1 |
# -*- Mode: Perl -*- |
2 |
# $Basename: index_yow $ |
3 |
# $Revision: 1.2 $ |
4 |
# Author : Ulrich Pfeifer |
5 |
# Created On : Fri Apr 7 13:45:50 2000 |
6 |
# Last Modified By: Ulrich Pfeifer |
7 |
# Last Modified On: Mon May 8 20:33:15 2000 |
8 |
# Language : CPerl |
9 |
# |
10 |
# (C) Copyright 2000, UUNET Deutschland GmbH, Germany |
11 |
# |
12 |
|
13 |
use strict; |
14 |
use File::Path; |
15 |
use DB_File; |
16 |
use Getopt::Long; |
17 |
use Cwd; |
18 |
|
19 |
require WAIT::Config; |
20 |
require WAIT::Database; |
21 |
require WAIT::Parse::Base; |
22 |
require WAIT::Document::Split; |
23 |
require WAIT::InvertedIndex; |
24 |
|
25 |
|
26 |
$DB_BTREE->{'cachesize'} = 200_000 ; |
27 |
|
28 |
my %OPT = (clean => 0, |
29 |
database => 'DB', |
30 |
dir => $WAIT::Config->{WAIT_home} || '/tmp', |
31 |
table => 'yow', |
32 |
); |
33 |
|
34 |
GetOptions(\%OPT, |
35 |
'clean!', |
36 |
'database=s', |
37 |
'dir=s', |
38 |
'table=s', |
39 |
) || die "Usage: ...\n"; |
40 |
|
41 |
if ($OPT{clean} and -d "$OPT{dir}/$OPT{database}") { |
42 |
my $tmp = WAIT::Database->open(name => $OPT{database}, |
43 |
'directory' => $OPT{dir}) |
44 |
or die "Could not open table $OPT{table}: $@\n"; |
45 |
my $tbl = $tmp->table(name => $OPT{table}); |
46 |
$tbl->drop if $tbl; |
47 |
rmtree("$OPT{dir}/$OPT{database}/$OPT{table}", 1, 1) |
48 |
if -d "$OPT{dir}/$OPT{database}/$OPT{table}"; |
49 |
$tmp->close; |
50 |
} |
51 |
|
52 |
my $db; |
53 |
unless (-d "$OPT{dir}/$OPT{database}") { |
54 |
$db = WAIT::Database->create(name => $OPT{database}, |
55 |
'directory' => $OPT{dir}) |
56 |
or die "Could not open database $OPT{database}: $@\n"; |
57 |
} |
58 |
else { |
59 |
$db = WAIT::Database->open(name => $OPT{database}, |
60 |
'directory' => $OPT{dir}) |
61 |
or die "Could not open table $OPT{table}: $@\n"; |
62 |
} |
63 |
|
64 |
my $layout = new WAIT::Parse::Base; |
65 |
|
66 |
my $stem = ['isotr', 'isolc', 'split2', 'stop', 'Stem']; |
67 |
my $text = [{ |
68 |
'prefix' => ['isotr', 'isolc'], |
69 |
'intervall' => ['isotr', 'isolc'], |
70 |
}, |
71 |
'isotr', 'isolc', 'split2', 'stop']; |
72 |
my $sound = ['isotr', 'isolc', 'split2', 'Soundex'],; |
73 |
|
74 |
my $cwd = cwd; |
75 |
|
76 |
my %D; |
77 |
my $access = tie %D, 'WAIT::Document::Split', 'sep', '\000', @ARGV, |
78 |
or die "Couldn't tie to file: $!\n"; |
79 |
|
80 |
my $tb = $db->create_table(name => $OPT{table}, |
81 |
attr => ['docid', 'headline', 'text'], |
82 |
layout => $layout, |
83 |
access => $access, |
84 |
invindex => |
85 |
[ |
86 |
'text' => $stem, 'text' => $text, |
87 |
] |
88 |
); |
89 |
die "Couldn't create table $OPT{table}: $@\n" unless $tb; |
90 |
|
91 |
my ($did, $value); |
92 |
while (($did, $value) = each %D) { |
93 |
my $record = $layout->split($value); |
94 |
my $headline = $record->{text}; |
95 |
$headline =~ s/\s+/ /sg; |
96 |
printf "%s\n", substr($headline,0,80); |
97 |
$tb->insert('docid' => $did, |
98 |
headline => $headline, |
99 |
%{$record}); |
100 |
} |
101 |
$tb->set(top=>1); |
102 |
$tb->close(); |
103 |
$db->close(); |
104 |
|
105 |
$WAIT::Config = $WAIT::Config; # make perl -w happy |
106 |
|
107 |
|
108 |
__END__ |
109 |
## ################################################################### |
110 |
## pod |
111 |
## ################################################################### |
112 |
|
113 |
=head1 NAME |
114 |
|
115 |
yow - generate an WAIT index for yow files |
116 |
|
117 |
=head1 SYNOPSIS |
118 |
|
119 |
B<yow> |
120 |
[B<-clean>] [B<-noclean>] |
121 |
[B<-database> I<dbname>] |
122 |
[B<-dir> I<directory>] |
123 |
[B<-table> I<table name>] |
124 |
|
125 |
=head1 DESCRIPTION |
126 |
|
127 |
Either indexes F<$WAIT/t/test.ste> (if called from directory F<$WAIT>) |
128 |
or F</usr/local/ls6/tex/bib/bibdb.ste>. |
129 |
|
130 |
=head1 OPTIONS |
131 |
|
132 |
=over 5 |
133 |
|
134 |
=item B<-clean> / B<-noclean> |
135 |
|
136 |
Clean the table before indexing. Default is B<off>. |
137 |
|
138 |
=item B<-database> I<dbname> |
139 |
|
140 |
Specify database name. Default is F<DB>. |
141 |
|
142 |
=item B<-dir> I<directory> |
143 |
|
144 |
Alternate directory where databases are located. Default is the |
145 |
directory specified during configuration of WAIT. |
146 |
|
147 |
=item B<-table> I<table name> |
148 |
|
149 |
Specify an alternate table name. Default is C<bibdb>. |
150 |
|
151 |
=head1 AUTHOR |
152 |
|
153 |
Ulrich Pfeifer E<lt>F<pfeifer@wait.de>E<gt> |
154 |
|