Changeset 28642 for main/trunk/greenstone2/perllib/plugouts/GreenstoneXMLPlugout.pm

Timestamp:

2013-11-19T12:06:05+13:00 (10 years ago)

Author:

kjdon

Message:

group processing code was GreenstoneXML format so moved it into GreenstoneXMLPlugout. tidying up the code. reordered options. group processing now writes out the correct archivesinf databases.

File:

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

main/trunk/greenstone2/perllib/plugouts/GreenstoneXMLPlugout.pm

@@ -40 +40 @@
 }
 
-my $arguments = [];
-
+my $arguments = [
+       { 'name' => "group_size", 
+    'desc' => "{BasePlugout.group_size}",
+    'type' => "int",
+        'deft' =>  "1",
+    'reqd' => "no",
+    'hiddengli' => "no"}
+    ];
 my $options = { 'name'     => "GreenstoneXMLPlugout",
         'desc'     => "{GreenstoneXMLPlugout.desc}",
         'abstract' => "no",
-        'inherits' => "yes" };
+        'inherits' => "yes",
+        'args'     => $arguments };
 
 sub new {
@@ -60 +67 @@
 }
 
+sub is_group {
+    my $self = shift (@_);
+    return ($self->{'group_size'} > 1); 
+}
+
 sub saveas {
     my $self = shift (@_);
-    my ($doc_obj,$doc_dir) = @_;
-
+    my ($doc_obj, $doc_dir) = @_;
     my $outhandler;
     my $output_file;
     if ($self->{'debug'}) {
     $outhandler = STDOUT;
-    # can we do the xslt and still do debug mode?
     }
     else {
-    my $output_dir = $self->get_output_dir();
-        if (!&FileUtils::directoryExists($output_dir))
-        {
-          &FileUtils::makeAllDirectories($output_dir);
-        }
+       
+    $self->process_assoc_files($doc_obj, $doc_dir, '');
+    $self->process_metafiles_metadata ($doc_obj);
+    
+    # open up the outhandler    
+    if ($self->is_group() && !$self->{'new_doc_dir'}) { 
+        # we already have a handle open ??
+        $outhandler = $self->{'group_outhandler'};
+    } else {
+        $output_file = &FileUtils::filenameConcatenate($self->{'output_dir'}, $doc_dir, "doc.xml");
+        # open the new handle
+        $self->open_xslt_pipe($output_file, $self->{'xslt_file'});
 
-    my $working_dir = &FileUtils::filenameConcatenate($output_dir, $doc_dir);
-        if (!&FileUtils::directoryExists($working_dir))
-        {
-          &FileUtils::makeAllDirectories($working_dir);
-        }
+        if (defined $self->{'xslt_writer'}){
+        $outhandler = $self->{'xslt_writer'};
+        }
+        else{
+        $outhandler = $self->get_output_handler($output_file);
+        }
+        
+        if ($self->is_group()) {
+        $self->{'group_outhandler'} = $outhandler;
+        }
+    }
+    } # else not debug
+    binmode($outhandler,":utf8");
 
-    $self->process_assoc_files ($doc_obj, $doc_dir, '');
-
-    $self->process_metafiles_metadata ($doc_obj);
-
-    $output_file = &FileUtils::filenameConcatenate($working_dir, "doc.xml");
-
-    $self->open_xslt_pipe($output_file, $self->{'xslt_file'});
-
-    if (defined $self->{'xslt_writer'}){
-        $outhandler = $self->{'xslt_writer'};
-    }
-    else{
-        $outhandler = $self->get_output_handler($output_file);
-    }
+    # only output the header if we have started a new doc
+    if (!$self->is_group() || $self->{'new_doc_dir'}) {
+    $self->output_xml_header($outhandler);
     }
 
-    binmode($outhandler,":utf8");
+    my $section_text = &docprint::get_section_xml($doc_obj,$doc_obj->get_top_section());
+    print $outhandler $section_text;
+ 
+    # only output the footer if we are not doing group stuff. The group file will be finished in close_group_output
+    if (!$self->is_group()) {
+    $self->output_xml_footer($outhandler);
+    }
 
-    $self->output_xml_header($outhandler,"Archive");
-    my $section_output = &docprint::get_section_xml($doc_obj, $doc_obj->get_top_section());
-    print $outhandler $section_output;
-    $self->output_xml_footer($outhandler,"Archive");
-
-    if (!$self->{'debug'}) {
+    # close off the output - in a group process situation, this will be done by close_group_output
+    if (!$self->is_group() && !$self->{'debug'}) {
     if (defined $self->{'xslt_writer'}){     
         $self->close_xslt_pipe(); 
@@ -113 +129 @@
         &FileUtils::closeFileHandle($output_file, \$outhandler) if defined $output_file;
     }
-    
-    $self->{'short_doc_file'} = &FileUtils::filenameConcatenate($doc_dir, "doc.xml");  
-    
-    $self->store_output_info_reference($doc_obj);
     }
+    $self->{'short_doc_file'} = &FileUtils::filenameConcatenate($doc_dir, "doc.xml");  
+    
+    $self->store_output_info_reference($doc_obj);
+    
 }
 
+sub output_xml_header {
+    my $self = shift (@_);
+    my ($outhandle) = @_;
+
+    print $outhandle '<?xml version="1.0" encoding="utf-8" standalone="no"?>' . "\n";
+    print $outhandle "<!DOCTYPE Archive SYSTEM \"http://greenstone.org/dtd/Archive/1.0/Archive.dtd\">\n";
+    print $outhandle "<Archive>\n";
+}
+
+sub output_xml_footer {
+    my $self = shift (@_);
+    my ($outhandle) = @_;
+
+    print $outhandle "</Archive>\n";
+}
+
+sub close_group_output
+{
+    my $self = shift(@_);
+ 
+    # make sure that the handle has been opened - it won't be if we failed
+    # to import any documents...
+    my $outhandle = $self->{'group_outhandler'};
+    if (defined(fileno($outhandle))) {
+    $self->output_xml_footer($outhandle);    
+    &FileUtils::closeFileHandle("", \$outhandle);
+    undef $self->{'group_outhandler'}
+    }
+
+    my $OID = $self->{'gs_OID'};
+    my $short_doc_file = $self->{'short_doc_file'};
+   
+    ### TODO - from here is old code. check that it is still valid.
+    if ($self->{'gzip'}) {
+    my $doc_file = $self->{'gs_filename'};
+    `gzip $doc_file`;
+    $doc_file .= ".gz";
+    $short_doc_file .= ".gz";
+    if (!&FileUtils::fileExists($doc_file)) {
+         my $outhandle = $self->{'output_handle'};
+        print $outhandle "error while gzipping: $doc_file doesn't exist\n";
+        return 0;
+    }
+    }
+
+    return 1;
+}