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;
use WebPAC::Input 0.03;
use WebPAC::Store 0.03;
use WebPAC::Normalize;
use WebPAC::Output::TT;
use WebPAC::Validate;
use YAML qw/LoadFile/;
use Getopt::Long;
use File::Path;
use Time::HiRes qw/time/;
use File::Slurp;
use MARC::Record 2.0;   # need 2.0 for utf-8 encoding see marcpm.sf.net
use MARC::Lint;
use Data::Dump qw/dump/;

=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-arc-lint>.

=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;

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,
);

$config = LoadFile($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;

my $lint = new MARC::Lint if ($marc_lint);

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 = new WebPAC::Lookup(
                        lookup_file => $input->{lookup},
                ) if ($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,
                        code_page => $config->{webpac}->{webpac_encoding},
                        limit => $limit || $input->{limit},
                        offset => $offset,
                        lookup => $lookup,
                        recode => $input->{recode},
                        stats => $stats,
                );
                $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_fh;
                        if (my $path = $normalize->{output}) {
                                open($marc_fh, '>', $path) ||
                                        $log->logdie("can't open MARC output $path: $!");

                                $log->info("Creating MARC export file $path", $marc_lint ? ' (with lint)' : '', "\n");
                        }

                        # 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 = WebPAC::Normalize::data_structure(
                                        row => $row,
                                        rules => $rules,
                                        lookup => $lookup ? $lookup->lookup_hash : undef,
                                        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_fh) {
                                        my $marc = new MARC::Record;
                                        $marc->encoding( 'utf-8' );
                                        my @marc_fields = WebPAC::Normalize::_get_marc_fields();
                                        if (! @marc_fields) {
                                                $log->warn("MARC record $mfn is empty, skipping");
                                        } else {
                                                $marc->add_fields( @marc_fields );
                                                if ($marc_lint) {
                                                        $lint->check_record( $marc );
                                                        my $err = join( "\n", $lint->warnings );
                                                        $log->error("MARC lint detected warning on MFN $mfn\n",
                                                                "Original imput row: ",dump($row), "\n",
                                                                "Normalized MARC row: ",dump(@marc_fields), "\n",
                                                                "MARC lint warnings: ",$err,"\n"
                                                        ) if ($err);
                                                }
                                                print $marc_fh $marc->as_usmarc;
                                        }
                                }

                                $total_rows++;
                        }

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

                        # close MARC file
                        close($marc_fh) if ($marc_fh);

                }

        }

        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	74	use WebPAC::Lookup;
11	dpavlin	287	use WebPAC::Input 0.03;
12	dpavlin	209	use WebPAC::Store 0.03;
13	dpavlin	536	use WebPAC::Normalize;
14	dpavlin	74	use WebPAC::Output::TT;
15	dpavlin	516	use WebPAC::Validate;
16	dpavlin	141	use YAML qw/LoadFile/;
17	dpavlin	301	use Getopt::Long;
18			use File::Path;
19	dpavlin	389	use Time::HiRes qw/time/;
20	dpavlin	492	use File::Slurp;
21	dpavlin	541	use MARC::Record 2.0; # need 2.0 for utf-8 encoding see marcpm.sf.net
22	dpavlin	556	use MARC::Lint;
23			use Data::Dump qw/dump/;
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			messages with C<--no-arc-lint>.
79
80	dpavlin	301	=back
81
82			=cut
83
84			my $offset;
85			my $limit;
86
87			my $clean = 0;
88			my $config = 'conf/config.yml';
89			my $debug = 0;
90	dpavlin	510	my $only_filter;
91	dpavlin	507	my $stats = 0;
92	dpavlin	516	my $validate_path;
93	dpavlin	552	my ($marc_normalize, $marc_output);
94	dpavlin	556	my $marc_lint = 1;
95	dpavlin	301
96			GetOptions(
97			"limit=i" => \$limit,
98			"offset=i" => \$offset,
99			"clean" => \$clean,
100	dpavlin	510	"one=s" => \$only_filter,
101			"only=s" => \$only_filter,
102	dpavlin	301	"config" => \$config,
103			"debug" => \$debug,
104	dpavlin	507	"stats" => \$stats,
105	dpavlin	516	"validate=s" => \$validate_path,
106	dpavlin	552	"marc-normalize=s" => \$marc_normalize,
107			"marc-output=s" => \$marc_output,
108	dpavlin	556	"marc-lint!" => \$marc_lint,
109	dpavlin	301	);
110
111			$config = LoadFile($config);
112
113	dpavlin	556	print "config = ",dump($config) if ($debug);
114	dpavlin	301
115	dpavlin	210	die "no databases in config file!\n" unless ($config->{databases});
116	dpavlin	431
117			my $log = _new WebPAC::Common()->_get_logger();
118	dpavlin	509	$log->info( "-" x 79 );
119	dpavlin	431
120	dpavlin	516	my $validate;
121			$validate = new WebPAC::Validate(
122			path => $validate_path,
123			) if ($validate_path);
124
125	dpavlin	430	my $use_indexer = $config->{use_indexer} \|\| 'hyperestraier';
126	dpavlin	509	if ($stats) {
127			$log->debug("option --stats disables update of indexing engine...");
128			$use_indexer = undef;
129			} else {
130			$log->info("using $use_indexer indexing engine...");
131			}
132	dpavlin	141
133	dpavlin	552	# disable indexing when creating marc
134			$use_indexer = undef if ($marc_normalize);
135
136	dpavlin	213	my $total_rows = 0;
137	dpavlin	389	my $start_t = time();
138	dpavlin	213
139	dpavlin	529	my @links;
140			my $indexer;
141
142	dpavlin	556	my $lint = new MARC::Lint if ($marc_lint);
143
144	dpavlin	210	while (my ($database, $db_config) = each %{ $config->{databases} }) {
145	dpavlin	141
146	dpavlin	512	my ($only_database,$only_input) = split(m#/#, $only_filter) if ($only_filter);
147	dpavlin	510	next if ($only_database && $database !~ m/$only_database/i);
148	dpavlin	335
149	dpavlin	509	if ($use_indexer) {
150			my $indexer_config = $config->{$use_indexer} \|\| $log->logdie("can't find '$use_indexer' part in confguration");
151			$indexer_config->{database} = $database;
152			$indexer_config->{clean} = $clean;
153			$indexer_config->{label} = $db_config->{name};
154	dpavlin	431
155	dpavlin	509	if ($use_indexer eq 'hyperestraier') {
156	dpavlin	255
157	dpavlin	509	# open Hyper Estraier database
158			use WebPAC::Output::Estraier '0.10';
159			$indexer = new WebPAC::Output::Estraier( %{ $indexer_config } );
160
161			} elsif ($use_indexer eq 'kinosearch') {
162	dpavlin	430
163	dpavlin	509	# open KinoSearch
164			use WebPAC::Output::KinoSearch;
165			$indexer_config->{clean} = 1 unless (-e $indexer_config->{index_path});
166			$indexer = new WebPAC::Output::KinoSearch( %{ $indexer_config } );
167	dpavlin	431
168	dpavlin	509	} else {
169			$log->logdie("unknown use_indexer: $use_indexer");
170			}
171
172			$log->logide("can't continue without valid indexer") unless ($indexer);
173	dpavlin	430	}
174
175
176	dpavlin	255	#
177			# now WebPAC::Store
178			#
179	dpavlin	210	my $abs_path = abs_path($0);
180			$abs_path =~ s#/[^/]*$#/#;
181	dpavlin	141
182	dpavlin	210	my $db_path = $config->{webpac}->{db_path} . '/' . $database;
183	dpavlin	74
184	dpavlin	301	if ($clean) {
185	dpavlin	511	$log->info("creating new database '$database' in $db_path");
186	dpavlin	301	rmtree( $db_path ) \|\| $log->warn("can't remove $db_path: $!");
187			} else {
188	dpavlin	511	$log->info("working on database '$database' in $db_path");
189	dpavlin	301	}
190	dpavlin	255
191	dpavlin	210	my $db = new WebPAC::Store(
192			path => $db_path,
193	dpavlin	217	database => $database,
194	dpavlin	301	debug => $debug,
195	dpavlin	210	);
196	dpavlin	74
197	dpavlin	233
198	dpavlin	213	#
199			# now, iterate through input formats
200			#
201	dpavlin	74
202	dpavlin	213	my @inputs;
203			if (ref($db_config->{input}) eq 'ARRAY') {
204			@inputs = @{ $db_config->{input} };
205	dpavlin	255	} elsif ($db_config->{input}) {
206			push @inputs, $db_config->{input};
207	dpavlin	213	} else {
208	dpavlin	255	$log->info("database $database doesn't have inputs defined");
209	dpavlin	213	}
210	dpavlin	74
211	dpavlin	286	my @supported_inputs = keys %{ $config->{webpac}->{inputs} };
212
213	dpavlin	213	foreach my $input (@inputs) {
214	dpavlin	233
215	dpavlin	512	next if ($only_input && ($input->{name} !~ m#$only_input#i && $input->{type} !~ m#$only_input#i));
216	dpavlin	510
217	dpavlin	233	my $type = lc($input->{type});
218
219	dpavlin	286	die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs));
220	dpavlin	233
221			my $lookup = new WebPAC::Lookup(
222			lookup_file => $input->{lookup},
223	dpavlin	520	) if ($input->{lookup});
224	dpavlin	233
225	dpavlin	286	my $input_module = $config->{webpac}->{inputs}->{$type};
226	dpavlin	74
227	dpavlin	523	$log->info("working on input '$input->{name}' in $input->{path} [type: $input->{type}] using $input_module",
228			$input->{lookup} ? "lookup '$input->{lookup}'" : ""
229			);
230	dpavlin	286
231	dpavlin	287	my $input_db = new WebPAC::Input(
232			module => $input_module,
233	dpavlin	213	code_page => $config->{webpac}->{webpac_encoding},
234	dpavlin	301	limit => $limit \|\| $input->{limit},
235			offset => $offset,
236	dpavlin	251	lookup => $lookup,
237	dpavlin	416	recode => $input->{recode},
238	dpavlin	507	stats => $stats,
239	dpavlin	287	);
240	dpavlin	286	$log->logdie("can't create input using $input_module") unless ($input);
241	dpavlin	113
242	dpavlin	287	my $maxmfn = $input_db->open(
243	dpavlin	285	path => $input->{path},
244	dpavlin	213	code_page => $input->{encoding}, # database encoding
245	dpavlin	523	%{ $input },
246	dpavlin	213	);
247	dpavlin	113
248	dpavlin	539	my @norm_array = ref($input->{normalize}) eq 'ARRAY' ?
249			@{ $input->{normalize} } : ( $input->{normalize} );
250	dpavlin	210
251	dpavlin	552	if ($marc_normalize) {
252			@norm_array = ( {
253			path => $marc_normalize,
254			output => $marc_output \|\| 'out/marc/' . $database . '-' . $input->{name} . '.marc',
255			} );
256			}
257
258	dpavlin	539	foreach my $normalize (@norm_array) {
259	dpavlin	492
260	dpavlin	539	my $normalize_path = $normalize->{path} \|\| $log->logdie("can't find normalize path in config");
261	dpavlin	269
262	dpavlin	539	$log->logdie("Found '$normalize_path' as normalization file which isn't supported any more!") unless ( $normalize_path =~ m!\.pl$!i );
263	dpavlin	210
264	dpavlin	539	my $rules = read_file( $normalize_path ) or die "can't open $normalize_path: $!";
265	dpavlin	213
266	dpavlin	539	$log->info("Using $normalize_path for normalization...");
267	dpavlin	213
268	dpavlin	540	my $marc_fh;
269	dpavlin	552	if (my $path = $normalize->{output}) {
270	dpavlin	540	open($marc_fh, '>', $path) \|\|
271			$log->logdie("can't open MARC output $path: $!");
272
273	dpavlin	556	$log->info("Creating MARC export file $path", $marc_lint ? ' (with lint)' : '', "\n");
274	dpavlin	540	}
275
276	dpavlin	539	# reset position in database
277			$input_db->seek(1);
278	dpavlin	290
279	dpavlin	539	foreach my $pos ( 0 ... $input_db->size ) {
280	dpavlin	516
281	dpavlin	539	my $row = $input_db->fetch \|\| next;
282	dpavlin	516
283	dpavlin	539	my $mfn = $row->{'000'}->[0];
284	dpavlin	213
285	dpavlin	539	if (! $mfn \|\| $mfn !~ m#^\d+$#) {
286			$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos");
287			$mfn = $pos;
288			push @{ $row->{'000'} }, $pos;
289			}
290	dpavlin	508
291	dpavlin	213
292	dpavlin	539	if ($validate) {
293			my @errors = $validate->validate_errors( $row );
294			$log->error( "MFN $mfn validation errors:\n", join("\n", @errors) ) if (@errors);
295			}
296	dpavlin	213
297	dpavlin	539
298			my $ds = WebPAC::Normalize::data_structure(
299			row => $row,
300			rules => $rules,
301			lookup => $lookup ? $lookup->lookup_hash : undef,
302	dpavlin	541	marc_encoding => 'utf-8',
303	dpavlin	539	);
304	dpavlin	507
305	dpavlin	539	$db->save_ds(
306			id => $mfn,
307			ds => $ds,
308			prefix => $input->{name},
309			) if ($ds && !$stats);
310	dpavlin	210
311	dpavlin	539	$indexer->add(
312			id => $input->{name} . "/" . $mfn,
313			ds => $ds,
314			type => $config->{$use_indexer}->{type},
315	dpavlin	540	) if ($indexer && $ds);
316	dpavlin	539
317	dpavlin	540	if ($marc_fh) {
318			my $marc = new MARC::Record;
319	dpavlin	541	$marc->encoding( 'utf-8' );
320	dpavlin	556	my @marc_fields = WebPAC::Normalize::_get_marc_fields();
321			if (! @marc_fields) {
322			$log->warn("MARC record $mfn is empty, skipping");
323			} else {
324			$marc->add_fields( @marc_fields );
325			if ($marc_lint) {
326			$lint->check_record( $marc );
327			my $err = join( "\n", $lint->warnings );
328			$log->error("MARC lint detected warning on MFN $mfn\n",
329			"Original imput row: ",dump($row), "\n",
330			"Normalized MARC row: ",dump(@marc_fields), "\n",
331			"MARC lint warnings: ",$err,"\n"
332			) if ($err);
333			}
334			print $marc_fh $marc->as_usmarc;
335			}
336	dpavlin	540	}
337
338	dpavlin	539	$total_rows++;
339			}
340
341			$log->info("statistics of fields usage:\n", $input_db->stats) if ($stats);
342
343	dpavlin	540	# close MARC file
344			close($marc_fh) if ($marc_fh);
345	dpavlin	539
346	dpavlin	540	}
347
348	dpavlin	539	}
349
350	dpavlin	509	eval { $indexer->finish } if ($indexer && $indexer->can('finish'));
351	dpavlin	434
352	dpavlin	389	my $dt = time() - $start_t;
353	dpavlin	511	$log->info("$total_rows records ", $indexer ? "indexed " : "",
354			sprintf("in %.2f sec [%.2f rec/sec]",
355	dpavlin	389	$dt, ($total_rows / $dt)
356			)
357			);
358	dpavlin	255
359	dpavlin	431	#
360			# add Hyper Estraier links to other databases
361			#
362	dpavlin	552	if (ref($db_config->{links}) eq 'ARRAY' && $use_indexer) {
363	dpavlin	431	foreach my $link (@{ $db_config->{links} }) {
364			if ($use_indexer eq 'hyperestraier') {
365	dpavlin	529	$log->info("saving link $database -> $link->{to} [$link->{credit}]");
366			push @links, {
367	dpavlin	430	from => $database,
368			to => $link->{to},
369			credit => $link->{credit},
370	dpavlin	529	};
371	dpavlin	431	} else {
372			$log->warn("NOT IMPLEMENTED WITH $use_indexer: adding link $database -> $link->{to} [$link->{credit}]");
373	dpavlin	430	}
374	dpavlin	255	}
375			}
376
377	dpavlin	210	}
378
379	dpavlin	529	foreach my $link (@links) {
380			$log->info("adding link $link->{from} -> $link->{to} [$link->{credit}]");
381			$indexer->add_link( %{ $link } );
382			}