Changeset 25846

Timestamp:

2012-06-28T11:11:31+12:00 (11 years ago)

Author:

sjm84

Message:

Some fixes and additions to the Solr perl code

Location:

gs3-extensions/solr/trunk/src

Files:

• bin/script/debug_solr_server.pl (added)
• perllib/solrbuilder.pm (modified) (2 diffs)
• perllib/solrbuildproc.pm (modified) (4 diffs)
• perllib/solrutil.pm (modified) (1 diff)

gs3-extensions/solr/trunk/src/perllib/solrbuilder.pm

@@ -229 +229 @@
     foreach my $ifm (@{$self->{'build_cfg'}->{'indexfieldmap'}}) {
 
-    my ($field) = ($ifm =~ m/^.*->(.*)$/);
-
-    # Need special case for Long/Lat
-    # ... but for now treat everything as of type string
-
-    $schema_insert_xml .= "    "; # indent
-    $schema_insert_xml .= "<field name=\"$field\" ";
-    $schema_insert_xml .=   "type=\"text_en_splitting\" indexed=\"true\" ";
-    $schema_insert_xml .=   "stored=\"false\" multiValued=\"true\" />\n"; 
+        my ($field) = ($ifm =~ m/^.*->(.*)$/);
+
+        $schema_insert_xml .= "    "; # indent
+        $schema_insert_xml .= "<field name=\"$field\" ";
+
+        if($field eq "LA" || $field eq "LO")
+        {
+            $schema_insert_xml .=   "type=\"location\" ";
+        }
+        elsif ($field ne "ZZ" && $field ne "TX")
+        {
+            $schema_insert_xml .=   "type=\"string\" ";
+        }
+        else
+        {
+            $schema_insert_xml .= "type=\"text_en_splitting\" ";
+        }
+        $schema_insert_xml .=  "indexed=\"true\" stored=\"false\" multiValued=\"true\" />\n"; 
     }
 
@@ -544 +553 @@
     &plugin::read ($self->{'pluginfo'}, $self->{'source_dir'},
            "", {}, {}, $self->{'buildproc'}, $self->{'maxdocs'}, 0, $self->{'gli'});
-
-
+    
     print $handle "</update>\n";
 

gs3-extensions/solr/trunk/src/perllib/solrbuildproc.pm

@@ -266 +266 @@
     my $gs2_docOID = $doc_obj->get_OID();
 
-
     my $start_doc;
     my $end_doc;
@@ -296 +295 @@
 
     # only output if working with doc level
-    my $text = $start_doc if ($sec_tag_name eq "");
-
+    # my $text = undef;
+    
+    my $text = ($sec_tag_name eq "") ? $start_doc : "";
+
+#     my $text = $start_doc if ($sec_tag_name eq "");
+      
     # get the text for this document
     my $section = $doc_obj->get_top_section();
@@ -318 +321 @@
         $start_sec .= "    <doc>\n";
         $start_sec .= "      <field name=\"docOID\">$sec_gs2_docOID</field>\n";
-
+    
         $end_sec    = "    </doc>\n";
         $end_sec   .= "  </add>\n"; 
@@ -441 +444 @@
         }
 
+        # Remove any leading or trailing white space
+        $new_text =~ s/\s+$//;
+        $new_text =~ s/^\s+//;
+    
+        
         if ($self->{'indexing_text'}) {
             # add the tag

gs3-extensions/solr/trunk/src/perllib/solrutil.pm

@@ -79 +79 @@
     $post_props .= " -Dcommit=yes";
     
-    my $post_java_cmd = "java $post_props -jar \"$full_post_jar\"";
+    my $post_java_cmd = "java -Xmx256M $post_props -jar \"$full_post_jar\"";
     
-##  print STDERR "**** post cmd = $post_java_cmd\n";
+    ##print STDERR "**** post cmd = $post_java_cmd\n";
     
     open (PIPEOUT, "| $post_java_cmd")