--- Webpacus/lib/Webpacus/Model/WebPAC.pm 2005/11/24 22:29:29 135
+++ Webpacus/lib/Webpacus/Model/WebPAC.pm 2006/02/19 20:16:11 404
@@ -6,11 +6,12 @@
use base qw/
Catalyst::Model
/;
-use Data::Dumper;
-use WebPAC::DB;
-use WebPAC::Output::TT;
-use WebPAC::Search::Estraier 0.02;
+use WebPAC::Store 0.08;
+use Search::Estraier 0.04;
use File::Slurp;
+use Time::HiRes qw/time/;
+use Encode qw/encode decode from_to/;
+use Template;
=head1 NAME
@@ -33,9 +34,13 @@
# configuration for hyper estraier full text search engine
hyperestraier:
- url: 'http://localhost:1978/node/webpac2'
+ masterurl: 'http://localhost:1978/node/webpac2'
+ defaultnode: 'webpac2'
+ defaultdepth: 1
user: 'admin'
passwd: 'admin'
+ hits_on_page: 100
+ hits_for_pager: 1000
webpac:
db_path: '/data/webpac2/db'
@@ -43,8 +48,6 @@
template: 'html_ffzg_results_short.tt'
# encoding comming from webpac
webpac_encoding: 'iso-8859-2'
- # encoding expected by Catalyst
- out_encoding: 'UTF-8'
=cut
@@ -60,122 +63,391 @@
my $est_cfg = $c->config->{hyperestraier};
$est_cfg->{'log'} = $log;
- $log->debug("using config:" . Dumper($est_cfg) );
+ $est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
- $self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );
+ $log->dumper($est_cfg, 'est_cfg');
- my $db_path = $c->config->{webpac}->{db_path};
- my $template_path = $c->config->{webpac}->{template_path};
- $self->{template_path} = $template_path;
+ if (! $est_cfg->{database}) {
+ my $defaultnode = $est_cfg->{defaultnode} || $log->logdie("can't find defaultnode in estraier configuration");
+ $log->info("using default node $defaultnode");
+ $est_cfg->{database} = $defaultnode;
+ }
+
+ my $url = $est_cfg->{masterurl} . '/node/' . $est_cfg->{database};
+
+ $log->info("opening Hyper Estraier index $url as $est_cfg->{'user'}");
+
+ $self->{est_node} = Search::Estraier::Node->new(
+ url => $url,
+ user => $est_cfg->{user},
+ passwd => $est_cfg->{passwd},
+ );
+
+ $log->fatal("can't create Search::Estraier::Node $url") unless ($self->{est_node});
+
+ # save config parametars in object
+ foreach my $f (qw/
+ db_path template_path hits_on_page webpac_encoding defaultdepth
+ masterurl defaultnode
+ /) {
+ $self->{$f} = $c->config->{hyperestraier}->{$f} ||
+ $c->config->{webpac}->{$f};
+ $log->debug("self->{$f} = " . $self->{$f});
+ }
+ my $db_path = $self->{db_path};
+ my $template_path = $self->{template_path};
$log->debug("using db path '$db_path', template path '$template_path'");
- $self->{db} = new WebPAC::DB(
+ $self->{db} = new WebPAC::Store(
path => $db_path,
read_only => 1,
- );
-
- $self->{out} = new WebPAC::Output::TT(
- include_path => $template_path,
- filters => { foo => sub { shift } },
+ database => $est_cfg->{database},
);
# default template from config.yaml
$self->{template} ||= $c->config->{webpac}->{template};
- $self->{iconv} = new Text::Iconv(
- $c->config->{webpac}->{webpac_encoding},
- $c->config->{webpac}->{out_encoding}
- );
-
$log->debug("converting encoding from webpac_encoding '" .
$c->config->{webpac}->{webpac_encoding} .
- "' to '" .
- $c->config->{webpac}->{out_encoding} .
"'"
);
+ $self->{databases} = $c->config->{databases} || $log->fatal("can't find databases in config");
+
+ # create Template toolkit instance
+ $self->{'tt'} = Template->new(
+ INCLUDE_PATH => $template_path,
+ FILTERS => {
+ dump_html => sub {
+ return unless (@_);
+ my $out;
+ my $i = 1;
+ foreach my $v (@_) {
+ $out .= qq{
} .
+ Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) .
+ qq{
};
+ $i++;
+ }
+ $out =~ s!/]*>!!gis if ($out);
+ return $out;
+ }
+ },
+ EVAL_PERL => 1,
+ );
+
return $self;
}
-=head2 iconv_on_save
+=head2 setup_site
- my $out = $m->iconv_on_save( $content );
+ $self->setup_site('site_name');
-Convert data saved to disk in Webpac encoding.
+Change node URL and database name according to site name (if available) or fallback
+to C from configuration.
=cut
-sub iconv_on_save {
+sub setup_site {
my $self = shift;
- $self->{iconv_save} ||= new Text::Iconv(
- $self->config->{webpac}->{out_encoding},
- $self->config->{webpac}->{webpac_encoding},
- );
+ my $site = shift || $self->{defaultnode};
- $self->{iconv_save}->convert( @_ );
-}
+ $self->{log}->fatal("setup_site can't find site or defaultnode") unless ($site);
+ my $url = $self->{masterurl} . '/node/' . $site;
+ $self->{est_node}->set_url( $url );
+ $self->{log}->debug("setup_site $site using $url");
+}
=head2 search
- my $m->search( 'query phrase', 'result_template.tt', \@add_attr );
+ my $m->search(
+ phrase => 'query phrase',
+ add_attr => \@add_attr
+ get_attr => [ '@uri' ],
+ max => 42,
+ template => 'result_template.tt',
+ depth => 1,
+ );
+
+All fields are standard C parametars except
+C which will (if specified) return results in HTML using
+selected template.
=cut
sub search {
- my ( $self, $query, $template, $add_attr ) = @_;
+ my $self = shift;
+
+ my $search_start_t = time();
+
+ my $args = {@_};
my $log = $self->{log};
- $log->debug("search model query: '$query', add_attr: '" . join("','", @{$add_attr}) . "'");
+ $log->dumper($args, 'args');
+
+ my $query = $args->{phrase} || $log->warn("no query phrase") && return;
+
+ my $template_filename = $args->{template} || $self->{template};
+
+ $args->{max} ||= $self->{'hits_for_pager'};
+ if (! $args->{max}) {
+ $args->{max} = 100;
+ $log->warn("max not set when calling model. Using default of $args->{max}");
+ }
+
+ my $times; # store some times for benchmarking
+
+ my $t = time();
+
+ # transfer depth of search
+ if (! $args->{depth}) {
+ my $default = $self->{defaultdepth} || $log->logdie("can't find defaultdepth in estraier configuration");
+ $args->{depth} = $default;
+ $log->warn("using default search depth $default");
+ }
+ $args->{depth} ||= 0;
+
+ $log->debug("searching for maximum $args->{max} results using depth $args->{depth} phrase: ", $query || '[none]');
+
+ #
+ # construct condition for Hyper Estraier
+ #
+ my $cond = Search::Estraier::Condition->new();
+ if ( ref($args->{add_attr}) eq 'ARRAY' ) {
+ $log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) );
+ map {
+ $cond->add_attr( $_ );
+ $log->debug(" + $_");
+ } @{ $args->{add_attr} };
+ };
+
+ $cond->set_phrase( $query ) if ($query);
+ $cond->set_options( $args->{options} ) if ($args->{options});
+ $cond->set_order( $args->{order} ) if ($args->{order});
+
+ my $max = $args->{max} || 7;
+ my $page = $args->{page} || 1;
+ if ($page < 1) {
+ $log->warn("page number $page < 1");
+ $page = 1;
+ }
+
+ $cond->set_max( $page * $max );
+
+ my $result = $self->{est_node}->search($cond, $args->{depth});
+ my $hits = $result->doc_num;
+
+ $times->{est} += time() - $t;
+
+ $log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );
+
+ $log->dumper($result->{hints}, 'result->hints' );
+
+ #
+ # fetch results
+ #
+
+ my @results;
+
+ for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
- my $template_filename = $template || $self->{template};
+ $t = time();
+
+ #$log->debug("get_doc($i)");
+ my $doc = $result->get_doc( $i );
+ if (! $doc) {
+ $log->warn("can't find result $i");
+ next;
+ }
+
+ my $hash;
+
+ foreach my $attr (@{ $args->{get_attr} }) {
+ my $val = $doc->attr( $attr );
+ #$log->debug("attr $attr = ", $val || 'undef');
+ $hash->{$attr} = $val if (defined($val));
+ }
+
+ $times->{hash} += time() - $t;
+
+ next unless ($hash);
+
+ if (! $args->{'template'}) {
+ push @results, $hash;
+ } else {
+ my ($database, $prefix, $id);
+
+ if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
+ ($database, $prefix,$id) = ($1,$2,$3);
+ } else {
+ $log->warn("can't decode database/prefix/id from " . $hash->{'@uri'});
+ next;
+ }
+
+ #$log->debug("load_ds( id => $id, prefix => '$prefix' )");
+
+ $t = time();
+
+ my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
+ if (! $ds) {
+ $log->error("can't load_ds( ${database}/${prefix}/${id} )");
+ next;
+ }
+
+ $times->{db} += time() - $t;
+
+ $t = time();
+
+ my $html = $self->apply(
+ template => $template_filename,
+ data => $ds,
+ record_uri => "${database}/${prefix}/${id}",
+ config => $self->{databases}->{$database},
+ );
+
+ $times->{apply} += time() - $t;
+
+ $t = time();
+
+ $html = decode($self->{webpac_encoding}, $html);
+
+ $times->{decode} += time() - $t;
+
+ push @results, $html;
+ }
+
+ }
+
+ $log->debug( sprintf(
+ "duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs",
+ $times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t,
+ ) );
+
+ return \@results;
+}
+
+=head2 record
+
+ my $html = $m->record(
+ mfn => 42,
+ template => 'foo.tt',
+ );
+
+This will load one record, convert it to html using C and return
+it.
+
+=cut
+
+sub record {
+ my $self = shift;
+
+ my $args = {@_};
+ my $log = $self->{log};
+ $log->dumper( $args, 'args' );
+
+ foreach my $f (qw/record_uri template/) {
+ $log->fatal("need $f") unless ($args->{$f});
+ }
+
+ my ($database, $prefix, $id);
+
+ if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
+ ($database, $prefix, $id) = ($1,$2,$3);
+ } else {
+ $log->error("can't parse $args->{record_uri} into prefix, database and uri");
+ return;
+ }
+
+ my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
+ if (! $ds) {
+ $log->error("can't load_ds( $database/$prefix/$id )");
+ return;
+ }
- my @results = $self->{est}->search(
- phrase => $query,
- get_attr => [ '@uri' ],
- max => 100,
- add_attr => $add_attr,
+ my $html = $self->apply(
+ template => $args->{template},
+ data => $ds,
+ record_uri => $args->{record_uri},
+ config => $self->{databases}->{$database},
);
- $log->debug("loading " . ($#results + 1) . " results");
+ $html = decode($self->{webpac_encoding}, $html);
- my @html_results;
+ return $html;
+}
+
+
+=head2 list_nodes
+
+ my @nodes = $m->list_nodes( 'site' );
+
+Return all databases which have records for selected site. Returned array of
+hashes has elements C and C