Changeset 11994 for trunk/gsdl/perllib/basebuildproc.pm

Timestamp:

2006-07-04T15:06:34+12:00 (18 years ago)

Author:

davidb

Message:

Improved support for incremental addition: instead of having to run the
classifier pass of buildcol.pl from scratch (i.e. read in all documents
from the archives folder) so correct browse structures are formed -- a
simple to implement strategy, but not very efficient -- the first layer
of a classifier structure is now reconstructed from the GDBM file. Then
the new files in the archives directory are added, and then finally the
completed browser structure is formed.

File:

• trunk/gsdl/perllib/basebuildproc.pm (modified) (4 diffs)

trunk/gsdl/perllib/basebuildproc.pm

@@ -296 +296 @@
     return if ($doctype ne "indexed_doc" && $doctype ne "info_doc");
 
-    my ($archivedir) = $filename =~ /^(.*?)(?:\/|\\)[^\/\\]*$/;
-    $archivedir = "" unless defined $archivedir;
-    $archivedir =~ s/\\/\//g;
-    $archivedir =~ s/^\/+//;
-    $archivedir =~ s/\/+$//;
-
-    # resolve the final filenames of the files associated with this document
-    $self->assoc_files ($doc_obj, $archivedir);
+    my $archivedir = "";
+
+    if (defined $filename)
+    {
+    # doc_obj derived directly from file
+
+    my ($dir) = $filename =~ /^(.*?)(?:\/|\\)[^\/\\]*$/;
+    $dir = "" unless defined $dir;
+    $dir =~ s/\\/\//g;
+    $dir =~ s/^\/+//;
+    $dir =~ s/\/+$//;
+
+    $archivedir = $dir;
+
+    # resolve the final filenames of the files associated with this document
+    $self->assoc_files ($doc_obj, $archivedir);
+    }
+    else
+    {
+    # doc_obj reconstructed from GDBM (has metadata, doc structure but no text)
+    my $top_section = $doc_obj->get_top_section();
+    $archivedir = $doc_obj->get_metadata_element($top_section,"archivedir");
+    }
+
 
     #GRB: moved 1/06/2004 from GRB01062004
@@ -340 +356 @@
     }
 
-    # output whether this node contains text
-    if ($doc_obj->get_text_length($section) > 0) {
-        print $handle "<hastxt>1\n";
-    } else {
-        print $handle "<hastxt>0\n";
+    # Output whether this node contains text
+    #
+    # If doc_obj reconstructed from GDBM file then no need to 
+    # explicitly add <hastxt> as this is preserved as metadata when
+    # the GDBM file is loaded in
+
+    if (defined $filename)
+    {
+        # doc_obj derived directly from file
+        if ($doc_obj->get_text_length($section) > 0) {
+        print $handle "<hastxt>1\n";
+        } else {
+        print $handle "<hastxt>0\n";
+        }
     }
 
@@ -380 +405 @@
     }
 
-    # output archivedir if at top level
-    if ($section eq $doc_obj->get_top_section()) {
-        print $handle "<archivedir>$archivedir\n";
+
+    # If doc_obj reconstructed from GDBM file then no need to 
+    # explicitly add <archivedir> as this is preserved as metadata when
+    # the GDBM file is loaded in
+
+    if (defined $filename)
+    {
+        # output archivedir if at top level
+        if ($section eq $doc_obj->get_top_section()) {
+        print $handle "<archivedir>$archivedir\n";
+        }
     }
 
@@ -389 +422 @@
         print $handle "<thistype>$thistype\n";
     }
+
 
     if ($self->{'gdbm_level'} eq "document") {