use strict;
use utf8;
use File::Basename;

my @files = glob($ARGV[0]);
my $outdir = $ARGV[1];
my $debug = $ARGV[2];

die "No output directory given\n" unless -d $outdir;
$outdir =~ s/\\/\//g;         # backslash to forward
$outdir =~ s/([^\/])$/$1\//;  # add final slash if missing

foreach my $file(@files){
  my $outfile = $outdir . '/' . basename($file);
  open(CSV, '<', $file)||die "Cannot open $file for read:$!\n";
  binmode CSV;
  open(OUT, '>', $outfile)||die "Cannot open $outfile for write:$!\n";
  while (my $line = <CSV>){
    $line =~ s/\x0D\x0A/\n/g; # binary, so we're still stuck with \r\n dos endings possibly - why are we using binary?
    if($line =~ /[^[:ascii:]]/){
      print "Before: $line\n" if $debug;
      
      # translations from octal sequence to ascii char
      $line =~ s/\302\267/./g;         # odd utf 'floating' point to ascii .
      $line =~ s/\342\200\230/'/g;     # left single curly quote to ascii '
      $line =~ s/\342\200\231/'/g;     # right single curly quote to ascii '
      $line =~ s/\342\200\223/-/g;     # em-dash to ascii -
      $line =~ s/\303\257/i/g;         # double-dot i to ascii i
      $line =~ s/\302\243/GBP/g;       # pound sign to GBP
      $line =~ s/\342\200\246/.../g;   # elipsis to ascii ...
      $line =~ s/\302\256/(a)/g;       # @ to (a)
      $line =~ s/\303\250/e/g;         # grave e to e
      $line =~ s/\303\251/e/g;         # acute e to e
      $line =~ s/\342\211\244/\>\=/g;  # utf >= to ascii >=
      $line =~ s/\342\211\245/\<\=/g;  # utf <= to ascii <=
      $line =~ s/\303\264/o/g;         # circumflex o (?!?) to ascii o
      $line =~ s/\302\240/\s/g;        # nbsp to sp
      $line =~ s/\302\263/\^3/g;       # superscript 3 to ^3
      $line =~ s/\302\262/\^2/g;       # superscript 2 to ^2
      $line =~ s/\302\260/ degrees/g;  # degrees symbol to word ' degrees'
      $line =~ s/\342\200\235/""/g;    # right double curly quote to ascii " (escaped for csv)
      $line =~ s/\342\200\234/""/g;    # left double curly quote to ascii " (escaped for csv)
      $line =~ s/\302\275/1\/2/g;      # utf 1/2 to ascii plain 1/2
      
      if($line =~ /[^[:ascii:]]/){
        $line =~ s/([^[:ascii:]])/'[' . (ord $1) . '\/' . (sprintf("0x%X", (ord $1))) . '\/' . (sprintf("%o", (ord $1))) . ']'/ge;
        print "Unhandled sequence: $line\n";
      }
      print "After: $line\n" if $debug;
    }
    print OUT "$line";
  }
}