multi extractions work now. holding seq# always included now

[migration-tools.git] / extract_holdings

diff --git a/extract_holdings b/extract_holdings
index 22bc9c3..0557416 100755 (executable)
--- a/extract_holdings
+++ b/extract_holdings
@@ -3,10 +3,18 @@ use strict;
 use warnings;
 
 use Getopt::Long;
-use Equinox::Migration::MapDrivenMARCXMLProc;
+use Equinox::Migration::MapDrivenMARCXMLProc 1.003;
 use Equinox::Migration::MARCXMLSampler;
 
-my $VERSION = '1.000';
+my $VERSION = '1.001';
+
+=pod
+
+TODO
+
+  * Have detail mode report on number of subfields per datafield
+
+=cut
 
 my $c = initialize();
 $| = 1;
@@ -21,40 +29,44 @@ sub extract_holdings {
     my ($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
-    open X, '>', ($c->{prefix} . "-HOLDINGS-privnotes.pg");
-    open Z, '>', ($c->{prefix} . "-HOLDINGS-pubnotes.pg");
-    select HOLDINGS;
+    # create multi files
+    my @multifiles = <*HOLDINGS-MULT*>;
+    unlink @multifiles;
+
+    my $i = 0; # record counter
+    my $j = 0; # holdings counter
+    my $multis = $m->get_multis;
+
+    while (  $m->{data}{recs}[$i] ) {
+        print HOLDINGS "BEGIN;\n\negid\thseq\t" unless $j;
+        my $rec = $m->{data}{recs}[$i];
 
-    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, $rec->{egid}; # slug in the egid first thing
-            print "BEGIN;\n\negid\t" if ($m->recno == 1);
+            push @out, $j;           # holding seq goes next
 
             # 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 HOLDINGS "l_", $m->name($tagid, $sub),"\t" unless $j;
             }
-            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";
+
+            # handle holdings multis
+            for my $sub ( sort keys %{$multis->{$tagid}} ) {
+                open MULT, ">>", ($c->{prefix} . "HOLDINGS-MULT" . "$tagid$sub.pg");
+                for my $value ( @{$rec->{tags}[$holdidx]{multi}{$sub}} ) 
+                  { print MULT join("\t", $rec->{egid}, $j, $value), "\n" }
+                close MULT;
             }
 
 
@@ -65,18 +77,19 @@ 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 $j;
                 }
             }
 
             # and dump it
-            print "\n" if ($m->recno == 1);
-            print join("\t", @out);
-            print "\n";
+            print HOLDINGS "\n" unless $j;
+            print HOLDINGS join("\t", @out);
+            print HOLDINGS "\n";
+            $j++;
         }
-
+        $i++;
+        print "\r$i";
     }
-    select STDOUT;
     print "\n";
 }
 
@@ -85,6 +98,7 @@ sub extract_holdings {
 sub run_samples {
     my ($c) = @_;
     my $s;
+    print "Parsing records for sampling... ";
     if ($c->{samplemap}) {
         $s = Equinox::Migration::MARCXMLSampler->new( marcfile => $c->{marcfile},
                                                       mapfile  => $c->{samplemap});
@@ -94,8 +108,6 @@ sub run_samples {
     } else {
         $s = Equinox::Migration::MARCXMLSampler->new( marcfile => $c->{marcfile} );
     }
-    print "Parsing records for sampling... ";
-    $s->parse_records;
 
     dump_sample_overview($c, $s) if $c->{sample};
     dump_sample_detail($c, $s) if ($c->{samplemap} or $c->{samplestr});
@@ -105,22 +117,29 @@ sub dump_sample_detail {
     my ($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 {
@@ -176,7 +195,7 @@ 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};