#!/usr/bin/perl
# This will replace annoying character's with normal ones
# run in Terminal by typing
# perl program.pl oldfilename:newfilename
# version 2, for use with an applescript :)
# version 3, for handling Mac Roman characters (in addition to the plain text)
# also catching MacRoman newlines (in addition to the plain text new lines)

use utf8;
$input = shift or die "Require original_filename:new_filename.\n"; chomp $input;
$input =~ /(\S*):(\S*)/;
$filename = $1;
$outfile = $2;
#print "$filename\n";
#print "$outfile\n";

#Things we pull out
$rightquote = "\322";
$macromanrightquote = "<D2>";

$leftquote = "\323";
$macromanleftquote = "<D3>";

$apost = "\325";
$macromanapost = "<D5>";

$longdash = "\321";
$macromanlongdash = "<D1>";

$shortdash = "\320";
$macromanshortdash = "<D0>";

$l_bullet = "l ";
$colon_bullet = "\; ";

#Things we put in
$regulardouble = "\x22";
$regularsingle = "\x27";
$regulardash = "\x2D";
$correct_bullet = "<li>";



open (OUTFILE, ">$outfile");

open (FILE, "$filename");

while  ($line = <FILE>) {
        utf8::upgrade($line);
        #print $line;
        #$line =~ s/A/a/g;
        $line =~ s/$apost/$regularsingle/g;
        $line =~ s/$macromanapost/$regularsingle/g;
        $line =~ s/$rightquote/$regulardouble/g;
        $line =~ s/$macromanrightquote/$regulardouble/g;
        $line =~ s/$leftquote/$regulardouble/g;
        $line =~ s/$macromanleftquote/$regulardouble/g; 
	$line =~ s/$longdash/$regulardash$regulardash/g;
	$line =~ s/$macromanlongdash/$regulardash$regulardash/g;
	$line =~ s/$shortdash/$regulardash/g;   
	$line =~ s/$macromanshortdash/$regulardash/g;  
	$line =~ s/^$l_bullet/$correct_bullet/g;
	$line =~ s/\r$l_bullet/\n$correct_bullet/g;  #converts from Mac returns with bullets to Unix returns with bullets
    $line =~ s/\r$colon_bullet/\n$correct_bullet/g;  #converts from Mac returns with bullets to Unix returns with bullets
    $line =~ s/\r/\n/g;  #converts from Mac returns to Unix returns
	$line =~ s/^$colon_bullet/$correct_bullet/g;
        print OUTFILE $line;
}
close (FILE);

close (OUTFILE);