#!/usr/local/bin/perl
#
sub CHK_CMD_LINE {
if ( $#ARGV < 0 )
{
print "Usage: ", $0, " inputfiles \n";
exit 0;
}
}
sub INIT_VAR {
$docno = "IR4873";
$title = "NIST Interagency Report 4873: Automatic Indexing";
$subtitle = "Automatic Indexing";
$type = "chapter";
$page_chapter = "1";
$page_number = "1";
$author1_first = "Donna";
$author1_last = "Harman";
$author_publisher = "National Institute of Standards and Technology";
$author_editor1_last = "";
$author_editor1_first = "";
$copyright_month = "July";
$copyright_day = "";
$copyright_year = "1992";
$copyright_statement = "";
$copyright_ownership = "National Institute of Standards and Technology";
}
sub INSERT_TOP_METADATA {
print ( OUTPUT " \n");
print ( OUTPUT " $docno \n");
print ( OUTPUT " $title \n");
print ( OUTPUT " $subtitle \n");
print ( OUTPUT " $type \n");
print ( OUTPUT " \n");
print ( OUTPUT " $author1_first $author1_last \n");
print ( OUTPUT " $author_publisher \n");
print ( OUTPUT " $author_editor1_first $author_editor1_last \n");
print ( OUTPUT " \n");
print ( OUTPUT "$copyright_statement \n");
print ( OUTPUT " \n");
print ( OUTPUT " \n");
}
sub INSERT_BOTTOM_METADATA {
print ( OUTPUT " \n");
print ( OUTPUT " \n");
print ( OUTPUT " \n");
}
sub CLEAN_DATA {
if ($_ =~ / /) # only procced if line contains a space
{
$_ =~ s/~/[OCRerr]/g;
$_ =~ s/&/&/g;
$_ =~ s//>/g;
}
}
&CHK_CMD_LINE;
&INIT_VAR;
for $arg (@ARGV)
{
open( INPUT, $arg ) or die "Can't open input file: $arg";
open( OUTPUT, ">$arg.rev") or die "can't open output file: $arg.rev";
&INSERT_TOP_METADATA;
while ()
{
&CLEAN_DATA;
print ( OUTPUT $_);
}
&INSERT_BOTTOM_METADATA;
close( INPUT );
$page_number++;
close( OUTPUT );
}