source: xtideuniversalbios/trunk/XTIDE_Universal_BIOS/Src/StringsCompress.pl @ 196

#======================================================================================
#
# Project name  :   XTIDE Universal BIOS
#
# Authors       :   Greg Lindhorst
#                   gregli@hotmail.com
#
# Description   :   Script for compiling and compressing strings for 
#                   use by DisplayFormatCompressed.asm.  See the header of that file
#                   for a description of the compression scheme.
#
# Usage         :   stdin:  Listing of strings.asm, 
#                           assembled with MODULE_STRINGS_COMPRESSED_PRECOMPRESS.
#                           We used the listing so that the assembler can take care of
#                           resolving %define and EQU symbol definitions.
#
#                   stdout: StringsCompressed.asm,
#                           plug replacement for Strings.asm (included by Main.asm)
#
#                   Also see the XTIDE makefile for building StringsCompressed.asm
#

#----------------------------------------------------------------------
#
# Translated and Format characters
#
# DisplayFormatCompressed can only deal with characters in one of the following categories:
#  1. Those in the Translate associative array
#  2. Those in the Format associative array
#  3. Characters between $normal_base and $normal_base+0x40
#  4. Null characters (marking the end of strings)
#  5. The special string LF,CR
#
# If a character or format read at the input cannot be found in one of the above categories, 
# it must be added here before this script will accept it (and DisplayFormatCompressed can 
# display it).
#
# Note that these tables are not present in DisplayFormatCompressed, and do not need to
# updated there.  Needed information is put in the compression output that it reads.
#
$translate{ord(' ')} = 0;
$translate{172}      = 1;     # ONE_QUARTER
$translate{171}      = 2;     # ONE_HALF
$translate{179}      = 3;     # SINGLE_VERTICAL
$translate{175}      = 4;     # ANGLE_QUOTE_RIGHT
$translate{ord('!')} = 5;
$translate{ord('"')} = 6;
$translate{ord(',')} = 7;
$translate{ord('-')} = 8;
$translate{ord('.')} = 9;
$translate{ord('/')} = 10;
$translate{ord('1')} = 11;    
$translate{ord('2')} = 12;
$translate{ord('3')} = 13;
$translate{ord('4')} = 14;
$translate{ord('5')} = 15;
$translate{ord('6')} = 16;
$translate{ord('8')} = 17;
$translate{200}      = 18;    # DOUBLE_BOTTOM_LEFT_CORNER
$translate{181}      = 19;    # DOUBLE_LEFT_HORIZONTAL_TO_SINGLE_VERTICAL

#
# Formats begin immediately after the last Translated character (they are in the same table)
#
$format_begin = 20;

$format{"s"}   = 20;        # n/a
$format{"c"}   = 21;        # n/a
$format{"2-I"} = 22;        # must be even
$format{"u"}   = 23;        # must be odd
$format{"5-u"} = 24;        # must be even
$format{"x"}   = 25;        # must be odd
$format{"5-x"} = 26;        # must be even
$format{"nl"}  = 27;        # n/a
$format{"2-u"} = 28;        # must be even
$format{"A"}   = 29;        # n/a

# NOTE: The last $format cannot exceed 31 (stored in a 5-bit quantity).

#
# Starting point for the "normal" range, typically around 0x40 to cover upper and lower case
# letters.  If lower case 'z' is not used, 0x3a can be a good choice as it adds ':' to the 
# front end.
#
$normal_base = 0x3a;

#
# High order code bits, determining which type of character we have (translated or not) and
# if a space or null should come after this character.
#
$code_space = 0xc0;
$code_null = 0x80;
$code_normal = 0x40;
$code_translate = 0x00;

#
# Bit used if it is a translated byte
#
$code_translate_null = 0x00;
$code_translate_normal = 0x20;

print ";;;======================================================================\n";
print ";;;\n";
print ";;; This file is generated by StringsCompress.pl from source in Strings.asm\n";
print ";;; DO NOT EDIT DIRECTLY - See the maekfile for how to rebuild this file.\n";
print ";;; This file only needs to be rebuilt if Strings.asm is changed.\n";
print ";;;\n";
print ";;;======================================================================\n\n";

#
# Loop through lines of the listing, looking for 'db' lines (and dealing with continuations)
# and compressing each line as it is encountered.
#
while(<>)
{
    #
    # The <number> indicates a line from an include file, do not include in the output
    #
    if( /^\s*\d+\s*\<\d\>/ )
    {
    }

    #
    # a 'db' line, with or without a label
    #
    elsif( /^\s*\d+\s[0-9A-F]+\s([0-9A-F]+)(-?)\s+([a-z0-9_]+:)?(\s+)(db\s+)(.*)/i )
    {
        $bytes = $1;
        $continuation = $2;
        $label = $3;
        $spacing = $4;
        $db = $5;
        $string = $6;

        print $label.$spacing."; ".$db.$string."\n";

        if( $continuation eq "-" )
        {
            do
            {
                $_ = <>;
                /^\s*\d+\s[0-9A-F]+\s([0-9A-F]+)(\-?)/i || die "parse error on continuation";
                $bytes .= $1;
                $continuation = $2;
            }
            while( $continuation eq "-" );
        }

        &processString( $bytes, $label.$spacing, $db );
    }
    
    # 
    # everything else, copy to the output as is
    #
    elsif( /^\s*\d+\s*(.*)$/ )
    {
        print $1."\n";
    }
}

print ";;; end of strings.asm\n\n";

#--------------------------------------------------------------------------------
#
# Output constants and the TranslatesAndFormats table
#

print "StringsCompressed_NormalBase     equ   ".$normal_base."\n\n";

print "StringsCompressed_FormatsBegin   equ   ".$format_begin."\n\n";

print "StringsCompressed_TranslatesAndFormats: \n";

foreach $f (keys(%translate))
{
    $translate_index[$translate{$f}] = $f;
    $used{$f} || die "translate $f unused\n";
    $translate{$f} <= 31 || die $translate{$f}.": translate codes must be below 32";
}

for( $g = 0; $translate_index[$g]; $g++ )
{
    print "        db     ".$translate_index[$g]."  ; ".$g."\n";
}

foreach $f (keys(%format))
{
    $n = $f;
    $n =~ s/\-/_/g;
    $format_index[$format{$f}] = "DisplayFormatCompressed_Format_".$n;
    $used{$f} || die "format $f unused\n";
    $format{$f} <= 31 || die $format{$f}.": format codes must be below 32";
}

for( $t = $format_begin; $format_index[$t]; $t++ )
{
    print "        db     (DisplayFormatCompressed_BaseFormatOffset - ".$format_index[$t].")    ; ".$t."\n";
}

print "\n";

# 
# Ensure that branch targets are within reach
#
print "%ifndef CHECK_FOR_UNUSED_ENTRYPOINTS\n";
for( $t = $format_begin; $format_index[$t]; $t++ )
{
    print "%if DisplayFormatCompressed_BaseFormatOffset < $format_index[$t] || DisplayFormatCompressed_BaseFormatOffset - $format_index[$t] > 255\n";
    print "%error \"".$format_index[$t]." is out of range of DisplayFormatCompressed_BaseFormatOffset\"\n";
    print "%endif\n";
}
print "%endif\n";

#--------------------------------------------------------------------------------
#
# Output usage statistics
#

print "\n;; translated usage stats\n";
foreach $f (keys(%special))
{
    print ";; ".$f.":".$used{$f}."\n";
    $translate_count++;
}
print ";; total translated: ".$translate_count."\n";

print "\n;; format usage stats\n";
$format_count = 0;
foreach $f (keys(%format))
{
    print ";; ".$f.":".$used{$f}."\n";
    $format_count++;
}
print ";; total format: ".$format_count."\n";

print "\n;; alphabet usage stats\n";

$used_count = 0;
for( $t = $normal_base; $t < $normal_base + 0x40; $t++ )
{
    print ";; ".$t.",".chr($t).":".$used{$t}."\n";
    if( $used{$t} )
    {
        $used_count++;
    }
}
print ";; alphabet used count: ".$used_count."\n";

#--------------------------------------------------------------------------------
#
# processString does the real compression work...
#

sub processString
{
    $chars = $_[0];
    $label = $_[1];
    $db = $_[2];

    $label =~ s/[a-z0-9_:]/ /ig;      # replace with spaces for proper output spacing

    #
    # Copy numeric bytes out of hexadecimal pairs in the listing
    #
    $#v = 0;

    $orig = "";
    for( $g = 0; $g < length($chars); $g += 2 )
    {
        $i = $g/2;
        $v[$i] = hex(substr($chars,$g,2));
        $orig .= sprintf( ($v[$i] > 0x9f ? ", %03xh" : ",  %02xh"), $v[$i] );
    }
    $v[length($chars)/2] = 0xff;      # guard byte to avoid thinking going past the end of 
                                      # the string is a null

    $output = "";
    #
    # Loop through bytes...
    # looking ahead as needed for possible space and null optimizations, compiling formats
    #
    for( $g = 0; $g < $#v-1; $g++ )    # -1 for the guard byte
    {
        #
        # Special translation of LF,CR to a format
        #
        if( $v[$g] == 10 && $v[$g+1] == 13 )
        {
            $g++;
            $post = $code_translate;
            $code = $format{"nl"};
            $used{"nl"}++;
        }

        #
        # Format operators
        #
        elsif( $v[$g] == 0x25 )    # "%"
        {
            $fo = "";
            $g++;
            if( $v[$g] >= ord("0") && $v[$g] <= ord("9") )
            {
                $fo = $fo.chr($v[$g]);
                $g++;
            }
            if( $v[$g] == ord("-") )
            {
                $fo = $fo.chr($v[$g]);
                $g++;
            }
            $fo = $fo.chr($v[$g]);

            $format{$fo} || die "unknown format operator: '".$fo."'\n";

            $code = $format{$fo};
            $post = $code_translate;
            $used{$fo}++;
        }

        #
        # Translated characters
        #
        elsif( $v[$g] == 32 || $translate{$v[$g]} )
        {
            $post = $code_translate;
            $code = $translate{$v[$g]};
            $used{$v[$g]}++;
        }
        
        #
        # "normal" characters (alphabet, and ASCII characters around the alphabet)
        #
        elsif( $v[$g] >= $normal_base && $v[$g] < ($normal_base+0x40) )
        {
            $used{$v[$g]}++;

            $post = $code_normal;
            $code = $v[$g] - $normal_base;
        }

        #
        # Not found
        # 
        else
        {
            die $v[$g].": no translation or format, and out of normal range - may need to be added\n";
        }

        if( $post == $code_translate )
        {
            #
            # NULL optimization (space optimization not possible on translate/format)
            #
            if( $v[$g+1] == 0 )
            {
                $g++;
                $post = $post | $code_translate_null;
            }
            else
            {
                $post = $post | $code_translate_normal;
            }
        }
        else # $post == $code_normal
        {
            #
            # Space optimization
            #
            if( $v[$g+1] == 0x20 && $v[$g+2] != 0 )
            {
                # can't take this optimization if the next byte is a null, 
                # since we can't have both a postfix space and null
                $g++;
                $post = $code_space;
            }

            #
            # NULL optimization
            #
            elsif( $v[$g+1] == 0 )
            {
                $g++;
                $post = $code_null;
            }
        }

        $code = $code | $post;
        $output .= sprintf( ($code > 0x9f ? ", %03xh" : ",  %02xh"), $code );
    }

    print $label."; ".$db.substr($orig,2)."    ; uncompressed\n";
    print $label."  ".$db.substr($output,2);
    for( $t = length($output); $t < length($orig); $t++ )
    {
        print " ";
    }
    print "    ; compressed\n\n";
}