X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=Equinox-Migration%2Flib%2FEquinox%2FMigration%2FMapDrivenMARCXMLProc.pm;h=9844daf68f02ef213b23c7d195801e177ea3cb4f;hp=ab053bba0fb782367d4ada7750f612ac5925e935;hb=ca9dd06ee372e59b4fe0ecdcbe24cf7468d8577f;hpb=a85d1ead29a140aefd716b303f0f0e67a96cee75

diff --git a/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm b/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm
index ab053bb..9844daf 100644
--- a/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm
+++ b/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm
@@ -6,6 +6,16 @@ use strict;
 use XML::Twig;
 use Equinox::Migration::SubfieldMapper 1.002;
 
+# 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
 
 Equinox::Migration::MapDrivenMARCXMLProc
@@ -84,6 +94,13 @@ sub new {
         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;
 }
 
@@ -166,7 +183,8 @@ sub process_subs {
         # 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}{rcnt}++ unless ( defined $u->{$tag}{$code}{last} and
+                                           $u->{$tag}{$code}{last} == $self->{data}{rptr} );
         $u->{$tag}{$code}{last} = $self->{data}{rptr};
         return;
     }
@@ -182,10 +200,34 @@ sub process_subs {
             push @{$dataf->{multi}{$name}}, $sub->text;
         }
     } else {
+        die "Multiple occurances of a non-multi field: \n"
+          if (defined $dataf->{uni}{$code});
         $dataf->{uni}{$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
+
+=head2 required
+
 =head1 PARSED RECORDS
 
 Given:
@@ -255,14 +297,30 @@ 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 },
-                  sub_code2 => { value => VALUE, count => COUNT },
+                  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
 
 Shawn Boyette, C<< <sboyette at esilibrary.com> >>