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; |
|
use WebPAC::Output::Estraier 0.08; |
|
16 |
use YAML qw/LoadFile/; |
use YAML qw/LoadFile/; |
17 |
use Getopt::Long; |
use Getopt::Long; |
18 |
use File::Path; |
use File::Path; |
19 |
|
use Time::HiRes qw/time/; |
20 |
|
|
21 |
=head1 NAME |
=head1 NAME |
22 |
|
|
40 |
|
|
41 |
remove database and Hyper Estraier index before indexing |
remove database and Hyper Estraier index before indexing |
42 |
|
|
43 |
=item --one=database_name |
=item --only=database_name |
44 |
|
|
45 |
reindex just single database |
reindex just single database (legacy name is --one) |
46 |
|
|
47 |
=item --config conf/config.yml |
=item --config conf/config.yml |
48 |
|
|
58 |
my $clean = 0; |
my $clean = 0; |
59 |
my $config = 'conf/config.yml'; |
my $config = 'conf/config.yml'; |
60 |
my $debug = 0; |
my $debug = 0; |
61 |
my $one_db_name; |
my $only_db_name; |
62 |
|
|
63 |
GetOptions( |
GetOptions( |
64 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
65 |
"offset=i" => \$offset, |
"offset=i" => \$offset, |
66 |
"clean" => \$clean, |
"clean" => \$clean, |
67 |
"one=s" => \$one_db_name, |
"one=s" => \$only_db_name, |
68 |
|
"only=s" => \$only_db_name, |
69 |
"config" => \$config, |
"config" => \$config, |
70 |
"debug" => \$debug, |
"debug" => \$debug, |
71 |
); |
); |
76 |
|
|
77 |
die "no databases in config file!\n" unless ($config->{databases}); |
die "no databases in config file!\n" unless ($config->{databases}); |
78 |
|
|
79 |
|
my $log = _new WebPAC::Common()->_get_logger(); |
80 |
|
|
81 |
|
my $use_indexer = $config->{use_indexer} || 'hyperestraier'; |
82 |
|
$log->info("using $use_indexer indexing engine..."); |
83 |
|
|
84 |
my $total_rows = 0; |
my $total_rows = 0; |
85 |
|
my $start_t = time(); |
86 |
|
|
87 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
88 |
|
|
89 |
next if ($one_db_name && $database !~ m/$one_db_name/i); |
next if ($only_db_name && $database !~ m/$only_db_name/i); |
90 |
|
|
91 |
my $log = _new WebPAC::Common()->_get_logger(); |
my $indexer; |
92 |
|
|
93 |
# |
my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration"); |
94 |
# open Hyper Estraier database |
$indexer_config->{database} = $database; |
95 |
# |
$indexer_config->{clean} = $clean; |
96 |
|
$indexer_config->{label} = $db_config->{name}; |
97 |
|
|
98 |
|
if ($use_indexer eq 'hyperestraier') { |
99 |
|
|
100 |
|
# open Hyper Estraier database |
101 |
|
use WebPAC::Output::Estraier '0.10'; |
102 |
|
$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } ); |
103 |
|
|
104 |
|
} elsif ($use_indexer eq 'kinosearch') { |
105 |
|
|
106 |
|
# open KinoSearch |
107 |
|
use WebPAC::Output::KinoSearch; |
108 |
|
$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } ); |
109 |
|
|
110 |
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
} else { |
111 |
$est_config->{database} = $database; |
$log->logdie("unknown use_indexer: $use_indexer"); |
112 |
$est_config->{clean} = $clean; |
} |
113 |
|
|
114 |
my $est = new WebPAC::Output::Estraier( %{ $est_config } ); |
$log->logide("can't continue without valid indexer") unless ($indexer); |
115 |
|
|
116 |
# |
# |
117 |
# now WebPAC::Store |
# now WebPAC::Store |
125 |
$log->info("creating new database $database in $db_path"); |
$log->info("creating new database $database in $db_path"); |
126 |
rmtree( $db_path ) || $log->warn("can't remove $db_path: $!"); |
rmtree( $db_path ) || $log->warn("can't remove $db_path: $!"); |
127 |
} else { |
} else { |
128 |
$log->info("working on $database in $db_path"); |
$log->debug("working on $database in $db_path"); |
129 |
} |
} |
130 |
|
|
131 |
my $db = new WebPAC::Store( |
my $db = new WebPAC::Store( |
162 |
|
|
163 |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
164 |
|
|
165 |
$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}'"); |
166 |
|
|
167 |
my $input_db = new WebPAC::Input( |
my $input_db = new WebPAC::Input( |
168 |
module => $input_module, |
module => $input_module, |
170 |
limit => $limit || $input->{limit}, |
limit => $limit || $input->{limit}, |
171 |
offset => $offset, |
offset => $offset, |
172 |
lookup => $lookup, |
lookup => $lookup, |
173 |
|
recode => $input->{recode}, |
174 |
); |
); |
175 |
$log->logdie("can't create input using $input_module") unless ($input); |
$log->logdie("can't create input using $input_module") unless ($input); |
176 |
|
|
215 |
|
|
216 |
my $ds = $n->data_structure($row); |
my $ds = $n->data_structure($row); |
217 |
|
|
218 |
$est->add( |
$indexer->add( |
219 |
id => $input->{name} . "/" . $mfn, |
id => $input->{name} . "/" . $mfn, |
220 |
ds => $ds, |
ds => $ds, |
221 |
type => $config->{hyperestraier}->{type}, |
type => $config->{$use_indexer}->{type}, |
222 |
); |
); |
223 |
|
|
224 |
$total_rows++; |
$total_rows++; |
226 |
|
|
227 |
}; |
}; |
228 |
|
|
229 |
$log->info("$total_rows records indexed"); |
my $dt = time() - $start_t; |
230 |
|
$log->info("$total_rows records indexed in " . |
231 |
|
sprintf("%.2f sec [%.2f rec/sec]", |
232 |
|
$dt, ($total_rows / $dt) |
233 |
|
) |
234 |
|
); |
235 |
|
|
236 |
# |
# |
237 |
# add Hyper Estraier links to other databases |
# add Hyper Estraier links to other databases |
238 |
# |
# |
239 |
if (ref($db_config->{links}) eq 'ARRAY') { |
if (ref($db_config->{links}) eq 'ARRAY') { |
240 |
foreach my $link (@{ $db_config->{links} }) { |
foreach my $link (@{ $db_config->{links} }) { |
241 |
$log->info("adding link $database -> $link->{to} [$link->{credit}]"); |
if ($use_indexer eq 'hyperestraier') { |
242 |
$est->add_link( |
$log->info("adding link $database -> $link->{to} [$link->{credit}]"); |
243 |
from => $database, |
$indexer->add_link( |
244 |
to => $link->{to}, |
from => $database, |
245 |
credit => $link->{credit}, |
to => $link->{to}, |
246 |
); |
credit => $link->{credit}, |
247 |
|
); |
248 |
|
} else { |
249 |
|
$log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]"); |
250 |
|
} |
251 |
} |
} |
252 |
} |
} |
253 |
|
|