dfhack/plugins/ruby/codegen.pl

#!/usr/bin/perl

use strict;
use warnings;

use XML::LibXML;

our @lines_rb;
my @lines_cpp;
my @include_cpp;
my %offsets;

sub indent_rb(&) {
    my ($sub) = @_;
    my @lines;
    {
        local @lines_rb;
        $sub->();
        @lines = map { "    " . $_ } @lines_rb;
    }
    push @lines_rb, @lines
}

sub rb_ucase {
    my ($name) = @_;
    return $name if ($name eq uc($name));
    return join("", map { ucfirst $_ } (split('_', $name)));
}

my %global_type_renderer = (
    'enum-type' => \&render_global_enum,
    'struct-type' => \&render_global_class,
    'class-type' => \&render_global_class,
    'bitfield-type' => \&render_global_bitfield,
);

my %item_renderer = (
    'global' => \&render_item_global,
    'number' => \&render_item_number,
    'container' => \&render_item_container,
    'compound' => \&render_item_compound,
    'pointer' => \&render_item_pointer,
    'static-array' => \&render_item_staticarray,
    'primitive' => \&render_item_primitive,
    'bytes' => \&render_item_bytes,
);


sub render_global_enum {
    my ($name, $type) = @_;

    my $rbname = rb_ucase($name);
    push @lines_rb, "class $rbname";
    indent_rb {
        render_enum_fields($type);
    };
    push @lines_rb, "end";
}
sub render_enum_fields {
    my ($type) = @_;

    my $value = -1;
    for my $item ($type->findnodes('child::enum-item')) {
        $value = $item->getAttribute('value') || ($value+1);
        my $elemname = $item->getAttribute('name'); # || "unk_$value";

        if ($elemname) {
            my $rbelemname = rb_ucase($elemname);
            push @lines_rb, "$rbelemname = $value";
        }
    }
}


sub render_global_bitfield {
    my ($name, $type) = @_;

    my $rbname = rb_ucase($name);
    push @lines_rb, "class $rbname < MemStruct";
    indent_rb {
        render_bitfield_fields($type);
    };
    push @lines_rb, "end";
}
sub render_bitfield_fields {
    my ($type) = @_;

    my $shift = 0;
    for my $field ($type->findnodes('child::ld:field')) {
        my $count = $field->getAttribute('count') || 1;
        my $name = $field->getAttribute('name');
        $name = $field->getAttribute('ld:anon-name') if (!$name);
        print "bitfield $name !number\n" if (!($field->getAttribute('ld:meta') eq 'number'));
        if ($count == 1) {
            push @lines_rb, "bit :$name, $shift" if ($name);
        } else {
            push @lines_rb, "bits :$name, $shift, $count" if ($name);
        }
        $shift += $count;
    }
}


my $cpp_var_counter = 0;
sub render_global_class {
    my ($name, $type) = @_;

    my $cppvar = "v_$cpp_var_counter";
    $cpp_var_counter++;
    push @lines_cpp, "}" if @include_cpp;
    push @lines_cpp, "void cpp_$name(FILE *fout) {";
    push @lines_cpp, "    df::$name *$cppvar = (df::$name*)moo;";
    push @include_cpp, $name;

    my $rbname = rb_ucase($name);
    my $parent = rb_ucase($type->getAttribute('inherits-from') || 'MemStruct');
    push @lines_rb, "class $rbname < $parent";
    indent_rb {
        render_struct_fields($type, "(*$cppvar)");
    };
    push @lines_rb, "end";
}
sub render_struct_fields {
    my ($type, $cppvar) = @_;

    for my $field ($type->findnodes('child::ld:field')) {
        my $name = $field->getAttribute('name');
        $name = $field->getAttribute('ld:anon-name') if (!$name);
        next if (!$name);
        my $offset = get_offset($cppvar, $name);

        push @lines_rb, "field(:$name, $offset) {";
        indent_rb {
            render_item($field, "$cppvar.$name");
        };
        push @lines_rb, "}";
    }
}

sub render_item {
    my ($item, $cppvar) = @_;
    return if (!$item);

    my $meta = $item->getAttribute('ld:meta');

    my $renderer = $item_renderer{$meta};
    if ($renderer) {
        $renderer->($item, $cppvar);
    } else {
        print "no render item $meta\n";
    }
}

sub render_item_global {
    my ($item, $cppvar) = @_;

    my $typename = $item->getAttribute('type-name');
    my $rbname = rb_ucase($typename);

    push @lines_rb, "global :$rbname";
}

sub render_item_number {
    my ($item, $cppvar) = @_;

    my $subtype = $item->getAttribute('ld:subtype');
    $subtype = $item->getAttribute('base-type') if ($subtype eq 'enum');
    $subtype = 'int32_t' if (!$subtype);

         if ($subtype eq 'int64_t') {
        push @lines_rb, 'number 64, true';
    } elsif ($subtype eq 'uint32_t') {
        push @lines_rb, 'number 32, false';
    } elsif ($subtype eq 'int32_t') {
        push @lines_rb, 'number 32, true';
    } elsif ($subtype eq 'uint16_t') {
        push @lines_rb, 'number 16, false';
    } elsif ($subtype eq 'int16_t') {
        push @lines_rb, 'number 16, true';
    } elsif ($subtype eq 'uint8_t') {
        push @lines_rb, 'number 8, false';
    } elsif ($subtype eq 'int8_t') {
        push @lines_rb, 'number 8, false';
    } elsif ($subtype eq 'bool') {
        push @lines_rb, 'number 8, true';
    } elsif ($subtype eq 's-float') {
        push @lines_rb, 'float';
    } else {
        print "no render number $subtype\n";
    }
}

sub render_item_compound {
    my ($item, $cppvar) = @_;

    my $subtype = $item->getAttribute('ld:subtype');
    if (!$subtype || $subtype eq 'bitfield') {
        push @lines_rb, "compound {";
        indent_rb {
            if (!$subtype) {
                render_struct_fields($item, $cppvar);
            } else {
                render_bitfield_fields($item);
            }
        };
        push @lines_rb, "}"
    } elsif ($subtype eq 'enum') {
        # declare constants
        render_enum_fields($item);
        # actual field
        render_item_number($item, $cppvar);
    } else {
        print "no render compound $subtype\n";
    }
}

sub render_item_container {
    my ($item, $cppvar) = @_;

    my $subtype = $item->getAttribute('ld:subtype');
    my $rbmethod = join('_', split('-', $subtype));
    my $tg = $item->findnodes('child::ld:item')->[0];
    # df-linked-list has no list[0]
    if ($tg and $rbmethod ne 'df_linked_list') {
        my $tglen = get_tglen($tg, $cppvar);
        push @lines_rb, "$rbmethod($tglen) {";
        indent_rb {
            render_item($tg, "${cppvar}[0]");
        };
        push @lines_rb, "}";
    } else {
        push @lines_rb, "$rbmethod";
    }
}

sub render_item_pointer {
    my ($item, $cppvar) = @_;

    my $tg = $item->findnodes('child::ld:item')->[0];
    my $tglen = get_tglen($tg, $cppvar);
    push @lines_rb, "pointer($tglen) {";
    indent_rb {
        render_item($tg, "${cppvar}[0]");
    };
    push @lines_rb, "}";
}

sub render_item_staticarray {
    my ($item, $cppvar) = @_;

    my $count = $item->getAttribute('count');
    my $tg = $item->findnodes('child::ld:item')->[0];
    my $tglen = get_tglen($tg, $cppvar);
    push @lines_rb, "static_array($count, $tglen) {";
    indent_rb {
        render_item($tg, "${cppvar}[0]");
    };
    push @lines_rb, "}";
}

sub render_item_primitive {
    my ($item, $cppvar) = @_;

    my $subtype = $item->getAttribute('ld:subtype');
    if ($subtype eq 'stl-string') {
        push @lines_rb, "stl_string";
    } else {
        print "no render primitive $subtype\n";
    }
}

sub render_item_bytes {
    my ($item, $cppvar) = @_;

    my $subtype = $item->getAttribute('ld:subtype');
    if ($subtype eq 'padding') {
    } elsif ($subtype eq 'static-string') {
        my $size = $item->getAttribute('size');
        push @lines_rb, "static_string($size)";
    } else {
        print "no render bytes $subtype\n";
    }
}


sub get_offset {
    my ($cppvar, $fname) = @_;

    # GCC fails with this
    #return query_cpp("offsetof(typeof($cppvar), $fname)");
    return query_cpp("((char*)&$cppvar.$fname - (char*)&$cppvar)");
}

sub get_tglen {
    my ($tg, $cppvar) = @_;

    if (!$tg) {
        return 'nil';
    }

    my $meta = $tg->getAttribute('ld:meta');
    if ($meta eq 'number') {
        return $tg->getAttribute('ld:bits')/8;
    } elsif ($meta eq 'pointer') {
        return 4;
    } else {
        return query_cpp("sizeof(${cppvar}[0])");
    }
}

sub query_cpp {
    my ($query) = @_;

    my $ans = $offsets{$query};
    return $ans if (defined($ans));

    push @lines_cpp, "    fprintf(fout, \"%s = %d\\n\", \"$query\", $query);";
    return "'$query'";
}



my $input = $ARGV[0] || '../../library/include/df/codegen.out.xml';

# run once with output = 'ruby-autogen.cpp'
# compile
# execute, save output to 'ruby-autogen.offsets'
# re-run this script with output = 'ruby-autogen.rb' and offsetfile = 'ruby-autogen.offsets'
# delete binary
# delete offsets
my $output = $ARGV[1] or die "need output file";
my $offsetfile = $ARGV[2];

if ($offsetfile) {
    open OF, "<$offsetfile";
    while (my $line = <OF>) {
        chomp($line);
        my ($key, $val) = split(' = ', $line);
        $offsets{$key} = $val;
    }
    close OF;
}


my $doc = XML::LibXML->new()->parse_file($input);
my %global_types;
$global_types{$_->getAttribute('type-name')} = $_ foreach $doc->findnodes('/ld:data-definition/ld:global-type');

for my $name (sort { $a cmp $b } keys %global_types) {
    my $type = $global_types{$name};
    my $meta = $type->getAttribute('ld:meta');
    my $renderer = $global_type_renderer{$meta};
    if ($renderer) {
        $renderer->($name, $type);
    } else {
        print "no render global type $meta\n";
    }
}

for my $obj ($doc->findnodes('/ld:data-definition/ld:global-object')) {
    my $name = $obj->getAttribute('name');
    # TODO
}

open FH, ">$output";
if ($output =~ /\.cpp$/) {
    print FH "#include \"Core.h\"\n";
    print FH "#include \"Console.h\"\n";
    print FH "#include \"Export.h\"\n";
    print FH "#include \"PluginManager.h\"\n";
    print FH "#include \"DataDefs.h\"\n";
    print FH "#include \"df/$_.h\"\n" for @include_cpp;
    print FH "#include <stdio.h>\n";
    print FH "static void *moo[1024];\n";
    print FH "$_\n" for @lines_cpp;
    print FH "}\n";
    print FH "int main(int argc, char **argv) {\n";
    print FH "    FILE *fout;\n";
    print FH "    if (argc < 2) return 1;\n";
    # sometimes gcc will generate accesses to the structures at 0 just for a sizeof/offsetof, this works around the segfaults...
    print FH "    for (int i=0 ; i<1024 ; i++) moo[i] = &moo;\n";
    print FH "    fout = fopen(argv[1], \"w\");\n";
    print FH "    cpp_$_(fout);\n" for @include_cpp;
    print FH "    fclose(fout);\n";
    print FH "    return 0;\n";
    print FH "}\n";
} else {
    print FH "$_\n" for @lines_rb;
}
close FH;