#!/usr/bin/perl
#
# swishquery.pl
# by Reto Hersiczky
# Copyright (c) 2000 by futureLAB AG
#
# https://team.futurelab.ch/cgi-bin/swishquery.pl?keywords=abc&swishindex=%2Fusr%2Flocal%2Fswish-e%2Findex_team_futurelab.swish
#
#
#
#
#
use CGI;
use strict;
my $VERSION = '0.91';
# ----- FORWARD DECLARATIONS & PROTOTYPING
#
sub build_occurrences ($);
sub Debug ($);
$| = 1;
# ----- GLOBAL VARIABLES
#
my $max_result = 100;
my $max_size_for_pdf = 900000;
my $q = CGI->new;
# whereis swish
my $swishexec = "/usr/local/bin/swish-e";
my $keywords = $q->param('keywords');
my $swishindex = $q->param('swishindex');
# build the command string
my $command = "$swishexec -f \"$swishindex\" -w \"$keywords\" -m $max_result";
print "Content-type: text/html\n\n";
print qq~
~;
######################################################################
#
# MAIN
#
######################################################################
print qq~
Search Result for $keywords - V$VERSION
~;
my $match_counter = my $result_count = 0;
open(SWISH, "$command |") or print "ERROR $!";
while () {
if (/^\# Number of/i) {
$result_count = (split /\:/, $_)[1];
last;
}
}
print qq~
Query «$keywords» found $result_count matches
~;
while () {
last if /^\./;
my ($file, $title, $size) = split /\"/;
$file =~ s/ $//; # remove last space;
my $ranking = $file;
$ranking =~ s/^(\d+) .+/$1/;
$ranking = int($ranking / 10);
$ranking = 100 if $ranking > 100;
$ranking = 1 if $ranking < 1;
$file =~ s/^\d+ //; # remove 1st ranking code
if ($file =~ /[pdf|sdw|tar]$/i) {
# fix missing bytes with PDF docs
my $h_file = $file;
$h_file =~ s/^\/files\//\/home\/team\//;
$size = -s $h_file;
}
my $imageicon = "/img/ico_unknown.gif";
$imageicon = "/img/ico_pdf.gif" if $file =~ /pdf$/i;
$imageicon = "/img/ico_txt.gif" if $file =~ /txt$/i;
$imageicon = "/img/ico_ns.gif" if $file =~ /html?$/i;
$imageicon = "/img/ico_staroffice.gif" if $file =~ /sdw$/i;
$imageicon = "/img/ico_zip.gif" if $file =~ /zip$/i;
$imageicon = "/img/ico_tar.gif" if $file =~ /tar$/i;
print qq~
$title -
$ranking \% - $size bytes
~;
$file =~ s/^\/files\//\/home\/team\//;
my $details = "";
if ($file =~ /pdf$/i && $size > $max_size_for_pdf) {
$details = "[ PDF document > $max_size_for_pdf bytes ]";
}
elsif ($file =~ /gif$/i) {
$details = "[ GIF Image ]";
}
else {
Debug "N1006: build occurences..." if $q->param('debug');
$details = build_occurrences $file;
}
print $details;
my @f_elems = split /\//, $file; shift @f_elems, shift @f_elems, shift @f_elems;
my $link2cat = shift @f_elems;
print '
', "![\"$link2cat\"](\"/img/folderclosed.gif\")
[ $link2cat";
foreach (@f_elems) {
$link2cat .= "/$_";
print " : $_";
}
print ' ]
';
last if $match_counter++ > $max_result -2;
}
close SWISH;
print qq~
[ Home ]
~;
##################################################################
#
# SUB ROUTINES
#
##################################################################
# build the occurrences of keywords in file
sub build_occurrences ($) {
my $file = shift;
my @keyw_local = split / /, $keywords;
my $keyw_loc = length($keyw_local[0]) > length($keyw_local[1]) ? $keyw_local[0] : $keyw_local[1];
my $keyw_loc_2nd = length($keyw_local[0]) > length($keyw_local[1]) ? $keyw_local[1] : $keyw_local[0];
Debug "N1007: keyw_loc = '$keyw_loc' / keyw_loc_2nd = '$keyw_loc_2nd'" if $q->param('debug');
# remove asterisk anywhere
$keyw_loc =~ s/\*//;
my $content = "";
if ($file =~ /pdf$/i) {
open(PDFTOTEXT, "/usr/bin/pdftotext \"$file\" - 2>/dev/null |");
my @pdflines = ;
close PDFTOTEXT;
$content = join " ", @pdflines;
}
elsif ($file =~ /sdw$/i) {
open(SDWTOTEXT, "/usr/local/swish-e/filters-bin/sdw-filter.sh $file 2>/dev/null |");
my @lines = ;
close SDWTOTEXT;
$content = join " ", @lines;
}
elsif ($file =~ /zip$/i) {
open(TOTEXT, "/usr/local/swish-e/filters-bin/zip-filter.sh $file 2>/dev/null |");
my @lines = ;
close TOTEXT;
$content = join " ", @lines;
}
elsif ($file =~ /tar$/i) {
open(TOTEXT, "/usr/local/swish-e/filters-bin/tar-filter.sh $file 2>/dev/null |");
my @lines = ;
close TOTEXT;
$content = join " ", @lines;
}
else {
# parse text or html file
local $/ = undef; # local declaration is important
open(F, $file) or print $!;
$content = ;
close F;
$content =~ s/<[^>]*>/ /g if $file =~ /html?$/; # Remove HTML tags
}
Debug "N1003: length of \$content: " . length($content) if $q->param('debug');
$keyw_loc =~ tr/a-z/A-Z/ if length($keyw_loc) < 5; # uppercase for very short words
$keyw_loc_2nd =~ tr/a-z/A-Z/ if length($keyw_loc) < 5; # uppercase for very short words
# try to locate
my $pos = index($content, $keyw_loc, 0);
Debug "N1000: '$keyw_loc' found at position '$pos'" if $q->param('debug');
if ($pos < 0) {
# keyword(s) doesn't not match in content!
if ($pos < 0) {
$keyw_loc = ucfirst lc $keyw_loc; # first letter upper case
$pos = index($content, $keyw_loc, 0);
Debug "N1001: '$keyw_loc' found at position '$pos'" if $q->param('debug');
}
if ($pos < 0) {
$keyw_loc =~ tr/a-z/A-Z/;
$pos = index($content, $keyw_loc, 0);
Debug "N1002: '$keyw_loc' found at position '$pos'" if $q->param('debug');
}
if ($pos < 0) {
$keyw_loc =~ tr/A-Z/a-z/;
$pos = index($content, $keyw_loc, 0);
Debug "N1004: '$keyw_loc' found at position '$pos'" if $q->param('debug');
}
if ($pos < 0) {
$keyw_loc = $keyw_loc_2nd;
$pos = index($content, $keyw_loc, 0);
Debug "N1005: '$keyw_loc' found at position '$pos'" if $q->param('debug');
}
}
$pos -= 70; # move back NN characters
$pos = 0 if $pos < 0;
my $width = 700; # size of partial content
$width = length($content) - $pos if ($pos + $width) > length($content);
Debug "\$partial = substr(\$content, $pos, $width)" if $q->param('debug');
my $partial = substr($content, $pos, $width);
$partial =~ s/$keyw_loc/$keyw_loc<\/b>/i;
return $partial;
}
sub Debug ($) {
print "
DEBUG: $_[0]
\n";
}
# next free application codes
# E1000
# N1008