--- foxridge-archiver/makemeta-vlp.pl	2006/06/27 15:25:40	1.5
+++ foxridge-archiver/makemeta-vlp.pl	2017/03/16 17:00:43	1.10
@@ -1,4 +1,4 @@
-#!/usr/local/bin/perl -w
+#!/usr/bin/perl -w
 
 use strict;
 use XML::LibXML;
@@ -10,7 +10,7 @@ use MPIWGStor;
 $|=1;
 
 # program version
-my $version = "0.2.3 (27.6.2006 ROC)";
+my $version = "0.2.7 (27.8.2010 ROC)";
 my $help = 
 "use: makemeta-vlp [options] file.xml
 options:
@@ -19,6 +19,8 @@ options:
   -replace  replace existing index files
   -online-mode  mode for creating online/permanent files
   -archive-mode  mode for creating archive/data files
+  -access=free  adds free access tag for online-mode
+  -texttool adds texttool tag for online-mode
 ";
 logger("INFO", "makemeta-vlp $version");
 
@@ -27,7 +29,9 @@ logger("INFO", "makemeta-vlp $version");
 
 # generic mappings at top level
 my %gen_map = (
-    'Custom2_Language' => 'meta/lang'
+    'Custom2_Language' => 'meta/lang',
+    'productionComment' => 'meta/image-acquisition/production-comment',
+    'derivedFrom' => 'derived-from/archive-path'
     );
 # sub type switch tag
 my %type_map = (
@@ -117,6 +121,16 @@ my %subtype_map = (
 	'Pages' => 'meta/bib/number-of-pages',
 	'#Cover pages only, articles have been extracted' => 'meta/bib/comment'
     },
+    'Journal' => {
+	'_name' => 'report',
+	'Title' => 'meta/bib/title',
+	'SecondaryTitle' => 'meta/bib/institution',
+	'Author' => 'meta/bib/author',
+	'Place_Published' => 'meta/bib/city',
+	'Year' => 'meta/bib/year',
+	'Date' => 'meta/bib/date',
+	'Pages' => 'meta/bib/pages',
+    },
     'Magazine Article' => {
 	'_name' => 'magazine-article',
 	'Author' => 'meta/bib/author',
@@ -127,6 +141,15 @@ my %subtype_map = (
 	'Date' => 'meta/bib/issue-date',
 	'Pages' => 'meta/bib/pages'
     },
+    'Newspaper Article' => {
+	'_name' => 'newspaper-article',
+	'Author' => 'meta/bib/author',
+	'Title' => 'meta/bib/title',
+	'Year' => 'meta/bib/year',
+	'Secondary_Title' => 'meta/bib/newspaper',
+	'Date' => 'meta/bib/issue-date',
+	'Pages' => 'meta/bib/pages'
+    },
     'Report' => {
 	'_name' => 'report',
 	'Author' => 'meta/bib/author',
@@ -180,10 +203,14 @@ my %lang_map = (
     'Japanese' => 'ja',
     'Dutch' => 'nl',
     'Spanish' => 'es',
-    'Swedish' => 'sv'
+    'Swedish' => 'sv',
+    'Russian' => 'ru',
+    'Polish' => 'pl',
+    'Greek' => 'el'
     );
 # storage fields
 my $arch_id_field = 'ID';
+my $access_free_field = 'online';
 
 #######################################################
 # internal parameters
@@ -219,6 +246,15 @@ logger('DEBUG', "online_mode: $online_mo
 my $archive_mode = (exists $$args{'archive-mode'}) ? $$args{'archive-mode'} : 0;
 logger('DEBUG', "archive_mode: $archive_mode");
 
+# create texttool tag (online mode only)
+my $texttool = (exists $$args{'texttool'}) ? $$args{'texttool'} : 1;
+logger('DEBUG', "texttool: $texttool");
+# image dir for texttool
+my $texttool_img_dir = "pages";
+
+# access type
+my $access_type = (exists $$args{'access'}) ? $$args{'access'} : "";
+
 # index.meta namespace (not really implemented!)
 my $namespace = "";
 
@@ -266,7 +302,7 @@ sub find_arch_dir {
 
 sub find_permanent_dir {
     my ($input_node) = @_;
-    my $online_base = '/mpiwg/online/permanent';
+    my $online_base = $lib_online_dir;
     my $dest_id = sstrip($input_node->findvalue("fm:$arch_id_field"));
     if (! $dest_id) {
 	logger('ERROR', "no ID field for online permanent entry");
@@ -274,7 +310,11 @@ sub find_permanent_dir {
 	return;
     }
     my $dir = "$online_base/lit$dest_id";
-    return $dir;
+    if (-d $dir) {
+        logger('DEBUG', "directory $dir exists"); 
+        return $dir;
+    }
+    return;
 }
 
 
@@ -407,14 +447,38 @@ sub process_fm_entry {
       create_text_path('meta/acquisition/date', stime(time), $index_root, $namespace);
       create_text_path('meta/acquisition/provider/provider-id', 'vlp', $index_root, $namespace);
       create_text_path('meta/acquisition/provider/address', 'Max Planck Institute for the History of Science', $index_root, $namespace);
-      # image acquisition
-      create_text_path('meta/image-acquisition/device', 'Flatbed Scanner' , $index_root, $namespace);
-      create_text_path('meta/image-acquisition/image-type', 'Greyscale' , $index_root, $namespace);
-      create_text_path('meta/image-acquisition/production-comment', 'Raw scans in \'raw\' folder, cleaned pages in \'pages\' folder.' , $index_root, $namespace);
     }
     # media
     create_text_path('media-type', 'image', $index_root, $namespace);
     create_text_path('meta/content-type', 'scanned document', $index_root, $namespace);
+    # access
+    if ($access_type) {
+	if ($access_type eq "free") {
+	    create_element_path('meta/access-conditions/access@type=free', $index_root, $namespace);
+	} else {
+	    my $acc_tag = create_element_path('meta/access-conditions/access@type=institution', $index_root, $namespace);
+	    create_text_path('name', $access_type, $acc_tag, $namespace);
+	}
+    } elsif ($online_mode) {
+        # read access conditions from "online" field in DB dump
+        my $online = sstrip($input_node->findvalue("fm:$access_free_field"));
+        if ($online) {
+	    create_element_path('meta/access-conditions/access@type=free', $index_root, $namespace);
+	} else {
+	    my $acc_tag = create_element_path('meta/access-conditions/access@type=institution', $index_root, $namespace);
+	    create_text_path('name', 'mpiwg', $acc_tag, $namespace);
+	}
+    }
+
+    # texttool tag with image dir
+    if ($online_mode && $texttool) {
+	if ( -d "$doc_dir/$texttool_img_dir" ) {
+	    create_text_path('meta/texttool/image', $texttool_img_dir,$index_root, $namespace);
+	} else {
+            logger('WARNING', "page image directory missing!");
+            $warncnt++;
+        }
+    }
 
     # convert bib entries
     my $cnt = convert_bib($input_node, $index_root, $index_doc);