works with new MDMP

[migration-tools.git] / extract_holdings

diff --git a/extract_holdings b/extract_holdings
index 913aba1..cf3e4fd 100755 (executable)
--- a/extract_holdings
+++ b/extract_holdings
@@ -8,6 +8,14 @@ use Equinox::Migration::MARCXMLSampler;
 
 my $VERSION = '1.001';
 
+=pod
+
+TODO
+
+  * Have detail mode report on number of subfields per datafield
+
+=cut
+
 my $c = initialize();
 $| = 1;
 
@@ -21,31 +29,37 @@ sub extract_holdings {
     my ($c) = @_;
     print "Parsing records for extraction... ";
     my $m = Equinox::Migration::MapDrivenMARCXMLProc->new( marcfile => $c->{marcfile},
-                                                           mapfile  => $c->{map} );
+                                                           mapfile  => $c->{map},
+                                                           verbose  => 1,
+                                                         );
+    print "Writing holdings to output file(s)...\n";
     # open main holdings file
     open HOLDINGS, '>', ($c->{prefix} . "-HOLDINGS.pg");
     # open the files for multi mappings
-    # FIXME DO THIS
+    # FIXME make this actually key off multi fields
     open X, '>', ($c->{prefix} . "-HOLDINGS-privnotes.pg");
     open Z, '>', ($c->{prefix} . "-HOLDINGS-pubnotes.pg");
     select HOLDINGS;
 
-    while (my $rec = $m->parse_record) {
+    my $i = 0;
+    for my $rec ( @{$m->{data}{recs}} ) {
         # for each holdings tag in the record...
         for my $holdidx ( @{$rec->{tmap}{ $c->{holdings} }} ) {
             my $tagid = $rec->{tags}[$holdidx]{tag};
 
-            print STDOUT "\r", $m->recno;
+            print STDOUT "\r$i";
             my @out = ();            # clear the output buffer
             push @out, $rec->{egid}; # slug in the egid first thing
-            print "BEGIN;\n\negid\t" if ($m->recno == 1);
+            print "BEGIN;\n\n" unless $i;
 
             # grab the unary mappings and slug 'em in
             for my $sub ( sort keys %{$rec->{tags}[$holdidx]{uni}} ) {
                 push @out, $rec->{tags}[$holdidx]{uni}{$sub};
-                print "l_", $m->name($tagid, $sub),"\t"
-                  if ($m->recno == 1);
+                #print "l_", $m->name($tagid, $sub),"\t"
+                #  unless $i;
+                # FIXME column names should be made workable again
             }
+
             for my $x (@{$rec->{tags}[$holdidx]{multi}{x}} ) {
                 print X $rec->{egid}, "\t",
                   $rec->{tags}[$holdidx]{uni}{ $c->{copyid} },
@@ -65,16 +79,16 @@ sub extract_holdings {
                 for my $sub ( sort keys %{$rec->{tags}[$idx]{uni}} ) {
                     push @out, $rec->{tags}[$idx]{uni}{$sub};
                     print "l_", $m->name($rec->{tags}[$idx]{tag}, $sub), "\t"
-                      if ($m->recno == 1);
+                      unless $i;
                 }
             }
 
             # and dump it
-            print "\n" if ($m->recno == 1);
+            print "\n" if ($i == 1);
             print join("\t", @out);
             print "\n";
         }
-
+        $i++;
     }
     select STDOUT;
     print "\n";
@@ -105,20 +119,29 @@ sub dump_sample_detail {
     my ($c, $s) = @_;
     my $tags = $s->{data}{samp};
     my $count = $s->{data}{rcnt};
+    my $scnt  = $s->{data}{scnt};
 
     open DETAIL, '>', ($c->{prefix} . "-HOLDINGS-DETAIL.txt");
     select DETAIL;
-    for my $tag (sort keys %{$tags}) {
+    for my $tag (sort keys %{ $tags }) {
         print ">>>>> TAG $tag\n\n";
-        for my $subkey (sort keys %{$tags->{$tag}}) {
+        for my $subkey (sort keys %{ $tags->{$tag} }) {
             my $sub = $tags->{$tag}{$subkey};
             print "|| $subkey | ", $sub->{value}, " | ", 
               $sub->{count}, "/", $sub->{tcnt}, " |  ||\n";
         }
+        print "\n";
     }
-    print "\n";
-    select STDOUT;
     close DETAIL;
+    open SCOUNT, '>', ($c->{prefix} . "-HOLDINGS-SUBCOUNTS.txt");
+    select SCOUNT;
+    for my $tag (sort keys %{ $scnt }) {
+        print ">>>>> TAG $tag\n\n";
+        for my $len (sort keys %{ $scnt->{$tag} }) 
+          { print "|| $len | ", $scnt->{$tag}{$len}, " ||\n" }
+        print "\n";
+    }
+    select STDOUT;
 }
 
 sub dump_sample_overview {
@@ -174,7 +197,7 @@ sub initialize {
     show_help("Nothing to do!")
       unless ($c->{map} or $c->{sample} or $c->{samplemap} or $c->{samplestr});
     show_help("map, holdings, and copyid must be specified together!")
-      if ($c->{map} and !$c->{holdings} and !$c->{copyid});
+      if ($c->{map} and !($c->{holdings} and $c->{copyid}));
     show_version() if $c->{version};
 
     my @keys = keys %{$c};