#!/usr/local/bin/perl -w
# -*- cperl -*-
=head1 NAME

getRange.pl

=head1 SYNOPSYS

 RCS:$Id$

=head1 DESCRIPTION

=head1 HISTORY

 ORIGIN: created from templateApp.pl version 3.4 by Min-Yen Kan <kanmy@comp.nus.edu.sg>

 RCS:$Log$

=cut

require 5.0;
use Getopt::Std;
use strict 'vars';
# use diagnostics;

### USER customizable section
my $tmpfile .= $0; $tmpfile =~ s/[\.\/]//g;
$tmpfile .= $$ . time;
if ($tmpfile =~ /^([-\@\w.]+)$/) { $tmpfile = $1; }		      # untaint tmpfile variable
$tmpfile = "/tmp/" . $tmpfile;
$0 =~ /([^\/]+)$/; my $progname = $1;
my $outputVersion = "1.0";

# Citeseer sf / url conversion
my $citeseerPapersDir = "/export/bulk/citeseer/cs/papers/";
my %CSURL2FS = ('/' => 'zSz',
		'$' => 'zDz',
              '+' => 'zPz',
              '*' => 'zAz',
              '@' => 'zTz',
              '%' => 'zCz',
              '?' => 'zQz',
              '\\' => 'zBz',

              'zSz' => '/',
              'zDz' => '$',
		'zPz' => '+',
		'zAz' => '*',
		'zTz' => '@',
		'zCz' => '%',
		'zQz' => '?',
		'zBz' => '\\');

### END user customizable section

### Ctrl-C handler
sub quitHandler {
  print STDERR "\n# $progname fatal\t\tReceived a 'SIGINT'\n# $progname - exiting cleanly\n";
  exit;
}

### HELP Sub-procedure
sub Help {
  print STDERR "usage: $progname -h\t\t\t\t[invokes help]\n";
  print STDERR "       $progname -v\t\t\t\t[invokes version]\n";
  print STDERR "       $progname [-q] start-end csMetadata_tsv_file\n";
  print STDERR "Options:\n";
  print STDERR "\t-q\tQuiet Mode (don't echo license)\n";
  print STDERR "\n";
  print STDERR "Will accept input on STDIN as a single file.\n";
  print STDERR "\n";
}

### VERSION Sub-procedure
sub Version {
  if (system ("perldoc $0")) {
    die "Need \"perldoc\" in PATH to print version information";
  }
  exit;
}

sub License {
  print STDERR "# Copyright 2005 \251 by Min-Yen Kan\n";
}

###
### MAIN program
###

my $cmdLine = $0 . " " . join (" ", @ARGV);
if ($#ARGV == -1) { 		        # invoked with no arguments, possible error in execution? 
  Help(); exit(0);
}

$SIG{'INT'} = 'quitHandler';
getopts ('hqv');

our ($opt_q, $opt_v, $opt_h);
# use (!defined $opt_X) for options with arguments
if (!$opt_q) { License(); }		# call License, if asked for
if ($opt_v) { Version(); exit(0); }	# call Version, if asked for
if ($opt_h) { Help(); exit (0); }	# call help, if asked for

my $range = shift;
my ($start,$end) = split (/\-/, $range);
if (!$opt_q) { print STDERR "# starting at $start, ending at $end\n"; }

## standardize input stream (either STDIN on first arg on command line)
my $fh;
my $filename;
if ($filename = shift) {
 NEWFILE:
  if (!(-e $filename)) { die "# $progname crash\t\tFile \"$filename\" doesn't exist"; }
  open (*IF, $filename) || die "# $progname crash\t\tCan't open \"$filename\"";
  $fh = "IF";
} else {
  $filename = "<STDIN>";
  $fh = "STDIN";
}

`mkdir /tmp/$range`;
my @elts;
my $mode = 0;
my $numRecords = 0;
open (OF, ">/tmp/$range/list.tsv");
while (<$fh>) {
  if (/^\#/) { next; }			# skip comments
  elsif (/^\s+$/) { next; }		# skip blank lines
  else {
    chomp;
    @elts = split(/\t/);
    if ($mode == 0 && $elts[0] eq $start) {
      $mode = 1; print STDERR "# found start\n";
    }
    if ($mode == 1) {
      my $dir = $elts[6];
#      $dir =~ s/cs/cs_overflow/g;
      my $paperLoc = &noExtensionWildCard(&url2sf($elts[5]));
      print "cp \"$citeseerPapersDir/$dir/$paperLoc\".* /tmp/$range/\n";
      `cp "$citeseerPapersDir/$dir/$paperLoc".* /tmp/$range/`;
      print OF "$elts[0]\t$dir\t$elts[5]\t$paperLoc";
      $numRecords++;
    }
    if ($mode == 1 && $elts[0] eq $end) {
      $mode = 0;
      print STDERR "# found end -- $numRecords exported\n";
      last;
    }
    if ($elts[0] % 100 == 0) { print STDERR "[$elts[0]]"; }
  }
}
close (OF);
close ($fh);

# create zip file
`cd /tmp; zip -r /tmp/$range.zip $range`;
#`rm -Rf /tmp/$range`;

###
### END of main program
###

sub url2sf {
  my $l = shift;
  $l =~ s/([\/\$\+\*\@\%\?])/$CSURL2FS{$1}/g;
  return $l;
}

sub noExtensionWildCard {
  my $l = shift;
  $l =~ s/\.(pdf|ps|ps.gz|ps.z|z|pdf.z|pdf.gz)$//ig;
  return $l;
}