CPAN/script/bibdb.PL

#!/bin/sh -- # -*- perl -*- -w
eval 'exec perl -S $0 "$@"'
  if 0;

use strict;

use Config;
use File::Basename qw(fileparse);

my($file, $path) = fileparse($0);
$file =~ s!\.PL$!!i;
chdir($path)         or die "Couldn't chdir to `$path': $!\n";

print "Extracting $file\n";

open(OUT, "> $file") or die "Couldn't create `$file': $!\n";
print OUT "$Config{'startperl'} -w\n";
while (<DATA>) {
  print OUT
}
close(OUT)           or die "Couldn't close `$file': $!\n";

chmod(0755, $file)   or die "Couldn't chmod 744 on `$file': $!\n";

exec("$Config{'eunicefix'} $file") if $Config{'eunicefix'} ne ':';

__END__
######################### -*- Mode: Perl -*- #########################
##
## $Basename: bibdb.PL $
## $Revision: 1.4 $
##
## 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 befor indexing. Default is B<off>.

=item B<-database> I<dbname>

Specify database name. Default is F<DB>.

=item B<-dir> I<directory>

Alternate directory were 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-dortumund.de>E<gt>
1	ulpfr	10	#!/bin/sh -- # -- perl -- -w
2			eval 'exec perl -S $0 "$@"'
3			if 0;
4
5			use strict;
6
7			use Config;
8			use File::Basename qw(fileparse);
9
10			my($file, $path) = fileparse($0);
11			$file =~ s!\.PL$!!i;
12			chdir($path) or die "Couldn't chdir to `$path': $!\n";
13
14			print "Extracting $file\n";
15
16			open(OUT, "> $file") or die "Couldn't create `$file': $!\n";
17			print OUT "$Config{'startperl'} -w\n";
18			while (<DATA>) {
19			print OUT
20			}
21			close(OUT) or die "Couldn't close `$file': $!\n";
22
23			chmod(0755, $file) or die "Couldn't chmod 744 on `$file': $!\n";
24
25			exec("$Config{'eunicefix'} $file") if $Config{'eunicefix'} ne ':';
26
27			__END__
28			######################### -- Mode: Perl -- #########################
29			##
30			## $Basename: bibdb.PL $
31			## $Revision: 1.4 $
32			##
33			## Author : Ulrich Pfeifer
34			## Created On : Mon Sep 2 12:57:12 1996
35			##
36			## Last Modified By : Ulrich Pfeifer
37			## Last Modified On : Sun Nov 22 18:44:36 1998
38			##
39			## Copyright (c) 1996-1997, Ulrich Pfeifer
40			##
41			##
42			######################################################################
43
44			eval 'exec perl -S $0 "$@"'
45			if 0;
46
47
48			use strict;
49
50			use File::Path;
51			use DB_File;
52			use Getopt::Long;
53			use Cwd;
54
55			require WAIT::Config;
56			require WAIT::Database;
57			require WAIT::Parse::Bibdb;
58			require WAIT::Document::Split;
59			require WAIT::InvertedIndex;
60
61
62			$DB_BTREE->{'cachesize'} = 200_000 ;
63
64			my %OPT = (clean => 0,
65			database => 'DB',
66			dir => $WAIT::Config->{WAIT_home} \|\| '/tmp',
67			table => 'bibdb',
68			);
69
70			GetOptions(\%OPT,
71			'clean!',
72			'database=s',
73			'dir=s',
74			'table=s',
75			) \|\| die "Usage: ...\n";
76
77			if ($OPT{clean} and -d "$OPT{dir}/$OPT{database}") {
78			my $tmp = WAIT::Database->open(name => $OPT{database},
79			'directory' => $OPT{dir})
80			or die "Could not open table $OPT{table}: $@\n";
81			my $tbl = $tmp->table(name => $OPT{table});
82			$tbl->drop if $tbl;
83			rmtree("$OPT{dir}/$OPT{database}/$OPT{table}", 1, 1)
84			if -d "$OPT{dir}/$OPT{database}/$OPT{table}";
85			$tmp->close;
86			}
87
88			my $db;
89			unless (-d "$OPT{dir}/$OPT{database}") {
90			$db = WAIT::Database->create(name => $OPT{database},
91			'directory' => $OPT{dir})
92			or die "Could not open database $OPT{database}: $@\n";
93			}
94			else {
95			$db = WAIT::Database->open(name => $OPT{database},
96			'directory' => $OPT{dir})
97			or die "Could not open table $OPT{table}: $@\n";
98			}
99
100			my $layout = new WAIT::Parse::Bibdb;
101
102			my $stem = ['detex', 'isotr', 'isolc', 'split2', 'stop', 'Stem'];
103			my $text = [{
104			'prefix' => ['detex', 'isotr', 'isolc'],
105			'intervall' => ['detex', 'isotr', 'isolc'],
106			},
107			'detex', 'isotr', 'isolc', 'split2', 'stop'];
108			my $sound = ['detex', 'isotr', 'isolc', 'split2', 'Soundex'],;
109
110			my $cwd = cwd;
111			my($file) = grep -e $_, @ARGV, "$cwd/t/test.ste", '/usr/local/ls6/tex/bib/bibdb.ste';
112
113			my %D;
114			my $access = tie %D, 'WAIT::Document::Split', 'sep', '\f', $file
115			or die "Couldn't tie to $file: $!\n";
116
117			my $tb = $db->create_table(name => $OPT{table},
118			attr => ['docid', 'headline'],
119			layout => $layout,
120			access => $access,
121			invindex =>
122			[
123			'ti' => $stem, 'ti' => $text,
124			'jt' => $stem, 'jt' => $text,
125			'bt' => $stem, 'bt' => $text,
126			'sd' => $stem, 'sd' => $text,
127			'ft' => $stem, 'ft' => $text,
128			'ab' => $stem,
129			'au' => $text, 'au' => $sound,
130			'pn' => $text, 'pn' => $sound,
131			'cc' => [{'prefix' => ['cctr', 'isolc']},
132			'cctr', 'isolc', 'split3'],
133			'ed' => [{'intervall' => 1}, 'split6'],
134			'py' => [{'intervall' => 1}, 'split4'],
135			]
136			);
137			die "Couldn't create table $OPT{table}: $@\n" unless $tb;
138
139			my ($did, $value);
140			while (($did, $value) = each %D) {
141			my $record = $layout->split($value);
142			my $headline = sprintf("%-20s %s", ($record->{ck}\|\|''), $record->{ti});
143			printf "%s\n", substr($headline,0,80);
144			$tb->insert('docid' => $did,
145			headline => $headline,
146			%{$record});
147			}
148
149			$db->close();
150
151			$WAIT::Config = $WAIT::Config; # make perl -w happy
152
153
154			__END__
155			## ###################################################################
156			## pod
157			## ###################################################################
158
159			=head1 NAME
160
161			bibdb - generate an WAIT index for bibdb records
162
163			=head1 SYNOPSIS
164
165			B<bibdb>
166			[B<-clean>] [B<-noclean>]
167			[B<-database> I<dbname>]
168			[B<-dir> I<directory>]
169			[B<-table> I<table name>]
170
171			=head1 DESCRIPTION
172
173			Either indexes F<$WAIT/t/test.ste> (if called from directory F<$WAIT>)
174			or F</usr/local/ls6/tex/bib/bibdb.ste>.
175
176			=head1 OPTIONS
177
178			=over 5
179
180			=item B<-clean> / B<-noclean>
181
182			Clean the table befor indexing. Default is B<off>.
183
184			=item B<-database> I<dbname>
185
186			Specify database name. Default is F<DB>.
187
188			=item B<-dir> I<directory>
189
190			Alternate directory were databases are located. Default is the
191			directory specified during configuration of WAIT.
192
193			=item B<-table> I<table name>
194
195			Specify an alternate table name. Default is C<bibdb>.
196
197			=head1 AUTHOR
198
199			Ulrich Pfeifer E<lt>F<pfeifer@ls6.informatik.uni-dortumund.de>E<gt>