view release on metacpan or  search on metacpan

lib/Alvis/NLPPlatform/Convert.pm  view on Meta::CPAN

package Alvis::NLPPlatform::Convert;

#  The name
# of the directory if defined, in the proirity order, from the variable
# C<SPOOLDIR> in the section C<CONVERTERS>, C<SPOOLDIR> in the section
# C<alvis_connection>. If any name can be determine, 

#     if (defined $config->{"CONVERTERS"}->{"SPOOLDIR"}) {
# 	$ODIR = $config->{"CONVERTERS"}->{"SPOOLDIR"};
#     } else {
# 	if (defined $config->{"alvis_connection"}->{"SPOOLDIR"}) {
# 	    $ODIR = $config->{"alvis_connection"}->{"SPOOLDIR"};
# 	} else {


use strict;
use warnings;
use utf8;
no utf8;

use Alvis::NLPPlatform::Document;
use File::MMagic;
use File::Basename;
use File::Path qw(mkpath);
use File::Touch;

use Data::Dumper;
use Cwd;

our $VERSION=$Alvis::NLPPlatform::VERSION;

sub load_MagicNumber
{
    my $config = shift;

    my $mm = new File::MMagic; # use internal magic file

    print STDERR "Loading complementary magic number ... ";
    
    open FILEM, $config->{"CONVERTERS"}->{"SupplMagicFile"} or die "No such file or directory\n";

    my $line;
    while($line = <FILEM>) {
	chomp $line;
	$line =~ s/\s*\#.*//;
	$line =~ s/^\s*//;
	
	if ($line ne "") {
	    $mm->addMagicEntry($line);
	}
    }
    print STDERR "done\n";

    return($mm);
}

sub conversion_file_to_alvis_xml
{

#    warn join("|", @_) . "\n";
    my $file = shift;
    my $AlvisConv = shift;
    my $config = shift;
    my $mm = shift;


    my $outfile = $config->{ALVISTMP} . "/" . basename($file) . ".out";
    my $infile = $file;
    my $outfile_path;

    my $outdata;
    
    my $type = &get_type_file($infile, $mm);

    if (($type !~ /xml/i) && ($type !~ /html/i)) {

	print STDERR "Conversion of file ...\n";

# 	print Dumper $config;
	
	if ((defined $config->{"CONVERTERS"}->{$type}) && ($config->{"CONVERTERS"}->{$type} ne "")) {
	    
	    print STDERR "Converting to HTML ...\n" ;
	    

	    my $commandline = $config->{"CONVERTERS"}->{$type} . " \"$infile\" > \"$outfile\"";