args handling bug

[migration-tools.git] / extract_holdings

diff --git a/extract_holdings b/extract_holdings
index 58353ad..742c6e2 100755 (executable)
--- a/extract_holdings
+++ b/extract_holdings
@@ -6,6 +6,15 @@ use Getopt::Long;
 use Equinox::Migration::MapDrivenMARCXMLProc;
 use Equinox::Migration::MARCXMLSampler;
 
+my $VERSION = '1.001';
+
+=pod
+
+TODO
+
+  * Have detail mode report on number of subfields per datafield
+
+=cut
 
 my $c = initialize();
 $| = 1;
@@ -25,11 +34,15 @@ sub extract_holdings {
     open HOLDINGS, '>', ($c->{prefix} . "-HOLDINGS.pg");
     # open the files for multi mappings
     # FIXME DO THIS
+    open X, '>', ($c->{prefix} . "-HOLDINGS-privnotes.pg");
+    open Z, '>', ($c->{prefix} . "-HOLDINGS-pubnotes.pg");
     select HOLDINGS;
 
     while (my $rec = $m->parse_record) {
         # 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;
             my @out = ();            # clear the output buffer
             push @out, $rec->{egid}; # slug in the egid first thing
@@ -38,10 +51,22 @@ sub extract_holdings {
             # 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_", $m->name($rec->{tags}[$holdidx]{tag}, $sub),"\t"
+                print "l_", $m->name($tagid, $sub),"\t"
                   if ($m->recno == 1);
             }
 
+            for my $x (@{$rec->{tags}[$holdidx]{multi}{x}} ) {
+                print X $rec->{egid}, "\t",
+                  $rec->{tags}[$holdidx]{uni}{ $c->{copyid} },
+                    "\t$x\n";
+            }
+            for my $z (@{$rec->{tags}[$holdidx]{multi}{z}} ) {
+                print Z $rec->{egid}, "\t",
+                  $rec->{tags}[$holdidx]{uni}{ $c->{copyid} },
+                    "\t$z\n";
+            }
+
+
             # now get everything else in the mapping
             for my $othertag ( sort keys %{$rec->{tmap}} ) {
                 next if $othertag eq $c->{holdings};  # ignoring the holdings, o'course
@@ -89,22 +114,29 @@ sub dump_sample_detail {
     my ($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}) {
-        print ">>>>> TAG $tag\n";
-        for my $subkey (sort keys %{$tags->{$tag}}) {
+    for my $tag (sort keys %{ $tags }) {
+        print ">>>>> TAG $tag\n\n";
+        for my $subkey (sort keys %{ $tags->{$tag} }) {
             my $sub = $tags->{$tag}{$subkey};
-            print "  Subfield: $subkey\n";
-            print "  Sample:   '", $sub->{value}, "'\n";
-            print "  Count:    ", $sub->{count}, " in ", $sub->{tcnt}, " tags\n\n";
-            #print "(", int($sub->{count} / $sub->{rcnt}), "%)\n";
+            print "|| $subkey | ", $sub->{value}, " | ", 
+              $sub->{count}, "/", $sub->{tcnt}, " |  ||\n";
         }
+        print "\n";
     }
-    select STDOUT;
-    print "\n";
     close DETAIL;
+    open SCOUNT, '>', ($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 {
@@ -150,16 +182,21 @@ sub initialize {
                          'marcfile|m=s',
                          'map=s',
                          'holdings|h=i',
+                         'copyid|c=s',
                          'prefix|p=s',
+                         'version|v',
                          'help',
                        );
     show_help() unless $rc;
     show_help() if ($c->{help});
-    show_help("map and holdings must be specified together!")
-      if ($c->{map} and !$c->{holdings});
+    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}));
+    show_version() if $c->{version};
 
     my @keys = keys %{$c};
-    for my $key ('prefix')
+    for my $key ('prefix', 'marcfile')
       { push @missing, $key unless $c->{$key} }
     if (@missing) {
         print "Required option: ", join(', ', @missing), " missing!\n";
@@ -194,8 +231,11 @@ HOLDINGS EXTRACTION ARGUMENTS
   --map          E::M::SM map file which will be used to extract holdings data
                  from the input MARC file
   --holdings -h  Specifies actual holdings tag
+  --copyid   -c  Specifies subfield of holdings with unique copy identifier
 
   Both these must be given together.
 HELP
     exit;
 }
+
+sub show_version { print "extract_holdings v$VERSION\n"; exit }