macros.pl

   1 #!/usr/bin/perl
   2 #
   3 # macros.pl   produce macros.c from standard.mac
   4 #
   5 # The Netwide Assembler is copyright (C) 1996 Simon Tatham and
   6 # Julian Hall. All rights reserved. The software is
   7 # redistributable under the license given in the file "LICENSE"
   8 # distributed in the NASM archive.
   9
  10 require 'phash.ph';
  11 require 'pptok.ph';
  12
  13 use bytes;
  14
  15 my $fname;
  16 my $line = 0;
  17 my $index      = 0;
  18 my $tasm_count = 0;
  19
  20 #
  21 # Print out a string as a character array
  22 #
  23 sub charcify(@) {
  24     my $l = '';
  25     my $c, $o;
  26     foreach $o (unpack("C*", join('',@_))) {
  27         $c = pack("C", $o);
  28         if ($o < 32 || $o > 126 || $c eq '"' || $c eq "\\") {
  29             $l .= sprintf("(char)%3d,", $o);
  30         } else {
  31             $l .= "\'".$c."\',";
  32         }
  33     }
  34     return $l;
  35 }
  36
  37 #
  38 # Generate macros.c
  39 #
  40 open(OUT,">macros.c") or die "unable to open macros.c\n";
  41
  42 print OUT "/*\n";
  43 print OUT " * Do not edit - this file auto-generated by macros.pl from:\n";
  44 print OUT " *   ", join("\n *   ", @ARGV), "\n";
  45 print OUT " */\n";
  46 print OUT "\n";
  47 print OUT "#include \"tables.h\"\n";
  48 print OUT "#include \"nasmlib.h\"\n";
  49 print OUT "#include \"hashtbl.h\"\n";
  50 print OUT "\n";
  51 print OUT "const char nasm_stdmac[] = {";
  52
  53 my $npkg = 0;
  54 my @pkg_list   = ();
  55 my %pkg_number = ();
  56 my $pkg;
  57 my @out_list   = ();
  58 my $outfmt;
  59 my $lastname;
  60 my $z;
  61
  62 foreach $fname ( @ARGV ) {
  63     open(INPUT,$fname) or die "unable to open $fname\n";
  64     while (<INPUT>) {
  65         $line++;
  66         chomp;
  67         if (m/^\s*\*END\*TASM\*MACROS\*\s*$/) {
  68             $tasm_count = $index;
  69             print OUT "    /* End of TASM macros */\n";
  70         } elsif (m/^OUT:\s*(.*\S)\s*$/) {
  71             undef $pkg;
  72             my @out_alias = split(/\s+/, $1);
  73             printf OUT "        /* %4d */ 0\n", $index++;
  74             print OUT "};\n";
  75             $index = 0;
  76             printf OUT "const char %s_stdmac[] = {\n", $out_alias[0];
  77             print  OUT "    /* From $fname */\n";
  78             $lastname = $fname;
  79             push(@out_list, $out_alias[0]);
  80             $out_index{$out_alias[0]} = $index;
  81         } elsif (m/^USE:\s*(\S+)\s*$/) {
  82             $pkg = $1;
  83             if (defined($pkg_number{$pkg})) {
  84                 die "$0: $fname: duplicate package: $pkg\n";
  85             }
  86             printf OUT "        /* %4d */ 0\n", $index++;
  87             print OUT "};\n";
  88             $index = 0;
  89             printf OUT "static const char nasm_stdmac_%s[] = {\n", $pkg;
  90             print  OUT "    /* From $fname */\n";
  91             $lastname = $fname;
  92             push(@pkg_list, $pkg);
  93             $pkg_number{$pkg} = $npkg++;
  94             $z = pack("C", $pptok_hash{'%define'}+128)."__USE_\U$pkg\E__";
  95             printf OUT "        /* %4d */ %s0,\n", $index, charcify($z);
  96             $index += length($z)+1;
  97         } elsif (m/^\s*((\s*([^\"\';\s]+|\"[^\"]*\"|\'[^\']*\'))*)\s*(;.*)?$/) {
  98             my $s1, $s2, $pd, $ws;
  99             $s1 = $1;
 100             $s2 = '';
 101             while ($s1 =~ /(\%[a-zA-Z_][a-zA-Z0-9_]*)((\s+)(.*)|)$/) {
 102                 $s2 .= "$'";
 103                 $pd = $1;
 104                 $ws = $3;
 105                 $s1 = $4;
 106                 if (defined($pptok_hash{$pd}) &&
 107                     $pptok_hash{$pd} <= 127) {
 108                     $s2 .= pack("C", $pptok_hash{$pd}+128);
 109                 } else {
 110                     $s2 .= $pd.$ws;
 111                 }
 112             }
 113             $s2 .= $s1;
 114             if (length($s2) > 0) {
 115                 if ($lastname ne $fname) {
 116                     print OUT "\n    /* From $fname */\n";
 117                     $lastname = $fname;
 118                 }
 119                 printf OUT "        /* %4d */ %s0,\n",
 120                     $index, charcify($s2);
 121                 $index += length($s2)+1;
 122             }
 123         } else {
 124             die "$fname:$line:  error unterminated quote";
 125         }
 126     }
 127     close(INPUT);
 128 }
 129 printf OUT "        /* %4d */ 0\n};\n\n", $index++;
 130 print OUT "const char * const nasm_stdmac_after_tasm = ",
 131     "&nasm_stdmac[$tasm_count];\n\n";
 132
 133 my @hashinfo = gen_perfect_hash(\%pkg_number);
 134 if (!@hashinfo) {
 135     die "$0: no hash found\n";
 136 }
 137 # Paranoia...
 138 verify_hash_table(\%pkg_number, \@hashinfo);
 139 my ($n, $sv, $g) = @hashinfo;
 140 die if ($n & ($n-1));
 141
 142 print OUT "const char *nasm_stdmac_find_package(const char *package)\n";
 143 print OUT "{\n";
 144 print OUT "    static const struct {\n";
 145 print OUT "         const char *package;\n";
 146 print OUT "         const char *macros;\n";
 147 print OUT "    } packages[$npkg] = {\n";
 148 foreach $pkg (@pkg_list) {
 149     printf OUT "        { \"%s\", nasm_stdmac_%s },\n",
 150         $pkg, $pkg;
 151 }
 152 print OUT "    };\n";
 153
 154 # Put a large value in unused slots.  This makes it extremely unlikely
 155 # that any combination that involves unused slot will pass the range test.
 156 # This speeds up rejection of unrecognized tokens, i.e. identifiers.
 157 print OUT "#define UNUSED 16383\n";
 158
 159 print OUT "    static const int16_t hash1[$n] = {\n";
 160 for ($i = 0; $i < $n; $i++) {
 161     my $h = ${$g}[$i*2+0];
 162     print OUT "        ", defined($h) ? $h : 'UNUSED', ",\n";
 163 }
 164 print OUT "    };\n";
 165
 166 print OUT "    static const int16_t hash2[$n] = {\n";
 167 for ($i = 0; $i < $n; $i++) {
 168     my $h = ${$g}[$i*2+1];
 169     print OUT "        ", defined($h) ? $h : 'UNUSED', ",\n";
 170 }
 171 print OUT "    };\n";
 172
 173 print OUT  "    uint32_t k1, k2;\n";
 174 print OUT  "    uint64_t crc;\n";
 175 # For correct overflow behavior, "ix" should be unsigned of the same
 176 # width as the hash arrays.
 177 print OUT  "    uint16_t ix;\n";
 178 print OUT  "\n";
 179
 180 printf OUT "    crc = crc64i(UINT64_C(0x%08x%08x), package);\n",
 181     $$sv[0], $$sv[1];
 182 print  OUT "    k1 = (uint32_t)crc;\n";
 183 print  OUT "    k2 = (uint32_t)(crc >> 32);\n";
 184 print  OUT "\n";
 185 printf OUT "    ix = hash1[k1 & 0x%x] + hash2[k2 & 0x%x];\n", $n-1, $n-1;
 186 printf OUT "    if (ix >= %d)\n", scalar(@pkg_list);
 187 print OUT  "        return NULL;\n";
 188 print OUT  "\n";
 189 print OUT  "    if (nasm_stricmp(packages[ix].package, package))\n";
 190 print OUT  "        return NULL;\n";
 191 print OUT  "\n";
 192 print OUT  "    return packages[ix].macros;\n";
 193 print OUT  "}\n";
 194
 195 close(OUT);