mk-okey-map.pl

   1 #!/usr/bin/env perl
   2 require 5.008_001;
   3 use utf8;
   4 #@ Parse 'enum okeys' from nail.h and create okeys.h.  And see accmacvar.c.
   5 # Public Domain
   6
   7 # Acceptable "longest distance" from hash-modulo-index to key
   8 my $MAXDISTANCE_PENALTY = 5;
   9
  10 # Generate a more verbose output.  Not for shipout versions.
  11 my $VERB = 1;
  12
  13 my $OUT = 'okeys.h';
  14
  15 ##  --  >8  --  8<  --  ##
  16
  17 use diagnostics -verbose;
  18 use strict;
  19 use warnings;
  20
  21 use sigtrap qw(handler cleanup normal-signals);
  22
  23 my ($S, @ENTS, $CTOOL, $CTOOL_EXE) = ($VERB ? '   ' : '');
  24
  25 sub main_fun{
  26    if(@ARGV) {$VERB = 0; $S = ''}
  27
  28    parse_nail_h();
  29
  30    create_c_tool();
  31
  32    hash_em();
  33    dump_map();
  34
  35    reverser();
  36
  37    cleanup(undef);
  38    exit 0
  39 }
  40
  41 sub cleanup{
  42    die "$CTOOL_EXE: couldn't unlink: $^E"
  43       if $CTOOL_EXE && -f $CTOOL_EXE && 1 != unlink $CTOOL_EXE;
  44    die "$CTOOL: couldn't unlink: $^E"
  45       if $CTOOL && -f $CTOOL && 1 != unlink $CTOOL;
  46    die "Terminating due to signal $_[0]" if $_[0]
  47 };
  48
  49 sub parse_nail_h{
  50    die "nail.h: open: $^E" unless open F, '<', 'nail.h';
  51    my ($init) = (0);
  52    while(<F>){
  53       # Only want the enum okeys content
  54       if(/^enum okeys/) {$init = 1; next}
  55       if(/^};/) {if($init) {$init = 2; last}; next}
  56       $init || next;
  57
  58       # Ignore empty and comment lines
  59       /^$/ && next;
  60       /^\s*\/\*/ && next;
  61
  62       # An entry may have a comment with special directives
  63       /^\s*(\w+),?\s*(?:\/\*\s*(?:{(.*)})\s*\*\/\s*)?$/;
  64       next unless $1;
  65       my ($k, $x) = ($1, $2);
  66       my %vals;
  67       $vals{enum} = $k;
  68       $vals{bool} = ($k =~ /^ok_b/ ? 1 : 0);
  69       $k = $1 if $k =~ /^ok_[bv]_(.+)$/;
  70       $k =~ s/_/-/g;
  71       $vals{name} = $k;
  72       if($x){
  73          while($x && $x =~ /^([^,]+?)(?:,(.*))?$/){
  74             $x = $2;
  75             $1 =~ /([^=]+)=(.+)/;
  76             die "Unsupported special directive: $1"
  77                if($1 ne 'name' &&
  78                   $1 ne 'rdonly' && $1 ne 'nodel' && $1 ne 'notempty' &&
  79                      $1 ne 'nocntrls' && $1 ne 'num' && $1 ne 'posnum' &&
  80                   $1 ne 'vip' && $1 ne 'virt' &&
  81                   $1 ne 'env' && $1 ne 'import' &&
  82                   $1 ne 'i3val' && $1 ne 'defval');
  83             $vals{$1} = $2
  84          }
  85       }
  86       push @ENTS, \%vals
  87    }
  88    if($init != 2) {die 'nail.h does not have the expected content'}
  89    close F
  90 }
  91
  92 sub create_c_tool{
  93    $CTOOL = './tmp-okey-tool-' . $$ . '.c';
  94    $CTOOL_EXE = $CTOOL . '.exe';
  95
  96    die "$CTOOL: open: $^E" unless open F, '>', $CTOOL;
  97    # xxx optimize: could read lines and write lines in HASH_MODE..
  98    print F '#define MAX_DISTANCE_PENALTY ', $MAXDISTANCE_PENALTY, "\n";
  99 # >>>>>>>>>>>>>>>>>>>
 100    print F <<'_EOT';
 101 #define __CREATE_OKEY_MAP_PL
 102 #include <stdint.h>
 103 #include <stdlib.h>
 104 #include <stdio.h>
 105 #include <string.h>
 106
 107 #define n_NELEM(A) (sizeof(A) / sizeof(A[0]))
 108
 109 #define ui32_t uint32_t
 110 #define ui16_t uint16_t
 111 #define ui8_t uint8_t
 112
 113 enum a_amv_var_flags{
 114    a_AMV_VF_NONE = 0,
 115    a_AMV_VF_BOOL = 1<<0,      /* ok_b_* */
 116    a_AMV_VF_VIRT = 1<<1,      /* "Stateless" automatic variable */
 117    a_AMV_VF_RDONLY = 1<<2,    /* May not be set by user */
 118    a_AMV_VF_NODEL = 1<<3,     /* May not be deleted */
 119    a_AMV_VF_NOTEMPTY = 1<<4,  /* May not be assigned an empty value */
 120    a_AMV_VF_NOCNTRLS = 1<<5,  /* Value may not contain control characters */
 121    a_AMV_VF_NUM = 1<<6,       /* Value must be a 32-bit number */
 122    a_AMV_VF_POSNUM = 1<<7,    /* Value must be positive 32-bit number */
 123    a_AMV_VF_VIP = 1<<8,       /* Wants _var_check_vips() evaluation */
 124    a_AMV_VF_IMPORT = 1<<9,    /* Import ONLY from environ (before PS_STARTED) */
 125    a_AMV_VF_ENV = 1<<10,      /* Update environment on change */
 126    a_AMV_VF_I3VAL = 1<<11,    /* Has an initial value */
 127    a_AMV_VF_DEFVAL = 1<<12,   /* Has a default value */
 128    a_AMV_VF_LINKED = 1<<13,   /* `environ' linked */
 129    a_AMV_VF__MASK = (1<<(13+1)) - 1
 130 };
 131
 132 struct a_amv_var_map{
 133    ui32_t avm_hash;
 134    ui16_t avm_keyoff;
 135    ui16_t avm_flags;    /* enum a_amv_var_flags */
 136 };
 137
 138 #ifdef HASH_MODE
 139 /* NOTE: copied over verbatim from auxlily.c */
 140 static ui32_t
 141 torek_hash(char const *name){
 142    /* Chris Torek's hash.
 143     * NOTE: need to change *at least* mk-okey-map.pl when changing the
 144     * algorithm!! */
 145    ui32_t h = 0;
 146
 147    while(*name != '\0'){
 148       h *= 33;
 149       h += *name++;
 150    }
 151    return h;
 152 }
 153
 154 #else
 155   /* Include what has been written in HASH_MODE */
 156 # define n_CTA(A,S)
 157 # include "okeys.h"
 158
 159 static ui8_t seen_wraparound;
 160 static size_t longest_distance;
 161
 162 static size_t
 163 next_prime(size_t no){ /* blush (brute force) */
 164 jredo:
 165    ++no;
 166    for(size_t i = 3; i < no; i += 2)
 167       if(no % i == 0)
 168          goto jredo;
 169    return no;
 170 }
 171
 172 static size_t *
 173 reversy(size_t size){
 174    struct a_amv_var_map const *vmp = a_amv_var_map,
 175       *vmaxp = vmp + n_NELEM(a_amv_var_map);
 176    size_t ldist = 0, *arr;
 177
 178    arr = malloc(sizeof *arr * size);
 179    for(size_t i = 0; i < size; ++i)
 180       arr[i] = n_NELEM(a_amv_var_map);
 181
 182    seen_wraparound = 0;
 183    longest_distance = 0;
 184
 185    while(vmp < vmaxp){
 186       ui32_t hash = vmp->avm_hash, i = hash % size, l;
 187
 188       for(l = 0; arr[i] != n_NELEM(a_amv_var_map); ++l)
 189          if(++i == size){
 190             seen_wraparound = 1;
 191             i = 0;
 192          }
 193       if(l > longest_distance)
 194          longest_distance = l;
 195       arr[i] = (size_t)(vmp++ - a_amv_var_map);
 196    }
 197    return arr;
 198 }
 199 #endif /* !HASH_MODE */
 200
 201 int
 202 main(int argc, char **argv){
 203 #ifdef HASH_MODE
 204    size_t h = torek_hash(argv[1]);
 205
 206    printf("%lu\n", (unsigned long)h);
 207
 208 #else
 209    size_t *arr, size = n_NELEM(a_amv_var_map);
 210
 211    fprintf(stderr, "Starting reversy, okeys=%zu\n", size);
 212    for(;;){
 213       arr = reversy(size = next_prime(size));
 214       fprintf(stderr, " - size=%zu longest_distance=%zu seen_wraparound=%d\n",
 215          size, longest_distance, seen_wraparound);
 216       if(longest_distance <= MAX_DISTANCE_PENALTY)
 217          break;
 218       free(arr);
 219    }
 220
 221    printf(
 222       "#define a_AMV_VAR_REV_ILL %zuu\n"
 223       "#define a_AMV_VAR_REV_PRIME %zuu\n"
 224       "#define a_AMV_VAR_REV_LONGEST %zuu\n"
 225       "#define a_AMV_VAR_REV_WRAPAROUND %d\n"
 226       "static %s const a_amv_var_revmap[a_AMV_VAR_REV_PRIME] = {\n%s",
 227       n_NELEM(a_amv_var_map), size, longest_distance, seen_wraparound,
 228       argv[1], (argc > 2 ? "  " : ""));
 229    for(size_t i = 0; i < size; ++i)
 230       printf("%s%zuu", (i == 0 ? ""
 231          : (i % 10 == 0 ? (argc > 2 ? ",\n  " : ",\n")
 232             : (argc > 2 ? ", " : ","))),
 233          arr[i]);
 234    printf("\n};\n");
 235 #endif
 236    return 0;
 237 }
 238 _EOT
 239 # <<<<<<<<<<<<<<<<<<<
 240    close F
 241 }
 242
 243 sub hash_em{
 244    system("c99 -DHASH_MODE -I. -o $CTOOL_EXE $CTOOL");
 245
 246    foreach my $e (@ENTS){
 247       my $h = `$CTOOL_EXE $e->{name}`;
 248       chomp $h;
 249       $e->{hash} = $h
 250    }
 251 }
 252
 253 sub dump_map{
 254    die "$OUT: open: $^E" unless open F, '>', $OUT;
 255    print F "/*@ $OUT, generated by $0 on ", scalar gmtime(), ".\n",
 256       " *@ See accmacvar.c for more */\n\n";
 257
 258    print F 'static char const a_amv_var_names[] = {', "\n";
 259    my ($i, $alen) = (0, 0);
 260    my (%virts, %defvals, %i3vals);
 261    foreach my $e (@ENTS){
 262       $e->{keyoff} = $alen;
 263       my $k = $e->{name};
 264       my $l = length $k;
 265       my $a = join '\',\'', split(//, $k);
 266       my (@fa);
 267       if($e->{bool}) {push @fa, 'a_AMV_VF_BOOL'}
 268       if($e->{virt}){
 269          # Virtuals are implicitly rdonly and nodel
 270          $e->{rdonly} = $e->{nodel} = 1;
 271          $virts{$k} = $e;
 272          push @fa, 'a_AMV_VF_VIRT'
 273       }
 274       if($e->{i3val}){
 275          $i3vals{$k} = $e;
 276          push @fa, 'a_AMV_VF_I3VAL'
 277       }
 278       if($e->{defval}){
 279          $e->{notempty} = 1;
 280          $defvals{$k} = $e;
 281          push @fa, 'a_AMV_VF_DEFVAL'
 282       }
 283       if($e->{import}){
 284          $e->{env} = 1;
 285          push @fa, 'a_AMV_VF_IMPORT'
 286       }
 287       if($e->{rdonly}) {push @fa, 'a_AMV_VF_RDONLY'}
 288       if($e->{nodel}) {push @fa, 'a_AMV_VF_NODEL'}
 289       if($e->{notempty}) {push @fa, 'a_AMV_VF_NOTEMPTY'}
 290       if($e->{nocntrls}) {push @fa, 'a_AMV_VF_NOCNTRLS'}
 291       if($e->{num}) {push @fa, 'a_AMV_VF_NUM'}
 292       if($e->{posnum}) {push @fa, 'a_AMV_VF_POSNUM'}
 293       if($e->{vip}) {push @fa, 'a_AMV_VF_VIP'}
 294       if($e->{env}) {push @fa, 'a_AMV_VF_ENV'}
 295       $e->{flags} = \@fa;
 296       my $f = join('|', @fa);
 297       $f = ', ' . $f if length $f;
 298       print F "${S}/* $i. [$alen]+$l $k$f */\n" if $VERB;
 299       print F "${S}'$a','\\0',\n";
 300       ++$i;
 301       $alen += $l + 1
 302    }
 303    print F '};', "\n\n";
 304
 305    print F 'n_CTA(a_AMV_VF_NONE == 0, "Value not 0 as expected");', "\n";
 306    print F 'static struct a_amv_var_map const a_amv_var_map[] = {', "\n";
 307    foreach my $e (@ENTS){
 308       my $f = $VERB ? 'a_AMV_VF_NONE' : '0';
 309       my $fa = join '|', @{$e->{flags}};
 310       $f .= '|' . $fa if length $fa;
 311       my $n = $1 if $e->{enum} =~ /ok_._(.*)/;
 312       print F "${S}{$e->{hash}u, $e->{keyoff}u, $f},";
 313       if($VERB) {print F "${S}/* $n */\n"}
 314       else {print F "\n"}
 315    }
 316    print F '};', "\n\n";
 317
 318    # We have at least version stuff in here
 319    # The problem is that struct var uses a variable sized character buffer
 320    # which cannot be initialized in a conforming way :(
 321    print F <<_EOT;
 322 #ifndef __CREATE_OKEY_MAP_PL
 323 # ifdef HAVE_PUTENV
 324 #  define a_X(X) X
 325 # else
 326 #  define a_X(X)
 327 # endif
 328
 329 /* Unfortunately init of varsized buffer won't work: define "subclass"es */
 330 _EOT
 331    my @skeys = sort keys %virts;
 332
 333    foreach(@skeys){
 334       my $e = $virts{$_};
 335       $e->{vname} = $1 if $e->{enum} =~ /ok_._(.*)/;
 336       $e->{vstruct} = "var_virt_$e->{vname}";
 337       print F "static char const a_amv_$e->{vstruct}_val[] = {$e->{virt}};\n";
 338       print F "static struct{\n";
 339       print F "${S}struct a_amv_var *av_link;\n";
 340       print F "${S}char const *av_value;\n";
 341       print F "${S}a_X(char *av_env;)\n";
 342       print F "${S}ui16_t av_flags;\n";
 343       print F "${S}char const av_name[", length($e->{name}), " +1];\n";
 344       my $f = $VERB ? 'a_AMV_VF_NONE' : '0';
 345       my $fa = join '|', @{$e->{flags}};
 346       $f .= '|' . $fa if length $fa;
 347       print F "} const a_amv_$e->{vstruct} = ",
 348          "{NULL, a_amv_$e->{vstruct}_val, a_X(0 COMMA) $f, ",
 349          "\"$e->{name}\"};\n\n"
 350    }
 351    print F "# undef a_X\n";
 352
 353    print F "\n";
 354    print F '#define a_AMV_VAR_VIRTS_CNT ', scalar @skeys, "\n";
 355    print F 'static struct a_amv_var_virt const a_amv_var_virts[] = {', "\n";
 356    foreach(@skeys){
 357       my $e = $virts{$_};
 358       my $n = $1 if $e->{enum} =~ /ok_._(.*)/;
 359       print F "${S}{$e->{enum}, {0,}, (void const*)&a_amv_$e->{vstruct}},\n";
 360    }
 361    print F "};\n";
 362
 363    #
 364    @skeys = sort keys %i3vals;
 365
 366    print F "\n";
 367    print F '#define a_AMV_VAR_I3VALS_CNT ', scalar @skeys, "\n";
 368    print F 'static struct a_amv_var_defval const a_amv_var_i3vals[] = {', "\n";
 369    foreach(@skeys){
 370       my $e = $i3vals{$_};
 371       print F "${S}{", $e->{enum}, ', {0,}, ',
 372          (!$e->{bool} ? $e->{i3val} : "NULL"), "},\n"
 373    }
 374    print F "};\n";
 375
 376    #
 377    @skeys = sort keys %defvals;
 378
 379    print F "\n";
 380    print F '#define a_AMV_VAR_DEFVALS_CNT ', scalar @skeys, "\n";
 381    print F 'static struct a_amv_var_defval const a_amv_var_defvals[] = {', "\n";
 382    foreach(@skeys){
 383       my $e = $defvals{$_};
 384       print F "${S}{", $e->{enum}, ', {0,}, ',
 385          (!$e->{bool} ? $e->{defval} : "NULL"), "},\n"
 386    }
 387    print F "};\n";
 388
 389    print F "#endif /* __CREATE_OKEY_MAP_PL */\n\n";
 390
 391    die "$OUT: close: $^E" unless close F
 392 }
 393
 394 sub reverser{
 395    my $argv2 = $VERB ? ' verb' : '';
 396    system("c99 -I. -o $CTOOL_EXE $CTOOL");
 397    my $t = (@ENTS < 0xFF ? 'ui8_t' : (@ENTS < 0xFFFF ? 'ui16_t' : 'ui32_t'));
 398    `$CTOOL_EXE $t$argv2 >> $OUT`
 399 }
 400
 401 {package main; main_fun()}
 402
 403 # s-it-mode