'math_bad_output' );
 159                                 }
 160                         } elseif( !is_dir( $hashpath ) || !is_writable( $hashpath ) ) {
 161                                 return $this->_error( 'math_bad_output' );
 162                         }
 163
 164                         if( !rename( "$wgTmpDirectory/{$this->hash}.png", "$hashpath/{$this->hash}.png" ) ) {
 165                                 return $this->_error( 'math_output_error' );
 166                         }
 167
 168                         # Now save it back to the DB:
 169                         if ( !wfReadOnly() ) {
 170                                 $outmd5_sql = pack('H32', $this->hash);
 171
 172                                 $md5_sql = pack('H32', $this->md5); # Binary packed, not hex
 173
 174                                 $dbw = wfGetDB( DB_MASTER );
 175                                 $dbw->replace( 'math', array( 'math_inputhash' ),
 176                                   array(
 177                                         'math_inputhash' => $dbw->encodeBlob($md5_sql),
 178                                         'math_outputhash' => $dbw->encodeBlob($outmd5_sql),
 179                                         'math_html_conservativeness' => $this->conservativeness,
 180                                         'math_html' => $this->html,
 181                                         'math_mathml' => $this->mathml,
 182                                   ), __METHOD__
 183                                 );
 184                         }
 185
 186                         // If we're replacing an older version of the image, make sure it's current.
 187                         global $wgUseSquid;
 188                         if ( $wgUseSquid ) {
 189                                 $urls = array( $this->_mathImageUrl() );
 190                                 $u = new SquidUpdate( $urls );
 191                                 $u->doUpdate();
 192                         }
 193                 }
 194
 195                 return $this->_doRender();
 196         }
 197
 198         function _error( $msg, $append = '' ) {
 199                 $mf   = htmlspecialchars( wfMsg( 'math_failure' ) );
 200                 $errmsg = htmlspecialchars( wfMsg( $msg ) );
 201                 $source = htmlspecialchars( str_replace( "\n", ' ', $this->tex ) );
 202                 return "<strong class='error'>$mf ($errmsg$append): $source</strong>\n";
 203         }
 204
 205         function _recall() {
 206                 global $wgMathDirectory, $wgMathCheckFiles;
 207
 208                 $this->md5 = md5( $this->tex );
 209                 $dbr = wfGetDB( DB_SLAVE );
 210                 $rpage = $dbr->selectRow( 'math',
 211                         array( 'math_outputhash','math_html_conservativeness','math_html','math_mathml' ),
 212                         array( 'math_inputhash' => $dbr->encodeBlob(pack("H32", $this->md5))), # Binary packed, not hex
 213                         __METHOD__
 214                 );
 215
 216                 if( $rpage !== false ) {
 217                         # Tailing 0x20s can get dropped by the database, add it back on if necessary:
 218                         $xhash = unpack( 'H32md5', $dbr->decodeBlob($rpage->math_outputhash) . "                " );
 219                         $this->hash = $xhash ['md5'];
 220
 221                         $this->conservativeness = $rpage->math_html_conservativeness;
 222                         $this->html = $rpage->math_html;
 223                         $this->mathml = $rpage->math_mathml;
 224
 225                         $filename = $this->_getHashPath() . "/{$this->hash}.png";
 226
 227                         if( !$wgMathCheckFiles ) {
 228                                 // Short-circuit the file existence & migration checks
 229                                 return true;
 230                         }
 231
 232                         if( file_exists( $filename ) ) {
 233                                 if( filesize( $filename ) == 0 ) {
 234                                         // Some horrible error corrupted stuff :(
 235                                         @unlink( $filename );
 236                                 } else {
 237                                         return true;
 238                                 }
 239                         }
 240
 241                         if( file_exists( $wgMathDirectory . "/{$this->hash}.png" ) ) {
 242                                 $hashpath = $this->_getHashPath();
 243
 244                                 if( !file_exists( $hashpath ) ) {
 245                                         if( !@wfMkdirParents( $hashpath, 0755 ) ) {
 246                                                 return false;
 247                                         }
 248                                 } elseif( !is_dir( $hashpath ) || !is_writable( $hashpath ) ) {
 249                                         return false;
 250                                 }
 251                                 if ( function_exists( "link" ) ) {
 252                                         return link ( $wgMathDirectory . "/{$this->hash}.png",
 253                                                         $hashpath . "/{$this->hash}.png" );
 254                                 } else {
 255                                         return rename ( $wgMathDirectory . "/{$this->hash}.png",
 256                                                         $hashpath . "/{$this->hash}.png" );
 257                                 }
 258                         }
 259
 260                 }
 261
 262                 # Missing from the database and/or the render cache
 263                 return false;
 264         }
 265
 266         /**
 267          * Select among PNG, HTML, or MathML output depending on
 268          */
 269         function _doRender() {
 270                 if( $this->mode == MW_MATH_MATHML && $this->mathml != '' ) {
 271                         return Xml::tags( 'math',
 272                                 $this->_attribs( 'math',
 273                                         array( 'xmlns' => 'http://www.w3.org/1998/Math/MathML' ) ),
 274                                 $this->mathml );
 275                 }
 276                 if (($this->mode == MW_MATH_PNG) || ($this->html == '') ||
 277                    (($this->mode == MW_MATH_SIMPLE) && ($this->conservativeness != 2)) ||
 278                    (($this->mode == MW_MATH_MODERN || $this->mode == MW_MATH_MATHML) && ($this->conservativeness == 0))) {
 279                         return $this->_linkToMathImage();
 280                 } else {
 281                         return Xml::tags( 'span',
 282                                 $this->_attribs( 'span',
 283                                         array( 'class' => 'texhtml' ) ),
 284                                 $this->html );
 285                 }
 286         }
 287
 288         function _attribs( $tag, $defaults=array(), $overrides=array() ) {
 289                 $attribs = Sanitizer::validateTagAttributes( $this->params, $tag );
 290                 $attribs = Sanitizer::mergeAttributes( $defaults, $attribs );
 291                 $attribs = Sanitizer::mergeAttributes( $attribs, $overrides );
 292                 return $attribs;
 293         }
 294
 295         function _linkToMathImage() {
 296                 $url = $this->_mathImageUrl();
 297
 298                 return Xml::element( 'img',
 299                         $this->_attribs(
 300                                 'img',
 301                                 array(
 302                                         'class' => 'tex',
 303                                         'alt' => $this->tex ),
 304                                 array(
 305                                         'src' => $url ) ) );
 306         }
 307
 308         function _mathImageUrl() {
 309                 global $wgMathPath;
 310                 $dir = $this->_getHashSubPath();
 311                 return "$wgMathPath/$dir/{$this->hash}.png";
 312         }
 313
 314         function _getHashPath() {
 315                 global $wgMathDirectory;
 316                 $path = $wgMathDirectory .'/' . $this->_getHashSubPath();
 317                 wfDebug( "TeX: getHashPath, hash is: $this->hash, path is: $path\n" );
 318                 return $path;
 319         }
 320
 321         function _getHashSubPath() {
 322                 return substr($this->hash, 0, 1)
 323                                         .'/'. substr($this->hash, 1, 1)
 324                                         .'/'. substr($this->hash, 2, 1);
 325         }
 326
 327         public static function renderMath( $tex, $params=array() ) {
 328                 global $wgUser;
 329                 $math = new MathRenderer( $tex, $params );
 330                 $math->setOutputMode( $wgUser->getOption('math'));
 331                 return $math->render();
 332         }
 333 }