1 |
ulpfr |
55 |
#!/usr/bin/perl -w |
2 |
|
|
# -*- Mode: Perl -*- |
3 |
|
|
# $Basename$ |
4 |
laperla |
67 |
# $Revision: 1.5 $ |
5 |
ulpfr |
55 |
# Author : Ulrich Pfeifer |
6 |
|
|
# Created On : Mon Dec 31 13:57:11 2001 |
7 |
|
|
# Last Modified By: Ulrich Pfeifer |
8 |
ulpfr |
62 |
# Last Modified On: Fri Jan 4 15:59:20 2002 |
9 |
ulpfr |
55 |
# 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 |
laperla |
67 |
my %OPT = ( |
30 |
ulpfr |
55 |
database => 'DB', |
31 |
|
|
dir => $WAIT::Config->{WAIT_home} || '/tmp', |
32 |
|
|
table => 'ora', |
33 |
|
|
); |
34 |
|
|
|
35 |
|
|
GetOptions(\%OPT, |
36 |
|
|
'database=s', |
37 |
|
|
'dir=s', |
38 |
|
|
'table=s', |
39 |
|
|
) || die "Usage: ...\n"; |
40 |
|
|
|
41 |
|
|
my $db; |
42 |
|
|
unless (-d "$OPT{dir}/$OPT{database}") { |
43 |
|
|
$db = WAIT::Database->create(name => $OPT{database}, |
44 |
|
|
'directory' => $OPT{dir}) |
45 |
|
|
or die "Could not open database $OPT{database}: $@\n"; |
46 |
|
|
} |
47 |
|
|
else { |
48 |
|
|
$db = WAIT::Database->open(name => $OPT{database}, |
49 |
|
|
'directory' => $OPT{dir}) |
50 |
|
|
or die "Could not open table $OPT{table}: $@\n"; |
51 |
|
|
} |
52 |
|
|
|
53 |
|
|
my $layout = new WAIT::Parse::Ora; |
54 |
|
|
|
55 |
|
|
my $stem = ['isotr', 'isolc', 'split2', 'stop', 'Stem']; |
56 |
|
|
my $text = [{ |
57 |
|
|
'prefix' => ['isotr', 'isolc'], |
58 |
|
|
'intervall' => ['isotr', 'isolc'], |
59 |
|
|
}, |
60 |
|
|
'isotr', 'isolc', 'split2', 'stop']; |
61 |
|
|
my $sound = ['isotr', 'isolc', 'split2', 'Soundex'],; |
62 |
|
|
|
63 |
|
|
my $cwd = cwd; |
64 |
|
|
|
65 |
|
|
my %D; |
66 |
|
|
my $access = tie %D, 'WAIT::Document::Ora', @ARGV, |
67 |
|
|
or die "Couldn't tie to file: $!\n"; |
68 |
|
|
|
69 |
|
|
my $tb = $db->create_table(name => $OPT{table}, |
70 |
ulpfr |
62 |
attr => ['author', 'isbn', 'title', |
71 |
ulpfr |
55 |
'headline', 'docid'], |
72 |
|
|
layout => $layout, |
73 |
|
|
access => $access, |
74 |
|
|
invindex => |
75 |
|
|
[ |
76 |
|
|
'title' => $stem, |
77 |
ulpfr |
59 |
'about' => $stem, |
78 |
ulpfr |
55 |
'text' => $text, |
79 |
|
|
'author' => $text, |
80 |
laperla |
65 |
'colophon' => $text, |
81 |
ulpfr |
62 |
'author' => $sound, |
82 |
|
|
'isbn' => $text, |
83 |
ulpfr |
55 |
] |
84 |
|
|
); |
85 |
|
|
die "Couldn't create table $OPT{table}: $@\n" unless $tb; |
86 |
|
|
|
87 |
|
|
my ($did, $value); |
88 |
|
|
while (($did, $value) = each %D) { |
89 |
|
|
my $record = $layout->split($value); |
90 |
|
|
my $headline = $record->{title}; |
91 |
|
|
$headline =~ s/\s+/ /sg; |
92 |
ulpfr |
62 |
printf "%15s %s\n", $record->{isbn}, substr($headline,0,60); |
93 |
ulpfr |
55 |
$tb->insert('docid' => $did, |
94 |
|
|
headline => $headline, |
95 |
|
|
%{$record}); |
96 |
|
|
} |
97 |
|
|
$tb->set(top=>1); |
98 |
|
|
$tb->close(); |
99 |
|
|
$db->close(); |
100 |
|
|
|
101 |
|
|
$WAIT::Config = $WAIT::Config; # make perl -w happy |
102 |
|
|
|
103 |
|
|
|
104 |
|
|
__END__ |
105 |
|
|
## ################################################################### |
106 |
|
|
## pod |
107 |
|
|
## ################################################################### |
108 |
|
|
|
109 |
|
|
=head1 NAME |
110 |
|
|
|
111 |
|
|
index_ora - generate an WAIT index for O'Reilly catalog |
112 |
|
|
|
113 |
|
|
=head1 SYNOPSIS |
114 |
|
|
|
115 |
|
|
B<index_ora> |
116 |
|
|
[B<-database> I<dbname>] |
117 |
|
|
[B<-dir> I<directory>] |
118 |
|
|
[B<-table> I<table name>] |
119 |
|
|
I<directory> |
120 |
|
|
|
121 |
|
|
=head1 DESCRIPTION |
122 |
|
|
|
123 |
|
|
=head1 OPTIONS |
124 |
|
|
|
125 |
|
|
=over 5 |
126 |
|
|
|
127 |
|
|
=item B<-database> I<dbname> |
128 |
|
|
|
129 |
|
|
Specify database name. Default is F<DB>. |
130 |
|
|
|
131 |
|
|
=item B<-dir> I<directory> |
132 |
|
|
|
133 |
|
|
Alternate directory where databases are located. Default is the |
134 |
|
|
directory specified during configuration of WAIT. |
135 |
|
|
|
136 |
|
|
=item B<-table> I<table name> |
137 |
|
|
|
138 |
|
|
Specify an alternate table name. Default is C<ora>. |
139 |
|
|
|
140 |
|
|
=head1 AUTHOR |
141 |
|
|
|
142 |
|
|
Ulrich Pfeifer E<lt>F<pfeifer@wait.de>E<gt> |
143 |
|
|
|