#!perl -w
# If like me you search since a long time ago the solution to 
# transform WORD DOCUMENT (extension word,rtf ...)in simple text document
# you will be interest by this piece of code
# All right it make a dependance with windows but it to good result
# 
# For this , Just using APACHE / PERL on WIN32 with WORD 97,98 or 2000
# 

use CGI ;
use Win32::OLE;
use Win32::OLE::Const 'Microsoft Word';

my $query=new CGI;
print $query->header;
my $filepath=$query->param('document');
my $html=$query->param('html');
my ( $filename , $inputFileName ) ;
if ( $filepath ne '' ) {    
    	if ($filepath =~ /([^\/\\]+)$/)
    		{ $filename = $1 ; }
    	else
	    	{ $filename= $filepath ;}
	$filename =~ s/\s+//g;
	$inputFileName= "c:\\doctotexte\\$filename";    
	if (!open(WFD,">$inputFileName"))
		{ print "Error last recording file on HD";
  		exit(0);
		}
	while ($bytes_read=read($filepath,$buff,2096))
		{
		binmode WFD;
		print WFD $buff;
		}
	close(WFD);
	my($outputFileName)	= "c:\\doctotexte\\$filename.doc";
	die("Can't find $inputFileName\n") if (! -e $inputFileName);
	unlink($outputFileName);
	my($word) = Win32::OLE -> new('Word.Application', 'Quit');
	my($doc)  = $word -> Documents -> Open($inputFileName);
	$word -> {DisplayAlerts} = 0;	# Stop msg box: Do you wish to save...?
	$word -> {Visible}	 = 1;	# Watch what happens.
	$doc -> SaveAs($outputFileName, wdFormatTextLineBreaks);
	$doc -> Close();
	$word -> Quit();
	# Success.
	open ( FIC , $outputFileName ) ;
	my $texte = join ( '' , <FIC> ) ;
	close ( FIC ) ;
	if ($html eq 'on' ) { 
		$texte =~ s/\n/<br>\n/sgi;
	}
	print $texte ;
	unlink($inputFileName);
	unlink($outputFileName);
} else {
	print 'No Attachment in document field' ;	
}
# --------------------------------------------------------------------