Changeset 23280 for main/trunk

Timestamp:

2010-11-03T10:48:50+13:00 (13 years ago)

Author:

kjdon

Message:

fixed this plugin up for incremental import. need to set gsdlzipfilename for all files unpacked into tmp dir. Also, if incremental, need to add all the files unpacked to the new files list so they get processed. If the zip file hasn't changed, then won't get into this read method.

File:

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

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

@@ -85 +85 @@
 }
 
+sub begin {
+    my $self = shift (@_);
+    my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
+
+    # Are we actually incremental and doing import?
+    my $proc_package_name = ref $processor;
+    if ($proc_package_name !~ /buildproc$/ && $self->{'incremental'} == 1) {
+        # Get the infodbtype value for this collection from the arcinfo object
+        my $infodbtype = $processor->getoutputinfo()->{'infodbtype'};
+    $infodbtype = "gdbm" if $infodbtype eq "gdbm-txtgz"; # in archives, cannot use txtgz version
+    my $output_dir = $processor->getoutputdir();
+        my $archives_inf = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $output_dir);
+
+    if ( -e $archives_inf ) {
+        $self->{'actually_incremental'} = 1;
+    }
+    }
+   
+    
+}
 # this is a recursive plugin
 sub is_recursive {
@@ -108 +128 @@
     return undef unless $self->can_process_this_file($filename_full_path);
     
-    my $tmpdir = &util::get_tmp_filename ();
+    my $tmpdir = $file;
+    $tmpdir =~ s/\.[^\.]*//;
+    $tmpdir = &util::rename_file($tmpdir, $self->{'file_rename_method'});
+    $tmpdir = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "tmp", $tmpdir);
     &util::mk_all_dir ($tmpdir);
     
@@ -136 +159 @@
     &plugin::file_block_read ($pluginfo, "", $tmpdir,
                   $block_hash, $metadata, $gli);
-    my $numdocs = &plugin::read ($pluginfo, "", $tmpdir, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli);
+
+    # if we are incremental, then we need to add all the files in the tmp folder into the new_files list otherwise they won't get processed.
+    if ($self->{'actually_incremental'}) {
+    my @file_list = ();
+    &inexport::add_dir_contents_to_list($tmpdir, \@file_list);
+    foreach my $file (@file_list) {
+        $block_hash->{'new_files'}->{$file} = 1;
+    }
+    }
+    # all files in the tmp folder need to get the gsdlzipfilenmae metadata
+    my $this_metadata = {};
+    $this_metadata->{"gsdlzipfilename"} = $filename_full_path;
+    &metadatautil::combine_metadata_structures($this_metadata, $metadata);
+    my $numdocs = &plugin::read ($pluginfo, "", $tmpdir, $block_hash, $this_metadata, $processor, $maxdocs, $total_count, $gli);
     &util::rm_r ($tmpdir);
     
@@ -191 +227 @@
 }
 
+
+
 1;