Changeset 31190 for main/trunk/greenstone2

Timestamp:

2016-12-09T21:37:52+13:00 (7 years ago)

Author:

ak19

Message:

First major commit to do with the new oaiinfo db that keeps track of which oids are deleted. Still need to fix up issue with the new remove and rename subroutines of dbutil's jdbm not being called to clean up *.lg log file associated with main db file. Still need to clean up unused methods in oaiinfo, remove debugging and test agains GS3 incr-build-with-manifest tutorial. (Previous 3 commits were commits, not all related.)

Location:

main/trunk/greenstone2

Files:

• bin/script/activate.pl (modified) (2 diffs)
• perllib/basebuilder.pm (modified) (3 diffs)
• perllib/inexport.pm (modified) (3 diffs)
• perllib/oaiinfo.pm (added)

main/trunk/greenstone2/bin/script/activate.pl

@@ -92 +92 @@
 # Greenstone modules
 use colcfg;
+use oaiinfo;
 use scriptutil;
 use servercontrol;
@@ -354 +355 @@
     }
     }
+
+
+    # now we've moved building to index, move tmp oaidb to live oaidb in parallel
+    my $oai_info = new oaiinfo($collect_cfg_filename, $collectcfg->{'infodbtype'});
+    $oai_info->activate_collection();
+
     
     if ($buildtype eq "solr") {

main/trunk/greenstone2/perllib/basebuilder.pm

@@ -29 +29 @@
 no strict 'refs'; # allow filehandles to be variables and viceversa
 
+use arcinfo;
 use classify;
 use cfgread;
 use colcfg;
 use dbutil;
+use oaiinfo;
 use plugin;
 use util;
@@ -92 +94 @@
 
     my $colcfgname = &colcfg::get_collect_cfg_name($outhandle, $gs_mode);
+    $self->{'colcfgname'} = $colcfgname;
     $self->{'collect_cfg'} = &colcfg::read_collection_cfg ($colcfgname, $gs_mode);
 
@@ -169 +172 @@
     }
 
+
+    # Prepare to work with the <collection>/etc/oai-inf.<db> that keeps track
+    # of the OAI identifiers with their time stamps and deleted status.
+    #
+    # At this stage of working with the oai info db, we don't care whether we have a
+    # manifest or are otherwise incremental, or whether we're doing removeold (full rebuild).
+    # Because we've already dealt with that during the import stage. From here on, we pretend
+    # we're incremental, since the oai info db should just do what archiveinfo contains.
+    # This is because "building is always incremental" where oai info db is concerned.
+    
+    my $archivedir = $self->{'source_dir'};
+    my $oai_info = new oaiinfo($self->{'colcfgname'}, $self->{'collect_cfg'}->{'infodbtype'});    
+    $oai_info->building_stage_before_indexing($archivedir);
+
+
     # check incremental against whether builder can cope or not.
     if ($self->{'incremental'} && !$self->is_incremental_capable()) {

main/trunk/greenstone2/perllib/inexport.pm

@@ -35 +35 @@
 use dbutil;
 use doc;
+use oaiinfo;
 use plugin;
 use plugout;
@@ -354 +355 @@
     my $gs_mode = "gs".$self->{'gs_version'}; #gs2 or gs3
     my $config_filename = &colcfg::get_collect_cfg_name($out, $gs_mode);
+
+    # store the config file's name, so oaiinfo object constructor can be instantiated with it
+    $self->{'config_filename'} = $config_filename;
+
     my $collectcfg = &colcfg::read_collection_cfg ($config_filename, $gs_mode);
 
@@ -725 +730 @@
       print STDERR "Skipping global file scan due to manifest and complexmeta configuration\n";
     }
+
+
+    # Prepare to work with the <collection>/etc/oai-inf.<db> that keeps track
+    # of the OAI identifiers with their time stamps and deleted status.    
+    my $oai_info = new oaiinfo(&util::get_parent_folder($self->{'importdir'}), $collectcfg->{'infodbtype'});
+    my $have_manifest = ($manifest eq '') ? 0 : 1;    
+    $oai_info->import_stage($removeold, $have_manifest);
+
 
     if ($manifest ne "") {