Changeset 32512 for main/trunk/greenstone2/perllib/docprint.pm

Timestamp:

2018-10-11T19:31:44+13:00 (4 years ago)

Author:

ak19

Message:

Refactoring some existing perl code as per Dr Bainbridge's ideas on this, so that MySQLPlugout, which inherits from GreenstoneXMLPlugout, can be structured better for reading.

File:

• main/trunk/greenstone2/perllib/docprint.pm (modified) (3 diffs)

main/trunk/greenstone2/perllib/docprint.pm

@@ -30 +30 @@
 package docprint;
 
+use constant OUTPUT_META_ONLY => 1;
+use constant OUTPUT_TEXT_ONLY => 2;
+use constant OUTPUT_ALL => 3;
+
 use strict;
 
 sub get_section_xml {
+    return &get_section_xml_from_root(@_);
+}
+
+sub get_section_xml_from_root {
+    my ($doc_obj, $options) = @_;
+    return &recursive_get_section_xml($doc_obj,$doc_obj->get_top_section(), $options);
+}
+
+sub recursive_get_section_xml {    
+    my ($doc_obj, $section, $options) = @_;
     
-    my ($doc_obj, $section) = @_;
-
+    # 'output' can be OUTPUT_ALL|OUTPUT_META_ONLY|OUTPUT_TEXT_ONLY
+    # If not provided, it defaults to OUTPUT_ALL.
+    # If OUTPUT_ALL, the metadata and full text both go into doc.xml
+    # If OUTPUT_META_ONLY, the metadata goes into doc.xml and full text goes elsewhere (mysql db).
+    # If OUTPUT_TEXT_ONLY, the full text goes into doc.xml and metadata goes elsewhere (mysql db).
+    # In the last two cases, an XML comment is left behind to indicate that the "missing" doc
+    # information is stored elsewhere.
+    if(!defined $options) { 
+    $options = {'output' => OUTPUT_ALL };
+    }
+    
     my $section_ptr = $doc_obj->_lookup_section ($section);
     return "" unless defined $section_ptr;
@@ -41 +64 @@
     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] . '">' . $escaped_value . "</Metadata>\n";
+
+    if($options->{'output'} == OUTPUT_ALL || $options->{'output'} == OUTPUT_META_ONLY) {
+    # 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";
+    }
+    } else {
+    $all_text .= "<!-- metadata is stored elsewhere (MySQL database) -->\n";
     }
 
@@ -52 +79 @@
     # output the text
     $all_text .= "  <Content>";
-    $all_text .= &escape_text($section_ptr->{'text'});
+    if($options->{'output'} == OUTPUT_ALL || $options->{'output'} == OUTPUT_TEXT_ONLY) {
+    $all_text .= &escape_text($section_ptr->{'text'});
+    } else {
+    $all_text .= "<!-- full text is stored elsewhere (MySQL database) -->\n";
+    }
     $all_text .= "</Content>\n";
     
     # output all the subsections
     foreach my $subsection (@{$section_ptr->{'subsection_order'}}) {
-    $all_text .= &get_section_xml($doc_obj, "$section.$subsection");
+    $all_text .= &recursive_get_section_xml($doc_obj, "$section.$subsection");
     }