Webpacus/Model/WebPAC.pm

package Webpacus::Model::WebPAC;

use strict;
use warnings;
use lib '/data/webpac2/lib';
use base qw/
        Catalyst::Model
/;
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;
use Data::Dumper;

=head1 NAME

Webpacus::Model::WebPAC - Catalyst Model

=head1 SYNOPSIS

See L<Webpacus> and L<WebPAC>.

=head1 DESCRIPTION

Catalyst Model for access to WebPAC data.

=head2 new

Configuration for hyperestraier in C<config.yaml> like this:

 --- #YAML:1.0
 # DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!

 # configuration for hyper estraier full text search engine
 hyperestraier:
  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'
  template_path: '/data/webpac2/conf/output/tt'
  template: 'html_ffzg_results_short.tt'
  # encoding comming from webpac
  webpac_encoding: 'iso-8859-2'

=cut

sub new {
        my ( $self, $c, $config ) = @_;

        $self = $self->NEXT::new($c, $config);
        $self->config($config);

        my $log = $c->log;
        $self->{log} = $log;

        my $est_cfg = $c->config->{hyperestraier};
        $est_cfg->{'log'} = $log;

        $est_cfg->{encoding} = $est_cfg->{catalyst_encoding} || $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");

        $log->debug("using config:" . Dumper($est_cfg) );

        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
                /) {
                $self->{$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::Store(
                path => $db_path,
                read_only => 1,
                database => $est_cfg->{database},
        );

        # default template from config.yaml
        $self->{template} ||= $c->config->{webpac}->{template};

        $log->debug("converting encoding from webpac_encoding '" .
                $c->config->{webpac}->{webpac_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 (@_) {
                                        $out .= qq{<div id="dump_$i">} .
                                                Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) .
                                                qq{</div>};
                                        $i++;
                                }
                                $out =~ s!<table[^>/]*>!<table class="dump">!gis if ($out);
                                return $out;
                        }
                },
                EVAL_PERL => 1,
        );

        return $self;

}

=head2 setup_site

 $self->setup_site('site_name');

Change node URL and database name according to site name (if available)

=cut

sub setup_site {
        my $self = shift;

        my $site = shift || return;

        my $url = $self->{masterurl} . '/node/' . $site;
        $self->{est_node}->set_url( $url );
        $self->{log}->debug("setup_site $site");
}

=head2 search

  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<WebPAC::Search::Estraier> parametars except
C<template> which will (if specified) return results in HTML using
selected template.

=cut

sub search {
        my $self = shift;

        my $search_start_t = time();

        my $args = {@_};

        my $log = $self->{log};

        $log->debug("search args: " . Dumper( $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}) );

        $log->debug( "hints: " . Dumper($result->{hints}) );

        #
        # fetch results
        #

        my @results;

        for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {

                $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;

                        #$log->debug( "ds = " . Dumper( \@html_results ) );

                        $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( '@results = ' . Dumper( \@results ) );

        $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<template> and return
it.

=cut

sub record {
        my $self = shift;

        my $args = {@_};
        my $log = $self->{log};
        $log->debug("record args: " . Dumper( $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 $html = $self->apply(
                template => $args->{template},
                data => $ds,
                record_uri => $args->{record_uri},
                config => $self->{databases}->{$database},
        );

        $html = decode($self->{webpac_encoding}, $html);

        return $html;
}


=head2 save_html

  $m->save_html( '/full/path/to/file', $content );

It will use C<Encode> to convert content encoding back to
Webpac codepage, recode JavaScript Unicode entities (%u1234),
strip extra newlines at beginning and end, and save to 
C</full/path/to/file.new> and if that succeeds, just rename
it over original file which should be atomic on filesystem level.

=cut

sub save_html {
        my ($self, $path, $content) = @_;

        # FIXME Should this be UTF-8 or someting?
        my $js_encoding = $self->{webpac_encoding};
        $js_encoding = 'UTF-16';

        sub _conv_js {
                return '0x' . $_[1];
                return encode($_[0], chr(hex($_[1])));
        }
        #$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
        $content =~ s/^[\n\r]+//s;
        $content =~ s/[\n\r]+$/\n/s;
        $content =~ s/\n\r/\n/gs;

        my $disk_encoding = $self->{webpac_encoding} || 'utf-8';
        $self->{log}->debug("convert encoding to $disk_encoding");
        from_to($content, 'utf-8', $disk_encoding) || $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");

        write_file($path . '.new', {binmode => ':raw' }, $content) || die "can't save ${path}.new $!";
        rename $path . '.new', $path || die "can't rename to $path: $!";
}

=head2 load_html

  my $html = $m->load_html('/full/path/to/file');

This will convert file from Webpac encoding to Catalyst and
convert that data to escaped HTML (for sending into
C<< <textarea/> >> tags in html.

=cut

sub load_html {
        my ($self, $path) = @_;

        die "no path?" unless ($path);

        my $content = read_file($path, {binmode => ':raw' }) || die "can't read $path: $!";

        return decode($self->{webpac_encoding}, $content);
}


=head2 apply

Create output from in-memory data structure using Template Toolkit template.

 my $text = $tt->apply(
        template => 'text.tt',
        data => $ds,
        record_uri => 'database/prefix/mfn',
 );

It also has follwing template toolikit filter routies defined:

=cut

sub apply {
        my $self = shift;

        my $args = {@_};

        my $log = $self->{log} || die "no log?";

        foreach my $a (qw/template data/) {
                $log->fatal("need $a") unless ($args->{$a});
        }

=head3 tt_filter_type

filter to return values of specified from $ds, usage from TT template is in form
C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:

  [% d('Title') %]
  [% d('Author',', ' %]

=cut

        sub tt_filter_type {
                my ($data,$type) = @_;
                
                die "no data?" unless ($data);
                $type ||= 'display';

                my $default_delimiter = {
                        'display' => '&#182;<br/>',
                        'index' => '\n',
                };

                return sub {

                        my ($name,$join) = @_;

                        die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
                        # Hm? Should we die here?
                        return unless ($name);

                        my $item = $data->{'data'}->{$name} || return;

                        my $v = $item->{$type} || return;

                        if (ref($v) eq 'ARRAY') {
                                if ($#{$v} == 0) {
                                        $v = $v->[0];
                                } else {
                                        $join = $default_delimiter->{$type} unless defined($join);
                                        $v = join($join, @{$v});
                                }
                        } else {
                                warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
                        }

                        return $v;
                }
        }

        $args->{'d'} = tt_filter_type($args, 'display');
        $args->{'display'} = tt_filter_type($args, 'display');

=head3 tt_filter_search

filter to return links to search, usage in TT:

  [% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]

=cut

        sub tt_filter_search {

                my ($data) = @_;

                die "no data?" unless ($data);
                
                return sub {

                        my ($display,$search,$delimiter,$template) = @_;
                        
                        # default delimiter
                        $delimiter ||= '&#182;<br/>',

                        die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
                        # Hm? Should we die here?
                        return unless ($display);

                        my $item = $data->{'data'}->{$display} || return;

                        return unless($item->{'display'});
                        if (! $item->{'search'}) {
                                warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
                                return;
                        }

                        my @warn;
                        foreach my $type (qw/display search/) {
                                push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
                        }

                        if (@warn) {
                                warn("TT filter search(): " . join(",", @warn) . ", skipping");
                                return;
                        }
                        my @html;

                        my $d_el = $#{ $item->{'display'} };
                        my $s_el = $#{ $item->{'search'} }; 

                        # easy, both fields have same number of elements or there is just
                        # one search and multiple display
                        if ( $d_el == $s_el || $s_el == 0 ) {

                                foreach my $i ( 0 .. $d_el ) {

                                        my $s;
                                        if ($s_el > 0) {
                                                $s = $item->{'search'}->[$i] or warn "can't find value $i for type search in field $search";
                                        } else {
                                                $s = $item->{'search'}->[0];
                                        }
                                        #$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
                                        $s = __quotemeta( $s );

                                        my $d = $item->{'display'}->[$i] or warn "can't find value $i for type display in field $display";

                                        my $template_arg = '';
                                        $template_arg = qq{,'$template'} if ($template);

                                        push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
                                }

                                return join($delimiter, @html);
                        } else {
                                my $html = qq{<div class="notice">WARNING: we should really support if there is $d_el display elements and $s_el search elements, but currently there is no nice way to do so, so we will just display values</div>};
                                my $v = $item->{'display'};

                                if ($#{$v} == 0) {
                                        $html .= $v->[0];
                                } else {
                                        $html .= join($delimiter, @{$v});
                                }
                                return $html;
                        }
                }
        }

        $args->{'search'} = tt_filter_search($args);

=head3 load_rec

Used mostly for onClick events like this:

  <a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>

It will automatically do sanity checking and create correct JavaScript code.

=cut

        $args->{'load_rec'} = sub {
                my @errors;

                my $record_uri = shift or push @errors, "record_uri missing";
                my $template = shift or push @errors, "template missing";

                if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
                        push @errors, "invalid format of record_uri: $record_uri";
                }

                if (@errors) {
                        return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
                } else {
                        return "load_rec('$record_uri','$template'); return false;";
                }
        };

=head3 load_template

Used to re-submit search request and load results in different template

  <a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>

=cut

        $args->{'load_template'} = sub {
                my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
                return "load_template($template); return false;";
        };

        my $out;

        $self->{'tt'}->process(
                $args->{'template'},
                $args,
                \$out
        ) || $log->error( "apply can't process template: ", $self->{'tt'}->error() );

        return $out;
}


=head2 __quotemeta

Helper to quote JavaScript-friendly characters

=cut

sub __quotemeta {
        local $_ = shift;
        $_ = decode('iso-8859-2', $_);

        s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
        {
                use bytes;  
                s<((?:[^ \x21-\x7E]|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
        }

        s/\\x09/\\t/g;
        s/\\x0A/\\n/g;
        s/\\x0D/\\r/g;
        s/"/\\"/g;
        s/\\x5C/\\\\/g;

        return $_;
}


=head1 AUTHOR

Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>

=head1 LICENSE

This library is free software, you can redistribute it and/or modify
it under the same terms as Perl itself.

=cut

1;
1	dpavlin	92	package Webpacus::Model::WebPAC;
2
3			use strict;
4			use warnings;
5			use lib '/data/webpac2/lib';
6	dpavlin	93	use base qw/
7			Catalyst::Model
8			/;
9	dpavlin	237	use WebPAC::Store 0.08;
10	dpavlin	382	use Search::Estraier 0.04;
11	dpavlin	135	use File::Slurp;
12	dpavlin	378	use Time::HiRes qw/time/;
13	dpavlin	348	use Encode qw/encode decode from_to/;
14	dpavlin	379	use Template;
15	dpavlin	378	use Data::Dumper;
16	dpavlin	92
17			=head1 NAME
18
19			Webpacus::Model::WebPAC - Catalyst Model
20
21			=head1 SYNOPSIS
22
23			See L<Webpacus> and L<WebPAC>.
24
25			=head1 DESCRIPTION
26
27			Catalyst Model for access to WebPAC data.
28
29			=head2 new
30
31			Configuration for hyperestraier in C<config.yaml> like this:
32
33			--- #YAML:1.0
34			# DO NOT USE TABS FOR INDENTATION OR label/value SEPARATION!!!
35
36			# configuration for hyper estraier full text search engine
37			hyperestraier:
38	dpavlin	222	masterurl: 'http://localhost:1978/node/webpac2'
39			defaultnode: 'webpac2'
40			defaultdepth: 1
41	dpavlin	96	user: 'admin'
42			passwd: 'admin'
43	dpavlin	143	hits_on_page: 100
44	dpavlin	305	hits_for_pager: 1000
45	dpavlin	92
46	dpavlin	96	webpac:
47			db_path: '/data/webpac2/db'
48			template_path: '/data/webpac2/conf/output/tt'
49			template: 'html_ffzg_results_short.tt'
50			# encoding comming from webpac
51			webpac_encoding: 'iso-8859-2'
52
53	dpavlin	92	=cut
54
55			sub new {
56			my ( $self, $c, $config ) = @_;
57
58			$self = $self->NEXT::new($c, $config);
59			$self->config($config);
60
61			my $log = $c->log;
62	dpavlin	94	$self->{log} = $log;
63	dpavlin	92
64	dpavlin	93	my $est_cfg = $c->config->{hyperestraier};
65			$est_cfg->{'log'} = $log;
66	dpavlin	92
67	dpavlin	271	$est_cfg->{encoding} = $est_cfg->{catalyst_encoding} \|\| $c->config->{catalyst_encoding} or $c->log->fatal("can't find catalyst_encoding");
68	dpavlin	142
69	dpavlin	93	$log->debug("using config:" . Dumper($est_cfg) );
70	dpavlin	92
71	dpavlin	222	if (! $est_cfg->{database}) {
72			my $defaultnode = $est_cfg->{defaultnode} \|\| $log->logdie("can't find defaultnode in estraier configuration");
73			$log->info("using default node $defaultnode");
74			$est_cfg->{database} = $defaultnode;
75			}
76
77	dpavlin	382	my $url = $est_cfg->{masterurl} . '/node/' . $est_cfg->{database};
78	dpavlin	92
79	dpavlin	382	$log->info("opening Hyper Estraier index $url as $est_cfg->{'user'}");
80
81			$self->{est_node} = Search::Estraier::Node->new(
82			url => $url,
83			user => $est_cfg->{user},
84			passwd => $est_cfg->{passwd},
85			);
86
87			$log->fatal("can't create Search::Estraier::Node $url") unless ($self->{est_node});
88
89	dpavlin	167	# save config parametars in object
90	dpavlin	399	foreach my $f (qw/
91			db_path template_path hits_on_page webpac_encoding defaultdepth
92			masterurl
93			/) {
94	dpavlin	167	$self->{$f} = $c->config->{hyperestraier}->{$f} \|\|
95			$c->config->{webpac}->{$f};
96			$log->debug("self->{$f} = " . $self->{$f});
97			}
98			my $db_path = $self->{db_path};
99			my $template_path = $self->{template_path};
100	dpavlin	95
101			$log->debug("using db path '$db_path', template path '$template_path'");
102
103	dpavlin	222	$self->{db} = new WebPAC::Store(
104	dpavlin	95	path => $db_path,
105			read_only => 1,
106	dpavlin	224	database => $est_cfg->{database},
107	dpavlin	95	);
108
109	dpavlin	101	# default template from config.yaml
110	dpavlin	95	$self->{template} \|\|= $c->config->{webpac}->{template};
111
112	dpavlin	100	$log->debug("converting encoding from webpac_encoding '" .
113			$c->config->{webpac}->{webpac_encoding} .
114	dpavlin	99	"'"
115			);
116	dpavlin	96
117	dpavlin	382	$self->{databases} = $c->config->{databases} \|\| $log->fatal("can't find databases in config");
118	dpavlin	155
119	dpavlin	379	# create Template toolkit instance
120			$self->{'tt'} = Template->new(
121			INCLUDE_PATH => $template_path,
122			FILTERS => {
123			dump_html => sub {
124			return unless (@_);
125			my $out;
126			my $i = 1;
127			foreach my $v (@_) {
128			$out .= qq{<div id="dump_$i">} .
129			Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) .
130			qq{</div>};
131			$i++;
132			}
133			$out =~ s!<table[^>/]*>!<table class="dump">!gis if ($out);
134			return $out;
135			}
136			},
137			EVAL_PERL => 1,
138			);
139	dpavlin	378
140	dpavlin	92	return $self;
141
142			}
143
144	dpavlin	399	=head2 setup_site
145	dpavlin	135
146	dpavlin	399	$self->setup_site('site_name');
147
148			Change node URL and database name according to site name (if available)
149
150			=cut
151
152			sub setup_site {
153			my $self = shift;
154
155			my $site = shift \|\| return;
156
157			my $url = $self->{masterurl} . '/node/' . $site;
158			$self->{est_node}->set_url( $url );
159			$self->{log}->debug("setup_site $site");
160			}
161
162	dpavlin	135	=head2 search
163
164	dpavlin	150	my $m->search(
165			phrase => 'query phrase',
166	dpavlin	155	add_attr => \@add_attr
167			get_attr => [ '@uri' ],
168			max => 42,
169	dpavlin	150	template => 'result_template.tt',
170	dpavlin	222	depth => 1,
171	dpavlin	150	);
172	dpavlin	135
173	dpavlin	155	All fields are standard C<WebPAC::Search::Estraier> parametars except
174			C<template> which will (if specified) return results in HTML using
175			selected template.
176
177	dpavlin	135	=cut
178
179	dpavlin	93	sub search {
180	dpavlin	150	my $self = shift;
181	dpavlin	93
182	dpavlin	378	my $search_start_t = time();
183
184	dpavlin	150	my $args = {@_};
185
186	dpavlin	95	my $log = $self->{log};
187	dpavlin	94
188	dpavlin	351	$log->debug("search args: " . Dumper( $args ));
189	dpavlin	95
190	dpavlin	150	my $query = $args->{phrase} \|\| $log->warn("no query phrase") && return;
191	dpavlin	95
192	dpavlin	150	my $template_filename = $args->{template} \|\| $self->{template};
193
194	dpavlin	305	$args->{max} \|\|= $self->{'hits_for_pager'};
195	dpavlin	155	if (! $args->{max}) {
196	dpavlin	305	$args->{max} = 100;
197			$log->warn("max not set when calling model. Using default of $args->{max}");
198	dpavlin	155	}
199	dpavlin	93
200	dpavlin	155	my $times; # store some times for benchmarking
201
202			my $t = time();
203
204	dpavlin	222	# transfer depth of search
205			if (! $args->{depth}) {
206			my $default = $self->{defaultdepth} \|\| $log->logdie("can't find defaultdepth in estraier configuration");
207			$args->{depth} = $default;
208			$log->warn("using default search depth $default");
209			}
210	dpavlin	383	$args->{depth} \|\|= 0;
211	dpavlin	222
212	dpavlin	383	$log->debug("searching for maximum $args->{max} results using depth $args->{depth} phrase: ", $query \|\| '[none]');
213	dpavlin	155
214	dpavlin	382	#
215			# construct condition for Hyper Estraier
216			#
217			my $cond = Search::Estraier::Condition->new();
218			if ( ref($args->{add_attr}) eq 'ARRAY' ) {
219			$log->debug("adding search attributes: " . join(", ", @{ $args->{add_attr} }) );
220			map {
221	dpavlin	383	$cond->add_attr( $_ );
222	dpavlin	382	$log->debug(" + $_");
223			} @{ $args->{add_attr} };
224			};
225
226			$cond->set_phrase( $query ) if ($query);
227			$cond->set_options( $args->{options} ) if ($args->{options});
228			$cond->set_order( $args->{order} ) if ($args->{order});
229
230			my $max = $args->{max} \|\| 7;
231			my $page = $args->{page} \|\| 1;
232			if ($page < 1) {
233			$log->warn("page number $page < 1");
234			$page = 1;
235			}
236
237			$cond->set_max( $page * $max );
238	dpavlin	99
239	dpavlin	383	my $result = $self->{est_node}->search($cond, $args->{depth});
240	dpavlin	382	my $hits = $result->doc_num;
241
242	dpavlin	384	$times->{est} += time() - $t;
243
244	dpavlin	380	$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );
245	dpavlin	150
246	dpavlin	385	$log->debug( "hints: " . Dumper($result->{hints}) );
247
248	dpavlin	155	#
249	dpavlin	382	# fetch results
250	dpavlin	155	#
251
252	dpavlin	382	my @results;
253	dpavlin	100
254	dpavlin	382	for my $i ( (($page - 1) * $max) .. ( $hits - 1 ) ) {
255	dpavlin	95
256	dpavlin	382	$t = time();
257
258			#$log->debug("get_doc($i)");
259			my $doc = $result->get_doc( $i );
260			if (! $doc) {
261			$log->warn("can't find result $i");
262	dpavlin	224	next;
263			}
264	dpavlin	95
265	dpavlin	382	my $hash;
266	dpavlin	95
267	dpavlin	382	foreach my $attr (@{ $args->{get_attr} }) {
268			my $val = $doc->attr( $attr );
269			#$log->debug("attr $attr = ", $val \|\| 'undef');
270			$hash->{$attr} = $val if (defined($val));
271	dpavlin	242	}
272	dpavlin	155
273	dpavlin	382	$times->{hash} += time() - $t;
274	dpavlin	155
275	dpavlin	382	next unless ($hash);
276	dpavlin	115
277	dpavlin	382	if (! $args->{'template'}) {
278			push @results, $hash;
279			} else {
280			my ($database, $prefix, $id);
281	dpavlin	155
282	dpavlin	382	if ( $hash->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
283			($database, $prefix,$id) = ($1,$2,$3);
284			} else {
285			$log->warn("can't decode database/prefix/id from " . $hash->{'@uri'});
286			next;
287			}
288	dpavlin	100
289	dpavlin	382	#$log->debug("load_ds( id => $id, prefix => '$prefix' )");
290	dpavlin	155
291	dpavlin	382	$t = time();
292	dpavlin	155
293	dpavlin	382	my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
294			if (! $ds) {
295			$log->error("can't load_ds( ${database}/${prefix}/${id} )");
296			next;
297			}
298	dpavlin	100
299	dpavlin	382	$times->{db} += time() - $t;
300	dpavlin	100
301	dpavlin	382	#$log->debug( "ds = " . Dumper( \@html_results ) );
302
303			$t = time();
304
305			my $html = $self->apply(
306			template => $template_filename,
307			data => $ds,
308			record_uri => "${database}/${prefix}/${id}",
309			config => $self->{databases}->{$database},
310			);
311
312			$times->{apply} += time() - $t;
313
314			$t = time();
315
316			$html = decode($self->{webpac_encoding}, $html);
317
318			$times->{decode} += time() - $t;
319
320			push @results, $html;
321			}
322
323	dpavlin	95	}
324
325	dpavlin	383	#$log->debug( '@results = ' . Dumper( \@results ) );
326	dpavlin	115
327	dpavlin	155	$log->debug( sprintf(
328	dpavlin	382	"duration breakdown: estraier %.6fs, hash %.6fs, store %.6fs, apply %.6fs, decode %.06f, total: %.6fs",
329			$times->{est}, $times->{hash}, $times->{db}, $times->{apply}, $times->{decode}, time() - $search_start_t,
330	dpavlin	155	) );
331
332	dpavlin	382	return \@results;
333	dpavlin	93	}
334
335	dpavlin	165	=head2 record
336
337			my $html = $m->record(
338			mfn => 42,
339			template => 'foo.tt',
340			);
341
342			This will load one record, convert it to html using C<template> and return
343			it.
344
345			=cut
346
347			sub record {
348			my $self = shift;
349
350			my $args = {@_};
351			my $log = $self->{log};
352	dpavlin	351	$log->debug("record args: " . Dumper( $args ));
353	dpavlin	165
354	dpavlin	242	foreach my $f (qw/record_uri template/) {
355	dpavlin	228	$log->fatal("need $f") unless ($args->{$f});
356	dpavlin	165	}
357
358	dpavlin	242	my ($database, $prefix, $id);
359	dpavlin	165
360	dpavlin	242	if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
361			($database, $prefix, $id) = ($1,$2,$3);
362			} else {
363			$log->error("can't parse $args->{record_uri} into prefix, database and uri");
364			return;
365			}
366	dpavlin	165
367	dpavlin	242	my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
368			if (! $ds) {
369			$log->error("can't load_ds( $database/$prefix/$id )");
370			return;
371			}
372
373	dpavlin	379	my $html = $self->apply(
374	dpavlin	165	template => $args->{template},
375			data => $ds,
376	dpavlin	242	record_uri => $args->{record_uri},
377	dpavlin	305	config => $self->{databases}->{$database},
378	dpavlin	165	);
379
380	dpavlin	348	$html = decode($self->{webpac_encoding}, $html);
381	dpavlin	165
382			return $html;
383			}
384
385	dpavlin	305
386	dpavlin	135	=head2 save_html
387	dpavlin	93
388	dpavlin	135	$m->save_html( '/full/path/to/file', $content );
389	dpavlin	93
390	dpavlin	348	It will use C<Encode> to convert content encoding back to
391	dpavlin	135	Webpac codepage, recode JavaScript Unicode entities (%u1234),
392			strip extra newlines at beginning and end, and save to
393			C</full/path/to/file.new> and if that succeeds, just rename
394			it over original file which should be atomic on filesystem level.
395
396			=cut
397
398			sub save_html {
399			my ($self, $path, $content) = @_;
400
401	dpavlin	348	# FIXME Should this be UTF-8 or someting?
402			my $js_encoding = $self->{webpac_encoding};
403			$js_encoding = 'UTF-16';
404
405	dpavlin	135	sub _conv_js {
406	dpavlin	348	return '0x' . $_[1];
407			return encode($_[0], chr(hex($_[1])));
408	dpavlin	135	}
409	dpavlin	348	#$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
410	dpavlin	135	$content =~ s/^[\n\r]+//s;
411			$content =~ s/[\n\r]+$/\n/s;
412	dpavlin	282	$content =~ s/\n\r/\n/gs;
413	dpavlin	135
414	dpavlin	348	my $disk_encoding = $self->{webpac_encoding} \|\| 'utf-8';
415			$self->{log}->debug("convert encoding to $disk_encoding");
416			from_to($content, 'utf-8', $disk_encoding) \|\| $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
417	dpavlin	179
418			write_file($path . '.new', {binmode => ':raw' }, $content) \|\| die "can't save ${path}.new $!";
419	dpavlin	135	rename $path . '.new', $path \|\| die "can't rename to $path: $!";
420			}
421
422			=head2 load_html
423
424			my $html = $m->load_html('/full/path/to/file');
425
426			This will convert file from Webpac encoding to Catalyst and
427			convert that data to escaped HTML (for sending into
428			C<< <textarea/> >> tags in html.
429
430			=cut
431
432			sub load_html {
433			my ($self, $path) = @_;
434
435			die "no path?" unless ($path);
436
437	dpavlin	179	my $content = read_file($path, {binmode => ':raw' }) \|\| die "can't read $path: $!";
438	dpavlin	135
439	dpavlin	348	return decode($self->{webpac_encoding}, $content);
440	dpavlin	135	}
441
442	dpavlin	379
443			=head2 apply
444
445			Create output from in-memory data structure using Template Toolkit template.
446
447			my $text = $tt->apply(
448			template => 'text.tt',
449			data => $ds,
450			record_uri => 'database/prefix/mfn',
451			);
452
453			It also has follwing template toolikit filter routies defined:
454
455			=cut
456
457			sub apply {
458			my $self = shift;
459
460			my $args = {@_};
461
462			my $log = $self->{log} \|\| die "no log?";
463
464			foreach my $a (qw/template data/) {
465	dpavlin	382	$log->fatal("need $a") unless ($args->{$a});
466	dpavlin	379	}
467
468			=head3 tt_filter_type
469
470			filter to return values of specified from $ds, usage from TT template is in form
471			C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:
472
473			[% d('Title') %]
474			[% d('Author',', ' %]
475
476			=cut
477
478			sub tt_filter_type {
479			my ($data,$type) = @_;
480
481			die "no data?" unless ($data);
482			$type \|\|= 'display';
483
484			my $default_delimiter = {
485			'display' => '¶<br/>',
486			'index' => '\n',
487			};
488
489			return sub {
490
491			my ($name,$join) = @_;
492
493			die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
494			# Hm? Should we die here?
495			return unless ($name);
496
497			my $item = $data->{'data'}->{$name} \|\| return;
498
499			my $v = $item->{$type} \|\| return;
500
501			if (ref($v) eq 'ARRAY') {
502			if ($#{$v} == 0) {
503			$v = $v->[0];
504			} else {
505			$join = $default_delimiter->{$type} unless defined($join);
506			$v = join($join, @{$v});
507			}
508			} else {
509			warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
510			}
511
512			return $v;
513			}
514			}
515
516			$args->{'d'} = tt_filter_type($args, 'display');
517			$args->{'display'} = tt_filter_type($args, 'display');
518
519			=head3 tt_filter_search
520
521			filter to return links to search, usage in TT:
522
523			[% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]
524
525			=cut
526
527			sub tt_filter_search {
528
529			my ($data) = @_;
530
531			die "no data?" unless ($data);
532
533			return sub {
534
535			my ($display,$search,$delimiter,$template) = @_;
536
537			# default delimiter
538			$delimiter \|\|= '¶<br/>',
539
540			die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
541			# Hm? Should we die here?
542			return unless ($display);
543
544			my $item = $data->{'data'}->{$display} \|\| return;
545
546			return unless($item->{'display'});
547			if (! $item->{'search'}) {
548			warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
549			return;
550			}
551
552			my @warn;
553			foreach my $type (qw/display search/) {
554			push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
555			}
556
557			if (@warn) {
558			warn("TT filter search(): " . join(",", @warn) . ", skipping");
559			return;
560			}
561			my @html;
562
563			my $d_el = $#{ $item->{'display'} };
564			my $s_el = $#{ $item->{'search'} };
565
566			# easy, both fields have same number of elements or there is just
567			# one search and multiple display
568			if ( $d_el == $s_el \|\| $s_el == 0 ) {
569
570			foreach my $i ( 0 .. $d_el ) {
571
572			my $s;
573			if ($s_el > 0) {
574	dpavlin	383	$s = $item->{'search'}->[$i] or warn "can't find value $i for type search in field $search";
575	dpavlin	379	} else {
576			$s = $item->{'search'}->[0];
577			}
578			#$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
579			$s = __quotemeta( $s );
580
581	dpavlin	383	my $d = $item->{'display'}->[$i] or warn "can't find value $i for type display in field $display";
582	dpavlin	379
583			my $template_arg = '';
584			$template_arg = qq{,'$template'} if ($template);
585
586			push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
587			}
588
589			return join($delimiter, @html);
590			} else {
591			my $html = qq{<div class="notice">WARNING: we should really support if there is $d_el display elements and $s_el search elements, but currently there is no nice way to do so, so we will just display values</div>};
592			my $v = $item->{'display'};
593
594			if ($#{$v} == 0) {
595			$html .= $v->[0];
596			} else {
597			$html .= join($delimiter, @{$v});
598			}
599			return $html;
600			}
601			}
602			}
603
604			$args->{'search'} = tt_filter_search($args);
605
606			=head3 load_rec
607
608			Used mostly for onClick events like this:
609
610			<a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>
611
612			It will automatically do sanity checking and create correct JavaScript code.
613
614			=cut
615
616			$args->{'load_rec'} = sub {
617			my @errors;
618
619			my $record_uri = shift or push @errors, "record_uri missing";
620			my $template = shift or push @errors, "template missing";
621
622			if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
623			push @errors, "invalid format of record_uri: $record_uri";
624			}
625
626			if (@errors) {
627			return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
628			} else {
629			return "load_rec('$record_uri','$template'); return false;";
630			}
631			};
632
633			=head3 load_template
634
635			Used to re-submit search request and load results in different template
636
637			<a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>
638
639			=cut
640
641			$args->{'load_template'} = sub {
642			my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
643			return "load_template($template); return false;";
644			};
645
646			my $out;
647
648			$self->{'tt'}->process(
649			$args->{'template'},
650			$args,
651			\$out
652			) \|\| $log->error( "apply can't process template: ", $self->{'tt'}->error() );
653
654			return $out;
655			}
656
657
658			=head2 __quotemeta
659
660			Helper to quote JavaScript-friendly characters
661
662			=cut
663
664			sub __quotemeta {
665			local $_ = shift;
666			$_ = decode('iso-8859-2', $_);
667
668			s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
669			{
670			use bytes;
671			s<((?:[^ \x21-\x7E]\|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
672			}
673
674			s/\\x09/\\t/g;
675			s/\\x0A/\\n/g;
676			s/\\x0D/\\r/g;
677			s/"/\\"/g;
678			s/\\x5C/\\\\/g;
679
680			return $_;
681			}
682
683
684
685	dpavlin	92	=head1 AUTHOR
686
687	dpavlin	348	Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
688	dpavlin	92
689			=head1 LICENSE
690
691			This library is free software, you can redistribute it and/or modify
692			it under the same terms as Perl itself.
693
694			=cut
695
696			1;