#!/usr/local/bin/perl

# soif2mwi - convert SOIF record on STDIN to NWI record on STDOUT
#
# Author: andy powell <a.powell@ukoln.ac.uk>
#
# $Id: soif2nwi,v 1.1 1997/06/24 22:07:18 lisap Exp $

$ENV{'HARVEST_HOME'} = "/usr/local/harvest" if (!defined($ENV{'HARVEST_HOME'}));
unshift(@INC, "$ENV{'HARVEST_HOME'}/lib");      # use local files
require "soif.pl";
use HTML::Entities (); # from LWP

# SOIF to NWI mappings...
# SOIF				NWI
# ----				---
# Abstract
# Author
# Body				<ip>
# Description
# File-Size			<si><by>
# Full-Text			<ip>
# Gatherer-Host
# Gatherer-Name
# Gatherer-Port
# Gatherer-Version
# Headings			<lsi><dh>
# Update-Time
# Keywords
# Last-Modification-Time	<dm>
# MD5				<si><md5>
# Refresh-Rate
# Time-to-Live
# Title				<ti>
# Type				<av><ty>
# URL-References
# 
# URL				<av><avli>

%Mappings = (
    'body', 'ip',
    'file-size', 'by',
    'full-text', 'ip',
    'headings', 'dh',
    'last-modification-time', 'dm',
    'md5', 'md5',
    'title', 'ti',
    'type', 'ty',
);

($ttype,$url,%SOIF) = &soif'parse;
$url =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C", hex($1))/eg;

foreach (keys SOIF) {
    $value = $SOIF{$_};
    tr/A-Z/a-z/;
#    if ($soif'debug == 1) {
#        $value =~ s/\s+/ /g;
#	print "SOIF-$_ = $value\n";
#    }
    next unless ($Mappings{$_});
    $value =~ s/\cM/ /;
    $value =~ s/\s+/ /g;
    if ($_ =~ /-time$/) {
	$value = localtime($value);
    }
    $value = HTML::Entities::encode_entities($value);
    $NWI{$Mappings{$_}} = $value;
}

$NWI{'avli'} = $url;

unless ($NWI{'ti'}) {
    if ($NWI{'dh'}) {
	$NWI{'ti'} = $NWI{'dh'};
	$NWI{'ti'} =~ s/\n.*//;
    }
}

print "<nwi>\n";
print "<ti> ".$NWI{'ti'}."\n" if $NWI{'ti'};
print "<dm> ".$NWI{'dm'}."\n" if $NWI{'dm'};
if ($NWI{'by'} || $NWI{'md5'}) {
    print "<si>\n";
    print " <by> ".$NWI{'by'}."\n" if $NWI{'by'};
    print " <md5> ".$NWI{'md5'}."\n" if $NWI{'md5'};
    print "</si>\n";
}
if ($NWI{'avli'} || $NWI{'ty'}) {
    print "<av>\n";
    print " <avli> ".$NWI{'avli'}."\n" if $NWI{'avli'};
    print " <ty> ".$NWI{'ty'}."\n" if $NWI{'ty'};
    print "</av>\n";
}
# Ignore Headings (<lsi><dh>) for now.
if ($NWI{'ip'}) {
    print "<ip>\n";
    write;
    print "</ip>\n";
}
print "</nwi>\n";

exit 0;

format =
^<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<~~
$NWI{'ip'}
.