Annotation of /trunk/script/bibdb

#!/usr/local/perl5.005_56.Mar06/bin/perl -w
######################### -*- Mode: Perl -*- #########################
##
## $Basename: bibdb $
## $Revision: 1.6 $
##
## Author           : Ulrich Pfeifer
## Created On       : Mon Sep  2 12:57:12 1996
##
## Last Modified By : Ulrich Pfeifer
## Last Modified On : Sun Nov 22 18:44:36 1998
## 
## Copyright (c) 1996-1997, Ulrich Pfeifer
## 
## 
######################################################################

eval 'exec perl -S $0 "$@"'
  if 0;


use strict;

use File::Path;
use DB_File;
use Getopt::Long;
use Cwd;

require WAIT::Config;
require WAIT::Database;
require WAIT::Parse::Bibdb;
require WAIT::Document::Split;
require WAIT::InvertedIndex;


$DB_BTREE->{'cachesize'} = 200_000 ;

my %OPT = (clean    => 0,
           database => 'DB',
           dir      => $WAIT::Config->{WAIT_home} || '/tmp',
           table    => 'bibdb',
          );

GetOptions(\%OPT,
           'clean!',
           'database=s',
           'dir=s',
           'table=s',
          ) || die "Usage: ...\n";

if ($OPT{clean} and -d "$OPT{dir}/$OPT{database}") {
  my $tmp = WAIT::Database->open(name        => $OPT{database},
                                 'directory' => $OPT{dir})
    or die "Could not open table $OPT{table}: $@\n";
  my $tbl = $tmp->table(name => $OPT{table});
  $tbl->drop if $tbl;
  rmtree("$OPT{dir}/$OPT{database}/$OPT{table}", 1, 1)
    if -d "$OPT{dir}/$OPT{database}/$OPT{table}";
  $tmp->close;
}

my $db;
unless (-d "$OPT{dir}/$OPT{database}") {
  $db = WAIT::Database->create(name       => $OPT{database},
                              'directory' => $OPT{dir})
    or die "Could not open database $OPT{database}: $@\n";
}
else {
  $db = WAIT::Database->open(name        => $OPT{database},
                             'directory' => $OPT{dir})
    or die "Could not open table $OPT{table}: $@\n";
}

my $layout = new WAIT::Parse::Bibdb;

my $stem  = ['detex', 'isotr', 'isolc', 'split2', 'stop', 'Stem'];
my $text  = [{
              'prefix'    => ['detex', 'isotr', 'isolc'],
              'intervall' => ['detex', 'isotr', 'isolc'],
             },
             'detex', 'isotr', 'isolc', 'split2', 'stop'];
my $sound = ['detex', 'isotr', 'isolc', 'split2', 'Soundex'],;

my $cwd = cwd;
my($file) = grep -e $_, @ARGV, "$cwd/t/test.ste", '/usr/local/ls6/tex/bib/bibdb.ste';

my %D;
my $access = tie %D, 'WAIT::Document::Split', 'sep', '\f', $file
  or die "Couldn't tie to $file: $!\n";

my $tb = $db->create_table(name     => $OPT{table},
                           attr     => ['docid', 'headline'],
                           layout   => $layout,
                           access   => $access,
                           invindex =>
                           [
                            'ti' => $stem, 'ti' => $text,
                            'jt' => $stem, 'jt' => $text,
                            'bt' => $stem, 'bt' => $text,
                            'sd' => $stem, 'sd' => $text,
                            'ft' => $stem, 'ft' => $text,
                            'ab' => $stem,
                            'au' => $text, 'au' => $sound,
                            'pn' => $text, 'pn' => $sound,
                            'cc' => [{'prefix' => ['cctr',  'isolc']},
                                     'cctr',  'isolc', 'split3'],
                            'ed' => [{'intervall' => 1}, 'split6'],
                            'py' => [{'intervall' => 1}, 'split4'],
                           ]
                          );
die "Couldn't create table $OPT{table}: $@\n" unless $tb;

my ($did, $value);
while (($did, $value) = each %D) {
  my $record = $layout->split($value);
  my $headline = sprintf("%-20s %s", ($record->{ck}||''), $record->{ti});
  printf "%s\n", substr($headline,0,80);
  $tb->insert('docid'  => $did,
              headline => $headline,
              %{$record});
}

$db->close();

$WAIT::Config = $WAIT::Config; # make perl -w happy


__END__
## ###################################################################
## pod
## ###################################################################

=head1 NAME

bibdb - generate an WAIT index for bibdb records

=head1 SYNOPSIS

B<bibdb>
[B<-clean>] [B<-noclean>]
[B<-database> I<dbname>]
[B<-dir> I<directory>]
[B<-table> I<table name>]

=head1 DESCRIPTION

Either indexes F<$WAIT/t/test.ste> (if called from directory F<$WAIT>)
or F</usr/local/ls6/tex/bib/bibdb.ste>.

=head1 OPTIONS

=over 5

=item B<-clean> / B<-noclean>

Clean the table before indexing. Default is B<off>.

=item B<-database> I<dbname>

Specify database name. Default is F<DB>.

=item B<-dir> I<directory>

Alternate directory where databases are located. Default is the
directory specified during configuration of WAIT.

=item B<-table> I<table name>

Specify an alternate table name. Default is C<bibdb>.

=head1 AUTHOR

Ulrich Pfeifer E<lt>F<pfeifer@ls6.informatik.uni-dortmund.de>E<gt>
1	ulpfr	10	#!/usr/local/perl5.005_56.Mar06/bin/perl -w
2			######################### -- Mode: Perl -- #########################
3			##
4			## $Basename: bibdb $
5			## $Revision: 1.6 $
6			##
7			## Author : Ulrich Pfeifer
8			## Created On : Mon Sep 2 12:57:12 1996
9			##
10			## Last Modified By : Ulrich Pfeifer
11			## Last Modified On : Sun Nov 22 18:44:36 1998
12			##
13			## Copyright (c) 1996-1997, Ulrich Pfeifer
14			##
15			##
16			######################################################################
17
18			eval 'exec perl -S $0 "$@"'
19			if 0;
20
21
22			use strict;
23
24			use File::Path;
25			use DB_File;
26			use Getopt::Long;
27			use Cwd;
28
29			require WAIT::Config;
30			require WAIT::Database;
31			require WAIT::Parse::Bibdb;
32			require WAIT::Document::Split;
33			require WAIT::InvertedIndex;
34
35
36			$DB_BTREE->{'cachesize'} = 200_000 ;
37
38			my %OPT = (clean => 0,
39			database => 'DB',
40			dir => $WAIT::Config->{WAIT_home} \|\| '/tmp',
41			table => 'bibdb',
42			);
43
44			GetOptions(\%OPT,
45			'clean!',
46			'database=s',
47			'dir=s',
48			'table=s',
49			) \|\| die "Usage: ...\n";
50
51			if ($OPT{clean} and -d "$OPT{dir}/$OPT{database}") {
52			my $tmp = WAIT::Database->open(name => $OPT{database},
53			'directory' => $OPT{dir})
54			or die "Could not open table $OPT{table}: $@\n";
55			my $tbl = $tmp->table(name => $OPT{table});
56			$tbl->drop if $tbl;
57			rmtree("$OPT{dir}/$OPT{database}/$OPT{table}", 1, 1)
58			if -d "$OPT{dir}/$OPT{database}/$OPT{table}";
59			$tmp->close;
60			}
61
62			my $db;
63			unless (-d "$OPT{dir}/$OPT{database}") {
64			$db = WAIT::Database->create(name => $OPT{database},
65			'directory' => $OPT{dir})
66			or die "Could not open database $OPT{database}: $@\n";
67			}
68			else {
69			$db = WAIT::Database->open(name => $OPT{database},
70			'directory' => $OPT{dir})
71			or die "Could not open table $OPT{table}: $@\n";
72			}
73
74			my $layout = new WAIT::Parse::Bibdb;
75
76			my $stem = ['detex', 'isotr', 'isolc', 'split2', 'stop', 'Stem'];
77			my $text = [{
78			'prefix' => ['detex', 'isotr', 'isolc'],
79			'intervall' => ['detex', 'isotr', 'isolc'],
80			},
81			'detex', 'isotr', 'isolc', 'split2', 'stop'];
82			my $sound = ['detex', 'isotr', 'isolc', 'split2', 'Soundex'],;
83
84			my $cwd = cwd;
85			my($file) = grep -e $_, @ARGV, "$cwd/t/test.ste", '/usr/local/ls6/tex/bib/bibdb.ste';
86
87			my %D;
88			my $access = tie %D, 'WAIT::Document::Split', 'sep', '\f', $file
89			or die "Couldn't tie to $file: $!\n";
90
91			my $tb = $db->create_table(name => $OPT{table},
92			attr => ['docid', 'headline'],
93			layout => $layout,
94			access => $access,
95			invindex =>
96			[
97			'ti' => $stem, 'ti' => $text,
98			'jt' => $stem, 'jt' => $text,
99			'bt' => $stem, 'bt' => $text,
100			'sd' => $stem, 'sd' => $text,
101			'ft' => $stem, 'ft' => $text,
102			'ab' => $stem,
103			'au' => $text, 'au' => $sound,
104			'pn' => $text, 'pn' => $sound,
105			'cc' => [{'prefix' => ['cctr', 'isolc']},
106			'cctr', 'isolc', 'split3'],
107			'ed' => [{'intervall' => 1}, 'split6'],
108			'py' => [{'intervall' => 1}, 'split4'],
109			]
110			);
111			die "Couldn't create table $OPT{table}: $@\n" unless $tb;
112
113			my ($did, $value);
114			while (($did, $value) = each %D) {
115			my $record = $layout->split($value);
116			my $headline = sprintf("%-20s %s", ($record->{ck}\|\|''), $record->{ti});
117			printf "%s\n", substr($headline,0,80);
118			$tb->insert('docid' => $did,
119			headline => $headline,
120			%{$record});
121			}
122
123			$db->close();
124
125			$WAIT::Config = $WAIT::Config; # make perl -w happy
126
127
128			__END__
129			## ###################################################################
130			## pod
131			## ###################################################################
132
133			=head1 NAME
134
135			bibdb - generate an WAIT index for bibdb records
136
137			=head1 SYNOPSIS
138
139			B<bibdb>
140			[B<-clean>] [B<-noclean>]
141			[B<-database> I<dbname>]
142			[B<-dir> I<directory>]
143			[B<-table> I<table name>]
144
145			=head1 DESCRIPTION
146
147			Either indexes F<$WAIT/t/test.ste> (if called from directory F<$WAIT>)
148			or F</usr/local/ls6/tex/bib/bibdb.ste>.
149
150			=head1 OPTIONS
151
152			=over 5
153
154			=item B<-clean> / B<-noclean>
155
156			Clean the table before indexing. Default is B<off>.
157
158			=item B<-database> I<dbname>
159
160			Specify database name. Default is F<DB>.
161
162			=item B<-dir> I<directory>
163
164			Alternate directory where databases are located. Default is the
165			directory specified during configuration of WAIT.
166
167			=item B<-table> I<table name>
168
169			Specify an alternate table name. Default is C<bibdb>.
170
171			=head1 AUTHOR
172
173			Ulrich Pfeifer E<lt>F<pfeifer@ls6.informatik.uni-dortmund.de>E<gt>