#!/bin/perl -w

use strict;
use File::Temp ();


#%%fid=fopen('./docs.dat','w');
#%%for i=1:length(docs)
#%%  fprintf(fid,'%d ',docs{i});
#%%  fprintf(fid,'\n');
#%%end
#%%fclose(fid);
#%%eval('!perl ./helperFunctions/docHist.pl ./docs.dat tf');
#

my $tmpfh = new File::Temp( UNLINK => 0, SUFFIX => '.dat' );
my $tmpfn = $tmpfh->filename;

my $row_cnt=0;
open(DOCS,$ARGV[0]) or die "couldn't open $ARGV[0]:$!";

while(<DOCS>){
    chomp;
    ++$row_cnt;
    my %termhist;
    map {++$termhist{$_}} split(" ",$_);
    foreach my $word (sort keys %termhist) {
        print $tmpfh "$row_cnt\t$word\t$termhist{$word}\n";
    }
}
close(DOCS);
close($tmpfh);

open(MATLAB,"|-","/usr/local/bin/matlab -nodesktop -nosplash >/dev/null")
    or die "couldn't open matlab: $!";
print MATLAB <<InlineCode;
% create term histogram for each doc (sparse) matrix
termHisttmp=load('$tmpfn');
$ARGV[1]=spconvert(termHisttmp);
save '$ARGV[1].mat' $ARGV[1];
clear
InlineCode
close(MATLAB);
unlink $tmpfn;