Changeset 28637

Timestamp:

2013-11-19T11:10:11+13:00 (10 years ago)

Author:

kjdon

Message:

added an extra field to the database: group-position. When we are processing documents into grouped doc.xml files, then this field will give the position in the doc.xml file, starting with the first document at 1. When we are reading the database to find the list of files to process for indexing, we must not process items where the group-position is > 1 - we have already seen this doc.xml file once, don't process it again.

File:

• main/trunk/greenstone2/perllib/arcinfo.pm (modified) (7 diffs)

main/trunk/greenstone2/perllib/arcinfo.pm

@@ -40 +40 @@
 use constant INFO_STATUS_INDEX  => 1;
 
+use constant INFO_GROUPPOS_INDEX  => 3;
 use strict;
 
@@ -112 +113 @@
     my ($index_status) = ($vals=~/^<index-status>(.*)$/m);
     my ($sortmeta) = ($vals=~/^<sort-meta>(.*)$/m);
-    $self->add_info ($oid,$doc_file,$index_status,$sortmeta);
+    my ($group_position) = ($vals=~/^<group-position>(.*)$/m);
+    $self->add_info ($oid,$doc_file,$index_status,$sortmeta, $group_position);
     }
 }
@@ -268 +270 @@
     my $self = shift (@_);
     my ($filename) = @_;
-
     if ($filename =~ m/(contents)|(\.inf)$/) {
     $self->_save_info_txt($filename);
@@ -298 +299 @@
 sub add_info {
     my $self = shift (@_);
-    my ($OID, $doc_file, $index_status, $sortmeta) = @_;
+    my ($OID, $doc_file, $index_status, $sortmeta, $group_position) = @_;
     $sortmeta = "" unless defined $sortmeta;
     $index_status = "I" unless defined $index_status; # I = needs indexing
-
     if (! defined($OID)) {
     # only happens when no files can be processed?
@@ -331 +331 @@
     }
 
-    $self->{'info'}->{$OID} = [$doc_file,$index_status,$sortmeta];
+    $self->{'info'}->{$OID} = [$doc_file,$index_status,$sortmeta, $group_position];
     push (@{$self->{'order'}}, [$OID, $sortmeta]); # ORDER_OID_INDEX and ORDER_SORT_INDEX
 
@@ -364 +364 @@
 }
 
-
+sub get_group_position {
+    my $self = shift (@_);
+    my ($OID) = @_;
+
+    my $group_position = undef;
+    my $OID_info = $self->{'info'}->{$OID};
+    if (defined $OID_info) {
+    $group_position = $OID_info->[INFO_GROUPPOS_INDEX];
+    }
+    else {
+    die "Unable to find document id $OID\n";
+    }
+    return $group_position;
+    
+}
 sub add_reverseinfo {
     my $self = shift (@_);
@@ -466 +480 @@
 
 
-# returns a list of the form [doc_file,index_status,$sort_meta]
+# returns a list of the form [doc_file,index_status,$sort_meta, $group_position]
 sub get_info {
     my $self = shift (@_);