use strict;
use XML::Twig;
-use Equinox::Migration::SubfieldMapper 1.002;
+use Equinox::Migration::SubfieldMapper 1.003;
+
+# FIXME
+#
+# sample functionality should be extracted into a new module which
+# uses E::M::SM to drive sampling of individual datafields, and
+# reports ALL datafields which occur
+#
+# --sample should give the list of all datafields
+# --samplefile should take a SM map as teh argument and introspect the mapped datafields
+
=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
},
}, $class;
die "Can't open marc file: $!\n";
}
- # if we have a sample arg, set up the sample set and umap hash
- if (defined $args{sample}) {
- for my $s ( @{$args{sample}})
- { $self->{data}{stag}{$s} = 1 }
- $self->{data}{umap} = {};
- }
-
return $self;
}
$record->purge;
$self->{data}{rptr}++;
+ # check for required fields
+ $self->check_required;
+
return $self->{data}{crec};
}
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 ( defined $u->{$tag}{$code}{last} and
- $u->{$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
+ # handle modifiers
if (my $mods = $map->mods($field)) {
if ($mods->{multi}) {
my $name = $tag . $code;
push @{$dataf->{multi}{$name}}, $sub->text;
+ return;
+ }
+ }
+
+ die "Multiple occurances of a non-multi field: $tag$code at rec ",
+ ($self->{data}{rptr} + 1),"\n" if (defined $dataf->{uni}{$code});
+ $dataf->{uni}{$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;
+
+ $found = 1 if ($crec->{bib}{($tag_id . $code)});
+ 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 bib
+
+The C<bib> modifier declares that a mapping is "bib-level", and should
+be encountered once per B<record> instead of once per B<datafield> --
+which is another way of saying that it occurs in a non-repeating
+datafield or in a controlfield.
+
+=head2 required
+
+By default, if a mapping does not occur in a datafield (or record, in
+the case of C<bib> mappings), 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:
record struct. Oppose subfields which are not mapped, which will be
C<undef>.
-=head1 UNMAPPED TAGS
-
-If the C<sample> argument is passed to L</new>, there will also be a
-structure which holds data about unmapped subfields encountered in
-mapped tags which are also in the declared sample set. This
-information is collected over the life of the object and is not reset
-for every record processed (as the current record data neccessarily
-is).
-
- { tag_id => {
- sub_code => { value => VALUE,
- count => COUNT,
- rcnt => RCOUNT
- },
- ...
- },
- ...
- }
-
-For each mapped tag, for each unmapped subfield, there is a hash of
-data about that subfield containing
-
- * value - A sample of the subfield text
- * count - Total number of times the subfield was seen
- * rcnt - The number of records the subfield was seen in
=head1 AUTHOR