10 |
tag search display |
tag search display |
11 |
marc marc_indicators marc_repeatable_subfield |
marc marc_indicators marc_repeatable_subfield |
12 |
marc_compose marc_leader |
marc_compose marc_leader |
13 |
marc_duplicate marc_remove |
marc_duplicate marc_remove marc_count |
14 |
marc_original_order |
marc_original_order |
15 |
|
|
16 |
rec1 rec2 rec |
rec1 rec2 rec |
19 |
save_into_lookup |
save_into_lookup |
20 |
|
|
21 |
split_rec_on |
split_rec_on |
22 |
|
|
23 |
|
get set |
24 |
|
count |
25 |
/; |
/; |
26 |
|
|
27 |
use warnings; |
use warnings; |
42 |
|
|
43 |
=head1 VERSION |
=head1 VERSION |
44 |
|
|
45 |
Version 0.22 |
Version 0.27 |
46 |
|
|
47 |
=cut |
=cut |
48 |
|
|
49 |
our $VERSION = '0.22'; |
our $VERSION = '0.27'; |
50 |
|
|
51 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
52 |
|
|
167 |
|
|
168 |
=cut |
=cut |
169 |
|
|
170 |
my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators); |
my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader); |
171 |
my ($marc_record_offset, $marc_fetch_offset) = (0, 0); |
my ($marc_record_offset, $marc_fetch_offset) = (0, 0); |
172 |
|
|
173 |
sub _get_ds { |
sub _get_ds { |
184 |
|
|
185 |
sub _clean_ds { |
sub _clean_ds { |
186 |
my $a = {@_}; |
my $a = {@_}; |
187 |
($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators) = (); |
($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader) = (); |
188 |
($marc_record_offset, $marc_fetch_offset) = (0,0); |
($marc_record_offset, $marc_fetch_offset) = (0,0); |
189 |
$marc_encoding = $a->{marc_encoding}; |
$marc_encoding = $a->{marc_encoding}; |
190 |
} |
} |
289 |
|
|
290 |
=cut |
=cut |
291 |
|
|
292 |
|
my $fetch_pos; |
293 |
|
|
294 |
sub _get_marc_fields { |
sub _get_marc_fields { |
295 |
|
|
296 |
my $arg = {@_}; |
my $arg = {@_}; |
297 |
warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2); |
warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2); |
298 |
my $offset = $marc_fetch_offset; |
$fetch_pos = $marc_fetch_offset; |
299 |
if ($arg->{offset}) { |
if ($arg->{offset}) { |
300 |
$offset = $arg->{offset}; |
$fetch_pos = $arg->{offset}; |
301 |
} elsif($arg->{fetch_next}) { |
} elsif($arg->{fetch_next}) { |
302 |
$marc_fetch_offset++; |
$marc_fetch_offset++; |
303 |
} |
} |
306 |
|
|
307 |
warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2); |
warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2); |
308 |
|
|
309 |
my $marc_rec = $marc_record->[ $offset ]; |
my $marc_rec = $marc_record->[ $fetch_pos ]; |
310 |
|
|
311 |
warn "## _get_marc_fields (at offset: $offset) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1); |
warn "## _get_marc_fields (at offset: $fetch_pos) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1); |
312 |
|
|
313 |
return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0); |
return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0); |
314 |
|
|
329 |
|
|
330 |
if ($debug) { |
if ($debug) { |
331 |
warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield ); |
warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield ); |
332 |
warn "## marc_record[$offset] = ", dump( $marc_rec ), $/; |
warn "## marc_record[$fetch_pos] = ", dump( $marc_rec ), $/; |
333 |
warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/; |
warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/; |
334 |
warn "## subfield count = ", dump( $u ), $/; |
warn "## subfield count = ", dump( $u ), $/; |
335 |
} |
} |
410 |
return \@m; |
return \@m; |
411 |
} |
} |
412 |
|
|
413 |
|
=head2 _get_marc_leader |
414 |
|
|
415 |
|
Return leader from currently fetched record by L</_get_marc_fields> |
416 |
|
|
417 |
|
print WebPAC::Normalize::_get_marc_leader(); |
418 |
|
|
419 |
|
=cut |
420 |
|
|
421 |
|
sub _get_marc_leader { |
422 |
|
die "no fetch_pos, did you called _get_marc_fields first?" unless ( defined( $fetch_pos ) ); |
423 |
|
return $marc_leader->[ $fetch_pos ]; |
424 |
|
} |
425 |
|
|
426 |
=head2 _debug |
=head2 _debug |
427 |
|
|
428 |
Change level of debug warnings |
Change level of debug warnings |
505 |
my ($offset,$value) = @_; |
my ($offset,$value) = @_; |
506 |
|
|
507 |
if ($offset) { |
if ($offset) { |
508 |
$out->{' leader'}->{ $offset } = $value; |
$marc_leader->[ $marc_record_offset ]->{ $offset } = $value; |
509 |
} else { |
} else { |
510 |
return $out->{' leader'}; |
|
511 |
|
if (defined($marc_leader)) { |
512 |
|
die "marc_leader not array = ", dump( $marc_leader ) unless (ref($marc_leader) eq 'ARRAY'); |
513 |
|
return $marc_leader->[ $marc_record_offset ]; |
514 |
|
} else { |
515 |
|
return; |
516 |
|
} |
517 |
} |
} |
518 |
} |
} |
519 |
|
|
645 |
my $m = $marc_record->[ -1 ]; |
my $m = $marc_record->[ -1 ]; |
646 |
die "can't duplicate record which isn't defined" unless ($m); |
die "can't duplicate record which isn't defined" unless ($m); |
647 |
push @{ $marc_record }, dclone( $m ); |
push @{ $marc_record }, dclone( $m ); |
648 |
warn "## marc_duplicate = ", dump(@$marc_record), $/ if ($debug > 1); |
push @{ $marc_leader }, dclone( marc_leader() ); |
649 |
|
warn "## marc_duplicate = ", dump(@$marc_leader, @$marc_record), $/ if ($debug > 1); |
650 |
$marc_record_offset = $#{ $marc_record }; |
$marc_record_offset = $#{ $marc_record }; |
651 |
warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1); |
warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1); |
652 |
|
|
653 |
} |
} |
654 |
|
|
655 |
=head2 marc_remove |
=head2 marc_remove |
661 |
|
|
662 |
This will erase field C<200> or C<200^a> from current MARC record. |
This will erase field C<200> or C<200^a> from current MARC record. |
663 |
|
|
664 |
|
marc_remove('*'); |
665 |
|
|
666 |
|
Will remove all fields in current MARC record. |
667 |
|
|
668 |
This is useful after calling C<marc_duplicate> or on it's own (but, you |
This is useful after calling C<marc_duplicate> or on it's own (but, you |
669 |
should probably just remove that subfield definition if you are not |
should probably just remove that subfield definition if you are not |
670 |
using C<marc_duplicate>). |
using C<marc_duplicate>). |
682 |
|
|
683 |
warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2); |
warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2); |
684 |
|
|
685 |
my $i = 0; |
if ($f eq '*') { |
686 |
foreach ( 0 .. $#{ $marc } ) { |
|
687 |
last unless (defined $marc->[$i]); |
delete( $marc_record->[ $marc_record_offset ] ); |
688 |
warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3); |
warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1); |
689 |
if ($marc->[$i]->[0] eq $f) { |
|
690 |
if (! defined $sf) { |
} else { |
691 |
# remove whole field |
|
692 |
splice @$marc, $i, 1; |
my $i = 0; |
693 |
warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3); |
foreach ( 0 .. $#{ $marc } ) { |
694 |
$i--; |
last unless (defined $marc->[$i]); |
695 |
} else { |
warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3); |
696 |
foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) { |
if ($marc->[$i]->[0] eq $f) { |
697 |
my $o = ($j * 2) + 3; |
if (! defined $sf) { |
698 |
if ($marc->[$i]->[$o] eq $sf) { |
# remove whole field |
699 |
# remove subfield |
splice @$marc, $i, 1; |
700 |
splice @{$marc->[$i]}, $o, 2; |
warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3); |
701 |
warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3); |
$i--; |
702 |
# is record now empty? |
} else { |
703 |
if ($#{ $marc->[$i] } == 2) { |
foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) { |
704 |
splice @$marc, $i, 1; |
my $o = ($j * 2) + 3; |
705 |
warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3); |
if ($marc->[$i]->[$o] eq $sf) { |
706 |
$i--; |
# remove subfield |
707 |
}; |
splice @{$marc->[$i]}, $o, 2; |
708 |
|
warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3); |
709 |
|
# is record now empty? |
710 |
|
if ($#{ $marc->[$i] } == 2) { |
711 |
|
splice @$marc, $i, 1; |
712 |
|
warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3); |
713 |
|
$i--; |
714 |
|
}; |
715 |
|
} |
716 |
} |
} |
717 |
} |
} |
718 |
} |
} |
719 |
|
$i++; |
720 |
} |
} |
|
$i++; |
|
|
} |
|
721 |
|
|
722 |
warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2); |
warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2); |
723 |
|
|
724 |
$marc_record->[ $marc_record_offset ] = $marc; |
$marc_record->[ $marc_record_offset ] = $marc; |
725 |
|
} |
726 |
|
|
727 |
warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1); |
warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1); |
728 |
} |
} |
797 |
warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1); |
warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1); |
798 |
} |
} |
799 |
|
|
800 |
|
=head2 marc_count |
801 |
|
|
802 |
|
Return number of MARC records created using L</marc_duplicate>. |
803 |
|
|
804 |
|
print "created ", marc_count(), " records"; |
805 |
|
|
806 |
|
=cut |
807 |
|
|
808 |
|
sub marc_count { |
809 |
|
return $#{ $marc_record }; |
810 |
|
} |
811 |
|
|
812 |
|
|
813 |
=head1 Functions to extract data from input |
=head1 Functions to extract data from input |
814 |
|
|
930 |
@v = rec('200') |
@v = rec('200') |
931 |
@v = rec('200','a') |
@v = rec('200','a') |
932 |
|
|
933 |
|
If rec() returns just single value, it will |
934 |
|
return scalar, not array. |
935 |
|
|
936 |
=cut |
=cut |
937 |
|
|
938 |
sub rec { |
sub rec { |
942 |
} elsif ($#_ == 1) { |
} elsif ($#_ == 1) { |
943 |
@out = rec2(@_); |
@out = rec2(@_); |
944 |
} |
} |
945 |
if (@out) { |
if ($#out == 0 && ! wantarray) { |
946 |
|
return $out[0]; |
947 |
|
} elsif (@out) { |
948 |
return @out; |
return @out; |
949 |
} else { |
} else { |
950 |
return ''; |
return ''; |
1072 |
sub lookup { |
sub lookup { |
1073 |
my ($what, $database, $input, $key, $having) = @_; |
my ($what, $database, $input, $key, $having) = @_; |
1074 |
|
|
1075 |
confess "lookup needs 5 arguments: what, database, input, key, having" unless ($#_ == 4); |
confess "lookup needs 5 arguments: what, database, input, key, having\n" unless ($#_ == 4); |
1076 |
|
|
1077 |
warn "## lookup ($database, $input, $key)", $/ if ($debug > 1); |
warn "## lookup ($database, $input, $key)", $/ if ($debug > 1); |
1078 |
return unless (defined($lookup->{$database}->{$input}->{$key})); |
return unless (defined($lookup->{$database}->{$input}->{$key})); |
1086 |
|
|
1087 |
foreach my $h ( @having ) { |
foreach my $h ( @having ) { |
1088 |
if (defined($lookup->{$database}->{$input}->{$key}->{$h})) { |
if (defined($lookup->{$database}->{$input}->{$key}->{$h})) { |
1089 |
warn "lookup for $database/$input/$key/$h return ",dump($lookup->{$database}->{$input}->{$key}->{$h}),"\n"; |
warn "lookup for $database/$input/$key/$h return ",dump($lookup->{$database}->{$input}->{$key}->{$h}),"\n" if ($debug); |
1090 |
$mfns->{$_}++ foreach keys %{ $lookup->{$database}->{$input}->{$key}->{$h} }; |
$mfns->{$_}++ foreach keys %{ $lookup->{$database}->{$input}->{$key}->{$h} }; |
1091 |
} |
} |
1092 |
} |
} |
1095 |
|
|
1096 |
my @mfns = sort keys %$mfns; |
my @mfns = sort keys %$mfns; |
1097 |
|
|
1098 |
warn "# lookup loading $database/$input/$key mfn ", join(",",@mfns)," having ",dump(@having),"\n"; |
warn "# lookup loading $database/$input/$key mfn ", join(",",@mfns)," having ",dump(@having),"\n" if ($debug); |
1099 |
|
|
1100 |
my $old_rec = $rec; |
my $old_rec = $rec; |
1101 |
my @out; |
my @out; |
1103 |
foreach my $mfn (@mfns) { |
foreach my $mfn (@mfns) { |
1104 |
$rec = $load_row_coderef->( $database, $input, $mfn ); |
$rec = $load_row_coderef->( $database, $input, $mfn ); |
1105 |
|
|
1106 |
warn "got $database/$input/$mfn = ", dump($rec), $/; |
warn "got $database/$input/$mfn = ", dump($rec), $/ if ($debug); |
1107 |
|
|
1108 |
my @vals = $what->(); |
my @vals = $what->(); |
1109 |
|
|
1110 |
push @out, ( @vals ); |
push @out, ( @vals ); |
1111 |
|
|
1112 |
warn "lookup for mfn $mfn returned ", dump(@vals), $/; |
warn "lookup for mfn $mfn returned ", dump(@vals), $/ if ($debug); |
1113 |
} |
} |
1114 |
|
|
1115 |
# if (ref($lookup->{$k}) eq 'ARRAY') { |
# if (ref($lookup->{$k}) eq 'ARRAY') { |
1120 |
|
|
1121 |
$rec = $old_rec; |
$rec = $old_rec; |
1122 |
|
|
1123 |
warn "## lookup returns = ", dump(@out), $/; |
warn "## lookup returns = ", dump(@out), $/ if ($debug); |
1124 |
|
|
1125 |
return @out; |
if ($#out == 0) { |
1126 |
|
return $out[0]; |
1127 |
|
} else { |
1128 |
|
return @out; |
1129 |
|
} |
1130 |
} |
} |
1131 |
|
|
1132 |
=head2 save_into_lookup |
=head2 save_into_lookup |
1298 |
} |
} |
1299 |
} |
} |
1300 |
|
|
1301 |
|
my $hash; |
1302 |
|
|
1303 |
|
=head2 set |
1304 |
|
|
1305 |
|
set( key => 'value' ); |
1306 |
|
|
1307 |
|
=cut |
1308 |
|
|
1309 |
|
sub set { |
1310 |
|
my ($k,$v) = @_; |
1311 |
|
warn "## set ( $k => ", dump($v), " )", $/ if ( $debug ); |
1312 |
|
$hash->{$k} = $v; |
1313 |
|
}; |
1314 |
|
|
1315 |
|
=head2 get |
1316 |
|
|
1317 |
|
get( 'key' ); |
1318 |
|
|
1319 |
|
=cut |
1320 |
|
|
1321 |
|
sub get { |
1322 |
|
my $k = shift || return; |
1323 |
|
my $v = $hash->{$k}; |
1324 |
|
warn "## get $k = ", dump( $v ), $/ if ( $debug ); |
1325 |
|
return $v; |
1326 |
|
} |
1327 |
|
|
1328 |
|
=head2 count |
1329 |
|
|
1330 |
|
if ( count( @result ) == 1 ) { |
1331 |
|
# do something if only 1 result is there |
1332 |
|
} |
1333 |
|
|
1334 |
|
=cut |
1335 |
|
|
1336 |
|
sub count { |
1337 |
|
warn "## count ",dump(@_),$/ if ( $debug ); |
1338 |
|
return @_ . ''; |
1339 |
|
} |
1340 |
|
|
1341 |
# END |
# END |
1342 |
1; |
1; |