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 WebPAC::Search::Estraier 0.05;
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;
        }

        $self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );

        # 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->error("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");
        }

        my @results = $self->{est}->search( %{ $args } );

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

        my $hits = $#results + 1;

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

        # just return results?
        return @results unless ($args->{'template'});

        #
        # construct HTML results
        #

        my @html_results;

        for my $i ( 0 .. $#results ) {

                my ($database, $prefix, $id);
                if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
                        ($database, $prefix,$id) = ($1,$2,$3);
                } else {
                        $log->warn("can't decode database/prefix/id from " .  $results[$i]->{'@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->{out} += time() - $t;

                $t = time();

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

                push @html_results, $html;

        }

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

        $log->debug( sprintf(
                "duration breakdown: store %.6fs, apply %.6fs, total: %.6fs",
                $times->{db}, $times->{out}, time() - $search_start_t,
        ) );

        return \@html_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->logconfess("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	222	use WebPAC::Search::Estraier 0.05;
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	93	$self->{est} = new WebPAC::Search::Estraier( %{ $est_cfg } );
78	dpavlin	92
79	dpavlin	167	# save config parametars in object
80	dpavlin	348	foreach my $f (qw/db_path template_path hits_on_page webpac_encoding defaultdepth/) {
81	dpavlin	167	$self->{$f} = $c->config->{hyperestraier}->{$f} \|\|
82			$c->config->{webpac}->{$f};
83			$log->debug("self->{$f} = " . $self->{$f});
84			}
85			my $db_path = $self->{db_path};
86			my $template_path = $self->{template_path};
87	dpavlin	95
88			$log->debug("using db path '$db_path', template path '$template_path'");
89
90	dpavlin	222	$self->{db} = new WebPAC::Store(
91	dpavlin	95	path => $db_path,
92			read_only => 1,
93	dpavlin	224	database => $est_cfg->{database},
94	dpavlin	95	);
95
96	dpavlin	101	# default template from config.yaml
97	dpavlin	95	$self->{template} \|\|= $c->config->{webpac}->{template};
98
99	dpavlin	100	$log->debug("converting encoding from webpac_encoding '" .
100			$c->config->{webpac}->{webpac_encoding} .
101	dpavlin	99	"'"
102			);
103	dpavlin	96
104	dpavlin	305	$self->{databases} = $c->config->{databases} \|\| $log->error("can't find databases in config");
105	dpavlin	155
106	dpavlin	379	# create Template toolkit instance
107			$self->{'tt'} = Template->new(
108			INCLUDE_PATH => $template_path,
109			FILTERS => {
110			dump_html => sub {
111			return unless (@_);
112			my $out;
113			my $i = 1;
114			foreach my $v (@_) {
115			$out .= qq{<div id="dump_$i">} .
116			Data::HTMLDumper->Dump([ $v ],[ "v$i" ]) .
117			qq{</div>};
118			$i++;
119			}
120			$out =~ s!<table[^>/]*>!<table class="dump">!gis if ($out);
121			return $out;
122			}
123			},
124			EVAL_PERL => 1,
125			);
126	dpavlin	378
127	dpavlin	92	return $self;
128
129			}
130
131	dpavlin	135
132			=head2 search
133
134	dpavlin	150	my $m->search(
135			phrase => 'query phrase',
136	dpavlin	155	add_attr => \@add_attr
137			get_attr => [ '@uri' ],
138			max => 42,
139	dpavlin	150	template => 'result_template.tt',
140	dpavlin	222	depth => 1,
141	dpavlin	150	);
142	dpavlin	135
143	dpavlin	155	All fields are standard C<WebPAC::Search::Estraier> parametars except
144			C<template> which will (if specified) return results in HTML using
145			selected template.
146
147	dpavlin	135	=cut
148
149	dpavlin	93	sub search {
150	dpavlin	150	my $self = shift;
151	dpavlin	93
152	dpavlin	378	my $search_start_t = time();
153
154	dpavlin	150	my $args = {@_};
155
156	dpavlin	95	my $log = $self->{log};
157	dpavlin	94
158	dpavlin	351	$log->debug("search args: " . Dumper( $args ));
159	dpavlin	95
160	dpavlin	150	my $query = $args->{phrase} \|\| $log->warn("no query phrase") && return;
161	dpavlin	95
162	dpavlin	155	$log->debug("search model query: '$query'");
163			if ($args->{add_attr}) {
164			$log->debug(" + add_attr: " .
165			join("','", @{ $args->{add_attr} })
166			);
167			}
168	dpavlin	150
169			my $template_filename = $args->{template} \|\| $self->{template};
170
171	dpavlin	305	$args->{max} \|\|= $self->{'hits_for_pager'};
172	dpavlin	155	if (! $args->{max}) {
173	dpavlin	305	$args->{max} = 100;
174			$log->warn("max not set when calling model. Using default of $args->{max}");
175	dpavlin	155	}
176	dpavlin	93
177	dpavlin	155	my $times; # store some times for benchmarking
178
179			my $t = time();
180
181	dpavlin	222	# transfer depth of search
182			if (! $args->{depth}) {
183			my $default = $self->{defaultdepth} \|\| $log->logdie("can't find defaultdepth in estraier configuration");
184			$args->{depth} = $default;
185			$log->warn("using default search depth $default");
186			}
187
188	dpavlin	155	my @results = $self->{est}->search( %{ $args } );
189
190			$times->{est} += time() - $t;
191
192	dpavlin	150	my $hits = $#results + 1;
193	dpavlin	99
194	dpavlin	380	$log->debug( sprintf("search took %.6fs and returned $hits hits.", $times->{est}) );
195	dpavlin	150
196	dpavlin	155	# just return results?
197			return @results unless ($args->{'template'});
198
199			#
200			# construct HTML results
201			#
202
203	dpavlin	100	my @html_results;
204
205	dpavlin	95	for my $i ( 0 .. $#results ) {
206
207	dpavlin	237	my ($database, $prefix, $id);
208	dpavlin	293	if ( $results[$i]->{'@uri'} =~ m!/([^/]+)/([^/]+)/(\d+)$!) {
209	dpavlin	237	($database, $prefix,$id) = ($1,$2,$3);
210	dpavlin	224	} else {
211	dpavlin	293	$log->warn("can't decode database/prefix/id from " . $results[$i]->{'@uri'});
212	dpavlin	224	next;
213			}
214	dpavlin	95
215	dpavlin	224	#$log->debug("load_ds( id => $id, prefix => '$prefix' )");
216	dpavlin	95
217	dpavlin	155	$t = time();
218
219	dpavlin	242	my $ds = $self->{db}->load_ds( database => $database, prefix => $prefix, id => $id );
220			if (! $ds) {
221	dpavlin	293	$log->error("can't load_ds( ${database}/${prefix}/${id} )");
222	dpavlin	242	next;
223			}
224	dpavlin	155
225			$times->{db} += time() - $t;
226
227	dpavlin	115	#$log->debug( "ds = " . Dumper( \@html_results ) );
228
229	dpavlin	155	$t = time();
230
231	dpavlin	379	my $html = $self->apply(
232			template => $template_filename,
233			data => $ds,
234			record_uri => "${database}/${prefix}/${id}",
235			config => $self->{databases}->{$database},
236			);
237	dpavlin	100
238	dpavlin	155	$times->{out} += time() - $t;
239
240			$t = time();
241
242	dpavlin	348	$html = decode($self->{webpac_encoding}, $html);
243	dpavlin	100
244			push @html_results, $html;
245
246	dpavlin	95	}
247
248	dpavlin	115	#$log->debug( '@html_results = ' . Dumper( \@html_results ) );
249
250	dpavlin	155	$log->debug( sprintf(
251	dpavlin	378	"duration breakdown: store %.6fs, apply %.6fs, total: %.6fs",
252			$times->{db}, $times->{out}, time() - $search_start_t,
253	dpavlin	155	) );
254
255	dpavlin	100	return \@html_results;
256	dpavlin	93	}
257
258	dpavlin	165	=head2 record
259
260			my $html = $m->record(
261			mfn => 42,
262			template => 'foo.tt',
263			);
264
265			This will load one record, convert it to html using C<template> and return
266			it.
267
268			=cut
269
270			sub record {
271			my $self = shift;
272
273			my $args = {@_};
274			my $log = $self->{log};
275	dpavlin	351	$log->debug("record args: " . Dumper( $args ));
276	dpavlin	165
277	dpavlin	242	foreach my $f (qw/record_uri template/) {
278	dpavlin	228	$log->fatal("need $f") unless ($args->{$f});
279	dpavlin	165	}
280
281	dpavlin	242	my ($database, $prefix, $id);
282	dpavlin	165
283	dpavlin	242	if ($args->{record_uri} =~ m#^([^/]+)/([^/]+)/([^/]+)$#) {
284			($database, $prefix, $id) = ($1,$2,$3);
285			} else {
286			$log->error("can't parse $args->{record_uri} into prefix, database and uri");
287			return;
288			}
289	dpavlin	165
290	dpavlin	242	my $ds = $self->{db}->load_ds( id => $id, prefix => $prefix, database => $database );
291			if (! $ds) {
292			$log->error("can't load_ds( $database/$prefix/$id )");
293			return;
294			}
295
296	dpavlin	379	my $html = $self->apply(
297	dpavlin	165	template => $args->{template},
298			data => $ds,
299	dpavlin	242	record_uri => $args->{record_uri},
300	dpavlin	305	config => $self->{databases}->{$database},
301	dpavlin	165	);
302
303	dpavlin	348	$html = decode($self->{webpac_encoding}, $html);
304	dpavlin	165
305			return $html;
306			}
307
308	dpavlin	305
309	dpavlin	135	=head2 save_html
310	dpavlin	93
311	dpavlin	135	$m->save_html( '/full/path/to/file', $content );
312	dpavlin	93
313	dpavlin	348	It will use C<Encode> to convert content encoding back to
314	dpavlin	135	Webpac codepage, recode JavaScript Unicode entities (%u1234),
315			strip extra newlines at beginning and end, and save to
316			C</full/path/to/file.new> and if that succeeds, just rename
317			it over original file which should be atomic on filesystem level.
318
319			=cut
320
321			sub save_html {
322			my ($self, $path, $content) = @_;
323
324	dpavlin	348	# FIXME Should this be UTF-8 or someting?
325			my $js_encoding = $self->{webpac_encoding};
326			$js_encoding = 'UTF-16';
327
328	dpavlin	135	sub _conv_js {
329	dpavlin	348	return '0x' . $_[1];
330			return encode($_[0], chr(hex($_[1])));
331	dpavlin	135	}
332	dpavlin	348	#$content =~ s/%u([a-fA-F0-9]{4})/_conv_js($js_encoding,$1)/gex;
333	dpavlin	135	$content =~ s/^[\n\r]+//s;
334			$content =~ s/[\n\r]+$/\n/s;
335	dpavlin	282	$content =~ s/\n\r/\n/gs;
336	dpavlin	135
337	dpavlin	348	my $disk_encoding = $self->{webpac_encoding} \|\| 'utf-8';
338			$self->{log}->debug("convert encoding to $disk_encoding");
339			from_to($content, 'utf-8', $disk_encoding) \|\| $self->{log}->warn("encoding from utf-8 to $disk_encoding failed for: $content");
340	dpavlin	179
341			write_file($path . '.new', {binmode => ':raw' }, $content) \|\| die "can't save ${path}.new $!";
342	dpavlin	135	rename $path . '.new', $path \|\| die "can't rename to $path: $!";
343			}
344
345			=head2 load_html
346
347			my $html = $m->load_html('/full/path/to/file');
348
349			This will convert file from Webpac encoding to Catalyst and
350			convert that data to escaped HTML (for sending into
351			C<< <textarea/> >> tags in html.
352
353			=cut
354
355			sub load_html {
356			my ($self, $path) = @_;
357
358			die "no path?" unless ($path);
359
360	dpavlin	179	my $content = read_file($path, {binmode => ':raw' }) \|\| die "can't read $path: $!";
361	dpavlin	135
362	dpavlin	348	return decode($self->{webpac_encoding}, $content);
363	dpavlin	135	}
364
365	dpavlin	379
366			=head2 apply
367
368			Create output from in-memory data structure using Template Toolkit template.
369
370			my $text = $tt->apply(
371			template => 'text.tt',
372			data => $ds,
373			record_uri => 'database/prefix/mfn',
374			);
375
376			It also has follwing template toolikit filter routies defined:
377
378			=cut
379
380			sub apply {
381			my $self = shift;
382
383			my $args = {@_};
384
385			my $log = $self->{log} \|\| die "no log?";
386
387			foreach my $a (qw/template data/) {
388			$log->logconfess("need $a") unless ($args->{$a});
389			}
390
391			=head3 tt_filter_type
392
393			filter to return values of specified from $ds, usage from TT template is in form
394			C<d('FieldName','delimiter')>, where C<delimiter> is optional, like this:
395
396			[% d('Title') %]
397			[% d('Author',', ' %]
398
399			=cut
400
401			sub tt_filter_type {
402			my ($data,$type) = @_;
403
404			die "no data?" unless ($data);
405			$type \|\|= 'display';
406
407			my $default_delimiter = {
408			'display' => '¶<br/>',
409			'index' => '\n',
410			};
411
412			return sub {
413
414			my ($name,$join) = @_;
415
416			die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
417			# Hm? Should we die here?
418			return unless ($name);
419
420			my $item = $data->{'data'}->{$name} \|\| return;
421
422			my $v = $item->{$type} \|\| return;
423
424			if (ref($v) eq 'ARRAY') {
425			if ($#{$v} == 0) {
426			$v = $v->[0];
427			} else {
428			$join = $default_delimiter->{$type} unless defined($join);
429			$v = join($join, @{$v});
430			}
431			} else {
432			warn("TT filter $type(): field $name values aren't ARRAY, ignoring");
433			}
434
435			return $v;
436			}
437			}
438
439			$args->{'d'} = tt_filter_type($args, 'display');
440			$args->{'display'} = tt_filter_type($args, 'display');
441
442			=head3 tt_filter_search
443
444			filter to return links to search, usage in TT:
445
446			[% search('FieldToDisplay','FieldToSearch','optional delimiter', 'optional_template.tt') %]
447
448			=cut
449
450			sub tt_filter_search {
451
452			my ($data) = @_;
453
454			die "no data?" unless ($data);
455
456			return sub {
457
458			my ($display,$search,$delimiter,$template) = @_;
459
460			# default delimiter
461			$delimiter \|\|= '¶<br/>',
462
463			die "no data hash" unless ($data->{'data'} && ref($data->{'data'}) eq 'HASH');
464			# Hm? Should we die here?
465			return unless ($display);
466
467			my $item = $data->{'data'}->{$display} \|\| return;
468
469			return unless($item->{'display'});
470			if (! $item->{'search'}) {
471			warn "error in TT template: field $display didn't insert anything into search, use d('$display') and not search('$display'...)";
472			return;
473			}
474
475			my @warn;
476			foreach my $type (qw/display search/) {
477			push @warn, "field $display type $type values aren't ARRAY" unless (ref($item->{$type}) eq 'ARRAY');
478			}
479
480			if (@warn) {
481			warn("TT filter search(): " . join(",", @warn) . ", skipping");
482			return;
483			}
484			my @html;
485
486			my $d_el = $#{ $item->{'display'} };
487			my $s_el = $#{ $item->{'search'} };
488
489			# easy, both fields have same number of elements or there is just
490			# one search and multiple display
491			if ( $d_el == $s_el \|\| $s_el == 0 ) {
492
493			foreach my $i ( 0 .. $d_el ) {
494
495			my $s;
496			if ($s_el > 0) {
497			$s = $item->{'search'}->[$i] \|\| die "can't find value $i for type search in field $search";
498			} else {
499			$s = $item->{'search'}->[0];
500			}
501			#$s =~ s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
502			$s = __quotemeta( $s );
503
504			my $d = $item->{'display'}->[$i] \|\| die "can't find value $i for type display in field $display";
505
506			my $template_arg = '';
507			$template_arg = qq{,'$template'} if ($template);
508
509			push @html, qq{<a href="#" onclick="return search_via_link('$search','$s'${template_arg})">$d</a>};
510			}
511
512			return join($delimiter, @html);
513			} else {
514			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>};
515			my $v = $item->{'display'};
516
517			if ($#{$v} == 0) {
518			$html .= $v->[0];
519			} else {
520			$html .= join($delimiter, @{$v});
521			}
522			return $html;
523			}
524			}
525			}
526
527			$args->{'search'} = tt_filter_search($args);
528
529			=head3 load_rec
530
531			Used mostly for onClick events like this:
532
533			<a href="#" onClick="[% load_rec( record_uri, 'template_name.tt') %]>foo</a>
534
535			It will automatically do sanity checking and create correct JavaScript code.
536
537			=cut
538
539			$args->{'load_rec'} = sub {
540			my @errors;
541
542			my $record_uri = shift or push @errors, "record_uri missing";
543			my $template = shift or push @errors, "template missing";
544
545			if ($record_uri !~ m#^[^/]+/[^/]+/[^/]+$#) {
546			push @errors, "invalid format of record_uri: $record_uri";
547			}
548
549			if (@errors) {
550			return "Logger.error('errors in load_rec: " . join(", ", @errors) . "'); return false;";
551			} else {
552			return "load_rec('$record_uri','$template'); return false;";
553			}
554			};
555
556			=head3 load_template
557
558			Used to re-submit search request and load results in different template
559
560			<a href="#" onClick="[% load_template( 'template_name.tt' ) %]">bar</a>
561
562			=cut
563
564			$args->{'load_template'} = sub {
565			my $template = shift or return "Logger.error('load_template missing template name!'); return false;";
566			return "load_template($template); return false;";
567			};
568
569			my $out;
570
571			$self->{'tt'}->process(
572			$args->{'template'},
573			$args,
574			\$out
575			) \|\| $log->error( "apply can't process template: ", $self->{'tt'}->error() );
576
577			return $out;
578			}
579
580
581			=head2 __quotemeta
582
583			Helper to quote JavaScript-friendly characters
584
585			=cut
586
587			sub __quotemeta {
588			local $_ = shift;
589			$_ = decode('iso-8859-2', $_);
590
591			s<([\x{0080}-\x{fffd}]+)>{sprintf '\u%0*v4X', '\u', $1}ge if ( Encode::is_utf8($_) );
592			{
593			use bytes;
594			s<((?:[^ \x21-\x7E]\|(?:\\(?!u)))+)>{sprintf '\x%0*v2X', '\x', $1}ge;
595			}
596
597			s/\\x09/\\t/g;
598			s/\\x0A/\\n/g;
599			s/\\x0D/\\r/g;
600			s/"/\\"/g;
601			s/\\x5C/\\\\/g;
602
603			return $_;
604			}
605
606
607
608	dpavlin	92	=head1 AUTHOR
609
610	dpavlin	348	Dobrica Pavlinusic C<< <dpavlin@rot13.org> >>
611	dpavlin	92
612			=head1 LICENSE
613
614			This library is free software, you can redistribute it and/or modify
615			it under the same terms as Perl itself.
616
617			=cut
618
619			1;