Changeset 18456 for gsdl/trunk/perllib/basebuildproc.pm

Timestamp:

2009-02-03T09:48:19+13:00 (15 years ago)

Author:

davidb

Message:

Additions to support the deleting of documents from the index. Only works for indexers that support incremental building, e.g. lucene

File:

• gsdl/trunk/perllib/basebuildproc.pm (modified) (6 diffs)

gsdl/trunk/perllib/basebuildproc.pm

@@ -35 +35 @@
 use doc;
 use docproc;
-use strict; no strict 'subs';
+use strict; 
+no strict 'subs';
+no strict 'refs';
 use util;
 
@@ -366 +368 @@
 
 
-sub infodb {
-    my $self = shift (@_);
-    my ($doc_obj, $filename) = @_;
+sub infodbedit {
+    my $self = shift (@_);
+    my ($doc_obj, $filename, $edit_mode) = @_;
 
     # only output this document if it is a "indexed_doc" or "info_doc" (database only) document
@@ -396 +398 @@
     }
 
-    #add this document to the browse structure
-    push(@{$self->{'doclist'}},$doc_obj->get_OID()) 
-    unless ($doctype eq "classification");
+    if (($edit_mode eq "add") || ($edit_mode eq "reindex")) {
+    #add this document to the browse structure
+    push(@{$self->{'doclist'}},$doc_obj->get_OID()) 
+        unless ($doctype eq "classification");
+    }
+    else {
+    # delete => remove this doc from browse structure
+    my $del_doc_oid = $doc_obj->get_OID();
+
+    my @filtered_doc_list = ();
+    foreach my $oid (@{$self->{'doclist'}}) {
+        push(@filtered_doc_list,$oid) if ($oid ne $del_doc_oid);
+    }
+    $self->{'doclist'} = \@filtered_doc_list;
+    }
+
 
     # classify this document
-    &classify::classify_doc ($self->{'classifiers'}, $doc_obj);
-
-    # this is another document
-    $self->{'num_docs'} += 1 unless ($doctype eq "classification");
+    &classify::classify_doc ($self->{'classifiers'}, $doc_obj, $edit_mode);
+
+    if (($edit_mode eq "add") || ($edit_mode eq "reindex")) {
+    # this is another document
+    $self->{'num_docs'} += 1 unless ($doctype eq "classification");
+    }
+    else {
+    # delete
+    $self->{'num_docs'} -= 1 unless ($doctype eq "classification");
+    return;
+    }
 
     # is this a paged or a hierarchical document
@@ -563 +585 @@
 
 
+
+
+sub infodb {
+    my $self = shift (@_);
+    my ($doc_obj, $filename) = @_;
+
+    $self->infodbedit($doc_obj,$filename,"add");
+}
+
+sub infodbreindex {
+    my $self = shift (@_);
+    my ($doc_obj, $filename) = @_;
+
+    $self->infodbedit($doc_obj,$filename,"reindex");
+}
+
+sub infodbdelete {
+    my $self = shift (@_);
+    my ($doc_obj, $filename) = @_;
+
+    $self->infodbedit($doc_obj,$filename,"delete");
+}
+
+
 sub text {
     my $self = shift (@_);
@@ -571 +617 @@
     die "\n";
 }
+
+sub textreindex
+{
+    my $self = shift @_;
+
+    my $outhandle = $self->{'outhandle'};
+    print $outhandle "basebuildproc::textreindex function must be implemented in sub classes\n";
+    if (!$self->is_incremental_capable()) {
+
+    print $outhandle "  This operation is only possible with indexing tools with that support\n";
+    print $outhandle "  incremental building\n";
+    }
+    die "\n";
+}
+
+sub textdelete
+{
+    my $self = shift @_;
+
+    my $outhandle = $self->{'outhandle'};
+    print $outhandle "basebuildproc::textdelete function must be implemented in sub classes\n";
+    if (!$self->is_incremental_capable()) {
+
+    print $outhandle "  This operation is only possible with indexing tools with that support\n";
+    print $outhandle "  incremental building\n";
+    }
+    die "\n";
+}
+
 
 # should the document be indexed - according to the subcollection and language
@@ -692 +767 @@
 }
 
-sub assoc_files() {
+sub assoc_files 
+{
     my $self = shift (@_);
     my ($doc_obj, $archivedir) = @_;