#!/usr/bin/perl
# -*- coding: iso-8859-15 -*-
###############################################################################
=pod

=item B<bibstats> --file foo.mrc

Reads through a marc file to generate statistical information about the file 
for quick analysis.

--uri_threshold defaults to 1, only shows URI values with more than that 
frequency

--ingore_filetype true will have it not care what file returns as the type and 
always treat it as marc21
=back

=cut

###############################################################################

use strict;
use warnings;

use Data::Dumper;
use Env qw(
    HOME PGHOST PGPORT PGUSER PGDATABASE MIGSCHEMA
    MIGBASEWORKDIR MIGBASEGITDIR MIGGITDIR MIGWORKDIR
);
use Pod::Usage;
use Switch;
use Getopt::Long;
use MARC::Batch;
use MARC::Record;
use MARC::Field;
use Cwd 'abs_path';
use Cwd qw(getcwd);
use FindBin;
my $mig_bin = "$FindBin::Bin/";
use lib "$FindBin::Bin/";
use EMig;
use open ':encoding(utf8)';

pod2usage(-verbose => 2) if defined $ARGV[0] && $ARGV[0] eq '--help';
pod2usage(-verbose => 1) if ! $ARGV[1];

my $file;
my $uri_threshold = 1;
my $p_holding_code;
my $p_barcode_subfield;
my $p_ils_name = 'Runtime ILS';
my $holding_threshold = 50;
my $p_ignore_filetype = 'false';

my $ret = GetOptions(
    'file:s'           		=> \$file,
	'uri_threshold:i'  		=> \$uri_threshold,
	'holding_code:s'   		=> \$p_holding_code,
	'barcode:s'        		=> \$p_barcode_subfield,
	'ignore_filetype:s'		=> \$p_ignore_filetype,
	'ils_name:s'	   		=> \$p_ils_name,
	'holding_threshold:s'	=> \$holding_threshold
);

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.'); }
}

my @ilses = (
	['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;



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 $i = 0;
my $uri_count = 0;
my $uri_valid_count = 0;
my $uri_sub9_count = 0;
my $author_sub0 = 0;
my $title_sub0 = 0;
my @uris;
my @fields;
my @codes;
my @holding_code_strings;
my %holding_counts;
my %barcode_counts;

foreach (@ilses) { 
	$holding_counts{@$_[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);
		my $l = scalar @holding_fields;
		my $v = $holding_counts{$ils};
		if ($l) { $holding_counts{$ils} = $v + $l; }
	}
    #process 856s
	@fields = $record->field('856');
	my $ldr = substr $record->leader(), 9, 1;
	push @codes, $ldr;
	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:\/\///;
            $ustring =~ s/ftp:\/\///;
			$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('245');
    foreach my $f (@fields) {
        my $t = $f->subfield('0');
        if (defined $t) { $author_sub0++; }
    }
    if(($i % 1000) == 0) { print "Processing bib $i.\n"; }
}

my %uri_counts;
$uri_counts{$_}++ for @uris;

my %code_counts;
$code_counts{$_}++ for @codes;

print "\n$filetype\n";
print "$i bibs read in file\n\n";

print "=== Leader 09 codes\n";
foreach my $key (keys %code_counts) {
    my $value = $code_counts{$key};
    print "=== $key   $value\n"; 
}
print "\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";
print "$uri_sub9_count 856 fields have subfield 9s\n";
print "$title_sub0 100 fields have a subfield 0\n";
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 "Could be $key $holding_counts{$key} holdings tags\n"; }
}

print "\nURI 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"; } 
}

close $file;

########### functions

sub abort {
    my $msg = shift;
    print STDERR "$0: $msg", "\n";
    exit 1;
}