#!/usr/bin/perl -w

# Read raw font data from atari rom images, plus a few hand-drawn
# "text bitmaps" from __DATA__.

# Write 3 bitmap font "txt" files in the format txt2psf expects, then
# runs txt2psf on them. Results in 3 fonts:

# fauxtari-8.psf - 8x8 native size
# fauxtari-16.psf - 16x16 scaled up
# fauxtari-24.psf - 24x24 scaled up

# The fiddly bits of this are getting the Unicode mappings correct.

# ROM dumps are mapped to Atari address space at $C000, the regular
# charset starts at $E000, so it's at offset $2000 (8192) in the
# image. The XL international set is at $CC00, or offset $0C00 (3072).
# In the Arabic ROM, the Arabic font takes the place of the standard
# charset. All the charsets are 1K in size.

# Arabic ROM, plus some info about it and a mention of the Hebrew ROM:
# https://www.savetz.com/vintagecomputers/arabic65xe/

# TODO;
# BDF fonts too: psf2bdf --iso10646 --fontname <blah> ...
# Arabic.

use bytes;

$fontname = "fauxtari";

sub byte2line {
	my $t = shift;
	$t = sprintf("%08b", $t);
	$t =~ y/0/-/;
	$t =~ y/1/#/;
	return $t;
}

sub scale_line {
	my $line = shift;
	my $scale = shift;
	my $one = '#' x $scale;
	my $zero = '-' x $scale;

	$line =~ s/#/$one/g;
	$line =~ s/-/$zero/g;
	return ($line x $scale);
}

sub chr2output {
	my $codepoint = shift;
	my $bytes = shift;
	my $scale = shift || 1;
	my $unicode = "";

	if(!ref $codepoint) {
		$codepoint = [ $codepoint ];
	}

	$unicode .= sprintf("[%04x];", $_) for @$codepoint;

	my $result = sprintf("%%\nUnicode: %s\nBitmap: \\\n", $unicode);
	for(0..7) {
		my $byte = $bytes->[$_];
		my $line = byte2line($byte);
		$result .= scale_line($line, $scale);
		$result .= " \\" unless $_ == 7;
		$result .= "\n";
	}
	return $result;
}

sub internal2byte {
	my $t = shift;
	$t =~ y/-/0/;
	$t =~ y/#/1/;
	return pack("B*", $t);
}

sub psf2txt_header {
	my $charcount = shift;
	my $scale = shift;
	my $w = 8 * $scale;
	my $h = 8 * $scale;
	return <<EOF;
%PSF2
Version: 0
Flags: 1
Length: $charcount
Width: $w
Height: $h
EOF
}

sub read_rom {
	my $data;
	my $filename = shift;
	my $offset = shift;
	my $charcount = shift;
	my $len = $charcount * 8;
	open my $fh, '<', $filename;
	read($fh, $data, 16384);
	close $fh;
	return substr($data, $offset, $len);
}

# position in the map (array index) is the raw glyph number, in
# the order we read the font data.
# value is the Unicode codepoint. if there are multiple codepoints
# for the glyph, make the value an array ref (see 0x1B below, the
# Atari Escape glyph is both 0x0118 aka LATIN CAPITAL LETTER E WITH OGONEK,
# and 0x241b aka SYMBOL FOR ESCAPE).
sub setup_map {
	our @map = (
			0x2665, # 0x00
			0x2523, # 0x01
			0x2503, # 0x02
			0x251b, # 0x03
			0x252b, # 0x04
			0x2513, # 0x05
			0x2571, # 0x06
			0x2572, # 0x07
			0x25E2, # 0x08
			0x2597, # 0x09
			0x25E3, # 0x0A
			0x259D, # 0x0B
			0x2598, # 0x0C
			0x2594, # 0x0D
			0x2581, # 0x0E
			0x2596, # 0x0F
			0x2663, # 0x10
			0x250F, # 0x11
			0x2501, # 0x12
			0x254B, # 0x13
			0x25CF, # 0x14
			0x2584, # 0x15
			0x258E, # 0x16
			0x2533, # 0x17
			0x253B, # 0x18
			0x258C, # 0x19
			0x2517, # 0x1A
			[ 0x0118, 0x241b ], # 0x1B
			0x2191, # 0x1C
			0x2193, # 0x1D
			0x2190, # 0x1E
			0x2192, # 0x1F
			[ 0x0020, 0xa0 ], # 0x20 (plus &nbsp;)
			0x0021, # 0x21
			0x0022, # 0x22
			0x0023, # 0x23
			0x0024, # 0x24
			0x0025, # 0x25
			0x0026, # 0x26
			0x0027, # 0x27
			0x0028, # 0x28
			0x0029, # 0x29
			0x002A, # 0x2A
			0x002B, # 0x2B
			0x002C, # 0x2C
			[ 0x002D, 0x00ad, 0x2013, 0x2014 ], # 0x2D (plus soft hyphen, en, em)
			0x002E, # 0x2E
			0x002F, # 0x2F
			0x0030, # 0x30
			0x0031, # 0x31
			0x0032, # 0x32
			0x0033, # 0x33
			0x0034, # 0x34
			0x0035, # 0x35
			0x0036, # 0x36
			0x0037, # 0x37
			0x0038, # 0x38
			0x0039, # 0x39
			0x003A, # 0x3A
			0x003B, # 0x3B
			0x003C, # 0x3C
			0x003D, # 0x3D
			0x003E, # 0x3E
			0x003F, # 0x3F
			0x0040, # 0x40
			0x0041, # 0x41
			0x0042, # 0x42
			0x0043, # 0x43
			0x0044, # 0x44
			0x0045, # 0x45
			0x0046, # 0x46
			0x0047, # 0x47
			0x0048, # 0x48
			0x0049, # 0x49
			0x004A, # 0x4A
			0x004B, # 0x4B
			0x004C, # 0x4C
			0x004D, # 0x4D
			0x004E, # 0x4E
			0x004F, # 0x4F
			0x0050, # 0x50
			0x0051, # 0x51
			0x0052, # 0x52
			0x0053, # 0x53
			0x0054, # 0x54
			0x0055, # 0x55
			0x0056, # 0x56
			0x0057, # 0x57
			0x0058, # 0x58
			0x0059, # 0x59
			0x005A, # 0x5A
			0x005B, # 0x5B
			0x005C, # 0x5C
			0x005D, # 0x5D
			0x005E, # 0x5E
			0x005F, # 0x5F
			0x25C6, # 0x60
			0x0061, # 0x61
			0x0062, # 0x62
			0x0063, # 0x63
			0x0064, # 0x64
			0x0065, # 0x65
			0x0066, # 0x66
			0x0067, # 0x67
			0x0068, # 0x68
			0x0069, # 0x69
			0x006A, # 0x6A
			0x006B, # 0x6B
			0x006C, # 0x6C
			0x006D, # 0x6D
			0x006E, # 0x6E
			0x006F, # 0x6F
			0x0070, # 0x70
			0x0071, # 0x71
			0x0072, # 0x72
			0x0073, # 0x73
			0x0074, # 0x74
			0x0075, # 0x75
			0x0076, # 0x76
			0x0077, # 0x77
			0x0078, # 0x78
			0x0079, # 0x79
			0x007A, # 0x7A
			0x2660, # 0x7B
			[ 0x007C, 0x0622, 0xFE81, 0xFE82 ], # 0x7C, pipe + Arabic ﺁ
			0x21B0, # 0x7D
			0x25C0, # 0x7E
			0x25B6, # 0x7F

# Next, 28 "international" characters from the XL ROM $CC00 area
# á
			0x00e1, # 0x80
# ù
			0x00f9, # 0x81
# Ñ
			0x00d1, # 0x82
# É
			0x00c9, # 0x83
# ç
			0x00e7, # 0x84
# ô (or is it?)
			0x00f4, # 0x85
# ò
			0x00f2, # 0x86
# ì
			0x00ec, # 0x87
# £
			0x00a3, # 0x88
# ï
			0x00ef, # 0x89
# ü
			0x00fc, # 0x8a
# ä
			0x00e4, # 0x8b
# Ö
			0x00d6, # 0x8c
# ú
			0x00fa, # 0x8d
# ó
			0x00f3, # 0x8e
# ö
			0x00f6, # 0x8f
# Ü
			0x00dc, # 0x90
# å
			0x00e5, # 0x91
# û (?)
			0x00fb, # 0x92
# î
			0x00ee, # 0x93
# é
			0x00e9, # 0x94
# è
			0x00e8, # 0x95
# ñ
			0x00f1, # 0x96
# ê
			0x00ea, # 0x97
# ȧ
			0x0227, # 0x98
# à
			0x00e0, # 0x99
# ¡
			0x00a1, # 0x9a
# Ä
			0x00c4, # 0x9b

# Hebrew ROM is probably a 3rd-party hack, AFAIK never released by
# Atari, but I'm told the glyphs look good (by someone who reads Hebrew),
# so include it here.
			0x05d0, # 0x9c
			0x05d1, # 0x9d
			0x05d2, # 0x9e
			0x05d3, # 0x9f
			0x05d4, # 0xa0
			0x05d5, # 0xa1
			0x05d6, # 0xa2
			0x05d7, # 0xa3
			0x05d8, # 0xa4
			0x05d9, # 0xa5
			0x05da, # 0xa6
			0x05db, # 0xa7
			0x05dc, # 0xa8
			0x05dd, # 0xa9
			0x05de, # 0xaa
			0x05df, # 0xab
			0x05e0, # 0xac
			0x05e1, # 0xad
			0x05e2, # 0xae
			0x05e3, # 0xaf
			0x05e4, # 0xb0
			0x05e5, # 0xb1
			0x05e6, # 0xb2
			0x05e7, # 0xb3
			0x05e8, # 0xb4
			0x05e9, # 0xb5
			0x05ea, # 0xb6

# Polish ROM. Probably a 3rd-party hack, but there are a lot of Polish
# Atari users out there.
			0x0179, # 0xb7
			0x0105, # 0xb8
			0x017a, # 0xb9
			0x0107, # 0xba
			0x015a, # 0xbb
			0x0119, # 0xbc
# German sharp S, not Polish at all:
			0x00df, # 0xbd
			0x0141, # 0xbe
			0x0142, # 0xbf
			0x0143, # 0xc0
			0x0144, # 0xc1
			0x00f3, # 0xc2
			0x00d3, # 0xc3
# 0xc3 weird-looking V?
# next glyph is E with ogonek, we already have it as ESC
			0x015b, # 0xc4
			0x0106, # 0xc5
			0x0104, # 0xc6
			0x017b, # 0xc7
			0x017c, # 0xc8

# The rest are from the Arabic XE ROM
			0x0660, # eastern arabic numeral 0
			0x0661, # eastern arabic numeral 1
			0x0662, # eastern arabic numeral 2
			0x0663, # eastern arabic numeral 3
			0x0664, # eastern arabic numeral 4
			0x0665, # eastern arabic numeral 5
			0x0666, # eastern arabic numeral 6
			0x0667, # eastern arabic numeral 7
			0x0668, # eastern arabic numeral 8
			0x0669, # eastern arabic numeral 9

# TODO: figure out the Arabic letter mappings. Might need someone who
# actually reads Arabic to make sense of them... what I've got is just
# a guess.
			[ 0x0624, 0xfeb5, 0xfeb6 ], # 0x3c, maybe?

			[ 0x0634, 0xFEB5, 0xFEB6, 0xFEB8 ], # 0x41
			[ 0x0648, 0xFEED, 0xFEEE ], # 0x42
			[ 0x0629, 0xFE93, 0xFE94 ], # 0x43
			[ 0x064A, 0xFEF1, 0xFEF2, 0xFEF4 ], # 0x44
			[ 0x062B, 0xFE99, 0xFE9A, 0xFE9C ], # 0x45
			[ 0x0628, 0xFE8F, 0xFE90, 0xFE92 ], # 0x46
			[ 0x0644, 0xFEDD, 0xFEDE, 0xFEE0 ], # 0x47
			[ 0x0621, 0xfe80 ], # 0x48
			[ 0x0647, 0xFEE9, 0xFEEA, 0xFEEC ], # 0x49
			[ 0x062A, 0xFE95, 0xFE96, 0xFE98 ], # 0x4a
			[ 0x0646, 0xFEE5, 0xFEE6, 0xFEE8 ], # 0x4b
			[ 0x0645, 0xFEE1, 0xFEE2, 0xFEE4 ], # 0x4c
			[ 0x062C, 0xFE9D, 0xFE9E, 0xFEA0 ], # 0x4d
			# XXX don't know what 0x4e is
			[ 0x062E, 0xFEA5, 0xFEA6, 0xFEA8 ], # 0x4f
			[ 0x062D, 0xFEA1, 0xFEA2, 0xFEA4 ], # 0x50
			[ 0x0636, 0xFEBD, 0xFEBE, 0xFEC0 ], # 0x51
			[ 0x0642, 0xFED5, 0xFED6, 0xFED8 ], # 0x52
			[ 0x0633, 0xFEB1, 0xFEB2, 0xFEB4 ], # 0x53
			[ 0x0641, 0xFED1, 0xFED2, 0xFED4 ], # 0x54
			[ 0x0639, 0xFEC9, 0xFECA, 0xFECC ], # 0x55
			[ 0x0643, 0xFED9, 0xFEDA, 0xFEDC ], # 0x56
			[ 0x0635, 0xFEB9, 0xFEBA, 0xFEBC ], # 0x57
			# XXX don't know what 0x58 is
			[ 0x063A, 0xFECD, 0xFECE, 0xFED0 ], # 0x59
			[ 0x0649, 0xFEEF, 0xFEF0 ], # 0x5a
			[ 0x0632, 0xFEAF, 0xFEB0 ], # 0x60
			0xFEB7, # 0x61
			# XXX 0x62 isn't really both ﺩ and ﺭ but where's the Atari's ﺭ at?
			[ 0x062F, 0xFEA9, 0xFEAA, 0x0631, 0xFEAD, 0xFEAE ], # 0x62
			[ 0x0630, 0xFEAB, 0xFEAC ], # 0x63
			0xFEF3, # 0x64
			0xFE9B, # 0x65
			0xFE91, # 0x66
			0xFEDF, # 0x67
			[ 0x0627, 0xFE8D, 0xFE8E ], # 0x68
			0xFEEB, # 0x69
			0xFE97, # 0x6a
			0xFEE7, # 0x6b
			0xFEE3, # 0x6c
			0xFE9F, # 0x6d
			[ 0xfef7, 0xf3f8 ], # 0x6e
			0xFEA7, # 0x6f
			0xFEA3, # 0x70
			0xFEBF, # 0x71
			0xFED7, # 0x72
			0xFEB3, # 0x73
			0xFED3, # 0x74
			0xFECB, # 0x75
			0xFEDB, # 0x76
			0xFEBB, # 0x77
			[ 0x0637, 0xFEC1, 0xFEC2, 0xFEC4, 0xFEC3 ], # 0x78
			0xFECF, # 0x79
			[ 0x0638, 0xFEC5, 0xFEC6, 0xFEC8, 0xFEC7 ], # 0x7a
			0x061F, # 0x7f
			);
}

### main()
setup_map();

@scale1 = ();
@scale2 = ();
@scale3 = ();

#$raw = read_rom("atarixl.rom", 0x2000, 128);
$raw = read_rom("atarixl.rom", 0x2200, 32);
$raw .= read_rom("atarixl.rom", 0x2000, 32);
$raw .= read_rom("atarixl.rom", 0x2100, 32);
$raw .= read_rom("atarixl.rom", 0x2300, 32);
$raw .= read_rom("atarixl.rom", 0xe00, 26);
$raw .= read_rom("atarixl.rom", 0xf00, 1);
$raw .= read_rom("atarixl.rom", 0xfd8, 1);
$raw .= read_rom("xl_hebrew.rom", 0xe00, 27);
$raw .= read_rom("xl_polish.rom", 0xe00, 6);
$raw .= read_rom("xl_polish.rom", 0xe50, 7);
$raw .= read_rom("xl_polish.rom", 0xe98, 1);
$raw .= read_rom("xl_polish.rom", 0xeb0, 3);
$raw .= read_rom("xl_polish.rom", 0xed0, 1);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*15, 10);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*0x3c, 1);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*0x41, 13);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*0x4f, 9);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*0x59, 2);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*0x60, 27);
$raw .= read_rom("xe_arabic.rom", 0x2000+8*0x7f, 1);
#warn length($raw) / 8;
##$raw = read_rom("arabicxe.rom", 0x2000, 128);
#$cnt = 0x9a;
$cnt = 0;
while($raw =~ /(.{8})/gc) {
	my @bytes;
	my $got = $1;
	my $chr = $map[$cnt++];
	die "$0: incomplete map" unless defined $chr;
	#warn $got;
	push @bytes, ord(substr($got, $_, 1)) for(0..7);
	#warn $_ for @bytes;
	push @scale1, chr2output($chr, \@bytes, 1);
	push @scale2, chr2output($chr, \@bytes, 2);
	push @scale3, chr2output($chr, \@bytes, 3);
}

while(<DATA>) {
	push @scale1, $_;
	$cnt++ if /^%/;
	if(/^\s/) {
		push @scale2, scale_line($_, 2);
		push @scale3, scale_line($_, 3);
	} else {
		push @scale2, $_;
		push @scale3, $_;
	}
}

#warn "$cnt characters\n";
#warn "(padding to 256 characters)\n" unless $cnt >= 256;
#warn "(padding to 512 characters)\n" unless $cnt >= 512;

if($cnt > 512) {
	die "$0: too many characters ($cnt > 512)";
}

while($cnt < 512) {
	my $fake = "%\nUnicode: [0000];\nBitmap: ";
	push @scale1, $fake . (("-" x 8) x 8) . "\n";
	push @scale2, $fake . (("-" x 16) x 16) . "\n";
	push @scale3, $fake . (("-" x 24) x 24) . "\n";
	$cnt++;
}

#warn "$cnt characters with padding\n";
sub mkfonts {
	my $px = shift;
	my $scaled_data = shift;
	my $scale = shift;

	open $fh, '>', "$fontname-$px.txt" or die $!;
	print $fh psf2txt_header($cnt, $scale);
	print $fh $_ for(@$scaled_data);
	close $fh;
	system("txt2psf $fontname-$px.txt $fontname-$px.psf");
	system("psf2bdf --iso10646 --fontname=$fontname-$px $fontname-$px.psf | perl ./fixbdf.pl $px > $fontname-$px.bdf");
}

mkfonts(8, \@scale1, 1);
mkfonts(16, \@scale2, 2);
mkfonts(24, \@scale3, 3);

exit 0;

__DATA__
%
// backtick
Unicode: [0060];
Bitmap: \
        -------- \
        --##---- \
        --##---- \
        ---##--- \
        -------- \
        -------- \
        -------- \
        --------
%
// curlies
Unicode: [007b];
Bitmap: \
        ----##-- \
        ---##--- \
        ---##--- \
        --##---- \
        ---##--- \
        ---##--- \
        ----##-- \
        --------
%
Unicode: [007d];
Bitmap: \
        --##---- \
        ---##--- \
        ---##--- \
        ----##-- \
        ---##--- \
        ---##--- \
        --##---- \
        --------
%
// tilde
Unicode: [007e];
Bitmap: \
        -------- \
        -###--## \
        ##-##-## \
        ##--###- \
        -------- \
        -------- \
        -------- \
        --------
%
// euro
Unicode: [20ac];
Bitmap: \
        ---####- \
        --##---- \
        -#####-- \
        --##---- \
        -#####-- \
        --##---- \
        ---####- \
        --------
%
// spanish left-quote
Unicode: [00ab];
Bitmap: \
        -------- \
        -------- \
        -------- \
        -##--##- \
        ##--##-- \
        -##--##- \
        -------- \
        --------
%
// spanish right-quote
Unicode: [00bb];
Bitmap: \
        -------- \
        -------- \
        -------- \
        ##--##-- \
        -##--##- \
        ##--##-- \
        -------- \
        --------
%
// spanish inverted question mark
Unicode: [00bf];
Bitmap: \
        -------- \
        ---##--- \
        -------- \
        ---##--- \
        ----##-- \
        -##--##- \
        --####-- \
        --------
%
// copyright
Unicode: [00a9];
Bitmap: \
        -#####-- \
        #-----#- \
        #--##-#- \
        #-#---#- \
        #--##-#- \
        #-----#- \
        -#####-- \
        --------
%
// degrees
Unicode: [00b0];[00ba];
Bitmap: \
        -------- \
        ---##--- \
        --#--#-- \
        ---##--- \
        -------- \
        -------- \
        -------- \
        --------