13 |
use WebPAC::Store 0.03; |
use WebPAC::Store 0.03; |
14 |
use WebPAC::Normalize::XML; |
use WebPAC::Normalize::XML; |
15 |
use WebPAC::Output::TT; |
use WebPAC::Output::TT; |
16 |
use WebPAC::Output::Estraier 0.08; |
use WebPAC::Output::Estraier '0.10'; |
17 |
use YAML qw/LoadFile/; |
use YAML qw/LoadFile/; |
18 |
use Getopt::Long; |
use Getopt::Long; |
19 |
use File::Path; |
use File::Path; |
20 |
|
use Time::HiRes qw/time/; |
21 |
|
|
22 |
=head1 NAME |
=head1 NAME |
23 |
|
|
41 |
|
|
42 |
remove database and Hyper Estraier index before indexing |
remove database and Hyper Estraier index before indexing |
43 |
|
|
44 |
=item --one=database_name |
=item --only=database_name |
45 |
|
|
46 |
reindex just single database |
reindex just single database (legacy name is --one) |
47 |
|
|
48 |
=item --config conf/config.yml |
=item --config conf/config.yml |
49 |
|
|
59 |
my $clean = 0; |
my $clean = 0; |
60 |
my $config = 'conf/config.yml'; |
my $config = 'conf/config.yml'; |
61 |
my $debug = 0; |
my $debug = 0; |
62 |
my $one_db_name; |
my $only_db_name; |
63 |
|
|
64 |
GetOptions( |
GetOptions( |
65 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
66 |
"offset=i" => \$offset, |
"offset=i" => \$offset, |
67 |
"clean" => \$clean, |
"clean" => \$clean, |
68 |
"one=s" => \$one_db_name, |
"one=s" => \$only_db_name, |
69 |
|
"only=s" => \$only_db_name, |
70 |
"config" => \$config, |
"config" => \$config, |
71 |
"debug" => \$debug, |
"debug" => \$debug, |
72 |
); |
); |
78 |
die "no databases in config file!\n" unless ($config->{databases}); |
die "no databases in config file!\n" unless ($config->{databases}); |
79 |
|
|
80 |
my $total_rows = 0; |
my $total_rows = 0; |
81 |
|
my $start_t = time(); |
82 |
|
|
83 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
84 |
|
|
85 |
next if ($one_db_name && $database !~ m/$one_db_name/i); |
next if ($only_db_name && $database !~ m/$only_db_name/i); |
86 |
|
|
87 |
my $log = _new WebPAC::Common()->_get_logger(); |
my $log = _new WebPAC::Common()->_get_logger(); |
88 |
|
|
93 |
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
94 |
$est_config->{database} = $database; |
$est_config->{database} = $database; |
95 |
$est_config->{clean} = $clean; |
$est_config->{clean} = $clean; |
96 |
|
$est_config->{label} = $db_config->{name}; |
97 |
|
|
98 |
my $est = new WebPAC::Output::Estraier( %{ $est_config } ); |
my $est = new WebPAC::Output::Estraier( %{ $est_config } ); |
99 |
|
|
109 |
$log->info("creating new database $database in $db_path"); |
$log->info("creating new database $database in $db_path"); |
110 |
rmtree( $db_path ) || $log->warn("can't remove $db_path: $!"); |
rmtree( $db_path ) || $log->warn("can't remove $db_path: $!"); |
111 |
} else { |
} else { |
112 |
$log->info("working on $database in $db_path"); |
$log->debug("working on $database in $db_path"); |
113 |
} |
} |
114 |
|
|
115 |
my $db = new WebPAC::Store( |
my $db = new WebPAC::Store( |
146 |
|
|
147 |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
148 |
|
|
149 |
$log->info("working on input $input->{path} [$input->{type}] using $input_module"); |
$log->info("working on input '$input->{path}' [$input->{type}] using $input_module lookup '$input->{lookup}'"); |
150 |
|
|
151 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
152 |
module => $input_module, |
module => $input_module, |
154 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
155 |
offset => $offset, |
offset => $offset, |
156 |
lookup => $lookup, |
lookup => $lookup, |
157 |
|
recode => $input->{recode}, |
158 |
); |
); |
159 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
160 |
|
|
210 |
|
|
211 |
}; |
}; |
212 |
|
|
213 |
$log->info("$total_rows records indexed"); |
my $dt = time() - $start_t; |
214 |
|
$log->info("$total_rows records indexed in " . |
215 |
|
sprintf("%.2f sec [%.2f rec/sec]", |
216 |
|
$dt, ($total_rows / $dt) |
217 |
|
) |
218 |
|
); |
219 |
|
|
220 |
# |
# |
221 |
# add Hyper Estraier links to other databases |
# add Hyper Estraier links to other databases |