Changeset 16102

Timestamp:

2008-06-23T10:54:24+12:00 (16 years ago)

Author:

davidb

Message:

Some minor adjustments to ingesting documents into a Fedoar repository: it now descends into all the folders generated in archives/export; and it tests for ImagePlugin (rather than the older ImagePlug) name;

Location:

gsdl/trunk

Files:

• bin/script/g2f-buildcol.pl (modified) (2 diffs)
• bin/script/g2f-import.pl (modified) (2 diffs)
• perllib/g2futil.pm (modified) (2 diffs)
• perllib/plugouts/FedoraMETSPlugout.pm (modified) (3 diffs)

gsdl/trunk/bin/script/g2f-buildcol.pl

@@ -205 +205 @@
 
     if (opendir(DIR, $export_dir)) {
-    my @hash_dirs = grep { /\.dir$/ } readdir(DIR);
     closedir DIR;
+    ## my @hash_dirs = grep { /\.dir$/ } readdir(DIR);
+    my @hash_dirs = &g2futil::get_all_hash_dirs($export_dir);
 
 
@@ -213 +214 @@
 
         my $docmets_filename 
-        = &util::filename_cat($export_dir,$hd,"docmets.xml");
+        = &util::filename_cat($hd,"docmets.xml");
 
         print STDERR "<Build>\n" if $gli;

gsdl/trunk/bin/script/g2f-import.pl

@@ -204 +204 @@
     if ( -e $export_dir ) {
         print "***\n";
-    print "* Removing existing Greenstone $gs_col objects from Fedora $pid_namespace\n";
+    print "* Updating existing Greenstone $gs_col objects from Fedora $pid_namespace\n";
         print "***\n";
 
     # readdir
     if (opendir(DIR, $export_dir)) {
-        my @hash_dirs = grep { /\.dir$/ } readdir(DIR);
+
         closedir DIR;
-
+        my @hash_dirs = &g2futil::get_all_hash_dirs($export_dir,$maxdocs);
 
         # for each hash dir, purge its respective PID
-        foreach my $hd (@hash_dirs) {
-
-        my $full_hd = &util::filename_cat($export_dir,$hd);
+        foreach my $full_hd (@hash_dirs) {
+
         my $hash_id = &g2futil::get_hash_id($full_hd);
 
@@ -226 +225 @@
 
         if ($dsinfo_status == 0) {
-            print "  $pid being removed.\n";        
+            print "  $pid being updated.\n";        
             &g2futil::run_purge($pid,$options);
         }

gsdl/trunk/perllib/g2futil.pm

@@ -204 +204 @@
 
 
+sub rec_get_all_hash_dirs
+{
+    my ($full_dir,$all_dirs) = @_;
+
+    if (opendir(DIR, $full_dir)) {
+    my @sub_dirs = grep { ($_ !~ /^\./) && (-d &util::filename_cat($full_dir,$_)) } readdir(DIR);
+    closedir DIR;
+
+    my @hash_dirs = grep { $_ =~ m/\.dir$/ } @sub_dirs;
+    my @rec_dirs = grep { $_ !~ m/\.dir$/ } @sub_dirs;
+    
+    foreach my $hd (@hash_dirs) {
+        my $full_hash_dir = &util::filename_cat($full_dir,$hd);
+        push(@$all_dirs,$full_hash_dir);
+    }
+
+    foreach my $rd (@rec_dirs) {
+        my $full_rec_dir = &util::filename_cat($full_dir,$rd);
+        rec_get_all_hash_dirs($full_rec_dir,$all_dirs);
+    }       
+    }
+}
+
+sub get_all_hash_dirs
+{
+    my ($start_dir,$maxdocs) = @_;
+    
+    my @all_dirs = ();
+    rec_get_all_hash_dirs($start_dir,\@all_dirs);
+
+    if ((defined $maxdocs) && ($maxdocs ne "")) {
+    my @maxdoc_dirs = ();
+    for (my $i=0; $i<$maxdocs; $i++) {
+        push(@maxdoc_dirs,shift(@all_dirs));
+    }
+    @all_dirs = @maxdoc_dirs;
+    }
+
+    return @all_dirs;
+}
 
 sub get_hash_id
@@ -296 +336 @@
     close(FIN); # close the file
     if($xml_contents eq $gsdlXMLcontents) { 
-        print STDERR "The old gsdl.xml file already contains the same.\n";
+        print STDERR "Fedora links to FLI import folder through gsdl.xml.\n";
         # it already contains what we want, we're done
         return "gsdl.xml";

gsdl/trunk/perllib/plugouts/FedoraMETSPlugout.pm

@@ -201 +201 @@
 
     my $section = $doc_obj->get_top_section();
-
-    my $doc_txt_file = &util::filename_cat ($working_dir,"doctoc.xml"); 
-    
-    $self->open_xslt_pipe($doc_txt_file,$self->{'xslt_txt'});
-
-    my $outhandler; 
-
-    if (defined $self->{'xslt_writer'}){
-    $outhandler = $self->{'xslt_writer'};
-    }
-    else{
-    $outhandler = $self->get_output_handler($doc_txt_file);
-    } 
-
-    print $outhandler $self->buffer_toc($doc_obj, $working_dir, $section, 0);
-
-    if (defined $self->{'xslt_writer'}){     
-    $self->close_xslt_pipe(); 
-    }
-    else{
-    close($outhandler);
-   } 
+    my $section_ptr=$doc_obj->_lookup_section($section);
+    my $num_subsections = scalar(@{$section_ptr->{'subsection_order'}});
+
+    # If num_subsections is 0, then there is no nested TOC
+
+    if ($num_subsections>0) {
+
+    my $doc_txt_file = &util::filename_cat ($working_dir,"doctoc.xml");
+    
+    $self->open_xslt_pipe($doc_txt_file,$self->{'xslt_txt'});
+    
+    my $outhandler; 
+    
+    if (defined $self->{'xslt_writer'}){
+        $outhandler = $self->{'xslt_writer'};
+    }
+    else{
+        $outhandler = $self->get_output_handler($doc_txt_file);
+    } 
+    print $outhandler $self->buffer_toc($doc_obj, $working_dir, $section, 0);
+    
+    if (defined $self->{'xslt_writer'}){     
+        $self->close_xslt_pipe(); 
+    }
+    else{
+        close($outhandler);
+    } 
+    }
 
 }
@@ -245 +251 @@
     my $plugin_type = $doc_obj->get_metadata_element($top_section,"Plugin");
 
-    if ((defined $plugin_type) && ($plugin_type eq "ImagePlug"))
+    if ((defined $plugin_type) && ($plugin_type eq "ImagePlugin"))
     {
 
@@ -284 +290 @@
 
     # Generate Filestream for Table of Contents (TOC)
-    print $handle $self->buffer_mets_fileSection_toc($doc_obj,$section,$working_dir);
+    my $section_ptr=$doc_obj->_lookup_section($section);
+    my $num_subsections = scalar(@{$section_ptr->{'subsection_order'}});
+
+    # If num_subsections is 0, then there is no nested TOC
+
+    if ($num_subsections>0) {
+    print $handle $self->buffer_mets_fileSection_toc($doc_obj,$section,$working_dir);
+    }
 
     # print out the fileSection by sections