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/) {
                $self->{$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 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;

        $log->debug("search model query: '$query'");
        if ($args->{add_attr}) {
                $log->debug(" + add_attr: " .
                        join("','", @{ $args->{add_attr} })
                );
        }

        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");
        }

        $log->debug("searching for maximum $args->{max} results using depth $args->{depth}");

        #
        # 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( _convert( $_ ) );
                        $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;
        }

        $times->{est} += time() - $t;

        $cond->set_max( $page * $max );

        my $result = $self->{est_node}->search($cond, ( $args->{depth} || 0 ));
        my $hits = $result->doc_num;

        $log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );

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