--- nsdl/nsdlloncapaorg/harvester.pl	2003/07/28 20:14:17	1.2
+++ nsdl/nsdlloncapaorg/harvester.pl	2003/07/29 14:37:51	1.4
@@ -12,6 +12,9 @@ use strict;
 use LWP::UserAgent;
 use Getopt::Std;
 use Digest::MD5 qw(md5_hex);
+use IO::File;
+
+my $basepath='/home/httpd/cgi-bin/OAI-XMLFile/XMLFile/nsdlexport/data';
 
 my $pub_month;
 my $pub_year;
@@ -60,6 +63,7 @@ if ( $response->is_success ) {
 #}
 
 my %records = ();;
+
 foreach my $metadata (@loncapa) {
 	chomp $metadata;
 	$metadata=~s/[^\w\d\s\.\;\:\,\|\/]/ /gs;
@@ -82,8 +86,10 @@ foreach my $metadata (@loncapa) {
 	next if ( ($subject eq 'Sample') || ($subject eq 'Something') );
 	my $resourceurl = 'http://nsdl.lon-capa.org' . $tkline[3];
         my $baseid=$tkline[3];
+	my ($adom,$auname)=($baseid=~/^\/res\/(\w+)\/(\w+)\//);
 	$baseid=~s/\W/\_/g;
 	$baseid=~s/^\_res\_//g;
+	my $fileid=md5_hex($baseid);
 
 	next if ( $resourceurl =~ /(.*)\/demo\/(.*)/ );
 	my $keywords = $tkline[4];
@@ -128,8 +134,9 @@ foreach my $metadata (@loncapa) {
 	next if ( $language ne 'seniso');
 	my $primary_language='en-US';
 	my $creation_date = $tkline[10];
-	my ($pub_year,$pub_month,$pub_day) = ( $creation_date =~ /^(\d{4})-(\d{2})-(\d{2})\s(\d{2}):(\d{2}):(\d{2})$/ );
+	my ($pub_year,$pub_month,$pub_day) = ( $creation_date =~ /^(\d{4}) (\d{2}) (\d{2})\s(\d{2}):(\d{2}):(\d{2})$/ );
 	my $revision_date = $tkline[11];
+	my ($rev_year,$rev_month,$rev_day) = ( $revision_date =~ /^(\d{4}) (\d{2}) (\d{2})\s(\d{2}):(\d{2}):(\d{2})$/ );
 	my $owner = $tkline[12];
 	my $rights_description;
 	my $copyright = $tkline[13]; # public,domain,default,private (skip if private and domain)
@@ -147,18 +154,32 @@ foreach my $metadata (@loncapa) {
 	# Private means open only to author of material
 	next if ( $copyright eq 'private');
 	my $platform = "5";     # HTML Browser (not specified but construed from metadata)
-	print (<<ENDMETA);
-<rdf about="lon-capa.nsdl.collections/$baseid">
-    <dc:title>$title</dc:title>
-    <dc:creator>$author_fname $author_lname</dc:creator>
-    <dc:subject>$keywords</dc:subject>
-    <dc:subject>$subject</dc:subject>
-    <dc:identifier scheme="URI">$resourceurl</dc:identifier>
-    <dc:language>$primary_language</dc:language>
-    <dc:description>$abstract<dc:description>
-    <dc:date>$revision_date</dc:date>
-</rdf>
-
+#
+# Create path
+#
+	unless (-e $basepath.'/'.$adom) { mkdir($basepath.'/'.$adom); }
+	unless (-e $basepath.'/'.$adom.'/'.$auname) { 
+	    mkdir($basepath.'/'.$adom.'/'.$auname) || die 'Could not create '.$basepath.'/'.$adom.'/'.$auname;
+	}
+	open(XML,'>'.$basepath.'/'.$adom.'/'.$auname.'/'.$baseid.'.xml');
+	print XML (<<ENDMETA);
+<?xml version="1.0" encoding="UTF-8"?>
+
+<oaidc:dc xmlns="http://purl.org/dc/elements/1.1/" 
+          xmlns:oaidc="http://www.openarchives.org/OAI/2.0/oai_dc/" 
+          xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
+          xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ 
+                              http://www.openarchives.org/OAI/2.0/oai_dc.xsd"
+>
+    <title>$title</title>
+    <creator>$author_fname $author_lname</creator>
+    <identifier>$resourceurl</identifier>
+    <subject>$keywords</subject>
+    <subject>$subject</subject>
+    <language>$primary_language</language>
+    <description>$abstract</description>
+    <date>$rev_year-$rev_month-$rev_day</date>
+</oaidc:dc>
 ENDMETA
+      close (XML);
 }
-