pptok.pl

   1 #!/usr/bin/perl
   2 #
   3 # Produce pptok.c and pptok.h from pptok.dat
   4 #
   5
   6 require 'phash.ph';
   7
   8 my($what, $in, $out) = @ARGV;
   9
  10 #
  11 # Read pptok.dat
  12 #
  13 open(IN, "< $in") or die "$0: cannot open: $in\n";
  14 while (defined($line = <IN>)) {
  15     chomp $line;
  16     $line =~ s/^\s+//;          # Remove leading whitespace
  17     $line =~ s/\s*\#.*$//;      # Remove comments and trailing whitespace
  18     next if ($line eq '');
  19
  20     if ($line =~ /^\%(.*)\*$/) {
  21         push(@cctok, $1);
  22     } elsif ($line =~ /^\%(.*)$/) {
  23         push(@pptok, $1);
  24     } elsif ($line =~ /^\*(.*)$/) {
  25         push(@cond, $1);
  26     }
  27 }
  28 close(IN);
  29
  30 @cctok = sort @cctok;
  31 @cond = sort @cond;
  32 @pptok = sort @pptok;
  33
  34 # Generate the expanded list including conditionals.  The conditionals
  35 # are at the beginning, padded to a power of 2, with the inverses
  36 # interspersed; this allows a simple mask to pick out the condition.
  37
  38 while ((scalar @cond) & (scalar @cond)-1) {
  39     push(@cond, undef);
  40 }
  41
  42 @cptok = ();
  43 foreach $ct (@cctok) {
  44     foreach $cc (@cond) {
  45         if (defined($cc)) {
  46             push(@cptok, $ct.$cc);
  47             push(@cptok, $ct.'n'.$cc);
  48         } else {
  49             push(@cptok, undef, undef);
  50         }
  51     }
  52 }
  53 $first_uncond = $pptok[0];
  54 @pptok = (@cptok, @pptok);
  55
  56 open(OUT, "> $out") or die "$0: cannot open: $out\n";
  57 print OUT "/* Automatically generated from $in by $0 */\n";
  58 print OUT "/* Do not edit */\n";
  59 print OUT "\n";
  60
  61 #
  62 # Output pptok.h
  63 #
  64 if ($what eq 'h') {
  65     print OUT "enum preproc_token {\n";
  66     $n = 0;
  67     foreach $pt (@pptok) {
  68         if (defined($pt)) {
  69             printf OUT "    %-16s = %3d,\n", "PP_\U$pt\E", $n;
  70         }
  71         $n++;
  72     }
  73     printf OUT "    %-16s = %3d\n", 'PP_INVALID', -1;
  74     print OUT "};\n";
  75     print OUT "\n";
  76
  77     print  OUT "enum pp_conditional {\n";
  78     $n = 0;
  79     foreach $cc (@cond) {
  80         if (defined($cc)) {
  81             printf OUT "    %-16s = %3d,\n", "PPC_IF\U$cc\E", $n;
  82         }
  83         $n += 2;
  84     }
  85     print  OUT "};\n\n";
  86
  87     printf OUT "#define PP_COND(x)     ((enum pp_conditional)((x) & 0x%x))\n",
  88         (scalar(@cond)-1) << 1;
  89     print  OUT "#define PP_IS_COND(x)  ((unsigned int)(x) < PP_\U$first_uncond\E)\n";
  90     print  OUT "#define PP_NEGATIVE(x) ((x) & 1)\n";
  91     print  OUT "\n";
  92
  93     foreach $ct (@cctok) {
  94         print OUT "#define CASE_PP_\U$ct\E";
  95         $pref = " \\\n";
  96         foreach $cc (@cond) {
  97             if (defined($cc)) {
  98                 print OUT "$pref\tcase PP_\U${ct}${cc}\E: \\\n";
  99                 print OUT "\tcase PP_\U${ct}N${cc}\E";
 100                 $pref = ":\\\n";
 101             }
 102         }
 103         print OUT "\n";         # No colon or newline on the last one
 104     }
 105 }
 106
 107 #
 108 # Output pptok.c
 109 #
 110 if ($what eq 'c') {
 111     my %tokens = ();
 112     my @tokendata = ();
 113
 114     my $n = 0;
 115     foreach $pt (@pptok) {
 116         if (defined($pt)) {
 117             $tokens{'%'.$pt} = $n;
 118             if ($pt =~ /[\@\[\]\\_]/) {
 119                 # Fail on characters which look like upper-case letters
 120                 # to the quick-and-dirty downcasing in the prehash
 121                 # (see below)
 122                 die "$in: invalid character in token: $pt";
 123             }
 124         }
 125         $n++;
 126     }
 127
 128     my @hashinfo = gen_perfect_hash(\%tokens);
 129     if (!defined(@hashinfo)) {
 130         die "$0: no hash found\n";
 131     }
 132
 133     # Paranoia...
 134     verify_hash_table(\%tokens, \@hashinfo);
 135
 136     ($n, $sv, $g) = @hashinfo;
 137     $sv2 = $sv+2;
 138
 139     die if ($n & ($n-1));
 140
 141     print OUT "#include <inttypes.h>\n";
 142     print OUT "#include <ctype.h>\n";
 143     print OUT "#include \"nasmlib.h\"\n";
 144     print OUT "#include \"preproc.h\"\n";
 145     print OUT "\n";
 146
 147     print OUT "#define rot(x,y) (((uint32_t)(x) << (y))+((uint32_t)(x) >> (32-(y))))\n";
 148     print OUT "\n";
 149
 150     # Note that this is global.
 151     printf OUT "const char * const pp_directives[%d] = {\n", scalar(@pptok);
 152     foreach $d (@pptok) {
 153         if (defined($d)) {
 154             print OUT "    \"%$d\",\n";
 155         } else {
 156             print OUT "    NULL,\n";
 157         }
 158     }
 159     print OUT  "};\n";
 160
 161     print OUT "enum preproc_token pp_token_hash(const char *token)\n";
 162     print OUT "{\n";
 163
 164     # Put a large value in unused slots.  This makes it extremely unlikely
 165     # that any combination that involves unused slot will pass the range test.
 166     # This speeds up rejection of unrecognized tokens, i.e. identifiers.
 167     print OUT "#define UNUSED 16383\n";
 168
 169     print OUT "    static const int16_t hash1[$n] = {\n";
 170     for ($i = 0; $i < $n; $i++) {
 171         my $h = ${$g}[$i*2+0];
 172         print OUT "        ", defined($h) ? $h : 'UNUSED', ",\n";
 173     }
 174     print OUT "    };\n";
 175
 176     print OUT "    static const int16_t hash2[$n] = {\n";
 177     for ($i = 0; $i < $n; $i++) {
 178         my $h = ${$g}[$i*2+1];
 179         print OUT "        ", defined($h) ? $h : 'UNUSED', ",\n";
 180     }
 181     print OUT "    };\n";
 182
 183     print OUT  "    uint32_t k1 = 0, k2 = 0;\n";
 184     print OUT  "    uint8_t c;\n";
 185     # For correct overflow behavior, "ix" should be unsigned of the same
 186     # width as the hash arrays.
 187     print OUT  "    uint16_t ix;\n";
 188     print OUT  "    const char *p = token;\n";
 189     print OUT  "\n";
 190
 191     print OUT  "    while ((c = *p++) != 0) {\n";
 192     print OUT  "        uint32_t kn1, kn2;\n";
 193     print OUT  "        c |= 0x20; /* convert to lower case */\n";
 194     printf OUT "        kn1 = rot(k1,%2d) - rot(k2,%2d) + c;\n", ${$sv}[0], ${$sv}[1];
 195     printf OUT "        kn2 = rot(k2,%2d) - rot(k1,%2d) + c;\n", ${$sv}[2], ${$sv}[3];
 196     print OUT  "        k1 = kn1; k2 = kn2;\n";
 197     print OUT  "    }\n";
 198     print OUT  "\n";
 199     printf OUT "    ix = hash1[k1 & 0x%x] + hash2[k2 & 0x%x];\n", $n-1, $n-1;
 200     printf OUT "    if (ix >= %d)\n", scalar(@pptok);
 201     print OUT  "        return PP_INVALID;\n";
 202     print OUT  "\n";
 203
 204     print OUT  "    if (!pp_directives[ix] || nasm_stricmp(pp_directives[ix], token))\n";
 205     print OUT  "        return PP_INVALID;\n";
 206     print OUT  "\n";
 207     print OUT  "    return ix;\n";
 208     print OUT  "}\n";
 209 }