substitute

#!/usr/bin/perl

# $Id: substitute,v 1.5 2006/08/27 13:43:03 jonathan Exp $

use strict;
use Getopt::Std;

# Note: this should be augmented so that you can do subs in more than one column at a time

my $usage = "Usage: substitute [-k column] [-v] datafile -m mapfile\n";
$usage .=   "       take key values from mapfile and substitute them into column k of datafile\n";
$usage .=   "       with no column given, all substitutions possible are made\n";
$usage .=   "       -v instead of \"key value\", read mapfile as \"value key\"\n";

$usage .=   "       -x if key is not defined in a column, ignore line\n";

$usage .=   "       -C file is not in columns, just substitute all values in map for anything that matches in the file\n";


my %para = ();

getopts('k:vxCm:', \%para);

my $dataFile = $ARGV[0];

my $mapFile = '';
if (!defined($para{'m'})) {
  die ($usage);
}
$mapFile = $para{'m'};

my %map = ();

open MAP, "$mapFile" or die ("Cannot open $mapFile\n");
while (my $line = <MAP>) {
    my (@items) = split (/\s+/, $line);
    if (!defined($para{'v'})) {
	$map{$items[0]} = $items[1];
    } else {
	$map{$items[1]} = $items[0];
    }
}
close MAP;

my $lineCount = 0;
#open DATA, "$dataFile" or die ("Cannot open $dataFile\n");
if (!$para{'C'}) {
    while (my $line = <>) {
	my (@items) = split (/\s+/, $line);
	my $out = '';
	my $validLine = 1;
	for (my $i = 0; $validLine && $i <= $#items; $i++) {
	    if ((defined($para{'k'}) && $i == $para{'k'}-1) ||
		(!defined($para{'k'}))) {
		if (defined ($map{$items[$i]})) {
		    $out .= $map{$items[$i]};
		} else {
		    if (defined($para{'x'})) {
			$validLine = 0;
		    } else {
			$out .= $items[$i];
		    }
		}
	    } else {
		$out .= $items[$i];
	    }
	    if ($i != $#items) {
		$out .= " ";
	    }
	}

	if ($validLine == 1) {
	    print "$out\n";
	}
    }
} else {
    my $whole = '';
    while (my $line = <>) {
#    while (my $line = <DATA>) {
	$whole .= $line;
    }
    foreach my $key (keys %map) {
	$whole =~ s/$key/$map{$key}/g;
    }
    print "$whole";
}

#close DATA;