1 |
#!/usr/bin/perl -w |
2 |
# -*- Mode: Perl -*- |
3 |
# $Basename$ |
4 |
# $Revision: 1.6 $ |
5 |
# Author : Ulrich Pfeifer |
6 |
# Created On : Mon Dec 31 13:57:11 2001 |
7 |
# Last Modified By: Ulrich Pfeifer |
8 |
# Last Modified On: Fri Jan 4 15:59:20 2002 |
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 = ( |
30 |
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 @localtime = localtime; |
42 |
$localtime[5] += 1900; |
43 |
$localtime[4]++; |
44 |
my $jobid = sprintf "%04s-%02s-%02s_%02s:%02s_%d", @localtime[5,4,3,2,1], $$; |
45 |
my $db = WAIT::Database->create(name => "$OPT{database}-$jobid", |
46 |
directory => $OPT{dir}) |
47 |
or die "Could not create database $OPT{database}: $@\n"; |
48 |
|
49 |
my $layout = new WAIT::Parse::Ora; |
50 |
|
51 |
my $stem = ['isotr', 'isolc', 'split2', 'stop', 'Stem']; |
52 |
my $text = [{ |
53 |
'prefix' => ['isotr', 'isolc'], |
54 |
'intervall' => ['isotr', 'isolc'], |
55 |
}, |
56 |
'isotr', 'isolc', 'split2', 'stop']; |
57 |
my $sound = ['isotr', 'isolc', 'split2', 'Soundex'],; |
58 |
|
59 |
my $cwd = cwd; |
60 |
|
61 |
my %D; |
62 |
my $access = tie %D, 'WAIT::Document::Ora', @ARGV, |
63 |
or die "Couldn't tie to file: $!\n"; |
64 |
|
65 |
my $tb = $db->create_table(name => $OPT{table}, |
66 |
attr => ['author', 'isbn', 'title', |
67 |
'headline', 'docid'], |
68 |
layout => $layout, |
69 |
access => $access, |
70 |
invindex => |
71 |
[ |
72 |
'title' => $stem, |
73 |
'about' => $stem, |
74 |
'text' => $text, |
75 |
'author' => $text, |
76 |
'colophon' => $text, |
77 |
'author' => $sound, |
78 |
'isbn' => $text, |
79 |
] |
80 |
); |
81 |
die "Couldn't create table $OPT{table}: $@\n" unless $tb; |
82 |
|
83 |
my ($did, $value); |
84 |
while (($did, $value) = each %D) { |
85 |
my $record = $layout->split($value); |
86 |
my $headline = $record->{title}; |
87 |
$headline =~ s/\s+/ /sg; |
88 |
printf "%15s %s\n", $record->{isbn}, substr($headline,0,60); |
89 |
$tb->insert('docid' => $did, |
90 |
headline => $headline, |
91 |
%{$record}); |
92 |
} |
93 |
$tb->set(top=>1); |
94 |
$tb->close(); |
95 |
$db->close(); |
96 |
|
97 |
# Now we have a new database with a very long name and we want that |
98 |
# database to be accessible with the $OPT{database} name |
99 |
|
100 |
use File::Spec; |
101 |
my $long_dir = "$OPT{database}-$jobid"; |
102 |
my $want_dir = File::Spec->catdir($OPT{dir}, $OPT{database}); |
103 |
my $prel_slink = File::Spec->catdir($OPT{dir}, "$OPT{database}-$$"); |
104 |
unlink $prel_slink; # may fail |
105 |
symlink $long_dir, $prel_slink or die "Could not symlink $long_dir, $prel_slink: $!"; |
106 |
rename $prel_slink, $want_dir or die "Could not rename $prel_slink, $want_dir: $!"; |
107 |
|
108 |
$WAIT::Config = $WAIT::Config; # make perl -w happy |
109 |
|
110 |
|
111 |
__END__ |
112 |
## ################################################################### |
113 |
## pod |
114 |
## ################################################################### |
115 |
|
116 |
=head1 NAME |
117 |
|
118 |
index_ora - generate an WAIT index for O'Reilly catalog |
119 |
|
120 |
=head1 SYNOPSIS |
121 |
|
122 |
B<index_ora> |
123 |
[B<-database> I<dbname>] |
124 |
[B<-dir> I<directory>] |
125 |
[B<-table> I<table name>] |
126 |
I<directory> |
127 |
|
128 |
=head1 DESCRIPTION |
129 |
|
130 |
=head1 OPTIONS |
131 |
|
132 |
=over 5 |
133 |
|
134 |
=item B<-database> I<dbname> |
135 |
|
136 |
Specify database name. Default is F<DB>. |
137 |
|
138 |
=item B<-dir> I<directory> |
139 |
|
140 |
Alternate directory where databases are located. Default is the |
141 |
directory specified during configuration of WAIT. |
142 |
|
143 |
=item B<-table> I<table name> |
144 |
|
145 |
Specify an alternate table name. Default is C<ora>. |
146 |
|
147 |
=head1 AUTHOR |
148 |
|
149 |
Ulrich Pfeifer E<lt>F<pfeifer@wait.de>E<gt> |
150 |
|