Changeset 25961

Timestamp:

2012-07-17T15:22:12+12:00 (11 years ago)

Author:

kjdon

Message:

more cunning document types. gs3 has a new one, pagedhierarchy - for documents with internal structure and sequences of pages. new documenttype option, auto, will select the most appropriate doc type for each document. paged/hierarchy for gs2, paged/pagedhierarchy for gs3. documenttype option is displayed differently for gs2/3 modes.

File:

• main/trunk/greenstone2/perllib/plugins/PagedImagePlugin.pm (modified) (9 diffs)

main/trunk/greenstone2/perllib/plugins/PagedImagePlugin.pm

@@ -147 +147 @@
 }
 
-my $type_list =
-    [ { 'name' => "paged",
-        'desc' => "{PagedImagePlugin.documenttype.paged}" },
+my $gs2_type_list =
+    [ { 'name' => "auto",
+    'desc' => "{PagedImagePlugin.documenttype.auto2}" },
+      { 'name' => "paged",
+        'desc' => "{PagedImagePlugin.documenttype.paged2}" },
       { 'name' => "hierarchy",
-        'desc' => "{PagedImagePlugin.documenttype.hierarchy}" } ];
+        'desc' => "{PagedImagePlugin.documenttype.hierarchy}" }
+    ];
+
+my $gs3_type_list =     
+    [ { 'name' => "auto",
+    'desc' => "{PagedImagePlugin.documenttype.auto3}" },
+      { 'name' => "paged",
+        'desc' => "{PagedImagePlugin.documenttype.paged3}" },
+      { 'name' => "hierarchy",
+        'desc' => "{PagedImagePlugin.documenttype.hierarchy}" }, 
+      { 'name' => "pagedhierarchy",
+        'desc' => "{PagedImagePlugin.documenttype.pagedhierarchy}" }
+    ];
 
 my $arguments =
@@ -167 +181 @@
     'type' => "flag",
     'reqd' => "no" },
-      { 'name' => "documenttype",
-    'desc' => "{PagedImagePlugin.documenttype}",
-    'type' => "enum",
-    'list' => $type_list,
-    'deft' => "paged",
-    'reqd' => "no" },
+#      { 'name' => "documenttype",
+#   'desc' => "{PagedImagePlugin.documenttype}",
+#   'type' => "enum",
+#   'list' => $type_list,
+#   'deft' => "auto",
+#   'reqd' => "no" },
       {'name' => "processing_tmp_files",
        'desc' => "{BasePlugin.processing_tmp_files}",
        'type' => "flag",
        'hiddengli' => "yes"}
-];
-
+    ];
+
+my $doc_type_opt = { 'name' => "documenttype",
+             'desc' => "{PagedImagePlugin.documenttype}",
+             'type' => "enum",
+             'deft' => "auto",
+             'reqd' => "no" };
 
 my $options = { 'name'     => "PagedImagePlugin",
@@ -191 +210 @@
     push(@$pluginlist, $class);
 
+    push(@{$hashArgOptLists->{"OptList"}},$options);
+   
+    my $imc_self = new ImageConverter($pluginlist, $inputargs, $hashArgOptLists);
+    
+    # we can use this plugin to check gs3 version
+    if ($imc_self->{'gs_version'} eq "3") {
+    $doc_type_opt->{'list'} = $gs3_type_list;
+    }
+    else {
+    $doc_type_opt->{'list'} = $gs2_type_list;
+    }
+    push(@$arguments,$doc_type_opt);
+    # now we add the args to the list for parsing
     push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
-    push(@{$hashArgOptLists->{"OptList"}},$options);
-    
-    my $imc_self = new ImageConverter($pluginlist, $inputargs, $hashArgOptLists);
+    
     my $rtf_self = new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists, 1);
     my $rxf_self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
@@ -212 +242 @@
     $rxf_self->{'parser'}->{'PluginObj'} = $self;
 
+    print STDERR "doc type = $self->{'documenttype'}\n";
     return bless $self, $class;
 }
@@ -449 +480 @@
     $self->{'current_section'} = $doc_obj->get_top_section();
     } elsif ($element eq "PageGroup" || $element eq "Page") {
+    if ($element eq "PageGroup") {
+        $self->{'has_internal_structure'} = 1;
+    }
     # create a new section as a child
     $self->{'current_section'} = $doc_obj->insert_section($doc_obj->get_end_child($self->{'current_section'}));
@@ -529 +563 @@
     $self->{'xml_file_dir'} = $dir;
     $self->{'num_pages'} = 0;
+    $self->{'has_internal_structure'} = 0;
 
 }
@@ -536 +571 @@
     my $doc_obj = $self->{'doc_obj'};
     
+    my $topsection = $doc_obj->get_top_section();
+
     # add numpages metadata
-    my $topsection = $doc_obj->get_top_section();
-
     $doc_obj->set_utf8_metadata_element ($topsection, 'NumPages', $self->{'num_pages'});
+
+    # set the document type
+    print STDERR "close doc, doc type = $self->{'documenttype'}\n";
+    my $final_doc_type = "";
+    if ($self->{'documenttype'} eq "auto") {
+    if ($self->{'has_internal_structure'}) {
+        if ($self->{'gs_version'} eq "3") {
+        $final_doc_type = "pagedhierarchy";
+        }
+        else {
+        $final_doc_type = "hierarchy";
+        }
+    } else {
+        $final_doc_type = "paged";
+    }
+    } else {
+    # set to what doc type option was set to
+    $final_doc_type = $self->{'documenttype'};
+    }
+    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", $final_doc_type);
+    ### capiatalisation????
+#    if ($self->{'documenttype'} eq 'paged') {
+    # set the gsdlthistype metadata to Paged - this ensures this document will
+    # be treated as a Paged doc, even if Titles are not numeric
+#   $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "Paged");
+#    } else {
+#   $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "Hierarchy");
+#    }
 
     $doc_obj->set_utf8_metadata_element($topsection,"MaxImageWidth",$self->{'MaxImageWidth'});
@@ -554 +617 @@
 
     my $topsection = $doc_obj->get_top_section();
-
-    if ($self->{'documenttype'} eq 'paged') {
-    # set the gsdlthistype metadata to Paged - this ensures this document will
-    # be treated as a Paged doc, even if Titles are not numeric
-    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "Paged");
-    } else {
-    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "Hierarchy");
-    }
 
     my $plugin_filename_encoding = $self->{'filename_encoding'};
@@ -631 +686 @@
     $self->set_initial_doc_fields($doc_obj, $filename_full_path, $processor, $metadata);
     my $topsection = $doc_obj->get_top_section();
+    # simple item files are always paged unless user specified
+    if ($self->{'documenttype'} eq "auto") {
+    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", "paged");
+    } else {
+    $doc_obj->set_utf8_metadata_element ($topsection, "gsdlthistype", $self->{'documenttype'});
+    }
     open (ITEMFILE, $filename_full_path) || die "couldn't open $filename_full_path\n";
     my $line = "";