loncom/xml/lontexconvert.pm - view

File: [LON-CAPA] / loncom / xml / lontexconvert.pm
Revision 1.110: download - view: text, annotated - select for diffs
Tue Feb 28 10:38:14 2012 UTC (12 years, 3 months ago) by foxr
Branches: MAIN
CVS tags: HEAD

BZ 5837 -
  1. Add table of conversion string -> unicode equivalents and
     sub to do a conversion.
  2. Use this to turn the entities generated by tth to unicode bracket
segment characters directly.

1: # The LearningOnline Network with CAPA 2: # TeX Conversion Module 3: # 4: # $Id: lontexconvert.pm,v 1.110 2012/02/28 10:38:14 foxr Exp $ 5: # 6: # Copyright Michigan State University Board of Trustees 7: # 8: # This file is part of the LearningOnline Network with CAPA (LON-CAPA). 9: # 10: # LON-CAPA is free software; you can redistribute it and/or modify 11: # it under the terms of the GNU General Public License as published by 12: # the Free Software Foundation; either version 2 of the License, or 13: # (at your option) any later version. 14: # 15: # LON-CAPA is distributed in the hope that it will be useful, 16: # but WITHOUT ANY WARRANTY; without even the implied warranty of 17: # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 18: # GNU General Public License for more details. 19: # 20: # You should have received a copy of the GNU General Public License 21: # along with LON-CAPA; if not, write to the Free Software 22: # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA 23: # 24: # /home/httpd/html/adm/gpl.txt 25: # 26: # http://www.lon-capa.org/ 27: # 28: # Copyright for TtHfunc and TtMfunc by Ian Hutchinson. 29: # TtHfunc and TtMfunc (the "Code") may be compiled and linked into 30: # binary executable programs or libraries distributed by the 31: # Michigan State University (the "Licensee"), but any binaries so 32: # distributed are hereby licensed only for use in the context 33: # of a program or computational system for which the Licensee is the 34: # primary author or distributor, and which performs substantial 35: # additional tasks beyond the translation of (La)TeX into HTML. 36: # The C source of the Code may not be distributed by the Licensee 37: # to any other parties under any circumstances. 38: # 39: 40: package Apache::lontexconvert; 41: 42: use strict; 43: use tth(); 44: use vars qw($errorstring); 45: #use Apache::lonxml(); 46: use Apache::lonlocal; 47: use Apache::lonnet; 48: use lib '/home/httpd/lib/perl/'; 49: use LONCAPA; 50: use URI::Escape; 51: use IO::Socket::INET; 52: 53: 54: # 55: # Table of substitutions to unicode characters. 56: # 57: my %unicode_translations = ( 58: '\rightleftharpoons' => 0x21cc, 59: '' => 0x23a1, 60: '' => 0x23a2, 61: '' => 0x23a3, 62: '' => 0x23a4, 63: '' => 0x23a5, 64: '' => 0x23a6 65: ); 66: 67: ## 68: # Utility to convert elements of a string to unicode: 69: # 70: # @param input - Input string 71: # @param pattern - Pattern to convert 72: # @param unicode - Unicode to substitute for pattern. 73: # 74: # @return string - resulting string. 75: # 76: sub unicode_subst { 77: my ($input, $pattern, $unicode) = @_; 78: 79: my $char = pack('U', $unicode); 80: 81: $input =~ s/$pattern/$char/g; 82: 83: return $input; 84: } 85: 86: # ====================================================================== Header 87: 88: sub init_tth { 89: my $options=$env{'course.'.$env{'request.course.id'}.'.tthoptions'}; 90: if ($options =~ /\S/) { 91: $options = ' '.$options; 92: } else { 93: undef($options); 94: } 95: if ($env{'browser.mathml'}) { 96: &tth::ttminit(); 97: if ($env{'browser.unicode'}) { 98: &tth::ttmoptions('-L -u1'.$options); 99: } else { 100: &tth::ttmoptions('-L -u0'.$options); 101: } 102: } else { 103: &tth::tthinit(); 104: if ($env{'browser.unicode'}) { 105: &tth::tthoptions('-L -u1'.$options); 106: } else { 107: &tth::tthoptions('-L -u0'.$options); 108: } 109: } 110: } 111: 112: # ================================================================== Conversion 113: 114: $Apache::lontexconvert::messedup=0; 115: 116: 117: sub convert_real { 118: my ($texstring)=@_; 119: my ($xmlstring,$errorstring); 120: local $SIG{SEGV}=sub { $Apache::lontexconvert::messedup=1; die; }; 121: local $SIG{ALRM}=sub { 122: &Apache::lonnet::logthis("ALRM"); 123: $xmlstring='['.&mt("TeX unconverted due to errors").']'; 124: $Apache::lontexconvert::messedup=1; 125: die &mt("TeX unconverted due to errors"); 126: }; 127: &Apache::lonxml::start_alarm(); 128: if ($env{'browser.mathml'}) { 129: &Apache::lonnet::logthis("mathml"); 130: $xmlstring=&tth::ttm($$texstring); 131: $xmlstring=~s/\<math\>/\<math xmlns=\"\&mathns\;\"\>/g; 132: $xmlstring=~s/\<br\>/\<br\/\>/g; 133: $xmlstring=~s/\<p\>/\<p\>\<\/p\>/g; 134: $errorstring.=&tth::ttmerror(); 135: } else { 136: &Apache::lonnet::logthis("tth"); 137: $xmlstring=&tth::tth($$texstring); 138: $errorstring.=&tth::ttherror(); 139: $xmlstring=~s-</font(\s*)>-</font>-g; 140: } 141: $xmlstring=~s/^\s*\<br clear\=\"all\"/\<br/s; 142: $xmlstring=~s/^\s*//; 143: $xmlstring=~s/\s*$//; 144: &Apache::lonxml::end_alarm(); 145: 146: # 147: # Several strings produced by tth require 148: # transliteration -> unicode equivalents to render reliably 149: # in browsers. %unicode_translations is a table of 150: # string->substitution which we now apply: 151: 152: foreach my $pattern (keys(%unicode_translations)) { 153: my $unicode = $unicode_translations{$pattern}; 154: $xmlstring = &unicode_subst($xmlstring, $pattern, $unicode); 155: } 156: 157: 158: return ($xmlstring,$errorstring); 159: } 160: 161: sub tth_converted { 162: my $texstring=shift; 163: my $xmlstring='['.&mt('UNDISPLAYABLE').']'; 164: if ($Apache::lontexconvert::messedup) { 165: return '['.&mt('TeX unconverted due to previous errors').']'; 166: } 167: $$texstring ='\\documentstyle{article}'.$$texstring; 168: 169: eval(<<'ENDCONV'); 170: ($xmlstring,$errorstring)=&convert_real($texstring) 171: ENDCONV 172: if ($@) { 173: $errorstring.=&mt("Evaluation Error: ").$@; 174: $Apache::lontexconvert::messedup=1; 175: } 176: if ($Apache::lontexconvert::messedup || &tth::tthmessedup() || 177: $errorstring) { 178: &Apache::lonnet::logthis("Trying to kill myself"); 179: $Apache::lontexconvert::messedup=1; 180: if (ref($Apache::lonxml::request)) { 181: $Apache::lonxml::request->child_terminate(); 182: } else { 183: my $request; 184: eval { $request=Apache->request; }; 185: if (!$request) { 186: eval { $request=Apache2::RequestUtil->request; }; 187: } 188: if ($request) { 189: $request->child_terminate(); 190: } else { 191: &Apache::lonnet::logthis("Unable to find a request to do child_terminate on"); 192: } 193: } 194: } 195: return $xmlstring; 196: } 197: 198: sub clean_out_math_mode { 199: my ($texstring)=@_; 200: $$texstring=~s/(?<!\\)\$//g; 201: $$texstring=~s/\\[\)\(\]\[]//g; 202: $$texstring=~s/\\ensuremath//g; 203: return ''; 204: } 205: 206: sub displaystyle { 207: my ($texstring)=@_; 208: #has a $$ or \[ or \displaystyle or eqnarray in it, guessinng it's display mode 209: if ($$texstring=~/[^\\]\$\$/ || 210: $$texstring=~/\\\[/ || 211: $$texstring=~/\\displaystyle/ || 212: $$texstring=~/eqnarray/ 213: ) { return 1; } 214: return 0; 215: } 216: 217: sub MathJax_converted { 218: my $texstring=shift; 219: my $tag='math/tex;'; 220: if (&displaystyle($texstring)) { $tag='math/tex; mode=display'; } 221: &clean_out_math_mode($texstring); 222: return &MathJax_header(). 223: '<script type="'.$tag.'">'.$$texstring.'</script>'; 224: } 225: 226: { 227: #Relies heavily on the previous jsMath installation 228: my @MathJax_sent_header; 229: sub MathJax_reset { 230: undef(@MathJax_sent_header); 231: } 232: sub MathJax_push { 233: push(@MathJax_sent_header,0); 234: } 235: sub MathJax_header { 236: if (!@MathJax_sent_header) { 237: &Apache::lonnet::logthis("mismatched calls of MathJax_header and MathJax_process"); 238: return ''; 239: } 240: return '' if $MathJax_sent_header[-1]; 241: $MathJax_sent_header[-1]=1; 242: return 243: '<script type="text/javascript" src="/adm/MathJax/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>'."\n"; 244: } 245: #sub MathJax_process { 246: # my $state = pop(@MathJax_sent_header); 247: # return '' if !$state; 248: # return "\n". 249: # '<script type="text/javascript">MathJax.Process()</script>'."\n"; 250: #} 251: #sub MathJax_state { 252: # my ($level) = @_; 253: # return $MathJax_sent_header[$level]; 254: #} 255: } 256: 257: 258: sub jsMath_converted { 259: my $texstring=shift; 260: my $tag='span'; 261: if (&displaystyle($texstring)) { $tag='div'; } 262: &clean_out_math_mode($texstring); 263: return &jsMath_header(). 264: '<'.$tag.' class="math">'.$$texstring.'</'.$tag.'>'; 265: } 266: 267: { 268: my @jsMath_sent_header; 269: sub jsMath_reset { 270: undef(@jsMath_sent_header); 271: } 272: sub jsMath_push { 273: push(@jsMath_sent_header,0); 274: } 275: sub jsMath_header { 276: if (!@jsMath_sent_header) { 277: &Apache::lonnet::logthis("mismatched calls of jsMath_header and jsMath_process"); 278: return ''; 279: } 280: return '' if $jsMath_sent_header[-1]; 281: $jsMath_sent_header[-1]=1; 282: return 283: '<script type="text/javascript"> 284: function NoFontMessage () {} 285: jsMath = {Parser: {prototype: {environments: {\'eqnarray*\' :[\'Array\',null,null,\'rcl\',[5/18,5/18],3,\'D\']}}}}; 286: </script>'."\n". 287: '<script type="text/javascript" src="/adm/jsMath/jsMath.js"></script>'."\n"; 288: } 289: sub jsMath_process { 290: my $state = pop(@jsMath_sent_header); 291: return '' if !$state; 292: return "\n". 293: '<script type="text/javascript">jsMath.Process()</script>'."\n"; 294: } 295: sub jsMath_state { 296: my ($level) = @_; 297: return $jsMath_sent_header[$level]; 298: } 299: } 300: 301: sub tex_engine { 302: if (exists($env{'form.texengine'})) { 303: if ($env{'form.texengine'} ne '') { 304: return $env{'form.texengine'}; 305: } 306: } 307: if ($env{'request.course.id'} 308: && exists($env{'course.'.$env{'request.course.id'}.'.texengine'})) { 309: return $env{'course.'.$env{'request.course.id'}.'.texengine'}; 310: } 311: if (exists($env{'environment.texengine'})) { 312: return $env{'environment.texengine'}; 313: } 314: return 'tth'; 315: } 316: 317: sub init_math_support { 318: my ($inherit_jsmath) = @_; 319: &init_tth(); 320: &Apache::lontexconvert::jsMath_push(); 321: if (lc(&tex_engine()) eq 'jsmath' || 322: ($inherit_jsmath && &jsMath_state(-2))) { 323: return &Apache::lontexconvert::jsMath_header(); 324: } 325: &Apache::lontexconvert::MathJax_push(); 326: if (lc(&tex_engine()) eq 'mathjax') { # || 327: #($inherit_jsmath && &jsMath_state(-2))) { 328: return &Apache::lontexconvert::MathJax_header(); 329: } 330: return; 331: } 332: 333: sub mimetex_valign { 334: my ($esc_texstring)=@_; 335: my $valign = 0; 336: my $path = '/cgi-bin/mimetex.cgi?'.$esc_texstring; 337: my $socket; 338: &Apache::lonxml::start_alarm(); 339: $socket = IO::Socket::INET->new(PeerAddr => 'localhost', 340: PeerPort => 'http(80)', 341: Proto => 'tcp'); 342: if ($socket) { 343: my $headreq = "HEAD $path HTTP/1.0\r\n\r\n"; 344: print $socket $headreq; 345: while (<$socket>) { 346: if (/Vertical\-Align\:\s*?([\-\d]+)/) { 347: $valign = $1; 348: } 349: } 350: $socket->close(); 351: } 352: &Apache::lonxml::end_alarm(); 353: return $valign; 354: } 355: 356: sub mimetex_converted { 357: my $texstring=shift; 358: 359: # Alt-Argument for screen readers 360: my $alt_string=$$texstring; 361: $alt_string=~s/\"/\'\'/g; 362: 363: # Is this displaystyle? 364: 365: my $displaystyle=&displaystyle($texstring); 366: 367: # Remove math environment delimiters 368: 369: &clean_out_math_mode($texstring); 370: 371: if ($displaystyle) { 372: $$texstring='\\displaystyle \\Large '.$$texstring; 373: } 374: my $esc_texstring = &uri_escape($$texstring); 375: my $valign = &mimetex_valign($esc_texstring); 376: my $result='<img src="/cgi-bin/mimetex.cgi?'.$esc_texstring.'" style="vertical-align:'.$valign.'px" alt="'.$alt_string.'" />'; 377: if ($displaystyle) { 378: $result='<div style="text-align:center">'.$result.'</div>'; 379: } 380: return $result; 381: } 382: 383: sub converted { 384: my ($string,$mode)=@_; 385: if ($mode eq '') { $mode = &tex_engine(); } 386: if ($mode =~ /tth/i) { 387: return &tth_converted($string); 388: } elsif ($mode =~ /jsmath/i) { 389: return &jsMath_converted($string); 390: } elsif ($mode =~ /mathjax/i) { 391: return &MathJax_converted($string); 392: } elsif ($mode =~ /mimetex/i) { 393: return &mimetex_converted($string); 394: } elsif ($mode =~ /raw/i) { 395: return $$string; 396: } 397: return &tth_converted($string); 398: } 399: 400: # ------------------------------------------------------------ Message display 401: 402: sub to_convert { 403: my ($string) = @_; 404: &init_tth(); 405: $string=~s/\<br\s*\/?\>/ /gs; 406: # $string=~s/\s/ /gs; 407: $string=&HTML::Entities::decode($string); 408: return &converted(\$string); 409: } 410: 411: sub smiley { 412: my $expression=shift; 413: my %smileys=( 414: '\:\-*\)' => 'face-smile.png', 415: '8\-\)' => 'face-cool.png', 416: '8\-(I|\|)' => 'face-glasses.png', 417: '\:\-(I|\|)' => 'face-plain.png', 418: '\:\-(o|O|)' => 'face-surprise.png', 419: ':\-$' => 'face-sad.png', 420: '\;\-$' => 'face-wink.png', 421: '\:\-(P|p)' => 'face-raspberry.png', 422: '\:\-(\\\|\\/)' => 'face-uncertain.png', 423: '\:\-D' => 'face-smile-big.png', 424: '\:\-(C|\@)' => 'face-angry.png', 425: '\:(\'|\`)\-*$' => 'face-crying.png', 426: '\:\-(X|x|\#)' => 'face-quiet.png', 427: '\:\-(s|S)' => 'face-uncertain.png', 428: '\:\-\$' => 'face-embarrassed.png', 429: '\:\-\*' => 'face-kiss.png', 430: '\+O\(' => 'face-sick.png', 431: '(\&lt\;3|\(heart$)' => 'heart.png', 432: '$rose$' => 'rose.png', 433: '$pizza$' => 'food-pizza.png', 434: '$cake$' => 'food-cake.png', 435: '$ninja$' => 'face-ninja.png', 436: '$pirate$' => 'face-pirate.png', 437: '$(agree|yes)$' => 'opinion-agree.png', 438: '$(disagree|nay)$' => 'opinion-disagree.png', 439: '(o|O)\-\)' => 'face-angel.png', 440: ); 441: my $iconpath=$Apache::lonnet::perlvar{'lonIconsURL'}; 442: foreach my $smiley (keys(%smileys)) { 443: $expression=~s/$smiley/\<img src="$iconpath\/$smileys{$smiley}" \/\>/gs; 444: } 445: return $expression; 446: } 447: 448: sub msgtexconverted { 449: my ($message,$email) = @_; 450: $errorstring=''; 451: my $outmessage=''; 452: my $tex=0; 453: foreach my $fragment (split(/(?:\&lt\;|\<)\/*m\s*(?:\&gt\;|\>)/i,$message)) { 454: if ($tex) { 455: if ($email) { 456: $outmessage.='</pre><tt>'.&to_convert($fragment).'</tt><pre>'; 457: $tex=0; 458: } else { 459: $outmessage.=&to_convert($fragment); 460: $tex=0; 461: } 462: } else { 463: $outmessage.=&smiley($fragment); 464: $tex=1; 465: } 466: } 467: $message=$outmessage; $outmessage=''; $tex=0; 468: foreach my $fragment (split(/(?:\&lt\;|\<)\/*algebra\s*(?:\&gt\;|\>)/i, 469: $message)) { 470: if ($tex) { 471: my $algebra = &algebra($fragment, 'web', undef, undef, undef, undef, 'tth'); 472: if ($email) { 473: $outmessage.='</pre><tt>'.$algebra.'</tt><pre>'; 474: $tex=0; 475: } else { 476: $outmessage.=$algebra; 477: $tex=0; 478: } 479: } else { 480: $outmessage.=$fragment; 481: $tex=1; 482: } 483: } 484: if (wantarray) { 485: return ($outmessage,$errorstring); 486: } else { 487: return $outmessage.$errorstring; 488: } 489: } 490: 491: sub algebra { 492: use AlgParser; 493: my ($string,$target,$style,$parstack,$safeeval,$tth)=@_; 494: my $parser = new AlgParserWithImplicitExpand; 495: if ($tth eq 'tth') {&init_tth();} 496: $string=&prepare_algebra($string); 497: my $ret = $parser->parse($string); 498: my $result='['.&mt('Algebra unconverted due to previous errors').']'; 499: if ( ref($ret) ) { 500: #$parser->tostring(); 501: $parser->normalize(); 502: my $latex=$parser->tolatex(); 503: $latex=&postprocess_algebra($latex); 504: if ($style eq 'display') { 505: $latex='$$'.$latex.'$$x'; 506: } else { 507: # style is 'inline' 508: $latex='\\ensuremath{'.$latex.'}'; 509: } 510: if ($target eq 'web' || $target eq 'analyze') { 511: my $display=&Apache::lonxml::get_param('display',$parstack,$safeeval); 512: $result = &converted(\$latex,$display); 513: # $result = &converted(\$latex); 514: } else { 515: $result = $latex; 516: } 517: } else { 518: &Apache::lonxml::error($parser->{'htmlerror'}); 519: } 520: } 521: 522: sub prepare_algebra { 523: my ($string)=@_; 524: 525: # makes the decision about what is a minus sign easier supposedly 526: $string =~ s/(\<\>|\<\=|\>\=[\=\>\<] *)-/$1 zeroplace -/g; 527: 528: return $string; 529: } 530: 531: sub postprocess_algebra { 532: my ($string)=@_; 533: 534: # moodle had these and I don't know why, ignoring them for now 535: # $string =~s/\\fun/ /g; 536: 537: # sqrt(3,4) means the 4 root of 3 538: $string =~s/\\sqrt{([^,]+),([^\}]+)}/\\sqrt[$2]{$1}/gs; 539: 540: # log(3,4) means the log base 4 of 3 541: $string =~s/\\log\\left$(.+?),(.+?)\\right$/\\log_{$2}\\left($1\\right)/gs; 542: 543: # log(3,4) means the log base 4 of 3 544: $string =~s/\$(?:sin|cos|tan|sec|csc|cot)(?:h)?)\\left\((.+?),(.+?)\\right$/\\$1^{$3}\\left($2\\right)/gs; 545: 546: # int(3,a,b) integral from a to b of 3 547: $string =~s/\\int\\left$(.+?),(.+?),(.+?)\\right$/\\int_{$2}^{$3}\\left($1\\right)/gs; 548: 549: # int( ... dx) -> ... 550: $string =~s/\\int\\left$(.+?)d[a-z]\\right$/$1/gs; 551: 552: # 553: $string =~s/\\lim\\left$(.+?),(.+?),(.+?)\\right$/\\lim_{$2\\to $3}$1/gs; 554: return $string; 555: } 556: 1; 557: __END__ 558: 559: 560: =pod 561: 562: =head1 NAME 563: 564: Apache::lontexconvert; 565: 566: =head1 SYNOPSIS 567: 568: Access to tth/ttm 569: 570: This is part of the LearningOnline Network with CAPA project 571: described at http://www.lon-capa.org. 572: 573: 574: =head1 SUBROUTINES 575: 576: =over 577: 578: =item init_tth() 579: 580: Header 581: 582: =item convert_real() 583: 584: we need this routine because &converted can get called from inside 585: of the safespace (through &xmlparse('<m>stuff</m>') which doesn't 586: allow the opcode for alarm, so we need to compile this before we get 587: into the safe space since opcode checks only occur at compile time 588: 589: =item tth_converted() 590: 591: 592: =item clean_out_math_mode() 593: 594: 595: =item displaystyle() 596: 597: 598: =item jsMath_converted() 599: 600: =item MathJax_converted() 601: - Mimics the jsMath functionality 602: 603: =item tex_engine() 604: 605: 606: =item init_math_support() 607: 608: =item mimetex_valign() 609: 610: Makes a HEAD call to /cgi-bin/mimetex.cgi via IO:: to retrieve the 611: vertical alignment, before the subsequent call to mimetex_converted() 612: which generates the <img> tag and the corresponding image. 613: 614: Input: 1. $esc_texstring (escaped TeX to be rendered by mimetex). 615: Output: 1. $valign - number of pixels: positive or negative integer 616: which will be included in <img> tag for mimetex image to 617: support vertical alignment of image within a line of text. 618: 619: If a server is running SSL, and Apache rewrite rules are in place 620: to rewrite requests for http to https, modification will most likely 621: be needed for pass through for HEAD requests for /cgi-bin/mimetex.cgi. 622: 623: Example rewrite rules which rewrite all http traffic to https, 624: except HEAD requests for /cgi-bin/mimetex.cgi are: 625: 626: <IfModule mod_rewrite.c> 627: RewriteEngine On 628: RewriteLogLevel 0 629: 630: RewriteCond %{HTTPS} off 631: RewriteCond %{HTTP:Host} (.*) 632: RewriteCond %{REQUEST_METHOD} !HEAD 633: RewriteRule ^/(.*) https://%1/$1 [R=301,L] 634: 635: RewriteCond %{HTTPS} off 636: RewriteCond %{HTTP:Host} (.*) 637: RewriteCond %{REQUEST_METHOD} HEAD 638: RewriteCond %{REQUEST_URI} !^/cgi-bin/mimetex.cgi 639: RewriteRule ^/(.*) https://%1/$1 [R=301,L] 640: </IfModule> 641: 642: =item mimetex_converted() 643: 644: 645: =item converted() 646: 647: 648: =item to_convert() 649: 650: message display 651: 652: =item smiley() 653: 654: ??? 655: 656: =item msgtexconverted() 657: 658: =item algebra() 659: 660: =item prepare_algebra() 661: 662: =item postprocess_algebra() 663: 664: =back 665: 666: =cut 667: 668: 669: