|
Line 0
Link Here
|
| 0 |
- |
1 |
#!/usr/bin/perl |
|
|
2 |
use Modern::Perl; |
| 3 |
use C4::Context; |
| 4 |
use C4::AuthoritiesMarc; |
| 5 |
use C4::Biblio; |
| 6 |
use C4::Search; |
| 7 |
use C4::Charset; |
| 8 |
use C4::Heading; |
| 9 |
use Koha::SearchEngine; |
| 10 |
use Koha::SearchEngine::QueryBuilder; |
| 11 |
use Koha::Logger; |
| 12 |
|
| 13 |
use Koha::Authorities; |
| 14 |
|
| 15 |
use Getopt::Long; |
| 16 |
use YAML; |
| 17 |
use List::MoreUtils qw/uniq/; |
| 18 |
use Pod::Usage qw( pod2usage ); |
| 19 |
|
| 20 |
=head1 NAME |
| 21 |
|
| 22 |
misc/migration_tools/dedup_authorities.pl - Deduping authorities script |
| 23 |
|
| 24 |
=head1 SYNOPSIS |
| 25 |
|
| 26 |
dedup_authorities.pl [ -h ] [ -where="authid < 5000" ] -c [ -v ] [ -m d ] [ -a PERSO_NAME ] |
| 27 |
|
| 28 |
Options: |
| 29 |
-h --help display usage statement |
| 30 |
-v --verbose increase verbosity, can be repeated for greater verbosity |
| 31 |
-m --method method for choosing the reference authority, can be: date, used, or ppn (UNIMARC) |
| 32 |
can be repeated |
| 33 |
-w --where a SQL WHERE statement to limit the authority records checked |
| 34 |
-c --confirm without this parameter no changes will be made, script will run in test mode |
| 35 |
-a --authtypecode check only specified auth type, repeatable |
| 36 |
|
| 37 |
=head1 OPTIONS |
| 38 |
|
| 39 |
=over |
| 40 |
|
| 41 |
=item B<--method> |
| 42 |
|
| 43 |
Method(s) used to choose which authority to keep in case we found |
| 44 |
duplicates. |
| 45 |
<methods> is a string composed of letters describing what methods to use |
| 46 |
and in which order. |
| 47 |
Letters can be: |
| 48 |
date: keep the most recent authority (based on 005 field) |
| 49 |
used: keep the most used authority |
| 50 |
ppn: PPN (UNIMARC only), keep the authority with a ppn (when some |
| 51 |
authorities don't have one, based on 009 field) |
| 52 |
|
| 53 |
Example: |
| 54 |
-m ppn -m date -m used |
| 55 |
Among the authorities that have a PPN, keep the most recent, |
| 56 |
and if two (or more) have the same date in 005, keep the |
| 57 |
most used. |
| 58 |
|
| 59 |
Default is 'used' |
| 60 |
|
| 61 |
=item B<--where> |
| 62 |
|
| 63 |
limit the deduplication to SOME authorities only |
| 64 |
|
| 65 |
Example: |
| 66 |
-where="authid < 5000" |
| 67 |
will only auths with a low auth_id (old records) |
| 68 |
|
| 69 |
=item B<--verbose> |
| 70 |
|
| 71 |
display verbose loggin, can be repeated twice for more info |
| 72 |
|
| 73 |
|
| 74 |
=item B<--help> |
| 75 |
|
| 76 |
show usage information. |
| 77 |
|
| 78 |
=back |
| 79 |
|
| 80 |
=cut |
| 81 |
|
| 82 |
my @methods; |
| 83 |
my @authtypecodes; |
| 84 |
my $help = 0; |
| 85 |
my $confirm = 0; |
| 86 |
my $verbose = 0; |
| 87 |
my $wherestring = ""; |
| 88 |
my $debug = 0; |
| 89 |
|
| 90 |
my $result = GetOptions( |
| 91 |
"d|debug" => \$debug, |
| 92 |
"v|verbose+" => \$verbose, |
| 93 |
"c|confirm" => \$confirm, |
| 94 |
"h|help" => \$help, |
| 95 |
"w|where=s" => \$wherestring, |
| 96 |
"m|method=s" => \@methods, |
| 97 |
"a|authtypecode=s" => \@authtypecodes |
| 98 |
); |
| 99 |
|
| 100 |
pod2usage( -verbose => 2 ) if ($help); |
| 101 |
|
| 102 |
print "RUNNING IN TEST MODE, NO CHANGES WILL BE MADE\n" unless $confirm; |
| 103 |
$verbose = 1 unless ( $confirm || $verbose ); |
| 104 |
|
| 105 |
my @choose_subs; |
| 106 |
@methods = ('used') unless @methods; |
| 107 |
foreach my $method (@methods) { |
| 108 |
if ( $method eq 'date' ) { |
| 109 |
push @choose_subs, \&_get_date; |
| 110 |
} elsif ( $method eq 'ppn' ) { |
| 111 |
die 'PPN method is only valid for UNIMARC' |
| 112 |
unless ( C4::Context->preference('marcflavour') eq 'UNIMARC' ); |
| 113 |
push @choose_subs, \&_has_ppn; |
| 114 |
} elsif ( $method eq 'used' ) { |
| 115 |
push @choose_subs, \&_get_usage; |
| 116 |
} else { |
| 117 |
warn "Choose method '$method' is not supported"; |
| 118 |
} |
| 119 |
} |
| 120 |
|
| 121 |
my $dbh = C4::Context->dbh; |
| 122 |
|
| 123 |
$verbose and print "Fetching authtypecodes...\n"; |
| 124 |
my $params = undef; |
| 125 |
if (@authtypecodes) { |
| 126 |
$params = { authtypecode => { -in => \@authtypecodes } }; |
| 127 |
} |
| 128 |
my @auth_types = Koha::Authority::Types->search($params)->as_list; |
| 129 |
my %auth_match_headings = |
| 130 |
map { $_->authtypecode => $_->auth_tag_to_report } @auth_types; |
| 131 |
$verbose and print "Fetching authtypecodes done.\n"; |
| 132 |
|
| 133 |
my %biblios; |
| 134 |
my $seen; |
| 135 |
|
| 136 |
for my $authtype (@auth_types) { |
| 137 |
my $authtypecode = $authtype->authtypecode; |
| 138 |
my %duplicated; |
| 139 |
my $deleted = 0; |
| 140 |
my $updated_bibs = 0; |
| 141 |
my $i = 0; |
| 142 |
$verbose and print "Deduping authtype '$authtypecode' \n"; |
| 143 |
|
| 144 |
$verbose and print "Fetching authorities for '$authtypecode'... "; |
| 145 |
my $authorities = Koha::Authorities->search( { authtypecode => $authtypecode } ); |
| 146 |
$authorities = $authorities->search( \$wherestring ) if $wherestring; |
| 147 |
my $size = $authorities->count; |
| 148 |
$verbose and print "$size authorities found\n"; |
| 149 |
|
| 150 |
while ( my $authority = $authorities->next ) { |
| 151 |
next if defined $seen->{ $authority->authid }; |
| 152 |
$seen->{ $authority->authid } = 1; |
| 153 |
$i++; |
| 154 |
if ( $verbose >= 2 ) { |
| 155 |
my $percentage = sprintf( "%.2f", $i * 100 / $size ); |
| 156 |
print "Processing authority " . $authority->authid . " ($i/$size $percentage%)\n"; |
| 157 |
} elsif ( $verbose and ( $i % 100 ) == 0 ) { |
| 158 |
my $percentage = sprintf( "%.2f", $i * 100 / $size ); |
| 159 |
print "Progression for authtype '$authtypecode': $i/$size ($percentage%)\n"; |
| 160 |
} |
| 161 |
|
| 162 |
#authority was marked as duplicate |
| 163 |
next if defined $duplicated{ $authority->authid }; |
| 164 |
my $authrecord = C4::AuthoritiesMarc::GetAuthority( $authority->authid ); |
| 165 |
|
| 166 |
next unless $authrecord; |
| 167 |
C4::Charset::SetUTF8Flag($authrecord); |
| 168 |
|
| 169 |
$debug and print " Building query...\n"; |
| 170 |
my $field = $authrecord->field( $auth_match_headings{$authtypecode} ); |
| 171 |
unless ($field) { |
| 172 |
warn " Malformed authority record, no heading"; |
| 173 |
next; |
| 174 |
} |
| 175 |
unless ( $field->as_string ) { |
| 176 |
warn " Malformed authority record, blank heading"; |
| 177 |
next; |
| 178 |
} |
| 179 |
my $heading = C4::Heading->new_from_field( $field, undef, 1 ); #new auth heading |
| 180 |
my $search_term = $heading->search_form; |
| 181 |
$debug and print " Building query done\n"; |
| 182 |
$debug and print " $search_term\n"; |
| 183 |
|
| 184 |
$debug and print " Searching..."; |
| 185 |
|
| 186 |
my $builder = Koha::SearchEngine::QueryBuilder->new( { index => $Koha::SearchEngine::AUTHORITIES_INDEX } ); |
| 187 |
my $searcher = Koha::SearchEngine::Search->new( { index => $Koha::SearchEngine::AUTHORITIES_INDEX } ); |
| 188 |
my $query = $builder->build_authorities_query_compat( |
| 189 |
['match-heading'], [''], |
| 190 |
[''], ['exact'], [$search_term], $authtypecode, '' |
| 191 |
); |
| 192 |
my ( $results, $total ) = $searcher->search_auth_compat( $query, 0, 50, undef ); |
| 193 |
if ( !$results ) { |
| 194 |
$debug and warn " " . $@; |
| 195 |
$debug and warn " " . YAML::Dump($search_term); |
| 196 |
$debug and warn " " . $field->as_string; |
| 197 |
next; |
| 198 |
} |
| 199 |
|
| 200 |
$debug and warn " " . YAML::Dump($results); |
| 201 |
|
| 202 |
my @recordids = |
| 203 |
map { $_->{authid} != $authority->authid ? $_->{authid} : () } @$results; |
| 204 |
if ( !$results || scalar(@$results) < 1 || scalar @recordids < 1 ) { |
| 205 |
( $verbose >= 2 ) |
| 206 |
and print ' No duplicates found for ' . $heading->display_form . "\n"; |
| 207 |
next; |
| 208 |
} |
| 209 |
map { $seen->{$_} = 1 } @recordids; |
| 210 |
$debug and print " Searching done.\n"; |
| 211 |
|
| 212 |
$debug and print " Choosing records..."; |
| 213 |
my ( $recordid_to_keep, @recordids_to_merge ) = _choose_records( $authority->authid, @recordids ); |
| 214 |
$debug and print " Choosing records done.\n"; |
| 215 |
unless ( !$confirm or @recordids_to_merge == 0 ) { |
| 216 |
( $verbose >= 2 ) |
| 217 |
and print " Merging " . join( ',', @recordids_to_merge ) . " into $recordid_to_keep.\n"; |
| 218 |
for my $localauthid (@recordids_to_merge) { |
| 219 |
next if $recordid_to_keep == $localauthid; |
| 220 |
my $MARCto = C4::AuthoritiesMarc::GetAuthority($recordid_to_keep); |
| 221 |
my $editedbiblios = 0; |
| 222 |
eval { |
| 223 |
$editedbiblios = C4::AuthoritiesMarc::merge( |
| 224 |
{ |
| 225 |
mergefrom => $localauthid, |
| 226 |
MARCfrom => undef, |
| 227 |
mergeto => $recordid_to_keep, |
| 228 |
MARCto => $MARCto |
| 229 |
} |
| 230 |
); |
| 231 |
}; |
| 232 |
if ($@) { |
| 233 |
warn " Merging $localauthid into $recordid_to_keep failed :", |
| 234 |
$@; |
| 235 |
} else { |
| 236 |
print " Updated " . $editedbiblios . " biblios\n"; |
| 237 |
$updated_bibs += $editedbiblios; |
| 238 |
$duplicated{$localauthid} = 2; |
| 239 |
print " Deleting $localauthid\n"; |
| 240 |
C4::AuthoritiesMarc::DelAuthority( { authid => $localauthid, skip_merge => 1 } ); |
| 241 |
$deleted++; |
| 242 |
} |
| 243 |
} |
| 244 |
( $verbose >= 2 ) and print " Merge done.\n"; |
| 245 |
$duplicated{$recordid_to_keep} = 1; |
| 246 |
} elsif ( $verbose >= 2 ) { |
| 247 |
if ( @recordids_to_merge > 0 ) { |
| 248 |
print ' Would merge ' |
| 249 |
. join( ',', @recordids_to_merge ) |
| 250 |
. " into $recordid_to_keep (" |
| 251 |
. $heading->display_form . ")\n"; |
| 252 |
} else { |
| 253 |
print " No duplicates found for $recordid_to_keep\n"; |
| 254 |
} |
| 255 |
} |
| 256 |
} |
| 257 |
$verbose and print "End of deduping for authtype '$authtypecode'\n"; |
| 258 |
$verbose and print "Updated $updated_bibs biblios\n"; |
| 259 |
$verbose and print "Deleted $deleted authorities\n"; |
| 260 |
} |
| 261 |
|
| 262 |
# Update biblios |
| 263 |
my @biblios_to_update = grep { defined $biblios{$_} and $biblios{$_} == 1 } |
| 264 |
keys %biblios; |
| 265 |
if ( @biblios_to_update > 0 ) { |
| 266 |
} else { |
| 267 |
print "No biblios to update\n"; |
| 268 |
} |
| 269 |
|
| 270 |
exit 0; |
| 271 |
|
| 272 |
sub _get_id { |
| 273 |
my $record = shift; |
| 274 |
|
| 275 |
if ( $record and ( my $field = $record->field('001') ) ) { |
| 276 |
return $field->data(); |
| 277 |
} |
| 278 |
return 0; |
| 279 |
} |
| 280 |
|
| 281 |
sub _has_ppn { |
| 282 |
my $record = shift; |
| 283 |
|
| 284 |
if ( $record and ( my $field = $record->field('009') ) ) { |
| 285 |
return $field->data() ? 1 : 0; |
| 286 |
} |
| 287 |
return 0; |
| 288 |
} |
| 289 |
|
| 290 |
sub _get_date { |
| 291 |
my $record = shift; |
| 292 |
|
| 293 |
if ( $record and ( my $field = $record->field('005') ) ) { |
| 294 |
return $field->data(); |
| 295 |
} |
| 296 |
return 0; |
| 297 |
} |
| 298 |
|
| 299 |
sub _get_usage { |
| 300 |
my $record = shift; |
| 301 |
|
| 302 |
if ( $record and ( my $field = $record->field('001') ) ) { |
| 303 |
return Koha::Authorities->get_usage_count( { authid => $field->data() } ); |
| 304 |
} |
| 305 |
return 0; |
| 306 |
} |
| 307 |
|
| 308 |
=head2 _choose_records |
| 309 |
this function takes input of candidate record ids to merging |
| 310 |
and returns |
| 311 |
first the record to merge to |
| 312 |
and list of records to merge from |
| 313 |
=cut |
| 314 |
|
| 315 |
sub _choose_records { |
| 316 |
my @recordids = @_; |
| 317 |
|
| 318 |
my @records = map { C4::AuthoritiesMarc::GetAuthority($_) } @recordids; |
| 319 |
my @candidate_auths = @records; |
| 320 |
|
| 321 |
# See http://www.sysarch.com/Perl/sort_paper.html Schwartzian transform |
| 322 |
my @candidate_authids = |
| 323 |
map $_->[0] => sort { $b->[1] <=> $a->[1] || $b->[2] <=> $a->[2] || $b->[3] <=> $a->[3] } |
| 324 |
map [ _get_id($_), |
| 325 |
$choose_subs[0] ? $choose_subs[0]->($_) : 0, |
| 326 |
$choose_subs[1] ? $choose_subs[1]->($_) : 0, |
| 327 |
$choose_subs[2] ? $choose_subs[2]->($_) : 0 ] => ( $records[0], @candidate_auths ); |
| 328 |
|
| 329 |
return @candidate_authids; |
| 330 |
} |