pdfsearch.pl

   1 #!/usr/bin/perl
   2 use utf8;
   3 use Encode 'decode';
   4 use Encode 'encode';
   5 use CGI;
   6 use CGI::Carp qw(fatalsToBrowser);
   7 use WebService::Solr;
   8 use WebService::Solr::Query;
   9 use warnings;
  10 use Data::Dumper;
  11 use YAML::XS;
  12 use DBD::SQLite;
  13 use DBI;
  14 use URI::Escape;
  15 use HTML::Template::Pro;
  16
  17 #
  18 #   Libre10
  19 #
  20 #   Copyright 2013 yukikaze
  21 #   Licensed under the Apache License, Version 2.0 (the "License");
  22 #   you may not use this file except in compliance with the License.
  23 #   You may obtain a copy of the License at
  24 #
  25 #       http://www.apache.org/licenses/LICENSE-2.0
  26 #
  27 #   Unless required by applicable law or agreed to in writing, software
  28 #   distributed under the License is distributed on an "AS IS" BASIS,
  29 #   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  30 #   See the License for the specific language governing permissions and
  31 #   limitations under the License.
  32 #
  33 my $confdata = YAML::XS::LoadFile( -e 'libre10.conf' ? 'libre10.conf' : '/etc/libre10.conf' );
  34
  35 my $sqlpath = "/tmp/libre10.db";
  36 $sqlpath = $confdata->{dburl};
  37 $sqldbi =
  38   DBI->connect( "dbi:SQLite:dbname=$sqlpath", "", "",
  39         { RaiseError => 1, AutoCommit => 1 } );
  40 my $q           = new CGI;
  41 my $mode        = "";
  42 my $search_text = $q->param("search_text");
  43 $mode   = $q->param("mode");
  44 $width  = $q->param("width");
  45 $height = $q->param("height");
  46 $start  = $q->param("start");
  47 $rows   = $q->param("rows");
  48 if ( $start eq "" ) { $start = "0"; }
  49 if ( $rows  eq "" ) { $rows  = "10"; }
  50
  51 my $url      = $confdata->{solrurl};
  52 my $solr     = WebService::Solr->new($url);
  53 my $template = HTML::Template::Pro->new(
  54         filename          => "./template/desktop_top.tmpl",
  55         die_on_bad_params => 0
  56 );
  57 print $q->header( -type => "text/html", -charset => "utf-8" );
  58
  59 $template->param( mode   => $mode );
  60 $template->param( width  => $width );
  61 $template->param( height => $height );
  62 $template->param( start  => $start );
  63 $template->param( rows   => $rows );
  64
  65 $bstart = int($start) - int($rows);
  66 $nstart = int($start) + int($rows);
  67 if ( $bstart < 0 ) {
  68         $bstart = 0;
  69 }
  70 $blink =
  71 "pdfsearch.pl?mode=$mode&search_text=$search_text&width=$width&height=$height&start=$bstart&rows=$rows";
  72 $nlink =
  73 "pdfsearch.pl?mode=$mode&search_text=$search_text&width=$width&height=$height&start=$nstart&rows=$rows";
  74 $template->param( blink => $blink );
  75 $template->param( nlink => $nlink );
  76 if ( length $search_text > 0 ) {
  77         $template->param( text        => $search_text );
  78         $template->param( show_result => "true" );
  79         $template->param( search_top  => "0" );
  80         my $query = WebService::Solr::Query->new(
  81                 { text => decode( 'utf-8', $search_text ) } );
  82         my $response = $solr->search(
  83                 $query,
  84                 {
  85                         'rows'                        => $rows,
  86                         'start'                       => $start,
  87                         'group.limit'                 => '30',
  88                         'group.field'                 => 'title_group',
  89                         'group'                       => true,
  90                         'hl'                          => true,
  91                         'hl.fl'                       => 'text',
  92                         'hl.simple.pre'               => "<font color=red><em>",
  93                         'hl.simple.post'              => "</em></font>",
  94                         'hl.fragsize'                 => '300',
  95                         'hl.fragmenter'               => 'gap',
  96                         'hl.alternateField'           => 'text',
  97                         'hl.maxAlternateFieldLength'  => '300',
  98                         'hl.useFastVectorHighlighter' => true,
  99                 }
 100         );
 101
 102         my $hits               = $response->content;
 103         my %hits               = %$hits;
 104         my $response_header    = $hits{grouped}{title_group}{groups};
 105         my @response_header    = @$response_header;
 106         my @print_group_arrays = ();
 107         foreach my $group (@response_header) {
 108                 my $textarray         = $group->{doclist}{docs};
 109                 my @textarray         = @$textarray;
 110                 my %print_group_array = &printGroupResults(
 111                         $group->{doclist}{numFound},
 112                         $textarray[0]->{title_group},
 113                         $search_text, $textarray, $response->content->{highlighting}
 114                 );
 115                 push( @print_group_arrays, \%print_group_array );
 116         }
 117         $template->param( group => \@print_group_arrays );
 118 }
 119 else {
 120         $template->param( search_top => true );
 121 }
 122
 123 sub printGroupResults() {
 124         ( $hitsnum, $title, $search_text, $hits, $hits_hl ) = @_;
 125         my $result_group;
 126         my @print_page_arrays = ();
 127         $template->param( show_result => 1 );
 128         foreach $doc (@$hits) {
 129                 ( $pagenum, $path, $linkpath ) =
 130                   ( $doc->{page}, $doc->{path_id}, $doc->{path_id} );
 131                 my $findPageSQL =
 132                   "SELECT startpage,title_id from pdffile WHERE id = '$path';";
 133                 my $sth = $sqldbi->prepare($findPageSQL);
 134                 $sth->execute;
 135                 $dbarray     = $sth->fetchrow_arrayref;
 136                 $showpagenum = $pagenum + $dbarray->[0];
 137                 $ref =
 138                     "title="
 139                   . uri_escape( $dbarray->[1] )
 140                   . "&pagenum=$showpagenum&mode=$mode&width=$width&height=$height";
 141                 my %print_page_array = (
 142                         page    => $showpagenum,
 143                         hl_text => $hits_hl->{ $doc->{id} }{text}[0],
 144                         refaddr => $ref
 145                 );
 146                 @print_page_array = ($print_page_array);
 147                 push( @print_page_arrays, \%print_page_array );
 148         }
 149         %print_group_array = (
 150                 text          => $title,
 151                 search_num    => $hitsnum,
 152                 search_result => \@print_page_arrays
 153         );
 154         return %print_group_array;
 155 }
 156 print $template->output();
 157 $sqldbi . disconnect;