r1804@llin: dpavlin | 2009-04-24 18:51:47 +0200

 port webpac2.cgi to use swish-e as indexing engine
 added ProperyName data which contains JSON encoded data

git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1149 07558da8-63fa-0310-ba24-9fe276d99e06

diff --git a/lib/WebPAC/Output/SWISH.pm b/lib/WebPAC/Output/SWISH.pm
index 60ce87f..ab8fc22 100644 (file)
--- a/lib/WebPAC/Output/SWISH.pm
+++ b/lib/WebPAC/Output/SWISH.pm
@@ -16,6 +16,7 @@ __PACKAGE__->mk_accessors(qw(
 use File::Path qw/make_path/;
 use Data::Dump qw/dump/;
 use YAML;
+use JSON;
 
 
 =head1 NAME
@@ -89,6 +90,8 @@ IndexFile $dir/$database
 # Croatian ISO-8859-2 characters to unaccented equivalents
 #TranslateCharacters ¹©ðÐèÈæƾ® ssddcccczz
 
+# store data into index
+PropertyNames data
 
 # disable output
 ParserWarnLevel 0
@@ -146,6 +149,8 @@ sub add {
 
        my $xml = qq{<xml>};
 
+       my $data;
+
        foreach my $tag (@tags) {
 
                my $r = ref $ds->{$tag}->{$type};
@@ -159,8 +164,14 @@ sub add {
                $xml .= qq{<$tag><![CDATA[$vals]]></$tag>};
 
                $self->{stats}->{attr}->{$tag}++;
+
+               $data->{$tag} = $vals;
        }
 
+       # serialize to JSON instead of YAML because we will loose whitespace
+       $data = to_json($data);
+       $xml .= qq{<data><![CDATA[$data]]></data>};
+
        $xml .= qq{</xml>\n};
 
        my $len = length($xml);
@@ -170,6 +181,8 @@ sub add {
        print $fh "Path-Name: $uri\nContent-Length: $len\nDocument-Type: XML\n\n$xml" or
                die "can't add $uri: $@\n$xml";
 
+       warn "$xml\n";
+
        return 1;
 }
 

diff --git a/vhost/ecas/config.yml b/vhost/ecas/config.yml
index e45de32..e71d242 100644 (file)
--- a/vhost/ecas/config.yml
+++ b/vhost/ecas/config.yml
@@ -54,4 +54,7 @@ databases:
         path: '/data/FF/ecasopisi/blackwell/blackwell2.mrc'
         normalize:
           path: 'vhost/ecas/blackwell.pl'
+    output:
+      module: 'WebPAC::Output::SWISH'
+      database: 'ecas'
 

diff --git a/vhost/webpac2.cgi b/vhost/webpac2.cgi
index 1f77b8f..02a4b3c 100755 (executable)
--- a/vhost/webpac2.cgi
+++ b/vhost/webpac2.cgi
@@ -7,9 +7,10 @@ use CGI qw/:standard/;
 use CGI::Carp qw/fatalsToBrowser/;
 use File::Slurp;
 use YAML;
-use Search::Estraier;
 use Data::Page;
 use Data::Dump qw/dump/;
+use SWISH::API;
+use JSON;
 
 my $range_around = 5;
 my $entries_per_page = 30;
@@ -143,10 +144,8 @@ if ( my $search = param('search') ) {
 
        print qq|<div id="results">|;
 
-       my $node = Search::Estraier::Node->new(
-               url => $config->{hyperestraier}->{masterurl} . '/node/' . $database,
-               croak_on_error => 1,
-       );
+       my $swish = SWISH::API->new( "$dir/../var/swish/$database" );
+       $swish->abort_last_error if $swish->Error;
 
        param( 'entries_per_page', $entries_per_page ) unless param('entries_per_page'); # FIXME not needed?
        my $pager = Data::Page->new;
@@ -155,33 +154,32 @@ if ( my $search = param('search') ) {
 
        dump_yaml( 'pager', $pager );
 
-       my $cond = Search::Estraier::Condition->new( debug => $debug );
-       $cond->set_phrase( $search );
-       $cond->set_skip( $pager->skipped );
-       $cond->set_max(  $pager->entries_per_page );
+       my @search = ( "xml=$search" );
 
-       if ( my $op = param('attr_operator') ) {
-               $cond->add_attr( param('attr') . " $op " . param('search') );
-       }
+#      if ( my $op = param('attr_operator') ) {
+#              $cond->add_attr( param('attr') . " $op " . param('search') );
+#      }
 
-       foreach my $i ( param('only_input') ) {
-               $cond->add_attr( '@uri STRBW file:///search/' . $database . '/' . $i );
-       }
+       my @only_input = param('only_input');
 
+       push @search, '(' . join(') or (', @only_input) . ')' if @only_input;
 
-       my $nres = $node->search( $cond, 0 );
-       $pager->total_entries( $nres->hits );
+       my $q = '(' . join(') and (', @search) . ')';
+       warn "# query: $q\n";
+       my $swish_results = $swish->query( $q );
+       $swish->abort_last_error if $swish->Error;
 
-       dump_yaml( 'cond', $cond );
-       dump_yaml( 'nres', $nres );
+       dump_yaml( 'swish_results', $swish_results );
 
-       if ( ! $nres ) {
+       $pager->total_entries( $swish_results->hits );
+
+       if ( ! $pager->total_entries ) {
                my $no_results = "No results for search '%s'";
                printf qq|<div class="error">$no_results</div>\n\n|, $search;
        } else {
 
                my $results = "%d results for search '%s' showing results %d - %d on page %d";
-               printf qq|<div class="message">$results</div>\n\n|, $nres->hits, $search, $pager->first, $pager->last, $pager->current_page;
+               printf qq|<div class="message">$results</div>\n\n|, $pager->total_entries, $search, $pager->first, $pager->last, $pager->current_page;
 
                my $pager_html = join("\n", show_pager( $pager ));
 
@@ -190,19 +188,20 @@ if ( my $search = param('search') ) {
                my $start = $pager->first;
                print qq|<ol start=$start>\n|;
 
-               foreach my $i ( 1 .. $nres->doc_num ) {
-                       my $rdoc = $nres->get_doc( $i - 1 );
+               while ( my $result = $swish_results->next_result ) {
+                       my $data = from_json $result->property('data');
+
                        print qq|<li>|;
                        foreach my $attr ( @attr ) {
-                               my $v = $rdoc->attr( $attr );
-                               if ( defined $v && $html_markup && ! $html_markup_skip->{$attr} ) {
+                               next unless defined $data->{$attr};
+                               my $v = $data->{$attr};
+                               if ( $html_markup && ! $html_markup_skip->{$attr} ) {
                                        eval "\$v = $html_markup->$attr( \$v );";
                                        if ( $@ ) {
                                                warn "disable html markup for $attr: $@";
                                                $html_markup_skip->{$attr} = $@;
                                        }
                                }
-                               next unless defined $v;
                                print qq|<div><label>$attr</label><span class=$attr>$v</span></div>\n|;
                        }
                        print qq|</li>\n|;
@@ -217,7 +216,6 @@ if ( my $search = param('search') ) {
 
 }
 
-dump_yaml( 'estraier', $estraier );
 dump_yaml( 'db', $db );
 dump_yaml( 'html_markup_skip', $html_markup_skip );