#!/bin/perl
# A script to extract data from Covid19Clustermap20200331Japan-Detail.txt
# (Version 2 on DOI:10.17605/OSF.IO/52NVS 2025-03-15)
#Author: TANAKA Sigeto
# Graduate School of Arts and Letters, Tohoku University
#  Kawauti 27-1, Aoba Ku, Sendai, Miyagi, 980-8576 JAPAN
# WWW: http://www.sal.tohoku.ac.jp/~tsigeto/

$/ = "";
$\ = "\n" ;
$" = $,  = "\t" ;

%Nmonth = (
	Jan => 1, 
	Feb => 2, 
	Mar => 3, 
	Apr => 4, 
) ;

print qw( 
	id size municipality
	date0 date1
	category direction
	month day
) ;

while(<>){

	next unless /^\#\s*data/i ... eof() ;
	next unless /^ID/ ;

	my %field;
	foreach (split /\n/) {
		s/\s+$//;
		my($k,$v) = split /:\s*/, $_, 2 ;
		$field{$k} = $v  unless $v eq 'Source';
	}

	my $category = $field{Category};
	$category =~ s/\s*\(.*$// ;

	my $direction = $field{Direction};
	$direction =~ s/^(\w+).*$/$1/;

	my($m, $day) = $field{'Date(5th)'} =~ /^(Jan|Feb|Mar|Apr)\.\s*(\d+)/ ;
	my $month = $Nmonth{$m} ;
	
	print @field{ 'ID', 'Size', 'Municipality', 'Date(occurrence)', 'Date(1st)'}, $category, $direction, $month, $day ;
}