Changeset 8504

Timestamp:

2004-11-10T11:14:18+13:00 (19 years ago)

Author:

chi

Message:

Modification of METS format in order to be compatible with GS3. Also, add some new methods for the use of DSpace.

File:

• trunk/gsdl/perllib/doc.pm (modified) (14 diffs)

trunk/gsdl/perllib/doc.pm

@@ -286 +286 @@
   
     
-    my $filePath = $doc_Dir . '/doctxt.xml';
-    
+    #my $filePath = $doc_Dir . '/doctxt.xml';
+
+    my $filePath = 'doctxt.xml';
+
     #**output the fileSection details
     my $all_text = '  <mets:fileGrp ID="FILEGROUP_PRELUDE' . $section_num . '">'. "\n";
@@ -331 +333 @@
        my $escaped_value = &_escape_text($data->[1]);
        if ($data->[0] eq "gsdlsourcefilename") { 
-          ($dirPath) = $escaped_value =~ m/^(.*)[\/\\][^\/\\]*$/; 
-          
+          ($dirPath) = $escaped_value =~ m/^(.*)[\/\\][^\/\\]*$/;
+      
       $all_text .= '  <mets:fileGrp ID="default">'."\n";
           ++$fileID;
           $all_text .= '    <mets:file MIMETYPE="text/xml" ID="default.'.$fileID.'">'. "\n";
-          $all_text .= '      <mets:FLocat LOCTYPE="URL" xlink:href="file:'.$data->[1].'" ID="default.'.$fileID.'" />'."\n";
-
+        
+      $all_text .= '      <mets:FLocat LOCTYPE="URL" xlink:href="file:'.$data->[1].'" ID="default.'.$fileID.'" />'."\n";
+      
           $all_text .= "    </mets:file>\n";
        }
@@ -346 +349 @@
       my $assfilePath = $dirPath . '/'. $1;
           ++$fileID;
+
       $all_text .= '    <mets:file MIMETYPE="'.$2.'" ID="default.'.$fileID. '">'. "\n";
       $all_text .= '      <mets:FLocat LOCTYPE="URL" xlink:href="file:'.$assfilePath.'" ID="default.'. $fileID.'" />'."\n";
@@ -366 +370 @@
     $section="" unless defined $section;
     
-
+    
     my $section_ptr=$self->_lookup_section($section);
     return "" unless defined $section_ptr;
 
-    #**output fileSection by sections
-    my $section_num ="1". $section; 
-    
+
+    #***output fileSection by Sections
+    my $section_num ="1". $section;
+    my $dmd_num = $section_num;
+
+    #**output the dmdSection details  
+    if ($section_num eq "1") {
+    $dmd_num = "0";
+    }
+
     #**output the StruMap details
  
     my $all_text = '  <mets:div ID="DS'. $section_num .'" TYPE="Section" ORDER="'.$$order_numref++.'" ORDERLABEL="'. $section_num .'" LABEL="';
-    $all_text .= $section_num . '" DMDID="DM'.$section_num.'">'. "\n";
+
+    $all_text .= $section_num . '" DMDID="DM'.$dmd_num.'">'. "\n";
    
     $all_text .= '    <mets:fptr FILEID="FILEGROUP_PRELUDE'.$section_num.'" />'. "\n";
@@ -442 +454 @@
     #***convert section number
     my $section_num ="1". $section;
-      
+    my $dmd_num = $section_num;
+
     #**output the dmdSection details  
-    my $all_text = '<mets:dmdSec ID="DM'.$section_num.'" GROUPID="'.$section_num.'">'. "\n";
-    $all_text .= '  <mets:mdWrap MDType="gsdl" ID="'.$section_num.'">'."\n";
+    if ($section_num eq "1") {
+    $dmd_num = "0";
+    }
+    my $all_text = '<mets:dmdSec ID="DM'.$dmd_num.'" GROUPID="'.$section_num.'">'. "\n";
+    $all_text .= '  <mets:mdWrap MDType="gsdl3" ID="'.$section_num.'">'."\n";
     $all_text .= "    <mets:xmlData>\n";
     foreach my $data (@{$section_ptr->{'metadata'}}){
        my $escaped_value = &_escape_text($data->[1]);
-       $all_text .= '      <gsdl:Metadata name="'. $data->[0].'">'. $escaped_value. "</gsdl:Metadata>\n";
-    }
+       $all_text .= '      <gsdl3:Metadata name="'. $data->[0].'">'. $escaped_value. "</gsdl3:Metadata>\n";
+       if ($data->[0] eq "dc.Title") {
+       $all_text .= '      <gsdl3:Metadata name="Title">'. $escaped_value."</gsdl3:Metadata>\n";
+       }
+    }
+   
     $all_text .= "    </mets:xmlData>\n";
     $all_text .= "  </mets:mdWrap>\n";
@@ -471 +491 @@
 }
 
+#*** print out DSpace dublin_core metadata section
+sub output_dspace_section {
+    my $self = shift (@_);
+    my ($handle, $section) = @_;
+
+    my $section_ptr = $self->_lookup_section ($section);
+    return "" unless defined $section_ptr;
+
+    my $all_text = "<Section>\n";
+    $all_text .= "  <Description>\n";
+    
+    # output metadata
+    foreach my $data (@{$section_ptr->{'metadata'}}) {
+    my $escaped_value = &_escape_text($data->[1]);
+    $all_text .= '    <Metadata name="' . $data->[0] . '">' . $escaped_value . "</Metadata>\n";
+    }
+
+    $all_text .= "  </Description>\n";
+    $all_text .=  "</Section>\n";
+
+    # make sure no nasty control characters have snuck through
+    # (XML::Parser will barf on anything it doesn't consider to be
+    # valid UTF-8 text, including things like \c@, \cC etc.)
+    $all_text =~ s/[\x00-\x09\x0B\x0C\x0E-\x1F]//g;
+
+    return $all_text;
+}
 
 #*** print out doctxt.xml file
@@ -491 +538 @@
     print $handle "<mets:fileSec>\n";
     print $handle $self->buffer_mets_fileSection_section_xml($section,$doc_Dir);
+
     #***print out the whole fileSection
     print $handle $self->buffer_mets_fileWhole_section_xml($section); 
@@ -504 +552 @@
 }
 
+#*** print out dublin_core.xml file
+sub output_dc_section {
+    my $self = shift(@_);
+    my ($handle, $section, $doc_Dir) = @_;
+    
+    #***print out the dublin_core 
+    $section="" unless defined $section;
+    
+    my $section_ptr=$self->_lookup_section($section);
+    return "" unless defined $section_ptr;
+    my $all_text="";
+    foreach my $data (@{$section_ptr->{'metadata'}}){
+    my $escaped_value = &_escape_text($data->[1]);
+    if ($data->[0]=~ /^dc/) {
+        #$all_text .= '   <dcvalue element="'. $data->[0].'" qualifier="#####">'. $escaped_value. "</dcvalue>\n";
+        $all_text .= '   <dcvalue element="'. $data->[0].'">'. $escaped_value. "</dcvalue>\n";
+    } 
+    }
+    if ($all_text eq "") {
+    $all_text .= "   There is no Dublin Core metatdata in this document\n";
+    }   
+    $all_text =~ s/[\x00-\x09\x0B\x0C\x0E-\x1F]//g;
+
+    print $handle $all_text;
+}
 
 # look up the reference to the a particular section
@@ -537 +610 @@
     my $hashfile_exe = &util::filename_cat($ENV{'GSDLHOME'},"bin",
                        $ENV{'GSDLOS'},"hashfile$osexe");
+
     my $result = "NULL";
     
@@ -547 +621 @@
     print STDERR "doc::_calc_OID $hashfile_exe could not be found\n";
     }
-
     return "HASH$result";
 }
@@ -557 +630 @@
     my $self = shift (@_);
     my ($OID) = @_;
-
+    
     # if an OID wasn't provided claculate one
     if (!defined $OID) {
@@ -572 +645 @@
         # cos its quicker
         my $filename = $self->get_filename_for_hashing();
+
         if (defined($filename) && -e $filename) {
         $OID = $self->_calc_OID ($filename);
@@ -583 +657 @@
             close (OUTFILE);
         }
-        
         $OID = $self->_calc_OID ($filename);
         &util::rm ($filename);
@@ -1188 +1261 @@
     # remove all associated files with the same name
     $self->delete_assoc_file ($assoc_filename);
-
+    
     push (@{$self->{'associated_files'}}, 
       [$real_filename, $assoc_filename, $mime_type, $section]);