#!/usr/bin/perl -w 

$GBCODE_BYTE_OFFSET = 0xa1;
$GBCODE_UNIT_LEN = 94;
$GBCODE_CHAR_SIZE = 32; #16x16 bitmap -> 32 bytes

#read from: HZK16, input txt file, 
#generate: output.msg, output.idx, output.pnm file
sub process_kanji {
    my ($in_db, $in_txt, $out, $idx_h) = @_;
    my $cnt = 0;
    my @map;
    my $type = 0;

    open(IN_DB, "<$in_db") or die "Can't open $in_db: $!";
    open(IN_TXT, "<$in_txt") or die "Can't open $in_txt: $!";
    binmode(IN_TXT);
    open(OUT_MSG, ">$out.msg") or die "Can't open $out.msg: $!";
    open(OUT_PNM_TMP, ">$out.pnm.tmp") or die "Can't open $out.pnm.tmp: $!";

LINE: while (<IN_TXT>) {
        $type = 0;
        SWITCH: {
            if (/^m(\d+)/) {
                printf OUT_MSG "static unsigned char str_%02d [] = {", $1;
                $type = 1;
                last SWITCH;
            }
            if (/^c(\d+)/) {
                $type = 2;
                printf OUT_MSG "static unsigned char course%s [] = {", $1;
                last SWITCH;
            }
            if (/^s(\d+)/) {
                $type = 3;
                printf OUT_MSG "static unsigned char star_name_%s [] = {", $1;
                last SWITCH;
            }
            if (/^w(_\w+)/) {
                $type = 4;
                printf OUT_MSG "static unsigned char %s [] = {", $1;
                last SWITCH;
            }
            if (/^$/) { next LINE; }
            if (/^ +$/) { next LINE; }
        }

        while ($type!=0) {
            $c = getc(IN_TXT);
            if ($c eq '*') { 
                <IN_TXT>;
                printf OUT_MSG "0xff, 0xff, };\n\n";
                next LINE; 
            }

            SWITCH: {
                if ($c =~ /^\d/) {
                    printf OUT_MSG "0x00, 0x%x, ", $c;
                    last SWITCH;
                }
                if ($c eq "A") {
                    printf OUT_MSG "0x00, 0x54, ";
                    last SWITCH;
                }
                if ($c eq "{") {
                    printf OUT_MSG "0x00, 0x54, ";
                    last SWITCH;
                }
                if ($c eq "B") {
                    printf OUT_MSG "0x00, 0x55, ";
                    last SWITCH;
                }
                if ($c eq "}") {
                    printf OUT_MSG "0x00, 0x55, ";
                    last SWITCH;
                }
                if ($c eq "C") {
                    printf OUT_MSG "0x00, 0x56, ";
                    last SWITCH;
                }
                if ($c eq "Z") {
                    printf OUT_MSG "0x00, 0x57, ";
                    last SWITCH;
                }
                if ($c eq "<") {
                    printf OUT_MSG "0x00, 0x57, ";
                    last SWITCH;
                }
                if ($c eq "R") {
                    printf OUT_MSG "0x00, 0x58, ";
                    last SWITCH;
                }
                if ($c eq ">") {
                    printf OUT_MSG "0x00, 0x58, ";
                    last SWITCH;
                }
                if ($c eq "#") {
                    printf OUT_MSG "0x00, 0xf5, ";
                    last SWITCH;
                }
                if ($c eq "\"") {
                    printf OUT_MSG "0x00, 0xf6, ";
                    last SWITCH;
                    last SWITCH;
                }
                if ($c eq "@") {
                    printf OUT_MSG "0xff, 0xe0, ";
                    last SWITCH;
                }
                if ($c eq " ") {
                    printf OUT_MSG "0xff, 0x9e, ";
                    last SWITCH;
                }
                if ($c eq "^") {
                    $c = getc(IN_TXT);
                    if ($c eq "[") {
                        printf OUT_MSG "0x00, 0x56, 0x00, 0x52, ";
                    }
                    if ($c eq "]") {
                        printf OUT_MSG "0x00, 0x56, 0x00, 0x53, ";
                    }
                    if ($c eq "/") {
                        printf OUT_MSG "0x00, 0x56, 0x00, 0x51, ";
                    }
                    if ($c eq "\\") {
                        printf OUT_MSG "0x00, 0x56, 0x00, 0x50, ";
                    }
                    if ($c eq ",") {
                        printf OUT_MSG "0x00, 0x56, 0x00, 0x6f, ";
                    }
                    if ($c eq "\"") {
                        printf OUT_MSG "0x00, 0x56, 0x00, 0xf6, ";
                    }
                    last SWITCH;
                }

                if (vec($c, 0, 8)==0x0a) {
                    printf OUT_MSG "0xff, 0xfe, ";
                } else {

                $c2 = getc(IN_TXT);
                if (vec($c, 0, 8)==0x0d && vec($c2, 0, 8)==0x0a) {
                    printf OUT_MSG "0xff, 0xfe, ";
                } else {
                    $idx = (vec($c,0,8)-$GBCODE_BYTE_OFFSET)*$GBCODE_UNIT_LEN + (vec($c2,0,8)-$GBCODE_BYTE_OFFSET);
                    if (!$map[$idx]) {
                        $cnt ++;
                        $map[$idx] = $cnt;

                        $skip = $idx * $GBCODE_CHAR_SIZE;
                        seek(IN_DB, $skip, 0);
                        for ($i=0; $i<$GBCODE_CHAR_SIZE; $i++) {
                            $byte = getc(IN_DB);
                            for ($j=0; $j<8; $j++) {
                                printf OUT_PNM_TMP "%c", (vec($byte, 0, 8) & (0x80>>$j))==0 ? 0:0xff;
                            }
                        }
                    }

                    $b1 = (($map[$idx] + 255) >> 8) & 0xff;
                    $b2 = ($map[$idx] + 255) & 0xff;
                    printf OUT_MSG "0x%x, 0x%x, ", $b1, $b2;
                }
                }
                last SWITCH;
            }
        }
    }

    close IN_DB;
    close IN_TXT;
    close OUT_MSG;
    close OUT_PNM_TMP;

    `cp $idx_h $out.idx`;
    `chmod u+w $out.idx`;
    open(OUT_IDX, ">>$out.idx") or die "Can't open $out.idx: $!";
    printf OUT_IDX "\n";
    for ($i=0; $i<$cnt; $i++) {
        printf OUT_IDX "c_%d_l_txt, c_%d_r_txt, ", $i, $i;
    }
    printf OUT_IDX "\n};\n";
    close OUT_IDX;
    
    open(OUT_PNM, ">$out.pnm") or die "Can't open $out.pnm: $!";
    printf OUT_PNM "P5\n16 %d\n255\n", 16*$cnt;
    close OUT_PNM;
    `cat $out.pnm.tmp >> $out.pnm`;
    `rm $out.pnm.tmp`;

     return $cnt;
}

#split each chinese character into half, rotate each of the half
#clockwise 90 degrees, and flip 
sub make_pnm_map {
   my ($name, $height) = @_;
   my $cnt = 0;

   `pnmcut 15 $height 0 0 $name.pnm > $name.mod.pnm`;

    while ($cnt < $height) {
        `pnmcut 0 $cnt 16 16 $name.pnm > 1.pnm`;
        `pnmcut 0 0 8 16 1.pnm > 11.pnm`;
        `pnmcut 8 0 8 16 1.pnm > 12.pnm`;
        `pnmflip -lr -r270 11.pnm > a.pnm`;
        `pnmflip -lr -r270 12.pnm > b.pnm`;
        `pnmcat -tb $name.mod.pnm a.pnm b.pnm > foo.pnm`; 
        `mv foo.pnm $name.mod.pnm`; 

         $cnt = $cnt + 16;
         print "finished character ", $cnt/16, "\n";
    }

    $height = $height + 1;
    `pnmcut 0 1 16 $height $name.mod.pnm > foo.pnm`;
    `mv foo.pnm $name.mod.pnm`;
    `rm 1.pnm 11.pnm 12.pnm a.pnm b.pnm`;
}

#generate data file of the 8x8 font glyph in SuperMario
sub gen_font_glyph {
    my $name = shift;
    my $cnt = 0, $idx=0;

    open(INFILE, "<$name.mod.pnm") or die "Can't open $name.mod.pnm: $!";
    open(OUTFILE, ">$name.glyph") or die "Can't open $name.glyph: $!";

    for ($i=0; $i<3; $i++) {<INFILE>};
    while(read(INFILE, $bytes, 8) != 0) {
#        $b1 = 7.0 / 255.0 * vec($bytes, 0, 8);
#        if ($b1!=0) { $b1 = $b1<<1 | 0x01; } 
#        $b2 = 7.0 / 255.0 * vec($bytes, 1, 8);
#        if ($b2!=0) { $b2 = $b2<<1 | 0x01; }
#        $b = $b1<<4 | $b2;
        $b = 0;
        for ($j=0; $j<8; $j++) {
            $b1 = vec($bytes, $j, 8);
            if ($b1>127.0) {$b1 = 1;} else {$b1=0};
            $b = $b1<<(7-$j) | $b;
        } 

        if ($idx%16==0) {
            printf OUTFILE "static unsigned char c_%d_%s_txt[] = {\n", $cnt/2, ($cnt%2==0)?"l":"r";
            $cnt ++;
        }
        printf OUTFILE "0x%02x, ", $b;
        $idx++;
        if ($idx%2==0) {
            printf OUTFILE "\n";
        }
        if ($idx%16==0) {
            printf OUTFILE "};\n";
        }
    }
    close INFILE;
    close OUTFILE;

    print $cnt/2, " characters with ", 16*$cnt/1024.0, "K bytes\n";
}

die "Usage: zh_kanji.pl <Chinese GB font map> <input Chinese text file> <output file> <idx header file>\n       See README for more detail.\n" unless $#ARGV==3;

print "processing input Chinese text file\n";
$out = process_kanji($ARGV[0], $ARGV[1], $ARGV[2], $ARGV[3]);
print "number of unique Chinese characters used: ", $out, "\n";

print "making pnm map with height: ", $out*16, "\n";
$h = $out*16 - 1;
make_pnm_map($ARGV[2], $h);

print "writing data file for chinese font glyph\n";
gen_font_glyph($ARGV[2]);