Changeset 7902 for trunk/gsdl/perllib/docsave.pm

Timestamp:

2004-08-06T16:34:02+12:00 (20 years ago)

Author:

chi

Message:

Saving of documents (in archive format) extended to generate METS format
as alternative to GreenstoneArchive (GA) format. Controlled through
'import.pl -saveas METS ...'

File:

• trunk/gsdl/perllib/docsave.pm (modified) (6 diffs)

trunk/gsdl/perllib/docsave.pm

@@ -96 +96 @@
     my $OID = $doc_obj->get_OID();
     $OID = "NULL" unless defined $OID;
-    
+
+
     # get document's directory
     my $doc_dir = $self->get_doc_dir ($OID);
-    
+     
+    
     # copy all the associated files, add this information as metadata
     # to the document
@@ -106 +108 @@
     my $doc_file 
         = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.xml");
-    my $short_doc_file = &util::filename_cat ($doc_dir, "doc.xml");
-    
+      
+        #***define doctxt.xml file
+        my $doc_txt_file
+            = &util::filename_cat ($self->{'archive_dir'}, $doc_dir,"doctxt.xml");
+    my $working_dir
+            =&util::filename_cat ($self->{'archive_dir'}, $doc_dir);
+        
+        #***define docmets.xmlfile
+        my $doc_mets_file
+            = &util::filename_cat ($self->{'archive_dir'},$doc_dir, "docmets.xml");
+ 
+        my $short_doc_file = &util::filename_cat ($doc_dir, "doc.xml");
+    #my $short_txt_doc_file=&util::filename_cat ($doc_dir, "doctxt.xml");
+        my $short_mets_doc_file=&util::filename_cat ($doc_dir, "docmets.xml");
+    
+
     if (!open (OUTDOC, ">$doc_file")) {
-        print $outhandle "docsave::process could not write to file $doc_file\n";
+        print $outhandle "docsave::process could not write to file $doc_file\n";
         return;
     }
@@ -117 +133 @@
     $doc_obj->output_section('docsave::OUTDOC', $doc_obj->get_top_section());
     $self->output_xml_footer('docsave::OUTDOC');
-    close OUTDOC;
+
+        #***save the document without metadata:doctxt.xml: Chi-Yu Huang
+
+    if (!open(OUTDOC_TXT, ">$doc_txt_file")){
+            print $outhandle "docsave::process could not write to file $doc_mets_file\n";
+            return;
+        }
+
+        $self->output_txt_xml_header('docsave::OUTDOC_TXT');
+        $doc_obj->output_txt_section('docsave::OUTDOC_TXT', $doc_obj->get_top_section());
+    #$self->output_txt_xml_footer('docsave::OUTDOC_TXT');
+                
+        #***Convert doctxt.xml file to docmets.xml: Chi-Yu Huang
+        if (!open(OUTDOC_METS,">$doc_mets_file")){
+        print $outhandle "docsave::process could not write to file $doc_mets_file\n";
+            return;
+    }
+         
+        $self->output_mets_xml_header('docsave::OUTDOC_METS', $OID);
+        $doc_obj->output_mets_section('docsave::OUTDOC_METS',$doc_obj->get_top_section(), $working_dir);
+        $self->output_mets_xml_footer('docsave::OUTDOC_METS');
+
+        close OUTDOC;
+        close OUTDOC_TXT;
+    close OUTDOC_METS;
 
     if ($self->{'gzip'}) {
@@ -131 +171 @@
 
     # do the sortmeta thing
-    my ($metadata);
-    if (defined ($self->{'sortmeta'})) {
+    my ($metadata); if (defined ($self->{'sortmeta'})) {
         $metadata = $doc_obj->get_metadata_element ($doc_obj->get_top_section(), $self->{'sortmeta'});
     }
 
     # store reference in the archive_info
-    $self->{'archive_info'}->add_info($OID, $short_doc_file, $metadata);
-    }
-}
-
+    if ($self->{'saveas'} eq "METS"){
+        $self->{'archive_info'}->add_info($OID, $short_mets_doc_file, $metadata);
+    } else {
+        $self->{'archive_info'}->add_info($OID, $short_doc_file, $metadata);
+    }
+    }
+}
+
+      
 sub group_process {
     my $self = shift (@_);
@@ -283 +327 @@
 
     # store reference in the archive_info
-    $self->{'archive_info'}->add_info($OID, $short_doc_file);
-
+    if ($self->{'saveas'} eq "METS"){
+    $self->{'archive_info'}->add_info($OID, $short_mets_doc_file);
+    } else {
+    $self->{'archive_info'}->add_info($OID, $short_doc_file);
+    }
+   
     return 1;
 }
@@ -292 +340 @@
     my ($handle) = @_;
 
+    print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
+   
+    print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
+    print $handle "<Archive>\n";
+}
+
+sub output_xml_footer {
+    my $self = shift (@_);
+    my ($handle) = @_;
+
+    print $handle "</Archive>\n";
+}
+
+sub output_txt_xml_header{
+    my $self = shift (@_);
+    my ($handle) = @_;
     print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
     print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
-    print $handle "<Archive>\n";
-}
-
-sub output_xml_footer {
-    my $self = shift (@_);
-    my ($handle) = @_;
-
-    print $handle "</Archive>\n";
+}
+
+sub output_txt_xml_footer{
+    my $self = shift(@_);
+    my ($handle) = @_;
+    print $handle "<the end of the file>\n";
+}
+
+sub output_mets_xml_header(){
+    my $self = shift(@_);
+    my ($handle, $OID) = @_;
+    print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
+    print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
+    print $handle '<mets:mets OBJID="'. $OID. ':2">' . "\n";
+}
+
+sub output_mets_xml_footer() {
+    my $self = shift(@_);
+    my ($handle) = @_;
+    print $handle '</mets:mets>' . "\n";
 }
 
 1;
+
+
+