webpac2/trunk/run.pl

#!/usr/bin/perl -w

use strict;

use Cwd qw/abs_path/;
use File::Temp qw/tempdir/;
use lib './lib';

use WebPAC::Common 0.02;
use WebPAC::Lookup 0.03;
use WebPAC::Input 0.07;
use WebPAC::Store 0.03;
use WebPAC::Normalize 0.11;
use WebPAC::Output::TT;
use WebPAC::Validate;
use WebPAC::Output::MARC;
use YAML qw/LoadFile/;
use Getopt::Long;
use File::Path;
use Time::HiRes qw/time/;
use File::Slurp;
use Data::Dump qw/dump/;
use Storable qw/dclone/;

=head1 NAME

run.pl - start WebPAC indexing

B<this command will probably go away. Don't get used to it!>

Options:

=over 4

=item --offset 42

start loading (all) databases at offset 42

=item --limit 100

limit loading to 100 records

=item --clean

remove database and Hyper Estraier index before indexing

=item --only=database_name/input_filter

reindex just single database (legacy name is --one)

C</input_filter> is optional part which can be C<name>
or C<type> from input

=item --config conf/config.yml

path to YAML configuration file

=item --stats

disable indexing and dump statistics about field and subfield
usage for each input

=item --validate path/to/validation_file

turn on extra validation of imput records, see L<WebPAC::Validation>

=item --marc-normalize conf/normalize/mapping.pl

This option specifies normalisation file for MARC creation

=item --marc-output out/marc/test.marc

Optional path to output file

=item --marc-lint

By default turned on if C<--marc-normalize> is used. You can disable lint
messages with C<--no-marc-lint>.

=item --marc-dump

Force dump or input and marc record for debugging.

=back

=cut

my $offset;
my $limit;

my $clean = 0;
my $config = 'conf/config.yml';
my $debug = 0;
my $only_filter;
my $stats = 0;
my $validate_path;
my ($marc_normalize, $marc_output);
my $marc_lint = 1;
my $marc_dump = 0;

GetOptions(
        "limit=i" => \$limit,
        "offset=i" => \$offset,
        "clean" => \$clean,
        "one=s" => \$only_filter,
        "only=s" => \$only_filter,
        "config" => \$config,
        "debug+" => \$debug,
        "stats" => \$stats,
        "validate=s" => \$validate_path,
        "marc-normalize=s" => \$marc_normalize,
        "marc-output=s" => \$marc_output,
        "marc-lint!" => \$marc_lint,
        "marc-dump!" => \$marc_dump,
);

$config = LoadFile($config);

print "config = ",dump($config) if ($debug);

die "no databases in config file!\n" unless ($config->{databases});

my $log = _new WebPAC::Common()->_get_logger();
$log->info( "-" x 79 );

my $validate;
$validate = new WebPAC::Validate(
        path => $validate_path,
) if ($validate_path);

my $use_indexer = $config->{use_indexer} || 'hyperestraier';
if ($stats) {
        $log->debug("option --stats disables update of indexing engine...");
        $use_indexer = undef;
} else {
        $log->info("using $use_indexer indexing engine...");
}

# disable indexing when creating marc
$use_indexer = undef if ($marc_normalize);

my $total_rows = 0;
my $start_t = time();

my @links;
my $indexer;

while (my ($database, $db_config) = each %{ $config->{databases} }) {

        my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
        next if ($only_database && $database !~ m/$only_database/i);

        if ($use_indexer) {
                my $indexer_config = $config->{$use_indexer} || $log->logdie("can't find '$use_indexer' part in confguration");
                $indexer_config->{database} = $database;
                $indexer_config->{clean} = $clean;
                $indexer_config->{label} = $db_config->{name};

                if ($use_indexer eq 'hyperestraier') {

                        # open Hyper Estraier database
                        use WebPAC::Output::Estraier '0.10';
                        $indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
                
                } elsif ($use_indexer eq 'kinosearch') {

                        # open KinoSearch
                        use WebPAC::Output::KinoSearch;
                        $indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
                        $indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );

                } else {
                        $log->logdie("unknown use_indexer: $use_indexer");
                }

                $log->logide("can't continue without valid indexer") unless ($indexer);
        }


        #
        # now WebPAC::Store
        #
        my $abs_path = abs_path($0);
        $abs_path =~ s#/[^/]*$#/#;

        my $db_path = $config->{webpac}->{db_path} . '/' . $database;

        if ($clean) {
                $log->info("creating new database '$database' in $db_path");
                rmtree( $db_path ) || $log->warn("can't remove $db_path: $!");
        } else {
                $log->info("working on database '$database' in $db_path");
        }

        my $db = new WebPAC::Store(
                path => $db_path,
                database => $database,
                debug => $debug,
        );


        #
        # now, iterate through input formats
        #

        my @inputs;
        if (ref($db_config->{input}) eq 'ARRAY') {
                @inputs = @{ $db_config->{input} };
        } elsif ($db_config->{input}) {
                push @inputs, $db_config->{input};
        } else {
                $log->info("database $database doesn't have inputs defined");
        }

        my @supported_inputs = keys %{ $config->{webpac}->{inputs} };

        foreach my $input (@inputs) {

                next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));

                my $type = lc($input->{type});

                die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));

                my $lookup;
                if ($input->{lookup}) {
                        $lookup = new WebPAC::Lookup(
                                lookup_file => $input->{lookup},
                        );
                        delete( $input->{lookup} );
                }

                my $input_module = $config->{webpac}->{inputs}->{$type};

                $log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module",
                        $input->{lookup} ? "lookup '$input->{lookup}'" : ""
                );

                my $input_db = new WebPAC::Input(
                        module => $input_module,
                        encoding => $config->{webpac}->{webpac_encoding},
                        limit => $limit || $input->{limit},
                        offset => $offset,
                        lookup_coderef => sub {
                                my $rec = shift || return;
                                $lookup->add( $rec );
                        },
                        recode => $input->{recode},
                        stats => $stats,
                        modify_records => $input->{modify_records},
                );
                $log->logdie("can't create input using $input_module") unless ($input);

                my $maxmfn = $input_db->open(
                        path => $input->{path},
                        code_page => $input->{encoding},        # database encoding
                        %{ $input },
                );

                my @norm_array = ref($input->{normalize}) eq 'ARRAY' ?
                        @{ $input->{normalize} } : ( $input->{normalize} );

                if ($marc_normalize) {
                        @norm_array = ( {
                                path => $marc_normalize,
                                output => $marc_output || 'out/marc/' . $database . '-' . $input->{name} . '.marc',
                        } );
                }

                foreach my $normalize (@norm_array) {

                        my $normalize_path = $normalize->{path} || $log->logdie("can't find normalize path in config");

                        $log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i );

                        my $rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";

                        $log->info("Using $normalize_path for normalization...");

                        my $marc = new WebPAC::Output::MARC(
                                path => $normalize->{output},
                                lint => $marc_lint,
                                dump => $marc_dump,
                        ) if ($normalize->{output});

                        # reset position in database
                        $input_db->seek(1);

                        foreach my $pos ( 0 ... $input_db->size ) {

                                my $row = $input_db->fetch || next;

                                my $mfn = $row->{'000'}->[0];

                                if (! $mfn || $mfn !~ m#^\d+$#) {
                                        $log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
                                        $mfn = $pos;
                                        push @{ $row->{'000'} }, $pos;
                                }


                                if ($validate) {
                                        my @errors = $validate->validate_errors( $row );
                                        $log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
                                }

                                my $ds_config = dclone($db_config);

                                # default values -> database key
                                $ds_config->{_} = $database;

                                # current mfn
                                $ds_config->{_mfn} = $mfn;

                                # attach current input
                                $ds_config->{input} = $input;

                                my $ds = WebPAC::Normalize::data_structure(
                                        row => $row,
                                        rules => $rules,
                                        lookup => $lookup ? $lookup->lookup_hash : undef,
                                        config => $ds_config,
                                        marc_encoding => 'utf-8',
                                );

                                $db->save_ds(
                                        id => $mfn,
                                        ds => $ds,
                                        prefix => $input->{name},
                                ) if ($ds && !$stats);

                                $indexer->add(
                                        id => $input->{name} . "/" . $mfn,
                                        ds => $ds,
                                        type => $config->{$use_indexer}->{type},
                                ) if ($indexer && $ds);

                                if ($marc) {
                                        my $i = 0;

                                        while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) {
                                                $marc->add(
                                                        id => $mfn . ( $i ? "/$i" : '' ),
                                                        fields => $fields,
                                                        leader => WebPAC::Normalize::marc_leader(),
                                                        row => $row,
                                                );
                                                $i++;
                                        }

                                        $log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
                                }

                                $total_rows++;
                        }

                        $log->info("statistics of fields usage:\n", $input_db->stats) if ($stats);

                        # close MARC file
                        $marc->finish if ($marc);

                }

        }

        eval { $indexer->finish } if ($indexer && $indexer->can('finish'));

        my $dt = time() - $start_t;
        $log->info("$total_rows records ", $indexer ? "indexed " : "",
                sprintf("in %.2f sec [%.2f rec/sec]",
                        $dt, ($total_rows / $dt)
                )
        );

        #
        # add Hyper Estraier links to other databases
        #
        if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
                foreach my $link (@{ $db_config->{links} }) {
                        if ($use_indexer eq 'hyperestraier') {
                                $log->info("saving link $database -> $link->{to} [$link->{credit}]");
                                push @links, {
                                        from => $database,
                                        to => $link->{to},
                                        credit => $link->{credit},
                                };
                        } else {
                                $log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
                        }
                }
        }

}

foreach my $link (@links) {
        $log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
        $indexer->add_link( %{ $link } );
}
1	dpavlin	74	#!/usr/bin/perl -w
2
3			use strict;
4
5			use Cwd qw/abs_path/;
6			use File::Temp qw/tempdir/;
7			use lib './lib';
8
9	dpavlin	255	use WebPAC::Common 0.02;
10	dpavlin	588	use WebPAC::Lookup 0.03;
11	dpavlin	585	use WebPAC::Input 0.07;
12	dpavlin	209	use WebPAC::Store 0.03;
13	dpavlin	581	use WebPAC::Normalize 0.11;
14	dpavlin	74	use WebPAC::Output::TT;
15	dpavlin	516	use WebPAC::Validate;
16	dpavlin	578	use WebPAC::Output::MARC;
17	dpavlin	141	use YAML qw/LoadFile/;
18	dpavlin	301	use Getopt::Long;
19			use File::Path;
20	dpavlin	389	use Time::HiRes qw/time/;
21	dpavlin	492	use File::Slurp;
22	dpavlin	556	use Data::Dump qw/dump/;
23	dpavlin	595	use Storable qw/dclone/;
24	dpavlin	74
25	dpavlin	301	=head1 NAME
26	dpavlin	76
27	dpavlin	301	run.pl - start WebPAC indexing
28	dpavlin	141
29	dpavlin	301	B<this command will probably go away. Don't get used to it!>
30	dpavlin	141
31	dpavlin	301	Options:
32
33			=over 4
34
35			=item --offset 42
36
37			start loading (all) databases at offset 42
38
39			=item --limit 100
40
41			limit loading to 100 records
42
43			=item --clean
44
45			remove database and Hyper Estraier index before indexing
46
47	dpavlin	510	=item --only=database_name/input_filter
48	dpavlin	335
49	dpavlin	423	reindex just single database (legacy name is --one)
50	dpavlin	335
51	dpavlin	510	C</input_filter> is optional part which can be C<name>
52			or C<type> from input
53
54	dpavlin	301	=item --config conf/config.yml
55
56			path to YAML configuration file
57
58	dpavlin	507	=item --stats
59
60	dpavlin	509	disable indexing and dump statistics about field and subfield
61			usage for each input
62	dpavlin	507
63	dpavlin	516	=item --validate path/to/validation_file
64
65			turn on extra validation of imput records, see L<WebPAC::Validation>
66
67	dpavlin	552	=item --marc-normalize conf/normalize/mapping.pl
68
69			This option specifies normalisation file for MARC creation
70
71			=item --marc-output out/marc/test.marc
72
73			Optional path to output file
74
75	dpavlin	556	=item --marc-lint
76
77			By default turned on if C<--marc-normalize> is used. You can disable lint
78	dpavlin	558	messages with C<--no-marc-lint>.
79	dpavlin	556
80	dpavlin	559	=item --marc-dump
81
82			Force dump or input and marc record for debugging.
83
84	dpavlin	301	=back
85
86			=cut
87
88			my $offset;
89			my $limit;
90
91			my $clean = 0;
92			my $config = 'conf/config.yml';
93			my $debug = 0;
94	dpavlin	510	my $only_filter;
95	dpavlin	507	my $stats = 0;
96	dpavlin	516	my $validate_path;
97	dpavlin	552	my ($marc_normalize, $marc_output);
98	dpavlin	556	my $marc_lint = 1;
99	dpavlin	559	my $marc_dump = 0;
100	dpavlin	301
101			GetOptions(
102			"limit=i" => \$limit,
103			"offset=i" => \$offset,
104			"clean" => \$clean,
105	dpavlin	510	"one=s" => \$only_filter,
106			"only=s" => \$only_filter,
107	dpavlin	301	"config" => \$config,
108	dpavlin	560	"debug+" => \$debug,
109	dpavlin	507	"stats" => \$stats,
110	dpavlin	516	"validate=s" => \$validate_path,
111	dpavlin	552	"marc-normalize=s" => \$marc_normalize,
112			"marc-output=s" => \$marc_output,
113	dpavlin	556	"marc-lint!" => \$marc_lint,
114	dpavlin	559	"marc-dump!" => \$marc_dump,
115	dpavlin	301	);
116
117			$config = LoadFile($config);
118
119	dpavlin	556	print "config = ",dump($config) if ($debug);
120	dpavlin	301
121	dpavlin	210	die "no databases in config file!\n" unless ($config->{databases});
122	dpavlin	431
123			my $log = _new WebPAC::Common()->_get_logger();
124	dpavlin	509	$log->info( "-" x 79 );
125	dpavlin	431
126	dpavlin	516	my $validate;
127			$validate = new WebPAC::Validate(
128			path => $validate_path,
129			) if ($validate_path);
130
131	dpavlin	430	my $use_indexer = $config->{use_indexer} \|\| 'hyperestraier';
132	dpavlin	509	if ($stats) {
133			$log->debug("option --stats disables update of indexing engine...");
134			$use_indexer = undef;
135			} else {
136			$log->info("using $use_indexer indexing engine...");
137			}
138	dpavlin	141
139	dpavlin	552	# disable indexing when creating marc
140			$use_indexer = undef if ($marc_normalize);
141
142	dpavlin	213	my $total_rows = 0;
143	dpavlin	389	my $start_t = time();
144	dpavlin	213
145	dpavlin	529	my @links;
146			my $indexer;
147
148	dpavlin	210	while (my ($database, $db_config) = each %{ $config->{databases} }) {
149	dpavlin	141
150	dpavlin	512	my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
151	dpavlin	510	next if ($only_database && $database !~ m/$only_database/i);
152	dpavlin	335
153	dpavlin	509	if ($use_indexer) {
154			my $indexer_config = $config->{$use_indexer} \|\| $log->logdie("can't find '$use_indexer' part in confguration");
155			$indexer_config->{database} = $database;
156			$indexer_config->{clean} = $clean;
157			$indexer_config->{label} = $db_config->{name};
158	dpavlin	431
159	dpavlin	509	if ($use_indexer eq 'hyperestraier') {
160	dpavlin	255
161	dpavlin	509	# open Hyper Estraier database
162			use WebPAC::Output::Estraier '0.10';
163			$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
164
165			} elsif ($use_indexer eq 'kinosearch') {
166	dpavlin	430
167	dpavlin	509	# open KinoSearch
168			use WebPAC::Output::KinoSearch;
169			$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
170			$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
171	dpavlin	431
172	dpavlin	509	} else {
173			$log->logdie("unknown use_indexer: $use_indexer");
174			}
175
176			$log->logide("can't continue without valid indexer") unless ($indexer);
177	dpavlin	430	}
178
179
180	dpavlin	255	#
181			# now WebPAC::Store
182			#
183	dpavlin	210	my $abs_path = abs_path($0);
184			$abs_path =~ s#/[^/]*$#/#;
185	dpavlin	141
186	dpavlin	210	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
187	dpavlin	74
188	dpavlin	301	if ($clean) {
189	dpavlin	511	$log->info("creating new database '$database' in $db_path");
190	dpavlin	301	rmtree( $db_path ) \|\| $log->warn("can't remove $db_path: $!");
191			} else {
192	dpavlin	511	$log->info("working on database '$database' in $db_path");
193	dpavlin	301	}
194	dpavlin	255
195	dpavlin	210	my $db = new WebPAC::Store(
196			path => $db_path,
197	dpavlin	217	database => $database,
198	dpavlin	301	debug => $debug,
199	dpavlin	210	);
200	dpavlin	74
201	dpavlin	233
202	dpavlin	213	#
203			# now, iterate through input formats
204			#
205	dpavlin	74
206	dpavlin	213	my @inputs;
207			if (ref($db_config->{input}) eq 'ARRAY') {
208			@inputs = @{ $db_config->{input} };
209	dpavlin	255	} elsif ($db_config->{input}) {
210			push @inputs, $db_config->{input};
211	dpavlin	213	} else {
212	dpavlin	255	$log->info("database $database doesn't have inputs defined");
213	dpavlin	213	}
214	dpavlin	74
215	dpavlin	286	my @supported_inputs = keys %{ $config->{webpac}->{inputs} };
216
217	dpavlin	213	foreach my $input (@inputs) {
218	dpavlin	233
219	dpavlin	512	next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));
220	dpavlin	510
221	dpavlin	233	my $type = lc($input->{type});
222
223	dpavlin	286	die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
224	dpavlin	233
225	dpavlin	588	my $lookup;
226			if ($input->{lookup}) {
227			$lookup = new WebPAC::Lookup(
228			lookup_file => $input->{lookup},
229			);
230			delete( $input->{lookup} );
231			}
232	dpavlin	233
233	dpavlin	286	my $input_module = $config->{webpac}->{inputs}->{$type};
234	dpavlin	74
235	dpavlin	523	$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module",
236			$input->{lookup} ? "lookup '$input->{lookup}'" : ""
237			);
238	dpavlin	286
239	dpavlin	287	my $input_db = new WebPAC::Input(
240			module => $input_module,
241	dpavlin	585	encoding => $config->{webpac}->{webpac_encoding},
242	dpavlin	301	limit => $limit \|\| $input->{limit},
243			offset => $offset,
244	dpavlin	585	lookup_coderef => sub {
245			my $rec = shift \|\| return;
246			$lookup->add( $rec );
247			},
248	dpavlin	416	recode => $input->{recode},
249	dpavlin	507	stats => $stats,
250	dpavlin	600	modify_records => $input->{modify_records},
251	dpavlin	287	);
252	dpavlin	286	$log->logdie("can't create input using $input_module") unless ($input);
253	dpavlin	113
254	dpavlin	287	my $maxmfn = $input_db->open(
255	dpavlin	285	path => $input->{path},
256	dpavlin	213	code_page => $input->{encoding}, # database encoding
257	dpavlin	523	%{ $input },
258	dpavlin	213	);
259	dpavlin	113
260	dpavlin	539	my @norm_array = ref($input->{normalize}) eq 'ARRAY' ?
261			@{ $input->{normalize} } : ( $input->{normalize} );
262	dpavlin	210
263	dpavlin	552	if ($marc_normalize) {
264			@norm_array = ( {
265			path => $marc_normalize,
266			output => $marc_output \|\| 'out/marc/' . $database . '-' . $input->{name} . '.marc',
267			} );
268			}
269
270	dpavlin	539	foreach my $normalize (@norm_array) {
271	dpavlin	492
272	dpavlin	539	my $normalize_path = $normalize->{path} \|\| $log->logdie("can't find normalize path in config");
273	dpavlin	269
274	dpavlin	539	$log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i );
275	dpavlin	210
276	dpavlin	539	my $rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
277	dpavlin	213
278	dpavlin	539	$log->info("Using $normalize_path for normalization...");
279	dpavlin	213
280	dpavlin	578	my $marc = new WebPAC::Output::MARC(
281			path => $normalize->{output},
282			lint => $marc_lint,
283			dump => $marc_dump,
284			) if ($normalize->{output});
285	dpavlin	540
286	dpavlin	539	# reset position in database
287			$input_db->seek(1);
288	dpavlin	290
289	dpavlin	539	foreach my $pos ( 0 ... $input_db->size ) {
290	dpavlin	516
291	dpavlin	539	my $row = $input_db->fetch \|\| next;
292	dpavlin	516
293	dpavlin	539	my $mfn = $row->{'000'}->[0];
294	dpavlin	213
295	dpavlin	539	if (! $mfn \|\| $mfn !~ m#^\d+$#) {
296			$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
297			$mfn = $pos;
298			push @{ $row->{'000'} }, $pos;
299			}
300	dpavlin	508
301	dpavlin	213
302	dpavlin	539	if ($validate) {
303			my @errors = $validate->validate_errors( $row );
304			$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
305			}
306	dpavlin	213
307	dpavlin	595	my $ds_config = dclone($db_config);
308
309			# default values -> database key
310			$ds_config->{_} = $database;
311
312			# current mfn
313			$ds_config->{_mfn} = $mfn;
314
315			# attach current input
316			$ds_config->{input} = $input;
317
318	dpavlin	539	my $ds = WebPAC::Normalize::data_structure(
319			row => $row,
320			rules => $rules,
321			lookup => $lookup ? $lookup->lookup_hash : undef,
322	dpavlin	595	config => $ds_config,
323	dpavlin	541	marc_encoding => 'utf-8',
324	dpavlin	539	);
325	dpavlin	507
326	dpavlin	539	$db->save_ds(
327			id => $mfn,
328			ds => $ds,
329			prefix => $input->{name},
330			) if ($ds && !$stats);
331	dpavlin	210
332	dpavlin	539	$indexer->add(
333			id => $input->{name} . "/" . $mfn,
334			ds => $ds,
335			type => $config->{$use_indexer}->{type},
336	dpavlin	540	) if ($indexer && $ds);
337	dpavlin	539
338	dpavlin	580	if ($marc) {
339			my $i = 0;
340	dpavlin	559
341	dpavlin	580	while (my $fields = WebPAC::Normalize::_get_marc_fields( fetch_next => 1 ) ) {
342			$marc->add(
343			id => $mfn . ( $i ? "/$i" : '' ),
344			fields => $fields,
345			leader => WebPAC::Normalize::marc_leader(),
346			row => $row,
347			);
348			$i++;
349			}
350
351			$log->info("Created $i instances of MFN $mfn\n") if ($i > 1);
352			}
353
354	dpavlin	539	$total_rows++;
355			}
356
357			$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats);
358
359	dpavlin	540	# close MARC file
360	dpavlin	578	$marc->finish if ($marc);
361	dpavlin	539
362	dpavlin	540	}
363
364	dpavlin	539	}
365
366	dpavlin	509	eval { $indexer->finish } if ($indexer && $indexer->can('finish'));
367	dpavlin	434
368	dpavlin	389	my $dt = time() - $start_t;
369	dpavlin	511	$log->info("$total_rows records ", $indexer ? "indexed " : "",
370			sprintf("in %.2f sec [%.2f rec/sec]",
371	dpavlin	389	$dt, ($total_rows / $dt)
372			)
373			);
374	dpavlin	255
375	dpavlin	431	#
376			# add Hyper Estraier links to other databases
377			#
378	dpavlin	552	if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
379	dpavlin	431	foreach my $link (@{ $db_config->{links} }) {
380			if ($use_indexer eq 'hyperestraier') {
381	dpavlin	529	$log->info("saving link $database -> $link->{to} [$link->{credit}]");
382			push @links, {
383	dpavlin	430	from => $database,
384			to => $link->{to},
385			credit => $link->{credit},
386	dpavlin	529	};
387	dpavlin	431	} else {
388			$log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
389	dpavlin	430	}
390	dpavlin	255	}
391			}
392
393	dpavlin	210	}
394
395	dpavlin	529	foreach my $link (@links) {
396			$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
397			$indexer->add_link( %{ $link } );
398			}