Annotation of /trunk/script/bibdb

#!/usr/bin/perl -w

######################### -*- Mode: Perl -*- #########################
##
## $Basename: bibdb $
## $Revision: 1.7 $
##
## Author           : Ulrich Pfeifer
## Created On       : Mon Sep  2 12:57:12 1996
##
## Last Modified By : Ulrich Pfeifer
## Last Modified On : Fri Apr  7 13:45:36 2000
## 
## Copyright (c) 1996-1997, Ulrich Pfeifer
## 
## 
######################################################################

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

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>.

=back

=head1 AUTHOR

Ulrich Pfeifer E<lt>F<pfeifer@ls6.informatik.uni-dortmund.de>E<gt>
1	dpavlin	116	#!/usr/bin/perl -w
2
3	ulpfr	10	######################### -- Mode: Perl -- #########################
4			##
5			## $Basename: bibdb $
6	ulpfr	19	## $Revision: 1.7 $
7	ulpfr	10	##
8			## Author : Ulrich Pfeifer
9			## Created On : Mon Sep 2 12:57:12 1996
10			##
11			## Last Modified By : Ulrich Pfeifer
12	ulpfr	19	## Last Modified On : Fri Apr 7 13:45:36 2000
13	ulpfr	10	##
14			## Copyright (c) 1996-1997, Ulrich Pfeifer
15			##
16			##
17			######################################################################
18
19			use strict;
20			use File::Path;
21			use DB_File;
22			use Getopt::Long;
23			use Cwd;
24	dpavlin	116	use blib;
25	ulpfr	10
26			require WAIT::Config;
27			require WAIT::Database;
28			require WAIT::Parse::Bibdb;
29			require WAIT::Document::Split;
30			require WAIT::InvertedIndex;
31
32
33			$DB_BTREE->{'cachesize'} = 200_000 ;
34
35			my %OPT = (clean => 0,
36			database => 'DB',
37			dir => $WAIT::Config->{WAIT_home} \|\| '/tmp',
38			table => 'bibdb',
39			);
40
41			GetOptions(\%OPT,
42			'clean!',
43			'database=s',
44			'dir=s',
45			'table=s',
46			) \|\| die "Usage: ...\n";
47
48			if ($OPT{clean} and -d "$OPT{dir}/$OPT{database}") {
49			my $tmp = WAIT::Database->open(name => $OPT{database},
50			'directory' => $OPT{dir})
51			or die "Could not open table $OPT{table}: $@\n";
52			my $tbl = $tmp->table(name => $OPT{table});
53			$tbl->drop if $tbl;
54			rmtree("$OPT{dir}/$OPT{database}/$OPT{table}", 1, 1)
55			if -d "$OPT{dir}/$OPT{database}/$OPT{table}";
56			$tmp->close;
57			}
58
59			my $db;
60			unless (-d "$OPT{dir}/$OPT{database}") {
61			$db = WAIT::Database->create(name => $OPT{database},
62			'directory' => $OPT{dir})
63			or die "Could not open database $OPT{database}: $@\n";
64			}
65			else {
66			$db = WAIT::Database->open(name => $OPT{database},
67			'directory' => $OPT{dir})
68			or die "Could not open table $OPT{table}: $@\n";
69			}
70
71			my $layout = new WAIT::Parse::Bibdb;
72
73			my $stem = ['detex', 'isotr', 'isolc', 'split2', 'stop', 'Stem'];
74			my $text = [{
75			'prefix' => ['detex', 'isotr', 'isolc'],
76			'intervall' => ['detex', 'isotr', 'isolc'],
77			},
78			'detex', 'isotr', 'isolc', 'split2', 'stop'];
79			my $sound = ['detex', 'isotr', 'isolc', 'split2', 'Soundex'],;
80
81			my $cwd = cwd;
82			my($file) = grep -e $_, @ARGV, "$cwd/t/test.ste", '/usr/local/ls6/tex/bib/bibdb.ste';
83
84			my %D;
85			my $access = tie %D, 'WAIT::Document::Split', 'sep', '\f', $file
86			or die "Couldn't tie to $file: $!\n";
87
88			my $tb = $db->create_table(name => $OPT{table},
89			attr => ['docid', 'headline'],
90			layout => $layout,
91			access => $access,
92			invindex =>
93			[
94			'ti' => $stem, 'ti' => $text,
95			'jt' => $stem, 'jt' => $text,
96			'bt' => $stem, 'bt' => $text,
97			'sd' => $stem, 'sd' => $text,
98			'ft' => $stem, 'ft' => $text,
99			'ab' => $stem,
100			'au' => $text, 'au' => $sound,
101			'pn' => $text, 'pn' => $sound,
102			'cc' => [{'prefix' => ['cctr', 'isolc']},
103			'cctr', 'isolc', 'split3'],
104			'ed' => [{'intervall' => 1}, 'split6'],
105			'py' => [{'intervall' => 1}, 'split4'],
106			]
107			);
108			die "Couldn't create table $OPT{table}: $@\n" unless $tb;
109
110			my ($did, $value);
111			while (($did, $value) = each %D) {
112			my $record = $layout->split($value);
113			my $headline = sprintf("%-20s %s", ($record->{ck}\|\|''), $record->{ti});
114			printf "%s\n", substr($headline,0,80);
115			$tb->insert('docid' => $did,
116			headline => $headline,
117			%{$record});
118			}
119
120			$db->close();
121
122			$WAIT::Config = $WAIT::Config; # make perl -w happy
123
124
125			__END__
126			## ###################################################################
127			## pod
128			## ###################################################################
129
130			=head1 NAME
131
132			bibdb - generate an WAIT index for bibdb records
133
134			=head1 SYNOPSIS
135
136			B<bibdb>
137			[B<-clean>] [B<-noclean>]
138			[B<-database> I<dbname>]
139			[B<-dir> I<directory>]
140			[B<-table> I<table name>]
141
142			=head1 DESCRIPTION
143
144			Either indexes F<$WAIT/t/test.ste> (if called from directory F<$WAIT>)
145			or F</usr/local/ls6/tex/bib/bibdb.ste>.
146
147			=head1 OPTIONS
148
149			=over 5
150
151			=item B<-clean> / B<-noclean>
152
153			Clean the table before indexing. Default is B<off>.
154
155			=item B<-database> I<dbname>
156
157			Specify database name. Default is F<DB>.
158
159			=item B<-dir> I<directory>
160
161			Alternate directory where databases are located. Default is the
162			directory specified during configuration of WAIT.
163
164			=item B<-table> I<table name>
165
166			Specify an alternate table name. Default is C<bibdb>.
167
168	dpavlin	116	=back
169
170	ulpfr	10	=head1 AUTHOR
171
172			Ulrich Pfeifer E<lt>F<pfeifer@ls6.informatik.uni-dortmund.de>E<gt>