Changeset 22421 for main/trunk/greenstone2/perllib/inexport.pm

Timestamp:

2010-07-18T16:36:56+12:00 (14 years ago)

Author:

davidb

Message:

Continued work on refactoring code to have better shared support for import.pl and export.pl

File:

• main/trunk/greenstone2/perllib/inexport.pm (modified) (18 diffs)

main/trunk/greenstone2/perllib/inexport.pm

@@ -51 +51 @@
 {
     my $class = shift (@_);
-    my ($argv,$options) = @_;
-
-    my $self = { 'xml' => 0 };
+    my ($mode,$argv,$options,$opt_listall_options) = @_;
+
+    my $self = { 'xml' => 0, 'mode' => $mode };
 
     # general options available to all plugins
@@ -72 +72 @@
     }
 
+    if ($self->{'listall'}) {
+    if ($self->{'xml'}) {
+        &PrintUsage::print_xml_usage($opt_listall_options);
+    }
+    else
+    {
+        &PrintUsage::print_txt_usage($opt_listall_options,"{export.params}");
+    }
+    die "\n";
+    }
+
+
     if ($self->{'xml'}) {
         &PrintUsage::print_xml_usage($options);
@@ -151 +163 @@
 
     # Read in the collection configuration file.
-    my ($configfilename, $gs_mode) = &colcfg::get_collect_cfg_name($out);
-    my $collectcfg = &colcfg::read_collection_cfg ($configfilename, $gs_mode);
-
-    return $collectcfg;
+    my ($config_filename, $gs_mode) = &colcfg::get_collect_cfg_name($out);
+    my $collectcfg = &colcfg::read_collection_cfg ($config_filename, $gs_mode);
+
+    return ($config_filename,$collectcfg);
 }
 
@@ -160 +172 @@
 {
     my $self = shift @_;
-    my ($inexport_mode,$collectcfg) = @_;
+    my ($collectcfg) = @_;
+
+    my $inexport_mode = $self->{'mode'};
 
     my $verbosity  = $self->{'verbosity'};
-    print STDERR "**** verbosity = $verbosity\n\n\n";
-
     my $debug      = $self->{'debug'};
     my $importdir  = $self->{'importdir'};
@@ -218 +230 @@
     }
     }
+    $self->{'verbosity'} = $verbosity;
+
     if (defined $collectcfg->{'manifest'} && $self->{'manifest'} eq "") {
     $self->{'manifest'} = $collectcfg->{'manifest'};
@@ -235 +249 @@
     }
     }
-    if ($self->{'groupsize'} == 1) {
+
+    if ((defined $self->{'groupsize'}) && ($self->{'groupsize'} == 1)) {
     if (defined $collectcfg->{'groupsize'} && $collectcfg->{'groupsize'} =~ /\d+/) {
         $self->{'groupsize'} = $collectcfg->{'groupsize'};
@@ -287 +302 @@
        
     # check keepold and removeold
+    my $checkdir = ($inexport_mode eq "import") ? "archives" : "export";
+
     my ($removeold, $keepold, $incremental, $incremental_mode) 
     = &scriptutil::check_removeold_and_keepold($self->{'removeold'}, $self->{'keepold'}, 
-                           $self->{'incremental'}, "archives", 
+                           $self->{'incremental'}, $checkdir, 
                            $collectcfg);
 
@@ -301 +318 @@
 {
     my $self = shift @_;
-    my ($inexport_mode,$collectcfg) = @_;
+    my ($config_filename,$collectcfg) = @_;
+
+    my $inexport_mode = $self->{'mode'};
 
     my $verbosity   = $self->{'verbosity'};
@@ -330 +349 @@
     my $removesuffix = $self->{'removesuffix'};
 
-    my $gli         = $self->{'gli'};
-
-    print STDERR "<Import>\n" if $gli;
+    my $gli          = $self->{'gli'};
+
+    # related to export
+    my $xsltfile         = $self->{'xsltfile'};
+    my $group_marc       = $self->{'group_marc'};
+    my $mapping_file     = $self->{'mapping_file'};
+    my $xslt_mets        = $self->{'xslt_mets'};
+    my $xslt_txt         = $self->{'xslt_txt'};
+    my $fedora_namespace = $self->{'fedora_namespace'};
+
+    if ($inexport_mode eq "import") {
+    print STDERR "<Import>\n" if $gli;
+    }
+    else {
+    print STDERR "<export>\n" if $gli;
+    }
     
     my $manifest_lookup = new manifest($collectcfg->{'infodbtype'},$archivedir);
@@ -358 +390 @@
     #some global options for the plugins
     my @global_opts = ();
-
 
     my $pluginfo = &plugin::load_plugins ($plugins, $verbosity, $out, $faillog, \@global_opts, $incremental_mode);
@@ -403 +434 @@
 
     ####Use Plugout####
-    my ($plugout); 
-    if (defined $collectcfg->{'plugout'}) {
-    # If a plugout was specified in the collect.cfg file, assume it is sensible
-    # We can't check the name because it could be anything, if it is a custom plugout
-    $plugout = $collectcfg->{'plugout'};
-    }
-    else{
-    if ($saveas !~ /^(GreenstoneXML|GreenstoneMETS)$/) {
-        push @$plugout,"GreenstoneXMLPlugout";
+    my $plugout; 
+
+    if ($inexport_mode eq "import") {
+    if (defined $collectcfg->{'plugout'}) {
+        # If a plugout was specified in the collect.cfg file, assume it is sensible
+        # We can't check the name because it could be anything, if it is a custom plugout
+        $plugout = $collectcfg->{'plugout'};
     }
     else{
-        push @$plugout,$saveas."Plugout";
-    }
-    }
-
-    push @$plugout,("-output_info",$archive_info) if (defined $archive_info); 
-    push @$plugout,("-verbosity",$verbosity)      if (defined $verbosity);
-    push @$plugout,("-gzip_output")               if ($gzip);
-    push @$plugout,("-group_size",$groupsize)     if (defined $groupsize);
-    push @$plugout,("-output_handle",$out)        if (defined);
-    push @$plugout,("-debug")                     if ($debug);
-    
+        if ($saveas !~ /^(GreenstoneXML|GreenstoneMETS)$/) {
+        push @$plugout,"GreenstoneXMLPlugout";
+        }
+        else{
+        push @$plugout,$saveas."Plugout";
+        }
+    }
+    }
+    else {
+    if (defined $collectcfg->{'plugout'} && $collectcfg->{'plugout'} =~ /^(.*METS|DSpace|MARCXML)Plugout/) {
+        $plugout = $collectcfg->{'plugout'};
+    }
+    else{
+        if ($saveas !~ /^(GreenstoneMETS|FedoraMETS|DSpace|MARCXML)$/) {
+        push @$plugout,"GreenstoneMETSPlugout";
+        }
+        else{
+        push @$plugout,$saveas."Plugout";
+        }
+    }
+    }
+    
+    my $plugout_name = $plugout->[0];
+
+    push @$plugout,("-output_info",$archive_info)  if (defined $archive_info); 
+    push @$plugout,("-verbosity",$verbosity)       if (defined $verbosity);
+    push @$plugout,("-debug")                      if ($debug);
+    push @$plugout,("-group_size",$groupsize)      if (defined $groupsize);
+    push @$plugout,("-gzip_output")                if ($gzip);
+    push @$plugout,("-output_handle",$out)         if (defined $out);
+
+    push @$plugout,("-xslt_file",$xsltfile)        if (defined $xsltfile && $xsltfile ne "");
+
+    if ($plugout_name =~ m/^MARCXMLPlugout$/) {
+    push @$plugout,("-group")                      if ($group_marc);
+    push @$plugout,("-mapping_file",$mapping_file) if (defined $mapping_file && $mapping_file ne "");
+    }
+    if ($plugout_name =~ m/^.*METSPlugout$/) {
+    push @$plugout,("-xslt_mets",$xslt_mets)       if (defined $xslt_mets && $xslt_mets ne "");
+    push @$plugout,("-xslt_txt",$xslt_txt)         if (defined $xslt_txt && $xslt_txt ne "");
+    }
+
+    if ($plugout_name eq "FedoraMETSPlugout") {
+    push @$plugout,("-fedora_namespace",$fedora_namespace) if (defined $fedora_namespace && $fedora_namespace ne "");
+    }
+
+
     my $processor = &plugout::load_plugout($plugout);                        
     $processor->setoutputdir ($archivedir);
@@ -443 +508 @@
     &plugin::file_block_read($pluginfo, $importdir, "", $block_hash, $metadata, $gli);
 
-
     if ($incremental || $incremental_mode eq "onlyadd") {
 
         prime_doc_oid_count($archivedir);
-
 
         # Can now work out which files were new, already existed, and have
@@ -532 +595 @@
     &plugin::remove_some($pluginfo, $collectcfg->{'infodbtype'}, $archivedir, \@full_deleted_files);
     mark_docs_for_deletion($archive_info,{},
-                      \@full_deleted_files,
-                      $archivedir, $verbosity, "delete");
+                   \@full_deleted_files,
+                   $archivedir, $verbosity, "delete");
 
 
@@ -574 +637 @@
     }
 
+    if ($saveas eq "FedoraMETS") {
+    # create collection "doc obj" for Fedora that contains
+    # collection-level metadata
+    
+    my $doc_obj = new doc($config_filename,"nonindexed_doc","none");
+    $doc_obj->set_OID("collection");
+    
+    my $col_name = undef;
+    my $col_meta = $collectcfg->{'collectionmeta'};
+    
+    if (defined $col_meta) {        
+        store_collectionmeta($col_meta,"collectionname",$doc_obj); # in GS3 this is a collection's name
+        store_collectionmeta($col_meta,"collectionextra",$doc_obj); # in GS3 this is a collection's description     
+    }
+    $processor->process($doc_obj);
+    }
+
     &plugin::end($pluginfo, $processor);
 
@@ -586 +666 @@
     $processor->close_group_output() if $processor->is_group();
 
-# The following 'if' statement is in the export.pl version of the script,
-# The reason for the 'if' statement is now given in export.pl
-# Unclear at this point if the same should be done here
-##    if (($saveas =~ m/^.*METS$/) || ($saveas eq "MARC")) {
-    # Not all export types need this (e.g. DSpace)
-
-    # should we still do this in debug mode??
-
     # for backwards compatability with archvies.inf file
     if ($arcinfo_doc_filename =~ m/(contents)|(\.inf)$/) {
@@ -602 +674 @@
     }
 
-
-##    }
-
     return $pluginfo;
 }
@@ -612 +681 @@
 {
     my $self = shift @_;
-    my ($inexport_mode,$pluginfo) = @_;
+    my ($pluginfo) = @_;
+
+    my $inexport_mode = $self->{'mode'};
 
     my $statsfile = $self->{'statsfile'};
@@ -634 +705 @@
     &gsprintf($out, "\n");
     &gsprintf($out, "*********************************************\n");
-    &gsprintf($out, "{import.complete}\n");
+    &gsprintf($out, "{$inexport_mode.complete}\n");
     &gsprintf($out, "*********************************************\n");