begin split of full-text into own module

This will enable usage of other search engines

diff --git a/bin/BackupPC_updatedb b/bin/BackupPC_updatedb
index 9315026..69725af 100755 (executable)
--- a/bin/BackupPC_updatedb
+++ b/bin/BackupPC_updatedb
@@ -15,6 +15,8 @@ use BackupPC::SearchLib;
 use Cwd qw/abs_path/;
 use Data::Dump qw(dump);
 
+use BackupPC::Search::Estraier;
+
 use constant BPC_FTYPE_DIR => 5;
 use constant EST_CHUNK => 4096;
 
@@ -106,37 +108,20 @@ sub curr_time {
        return strftime($t_fmt,localtime());
 }
 
-my $hest_node;
-
 sub hest_update {
 
        my ($host_id, $share_id, $num) = @_;
 
        my $skip_check = $opt{j} && print STDERR "Skipping check for existing files -- this should be used only with initital import\n";
 
-       unless ($index_node_url && $index_node_url =~ m#^http://#) {
-               print STDERR "HyperEstraier support not enabled or index node invalid\n" if ($debug);
-               $index_node_url = 0;
-               return;
-       }
-
-       print curr_time," updating Hyper Estraier:";
+       print curr_time," updating fulltext:";
 
        my $t = time();
 
        my $offset = 0;
        my $added = 0;
 
-       if ($index_node_url) {
-               print " opening index $index_node_url";
-               $hest_node ||= Search::Estraier::Node->new(
-                       url => $index_node_url,
-                       user => 'admin',
-                       passwd => 'admin',
-                       croak_on_error => 1,
-               );
-               print " via node URL";
-       }
+       my $search = BackupPC::Search::Estraier->new( $index_node_url );
 
        my $results = 0;
 
@@ -196,37 +181,8 @@ sub hest_update {
                }
 
                while (my $row = $sth->fetchrow_hashref()) {
-
-                       my $uri = $row->{hname} . ':' . $row->{sname} . '#' . $row->{backupnum} . ' ' . $row->{filepath};
-                       if (! $skip_check && $hest_node) {
-                               my $id = $hest_node->uri_to_id($uri);
-                               next if ($id && $id == -1);
-                       }
-
-                       # create a document object 
-                       my $doc = Search::Estraier::Document->new;
-
-                       # add attributes to the document object 
-                       $doc->add_attr('@uri', $uri);
-
-                       foreach my $c (@{ $sth->{NAME} }) {
-                               print STDERR "attr $c = $row->{$c}\n" if ($debug > 2);
-                               $doc->add_attr($c, $row->{$c}) if (defined($row->{$c}));
-                       }
-
-                       #$doc->add_attr('@cdate', fmt_date($row->{'date'}));
-
-                       # add the body text to the document object 
-                       my $path = $row->{'filepath'};
-                       $doc->add_text($path);
-                       $path =~ s/(.)/$1 /g;
-                       $doc->add_hidden_text($path);
-
-                       print STDERR $doc->dump_draft,"\n" if ($debug > 1);
-
-                       # register the document object to the database
-                       $hest_node->put_doc($doc) if ($hest_node);
-
+                       next if $search->exists( $row );
+                       $search->add_doc( $row );
                        $added++;
                }
 
@@ -541,7 +497,6 @@ foreach my $host_key (@hosts) {
        $host_nr++;
        # get backups for a host
        my @backups = $bpc->BackupInfoRead($hostname);
-warn "XXXX ",dump(@backups);
        my $incs = scalar @backups;
 
        my $host_header = sprintf("host %s [%d/%d]: %d increments\n",

diff --git a/lib/BackupPC/Search/Estraier.pm b/lib/BackupPC/Search/Estraier.pm
new file mode 100644 (file)
index 0000000..ff2b287
--- /dev/null
+++ b/lib/BackupPC/Search/Estraier.pm
@@ -0,0 +1,70 @@
+package BackupPC::Search::Estraier;
+use warnings;
+use strict;
+
+use Search::Estraier;
+
+my $debug = $ENV{DEBUG} || 0;
+
+sub new {
+       my ( $class, $index_node_url ) = @_;
+
+       warn "# using $index_node_url";
+
+       my $self = bless {
+               node => Search::Estraier::Node->new(
+                       url => $index_node_url,
+                       user => 'admin',
+                       passwd => 'admin',
+                       croak_on_error => 1,
+               ),
+       }, $class;
+       return $self;
+}
+
+sub node { $_[0]->{node} };
+
+sub exists {
+       my ( $self, $row ) = @_;
+
+       my $uri = $row->{hname} . ':' . $row->{sname} . '#' . $row->{backupnum} . ' ' . $row->{filepath};
+       my $id = $self->node->uri_to_id($uri);
+       return $id && $id != -1;
+}
+
+sub add_doc {
+       my ( $self, $row ) = @_;
+
+       # create a document object 
+       my $doc = Search::Estraier::Document->new;
+
+       my $uri = $row->{hname} . ':' . $row->{sname} . '#' . $row->{backupnum} . ' ' . $row->{filepath};
+       # add attributes to the document object 
+       $doc->add_attr('@uri', $uri);
+
+       foreach my $c (keys %$row) {
+               print STDERR "attr $c = $row->{$c}\n" if ($debug > 2);
+               $doc->add_attr($c, $row->{$c}) if defined($row->{$c});
+       }
+
+       #$doc->add_attr('@cdate', fmt_date($row->{'date'}));
+
+       # add the body text to the document object 
+       my $path = $row->{'filepath'};
+       $doc->add_text($path);
+       $path =~ s/(.)/$1 /g;
+       $doc->add_hidden_text($path);
+
+       print STDERR $doc->dump_draft,"\n" if ($debug > 1);
+
+       # register the document object to the database
+       $self->node->put_doc($doc);
+
+}
+
+sub commit {
+       my $self = shift;
+       warn "# commit not needed";
+}
+
+1;