| 1 |
#!/usr/bin/perl -w |
| 2 |
# -*- Mode: Perl -*- |
| 3 |
# $Basename$ |
| 4 |
# $Revision: 1.11 $ |
| 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, Ulrich Pfeifer |
| 12 |
# |
| 13 |
|
| 14 |
use 5.007; |
| 15 |
|
| 16 |
use strict; |
| 17 |
use Devel::Peek qw(Dump); |
| 18 |
|
| 19 |
use File::Path; |
| 20 |
use DB_File; |
| 21 |
use Getopt::Long; |
| 22 |
use Cwd; |
| 23 |
|
| 24 |
BEGIN {require WAIT::Config;} |
| 25 |
use WAIT::Database; |
| 26 |
use WAIT::Parse::Ora; |
| 27 |
use WAIT::Document::Ora; |
| 28 |
use WAIT::InvertedIndex; |
| 29 |
use Data::Dumper; |
| 30 |
|
| 31 |
|
| 32 |
$DB_BTREE->{'cachesize'} = 200_000 ; |
| 33 |
|
| 34 |
my %OPT = ( |
| 35 |
database => 'oreilly_de_catalog', |
| 36 |
dir => '/usr/local/apache/data', |
| 37 |
table => 'ora', |
| 38 |
); |
| 39 |
|
| 40 |
GetOptions(\%OPT, |
| 41 |
'database=s', |
| 42 |
'dir=s', |
| 43 |
'table=s', |
| 44 |
) || die "Usage: ...\n"; |
| 45 |
|
| 46 |
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 |
|
| 54 |
my $layout = new WAIT::Parse::Ora; |
| 55 |
|
| 56 |
use lib "/usr/local/apache/lib"; |
| 57 |
use oreilly_de_catalog::wait_filter; |
| 58 |
|
| 59 |
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 |
# split6 is better than split13 or split10: it allows them to enter |
| 65 |
# shorter sequences when searching. |
| 66 |
my $isbn = ['split6', 'OR_isbn_20020127']; |
| 67 |
|
| 68 |
my $cwd = cwd; |
| 69 |
|
| 70 |
my $droot = shift or die "Usage: $0 <options> Document-Rootdirectories"; |
| 71 |
|
| 72 |
my %D; |
| 73 |
my $access = tie %D, 'WAIT::Document::Ora', $droot, |
| 74 |
or die "Couldn't tie to dir $droot: $!\n"; |
| 75 |
|
| 76 |
my $tb = $db->create_table(name => $OPT{table}, |
| 77 |
attr => ['author', 'isbn', 'title', |
| 78 |
'headline', 'docid'], |
| 79 |
layout => $layout, |
| 80 |
access => $access, |
| 81 |
invindex => |
| 82 |
[ |
| 83 |
'aboutauthor' => $text, |
| 84 |
'aboutauthor' => $wplus, |
| 85 |
'abouttranslator' => $text, |
| 86 |
'abouttranslator' => $wplus, |
| 87 |
'abstract' => $text, |
| 88 |
'abstract' => $wplus, |
| 89 |
'author' => $text, |
| 90 |
'colophon' => $text, |
| 91 |
'colophon' => $wplus, |
| 92 |
'desc' => $text, |
| 93 |
'desc' => $wplus, |
| 94 |
'inx' => $text, |
| 95 |
'inx' => $wplus, |
| 96 |
'isbn' => $isbn, |
| 97 |
'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 |
] |
| 108 |
); |
| 109 |
die "Couldn't create table $OPT{table}: $@\n" unless $tb; |
| 110 |
|
| 111 |
my ($did, $value); |
| 112 |
binmode STDOUT, ":utf8"; |
| 113 |
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 |
my $record = $layout->split($value); |
| 121 |
my $headline = $record->{title}; |
| 122 |
$headline =~ s/\s+/ /sg; |
| 123 |
# printf "%15s %s\n", $record->{isbn}, substr($headline,0,60); |
| 124 |
printf "%15s %s...\n", $did, substr($headline,0,60); |
| 125 |
$tb->insert('docid' => $did, |
| 126 |
headline => $headline, |
| 127 |
%{$record}); |
| 128 |
$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 |
} |
| 148 |
undef $ALL; |
| 149 |
$tb->set(top=>1); |
| 150 |
|
| 151 |
my $tritb = $db->create_table( |
| 152 |
name => "$OPT{table}_fallback", |
| 153 |
attr => [qw(docid headline)], # name |
| 154 |
# "headline" |
| 155 |
# only for |
| 156 |
# sman |
| 157 |
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 |
my $name = $idx->name; |
| 164 |
next if $name =~ /(_|\b)(mixedonly|Stem|Soundex)(\b|_)/; |
| 165 |
# irrelevant for alternatives |
| 166 |
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 |
# printf "%s\n", substr($headline,0,60); |
| 178 |
$tritb->insert(docid => $headline, headline => $headline); |
| 179 |
} |
| 180 |
$tritb->set(top=>1); |
| 181 |
$tritb->close; |
| 182 |
$tb->close(); |
| 183 |
$db->close(); |
| 184 |
|
| 185 |
# Atomically relinking symlink: now we have a new database with a very |
| 186 |
# 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 |
|
| 190 |
use File::Spec; |
| 191 |
chdir $OPT{dir} or die; |
| 192 |
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 |
|
| 201 |
$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 |
|
Parent Directory
| 
Revision Log