latexml 0.8.2-1 / usr / bin / latexmlmath

#!/usr/bin/perl -w
# /=====================================================================\ #
# |  latexmlmath                                                        | #
# | math conversion program                                             | #
# |=====================================================================| #
# | Part of LaTeXML:                                                    | #
# |  Public domain software, produced as part of work done by the       | #
# |  United States Government & not subject to copyright in the US.     | #
# |---------------------------------------------------------------------| #
# | Bruce Miller <bruce.miller@nist.gov>                        #_#     | #
# | http://dlmf.nist.gov/LaTeXML/                              (o o)    | #
# \=========================================================ooo==U==ooo=/ #
use strict;
use warnings;
use FindBin;
use lib "$FindBin::RealBin/../lib";
use Getopt::Long qw(:config no_ignore_case);
use Pod::Usage;
use LaTeXML;    # Currently, just for version information.
use LaTeXML::Core;
use LaTeXML::Util::Pathname;
use File::Spec;
use File::Temp qw(tempfile);
use LaTeXML::Post;
use LaTeXML::Post::Scan;
use LaTeXML::Post::CrossRef;
#use LaTeXML::Post::Writer;
use LaTeXML::Util::ObjectDB;

#**********************************************************************
# Parse command line

my ($verbosity, $strict, $noparse, $includestyles) = (-1, 0, 0, 0);
my ($destination, $help, $showversion) = ('', '');
my ($documentid);
my ($mathimage, $mathsvg, $mag) = (undef, undef, 1.75);
my ($pmml, $linelength, $plane1, $hackplane1) = (undef, undef, undef, undef);
my $xmath = undef;
my $cmml  = undef;
my $om    = undef;
my @paths = ('.');
my @preload;
GetOptions(
  # No provision for parallel output (yet)
  "mathimage=s"               => \$mathimage,
  "mathsvg=s"                 => \$mathsvg,
  "magnification=f"           => \$mag,
  "presentationmathml|pmml=s" => \$pmml,
  "linelength=i"              => \$linelength,
  "plane1!"                   => \$plane1,
  "hackplane1!"               => \$hackplane1,
  "contentmathml|cmml=s"      => \$cmml,
  "openmath|om=s"             => \$om,
  "XMath=s"                   => \$xmath,

  "noparse" => \$noparse,

  "preload=s"     => \@preload,
  "includestyles" => \$includestyles,
  "path=s"        => \@paths,
  "quiet"         => sub { $verbosity--; },
  "verbose"       => sub { $verbosity++; },
  "strict"        => \$strict,
  "VERSION"       => \$showversion,
  "debug=s"       => sub { no strict 'refs'; ${ 'LaTeXML::' . $_[1] . '::DEBUG' } = 1; },
  "documentid=s"  => \$documentid,
  "help"          => \$help,
  ) or pod2usage(-message => $LaTeXML::IDENTITY,
  -exitval => 1, -verbose => 0, -output => \*STDERR);
pod2usage(-message => $LaTeXML::IDENTITY, -exitval => 1, -verbose => 2, -output => \*STDOUT)
  if $help;
if ($showversion) { print STDERR "$LaTeXML::IDENTITY\n"; exit(1); }
pod2usage(-message => "$LaTeXML::IDENTITY\nMissing input TeX file",
  -exitval => 1, -verbose => 0, -output => \*STDERR) unless @ARGV;

binmode(STDERR, ":encoding(UTF-8)");
print STDERR "$LaTeXML::IDENTITY\n" if $verbosity > -1;

#======================================================================
# TeX Source
#======================================================================
# First read and digest whatever we're given.
my $tex = join(' ', @ARGV);
if ($tex eq '-') {
  { local $/ = undef; $tex = <>; } }

# We need to determine whether the TeX we're given needs to be wrapped in \[...\]
# Does it have $'s around it? Does it have a display math environment?
# The most elegant way would be to notice as soon as we start adding to the doc
# and switch to math mode if necessary, but that's tricky.
# Let's just try a manual hack, looking for known switches...
$tex =~ s/^\s+//;
$tex =~ s/\s+$//;

our $MATHENVS = 'math|displaymath|equation*?|eqnarray*?'
  . '|multline*?|align*?|falign*?|alignat*?|xalignat*?|xxalignat*?|gather*?';
if    (($tex =~ /^\$/)                     && ($tex =~ /\$$/))          { }    # Wrapped in $'s
elsif (($tex =~ /^\\\(/)                   && ($tex =~ /\\\)$/))        { }    # Wrapped in \(...\)
elsif (($tex =~ /^\\\[/)                   && ($tex =~ /\\\]$/))        { }    # Wrapped in \[...\]
elsif (($tex =~ /^\\begin\{($MATHENVS)\}/) && ($tex =~ /\\end\{$1\}$/)) { }
else {
  $tex = '\[ ' . $tex . ' \]'; }

my $texdoc = <<"EODoc";
literal:
\\documentclass{article}
\\begin{document}
\\newcounter{equation}
\\newcounter{Unequation}
$tex
\\end{document}
EODoc

# NOTE:
#   Should I do something to verify that there is ONLY one Math node ?
#   Eg: cmdline  a + b $ \badmacros $ c + d
#
#   What about alignment environments?
#     Should it be possible to give eqnarray or similar?
#     and if so, will they give multiple math nodes?

#======================================================================
# Digest the TeX
#======================================================================
@paths = map { pathname_canonical($_) } @paths;
if (my @baddirs = grep { !-d $_ } @paths) {
  warn "$LaTeXML::IDENTITY: these path directories do not exist: " . join(', ', @baddirs) . "\n"; }

my $latexml = LaTeXML::Core->new(preload => ['LaTeX.pool', @preload], searchpaths => [@paths],
  verbosity => $verbosity, strict => $strict,
  includeComments => 0, includeStyles => $includestyles,
  documentid      => $documentid,
  nomathparse     => $noparse);

my $digested = $latexml->digestFile($texdoc);

# NASTY Hack: to disable cleanup_Math
# cleanup_Math is useful within "real" documents, to recognize & simplify non-math
# that gets marked up within $..$ (a common TeX idiom).
# However, that simplification isn't appropriate for isolated "math" expressions
# especially within automation, where latexmlmath might be used.
my $model = $$latexml{state}->getModel;
$$model{tagprop}{'ltx:Math'}{afterClose}
  = [grep { $_ ne \&LaTeXML::Package::Pool::cleanup_Math }
    @{ $$model{tagprop}{'ltx:Math'}{afterClose} }];

my $converted = $digested && $latexml->convertDocument($digested);
my $document = $digested && LaTeXML::Post::Document->new($converted, nocache => 1);

#======================================================================
# Postprocess to convert the math to whatever desired forms.
#======================================================================
# Since we can't easily find & extract all the various formats at once,
# let's just process each one separately.

our $mmlURI = "http://www.w3.org/1998/Math/MathML";
our $omURI  = "http://www.openmath.org/OpenMath";

# Default to pmml to stdout, if nothing else specified.
$pmml = '-' unless (defined $mathimage || defined $mathsvg
  || defined $pmml || defined $cmml || defined $om || defined $xmath);

our %OPTIONS = ();

# I wonder if this stuff even will be needed,
# once fragid is sorted out?
my $DB = LaTeXML::Util::ObjectDB->new(%OPTIONS);
my $post = LaTeXML::Post->new(verbosity => $verbosity || 0);
($document) = $post->ProcessChain($document,
  LaTeXML::Post::Scan->new(db => $DB, %OPTIONS),
  LaTeXML::Post::CrossRef->new(db => $DB, %OPTIONS));

if ($mathimage) {
  my $imagetype = 'png';
  if ($mathimage =~ s/\.(png|gif|jpg|jpeg|svg)$//) {
    $imagetype = $1; }
  if ($mathimage eq '-') {
    die "I dont really want to write the image to STDOUT, sorry!"; }

  my ($result) = $post->ProcessChain(cloneDoc($document),
    DirectMathImages->new(magnification => $mag,
      imagename => $mathimage, imagetype => $imagetype,
      %OPTIONS));
  unlink('LaTeXML.cache'); }

if ($mathsvg) {
  my $imagetype = 'svg';
  if ($mathsvg eq '-') {
    my ($fh, $tmpfile) = tempfile();
    my ($result) = $post->ProcessChain(cloneDoc($document),
      DirectMathImages->new(magnification => $mag,
        imagename => $tmpfile, imagetype => $imagetype,
        %OPTIONS));
    my $SVG;
    open($SVG, '<', "$tmpfile.svg") or die "Couldn't read temporary svg output '$tmpfile': $!";
    while (<$SVG>) {
      print $_; }
    close($SVG);
    unlink "$tmpfile.svg"; }
  else {
    $mathsvg =~ s/\.svg$//;
    my ($result) = $post->ProcessChain(cloneDoc($document),
      DirectMathImages->new(magnification => $mag,
        imagename => $mathsvg, imagetype => $imagetype,
        %OPTIONS)); }
  unlink('LaTeXML.cache'); }

if ($pmml) {
  require LaTeXML::Post::MathML::Presentation;
  my ($result) = $post->ProcessChain(cloneDoc($document),
    LaTeXML::Post::MathML::Presentation->new(
      (defined $linelength ? (linelength => $linelength) : ()),
      (defined $plane1     ? (plane1     => $plane1)     : ()),
      (defined $hackplane1 ? (hackplane1 => $hackplane1) : ()),
      %OPTIONS));
  outputXML($result->findnode('//m:math'), $pmml, $mmlURI); }

if ($cmml) {
  require LaTeXML::Post::MathML::Content;
  my ($result) = $post->ProcessChain(cloneDoc($document),
    LaTeXML::Post::MathML::Content->new(%OPTIONS));
  outputXML($result->findnode('//m:math'), $cmml, $mmlURI); }

if ($om) {
  require LaTeXML::Post::OpenMath;
  my ($result) = $post->ProcessChain(cloneDoc($document),
    LaTeXML::Post::OpenMath->new(%OPTIONS));
  outputXML($result->findnode('//om:OMOBJ'), $om, $omURI); }

if ($xmath) {
  # extract the xmath
  my ($result) = $post->ProcessChain(cloneDoc($document));
  outputXML($result->findnode('//ltx:XMath'), $xmath); }

print STDERR "\nConversion complete\n" if $verbosity > -1;

#======================================================================
# Helpers
#======================================================================

sub cloneDoc {
  my ($document) = @_;
  my $clone = $document->getDocumentElement->cloneNode(1);
  foreach my $pi ($document->findnodes(".//processing-instruction('latexml')")) {
    $clone->appendChild($pi->cloneNode); }
  return $document->new($clone); }

sub outputXML {
  my ($xml, $xmldestination, $defaulturi) = @_;
  # Need to move the node to a new document to set up namespaces
  my $newdoc = XML::LibXML::Document->new("1.0", "UTF-8");
  $newdoc->setDocumentElement($xml);
  if (my $oldprefix = $defaulturi && $xml->lookupNamespacePrefix($defaulturi)) {
    $xml->setNamespaceDeclPrefix($oldprefix, undef); }
  my $serialized = $newdoc->toString(1);
  if ($xmldestination eq '-') {
    print $serialized; }
  else {
    $xmldestination = pathname_canonical($xmldestination);
    if (my $dir = pathname_directory($xmldestination)) {
      pathname_mkdir($dir) or die "Couldn't create destination directory $dir: $!"; }
    my $OUT;
    open($OUT, '>', $xmldestination) or die "Couldn't open output file $xmldestination: $!";
    print $OUT $serialized;
    close($OUT); }
  return; }

#======================================================================
# Variant of MathImages to write directly to destination
#======================================================================
package DirectMathImages;
use base qw(LaTeXML::Post::MathImages);

our $counter;

sub desiredResourcePathname {
  my ($self, $doc, $node, $source, $type) = @_;
  my $name = $$self{imagename};
  if ($counter) {    # Shouldn't get multiple images, but just in case...
    return $name . $counter++ . '.' . $type; }
  else {
    $counter++;
    return $name . '.' . $type; } }

#**********************************************************************
__END__

=head1 NAME

C<latexmlmath> - transforms a TeX/LaTeX math expression into various formats.

=head1 SYNOPSIS

latexmlmath [options] I<texmath>

 Options:
 --mathimage=file           converts to image in file
 --mathsvg=file             converts to svg image in file
 --magnification=mag        specifies magnification factor
 --presentationmathml=file  converts to Presentation MathML
 --pmml=file                alias for --presentationmathml
 --linelength=n             do linewrapping of pMML
 --contentmathml=file       convert to Content MathML
 --cmml=file                alias for --contentmathml
 --openmath=file            convert to OpenMath
 --om=file                  alias for --openmath
 --XMath=file               output LaTeXML's internal format
 --noparse                  disables parsing of math
                            (not useful for cMML or openmath)
 --preload=file             loads a style file.
 --includestyles            allows processing raw *.sty files
                            (normally it avoids this)
 --path=dir                 adds a search path for style files.
 --quiet                    reduces verbosity (can repeat)
 --verbose                  increases verbosity (can repeat)
 --strict                   be more strict about errors.
 --documentid=id            assign an id to the document root.
 --debug=package            enables debugging output for the
                            named package
 --VERSION                  show version number and exit.
 --help                     shows this help message.
 --                         ends options

If I<texmath> is '-', C<latexmlmath> reads the TeX from standard input.
If any of the output files are '-', the result is printed on standard output.

=head2 Input notes

Note that, unless you are reading I<texmath> from standard input,
the I<texmath> string will be processed by whatever shell you are using
before C<latexmlmath> even sees it.  This means that many so-called
meta characters, such as backslash and star,  may confuse the
shell or be changed.  Consequently, you will need to quote and/or
slashify the input appropriately.  Most particularly, C<\> will
need to be doubled to C<\\> for C<latexmlmath> to see it as a
control sequence.

Using C<--> to explicitly end the option list is useful for
cases when the math starts with a minus (and would otherwise be interpreted
as an option, probably an unrecognized one). Alternatively,
wrapping the I<texmath> with {} will hide the minus.

Simple examples:

  latexmlmath \\frac{-b\\pm\\sqrt{b^2-4ac}}{2a}
  echo "\\sqrt{b^2-4ac}" | latexmlmath --pmml=quad.mml -

=head1 OPTIONS AND ARGUMENTS

=head2 Conversion Options

These options specify what formats the math should be converted to.
In each case, the destination file is given.  Except for
mathimage, the file can be given as '-', in which case the
result is printed to standard output.

If no conversion option is specified, the default is to output
presentation MathML to standard output.

=over 4

=item C<--mathimage>=I<file>

Requests conversion to png images.

=item C<--mathsvg>=I<file>

Requests conversion to svg images.

=item C<--magnification=>I<factor>

Specifies the magnification used for math image. Default is 1.75.

=item C<--presentationmathml>=I<file>

Requests conversion to Presentation MathML.

=item C<--linelength>I<=number>

(Experimental) Line-breaks the generated Presentation
MathML so that it is no longer than I<number> `characters'.

=item C<--plane1>

Converts the content of Presentation MathML token elements to
the appropriate Unicode Plane-1 codepoints according to the selected font,
when applicable.

=item C<--hackplane1>

Converts the content of Presentation MathML token elements to
the appropriate Unicode Plane-1 codepoints according to the selected font,
but only for the mathvariants double-struck, fraktur and script.
This gives support for current (as of August 2009) versions of
Firefox and MathPlayer, provided a sufficient set of fonts is available (eg. STIX).

=item C<--contentmathml>=I<file>

Requests conversion to Content MathML.
B<Note> that this conversion is only partially implemented.

=item C<--openmath>=I<file>

Requests conversion to OpenMath.
B<Note> that this conversion is only partially implemented.

=item C<--XMath>=I<file>

Requests convertion to LaTeXML's internal format.

=back

=head1 Other Options

=over 4

=item C<--preload>=I<module>

Requests the loading of an optional module or package.  This may be useful if the TeX code
does not specificly require the module (eg. through input or usepackage).
For example, use C<--preload=LaTeX.pool> to force LaTeX mode.

=item C<--includestyles>

This optional allows processing of style files (files with extensions C<sty>,
C<cls>, C<clo>, C<cnf>).  By default, these files are ignored  unless a latexml
implementation of them is found (with an extension of C<ltxml>).

These style files generally fall into two classes:  Those
that merely affect document style are ignorable in the XML.
Others define new markup and document structure, often using
deeper LaTeX macros to achieve their ends.  Although the omission
will lead to other errors (missing macro definitions), it is
unlikely that processing the TeX code in the style file will
lead to a correct document.

=item C<--path>=I<dir>

Add I<dir> to the search paths used when searching for files, modules, style files, etc;
somewhat like TEXINPUTS.  This option can be repeated.

=item C<--documentid>=I<id>

Assigns an ID to the root element of the XML document.  This ID is generally
inherited as the prefix of ID's on all other elements within the document.
This is useful when constructing a site of multiple documents so that
all nodes have unique IDs.

=item C<--quiet>

Reduces the verbosity of output during processing, used twice is pretty silent.

=item C<--verbose>

Increases the verbosity of output during processing, used twice is pretty chatty.
Can be useful for getting more details when errors occur.

=item C<--strict>

Specifies a strict processing mode. By default, undefined control sequences and
invalid document constructs (that violate the DTD) give warning messages, but attempt
to continue processing.  Using --strict makes them generate fatal errors.

=item C<--VERSION>

Shows the version number of the LaTeXML package..

=item C<--debug>=I<package>

Enables debugging output for the named package. The package is given without the leading LaTeXML::.

=item C<--help>

Shows this help message.

=back

=head1 BUGS

This program runs much slower than would seem justified.
This is a result of the relatively slow initialization
including loading TeX and LaTeX macros and the schema.
Normally, this cost would be ammortized over large documents,
whereas, in this case, we're processing a single math expression.

=head1 SEE ALSO

L<latexml>, L<latexmlpost>, L<LaTeXML>

=cut