Changeset 26324

Timestamp:

2012-10-16T21:12:31+13:00 (12 years ago)

Author:

ak19

Message:

1. When purging documents from the fedora repository during import and buildcol, these documents are also removed from the Fedora GSearch index. Likewise, when documents are ingested into the Fedora repository during buildcol, these documents are also indexed by Fedora GSearch. 2. Added methods in g2futil.pm to be able to do this

Location:

main/trunk/greenstone2

Files:

• bin/script/g2f-buildcol.pl (modified) (3 diffs)
• bin/script/g2f-import.pl (modified) (2 diffs)
• perllib/g2futil.pm (modified) (2 diffs)

main/trunk/greenstone2/bin/script/g2f-buildcol.pl

@@ -208 +208 @@
     # => Ingest individually!
 
+    # set up fedoragsearch for updating the index upon ingesting documents
+    my $fedoragsearch_webapp = &g2futil::gsearch_webapp_folder();    
+
+    # need the username and password preset in order to run fedoraGSearch's RESTClient script
+    # this assumes that the fedoragsearch authentication details are the same as for fedora
+    if (defined $fedoragsearch_webapp) {    
+    $ENV{'fgsUserName'} = $options->{'username'};
+    $ENV{'fgsPassword'} = $options->{'password'};
+    }
+
     if (opendir(DIR, $export_dir)) {
     closedir DIR;
@@ -227 +237 @@
         
         if ($dsinfo_status == 0) {
-            print "  $pid being updated.\n";        
+            # first remove the doc from the gsearch index before removing it from the fedora repository
+            print "  deleting $pid from GSearch index\n";
+            &g2futil::run_delete_from_index($fedoragsearch_webapp,$pid,$options) if defined $fedoragsearch_webapp;
+
+            print "  $pid being updated.\n";
             &g2futil::run_purge($pid,$options);
         }
@@ -242 +256 @@
         print "Ingesting $docmets_filename\n";
 
-        &g2futil::run_ingest($docmets_filename,$options);
+        my $status = &g2futil::run_ingest($docmets_filename,$options);
+
+        # if the document was ingested into Fedora successfully, index it with GSearch next
+        if($status == 0) {
+        if(defined $hash_id) {
+            my $pid = "$pid_namespace:$gs_col-$hash_id";
+            # now update the fedoragsearch index with the newly ingested document
+            &g2futil::run_update_index($fedoragsearch_webapp,$pid,$options) if defined $fedoragsearch_webapp;
+        }
+        }
+
         print STDERR "</Build>\n" if $gli;
 

main/trunk/greenstone2/bin/script/g2f-import.pl

@@ -209 +209 @@
     print "* Updating existing Greenstone $gs_col objects from Fedora $pid_namespace\n";
         print "***\n";
+
+    # set up fedoragsearch for updating the index upon ingesting documents
+    my $fedoragsearch_webapp = &g2futil::gsearch_webapp_folder();
+    
+    # need the username and password preset in order to run fedoraGSearch's RESTClient script
+    # this assumes that the fedoragsearch authentication details are the same as for fedora
+    if (defined $fedoragsearch_webapp) {    
+        $ENV{'fgsUserName'} = $options->{'username'};
+        $ENV{'fgsPassword'} = $options->{'password'};
+    }
 
     # readdir
@@ -243 +253 @@
 
         if ($dsinfo_status == 0) {
+            # first remove the doc from the gsearch index before removing it from the fedora repository
+            print "  deleting $pid from GSearch index\n";
+            &g2futil::run_delete_from_index($fedoragsearch_webapp,$pid,$options) if defined $fedoragsearch_webapp;
+
             print "  $pid being updated.\n";        
             &g2futil::run_purge($pid,$options);

main/trunk/greenstone2/perllib/g2futil.pm

@@ -59 +59 @@
     my $script_ext = ($ENV{'GSDLOS'} =~ m/^windows/) ? ".bat" : ".sh";
 
-    if ($prog =~ m/^fedora-/) {
+    if ($prog =~ m/^fedora-/ || $prog =~ m/^run[A-Z]*Client/) { # fedora or fedoragsearch script
     $prog .= $script_ext;
     }
@@ -168 +168 @@
     return $status;
 }
+
+# runs fedora gsearch's runRESTClient.sh: updateIndex deletePID <PID>
+sub run_delete_from_index
+{
+    my ($fedoragsearch_webapp,$pid,$options) = @_;
+
+    my $verbosity = $options->{'verbosity'};
+
+    my $hostname = $options->{'hostname'};
+    my $port     = $options->{'port'};
+    my $username = $options->{'username'};
+    my $password = $options->{'password'};
+    my $protocol = $options->{'protocol'};
+
+    my $server = "$hostname:$port";
+    #$ENV{'fgsUserName'} = $options->{'username'};
+    #$ENV{'fgsPassword'} = $options->{'password'};    
+
+    #my $prog = &util::filename_cat($ENV{'FEDORA_GSEARCH'}, "runRESTClient.sh");
+    my $prog = &util::filename_cat($fedoragsearch_webapp, "client", "runRESTClient.sh");
+
+    my $gsearch_commands = "updateIndex deletePID";
+    my $arguments = "$server $gsearch_commands $pid";    
+
+    my $status = run_cmd($prog,$arguments,$verbosity);
+
+    return $status;
+}
+
+# runs fedora gsearch's runRESTClient.sh: updateIndex fromPID <PID>
+sub run_update_index
+{
+    my ($fedoragsearch_webapp,$pid,$options) = @_;
+
+    my $verbosity = $options->{'verbosity'};
+
+    my $hostname = $options->{'hostname'};
+    my $port     = $options->{'port'};
+    my $username = $options->{'username'};
+    my $password = $options->{'password'};
+    my $protocol = $options->{'protocol'};
+
+    my $server = "$hostname:$port";
+    #$ENV{'fgsUserName'} = $options->{'username'};
+    #$ENV{'fgsPassword'} = $options->{'password'};    
+
+    #my $prog = &util::filename_cat($ENV{'FEDORA_GSEARCH'}, "runRESTClient.sh");
+    my $prog = &util::filename_cat($fedoragsearch_webapp, "client", "runRESTClient.sh");
+    
+    my $gsearch_commands = "updateIndex fromPID";
+    my $arguments = "$server $gsearch_commands $pid";    
+
+    my $status = run_cmd($prog,$arguments,$verbosity);
+
+    return $status;
+}
+
+sub gsearch_webapp_folder
+{   
+    my $fedoragsearch_webapp = undef;
+    
+    # if GS3, first look for a fedoragsearch webapp installed in Greenstone's tomcat
+    if(defined $ENV{'GSDL3SRCHOME'}) {
+    $fedoragsearch_webapp = &util::filename_cat($ENV{'GSDL3SRCHOME'},"packages","tomcat","webapps","fedoragsearch");    
+    return $fedoragsearch_webapp if (&util::dir_exists($fedoragsearch_webapp));
+    }
+
+    # next look for a fedoragsearch webapp installed in Fedora's tomcat
+    if(defined $ENV{'FEDORA_HOME'}) {
+    $fedoragsearch_webapp =  &util::filename_cat($ENV{'FEDORA_HOME'},"tomcat","webapps","fedoragsearch");
+    return $fedoragsearch_webapp if (&util::dir_exists($fedoragsearch_webapp));
+    }
+
+    ## check for a user-defined $ENV{'FEDORA_GSEARCH'} variable first, which points to a gsearch webapp folder??
+
+    # assume no fedoragsearch
+    return $fedoragsearch_webapp; # undef
+}
+
 
 sub run_ingest