Changeset 19494

Timestamp:

2009-05-18T10:58:18+12:00 (15 years ago)

Author:

davidb

Message:

Supporting routines that exploit the new 'metafiles' structures, introduction to track which metadata.xml file a piece of metadata came from

Location:

gsdl/trunk/perllib

Files:

• doc.pm (modified) (1 diff)
• plugouts/BasePlugout.pm (modified) (6 diffs)
• plugouts/GreenstoneXMLPlugout.pm (modified) (1 diff)

gsdl/trunk/perllib/doc.pm

@@ -999 +999 @@
 }
 
+
+sub metadata_file {
+    my $self = shift (@_);
+    my ($real_filename, $filename) = @_;
+    
+    push (@{$self->{'metadata_files'}}, 
+      [$real_filename, $filename]);
+}
+
+sub get_meta_files {
+    my $self = shift (@_);
+
+    return $self->{'metadata_files'};
+}
+
 sub delete_section_assoc_files {
     my $self = shift (@_);

gsdl/trunk/perllib/plugouts/BasePlugout.pm

@@ -493 +493 @@
         print $outhandle "Writing associated files to $doc_dir\n";
     $self->process_assoc_files ($doc_obj, $doc_dir);
+
+    # look up 'gsdlmetafile' metadata and store that information
+    # explicitly in $doc_obj
+    $self->process_metafiles_metadata ($doc_obj);
     }
 
@@ -643 +647 @@
 
 
-sub archiveinf_gdbm
+sub process_metafiles_metadata 
 {
     my $self = shift (@_);
     my ($doc_obj) = @_;
 
-    my $verbosity = $self->{'verbosity'};
-
-    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
-    if (defined $collect_dir) {
-    my $dirsep_regexp = &util::get_os_dirsep();
-
-    if ($collect_dir !~ /$dirsep_regexp$/) {
-        # ensure there is a slash at the end
-        $collect_dir .= &util::get_dirsep(); 
-    }
-    }
-
-    my $oid = $doc_obj->get_OID();
-    my $source_filename = $doc_obj->get_source_filename();
-
-    my $working_info = $self->{'output_info'}; 
-    my $doc_info = $working_info->get_info($oid);
-    my ($doc_file,$index_status) = @$doc_info;
-
-    my $oid_files = { 'doc-file' => $doc_file,
-              'index-status' => $index_status,
-              'src-file' => $source_filename,
-              'assoc-files' => [] };
-    
-    my %reverse_lookups = ( $source_filename => "1" );
-
-    foreach my $assoc_file_rec (@{$doc_obj->get_assoc_files()}) {
-    my $real_filename = $assoc_file_rec->[0];
-    my $full_afile = $assoc_file_rec->[1];
+    my $top_section = $doc_obj->get_top_section();
+    my $metafiles = $doc_obj->get_metadata($top_section,"gsdlmetafile");
+
+    foreach my $metafile_pair (@$metafiles) {
+    my ($full_metafile,$metafile) = split(/:/,$metafile_pair);
+
+    $doc_obj->metadata_file($full_metafile,$metafile);
+    }
+
+    $doc_obj->delete_metadata($top_section,"gsdlmetafile");
+}
+
+sub archiveinf_files_to_field
+{
+    my $self = shift(@_);
+    my ($files,$field,$collect_dir,$oid_files,$reverse_lookups) = @_;
+
+    foreach my $file_rec (@$files) {
+    my $real_filename = $file_rec->[0];
+    my $full_file = $file_rec->[1];
 
     # for some reasons the image associate file has / before the full path
@@ -690 +685 @@
         }
 
-        $reverse_lookups{$real_filename} = 1;
-
-        push(@{$oid_files->{'assoc-files'}},$full_afile);
-
+        $reverse_lookups->{$real_filename} = 1;
+
+        push(@{$oid_files->{$field}},$full_file);
     }
     else {
@@ -699 +693 @@
     }
     }
+}
+
+sub archiveinf_gdbm
+{
+    my $self = shift (@_);
+    my ($doc_obj) = @_;
+
+    my $verbosity = $self->{'verbosity'};
+
+    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
+    if (defined $collect_dir) {
+    my $dirsep_regexp = &util::get_os_dirsep();
+
+    if ($collect_dir !~ /$dirsep_regexp$/) {
+        # ensure there is a slash at the end
+        $collect_dir .= &util::get_dirsep(); 
+    }
+    }
+
+    my $oid = $doc_obj->get_OID();
+    my $source_filename = $doc_obj->get_source_filename();
+
+    my $working_info = $self->{'output_info'}; 
+    my $doc_info = $working_info->get_info($oid);
+    my ($doc_file,$index_status) = @$doc_info;
+
+    my $oid_files = { 'doc-file' => $doc_file,
+              'index-status' => $index_status,
+              'src-file' => $source_filename,
+              'assoc-files' => [],
+              'meta-files'  => [] };
+    
+    my $reverse_lookups = { $source_filename => "1" };
+
+
+    $self->archiveinf_files_to_field($doc_obj->get_assoc_files(),"assoc-files",
+                     $collect_dir,$oid_files,$reverse_lookups);
+
+#    foreach my $assoc_file_rec (@{$doc_obj->get_assoc_files()}) {
+#   my $real_filename = $assoc_file_rec->[0];
+#   my $full_afile = $assoc_file_rec->[1];
+#
+#   # for some reasons the image associate file has / before the full path
+#   $real_filename =~ s/^\\(.*)/$1/i;
+#   if (-e $real_filename) {
+#
+#       if (defined $collect_dir) {
+#       my $collect_dir_re_safe = $collect_dir;
+#       $collect_dir_re_safe =~ s/\\/\\\\/g;
+#       $collect_dir_re_safe =~ s/\./\\./g;
+#
+#       $real_filename =~ s/^$collect_dir_re_safe//;
+#       }
+#
+#       $reverse_lookups->{$real_filename} = 1;
+#
+#       push(@{$oid_files->{'assoc-files'}},$full_afile);
+#   }
+#   else {
+#       print STDERR "Warning: archiveinf_gdbm()\n  $real_filename does not appear to be on the file system\n";
+#   }
+#    }
+
+    $self->archiveinf_files_to_field($doc_obj->get_meta_files(),"meta-files",
+                     $collect_dir,$oid_files,$reverse_lookups);
+
 
     # better not to commit to a particular db implementation, but
@@ -712 +772 @@
     $doc_db_text .= "<index-status>$oid_files->{'index-status'}\n";
     $doc_db_text .= "<src-file>$oid_files->{'src-file'}\n";
+
     foreach my $af (@{$oid_files->{'assoc-files'}}) {
     $doc_db_text .= "<assoc-file>$af\n";
     }
+
+    foreach my $mf (@{$oid_files->{'meta-files'}}) {
+    $doc_db_text .= "<meta-file>$mf\n";
+    }
+
     chomp($doc_db_text); # remove trailing \n
 
@@ -720 +786 @@
     &GDBMUtils::gdbmDatabaseSet($doc_db,$oid,$doc_db_text);
 
-    foreach my $rl (keys %reverse_lookups) {
+    foreach my $rl (keys %$reverse_lookups) {
     &GDBMUtils::gdbmDatabaseAppend($src_db,$rl,"<oid>$oid\n");
     }

gsdl/trunk/perllib/plugouts/GreenstoneXMLPlugout.pm

@@ -77 +77 @@
     $self->process_assoc_files ($doc_obj, $doc_dir, '');
     
+    $self->process_metafiles_metadata ($doc_obj);
+
     my $output_file = util::filename_cat ($working_dir, "doc.xml");