#!/usr/bin/perl -w
use strict;

my $FILE_IN  = $ARGV[0];
my $FILE_OUT = "patternmatcher.pl";
open FILE_IN,  $FILE_IN;
open FILE_OUT, ">$FILE_OUT";

print FILE_OUT "\#!/usr/bin/perl -w\n";
print FILE_OUT "\$seq = \$ARGV[0]\;\n";
print FILE_OUT "if (\n";

my $wild = "\\w\{0,2\}";
my $onN  = "X";
my $onN2 = "\[ATGCN\]";
my $name;
while (<FILE_IN>) {
    if ( $_ =~ /^\>(.+)$/ ) {
        $name = $1;
    }
    elsif ( $_ =~ /^(\w+)$/ ) {
        my $seq = $1;
        my $len = length $seq;
        print FILE_OUT "\# $name\n";

        for my $i ( 0 .. $len ) { #go through $seq base by base, replacing one character with the wild card
            my $newseq = $seq;
            substr( $newseq, $i, 1 ) = $wild;
            $newseq =~ s/N/$onN/go;  # these to lines are to allow Ns to match any character
            $newseq =~ s/X/$onN2/go; #
            print FILE_OUT "  \$seq =~ \/$newseq\/ or \n";
        }
        print FILE_OUT "  \$seq =~ \/ZZZ\/) \{\n"; #impossible match to close the if
        print FILE_OUT "    \$match=\$\&\;\n";
        print FILE_OUT "    \$name = \"$name\"\;\n";
        print FILE_OUT "    \$matchlength=length\(\$match\)\;\n";
        print FILE_OUT "    \$firstpos=index\(\$seq,\$match)\;\n";
        print FILE_OUT
          "    print \"\$name, \$match, \$matchlength, \$firstpos \\n\"\;\n";
        print FILE_OUT "    print \"Got it!\\n\"\;\n";
        print FILE_OUT "\} elsif \(\n";

    }
}
print FILE_OUT "\"false\"\)\{\n"; # false value to end the series of elsifs

print FILE_OUT "\}\n";
close FILE_IN;
close FILE_OUT;

system( "chmod", "u+x", $FILE_OUT );