use strict;
use XML::Twig;
-use Equinox::Migration::SubfieldMapper 1.002;
+use Equinox::Migration::SubfieldMapper 1.003;
+
=head1 NAME
my $m = Equinox::Migration::MapDrivenMARCXMLProc->new( mapfile => FILE,
marcfile => FILE );
-There is an optional third, argument, C<sample>, which specifies a
-arrayref of datafields to "sample" by reporting on subfields which are
-found in the data but not in the map.
-
- my $m = Equinox::Migration::MapDrivenMARCXMLProc->new( mapfile => FILE,
- marcfile => FILE,
- sample => \@TAGS
- );
-
-See L</UNMAPPED TAGS> for more info.
-
=cut
sub new {
my ($class, %args) = @_;
my $self = bless { mods => { multi => {},
- once => {},
+ bib => {},
required => {},
},
data => { recs => undef, # X::T record objects
rptr => 0, # next record pointer
crec => undef, # parsed record storage
- stag => undef, # list of tags to sample
- umap => undef, # unmapped data samples
+ tmap => undef, # tag_id-to-tag array map
},
}, $class;
# get the next record and wipe current parsed record
return 0 unless defined $self->{data}{recs}[ $self->{data}{rptr} ];
my $record = $self->{data}{recs}[ $self->{data}{rptr} ];
- $self->{data}{crec} = { egid => undef, bib => undef, tags => undef };
+ $self->{data}{crec} = { egid => undef, tags => undef };
+ $self->{data}{tmap} = {};
my @fields = $record->children;
for my $f (@fields)
$record->purge;
$self->{data}{rptr}++;
+ # check for required fields
+ $self->check_required;
+
return $self->{data}{crec};
}
my $map = $self->{map};
my $tag = $field->{'att'}->{'tag'};
my $crec = $self->{data}{crec};
+ my $tmap = $self->{data}{tmap};
# leader
unless (defined $tag) {
}
if ($map->has($tag)) {
push @{$crec->{tags}}, { tag => $tag, uni => undef, multi => undef };
+ push @{$tmap->{$tag}}, (@{$crec->{tags}} - 1);
my @subs = $field->children('subfield');
for my $sub (@subs)
{ $self->process_subs($tag, $sub) }
my $code = $sub->{'att'}->{'code'};
# handle unmapped tag/subs
- unless ($map->has($tag, $code)) {
- my $u = $self->{data}{umap};
- my $s = $self->{data}{stag};
- return unless (defined $s->{$tag});
-
- # set a value, total-seen count and records-seen-in count
- $u->{$tag}{$code}{value} = $sub->text unless defined $u->{$tag}{$code};
- $u->{$tag}{$code}{count}++;
- $u->{$tag}{$code}{rcnt}++ unless ($u->{$tag}{$code}{last} == $self->{data}{rptr});
- $u->{$tag}{$code}{last} = $self->{data}{rptr};
- return;
- }
+ return unless ($map->has($tag, $code));
# fetch our datafield struct and fieldname
my $dataf = $self->{data}{crec}{tags}[-1];
my $field = $map->field($tag, $code);
- # handle modifiers, or slug data in normally
+ # test filters
+ for my $filter ( @{$map->filters($field)} ) {
+ return if ($sub->text =~ /$filter/i);
+ }
+ # handle multi modifier
if (my $mods = $map->mods($field)) {
if ($mods->{multi}) {
my $name = $tag . $code;
push @{$dataf->{multi}{$name}}, $sub->text;
+ return;
+ }
+ }
+
+ # if this were a multi field, it would be handled already. make sure its a singleton
+ die "Multiple occurances of a non-multi field: $tag$code at rec ",
+ ($self->{data}{rptr} + 1),"\n" if (defined $dataf->{uni}{$code});
+
+ # everything seems okay
+ $dataf->{uni}{$code} = $sub->text;
+}
+
+
+sub check_required {
+ my ($self) = @_;
+ my $mods = $self->{map}->mods;
+ my $crec = $self->{data}{crec};
+
+ for my $tag_id (keys %{$mods->{required}}) {
+ for my $code (@{$mods->{required}{$tag_id}}) {
+ my $found = 0;
+
+ for my $tag (@{$crec->{tags}}) {
+ $found = 1 if ($tag->{multi}{($tag_id . $code)});
+ $found = 1 if ($tag->{uni}{$code});
+ }
+
+ die "Required mapping $tag_id$code not found in rec ",$self->{data}{rptr},"\n"
+ unless ($found);
}
- } else {
- $dataf->{uni}{$code} = $sub->text;
}
+
}
+=head1 MODIFIERS
+
+MapDrivenMARCXMLProc implements the following modifiers, and passes
+them to L<Equinox::Migration::SubfieldMapper>, meaning that specifying
+any other modifiers in a MDMP map file will cause a fatal error when
+it is processed.
+
+=head2 multi
+
+If a mapping is declared to be C<multi>, then MDMP expects to see more
+than one instance of that subfield per datafield, and the data is
+handled accordingly (see L</PARSED RECORDS> below).
+
+Occurring zero or one time is legal for a C<multi> mapping.
+
+A mapping which is not flagged as C<multi>, but which occurs more than
+once per datafield will cause a fatal error.
+
+=head2 required
+
+By default, if a mapping does not occur in a datafield, processing
+continues normally. if a mapping has the C<required> modifier,
+however, it must appear, or a fatal error will occur.
+
=head1 PARSED RECORDS
Given:
Then C<$rec> will look like:
{
- egid => evergreen_record_id,
- bib => {
- (tag_id . sub_code)1 => value1,
- (tag_id . sub_code)2 => value2,
- ...
- },
+ egid => evergreen_record_id,
tags => [
{
tag => tag_id,
}
That is, there is an C<egid> key which points to the Evergreen ID of
-that record, a C<bib> key which points to a hashref, and a C<tags>
-key which points to an arrayref.
-
-=head3 C<bib>
-
-A reference to a hash which holds extracted data which occurs only
-once per record (and is therefore "bib-level"; the default assumption
-is that a tag/subfield pair can occur multiple times per record). The
-keys are composed of tag id and subfield code, catenated
-(e.g. 901c). The values are the contents of that subfield of that tag.
-
-If there are no tags defined as bib-level in the mapfile, C<bib> will
-be C<undef>.
+that record, and a C<tags> key which points to an arrayref.
=head3 C<tags>
record struct. Oppose subfields which are not mapped, which will be
C<undef>.
-=head1 UNMAPPED TAGS
-
- { tag_id => {
- sub_code => { value => VALUE, count => COUNT },
- sub_code2 => { value => VALUE, count => COUNT },
- ...
- },
- ...
- }
=head1 AUTHOR