Call Number Normalizers

CREATE OR REPLACE FUNCTION asset.label_normalizer_mvlc(text, integer)
RETURNS text AS $BODY$
use strict;
use warnings;
    my $callnum = uc(shift); # All uppercase
    my $mode = shift;

    if ($mode) { # Dewey number expansion (second group(s) are left-aligned)
        $callnum =~ s/([0-9]+)((\.|\s+)([0-9]+))?/sprintf(" %015.15s", $1) . ($2 ? $3 . do {my $tmp = sprintf("%-15.15s", $4); $tmp =~ tr! !0!; $tmp . ' '; }: '.000000000000000 ')/eg;
    } else { # Normal number expansion (everything is right-aligned)
        $callnum =~ s/([0-9]+)/sprintf("%015.15s", $1)/eg;
    }
    $callnum =~ s/([0-9.]+)/ $1 /g; # Add spaces before/after numbers
    # This does *not* preserve . or _ intentionally
    $callnum =~ s/[^A-Z0-9]/ /g; # Replace special characters with spaces
    $callnum =~ s/ {2,}/ /g; # Collapse multiple spaces to 1
    $callnum =~ s/^\s//g; # Remove leading spaces
    $callnum =~ s/\s$//g; # Remove trailing spaces

    return $callnum;
$BODY$ LANGUAGE plperlu;

CREATE OR REPLACE FUNCTION asset.label_normalizer_mvlc_generic(text)
RETURNS text AS $BODY$
SELECT asset.label_normalizer_mvlc($1, 0);
$BODY$ LANGUAGE sql;

CREATE OR REPLACE FUNCTION asset.label_normalizer_mvlc_dewey(text)
RETURNS text AS $BODY$
SELECT asset.label_normalizer_mvlc($1, 1);
$BODY$ LANGUAGE sql;

UPDATE asset.call_number_class SET normalizer = 'asset.label_normalizer_mvlc_generic' WHERE id = 1;
UPDATE asset.call_number_class SET normalizer = 'asset.label_normalizer_mvlc_dewey' WHERE id = 2;
UPDATE asset.call_number SET label = label WHERE label_class IN (1,2);


Example output:
200.5 Dan
Orig Generic: 200.5 DAN
Orig Dewey:   200_500000000000000_DAN
New Generic:  000000000000200 000000000000005 DAN
New Dewey:    000000000000200 500000000000000 DAN

200.6 Dan
Orig Generic: 200.6 DAN
Orig Dewey:   200_600000000000000_DAN
New Generic:  000000000000200 000000000000006 DAN
New Dewey:    000000000000200 600000000000000 DAN

200.52 Dan
Orig Generic: 200.52 DAN
Orig Dewey:   200_520000000000000_DAN
New Generic:  000000000000200 000000000000052 DAN
New Dewey:    000000000000200 520000000000000 DAN

Disc 1-3
Orig Generic: DISC 1 3
Orig Dewey:   DISC_13
New Generic:  DISC 000000000000001 000000000000003
New Dewey:    DISC 000000000000001 000000000000000 000000000000003 000000000000000

Disc 10-12
Orig Generic: DISC 10 12
Orig Dewey:   DISC_1012
New Generic:  DISC 000000000000010 000000000000012
New Dewey:    DISC 000000000000010 000000000000000 000000000000012 000000000000000

Fic/King
Orig Generic: FIC KING
Orig Dewey:   FICKING
New Generic:  FIC KING
New Dewey:    FIC KING

REF 973.2 AME v.1
Orig Generic: REF 973.2 AME V.1
Orig Dewey:   REF_973_200000000000000_AME_V_1
New Generic:  REF 000000000000973 000000000000002 AME V 000000000000001
New Dewey:    REF 000000000000973 200000000000000 AME V 000000000000001 000000000000000

REF 973.2 AME v.2
Orig Generic: REF 973.2 AME V.2
Orig Dewey:   REF_973_200000000000000_AME_V_2
New Generic:  REF 000000000000973 000000000000002 AME V 000000000000002
New Dewey:    REF 000000000000973 200000000000000 AME V 000000000000002 000000000000000

REF 973.2 AME v.10
Orig Generic: REF 973.2 AME V.10
Orig Dewey:   REF_973_200000000000000_AME_V_10
New Generic:  REF 000000000000973 000000000000002 AME V 000000000000010
New Dewey:    REF 000000000000973 200000000000000 AME V 000000000000010 000000000000000

Season 1 Disc 1
Orig Generic: SEASON 1 DISC 1
Orig Dewey:   SEASON_1_DISC_100000000000000
New Generic:  SEASON 000000000000001 DISC 000000000000001
New Dewey:    SEASON 000000000000001 000000000000000 DISC 000000000000001 000000000000000

Season 1 Disc 2
Orig Generic: SEASON 1 DISC 2
Orig Dewey:   SEASON_1_DISC_200000000000000
New Generic:  SEASON 000000000000001 DISC 000000000000002
New Dewey:    SEASON 000000000000001 000000000000000 DISC 000000000000002 000000000000000

Season 1 Disc 10
Orig Generic: SEASON 1 DISC 10
Orig Dewey:   SEASON_1_DISC_100000000000000
New Generic:  SEASON 000000000000001 DISC 000000000000010
New Dewey:    SEASON 000000000000001 000000000000000 DISC 000000000000010 000000000000000


Example call numbers ordered by:

Orig Generic:
200.5 Dan
200.52 Dan
200.6 Dan
Disc 1-3
Disc 10-12
Fic/King
REF 973.2 AME v.1
REF 973.2 AME v.10
REF 973.2 AME v.2
Season 1 Disc 1
Season 1 Disc 10
Season 1 Disc 2

Orig Dewey:
200.5 Dan
200.52 Dan
200.6 Dan
Disc 10-12
Disc 1-3
Fic/King
REF 973.2 AME v.1
REF 973.2 AME v.10
REF 973.2 AME v.2
Season 1 Disc 1
Season 1 Disc 10
Season 1 Disc 2

New Generic:
200.5 Dan
200.6 Dan
200.52 Dan
Disc 1-3
Disc 10-12
Fic/King
REF 973.2 AME v.1
REF 973.2 AME v.2
REF 973.2 AME v.10
Season 1 Disc 1
Season 1 Disc 2
Season 1 Disc 10

New Dewey:
200.5 Dan
200.52 Dan
200.6 Dan
Disc 1-3
Disc 10-12
Fic/King
REF 973.2 AME v.1
REF 973.2 AME v.2
REF 973.2 AME v.10
Season 1 Disc 1
Season 1 Disc 2
Season 1 Disc 10