22 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
23 |
use Storable qw/dclone/; |
use Storable qw/dclone/; |
24 |
|
|
25 |
|
use Proc::Queue size => 1; |
26 |
|
use POSIX ":sys_wait_h"; # imports WNOHANG |
27 |
|
|
28 |
=head1 NAME |
=head1 NAME |
29 |
|
|
30 |
run.pl - start WebPAC indexing |
run.pl - start WebPAC indexing |
84 |
|
|
85 |
Force dump or input and marc record for debugging. |
Force dump or input and marc record for debugging. |
86 |
|
|
87 |
|
=item --parallel 4 |
88 |
|
|
89 |
|
Run databases in parallel (aproximatly same as number of processors in |
90 |
|
machine if you want to use full load) |
91 |
|
|
92 |
=back |
=back |
93 |
|
|
94 |
=cut |
=cut |
106 |
my $marc_lint = 1; |
my $marc_lint = 1; |
107 |
my $marc_dump = 0; |
my $marc_dump = 0; |
108 |
|
|
109 |
|
my $parallel = 0; |
110 |
|
|
111 |
GetOptions( |
GetOptions( |
112 |
"limit=i" => \$limit, |
"limit=i" => \$limit, |
113 |
"offset=i" => \$offset, |
"offset=i" => \$offset, |
122 |
"marc-output=s" => \$marc_output, |
"marc-output=s" => \$marc_output, |
123 |
"marc-lint!" => \$marc_lint, |
"marc-lint!" => \$marc_lint, |
124 |
"marc-dump!" => \$marc_dump, |
"marc-dump!" => \$marc_dump, |
125 |
|
"parallel=i" => \$parallel, |
126 |
); |
); |
127 |
|
|
128 |
$config = LoadFile($config); |
$config = LoadFile($config); |
156 |
my @links; |
my @links; |
157 |
my $indexer; |
my $indexer; |
158 |
|
|
159 |
|
if ($parallel) { |
160 |
|
$log->info("Using $parallel processes for speedup"); |
161 |
|
Proc::Queue::size($parallel); |
162 |
|
} |
163 |
|
|
164 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
165 |
|
|
166 |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter); |
167 |
next if ($only_database && $database !~ m/$only_database/i); |
next if ($only_database && $database !~ m/$only_database/i); |
168 |
|
|
169 |
|
if ($parallel) { |
170 |
|
my $f=fork; |
171 |
|
if(defined ($f) and $f==0) { |
172 |
|
$log->info("Created processes $$ for speedup"); |
173 |
|
} else { |
174 |
|
next; |
175 |
|
} |
176 |
|
} |
177 |
|
|
178 |
if ($use_indexer) { |
if ($use_indexer) { |
179 |
my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration"); |
my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration"); |
180 |
$indexer_config->{database} = $database; |
$indexer_config->{database} = $database; |
415 |
} |
} |
416 |
} |
} |
417 |
|
|
418 |
|
# end forked process |
419 |
|
if ($parallel) { |
420 |
|
$log->info("parallel process $$ finished"); |
421 |
|
exit(0); |
422 |
|
} |
423 |
|
|
424 |
|
} |
425 |
|
|
426 |
|
if ($parallel) { |
427 |
|
# wait all children to finish |
428 |
|
sleep(1) while wait != -1; |
429 |
|
$log->info("all parallel processes finished"); |
430 |
} |
} |
431 |
|
|
432 |
foreach my $link (@links) { |
foreach my $link (@links) { |
433 |
$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]"); |
$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]"); |
434 |
$indexer->add_link( %{ $link } ); |
$indexer->add_link( %{ $link } ); |
435 |
} |
} |
436 |
|
|