Annotation of /cvs-head/script/index_ora

#!/usr/bin/perl -w
#                              -*- Mode: Perl -*- 
# $Basename$
# $Revision: 1.11 $
# Author          : Ulrich Pfeifer
# Created On      : Mon Dec 31 13:57:11 2001
# Last Modified By: Ulrich Pfeifer
# Last Modified On: Fri Jan  4 15:59:20 2002
# Language        : CPerl
# 
# (C) Copyright 2001, Ulrich Pfeifer
# 

use 5.007;

use strict;
use Devel::Peek qw(Dump);

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

BEGIN {require WAIT::Config;}
use WAIT::Database;
use WAIT::Parse::Ora;
use WAIT::Document::Ora;
use WAIT::InvertedIndex;
use Data::Dumper;


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

my %OPT = (
           database => 'oreilly_de_catalog',
           dir      => '/usr/local/apache/data',
           table    => 'ora',
          );

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

my @localtime = localtime;
$localtime[5] += 1900;
$localtime[4]++;
my $jobid = sprintf "%04s-%02s-%02s_%02s:%02s_%d", @localtime[5,4,3,2,1], $$;
my $db = WAIT::Database->create(name      => "$OPT{database}-$jobid",
                                directory => $OPT{dir})
    or die "Could not create database $OPT{database}: $@\n";

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

use lib "/usr/local/apache/lib";
use oreilly_de_catalog::wait_filter;

my $stem  = ['OR_tr_20020124', 'OR_lc_20020125', 'split2', 'stop', 'Stem'];
my $text  = ['OR_tr_20020124', 'OR_lc_20020125', 'split2', 'stop'];
my $wplus = ['split2', 'OR_lc_20020125', 'OR_mixedonly_20020221'];
my $sound = ['OR_tr_20020124', 'OR_lc_20020125', 'split2', 'Soundex'];
my $trigr = ['OR_lc_20020125', 'OR_trigrams_20020125'];
# split6 is better than split13 or split10: it allows them to enter
# shorter sequences when searching.
my $isbn  = ['split6', 'OR_isbn_20020127'];

my $cwd = cwd;

my $droot = shift or die "Usage: $0 <options> Document-Rootdirectories";

my %D;
my $access = tie %D, 'WAIT::Document::Ora', $droot,
  or die "Couldn't tie to dir $droot: $!\n";

my $tb = $db->create_table(name     => $OPT{table},
                           attr     => ['author', 'isbn', 'title',
                                        'headline', 'docid'],
                           layout   => $layout,
                           access   => $access,
                           invindex =>
                           [
                            'aboutauthor'  => $text,
                            'aboutauthor'  => $wplus,
                            'abouttranslator'  => $text,
                            'abouttranslator'  => $wplus,
                            'abstract' => $text,
                            'abstract' => $wplus,
                            'author' => $text,
                            'colophon' => $text,
                            'colophon' => $wplus,
                            'desc'   => $text,
                            'desc'   => $wplus,
                            'inx'   => $text,
                            'inx'   => $wplus,
                            'isbn'   => $isbn,
                            'subtitle'  => $text,
                            'subtitle'  => $wplus,
                            'title'  => $text,
                            'title'  => $wplus,
                            'title_orig'  => $text,
                            'title_orig'  => $wplus,
                            'toc'   => $text,
                            'toc'   => $wplus,
                            'translator'  => $text,
                            'translator'  => $wplus,
                           ]
                          );
die "Couldn't create table $OPT{table}: $@\n" unless $tb;

my ($did, $value);
binmode STDOUT, ":utf8";
my $ALL;
my $traceALL = 0; # expensive
my $done = 0;
my $todo = keys %D;
my $lasttimeround = my $starttime = time;

DOC: while (($did, $value) = each %D) {
  my $record   = $layout->split($value);
  my $headline = $record->{title};
  $headline =~ s/\s+/ /sg;
  # printf "%15s %s\n", $record->{isbn}, substr($headline,0,60);
  printf "%15s %s...\n", $did, substr($headline,0,60);
  $tb->insert('docid'  => $did,
              headline => $headline,
              %{$record});
  $done++;
  my $spenttime = time - $starttime;
  my $averagetime = $spenttime/$done;
  my $left = $todo-$done;
  printf("%2d secs, %3d done, %3d left, %4d s done, %5.1f s avg, %4d s left\n",
               time - $lasttimeround,
               $done,
               $left,
               $spenttime,
               $averagetime,
               $left*$averagetime,
              );
  $lasttimeround = time;
  if ($traceALL) { # costs a lot when reaching the 100th file or so
    $ALL->{$did} = $record;
    open F, ">:utf8", "$OPT{dir}/$OPT{database}-$jobid/debug.dump" or die;
    print F Data::Dumper::Dumper($ALL);
    close F;
  }
}
undef $ALL;
$tb->set(top=>1);

my $tritb = $db->create_table(
                              name => "$OPT{table}_fallback",
                              attr => [qw(docid headline)], # name
                                                            # "headline"
                                                            # only for
                                                            # sman
                              invindex => [ headline => $trigr ],
                             );
my %dict;
for my $f ($tb->fields) {
  my(@idx) = @{$tb->table->{inverted}{$f} || []};
  for my $idx (@idx) {
    my $name = $idx->name;
    next if $name =~ /(_|\b)(mixedonly|Stem|Soundex)(\b|_)/;
              # irrelevant for alternatives
    my @keys = $idx->keys;
    @dict{@keys} = ();
  }
}
my @dictkeys = grep s/^p//, keys %dict;
my $maxdebug = 5;
for my $headline (@dictkeys) {
  if ($maxdebug && $headline =~ /[^\040-\177]/) {
    Dump $headline;
    $maxdebug--;
  }
  # printf "%s\n", substr($headline,0,60);
  $tritb->insert(docid => $headline, headline => $headline);
}
$tritb->set(top=>1);
$tritb->close;
$tb->close();
$db->close();

# Atomically relinking symlink: now we have a new database with a very
# long name "$OPT{database}-$jobid" (e.g.
# oreilly_de_catalog-2002-01-28_16:12_16467) and we want that database
# to be accessible with the oreilly_de_catalog name.

use File::Spec;
chdir $OPT{dir} or die;
my $dir    = "$OPT{database}-$jobid";
my $slwant = File::Spec->catdir($OPT{dir}, $OPT{database});
my $sltmp  = File::Spec->catdir($OPT{dir}, "$OPT{database}-$$");
unlink $sltmp; # may fail
symlink $dir, $sltmp or die "Couldn't symlink $dir, $sltmp: $!";
rename $sltmp, $slwant or die "Couldn't rename $sltmp, $slwant: $!";
warn "$slwant now points to $dir";
system("chmod 777 $slwant/*/read")==0 or die;

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


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

=head1 NAME

index_ora - generate an WAIT index for O'Reilly catalog

=head1 SYNOPSIS

B<index_ora>
[B<-database> I<dbname>]
[B<-dir> I<directory>]
[B<-table> I<table name>]
I<directory>

=head1 DESCRIPTION

=head1 OPTIONS

=over 5

=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<ora>.

=head1 AUTHOR

Ulrich Pfeifer E<lt>F<pfeifer@wait.de>E<gt>

1	ulpfr	55	#!/usr/bin/perl -w
2			# -- Mode: Perl --
3			# $Basename$
4	laperla	74	# $Revision: 1.11 $
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	laperla	73	# (C) Copyright 2001, Ulrich Pfeifer
12	ulpfr	55	#
13
14	laperla	69	use 5.007;
15
16	ulpfr	55	use strict;
17	laperla	71	use Devel::Peek qw(Dump);
18	laperla	69
19	ulpfr	55	use File::Path;
20			use DB_File;
21			use Getopt::Long;
22			use Cwd;
23
24	laperla	69	BEGIN {require WAIT::Config;}
25			use WAIT::Database;
26			use WAIT::Parse::Ora;
27			use WAIT::Document::Ora;
28			use WAIT::InvertedIndex;
29	laperla	74	use Data::Dumper;
30	ulpfr	55
31
32			$DB_BTREE->{'cachesize'} = 200_000 ;
33
34	laperla	67	my %OPT = (
35	laperla	73	database => 'oreilly_de_catalog',
36			dir => '/usr/local/apache/data',
37	ulpfr	55	table => 'ora',
38			);
39
40			GetOptions(\%OPT,
41			'database=s',
42			'dir=s',
43			'table=s',
44			) \|\| die "Usage: ...\n";
45
46	laperla	68	my @localtime = localtime;
47			$localtime[5] += 1900;
48			$localtime[4]++;
49			my $jobid = sprintf "%04s-%02s-%02s_%02s:%02s_%d", @localtime[5,4,3,2,1], $$;
50			my $db = WAIT::Database->create(name => "$OPT{database}-$jobid",
51			directory => $OPT{dir})
52			or die "Could not create database $OPT{database}: $@\n";
53	ulpfr	55
54			my $layout = new WAIT::Parse::Ora;
55
56	laperla	69	use lib "/usr/local/apache/lib";
57	laperla	74	use oreilly_de_catalog::wait_filter;
58	laperla	69
59	laperla	74	my $stem = ['OR_tr_20020124', 'OR_lc_20020125', 'split2', 'stop', 'Stem'];
60			my $text = ['OR_tr_20020124', 'OR_lc_20020125', 'split2', 'stop'];
61			my $wplus = ['split2', 'OR_lc_20020125', 'OR_mixedonly_20020221'];
62			my $sound = ['OR_tr_20020124', 'OR_lc_20020125', 'split2', 'Soundex'];
63			my $trigr = ['OR_lc_20020125', 'OR_trigrams_20020125'];
64	laperla	73	# split6 is better than split13 or split10: it allows them to enter
65			# shorter sequences when searching.
66	laperla	72	my $isbn = ['split6', 'OR_isbn_20020127'];
67	ulpfr	55
68			my $cwd = cwd;
69
70	laperla	73	my $droot = shift or die "Usage: $0 <options> Document-Rootdirectories";
71
72	ulpfr	55	my %D;
73	laperla	73	my $access = tie %D, 'WAIT::Document::Ora', $droot,
74			or die "Couldn't tie to dir $droot: $!\n";
75	ulpfr	55
76			my $tb = $db->create_table(name => $OPT{table},
77	ulpfr	62	attr => ['author', 'isbn', 'title',
78	ulpfr	55	'headline', 'docid'],
79			layout => $layout,
80			access => $access,
81			invindex =>
82			[
83	laperla	72	'aboutauthor' => $text,
84	laperla	73	'aboutauthor' => $wplus,
85	laperla	74	'abouttranslator' => $text,
86			'abouttranslator' => $wplus,
87	laperla	72	'abstract' => $text,
88	laperla	73	'abstract' => $wplus,
89	ulpfr	55	'author' => $text,
90	laperla	65	'colophon' => $text,
91	laperla	73	'colophon' => $wplus,
92	laperla	74	'desc' => $text,
93			'desc' => $wplus,
94			'inx' => $text,
95			'inx' => $wplus,
96	laperla	72	'isbn' => $isbn,
97	laperla	74	'subtitle' => $text,
98			'subtitle' => $wplus,
99			'title' => $text,
100			'title' => $wplus,
101			'title_orig' => $text,
102			'title_orig' => $wplus,
103			'toc' => $text,
104			'toc' => $wplus,
105			'translator' => $text,
106			'translator' => $wplus,
107	ulpfr	55	]
108			);
109			die "Couldn't create table $OPT{table}: $@\n" unless $tb;
110
111			my ($did, $value);
112	laperla	69	binmode STDOUT, ":utf8";
113	laperla	74	my $ALL;
114			my $traceALL = 0; # expensive
115			my $done = 0;
116			my $todo = keys %D;
117			my $lasttimeround = my $starttime = time;
118
119			DOC: while (($did, $value) = each %D) {
120	ulpfr	55	my $record = $layout->split($value);
121			my $headline = $record->{title};
122			$headline =~ s/\s+/ /sg;
123	laperla	74	# printf "%15s %s\n", $record->{isbn}, substr($headline,0,60);
124			printf "%15s %s...\n", $did, substr($headline,0,60);
125	ulpfr	55	$tb->insert('docid' => $did,
126			headline => $headline,
127			%{$record});
128	laperla	74	$done++;
129			my $spenttime = time - $starttime;
130			my $averagetime = $spenttime/$done;
131			my $left = $todo-$done;
132			printf("%2d secs, %3d done, %3d left, %4d s done, %5.1f s avg, %4d s left\n",
133			time - $lasttimeround,
134			$done,
135			$left,
136			$spenttime,
137			$averagetime,
138			$left*$averagetime,
139			);
140			$lasttimeround = time;
141			if ($traceALL) { # costs a lot when reaching the 100th file or so
142			$ALL->{$did} = $record;
143			open F, ">:utf8", "$OPT{dir}/$OPT{database}-$jobid/debug.dump" or die;
144			print F Data::Dumper::Dumper($ALL);
145			close F;
146			}
147	ulpfr	55	}
148	laperla	74	undef $ALL;
149	ulpfr	55	$tb->set(top=>1);
150	laperla	71
151			my $tritb = $db->create_table(
152			name => "$OPT{table}_fallback",
153	laperla	72	attr => [qw(docid headline)], # name
154			# "headline"
155			# only for
156			# sman
157	laperla	71	invindex => [ headline => $trigr ],
158			);
159			my %dict;
160			for my $f ($tb->fields) {
161			my(@idx) = @{$tb->table->{inverted}{$f} \|\| []};
162			for my $idx (@idx) {
163	laperla	72	my $name = $idx->name;
164	laperla	73	next if $name =~ /(_\|\b)(mixedonly\|Stem\|Soundex)(\b\|_)/;
165			# irrelevant for alternatives
166	laperla	71	my @keys = $idx->keys;
167			@dict{@keys} = ();
168			}
169			}
170			my @dictkeys = grep s/^p//, keys %dict;
171			my $maxdebug = 5;
172			for my $headline (@dictkeys) {
173			if ($maxdebug && $headline =~ /[^\040-\177]/) {
174			Dump $headline;
175			$maxdebug--;
176			}
177	laperla	72	# printf "%s\n", substr($headline,0,60);
178	laperla	71	$tritb->insert(docid => $headline, headline => $headline);
179			}
180			$tritb->set(top=>1);
181			$tritb->close;
182	ulpfr	55	$tb->close();
183			$db->close();
184
185	laperla	73	# Atomically relinking symlink: now we have a new database with a very
186	laperla	74	# long name "$OPT{database}-$jobid" (e.g.
187			# oreilly_de_catalog-2002-01-28_16:12_16467) and we want that database
188			# to be accessible with the oreilly_de_catalog name.
189	laperla	68
190			use File::Spec;
191	laperla	74	chdir $OPT{dir} or die;
192	laperla	73	my $dir = "$OPT{database}-$jobid";
193			my $slwant = File::Spec->catdir($OPT{dir}, $OPT{database});
194			my $sltmp = File::Spec->catdir($OPT{dir}, "$OPT{database}-$$");
195			unlink $sltmp; # may fail
196			symlink $dir, $sltmp or die "Couldn't symlink $dir, $sltmp: $!";
197			rename $sltmp, $slwant or die "Couldn't rename $sltmp, $slwant: $!";
198			warn "$slwant now points to $dir";
199			system("chmod 777 $slwant/*/read")==0 or die;
200	laperla	68
201	ulpfr	55	$WAIT::Config = $WAIT::Config; # make perl -w happy
202
203
204			__END__
205			## ###################################################################
206			## pod
207			## ###################################################################
208
209			=head1 NAME
210
211			index_ora - generate an WAIT index for O'Reilly catalog
212
213			=head1 SYNOPSIS
214
215			B<index_ora>
216			[B<-database> I<dbname>]
217			[B<-dir> I<directory>]
218			[B<-table> I<table name>]
219			I<directory>
220
221			=head1 DESCRIPTION
222
223			=head1 OPTIONS
224
225			=over 5
226
227			=item B<-database> I<dbname>
228
229			Specify database name. Default is F<DB>.
230
231			=item B<-dir> I<directory>
232
233			Alternate directory where databases are located. Default is the
234			directory specified during configuration of WAIT.
235
236			=item B<-table> I<table name>
237
238			Specify an alternate table name. Default is C<ora>.
239
240			=head1 AUTHOR
241
242			Ulrich Pfeifer E<lt>F<pfeifer@wait.de>E<gt>
243