#!/usr/local/bin/perl

# This script was designed to help create MARC 
# records for individual titles in the 
# Chadwyck-Healey databases that are made 
# available to VIVA insitutions by the 
# University of Virginia:
#
# - African American Poetry (1760-1900)
# - American Poetry Database
# - English Poetry Full-Text Database
# - English Verse Drama
#
# This script will take UVa's records and convert them 
# into the format that is understood by the freely available
# Perl module MARC.pm
#
# http://marcpm.sourceforge.net
#
# First you run this script on a system that supports perl 
# to convert the UVa file into MARMAKR format. And then 
# you use MARC.pm to create a properly formatted 
# USMARC record, which you should be able to load into your 
# system. If you don't want to use MARC.pm you can also used the
# Library of Congress MARCMaker tool.
#
# Let me know if you can't get it to work, or if you would 
# like help in customizing it. You can add fields that you 
# would like to be inserted into the MARC record by adding 
# them in the section beginning with line 136, and begins 
# with the words "put in 000 field..."
#
# Ed Summers
# ed@cheetahmail.com

($in,$out) = @ARGV;

unless ($in && $out) {
  print "input file : ";
  $in=<STDIN>;
  print "output file : ";
  $out=<STDIN>;
}

open (IN,"$in");
open (OUT,">$out");

$count=0;
READ: while (<IN>) {
   
   #count how many lines read in
   $count++;
      
   #change .245. to =245
   $_=~s/\.(\d{3})\./=$1/;
   
   #if there are no indicators then convert
   #ie. replace "=500  |" with "=500  \\|"
   $_=~s/(=\d{3}) +\|/$1  \\\\\|/;
   
   #if there are indicators then convert
   #ie. replace "=245 10|" with "=245  10|"
   $_=~s/(=\d{3}) ([\d ]{2})\|/$1  $2\|/;
   
   #convert missing first indicator to \
   #ie. replace "=245   1|" with "=245  \1|"
   $_=~s/   (\d)\}/  \\$1\|/;
   
   #convert missing second indicators to \
   #ie. replace "=245  1 |" with "=245  1\|"
   $_=~s/  (\d) \|/  $1\\\|/;
   
   #convert | diacritic to $ diacritic
   $_=~s/\|/\$/g;
   
   #add space to end of line
   $_=~s/\n$/ \n/;
   
   #remove all of the "[SGML-encoded machine-readable 
   #transcript]" GMD's and replace with "[computer file]" 
   #you can comment this section out if you would 
   #like to leave the GMD the way it is
   if (/\$h.*SGML.*/) {
      #if GMD is on one line
      if (/\$h.*transcript/) {
         $_=~s/\$h.*transcript(\.)?\]?/\$h\[computer file\]\1/;
         }
      #else if the GMD spans more than one line
      else {
         chop;
         $_.=<IN>;
         redo READ;
         }
      }
   
   #replace UVA's hotlink text to ODU's you can change this
   #by changing "Connect to this resource \(Authorized users 
   #only\)" to what you want. You will probably need to put 
   #a \ before any non-letter or number characters
   #$_=~s/Available only to UVa and VIVA users/Connect to this resource \(Authorized users only\)/;
   
   #if it the line contains a url then remove any pesky spaces
   #if ($_=~/http:\/\//) {
   #   $_=~s/ //g;
   #   }
   
   #remove indicator spaces and |a from OO6, 007 and 008
   if ($_=~/=00[678]/) {
      $_=~s/\\\\\$a//;
      }
   
   #remove information from 008 concerning computer file
   #since these records should use the book format (mostly
   #textual. you may want to comment this section out if you 
   #would like to keep your 008 field the way it is
   if ($_=~/=008/) {
      substr($_,25,16)="                ";
      }
      
   #if it is the first document boundary then don't print a newline
   if ($count == "1" and $_=~/DOCUMENT BOUNDARY/) {
      $record="1";
      next;
      }
   
   #if it is a new record increment the record counter and 
   #print out a newline to signal a new record
   elsif ($count != "1" and $_=~/DOCUMENT BOUNDARY/) {
      $record++;
      print OUT "\n";
      next;
      }
   
   #remove 001 field
   #elsif ($_=~/=001/) {
   #   next;
   #   }
   
   #put in 000 field that the marcmakr program understands
   #you can also use this area to add any marc fields you like..
   #they should be put in order when your record is processed
   elsif ($_=~/=000/) {
      print OUT "=LDR  00000nam\\\\2200000\\a\\4500\n";
      print OUT "=006  m        d\n";
      print OUT "=007  cr un \n";
      print OUT "=590  \\\\\$aElectronic access sponsored by VIVA, the Virtual Library of Virginia.\n";
      }
         
   #do not print out FORM=MRDF or FORM=MARC which isn't understood by the marcmakr program
   elsif ($_=~/FORM=MRDF/ or $_=~/FORM=MARC/) {
      next;
      }
      
   #do not print out hotlink to chadwyck healey
   #this section is commented out, but you can uncomment it
   #by removing the pound signs from the beginning of 
   #the next three lines
   #elsif ($_=~/=856  0\\\$achadwyck/) {
   #   $linkflag="1";
   #   next;
   #   }
      
   #do not print out hotlink about non-viva, non-odu users
   #this section is also commented out, but you can uncomment
   #it by removing the pound signs from the beginning of the 
   #next three lines.
   #elsif ($_=~/Non-UVa,non-VIVAusers/) {
   #   $linkflag="1";
   #   next;
   #   }
   
   #if the line is a carry over from one of the links that should
   #be ignored then do not print out
   elsif (not($_=~/^=/) and $linkflag=="1") {
      $linkflag="0";
      next;
      }
   
   #if the line is the 596 then ignore it
   elsif ($_=~/=596/) {
      next;
      }
   
   #print out the line!
   else {
      print OUT "$_"; }
   }

print "\n$record records processed\n";