#!/usr/local/perl -w
#printhh.perl - perl script for printing household data
#
open(INPUT,$ARGV[0]);         #file specified on command line
open(RUNINFO,">run_info");    #file for writing run information
open(ERROR,">error_records"); #file for writing $status==0 records
#
$lpid       = -1; #initialize "last" $pid value for reading 1st record
$lhhn       =  0; #initialize "last" $hhn value for reading 1st record
$records    =  0; #initialize counter for records read
$persons1   =  0; #initialize counter for persons with $status == 1
$persons0   =  0; #initialize counter for persons with $status == 0
$personsh   =  0; #initialize counter for persons in households printed
$households =  0; #initialize counter for households printed
@hh         = (); #initialize array for persons in household
#
while (<INPUT>) {
#	compute status code to identify invalid records ($status == 0)
#	because $pid doesn't always count up to $hhn as it should
	$pid = substr($_,16,2); #person's id number
	$hhn = substr($_,50,2); #number of persons in household
	if ($pid == 1) { $status = 1; }
	elsif ($pid > 1 &&
	       $pid == $lpid + 1 &&
	       $pid <= $hhn &&
	       $hhn == $lhhn ) { $status = 1; }
	else { $status = 0; }
	$lpid = $pid; $lhhn = $hhn;
#	print $status," ",$pid," ",$hhn,"\n"; #uncomment for checking
#
#	accumulate person records for each household ($status == 0)
#	in array @hh, then print information from this array
	$records++;
	if ($status == 1) {
		$persons1++;
		$hh[$pid-1] = $_;
		if ($pid == $hhn) { 
			$personsh = $personsh + scalar(@hh);
			$households++;
			foreach $person (@hh) {
				&print_person($person);
			}
			print "\n";
			@hh = ();
		}
	}
	else {
		$persons0++;
		print ERROR substr($_,1,15), " ",substr($_,16);
	}
#	no check here for $pid not reaching $hhn; household will not
#	be printed in this case
}
print RUNINFO 'Records read:                  ',$records,"\n";
print RUNINFO 'Persons with $status == 1:     ',$persons1,"\n";
print RUNINFO 'Persons with $status == 0:     ',$persons0,"\n";
print RUNINFO 'Households written:            ',$households,"\n";
print RUNINFO 'Persons in households written: ',$personsh,"\n";
$discrepancy = $persons1 - $personsh;
print RUNINFO 'Discrepancy:                   ',$discrepancy,"\n";
#
sub print_person {
	$rhh = substr($person,18,1);
	$rhh =~ tr/12345678/HSCDPGON/;
	$sex = substr($person,19,1); 
	$sex =~ tr/12/MF/;
	$yrb = substr($person,20,3); #year of birth
	$mnb = substr($person,23,2); #month of birth
	$yrb = 1000 + $yrb;
	$adj = $mnb eq "07" || $mnb eq "08" || $mnb eq "09" ;
	$adj = $adj || $mnb eq "10" || $mnb eq "11" || $mnb eq "12";
	$age = (1990 - $yrb) - $adj; 
	if (length($age) eq 1) {$age = "0".$age}
	$mar = substr($person,42,1);
	$mar =~ tr/01234/-SMWD/;
	$kids = substr($person,43,4);
	if ($sex eq 'M') { $kids = '----'; }
	if ($age <= 10)  { $kids = '----'; }
#
	print $rhh," ",$sex," ",$age," ",$mar," ",$kids,"\n";
}
#
#	$prv = substr($_, 0,2); #province
#	$prf = substr($_, 2,2); #prefecture
#	$cnt = substr($_, 4,2); #county 
#	$twn = substr($_, 6,3); #township
#	$vil = substr($_, 9,2); #village
#	$ced = substr($_,11,2); #census enumeration district
#	$hid = substr($_,13,3); #household id number
#	$pid = substr($_,16,2); #person id number
#	$rhh = substr($_,18,1); #relation to head of household
#	$sex = substr($_,19,1); #sex
#	$yrb = substr($_,20,3); #year of birth
#	$mnb = substr($_,23,2); #month of birth
#	$eth = substr($_,25,2); #ethnicity
#	$hhr = substr($_,27,1); #household registration
#	$tyr = substr($_,28,1); #type of registration
#	$mig = substr($_,29,2); #migration
#	$res = substr($_,31,1); #type of residence at origin
#	$rea = substr($_,32,1); #reason for migration
#	$eda = substr($_,33,1); #educational attainment
#	$eds = substr($_,34,1); #educational enrollment status
#	$ind = substr($_,35,3); #industry
#	$occ = substr($_,38,3); #occupation
#	$emp = substr($_,41,1); #employment status
#	$mar = substr($_,42,1); #marital status
#	$cbm = substr($_,43,1); #children born, male
#	$cbf = substr($_,44,1); #children born, female
#	$csm = substr($_,45,1); #children surviving, male
#	$csf = substr($_,46,1); #children surviving, female
#	$hht = substr($_,49,1); #household type
#	$hhn = substr($_,50,2); #number of persons in household