Changeset 27697

Timestamp:

2013-06-24T17:23:04+12:00 (11 years ago)

Author:

ak19

Message:

Dr Bainbridge fixed it so that the gdb files generated on Windows for diffcol match those on Linux. This actually involved changing the order in which docids appear in archiveinf-doc. This last needed the newly invented flag -sort to the ArchivesInfPlugin in combination with -sortmeta OID to import.pl

Location:

main/trunk/greenstone2/perllib

Files:

• arcinfo.pm (modified) (5 diffs)
• plugins/ArchivesInfPlugin.pm (modified) (2 diffs)
• plugouts/BasePlugout.pm (modified) (1 diff)
• strings.properties (modified) (2 diffs)

main/trunk/greenstone2/perllib/arcinfo.pm

@@ -67 +67 @@
         'reverse-info'=>{},
         'order'=>[],
-        'reverse_sort'=>0};
+        'reverse_sort'=>0,
+        'sort'=>0};
 
     return bless $self, $class;
@@ -323 +324 @@
 
     $self->{'info'}->{$OID} = [$doc_file,$index_status,$sortmeta];
-    push (@{$self->{'order'}}, [$OID, $sortmeta]);
+    push (@{$self->{'order'}}, [$OID, $sortmeta]); # ORDER_OID_INDEX and ORDER_SORT_INDEX
 
 
@@ -393 +394 @@
     $self->{'reverse_sort'} = 1;
 }
+sub sort 
+{
+    my $self = shift(@_);
+    $self->{'sort'} = 1;
+}
+
 
 # returns a list of the form [[OID, doc_file, index_status], ...]
@@ -404 +411 @@
     if ($self->{'reverse_sort'}) {
     @sorted_order = sort {$b->[ORDER_SORT_INDEX] cmp $a->[ORDER_SORT_INDEX]} @$order;
-    } else {
+    } elsif ($self->{'sort'}) {
     @sorted_order = sort {$a->[ORDER_SORT_INDEX] cmp $b->[ORDER_SORT_INDEX]} @$order;
+    } else { # not sorting, don't bother
+    @sorted_order = @$order;
     }
 
@@ -430 +439 @@
     if ($self->{'reverse_sort'}) {
     @sorted_order = sort {$b->[ORDER_SORT_INDEX] cmp $a->[ORDER_SORT_INDEX]} @$order;
-    } else {
+    } elsif ($self->{'sort'}) {
     @sorted_order = sort {$a->[ORDER_SORT_INDEX] cmp $b->[ORDER_SORT_INDEX]} @$order;
+    } else { # not sorting, don't bother
+    @sorted_order = @$order;
     }
 

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

@@ -52 +52 @@
     'reqd' => "no",
     'modegli' => "2" },
+    { 'name' => "sort",
+    'desc' => "{ArchivesInfPlugin.sort}",
+    'type' => "flag",
+    'reqd' => "no",
+    'modegli' => "2" }
 
          ];
@@ -244 +249 @@
     if ($self->{'reversesort'}) {
         $archive_info->reverse_sort();
+    } elsif ($self->{'sort'}) { 
+        $archive_info->sort();
     }
     

main/trunk/greenstone2/perllib/plugouts/BasePlugout.pm

@@ -477 +477 @@
     my $output_info = $self->{'output_info'};
     my $metaname = $self->{'sortmeta'};
+    
     if (!defined $metaname || $metaname !~ /\S/) {
-    $output_info->add_info($doc_obj->get_OID(),$self->{'short_doc_file'}, undef, "");
+    my $OID = $doc_obj->get_OID();
+    $output_info->add_info($OID,$self->{'short_doc_file'}, undef, "");
     return;
     }
+    
+    if ($metaname eq "OID") { # sort by OID
+    my $OID = $doc_obj->get_OID();
+    $output_info->add_info($OID,$self->{'short_doc_file'}, undef, $OID);
+    return;
+    }
     
     my $metadata = "";

main/trunk/greenstone2/perllib/strings.properties

@@ -314 +314 @@
 import.site:Site to find collect directory in (for Greenstone 3 installation).
 
-import.sortmeta:Sort documents alphabetically by metadata for building. Search results for boolean queries will be displayed in this order. This will be disabled if groupsize > 1. May be a commma separated list to sort by more than one metadata value.
+import.sortmeta:Sort documents alphabetically by metadata for building (specifying -sortmeta as 'OID' is a special case, and instructs Greenstone to use the document identifier for ordering). Search results for boolean queries will be displayed in the order determined by sortmeta. This will be disabled if groupsize > 1. May be a commma separated list to sort by more than one metadata value.
 
 import.statsfile:Filename or handle to print import statistics to.
@@ -742 +742 @@
 
 ArchivesInfPlugin.desc:Plugin which processes the archive info database (archiveinf-doc) which is generated by the import process. It passes each archive file listed in the database to the plugin pipeline to be processed by GreenstoneXMLPlugin.
-ArchivesInfPlugin.reversesort:Sort in reverse order. Useful if the -sortmeta option was used with import.pl.
+ArchivesInfPlugin.reversesort:Sort in reverse alphabetical order. Useful if the -sortmeta option was used with import.pl.
+ArchivesInfPlugin.sort:Sort in ascending alphabetical order. Useful if the -sortmeta option was used with import.pl.
 
 AutoExtractMetadata.desc: Base plugin that brings together all the extractor functionality from the extractor plugins.