Changeset 14935 for gsdl

Timestamp:

2007-12-20T21:55:31+13:00 (16 years ago)

Author:

davidb

Message:

export.pl script updated to be more consistent with its import.pl counterpart, i.e. support for -incremental and -manifest; also modificed to use plugouts FedoraMETSPlugout and GreenstoneMETSPlugout, rather than this being controlled through a -version option for METSPlugout.

File:

• gsdl/trunk/bin/script/export.pl (modified) (19 diffs)

gsdl/trunk/bin/script/export.pl

@@ -50 +50 @@
 use plugin;
 use plugout;
+use manifest;
 use util;
 use scriptutil;
 use FileHandle;
-use gsprintf;
+use gsprintf 'gsprintf';
 use printusage;
 use parse2;
@@ -61 +62 @@
     [ { 'name' => "hash",
         'desc' => "{import.OIDtype.hash}" },
+      { 'name' => "assigned",
+        'desc' => "{import.OIDtype.assigned}" },
       { 'name' => "incremental",
         'desc' => "{import.OIDtype.incremental}" },
-      { 'name' => "assigned",
-        'desc' => "{import.OIDtype.assigned}" },
       { 'name' => "dirname",
         'desc' => "{import.OIDtype.dirname}" } ];
@@ -72 +73 @@
     [ { 'name' => "DSpace",
         'desc' => "{export.saveas.DSpace}" },
-      { 'name' => "METS",
-        'desc' => "{export.saveas.METS}"},
+      { 'name' => "GreenstoneMETS",
+        'desc' => "{export.saveas.GreenstoneMETS}"},
+      { 'name' => "FedoraMETS",
+        'desc' => "{export.saveas.FedoraMETS}"},
       { 'name' => "GA",
         'desc' => "{export.saveas.GA}"},
@@ -89 +92 @@
 # modegli: The lowest detail mode this argument is visible at in GLI
 
-my $saveas_argument =       
+my $saveas_argument =
       { 'name' => "saveas",
     'desc' => "{export.saveas}",
@@ -102 +105 @@
     [ 
       $saveas_argument,
-      { 'name' => "saveas_version",
-    'desc' => "{export.saveas_version}",
-    'type' => "string",
-    'reqd' => "no",
-       'deft' => "greenstone" },
       { 'name' => "exportdir",
     'desc' => "{export.exportdir}",
@@ -128 +126 @@
     # parsearg left "" as default
     #'deft' => &util::filename_cat ($ENV{'GSDLHOME'}, "collect"),
-        'reqd' => "no",
+    'deft' => "",
+    'reqd' => "no",
+    'hiddengli' => "yes" },
+      { 'name' => "manifest",
+    'desc' => "{import.manifest}",
+    'type' => "string",
+    'deft' => "",
+    'reqd' => "no",
         'hiddengli' => "yes" },
       { 'name' => "listall",
@@ -145 +150 @@
     'reqd' => "no",
         'modegli' => "4" },
+      { 'name' => "incremental",
+    'desc' => "{import.incremental}",
+    'type' => "flag",
+    'hiddengli' => "yes" },
       { 'name' => "keepold",
     'desc' => "{export.keepold}",
     'type' => "flag",
     'reqd' => "no",
-        'hiddengli' => "yes" },
+    'hiddengli' => "yes" },
       { 'name' => "removeold",
     'desc' => "{export.removeold}",
@@ -207 +216 @@
     'reqd' => "no",
         'hiddengli' => "no" },
+      { 'name' => "fedora_namespace",
+    'desc' => "{FedoraMETSPlugout.fedora_namespace}",
+    'type' => "string",
+    'deft' => "greenstone",
+    'reqd' => "no",
+        'hiddengli' => "no" },
       { 'name' => "mapping_file",
     'desc' => "{MARCXMLPlugout.mapping_file}",
@@ -244 +259 @@
                 'args' => [ $saveas_argument ] };
 
-sub gsprintf
+#sub gsprintf
+#{
+#    return &gsprintf::gsprintf(@_);
+#}
+
+
+sub store_collectionmeta
 {
-    return &gsprintf::gsprintf(@_);
+    my ($collectionmeta,$field,$doc_obj) = @_;
+
+    my $section = $doc_obj->get_top_section();
+
+    my $field_hash = $collectionmeta->{$field};
+
+    foreach my $k (keys %$field_hash) 
+    {
+    my $val = $field_hash->{$k};
+
+
+### print STDERR "*** $k = $field_hash->{$k}\n";
+
+    my $md_label = "ex.$field";
+
+
+    if ($k =~ m/^\[l=(.*?)\]$/)
+    {
+
+        my $md_suffix = $1;
+        $md_label .= "^$md_suffix";
+    }
+
+
+    $doc_obj->add_utf8_metadata($section,$md_label, $val);
+    
+    if (($md_label eq "ex.collectionname^en") || ($md_label eq "ex.collectionname"))
+    {
+        $doc_obj->add_utf8_metadata($section,"dc.Title", $val);
+    }
+
+    }
 }
 
@@ -254 +306 @@
 sub main {
     # params
-    my ($language, $verbosity, $importdir, $exportdir, $site, $keepold, $listall,
-    $removeold, $saveas, $saveas_version, $debug, $OIDtype, $OIDmetadata,
-    $maxdocs, $statsfile, $xsltfile, $mapping_file, $out, $faillog, $gs_mode, $collectcfg,
-    $collectdir, $gli,$xslt_mets,$xslt_txt,$group_marc);
+    my ($language, $verbosity, $debug,
+    $importdir, $manifest, $incremental, $keepold,
+    $exportdir, $site, $listall,
+    $removeold, $saveas,
+    $OIDtype, $OIDmetadata,
+    $out, $faillog, $collectdir, $gli,
+    $gs_mode, $collectcfg,
+    $maxdocs, $statsfile, 
+    $mapping_file,
+    $xsltfile, 
+    $xslt_mets, $xslt_txt, $fedora_namespace, $group_marc);
+
     my $xml = 0;
     
     # other vars
-    my ($configfilename, $collection, $export_info_filename, $export_info, $processor, $pluginfo);
+    my ($configfilename, $collection, 
+    $export_info_filename, $export_info, 
+    $processor, $pluginfo);
+
     my $service = "export";
 
@@ -341 +404 @@
     eval {
         # get and check the collection name
-        print STDERR "**** site = $site\n";
         if (($collection = &colcfg::use_collection($site, $collect_name, $collectdir)) eq "") {
         &PrintUsage::print_txt_usage($options, "{export.params}");
@@ -372 +434 @@
         my $plugins = [];
         my @global_opts = ();
-        
+
         # Read in the collection configuration file.
         ($configfilename, $gs_mode) = &colcfg::get_collect_cfg_name($out);
-        
+
         if ($gs_mode eq "gs2") {
         $collectcfg = &colcfg::read_collect_cfg ($configfilename);
@@ -381 +443 @@
         $collectcfg = &colcfg::read_collection_cfg_xml ($configfilename);
         }
+
+        if (defined $collectcfg->{'importdir'} && $importdir eq "") {
+        $importdir = $collectcfg->{'importdir'};
+        }
+        if (defined $collectcfg->{'exportdir'} && $exportdir eq "") {
+        $exportdir = $collectcfg->{'exportdir'};
+        }
+
+        # fill in the default import and export directories if none
+        # were supplied, turn all \ into / and remove trailing /
+        $importdir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "import") if $importdir eq "";
+        $importdir =~ s/[\\\/]+/\//g;
+        $importdir =~ s/\/$//;
+        if (!-e $importdir) {
+        &gsprintf($out, "{import.no_import_dir}\n\n", $importdir);
+        die "\n";
+        }
+
+        $exportdir = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "export") if $exportdir eq "";
+        $exportdir =~ s/[\\\/]+/\//g;
+        $exportdir =~ s/\/$//;
 
         if (defined $collectcfg->{'plugin'}) {
@@ -392 +475 @@
             $verbosity = 2; # the default
         }
-        }
-        if (defined $collectcfg->{'importdir'} && $importdir eq "") {
-        $importdir = $collectcfg->{'importdir'};
-        }
-        if (defined $collectcfg->{'exportdir'} && $exportdir eq "") {
-        $exportdir = $collectcfg->{'exportdir'};
         }
 
@@ -476 +553 @@
         if ($saveas eq "DSpace"){
         $export_info_filename = &util::filename_cat ($exportdir, "contents");
-        } elsif ($saveas eq "METS" || $saveas eq "GA" || $saveas eq "MARC" ) {
+        } elsif ($saveas =~ m/^.*METS$/ || $saveas eq "GA" || $saveas eq "MARC" ) {
         $export_info_filename = &util::filename_cat ($exportdir, "export.inf");
         }
@@ -488 +565 @@
         }
         else{
-        if ($saveas !~ /^(GA|METS|DSpace|MARCXML)$/) {
+        if ($saveas !~ /^(GA|.*METS|DSpace|MARCXML)$/) {
             push @$plugout,"GAPlugout";
         }
@@ -505 +582 @@
         push @$plugout,("-output_handle",$out) if (defined $out);
         push @$plugout,("-xslt_file",$xsltfile) if (defined $xsltfile);
-        push @$plugout,("-group") if ($group_marc && $plugout_name =~ /^MARCXMLPlugout$/);
-        push @$plugout,("-mapping_file",$mapping_file) if (defined $mapping_file && $plugout_name =~ /^MARCXMLPlugout$/);
-        push @$plugout,("-saveas_version",$saveas_version) if (defined $saveas_version && $plugout_name =~ /^METSPlugout$/);
-        push @$plugout,("-xslt_mets",$xslt_mets) if (defined $xslt_mets && $plugout_name =~ /^METSPlugout$/);
-        push @$plugout,("-xslt_txt",$xslt_txt) if (defined $xslt_txt && $plugout_name =~ /^METSPlugout$/);
+        push @$plugout,("-group") if ($group_marc && $plugout_name =~ m/^MARCXMLPlugout$/);
+        push @$plugout,("-mapping_file",$mapping_file) if (defined $mapping_file && $plugout_name =~ m/^MARCXMLPlugout$/);
+        push @$plugout,("-xslt_mets",$xslt_mets) if (defined $xslt_mets && $plugout_name =~ m/^.*METSPlugout$/);
+        push @$plugout,("-xslt_txt",$xslt_txt) if (defined $xslt_txt && $plugout_name =~ m/^.*METSPlugout$/);
+        push @$plugout,("-fedora_namespace",$fedora_namespace) if (defined $fedora_namespace && $plugout_name eq "FedoraMETSPlugout");
+
         $processor = &plugout::load_plugout($plugout);    
         
@@ -521 +599 @@
         # process the import directory
         &plugin::read ($pluginfo, $importdir, "", {}, $processor, $maxdocs,0, $gli);
+
+        if ($saveas eq "FedoraMETS")
+        {
+        # create collection "doc obj" for Fedora that contains
+        # collection-level metadata
+
+        my $doc_obj = new doc($configfilename,"nonindexed_doc");
+        $doc_obj->set_OID("collection");
+
+        my $col_name = undef;
+        my $col_meta = $collectcfg->{'collectionmeta'};
+        if (defined $col_meta)
+        {
+            store_collectionmeta($col_meta,"collectionname",$doc_obj);
+            store_collectionmeta($col_meta,"collectioextra",$doc_obj);
+
+        }
+
+        $processor->process($doc_obj);
+        }
         
         &plugin::end($pluginfo, $processor);
@@ -529 +627 @@
         $processor->close_file_output() if $groupsize > 1;
         $processor->close_group_output() if $processor->is_group();
-        if ($saveas eq "METS") {
+        if ($saveas =~ m/^.*METS$/) {
         $export_info->save_info($export_info_filename);
         }