Changeset 2810

Timestamp:

2001-10-31T19:37:58+13:00 (22 years ago)

Author:

sjboddie

Message:

Created GAPlug (and XMLPlug base class) to replace the old GMLPlug.
Greenstone archives will now be stored as proper XML documents (with .xml
file extension) instead of the old .gml files.

Location:

trunk/gsdl/perllib

Files:

• doc.pm (modified) (1 diff)
• docsave.pm (modified) (10 diffs)
• plugins/GAPlug.pm (added)
• plugins/XMLPlug.pm (added)

trunk/gsdl/perllib/doc.pm

@@ -155 +155 @@
 }
 
-
-sub buffer_section_gml {
-    my $self = shift (@_);
-    my ($section, $suppress_subject_info) = @_;
-
-    $suppress_subject_info = 0 unless defined $suppress_subject_info;
-    my ($all_text,$data, $subsection);
-    
+sub buffer_section_xml {
+    my $self = shift (@_);
+    my ($section) = @_;
+
     my $section_ptr = $self->_lookup_section ($section);
-    my ($section_num) = $section =~ /(\d+)$/;
-    
     return "" unless defined $section_ptr;
 
-    # output the section header (including the section number
-    # and metadata)
-
-    $all_text = "<gsdlsection";
-    $all_text .= " gsdlnum=\"$section_num\"" if defined $section_num;
-    foreach $data (@{$section_ptr->{'metadata'}}) {
-    $all_text .= " $data->[0]=\"" . &_escape_text($data->[1]) . "\"" 
-        unless $suppress_subject_info && $data->[0] eq "Subject";
-    }
-    $all_text .= ">";
+    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";
 
     # output the text
+    $all_text .= "  <Content>\n";
     $all_text .= &_escape_text($section_ptr->{'text'});
-
+    $all_text .= "  </Content>\n";
+    
     # output all the subsections
-    foreach $subsection (@{$section_ptr->{'subsection_order'}}) {
-    $all_text .= $self->buffer_section_gml("$section.$subsection", 
-                           $suppress_subject_info);
-    }
-    
-    # output the closing tag
-    $all_text .=  "</gsdlsection>\n";
+    foreach my $subsection (@{$section_ptr->{'subsection_order'}}) {
+    $all_text .= $self->buffer_section_xml("$section.$subsection");
+    }
+    
+    $all_text .=  "</Section>\n";
 
     return $all_text;
 }
 
-
-sub buffer_section_xml {
-    my $self = shift (@_);
-    my ($section, $dtd_metadata, $suppress_subject_info) = @_;
-
-    $suppress_subject_info = 0 unless defined $suppress_subject_info;
-    my ($all_text, $data, $subsection);
-    
-    my $section_ptr = $self->_lookup_section ($section);
-    my ($section_num) = $section =~ /(\d+)$/;
-    
-    return "" unless defined $section_ptr;
-
-    # output the section header (including the section number
-    # and metadata)
-
-    $all_text .= "<gsdlsection";
-    $all_text .= " gsdlnum=\"$section_num\"" if defined $section_num;
-    $all_text .= ">\n";
-    
-    $all_text .= "  <metadata>\n";
-    
-    # output metadata
-    foreach $data (@{$section_ptr->{'metadata'}}) {
-    my $tag_name = $data->[0];
-
-    # a tagname beginning with '/' (like </srclink>) will cause problems
-    # so we'll escape any leading '/'
-    $tag_name =~ s/^\//&\#47;/; 
-
-    my $tag_value = &_escape_text($data->[1]);
-
-    unless ($suppress_subject_info && $tag_name eq "Subject")
-    {
-        if (defined $dtd_metadata)
-        {
-        $dtd_metadata->{$tag_name}++;
-        }
-
-        $all_text .= "    <$tag_name>$tag_value</$tag_name>\n";
-    }
-    }
-
-    $all_text .= "  </metadata>\n";
-
-    # output the text
-    $all_text .= &_escape_text($section_ptr->{'text'});
-
-    # output all the subsections
-    foreach $subsection (@{$section_ptr->{'subsection_order'}}) {
-    $all_text .= $self->buffer_section_xml("$section.$subsection", 
-                           $dtd_metadata,
-                           $suppress_subject_info);
-    }
-    
-    # output the closing tag
-    $all_text .=  "</gsdlsection>\n";
-
-    return $all_text;
-}
-
 sub output_section {
     my $self = shift (@_);
-    my ($handle, $section, $colname, $dtd_metadata, 
-    $suppress_subject_info) = @_;
-
-    my $all_text = $self->buffer_section_xml($section, $dtd_metadata, 
-                         $suppress_subject_info);
-
-    # xml header    
-    if (defined $collection)
-    {
-    my $xml_head
-        = '<? xml version="1.0" standalone="no" encoding="UTF-8" ?>';
-    $xml_head .= "\n<!DOCTYPE gsdl:$colname SYSTEM \"$colname.dtd\">\n";
-    $all_text = $xml_head.$all_text;
-    }
-
-    print $handle $all_text;
+    my ($handle, $section) = @_;
+
+    print $handle $self->buffer_section_xml($section);
 }
 

trunk/gsdl/perllib/docsave.pm

@@ -25 +25 @@
 
 # This document processor saves a document in the 
-# archives directory of a collection
+# archives directory of a collection (as xml)
 
 
@@ -53 +53 @@
     $self->{'groupsize'} = $groupsize;
     $self->{'gs_count'} = 0;
-
-    # keep an associate array of all metavalues used by collection to
-    # help generate the XML DTD 
-    $self->{'dtd_metadata'} = {}; 
 
     $self->{'outhandle'} = STDERR;
@@ -93 +89 @@
     
     } else {
-    # groupsize is 1 (i.e. one document per GML file) so sortmeta 
+    # groupsize is 1 (i.e. one document per XML file) so sortmeta 
     # may be used
     
@@ -107 +103 @@
     
     my $doc_file 
-        = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.gml");
-    my $short_doc_file = &util::filename_cat ($doc_dir, "doc.gml");
+        = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.xml");
+    my $short_doc_file = &util::filename_cat ($doc_dir, "doc.xml");
     
     if (!open (OUTDOC, ">$doc_file")) {
@@ -116 +112 @@
 
     # save this document
-    $doc_obj->output_section('docsave::OUTDOC', 
-                 $doc_obj->get_top_section(),
-                 $self->{'collection'},
-                 $self->{'dtd_metadata'},0);
-
+    $self->output_xml_header('docsave::OUTDOC');
+    $doc_obj->output_section('docsave::OUTDOC', $doc_obj->get_top_section());
+    $self->output_xml_footer('docsave::OUTDOC');
     close OUTDOC;
 
@@ -172 +166 @@
         # only if opening new file
         my $doc_file 
-        = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.gml");
-        my $short_doc_file = &util::filename_cat ($doc_dir, "doc.gml");
+        = &util::filename_cat ($self->{'archive_dir'}, $doc_dir, "doc.xml");
+        my $short_doc_file = &util::filename_cat ($doc_dir, "doc.xml");
         
         if ($gs_count>0)
@@ -187 +181 @@
         $self->{'gs_short_filename'} = $short_doc_file;
         $self->{'gs_OID'} = $OID;
+
+        $self->output_xml_header('docsave::OUTDOC');
     }
     }
 
     # save this document
-    $doc_obj->output_section('docsave::OUTDOC', 
-                 $doc_obj->get_top_section(),
-                 $self->{'collection'},
-                 $self->{'dtd_metadata'},0);
+    $doc_obj->output_section('docsave::OUTDOC', $doc_obj->get_top_section());
 
     $self->{'gs_count'}++;
@@ -210 +203 @@
     # same one.
     $doc_dir = $doc_info->[0];
-    $doc_dir =~ s/\/?doc\.gml(\.gz)?$//;
+    $doc_dir =~ s/\/?doc\.xml(\.gz)?$//;
     } else {
     # have to get a new document directory
@@ -264 +257 @@
 {
     my ($self) = @_;
-    
+
+    $self->output_xml_footer('docsave::OUTDOC');    
     close OUTDOC;
 
@@ -288 +282 @@
 }
 
+sub output_xml_header {
+    my $self = shift (@_);
+    my ($handle) = @_;
+
+    print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
+    print $handle '<!DOCTYPE GreenstoneArchive SYSTEM ' .
+    '"http://greenstone.org/dtd/GreenstoneArchive/1.0/GreenstoneArchive.dtd">' . "\n";
+    print $handle '<GreenstoneArchive>' . "\n";
+}
+
+sub output_xml_footer {
+    my $self = shift (@_);
+    my ($handle) = @_;
+
+    print $handle '</GreenstoneArchive>' . "\n";
+}
+
 1;