/tools/filters/condense_characters.pl

  1#! /usr/bin/perl -w
  2
  3use strict;
  4use warnings;
  5
  6# condenses all consecutive characters of one type
  7# convert_characters.pl [input] [character] [output]
  8
  9die "Check arguments" unless @ARGV == 3;
 10
 11my $inputfile = $ARGV[0];
 12my $character = $ARGV[1];
 13my $outputfile = $ARGV[2];
 14
 15
 16my $convert_from;
 17my $convert_to;
 18
 19
 20if ($character eq "s")
 21{
 22    $convert_from = '\s';
 23}
 24elsif ($character eq "T")
 25{
 26    $convert_from = '\t';
 27}
 28elsif ($character eq "Sp")
 29{
 30    $convert_from = " ";
 31}
 32elsif ($character eq "Dt")
 33{
 34    $convert_from = '\.';
 35}
 36elsif ($character eq "C")
 37{
 38    $convert_from = ",";
 39}
 40elsif ($character eq "D")
 41{
 42    $convert_from = "-";
 43}
 44elsif ($character eq "U")
 45{
 46    $convert_from = "_";
 47}
 48elsif ($character eq "P")
 49{
 50    $convert_from = '\|';
 51}
 52else
 53{
 54    die "Invalid value specified for convert from\n";
 55}
 56
 57
 58if ($character eq "T")
 59{
 60    $convert_to = "\t";
 61}
 62elsif ($character eq "Sp")
 63{
 64    $convert_to = " ";
 65}
 66elsif ($character eq "Dt")
 67{
 68    $convert_to = "\.";
 69}
 70elsif ($character eq "C")
 71{
 72    $convert_to = ",";
 73}
 74elsif ($character eq "D")
 75{
 76    $convert_to = "-";
 77}
 78elsif ($character eq "U")
 79{
 80    $convert_to = "_";
 81}
 82elsif ($character eq "P")
 83{
 84    $convert_to = "|";
 85}
 86else
 87{
 88    die "Invalid value specified for Convert to\n";
 89}
 90
 91my $fhIn;
 92open ($fhIn, "< $inputfile") or die "Cannot open source file";
 93
 94my $fhOut;
 95open ($fhOut, "> $outputfile");
 96
 97while (<$fhIn>)
 98{
 99    my $thisLine = $_;
100    chomp $thisLine;
101    $thisLine =~ s/${convert_from}+/$convert_to/g;
102    print $fhOut $thisLine,"\n";    
103}
104close ($fhIn) or die "Cannot close source file";
105close ($fhOut) or die "Cannot close output file";