| 1 |
ulpfr |
55 |
#!/usr/bin/perl -w |
| 2 |
|
|
# -*- Mode: Perl -*- |
| 3 |
|
|
# $Basename$ |
| 4 |
|
|
# $Revision: 1.1 $ |
| 5 |
|
|
# Author : Ulrich Pfeifer |
| 6 |
|
|
# Created On : Mon Dec 31 13:57:11 2001 |
| 7 |
|
|
# Last Modified By: Ulrich Pfeifer |
| 8 |
|
|
# Last Modified On: Mon Dec 31 14:37:43 2001 |
| 9 |
|
|
# Language : CPerl |
| 10 |
|
|
# |
| 11 |
|
|
# (C) Copyright 2001, UUNET Deutschland GmbH, Germany |
| 12 |
|
|
# |
| 13 |
|
|
|
| 14 |
|
|
use strict; |
| 15 |
|
|
use File::Path; |
| 16 |
|
|
use DB_File; |
| 17 |
|
|
use Getopt::Long; |
| 18 |
|
|
use Cwd; |
| 19 |
|
|
|
| 20 |
|
|
require WAIT::Config; |
| 21 |
|
|
require WAIT::Database; |
| 22 |
|
|
require WAIT::Parse::Ora; |
| 23 |
|
|
require WAIT::Document::Ora; |
| 24 |
|
|
require WAIT::InvertedIndex; |
| 25 |
|
|
|
| 26 |
|
|
|
| 27 |
|
|
$DB_BTREE->{'cachesize'} = 200_000 ; |
| 28 |
|
|
|
| 29 |
|
|
my %OPT = (clean => 0, |
| 30 |
|
|
database => 'DB', |
| 31 |
|
|
dir => $WAIT::Config->{WAIT_home} || '/tmp', |
| 32 |
|
|
table => 'ora', |
| 33 |
|
|
); |
| 34 |
|
|
|
| 35 |
|
|
GetOptions(\%OPT, |
| 36 |
|
|
'clean!', |
| 37 |
|
|
'database=s', |
| 38 |
|
|
'dir=s', |
| 39 |
|
|
'table=s', |
| 40 |
|
|
) || die "Usage: ...\n"; |
| 41 |
|
|
|
| 42 |
|
|
if ($OPT{clean} and -d "$OPT{dir}/$OPT{database}") { |
| 43 |
|
|
my $tmp = WAIT::Database->open(name => $OPT{database}, |
| 44 |
|
|
'directory' => $OPT{dir}) |
| 45 |
|
|
or die "Could not open table $OPT{table}: $@\n"; |
| 46 |
|
|
my $tbl = $tmp->table(name => $OPT{table}); |
| 47 |
|
|
$tbl->drop if $tbl; |
| 48 |
|
|
rmtree("$OPT{dir}/$OPT{database}/$OPT{table}", 1, 1) |
| 49 |
|
|
if -d "$OPT{dir}/$OPT{database}/$OPT{table}"; |
| 50 |
|
|
$tmp->close; |
| 51 |
|
|
} |
| 52 |
|
|
|
| 53 |
|
|
my $db; |
| 54 |
|
|
unless (-d "$OPT{dir}/$OPT{database}") { |
| 55 |
|
|
$db = WAIT::Database->create(name => $OPT{database}, |
| 56 |
|
|
'directory' => $OPT{dir}) |
| 57 |
|
|
or die "Could not open database $OPT{database}: $@\n"; |
| 58 |
|
|
} |
| 59 |
|
|
else { |
| 60 |
|
|
$db = WAIT::Database->open(name => $OPT{database}, |
| 61 |
|
|
'directory' => $OPT{dir}) |
| 62 |
|
|
or die "Could not open table $OPT{table}: $@\n"; |
| 63 |
|
|
} |
| 64 |
|
|
|
| 65 |
|
|
my $layout = new WAIT::Parse::Ora; |
| 66 |
|
|
|
| 67 |
|
|
my $stem = ['isotr', 'isolc', 'split2', 'stop', 'Stem']; |
| 68 |
|
|
my $text = [{ |
| 69 |
|
|
'prefix' => ['isotr', 'isolc'], |
| 70 |
|
|
'intervall' => ['isotr', 'isolc'], |
| 71 |
|
|
}, |
| 72 |
|
|
'isotr', 'isolc', 'split2', 'stop']; |
| 73 |
|
|
my $sound = ['isotr', 'isolc', 'split2', 'Soundex'],; |
| 74 |
|
|
|
| 75 |
|
|
my $cwd = cwd; |
| 76 |
|
|
|
| 77 |
|
|
my %D; |
| 78 |
|
|
my $access = tie %D, 'WAIT::Document::Ora', @ARGV, |
| 79 |
|
|
or die "Couldn't tie to file: $!\n"; |
| 80 |
|
|
|
| 81 |
|
|
my $tb = $db->create_table(name => $OPT{table}, |
| 82 |
|
|
attr => ['text', 'author', 'title', |
| 83 |
|
|
'headline', 'docid'], |
| 84 |
|
|
layout => $layout, |
| 85 |
|
|
access => $access, |
| 86 |
|
|
invindex => |
| 87 |
|
|
[ |
| 88 |
|
|
'title' => $stem, |
| 89 |
|
|
'text' => $text, |
| 90 |
|
|
'author' => $text, |
| 91 |
|
|
] |
| 92 |
|
|
); |
| 93 |
|
|
die "Couldn't create table $OPT{table}: $@\n" unless $tb; |
| 94 |
|
|
|
| 95 |
|
|
my ($did, $value); |
| 96 |
|
|
while (($did, $value) = each %D) { |
| 97 |
|
|
my $record = $layout->split($value); |
| 98 |
|
|
my $headline = $record->{title}; |
| 99 |
|
|
$headline =~ s/\s+/ /sg; |
| 100 |
|
|
printf "%s\n", substr($headline,0,80); |
| 101 |
|
|
$tb->insert('docid' => $did, |
| 102 |
|
|
headline => $headline, |
| 103 |
|
|
%{$record}); |
| 104 |
|
|
} |
| 105 |
|
|
$tb->set(top=>1); |
| 106 |
|
|
$tb->close(); |
| 107 |
|
|
$db->close(); |
| 108 |
|
|
|
| 109 |
|
|
$WAIT::Config = $WAIT::Config; # make perl -w happy |
| 110 |
|
|
|
| 111 |
|
|
|
| 112 |
|
|
__END__ |
| 113 |
|
|
## ################################################################### |
| 114 |
|
|
## pod |
| 115 |
|
|
## ################################################################### |
| 116 |
|
|
|
| 117 |
|
|
=head1 NAME |
| 118 |
|
|
|
| 119 |
|
|
index_ora - generate an WAIT index for O'Reilly catalog |
| 120 |
|
|
|
| 121 |
|
|
=head1 SYNOPSIS |
| 122 |
|
|
|
| 123 |
|
|
B<index_ora> |
| 124 |
|
|
[B<-clean>] [B<-noclean>] |
| 125 |
|
|
[B<-database> I<dbname>] |
| 126 |
|
|
[B<-dir> I<directory>] |
| 127 |
|
|
[B<-table> I<table name>] |
| 128 |
|
|
I<directory> |
| 129 |
|
|
|
| 130 |
|
|
=head1 DESCRIPTION |
| 131 |
|
|
|
| 132 |
|
|
=head1 OPTIONS |
| 133 |
|
|
|
| 134 |
|
|
=over 5 |
| 135 |
|
|
|
| 136 |
|
|
=item B<-clean> / B<-noclean> |
| 137 |
|
|
|
| 138 |
|
|
Clean the table before indexing. Default is B<off>. |
| 139 |
|
|
|
| 140 |
|
|
=item B<-database> I<dbname> |
| 141 |
|
|
|
| 142 |
|
|
Specify database name. Default is F<DB>. |
| 143 |
|
|
|
| 144 |
|
|
=item B<-dir> I<directory> |
| 145 |
|
|
|
| 146 |
|
|
Alternate directory where databases are located. Default is the |
| 147 |
|
|
directory specified during configuration of WAIT. |
| 148 |
|
|
|
| 149 |
|
|
=item B<-table> I<table name> |
| 150 |
|
|
|
| 151 |
|
|
Specify an alternate table name. Default is C<ora>. |
| 152 |
|
|
|
| 153 |
|
|
=head1 AUTHOR |
| 154 |
|
|
|
| 155 |
|
|
Ulrich Pfeifer E<lt>F<pfeifer@wait.de>E<gt> |
| 156 |
|
|
|
Parent Directory
| 
Revision Log