Changeset 17087 for gsdl/trunk/perllib/plugouts

Timestamp:

2008-08-29T13:10:39+12:00 (16 years ago)

Author:

davidb

Message:

Introduction of new GDBM alternative for archives.inf as step towards full incremental building. Information traditionally stored in archives.inf PLUS additional information that will help with working out what files have changed since last build, and what doc-id they hashed to is stored in two GDBM databases. For now these databases aren't read, but in the future ArchivesInfPlugin will be upgraded to use these to support these.

File:

• gsdl/trunk/perllib/plugouts/BasPlugout.pm (modified) (3 diffs)

gsdl/trunk/perllib/plugouts/BasPlugout.pm

@@ -35 +35 @@
 use printusage;
 use parse2;
+use GDBMUtils;
+
 
 # suppress the annoying "subroutine redefined" warning that various
@@ -402 +404 @@
     ##############################
     $self->saveas($doc_obj,$doc_dir);
-    
+    $self->archiveinf_gdbm($doc_obj,$doc_dir); 
+
 }
 
@@ -630 +633 @@
 }
 
+
+sub archiveinf_gdbm
+{
+    my $self = shift (@_);
+    my ($doc_obj) = @_;
+
+    my $verbosity = $self->{'verbosity'};
+
+    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
+    if (defined $collect_dir) {
+    my $dirsep_regexp = &util::get_os_dirsep();
+
+    if ($collect_dir !~ /$dirsep_regexp$/) {
+        # ensure there is a slash at the end
+        $collect_dir .= &util::get_dirsep(); 
+    }
+    }
+
+    my $oid = $doc_obj->get_OID();
+    my $source_filename = $doc_obj->get_source_filename();
+
+    my $oid_files = { 'src-file' => $source_filename,
+              'assoc-files' => [] };
+    
+    my @reverse_lookups = ($source_filename);
+
+    foreach my $assoc_file_rec (@{$doc_obj->get_assoc_files()}) {
+    my $real_filename = $assoc_file_rec->[0];
+    my $full_afile = $assoc_file_rec->[1];
+
+    # for some reasons the image associate file has / before the full path
+    $real_filename =~ s/^\\(.*)/$1/i;
+    if (-e $real_filename) {
+
+        if (defined $collect_dir) {
+        $real_filename =~ s/^$collect_dir//;
+        }
+
+        push(@reverse_lookups,$real_filename);
+
+        push(@{$oid_files->{'assoc-files'}},$full_afile);
+
+    }
+    else {
+        print STDERR "Warning: archiveinf_gdbm()\n  $real_filename does not appear to be on the file system\n";
+    }
+    }
+
+    # better not to commit to a particular db implementation, but
+    # for simplicity, will use GDBM for now.
+
+    my $output_dir = $self->{'output_dir'};
+    my $db_ext = &util::is_little_endian() ? ".ldb" : ".bdb";
+
+    my $doc_db = &util::filename_cat($output_dir,"archiveinf-doc$db_ext");
+    my $src_db = &util::filename_cat($output_dir,"archiveinf-src$db_ext");
+
+    my $doc_db_text = "";
+    $doc_db_text .= "<src-file>$oid_files->{'src-file'}\n";
+    foreach my $af (@{$oid_files->{'assoc-files'}}) {
+    $doc_db_text .= "<assoc-file>$af\n";
+    }
+    chomp($doc_db_text); # remove trailing \n
+
+    &GDBMUtils::gdbmDatabaseSet($doc_db,$oid,$doc_db_text);
+
+    foreach my $rl (@reverse_lookups) {
+    &GDBMUtils::gdbmDatabaseAppend($src_db,$rl,"<oid>$oid\n");
+    }
+    
+}
+
+
 sub set_sortmeta {
     my $self = shift (@_);