package WebPAC::Normalize;
+use Exporter 'import';
+@EXPORT = qw/
+ _set_rec _set_lookup
+ _get_ds _clean_ds
+ _debug
+ _pack_subfields_hash
+
+ tag search display
+ marc marc_indicators marc_repeatable_subfield
+ marc_compose marc_leader
+ marc_duplicate marc_remove
+ marc_original_order
+
+ rec1 rec2 rec
+ regex prefix suffix surround
+ first lookup join_with
+ save_into_lookup
+
+ split_rec_on
+/;
use warnings;
use strict;
-use base 'WebPAC::Common';
-use Data::Dumper;
+
+#use base qw/WebPAC::Common/;
+use Data::Dump qw/dump/;
+use Storable qw/dclone/;
+
+# debugging warn(s)
+my $debug = 0;
+
=head1 NAME
-WebPAC::Normalize - data mungling for normalisation
+WebPAC::Normalize - describe normalisaton rules using sets
=head1 VERSION
-Version 0.02
+Version 0.20
=cut
-our $VERSION = '0.02';
+our $VERSION = '0.20';
=head1 SYNOPSIS
-This package contains code that mungle data to produce normalized format.
+This module uses C<conf/normalize/*.pl> files to perform normalisation
+from input records using perl functions which are specialized for set
+processing.
+
+Sets are implemented as arrays, and normalisation file is valid perl, which
+means that you check it's validity before running WebPAC using
+C<perl -c normalize.pl>.
-It contains several assumptions:
+Normalisation can generate multiple output normalized data. For now, supported output
+types (on the left side of definition) are: C<tag>, C<display>, C<search> and
+C<marc>.
-=over
+=head1 FUNCTIONS
+
+Functions which start with C<_> are private and used by WebPAC internally.
+All other functions are available for use within normalisation rules.
-=item *
+=head2 data_structure
-format of fields is defined using C<v123^a> notation for repeatable fields
-or C<s123^a> for single (or first) value, where C<123> is field number and
-C<a> is subfield.
+Return data structure
-=item *
+ my $ds = WebPAC::Normalize::data_structure(
+ lookup => $lookup_variable,
+ row => $row,
+ rules => $normalize_pl_config,
+ marc_encoding => 'utf-8',
+ config => $config,
+ );
-source data records (C<$rec>) have unique identifiers in field C<000>
+Options C<row>, C<rules> and C<log> are mandatory while all
+other are optional.
-=item *
+This function will B<die> if normalizastion can't be evaled.
-optional C<eval{length('v123^a') == 3}> tag at B<beginning of format> will be
-perl code that is evaluated before producing output (value of field will be
-interpolated before that)
+Since this function isn't exported you have to call it with
+C<WebPAC::Normalize::data_structure>.
-=item *
+=cut
-optional C<filter{filter_name}> at B<begining of format> will apply perl
-code defined as code ref on format after field substitution to producing
-output
+sub data_structure {
+ my $arg = {@_};
-=item *
+ die "need row argument" unless ($arg->{row});
+ die "need normalisation argument" unless ($arg->{rules});
-optional C<lookup{...}> will be then performed. See C<WebPAC::Lookups>.
+ no strict 'subs';
+ _set_lookup( $arg->{lookup} ) if (defined( $arg->{lookup} ));
+ _set_rec( $arg->{row} );
+ _set_config( $arg->{config} ) if (defined( $arg->{config} ));
+ _clean_ds( %{ $arg } );
+ eval "$arg->{rules}";
+ die "error evaling $arg->{rules}: $@\n" if ($@);
-=item *
+ return _get_ds();
+}
-at end, optional C<format>s rules are resolved. Format rules are similar to
-C<sprintf> and can also contain C<lookup{...}> which is performed after
-values are inserted in format.
+=head2 _set_rec
-=back
+Set current record hash
-This also describes order in which transformations are applied (eval,
-filter, lookup, format) which is important to undestand when deciding how to
-solve your data mungling and normalisation process.
+ _set_rec( $rec );
+=cut
+my $rec;
+sub _set_rec {
+ $rec = shift or die "no record hash";
+}
-=head1 FUNCTIONS
+=head2 _set_config
-=head2 new
+Set current config hash
-Create new normalisation object
+ _set_config( $config );
- my $n = new WebPAC::Normalize::Something(
- filter => {
- 'filter_name_1' => sub {
- # filter code
- return length($_);
- }, ...
- },
- db => $db_obj,
- lookup_regex => $lookup->regex,
- lookup => $lookup_obj,
- );
+Magic keys are:
-Parametar C<filter> defines user supplied snippets of perl code which can
-be use with C<filter{...}> notation.
+=over 4
-Recommended parametar C<lookup_regex> is used to enable parsing of lookups
-in structures. If you pass this parametar, you must also pass C<lookup>
-which is C<WebPAC::Lookup> object.
+=item _
+
+Code of current database
+
+=item _mfn
+
+Current MFN
+
+=back
=cut
-sub new {
- my $class = shift;
- my $self = {@_};
- bless($self, $class);
+my $config;
+
+sub _set_config {
+ $config = shift;
+}
- my $r = $self->{'lookup_regex'} ? 1 : 0;
- my $l = $self->{'lookup'} ? 1 : 0;
+=head2 _get_ds
- my $log = $self->_get_logger();
+Return hash formatted as data structure
- # those two must be in pair
- if ( ($r & $l) != ($r || $l) ) {
- my $log = $self->_get_logger();
- $log->logdie("lookup_regex and lookup must be in pair");
- }
+ my $ds = _get_ds();
- $log->logdie("lookup must be WebPAC::Lookup object") if ($self->{'lookup'} && ! $self->{'lookup'}->isa('WebPAC::Lookup'));
+=cut
+
+my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators);
+my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
- $self ? return $self : return undef;
+sub _get_ds {
+ return $out;
}
+=head2 _clean_ds
-=head2 data_structure
+Clean data structure hash for next record
-Create in-memory data structure which represents normalized layout from
-C<conf/normalize/*.xml>.
+ _clean_ds();
+
+=cut
+
+sub _clean_ds {
+ my $a = {@_};
+ ($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators) = ();
+ ($marc_record_offset, $marc_fetch_offset) = (0,0);
+ $marc_encoding = $a->{marc_encoding};
+}
-This structures are used to produce output.
+=head2 _set_lookup
- my $ds = $webpac->data_structure($rec);
+Set current lookup hash
+
+ _set_lookup( $lookup );
=cut
-sub data_structure {
- my $self = shift;
+my $lookup;
- my $log = $self->_get_logger();
+sub _set_lookup {
+ $lookup = shift;
+}
- my $rec = shift;
- $log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
+=head2 _get_lookup
- $log->debug("data_structure rec = ", sub { Dumper($rec) });
+Get current lookup hash
- $log->logdie("need unique ID (mfn) in field 000 of record ", sub { Dumper($rec) } ) unless (defined($rec->{'000'}));
+ my $lookup = _get_lookup();
- my $mfn = $rec->{'000'}->[0] || $log->logdie("field 000 isn't array!");
+=cut
- my $cache_file;
+sub _get_lookup {
+ return $lookup;
+}
- if ($self->{'db'}) {
- my $ds = $self->{'db'}->load_ds( $mfn );
- $log->debug("load_ds( rec = ", sub { Dumper($rec) }, ") = ", sub { Dumper($ds) });
- return $ds if ($ds);
- $log->debug("cache miss, creating");
- }
+=head2 _get_marc_fields
- undef $self->{'currnet_filename'};
- undef $self->{'headline'};
+Get all fields defined by calls to C<marc>
- my @sorted_tags;
- if ($self->{tags_by_order}) {
- @sorted_tags = @{$self->{tags_by_order}};
- } else {
- @sorted_tags = sort { $self->_sort_by_order } keys %{$self->{'import_xml'}->{'indexer'}};
- $self->{tags_by_order} = \@sorted_tags;
- }
+ $marc->add_fields( WebPAC::Normalize:_get_marc_fields() );
- my $ds;
+We are using I<magic> which detect repeatable fields only from
+sequence of field/subfield data generated by normalization.
- $log->debug("tags: ",sub { join(", ",@sorted_tags) });
+Repeatable field is created when there is second occurence of same subfield or
+if any of indicators are different.
- foreach my $field (@sorted_tags) {
+This is sane for most cases. Something like:
- my $row;
+ 900a-1 900b-1 900c-1
+ 900a-2 900b-2
+ 900a-3
-#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}});
+will be created from any combination of:
- foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) {
- my $format;
+ 900a-1 900a-2 900a-3 900b-1 900b-2 900c-1
- $log->logdie("expected tag HASH and got $tag") unless (ref($tag) eq 'HASH');
- $format = $tag->{'value'} || $tag->{'content'};
+and following rules:
- $log->debug("format: $format");
+ marc('900','a', rec('200','a') );
+ marc('900','b', rec('200','b') );
+ marc('900','c', rec('200','c') );
- my @v;
- if ($self->{'lookup_regex'} && $format =~ $self->{'lookup_regex'}) {
- @v = $self->fill_in_to_arr($rec,$format);
- } else {
- @v = $self->parse_to_arr($rec,$format);
- }
- next if (! @v);
+which might not be what you have in mind. If you need repeatable subfield,
+define it using C<marc_repeatable_subfield> like this:
- if ($tag->{'sort'}) {
- @v = $self->sort_arr(@v);
- }
+ marc_repeatable_subfield('900','a');
+ marc('900','a', rec('200','a') );
+ marc('900','b', rec('200','b') );
+ marc('900','c', rec('200','c') );
- # use format?
- if ($tag->{'format_name'}) {
- @v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v;
- }
+will create:
- # delimiter will join repeatable fields
- if ($tag->{'delimiter'}) {
- @v = ( join($tag->{'delimiter'}, @v) );
- }
+ 900a-1 900a-2 900a-3 900b-1 900c-1
+ 900b-2
- # default types
- my @types = qw(display search);
- # override by type attribute
- @types = ( $tag->{'type'} ) if ($tag->{'type'});
+There is also support for returning next or specific using:
- foreach my $type (@types) {
- # append to previous line?
- $log->debug("type: $type ",sub { join(" ",@v) }, $row->{'append'} || 'no append');
- if ($tag->{'append'}) {
+ while (my $mf = WebPAC::Normalize:_get_marc_fields( fetch_next => 1 ) ) {
+ # do something with $mf
+ }
- # I will delimit appended part with
- # delimiter (or ,)
- my $d = $tag->{'delimiter'};
- # default delimiter
- $d ||= " ";
+will always return fields from next MARC record or
- my $last = pop @{$row->{$type}};
- $d = "" if (! $last);
- $last .= $d . join($d, @v);
- push @{$row->{$type}}, $last;
+ my $mf = WebPAC::Normalize::_get_marc_fields( offset => 42 );
- } else {
- push @{$row->{$type}}, @v;
- }
- }
+will return 42th copy record (if it exists).
+=cut
+
+sub _get_marc_fields {
+
+ my $arg = {@_};
+ warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);
+ my $offset = $marc_fetch_offset;
+ if ($arg->{offset}) {
+ $offset = $arg->{offset};
+ } elsif($arg->{fetch_next}) {
+ $marc_fetch_offset++;
+ }
+
+ return if (! $marc_record || ref($marc_record) ne 'ARRAY');
+
+ warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);
+
+ my $marc_rec = $marc_record->[ $offset ];
+
+ warn "## _get_marc_fields (at offset: $offset) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);
+
+ return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);
+
+ # first, sort all existing fields
+ # XXX might not be needed, but modern perl might randomize elements in hash
+ my @sorted_marc_record = sort {
+ $a->[0] . ( $a->[3] || '' ) cmp $b->[0] . ( $b->[3] || '')
+ } @{ $marc_rec };
+
+ @sorted_marc_record = @{ $marc_rec }; ### FIXME disable sorting
+
+ # output marc fields
+ my @m;
+
+ # count unique field-subfields (used for offset when walking to next subfield)
+ my $u;
+ map { $u->{ $_->[0] . ( $_->[3] || '') }++ } @sorted_marc_record;
+
+ if ($debug) {
+ warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );
+ warn "## marc_record[$offset] = ", dump( $marc_rec ), $/;
+ warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;
+ warn "## subfield count = ", dump( $u ), $/;
+ }
+
+ my $len = $#sorted_marc_record;
+ my $visited;
+ my $i = 0;
+ my $field;
+ foreach ( 0 .. $len ) {
+
+ # find next element which isn't visited
+ while ($visited->{$i}) {
+ $i = ($i + 1) % ($len + 1);
}
- if ($row) {
- $row->{'tag'} = $field;
+ # mark it visited
+ $visited->{$i}++;
- # TODO: name_sigular, name_plural
- my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'};
- my $row_name = $name ? $self->_x($name) : $field;
+ my $row = dclone( $sorted_marc_record[$i] );
- # post-sort all values in field
- if ($self->{'import_xml'}->{'indexer'}->{$field}->{'sort'}) {
- $log->warn("sort at field tag not implemented");
- }
+ # field and subfield which is key for
+ # marc_repeatable_subfield and u
+ my $fsf = $row->[0] . ( $row->[3] || '' );
+
+ if ($debug > 1) {
+
+ print "### field so far [", $#$field, "] : ", dump( $field ), " ", $field ? 'T' : 'F', $/;
+ print "### this [$i]: ", dump( $row ),$/;
+ print "### sf: ", $row->[3], " vs ", $field->[3],
+ $marc_repeatable_subfield->{ $row->[0] . $row->[3] } ? ' (repeatable)' : '', $/,
+ if ($#$field >= 0);
- $ds->{$row_name} = $row;
+ }
+
+ # if field exists
+ if ( $#$field >= 0 ) {
+ if (
+ $row->[0] ne $field->[0] || # field
+ $row->[1] ne $field->[1] || # i1
+ $row->[2] ne $field->[2] # i2
+ ) {
+ push @m, $field;
+ warn "## saved/1 ", dump( $field ),$/ if ($debug);
+ $field = $row;
+
+ } elsif (
+ ( $row->[3] lt $field->[-2] ) # subfield which is not next (e.g. a after c)
+ ||
+ ( $row->[3] eq $field->[-2] && # same subfield, but not repeatable
+ ! $marc_repeatable_subfield->{ $fsf }
+ )
+ ) {
+ push @m, $field;
+ warn "## saved/2 ", dump( $field ),$/ if ($debug);
+ $field = $row;
+
+ } else {
+ # append new subfields to existing field
+ push @$field, ( $row->[3], $row->[4] );
+ }
+ } else {
+ # insert first field
+ $field = $row;
+ }
- $log->debug("row $field: ",sub { Dumper($row) });
+ if (! $marc_repeatable_subfield->{ $fsf }) {
+ # make step to next subfield
+ $i = ($i + $u->{ $fsf } ) % ($len + 1);
}
+ }
+ if ($#$field >= 0) {
+ push @m, $field;
+ warn "## saved/3 ", dump( $field ),$/ if ($debug);
}
- $self->{'db'}->save_ds(
- id => $mfn,
- ds => $ds,
- ) if ($self->{'db'});
+ return \@m;
+}
+
+=head2 _debug
- $log->debug("ds: ", sub { Dumper($ds) });
+Change level of debug warnings
- $log->logconfess("data structure returned is not array any more!") if wantarray;
+ _debug( 2 );
- return $ds;
+=cut
+sub _debug {
+ my $l = shift;
+ return $debug unless defined($l);
+ warn "debug level $l",$/ if ($l > 0);
+ $debug = $l;
}
-=head2 parse
+=head1 Functions to create C<data_structure>
+
+Those functions generally have to first in your normalization file.
-Perform smart parsing of string, skipping delimiters for fields which aren't
-defined. It can also eval code in format starting with C<eval{...}> and
-return output or nothing depending on eval code.
+=head2 tag
+
+Define new tag for I<search> and I<display>.
+
+ tag('Title', rec('200','a') );
- my $text = $webpac->parse($rec,'eval{"v901^a" eq "Deskriptor"}descriptor: v250^a', $i);
=cut
-sub parse {
- my $self = shift;
+sub tag {
+ my $name = shift or die "tag needs name as first argument";
+ my @o = grep { defined($_) && $_ ne '' } @_;
+ return unless (@o);
+ $out->{$name}->{tag} = $name;
+ $out->{$name}->{search} = \@o;
+ $out->{$name}->{display} = \@o;
+}
- my ($rec, $format_utf8, $i) = @_;
+=head2 display
- return if (! $format_utf8);
+Define tag just for I<display>
- my $log = $self->_get_logger();
+ @v = display('Title', rec('200','a') );
- $log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
+=cut
- $i = 0 if (! $i);
+sub display {
+ my $name = shift or die "display needs name as first argument";
+ my @o = grep { defined($_) && $_ ne '' } @_;
+ return unless (@o);
+ $out->{$name}->{tag} = $name;
+ $out->{$name}->{display} = \@o;
+}
- my $format = $self->_x($format_utf8) || $log->logconfess("can't convert '$format_utf8' from UTF-8 to ",$self->{'code_page'});
+=head2 search
- my @out;
+Prepare values just for I<search>
- $log->debug("format: $format");
+ @v = search('Title', rec('200','a') );
- my $eval_code;
- # remove eval{...} from beginning
- $eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s);
+=cut
- my $filter_name;
- # remove filter{...} from beginning
- $filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s);
+sub search {
+ my $name = shift or die "search needs name as first argument";
+ my @o = grep { defined($_) && $_ ne '' } @_;
+ return unless (@o);
+ $out->{$name}->{tag} = $name;
+ $out->{$name}->{search} = \@o;
+}
- my $prefix;
- my $all_found=0;
+=head2 marc_leader
- while ($format =~ s/^(.*?)(v|s)(\d+)(?:\^(\w))?//s) {
+Setup fields within MARC leader or get leader
- my $del = $1 || '';
- $prefix ||= $del if ($all_found == 0);
+ marc_leader('05','c');
+ my $leader = marc_leader();
- # repeatable index
- my $r = $i;
- $r = 0 if (lc("$2") eq 's');
+=cut
- my $found = 0;
- my $tmp = $self->get_data(\$rec,$3,$4,$r,\$found);
+sub marc_leader {
+ my ($offset,$value) = @_;
- if ($found) {
- push @out, $del;
- push @out, $tmp;
- $all_found += $found;
- }
+ if ($offset) {
+ $out->{' leader'}->{ $offset } = $value;
+ } else {
+ return $out->{' leader'};
}
+}
- return if (! $all_found);
+=head2 marc
- my $out = join('',@out);
+Save value for MARC field
- if ($out) {
- # add rest of format (suffix)
- $out .= $format;
+ marc('900','a', rec('200','a') );
+ marc('001', rec('000') );
- # add prefix if not there
- $out = $prefix . $out if ($out !~ m/^\Q$prefix\E/);
+=cut
- $log->debug("result: $out");
- }
+sub marc {
+ my $f = shift or die "marc needs field";
+ die "marc field must be numer" unless ($f =~ /^\d+$/);
- if ($eval_code) {
- my $eval = $self->fill_in($rec,$eval_code,$i) || return;
- $log->debug("about to eval{$eval} format: $out");
- return if (! $self->_eval($eval));
- }
-
- if ($filter_name && $self->{'filter'}->{$filter_name}) {
- $log->debug("about to filter{$filter_name} format: $out");
- $out = $self->{'filter'}->{$filter_name}->($out);
- return unless(defined($out));
- $log->debug("filter result: $out");
+ my $sf;
+ if ($f >= 10) {
+ $sf = shift or die "marc needs subfield";
}
- return $out;
+ foreach (@_) {
+ my $v = $_; # make var read-write for Encode
+ next unless (defined($v) && $v !~ /^\s*$/);
+ my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
+ if (defined $sf) {
+ push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $i1, $i2, $sf => $v ];
+ } else {
+ push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
+ }
+ }
}
-=head2 parse_to_arr
+=head2 marc_repeatable_subfield
-Similar to C<parse>, but returns array of all repeatable fields
+Save values for MARC repetable subfield
- my @arr = $webpac->parse_to_arr($rec,'v250^a');
+ marc_repeatable_subfield('910', 'z', rec('909') );
=cut
-sub parse_to_arr {
- my $self = shift;
+sub marc_repeatable_subfield {
+ my ($f,$sf) = @_;
+ die "marc_repeatable_subfield need field and subfield!\n" unless ($f && $sf);
+ $marc_repeatable_subfield->{ $f . $sf }++;
+ marc(@_);
+}
- my ($rec, $format_utf8) = @_;
+=head2 marc_indicators
- my $log = $self->_get_logger();
+Set both indicators for MARC field
- $log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
- return if (! $format_utf8);
+ marc_indicators('900', ' ', 1);
- my $i = 0;
- my @arr;
+Any indicator value other than C<0-9> will be treated as undefined.
- while (my $v = $self->parse($rec,$format_utf8,$i++)) {
- push @arr, $v;
- }
+=cut
- $log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr);
+sub marc_indicators {
+ my $f = shift || die "marc_indicators need field!\n";
+ my ($i1,$i2) = @_;
+ die "marc_indicators($f, ...) need i1!\n" unless(defined($i1));
+ die "marc_indicators($f, $i1, ...) need i2!\n" unless(defined($i2));
- return @arr;
+ $i1 = ' ' if ($i1 !~ /^\d$/);
+ $i2 = ' ' if ($i2 !~ /^\d$/);
+ @{ $marc_indicators->{$f} } = ($i1,$i2);
}
+=head2 marc_compose
+
+Save values for each MARC subfield explicitly
+
+ marc_compose('900',
+ 'a', rec('200','a')
+ 'b', rec('201','a')
+ 'a', rec('200','b')
+ 'c', rec('200','c')
+ );
-=head2 fill_in
+If you specify C<+> for subfield, value will be appended
+to previous defined subfield.
-Workhourse of all: takes record from in-memory structure of database and
-strings with placeholders and returns string or array of with substituted
-values from record.
+=cut
- my $text = $webpac->fill_in($rec,'v250^a');
+sub marc_compose {
+ my $f = shift or die "marc_compose needs field";
+ die "marc_compose field must be numer" unless ($f =~ /^\d+$/);
-Optional argument is ordinal number for repeatable fields. By default,
-it's assume to be first repeatable field (fields are perl array, so first
-element is 0).
-Following example will read second value from repeatable field.
+ my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
+ my $m = [ $f, $i1, $i2 ];
- my $text = $webpac->fill_in($rec,'Title: v250^a',1);
+ warn "### marc_compose input subfields = ", dump(@_),$/ if ($debug > 2);
-This function B<does not> perform parsing of format to inteligenty skip
-delimiters before fields which aren't used.
+ if ($#_ % 2 != 1) {
+ die "ERROR: marc_compose",dump($f,@_)," not valid (must be even).\nDo you need to add first() or join() around some argument?\n";
+ }
-This method will automatically decode UTF-8 string to local code page
-if needed.
+ while (@_) {
+ my $sf = shift;
+ my $v = shift;
+
+ next unless (defined($v) && $v !~ /^\s*$/);
+ warn "## ++ marc_compose($f,$sf,$v) ", dump( $m ),$/ if ($debug > 1);
+ if ($sf ne '+') {
+ push @$m, ( $sf, $v );
+ } else {
+ $m->[ $#$m ] .= $v;
+ }
+ }
+
+ warn "## marc_compose current marc = ", dump( $m ),$/ if ($debug > 1);
+
+ push @{ $marc_record->[ $marc_record_offset ] }, $m if ($#{$m} > 2);
+}
+
+=head2 marc_duplicate
+
+Generate copy of current MARC record and continue working on copy
+
+ marc_duplicate();
+
+Copies can be accessed using C<< _get_marc_fields( fetch_next => 1 ) >> or
+C<< _get_marc_fields( offset => 42 ) >>.
=cut
-sub fill_in {
- my $self = shift;
+sub marc_duplicate {
+ my $m = $marc_record->[ -1 ];
+ die "can't duplicate record which isn't defined" unless ($m);
+ push @{ $marc_record }, dclone( $m );
+ warn "## marc_duplicate = ", dump(@$marc_record), $/ if ($debug > 1);
+ $marc_record_offset = $#{ $marc_record };
+ warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);
+}
- my $log = $self->_get_logger();
+=head2 marc_remove
- my $rec = shift || $log->logconfess("need data record");
- my $format = shift || $log->logconfess("need format to parse");
- # iteration (for repeatable fields)
- my $i = shift || 0;
+Remove some field or subfield from MARC record.
- $log->logdie("infitite loop in format $format") if ($i > ($self->{'max_mfn'} || 9999));
+ marc_remove('200');
+ marc_remove('200','a');
- # FIXME remove for speedup?
- $log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
+This will erase field C<200> or C<200^a> from current MARC record.
- if (utf8::is_utf8($format)) {
- $format = $self->_x($format);
- }
+This is useful after calling C<marc_duplicate> or on it's own (but, you
+should probably just remove that subfield definition if you are not
+using C<marc_duplicate>).
- my $found = 0;
+FIXME: support fields < 10.
- my $eval_code;
- # remove eval{...} from beginning
- $eval_code = $1 if ($format =~ s/^eval{([^}]+)}//s);
+=cut
- my $filter_name;
- # remove filter{...} from beginning
- $filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s);
+sub marc_remove {
+ my ($f, $sf) = @_;
- # do actual replacement of placeholders
- # repeatable fields
- $format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges;
- # non-repeatable fields
- $format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found)/ges;
+ die "marc_remove needs record number" unless defined($f);
- if ($found) {
- $log->debug("format: $format");
- if ($eval_code) {
- my $eval = $self->fill_in($rec,$eval_code,$i);
- return if (! $self->_eval($eval));
- }
- if ($filter_name && $self->{'filter'}->{$filter_name}) {
- $log->debug("filter '$filter_name' for $format");
- $format = $self->{'filter'}->{$filter_name}->($format);
- return unless(defined($format));
- $log->debug("filter result: $format");
- }
- # do we have lookups?
- if ($self->{'lookup'}) {
- if ($self->{'lookup'}->can('lookup')) {
- return $self->{'lookup'}->lookup($format);
+ my $marc = $marc_record->[ $marc_record_offset ];
+
+ warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2);
+
+ my $i = 0;
+ foreach ( 0 .. $#{ $marc } ) {
+ last unless (defined $marc->[$i]);
+ warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3);
+ if ($marc->[$i]->[0] eq $f) {
+ if (! defined $sf) {
+ # remove whole field
+ splice @$marc, $i, 1;
+ warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3);
+ $i--;
} else {
- $log->warn("Have lookup object but can't invoke lookup method");
+ foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) {
+ my $o = ($j * 2) + 3;
+ if ($marc->[$i]->[$o] eq $sf) {
+ # remove subfield
+ splice @{$marc->[$i]}, $o, 2;
+ warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3);
+ # is record now empty?
+ if ($#{ $marc->[$i] } == 2) {
+ splice @$marc, $i, 1;
+ warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3);
+ $i--;
+ };
+ }
+ }
}
- } else {
- return $format;
}
- } else {
- return;
+ $i++;
}
+
+ warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2);
+
+ $marc_record->[ $marc_record_offset ] = $marc;
+
+ warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
}
+=head2 marc_original_order
-=head2 fill_in_to_arr
+Copy all subfields preserving original order to marc field.
-Similar to C<fill_in>, but returns array of all repeatable fields. Usable
-for fields which have lookups, so they shouldn't be parsed but rather
-C<fill_id>ed.
+ marc_original_order( marc_field_number, original_input_field_number );
+
+Please note that field numbers are consistent with other commands (marc
+field number first), but somewhat counter-intuitive (destination and then
+source).
+
+You might want to use this command if you are just renaming subfields or
+using pre-processing modify_record in C<config.yml> and don't need any
+post-processing or want to preserve order of original subfields.
- my @arr = $webpac->fill_in_to_arr($rec,'[v900];;[v250^a]');
=cut
-sub fill_in_to_arr {
- my $self = shift;
+sub marc_original_order {
- my ($rec, $format_utf8) = @_;
+ my ($to, $from) = @_;
+ die "marc_original_order needs from and to fields\n" unless ($from && $to);
- my $log = $self->_get_logger();
+ return unless defined($rec->{$from});
- $log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
- return if (! $format_utf8);
+ my $r = $rec->{$from};
+ die "record field $from isn't array\n" unless (ref($r) eq 'ARRAY');
- my $i = 0;
- my @arr;
+ my ($i1,$i2) = defined($marc_indicators->{$to}) ? @{ $marc_indicators->{$to} } : (' ',' ');
+ warn "## marc_original_order($to,$from) source = ", dump( $r ),$/ if ($debug > 1);
- while (my @v = $self->fill_in($rec,$format_utf8,$i++)) {
- push @arr, @v;
- }
+ foreach my $d (@$r) {
- $log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr);
+ if (! defined($d->{subfields}) && ref($d->{subfields}) ne 'ARRAY') {
+ warn "# marc_original_order($to,$from): field $from doesn't have subfields specification\n";
+ next;
+ }
+
+ my @sfs = @{ $d->{subfields} };
- return @arr;
-}
+ die "field $from doesn't have even number of subfields specifications\n" unless($#sfs % 2 == 1);
+
+ warn "#--> d: ",dump($d), "\n#--> sfs: ",dump(@sfs),$/ if ($debug > 2);
+ my $m = [ $to, $i1, $i2 ];
-=head2 get_data
+ while (my $sf = shift @sfs) {
-Returns value from record.
+ warn "#--> sf: ",dump($sf), $/ if ($debug > 2);
+ my $offset = shift @sfs;
+ die "corrupted sufields specification for field $from\n" unless defined($offset);
- my $text = $self->get_data(\$rec,$f,$sf,$i,\$found);
+ my $v;
+ if (ref($d->{$sf}) eq 'ARRAY') {
+ $v = $d->{$sf}->[$offset] if (defined($d->{$sf}->[$offset]));
+ } elsif ($offset == 0) {
+ $v = $d->{$sf};
+ } else {
+ die "field $from subfield '$sf' need occurence $offset which doesn't exist", dump($d->{$sf});
+ }
+ push @$m, ( $sf, $v ) if (defined($v));
+ }
+
+ if ($#{$m} > 2) {
+ push @{ $marc_record->[ $marc_record_offset ] }, $m;
+ }
+ }
+
+ warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
+}
-Arguments are:
-record reference C<$rec>,
-field C<$f>,
-optional subfiled C<$sf>,
-index for repeatable values C<$i>.
-Optinal variable C<$found> will be incremeted if there
-is field.
+=head1 Functions to extract data from input
-Returns value or empty string.
+This function should be used inside functions to create C<data_structure> described
+above.
+
+=head2 _pack_subfields_hash
+
+ @subfields = _pack_subfields_hash( $h );
+ $subfields = _pack_subfields_hash( $h, 1 );
+
+Return each subfield value in array or pack them all together and return scalar
+with subfields (denoted by C<^>) and values.
=cut
-sub get_data {
- my $self = shift;
+sub _pack_subfields_hash {
- my ($rec,$f,$sf,$i,$found) = @_;
+ warn "## _pack_subfields_hash( ",dump(@_), " )\n" if ($debug > 1);
- if ($$rec->{$f}) {
- return '' if (! $$rec->{$f}->[$i]);
- no strict 'refs';
- if ($sf && $$rec->{$f}->[$i]->{$sf}) {
- $$found++ if (defined($$found));
- return $$rec->{$f}->[$i]->{$sf};
- } elsif (! $sf && $$rec->{$f}->[$i]) {
- $$found++ if (defined($$found));
- # it still might have subfield, just
- # not specified, so we'll dump all
- if ($$rec->{$f}->[$i] =~ /HASH/o) {
- my $out;
- foreach my $k (keys %{$$rec->{$f}->[$i]}) {
- $out .= $$rec->{$f}->[$i]->{$k}." ";
- }
- return $out;
+ my ($h,$include_subfields) = @_;
+
+ if ( defined($h->{subfields}) ) {
+ my $sfs = delete $h->{subfields} || die "no subfields?";
+ my @out;
+ while (@$sfs) {
+ my $sf = shift @$sfs;
+ push @out, '^' . $sf if ($include_subfields);
+ my $o = shift @$sfs;
+ if ($o == 0 && ref( $h->{$sf} ) ne 'ARRAY' ) {
+ # single element subfields are not arrays
+#warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
+
+ push @out, $h->{$sf};
} else {
- return $$rec->{$f}->[$i];
+#warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
+ push @out, $h->{$sf}->[$o];
}
+ }
+ if ($include_subfields) {
+ return join('', @out);
} else {
- return '';
+ return @out;
}
} else {
- return '';
+ if ($include_subfields) {
+ my $out = '';
+ foreach my $sf (sort keys %$h) {
+ if (ref($h->{$sf}) eq 'ARRAY') {
+ $out .= '^' . $sf . join('^' . $sf, @{ $h->{$sf} });
+ } else {
+ $out .= '^' . $sf . $h->{$sf};
+ }
+ }
+ return $out;
+ } else {
+ # FIXME this should probably be in alphabetical order instead of hash order
+ values %{$h};
+ }
}
}
+=head2 rec1
+
+Return all values in some field
+
+ @v = rec1('200')
+
+TODO: order of values is probably same as in source data, need to investigate that
-=head2 apply_format
+=cut
+
+sub rec1 {
+ my $f = shift;
+ warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
+ return unless (defined($rec) && defined($rec->{$f}));
+ warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
+ if (ref($rec->{$f}) eq 'ARRAY') {
+ my @out;
+ foreach my $h ( @{ $rec->{$f} } ) {
+ if (ref($h) eq 'HASH') {
+ push @out, ( _pack_subfields_hash( $h ) );
+ } else {
+ push @out, $h;
+ }
+ }
+ return @out;
+ } elsif( defined($rec->{$f}) ) {
+ return $rec->{$f};
+ }
+}
-Apply format specified in tag with C<format_name="name"> and
-C<format_delimiter=";;">.
+=head2 rec2
- my $text = $webpac->apply_format($format_name,$format_delimiter,$data);
+Return all values in specific field and subfield
-Formats can contain C<lookup{...}> if you need them.
+ @v = rec2('200','a')
=cut
-sub apply_format {
- my $self = shift;
+sub rec2 {
+ my $f = shift;
+ return unless (defined($rec && $rec->{$f}));
+ my $sf = shift;
+ warn "rec2($f,$sf) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
+ return map {
+ if (ref($_->{$sf}) eq 'ARRAY') {
+ @{ $_->{$sf} };
+ } else {
+ $_->{$sf};
+ }
+ } grep { ref($_) eq 'HASH' && $_->{$sf} } @{ $rec->{$f} };
+}
+
+=head2 rec
+
+syntaxtic sugar for
- my ($name,$delimiter,$data) = @_;
+ @v = rec('200')
+ @v = rec('200','a')
- my $log = $self->_get_logger();
+=cut
- if (! $self->{'import_xml'}->{'format'}->{$name}) {
- $log->warn("<format name=\"$name\"> is not defined in ",$self->{'import_xml_file'});
- return $data;
+sub rec {
+ my @out;
+ if ($#_ == 0) {
+ @out = rec1(@_);
+ } elsif ($#_ == 1) {
+ @out = rec2(@_);
}
+ if (@out) {
+ return @out;
+ } else {
+ return '';
+ }
+}
- $log->warn("no delimiter for format $name") if (! $delimiter);
+=head2 regex
- my $format = $self->_x($self->{'import_xml'}->{'format'}->{$name}->{'content'}) || $log->logdie("can't find format '$name'");
+Apply regex to some or all values
- my @data = split(/\Q$delimiter\E/, $data);
+ @v = regex( 's/foo/bar/g', @v );
- my $out = sprintf($format, @data);
- $log->debug("using format $name [$format] on $data to produce: $out");
+=cut
- if ($self->{'lookup_regex'} && $out =~ $self->{'lookup_regex'}) {
- return $self->{'lookup'}->lookup($out);
- } else {
- return $out;
+sub regex {
+ my $r = shift;
+ my @out;
+ #warn "r: $r\n", dump(\@_);
+ foreach my $t (@_) {
+ next unless ($t);
+ eval "\$t =~ $r";
+ push @out, $t if ($t && $t ne '');
}
+ return @out;
+}
+
+=head2 prefix
+Prefix all values with a string
+
+ @v = prefix( 'my_', @v );
+
+=cut
+
+sub prefix {
+ my $p = shift or return;
+ return map { $p . $_ } grep { defined($_) } @_;
+}
+
+=head2 suffix
+
+suffix all values with a string
+
+ @v = suffix( '_my', @v );
+
+=cut
+
+sub suffix {
+ my $s = shift or die "suffix needs string as first argument";
+ return map { $_ . $s } grep { defined($_) } @_;
}
-=head2 sort_arr
+=head2 surround
-Sort array ignoring case and html in data
+surround all values with a two strings
- my @sorted = $webpac->sort_arr(@unsorted);
+ @v = surround( 'prefix_', '_suffix', @v );
=cut
-sub sort_arr {
- my $self = shift;
+sub surround {
+ my $p = shift or die "surround need prefix as first argument";
+ my $s = shift or die "surround needs suffix as second argument";
+ return map { $p . $_ . $s } grep { defined($_) } @_;
+}
+
+=head2 first
+
+Return first element
+
+ $v = first( @v );
+
+=cut
+
+sub first {
+ my $r = shift;
+ return $r;
+}
+
+=head2 lookup
+
+Consult lookup hashes for some value
+
+ @v = lookup( $v );
+ @v = lookup( @v );
+
+FIXME B<currently this one is broken!>
+
+=cut
+
+sub lookup {
+ my $k = shift or return;
+ return unless (defined($lookup->{$k}));
+ if (ref($lookup->{$k}) eq 'ARRAY') {
+ return @{ $lookup->{$k} };
+ } else {
+ return $lookup->{$k};
+ }
+}
+
+=head2 save_into_lookup
+
+Save value into lookup.
- my $log = $self->_get_logger();
+ save_into_lookup($database,$input,$key,sub {
+ # code which produce one or more values
+ });
- # FIXME add Schwartzian Transformation?
+This function shouldn't be called directly, it's called from code created by L<WebPAC::Parser>.
- my @sorted = sort {
- $a =~ s#<[^>]+/*>##;
- $b =~ s#<[^>]+/*>##;
- lc($b) cmp lc($a)
- } @_;
- $log->debug("sorted values: ",sub { join(", ",@sorted) });
+=cut
- return @sorted;
+sub save_into_lookup {
+ my ($database,$input,$key,$coderef) = @_;
+ die "save_into_lookup needs database" unless defined($database);
+ die "save_into_lookup needs input" unless defined($input);
+ die "save_into_lookup needs key" unless defined($key);
+ die "save_into_lookup needs CODE" unless ( defined($coderef) && ref($coderef) eq 'CODE' );
+ my $mfn = $rec->{'000'}->[0] || die "mfn not defined or zero";
+ foreach my $v ( $coderef->() ) {
+ $lookup->{$database}->{$input}->{$key}->{$v}->{$mfn}++;
+ warn "# saved lookup $database/$input/$key [$v] $mfn\n"; #if ($debug > 1);
+ }
}
+=head2 config
-=head1 INTERNAL METHODS
+Consult config values stored in C<config.yml>
-=head2 _sort_by_order
+ # return database code (key under databases in yaml)
+ $database_code = config(); # use _ from hash
+ $database_name = config('name');
+ $database_input_name = config('input name');
+ $tag = config('input normalize tag');
-Sort xml tags data structure accoding to C<order=""> attribute.
+Up to three levels are supported.
=cut
-sub _sort_by_order {
- my $self = shift;
+sub config {
+ return unless ($config);
+
+ my $p = shift;
+
+ $p ||= '';
+
+ my $v;
+
+ warn "### getting config($p)\n" if ($debug > 1);
+
+ my @p = split(/\s+/,$p);
+ if ($#p < 0) {
+ $v = $config->{ '_' }; # special, database code
+ } else {
+
+ my $c = dclone( $config );
+
+ foreach my $k (@p) {
+ warn "### k: $k c = ",dump($c),$/ if ($debug > 1);
+ if (ref($c) eq 'ARRAY') {
+ $c = shift @$c;
+ warn "config($p) taking first occurence of '$k', probably not what you wanted!\n";
+ last;
+ }
+
+ if (! defined($c->{$k}) ) {
+ $c = undef;
+ last;
+ } else {
+ $c = $c->{$k};
+ }
+ }
+ $v = $c if ($c);
- my $va = $self->{'import_xml'}->{'indexer'}->{$a}->{'order'} ||
- $self->{'import_xml'}->{'indexer'}->{$a};
- my $vb = $self->{'import_xml'}->{'indexer'}->{$b}->{'order'} ||
- $self->{'import_xml'}->{'indexer'}->{$b};
+ }
- return $va <=> $vb;
+ warn "## config( '$p' ) = ",dump( $v ),$/ if ($v && $debug);
+ warn "config( '$p' ) is empty\n" if (! $v);
+
+ return $v;
}
-=head2 _x
+=head2 id
-Convert strings from C<conf/normalize/*.xml> encoding into application
-specific encoding (optinally specified using C<code_page> to C<new>
-constructor).
+Returns unique id of this record
- my $text = $n->_x('normalize text string');
+ $id = id();
-This is a stub so that other modules doesn't have to implement it.
+Returns C<42/2> for 2nd occurence of MFN 42.
=cut
-sub _x {
- my $self = shift;
- return shift;
+sub id {
+ my $mfn = $config->{_mfn} || die "no _mfn in config data";
+ return $mfn . $#{$marc_record} ? $#{$marc_record} + 1 : '';
}
+=head2 join_with
+
+Joins walues with some delimiter
+
+ $v = join_with(", ", @v);
+
+=cut
+
+sub join_with {
+ my $d = shift;
+ warn "### join_with('$d',",dump(@_),")\n" if ($debug > 2);
+ my $v = join($d, grep { defined($_) && $_ ne '' } @_);
+ return '' unless defined($v);
+ return $v;
+}
-=head1 AUTHOR
+=head2 split_rec_on
-Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
+Split record subfield on some regex and take one of parts out
-=head1 COPYRIGHT & LICENSE
+ $a_before_semi_column =
+ split_rec_on('200','a', /\s*;\s*/, $part);
-Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
+C<$part> is optional number of element. First element is
+B<1>, not 0!
-This program is free software; you can redistribute it and/or modify it
-under the same terms as Perl itself.
+If there is no C<$part> parameter or C<$part> is 0, this function will
+return all values produced by splitting.
=cut
-1; # End of WebPAC::DB
+sub split_rec_on {
+ die "split_rec_on need (fld,sf,regex[,part]" if ($#_ < 2);
+
+ my ($fld, $sf, $regex, $part) = @_;
+ warn "### regex ", ref($regex), $regex, $/ if ($debug > 2);
+
+ my @r = rec( $fld, $sf );
+ my $v = shift @r;
+ warn "### first rec($fld,$sf) = ",dump($v),$/ if ($debug > 2);
+
+ return '' if ( ! defined($v) || $v =~ /^\s*$/);
+
+ my @s = split( $regex, $v );
+ warn "## split_rec_on($fld,$sf,$regex,$part) = ",dump(@s),$/ if ($debug > 1);
+ if ($part && $part > 0) {
+ return $s[ $part - 1 ];
+ } else {
+ return @s;
+ }
+}
+
+# END
+1;
projects / webpac2 / blobdiff