X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=emig.d%2Fbin%2Fmig-bibstats;h=0161176adc8fb58ed0f20df3e04bf22947695439;hp=cfb382a4f8a8028289ed326ac41908e541d865c6;hb=1c61b119580d507b820b9d209ac56f7ad297cc7a;hpb=3aaa60eb13b14b0019b79fd2228cbc0a30e42d01

diff --git a/emig.d/bin/mig-bibstats b/emig.d/bin/mig-bibstats
index cfb382a..0161176 100755
--- a/emig.d/bin/mig-bibstats
+++ b/emig.d/bin/mig-bibstats
@@ -24,6 +24,10 @@ with one of the ILSes so provide the name
 --exportbarcodesfile will use this file name for a barcode export instead 
 of the generic 'barcodes_export.txt'
 
+--item_type_subfield will make use of --holding_code and provide a breakdown of
+bib types by item types.  If --branch_subfield is also provided, then the
+breakdown will be further subdivided by branch.
+
 =back
 =cut
 
@@ -60,6 +64,8 @@ my $file;
 my $uri_threshold = 1;
 my $p_holding_code;
 my $p_barcode_subfield;
+my $p_item_type_subfield;
+my $p_branch_subfield;
 my $p_ils_name = '';
 my $holding_threshold = 50;
 my $p_ignore_filetype = 'false';
@@ -69,15 +75,17 @@ my $exportbarcodes;
 my $exportbarcodesfile;
 
 my $ret = GetOptions(
-    'file:s'           		 => \$file,
-	'uri_threshold:i'  		 => \$uri_threshold,
-	'holding_code:s'   		 => \$p_holding_code,
-	'barcode_subfield:s'     => \$p_barcode_subfield,
-	'ignore_filetype:s'		 => \$p_ignore_filetype,
-	'ils:s'	        		 => \$p_ils_name,
-	'exportbarcodes:s'		 => \$exportbarcodes,
-	'exportbarcodesfile:s'   => \$exportbarcodesfile,
-	'holding_threshold:s'	 => \$holding_threshold
+    'file:s'                    => \$file,
+    'uri_threshold:i'           => \$uri_threshold,
+    'holding_code:s'            => \$p_holding_code,
+    'barcode_subfield:s'     => \$p_barcode_subfield,
+    'item_type_subfield:s'     => \$p_item_type_subfield,
+    'branch_subfield:s'     => \$p_branch_subfield,
+    'ignore_filetype:s'         => \$p_ignore_filetype,
+    'ils:s'                     => \$p_ils_name,
+    'exportbarcodes:s'         => \$exportbarcodes,
+    'exportbarcodesfile:s'   => \$exportbarcodesfile,
+    'holding_threshold:s'     => \$holding_threshold
 );
 
 if ($exportbarcodesfile and !defined $exportbarcodes) { abort('You have to provide an ILS name if you want a barcode export file.'); }
@@ -85,36 +93,49 @@ if ($exportbarcodesfile and !defined $exportbarcodes) { abort('You have to provi
 if ($p_holding_code and length $p_holding_code != 3) { abort('Holdings codes must be three characters.'); }
 
 if ($p_barcode_subfield) {
-	if (!defined $p_holding_code) { abort('A barcode field can not be used without a holding code.'); }
-	if (length $p_barcode_subfield != 1) { abort('Barcode subfields must be a single character code.'); }
+    if (!defined $p_holding_code) { abort('A barcode field can not be used without a holding code.'); }
+    if (length $p_barcode_subfield != 1) { abort('Barcode subfields must be a single character code.'); }
+}
+
+if ($p_item_type_subfield) {
+	if (!defined $p_holding_code) { abort('An item type field can not be used without a holding code.'); }
+	if (length $p_item_type_subfield != 1) { abort('Item type subfields must be a single character code.'); }
+}
+
+if ($p_branch_subfield) {
+	if (!defined $p_holding_code) { abort('A branch field can not be used without a holding code.'); }
+	if (length $p_branch_subfield != 1) { abort('Branch subfields must be a single character code.'); }
 }
 
 # ils name, holding tag, barcode subfield 
 my @ilses = (
-	['Mandarin','852','p'],
-	['Evergreen','852','p'],
-	['Polaris','852','p'],
-	['TLC','949','g'],
-	['Koha','952','p'],
-	['Sympony','999','i'],
+    ['Mandarin','852','p'],
+    ['Evergreen','852','p'],
+    ['Polaris','852','p'],
+    ['TLC','949','g'],
+    ['Koha','952','p'],
+    ['Sympony','999','i'],
     ['Destiny','852','p']
 );
 
 my @temp;
-if ($p_holding_code) {
-	push @temp, $p_ils_name;
-	push @temp, $p_holding_code;
-	if ($p_barcode_subfield) { push @temp, lc $p_barcode_subfield; }
-	push @ilses, [@temp];
+if (defined $p_holding_code && defined $p_ils_name && defined $p_barcode_subfield) {
+    push @temp, $p_ils_name;
+    push @temp, $p_holding_code;
+    if ($p_barcode_subfield) { push @temp, lc $p_barcode_subfield; }
+    push @ilses, [@temp];
 }
 
 #to do - add a check for exportbarcodes being in @ilses
 
-my $batch = MARC::Batch->new('USMARC', $file);
-$batch->strict_off();
 my $filetype = `file $file`;
-if ($filetype =~ m/MARC21/ or $p_ignore_filetype eq 'true') { print "$filetype.\n" }
-    else { abort("File is not MARC21."); }
+my $batch;
+if ($filetype =~ m/MARC21/) {
+    $batch = MARC::Batch->new( 'USMARC', $file );
+} else {
+    $batch = MARC::Batch->new( 'XML', $file );
+}
+$batch->strict_off();
 
 my $i = 0;
 my $uri_count = 0;
@@ -126,65 +147,84 @@ my @uris;
 my @fields;
 my @encodings;
 my @types;
+my %bib_types_by_item_type;
 my @holding_code_strings;
 my %holding_counts;
 my %barcode_counts;
 
 foreach (@ilses) { 
-	$holding_counts{@$_[0]} = 0; 
-	$barcode_counts{@$_[0]} = 0;
+    $holding_counts{@$_[0]} = 0; 
+    $barcode_counts{@$_[0]} = 0;
 }
 
 while ( my $record = $batch->next() ) {
     $i++;
-	#check holdings, bit time consuming but more future proof
-	foreach (@ilses) {
-		my $ils = @$_[0];
-		my $hcode = @$_[1];
-		my $barcode = @$_[2];
-		my @holding_fields = $record->field($hcode);
+    #check holdings, bit time consuming but more future proof
+    foreach (@ilses) {
+        my $ils = @$_[0];
+        my $hcode = @$_[1];
+        my $barcode = @$_[2];
+        my @holding_fields = $record->field($hcode);
         foreach my $hf (@holding_fields) {
-			my @h;
-			my $barcode_string = $hf->subfield($barcode);
-        	push @h, $ils;
-        	push @h, $barcode_string;
-			push @holdings, [@h];
-		}
-		my $l = scalar @holding_fields;
-		my $v = $holding_counts{$ils};
-		if ($l) { $holding_counts{$ils} = $v + $l; }
-	}
+            my @h;
+            my $barcode_string = $hf->subfield($barcode);
+            push @h, $ils;
+            push @h, $barcode_string;
+            push @holdings, [@h];
+        }
+        my $l = scalar @holding_fields;
+        my $v = $holding_counts{$ils};
+        if ($l) { $holding_counts{$ils} = $v + $l; }
+    }
     #process 856s
-	@fields = $record->field('856');
-	my $enc = substr $record->leader(), 9, 1;
-	push @encodings, $enc;
+    @fields = $record->field('856');
+    my $enc = substr $record->leader(), 9, 1;
+    push @encodings, $enc;
     my $type = substr $record->leader(), 6, 1;
     push @types, $type;
-	foreach my $f (@fields) {
-		my $u = $f->subfield('u');
+    # bib type by branch and by item type if item subfield (and optionally branch subfield) provided
+    if (defined $p_holding_code && defined $p_item_type_subfield) {
+        my @holding_fields = $record->field($p_holding_code);
+        foreach my $hf (@holding_fields) {
+            my $item_type = $hf->subfield($p_item_type_subfield) || '<missing item type subfield>';
+            my $branch = $p_branch_subfield ? $hf->subfield($p_branch_subfield) : 'default';
+            if (! defined $bib_types_by_item_type{ $branch }) {
+                $bib_types_by_item_type{ $branch } = {};
+            }
+            if (! defined $bib_types_by_item_type{ $branch }{ $type }) {
+                $bib_types_by_item_type{ $branch }{ $type } = {};
+            }
+            if (! defined $bib_types_by_item_type{ $branch }{ $type }{ $item_type }) {
+                $bib_types_by_item_type{ $branch }{ $type }{ $item_type } = 0;
+            }
+            $bib_types_by_item_type{ $branch }{ $type }{ $item_type }++;
+        }
+    }
+    foreach my $f (@fields) {
+        my $u = $f->subfield('u');
         my $n = $f->subfield('9');
         if (defined $n) { $uri_sub9_count++; }
-		if (defined $u) {
-			$uri_count++;
-			my $ind1 = $f->indicator('1');
-			my $ind2 = $f->indicator('2');
-			if ($ind1 eq '4') {
-				if ($ind2 eq '0' or $ind2 eq '1') { $uri_valid_count++; }
-			}
-			my $ustring = lc $f->as_string('u');
-			$ustring =~ s/http:\/\///;
+        if (defined $u) {
+            $uri_count++;
+            my $ind1 = $f->indicator('1');
+            my $ind2 = $f->indicator('2');
+            if ($ind1 eq '4') {
+                if ($ind2 eq '0' or $ind2 eq '1') { $uri_valid_count++; }
+            }
+            my $ustring = lc $f->as_string('u');
+            $ustring =~ s/http:\/\///;
             $ustring =~ s/ftp:\/\///;
-			$ustring =~ s/https:\/\///;
-			$ustring =~ s/\/.*//;
-			push @uris, $ustring;
-		}
-	}
+            $ustring =~ s/https:\/\///;
+            $ustring =~ s/\/.*//;
+            push @uris, $ustring;
+        }
+    }
     #check for authority linking on 100s and 245s, if present may need to scrub them
-	@fields = $record->field('100');
-	foreach my $f (@fields) {
-		my $t = $f->subfield('0');
-		if (defined $t) { $title_sub0++; }	
-	}
+    @fields = $record->field('100');
+    foreach my $f (@fields) {
+        my $t = $f->subfield('0');
+        if (defined $t) { $title_sub0++; }    
+    }
     @fields = $record->field('245');
     foreach my $f (@fields) {
         my $t = $f->subfield('0');
@@ -194,11 +234,11 @@ while ( my $record = $batch->next() ) {
 }
 
 foreach (@ilses) {
-	my $ils = @$_[0];
+    my $ils = @$_[0];
     my @temp_barcodes;
     foreach my $h (@holdings) {
-		my $temp_ils_name = @$h[0];
-    	if ($temp_ils_name eq $ils) { push @temp_barcodes, @$h[1]; }
+        my $temp_ils_name = @$h[0];
+        if ($temp_ils_name eq $ils) { push @temp_barcodes, @$h[1]; }
     }
     my @uniq_barcodes = uniq @temp_barcodes;;
     $barcode_counts{$ils} = scalar @uniq_barcodes;
@@ -231,6 +271,19 @@ foreach my $key (keys %type_counts) {
 }
 print "\n";
 
+if ($p_item_type_subfield) {
+    print "===== Branch / Leader 06 / Item Type\n";
+    foreach my $branch (keys %bib_types_by_item_type) {
+        foreach my $btype (keys %{ $bib_types_by_item_type{$branch} }) {
+            foreach my $itype (keys %{ $bib_types_by_item_type{$branch}{$btype} }) {
+                my $count = $bib_types_by_item_type{$branch}{$btype}{$itype};
+                print "$branch\t$btype (" . give_type($btype) . ")\t$itype\t$count\n";
+            }
+        }
+    }
+    print "\n";
+}
+
 print "===== Summary of Select Field Counts\n";
 print "  $uri_count 856 fields with a subfield u\n";
 print "  $uri_valid_count 856 fields with a subfield u and valid indicators\n";
@@ -240,14 +293,14 @@ print "  $author_sub0 245 fields have a subfield 0\n";
 
 print "\n===== Holdings Analysis\n";
 foreach my $key (keys %holding_counts) {
-	my $c = $holding_counts{$key};
-	if (((100/$i)*$c) >= $holding_threshold) { print "  $key $holding_counts{$key} holdings in $i bibs with $barcode_counts{$key} unique barcodes\n"; }
+    my $c = $holding_counts{$key};
+    if (((100/$i)*$c) >= $holding_threshold) { print "  $key $holding_counts{$key} holdings in $i bibs with $barcode_counts{$key} unique barcodes\n"; }
 }
 
 print "\n===== URI values are domains and filtered to only show those with more than $uri_threshold\n";
 foreach my $key (keys %uri_counts) {
-	my $value = $uri_counts{$key};
-	if ($value > $uri_threshold) { print "  $key   $value\n"; } 
+    my $value = $uri_counts{$key};
+    if ($value > $uri_threshold) { print "  $key   $value\n"; } 
 }
 
 if ($exportbarcodes) {
@@ -257,8 +310,8 @@ if ($exportbarcodes) {
     open my $out_fh, '>:utf8', $outfile or abort('can not open output file for barcode list');
     foreach my $h (@holdings) {
         my $temp_ils_name = @$h[0];
-		my $barcode = @$h[1];
-		if (!defined $barcode) { $barcode = 'no barcode found'; }
+        my $barcode = @$h[1];
+        if (!defined $barcode) { $barcode = 'no barcode found'; }
         if ($temp_ils_name eq $exportbarcodes) { print $out_fh "@$h[1]\n" }
     }
     close $out_fh;
@@ -275,7 +328,7 @@ sub abort {
 }
 
 sub give_type {
-	my $type = shift;
+    my $type = shift;
     if ($type eq 'a') { return 'Language material'; }
     if ($type eq 'c') { return 'Notated Music'; }
     if ($type eq 'd') { return 'Manuscript notated music'; }