Changeset 23167

Timestamp:

2010-10-19T14:14:18+13:00 (14 years ago)

Author:

davidb

Message:

GreenstoneXMLPlugin used to (or at least in theory used to) to be able to process files without them needing to be fed down the pipeline from ArchiveInfPlugin; however, this has not been tested for a long time. A very long time. It's certainly the case its not working now. For some new work, we need GreenstoneXMLPlugin to work correctly when processing doc.xml (and associated files all in the same folder) directly. For this new use, this occurs when it is placed in the *import* folder (part of the DL talkback scheme). The mods being commited here to the plugin give this capability back to the plugin.

File:

• main/trunk/greenstone2/perllib/plugins/GreenstoneXMLPlugin.pm (modified) (4 diffs)

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

@@ -33 +33 @@
 
 use Encode;
+use File::Basename;
 
 use ReadXMLFile;
@@ -80 +81 @@
     $self->{'metadata_value'} = "";
     $self->{'content'} = "";
+    $self->{'metadata_read_store'} = {};
 
 #    # Currently used to store information for previous values controls. In
@@ -87 +89 @@
     return bless $self, $class;
 }
+
+
+
+
+sub metadata_read {
+    my $self = shift (@_);
+    my ($pluginfo, $base_dir, $file, $block_hash, 
+    $extrametakeys, $extrametadata, $extrametafile,
+    $processor, $maxdocs, $gli) = @_;
+
+    my $outhandle = $self->{'outhandle'};
+
+    # can we process this file??
+    my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
+    return undef unless $self->can_process_this_file($filename_full_path);
+
+    $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
+    
+    print $outhandle "GreenstoneXMLlugin: setting up block list for $file\n"
+    if $self->{'verbosity'} > 1;
+
+    my $line;
+    if (open(GIN,"<:utf8",$filename_full_path)) {
+
+    while (defined($line=<GIN>)) {
+        if ($line =~ m@<Metadata\s+name="gsdlassocfile">([^:]*):(?:[^:]*):(?:[^:]*)</Metadata>@) {
+        my $gsdl_assoc_file = $1;
+
+        my $dirname = dirname($filename_full_path);
+        my $full_gsdl_assoc_filename = &util::filename_cat($dirname,$gsdl_assoc_file);
+        if ($self->{'verbosity'}>2) {
+            print $outhandle "  Storing block list item: $full_gsdl_assoc_filename\n";
+        }
+
+        $block_hash->{'file_blocks'}->{$full_gsdl_assoc_filename} = 1;
+        }
+    }
+
+    close(GIN);
+    }
+    else {
+    
+    print $outhandle "Error: Failed to open $file in GreenstoneXMLPlugin::metadata_read()\n";
+    print $outhandle "       $!\n";
+    }
+
+
+    $self->{'metadata_read_store'}->{$filename_full_path} = 1;
+
+    return 1;
+}
+
 
 sub xml_start_document {
@@ -221 +275 @@
     my $self = shift(@_);
 
+    my $filename = $self->{'filename'};
+
     # create a new document
-    $self->{'doc_obj'} = new doc ();
+    if (defined $self->{'metadata_read_store'}->{$filename}) {
+    # Being processed as part of *import* phase 
+    # (i.e. was in import directory)
+    $self->SUPER::open_document(@_);
+    delete $self->{'metadata_read_store'}->{$filename};
+    }
+    else {
+    # Otherwise being processed as part of the *buildcol* phase 
+    # (i.e. named directly by ArchiveInf plugin)
+    $self->{'doc_obj'} = new doc();
+    }
+
     $self->{'section'} = "";
 }