cholera_i

#!/usr/bin/perl -w

use strict;

my $preposition = $ARGV[0] ? $ARGV[0] :  'i';

my $dir = 'yearly' . ( $preposition ne 'i' ? $preposition : '');

system("mkdir $dir");
system("rm $dir/*.text*");

for my $year (1848 .. 1858) {
    system("touch ./$dir/$year\.text" );
}

if(open(my $pipe,'grep -aP \'[0123456789]\s((ch)|(k))olera\s' . $preposition . '\s\' trigrams/* | sort -n -k1.10|')) {
    while(my $line = <$pipe>) {
#	print $line;
	chomp $line;
	$line =~ m/^trigrams\/(\d+).text:.*/;
	my $week = $1;
	my $third_word = $line;
	if($third_word =~s/^.+olera\s+$preposition\s+//) {
	    my $year = `grep $week\.text year_by_weekly.text`;
	    chomp $year;
	    $year =~ s/\sweek.*$//;
	    next if $year =~ m/weekly/;
	    next if $year =~ m/1847|1859|1860/;
	    if(open(my $olera,">>./$dir/$year".'.text')) {
		print $olera "$third_word\n";
		close $olera;
	    } else {
		print STDERR 
	    }
	}
    }
} else {
    die "Cannot open $pipe: $!";
}