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 'virt' && $1 ne 'nolopts' &&
  79                   $1 ne 'rdonly' && $1 ne 'nodel' && $1 ne 'notempty' &&
  80                      $1 ne 'nocntrls' &&
  81                   $1 ne 'num' && $1 ne 'posnum' && $1 ne 'lower' &&
  82                   $1 ne 'vip' && $1 ne 'import' && $1 ne 'env' &&
  83                   $1 ne 'i3val' && $1 ne 'defval');
  84             $vals{$1} = $2
  85          }
  86       }
  87       push @ENTS, \%vals
  88    }
  89    if($init != 2) {die 'nail.h does not have the expected content'}
  90    close F
  91 }
  92
  93 sub create_c_tool{
  94    $CTOOL = './tmp-okey-tool-' . $$ . '.c';
  95    $CTOOL_EXE = $CTOOL . '.exe';
  96
  97    die "$CTOOL: open: $^E" unless open F, '>', $CTOOL;
  98    # xxx optimize: could read lines and write lines in HASH_MODE..
  99    print F '#define MAX_DISTANCE_PENALTY ', $MAXDISTANCE_PENALTY, "\n";
 100 # >>>>>>>>>>>>>>>>>>>
 101    print F <<'_EOT';
 102 #define __CREATE_OKEY_MAP_PL
 103 #include <stdint.h>
 104 #include <stdlib.h>
 105 #include <stdio.h>
 106 #include <string.h>
 107
 108 #define n_NELEM(A) (sizeof(A) / sizeof(A[0]))
 109
 110 #define ui32_t uint32_t
 111 #define ui16_t uint16_t
 112 #define ui8_t uint8_t
 113
 114 enum a_amv_var_flags{
 115    a_AMV_VF_NONE = 0,
 116    a_AMV_VF_BOOL = 1<<0,      /* ok_b_* */
 117    a_AMV_VF_VIRT = 1<<1,      /* "Stateless" automatic variable */
 118    a_AMV_VF_NOLOPTS = 1<<2,   /* May not be tracked by `localopts' */
 119    a_AMV_VF_RDONLY = 1<<3,    /* May not be set by user */
 120    a_AMV_VF_NODEL = 1<<4,     /* May not be deleted */
 121    a_AMV_VF_NOTEMPTY = 1<<5,  /* May not be assigned an empty value */
 122    a_AMV_VF_NOCNTRLS = 1<<6,  /* Value may not contain control characters */
 123    a_AMV_VF_NUM = 1<<7,       /* Value must be a 32-bit number */
 124    a_AMV_VF_POSNUM = 1<<8,    /* Value must be positive 32-bit number */
 125    a_AMV_VF_LOWER = 1<<9,     /* Values will be stored in a lowercase version */
 126    a_AMV_VF_VIP = 1<<10,      /* Wants _var_check_vips() evaluation */
 127    a_AMV_VF_IMPORT = 1<<11,   /* Import ONLY from environ (pre n_PSO_STARTED) */
 128    a_AMV_VF_ENV = 1<<12,      /* Update environment on change */
 129    a_AMV_VF_I3VAL = 1<<13,    /* Has an initial value */
 130    a_AMV_VF_DEFVAL = 1<<14,   /* Has a default value */
 131    a_AMV_VF_LINKED = 1<<15,   /* `environ' linked */
 132    a_AMV_VF__MASK = (1<<(15+1)) - 1
 133 };
 134
 135 struct a_amv_var_map{
 136    ui32_t avm_hash;
 137    ui16_t avm_keyoff;
 138    ui16_t avm_flags;    /* enum a_amv_var_flags */
 139 };
 140
 141 #ifdef HASH_MODE
 142 /* NOTE: copied over verbatim from auxlily.c */
 143 static ui32_t
 144 torek_hash(char const *name){
 145    /* Chris Torek's hash.
 146     * NOTE: need to change *at least* mk-okey-map.pl when changing the
 147     * algorithm!! */
 148    ui32_t h = 0;
 149
 150    while(*name != '\0'){
 151       h *= 33;
 152       h += *name++;
 153    }
 154    return h;
 155 }
 156
 157 #else
 158   /* Include what has been written in HASH_MODE */
 159 # define n_CTA(A,S)
 160 # include "okeys.h"
 161
 162 static ui8_t seen_wraparound;
 163 static size_t longest_distance;
 164
 165 static size_t
 166 next_prime(size_t no){ /* blush (brute force) */
 167 jredo:
 168    ++no;
 169    for(size_t i = 3; i < no; i += 2)
 170       if(no % i == 0)
 171          goto jredo;
 172    return no;
 173 }
 174
 175 static size_t *
 176 reversy(size_t size){
 177    struct a_amv_var_map const *vmp = a_amv_var_map,
 178       *vmaxp = vmp + n_NELEM(a_amv_var_map);
 179    size_t ldist = 0, *arr;
 180
 181    arr = malloc(sizeof *arr * size);
 182    for(size_t i = 0; i < size; ++i)
 183       arr[i] = n_NELEM(a_amv_var_map);
 184
 185    seen_wraparound = 0;
 186    longest_distance = 0;
 187
 188    while(vmp < vmaxp){
 189       ui32_t hash = vmp->avm_hash, i = hash % size, l;
 190
 191       for(l = 0; arr[i] != n_NELEM(a_amv_var_map); ++l)
 192          if(++i == size){
 193             seen_wraparound = 1;
 194             i = 0;
 195          }
 196       if(l > longest_distance)
 197          longest_distance = l;
 198       arr[i] = (size_t)(vmp++ - a_amv_var_map);
 199    }
 200    return arr;
 201 }
 202 #endif /* !HASH_MODE */
 203
 204 int
 205 main(int argc, char **argv){
 206 #ifdef HASH_MODE
 207    size_t h = torek_hash(argv[1]);
 208
 209    printf("%lu\n", (unsigned long)h);
 210
 211 #else
 212    size_t *arr, size = n_NELEM(a_amv_var_map);
 213
 214    fprintf(stderr, "Starting reversy, okeys=%zu\n", size);
 215    for(;;){
 216       arr = reversy(size = next_prime(size));
 217       fprintf(stderr, " - size=%zu longest_distance=%zu seen_wraparound=%d\n",
 218          size, longest_distance, seen_wraparound);
 219       if(longest_distance <= MAX_DISTANCE_PENALTY)
 220          break;
 221       free(arr);
 222    }
 223
 224    printf(
 225       "#define a_AMV_VAR_REV_ILL %zuu\n"
 226       "#define a_AMV_VAR_REV_PRIME %zuu\n"
 227       "#define a_AMV_VAR_REV_LONGEST %zuu\n"
 228       "#define a_AMV_VAR_REV_WRAPAROUND %d\n"
 229       "static %s const a_amv_var_revmap[a_AMV_VAR_REV_PRIME] = {\n%s",
 230       n_NELEM(a_amv_var_map), size, longest_distance, seen_wraparound,
 231       argv[1], (argc > 2 ? "  " : ""));
 232    for(size_t i = 0; i < size; ++i)
 233       printf("%s%zuu", (i == 0 ? ""
 234          : (i % 10 == 0 ? (argc > 2 ? ",\n  " : ",\n")
 235             : (argc > 2 ? ", " : ","))),
 236          arr[i]);
 237    printf("\n};\n");
 238 #endif
 239    return 0;
 240 }
 241 _EOT
 242 # <<<<<<<<<<<<<<<<<<<
 243    close F
 244 }
 245
 246 sub hash_em{
 247    system("c99 -DHASH_MODE -I. -o $CTOOL_EXE $CTOOL");
 248
 249    foreach my $e (@ENTS){
 250       my $h = `$CTOOL_EXE $e->{name}`;
 251       chomp $h;
 252       $e->{hash} = $h
 253    }
 254 }
 255
 256 sub dump_map{
 257    die "$OUT: open: $^E" unless open F, '>', $OUT;
 258    print F "/*@ $OUT, generated by $0 on ", scalar gmtime(), ".\n",
 259       " *@ See accmacvar.c for more */\n\n";
 260
 261    print F 'static char const a_amv_var_names[] = {', "\n";
 262    my ($i, $alen) = (0, 0);
 263    my (%virts, %defvals, %i3vals);
 264    foreach my $e (@ENTS){
 265       $e->{keyoff} = $alen;
 266       my $k = $e->{name};
 267       my $l = length $k;
 268       my $a = join '\',\'', split(//, $k);
 269       my (@fa);
 270       if($e->{bool}) {push @fa, 'a_AMV_VF_BOOL'}
 271       if($e->{virt}){
 272          # Virtuals are implicitly rdonly and nodel
 273          $e->{rdonly} = $e->{nodel} = 1;
 274          $virts{$k} = $e;
 275          push @fa, 'a_AMV_VF_VIRT'
 276       }
 277       if(defined $e->{i3val}){
 278          $i3vals{$k} = $e;
 279          push @fa, 'a_AMV_VF_I3VAL'
 280       }
 281       if($e->{defval}){
 282          $e->{notempty} = 1;
 283          $defvals{$k} = $e;
 284          push @fa, 'a_AMV_VF_DEFVAL'
 285       }
 286       if($e->{import}){
 287          $e->{env} = 1;
 288          push @fa, 'a_AMV_VF_IMPORT'
 289       }
 290       if($e->{nolopts}) {push @fa, 'a_AMV_VF_NOLOPTS'}
 291       if($e->{rdonly}) {push @fa, 'a_AMV_VF_RDONLY'}
 292       if($e->{nodel}) {push @fa, 'a_AMV_VF_NODEL'}
 293       if($e->{notempty}) {push @fa, 'a_AMV_VF_NOTEMPTY'}
 294       if($e->{nocntrls}) {push @fa, 'a_AMV_VF_NOCNTRLS'}
 295       if($e->{num}) {push @fa, 'a_AMV_VF_NUM'}
 296       if($e->{posnum}) {push @fa, 'a_AMV_VF_POSNUM'}
 297       if($e->{lower}) {push @fa, 'a_AMV_VF_LOWER'}
 298       if($e->{vip}) {push @fa, 'a_AMV_VF_VIP'}
 299       if($e->{env}) {push @fa, 'a_AMV_VF_ENV'}
 300       $e->{flags} = \@fa;
 301       my $f = join('|', @fa);
 302       $f = ', ' . $f if length $f;
 303       print F "${S}/* $i. [$alen]+$l $k$f */\n" if $VERB;
 304       print F "${S}'$a','\\0',\n";
 305       ++$i;
 306       $alen += $l + 1
 307    }
 308    print F '};', "\n\n";
 309
 310    print F 'n_CTA(a_AMV_VF_NONE == 0, "Value not 0 as expected");', "\n";
 311    print F 'static struct a_amv_var_map const a_amv_var_map[] = {', "\n";
 312    foreach my $e (@ENTS){
 313       my $f = $VERB ? 'a_AMV_VF_NONE' : '0';
 314       my $fa = join '|', @{$e->{flags}};
 315       $f .= '|' . $fa if length $fa;
 316       print F "${S}{$e->{hash}u, $e->{keyoff}u, $f},";
 317       if($VERB) {print F "${S}/* $e->{name} */\n"}
 318       else {print F "\n"}
 319    }
 320    print F '};', "\n\n";
 321
 322    # We have at least version stuff in here
 323    # The problem is that struct var uses a variable sized character buffer
 324    # which cannot be initialized in a conforming way :(
 325    print F <<_EOT;
 326 #ifndef __CREATE_OKEY_MAP_PL
 327 # ifdef HAVE_PUTENV
 328 #  define a_X(X) X
 329 # else
 330 #  define a_X(X)
 331 # endif
 332
 333 /* Unfortunately init of varsized buffer won't work: define "subclass"es */
 334 _EOT
 335    my @skeys = sort keys %virts;
 336
 337    foreach(@skeys){
 338       my $e = $virts{$_};
 339       $e->{vname} = $1 if $e->{enum} =~ /ok_._(.*)/;
 340       $e->{vstruct} = "var_virt_$e->{vname}";
 341       print F "static char const a_amv_$e->{vstruct}_val[] = {$e->{virt}};\n";
 342       print F "static struct{\n";
 343       print F "${S}struct a_amv_var *av_link;\n";
 344       print F "${S}char const *av_value;\n";
 345       print F "${S}a_X(char *av_env;)\n";
 346       print F "${S}ui16_t av_flags;\n";
 347       print F "${S}char const av_name[", length($e->{name}), " +1];\n";
 348       my $f = $VERB ? 'a_AMV_VF_NONE' : '0';
 349       my $fa = join '|', @{$e->{flags}};
 350       $f .= '|' . $fa if length $fa;
 351       print F "} const a_amv_$e->{vstruct} = ",
 352          "{NULL, a_amv_$e->{vstruct}_val, a_X(0 COMMA) $f, ",
 353          "\"$e->{name}\"};\n\n"
 354    }
 355    print F "# undef a_X\n";
 356
 357    print F "\n";
 358    print F '#define a_AMV_VAR_VIRTS_CNT ', scalar @skeys, "\n";
 359    print F 'static struct a_amv_var_virt const a_amv_var_virts[] = {', "\n";
 360    foreach(@skeys){
 361       my $e = $virts{$_};
 362       my $n = $1 if $e->{enum} =~ /ok_._(.*)/;
 363       print F "${S}{$e->{enum}, {0,}, (void const*)&a_amv_$e->{vstruct}},\n";
 364    }
 365    print F "};\n";
 366
 367    #
 368    @skeys = sort keys %i3vals;
 369
 370    print F "\n";
 371    print F '#define a_AMV_VAR_I3VALS_CNT ', scalar @skeys, "\n";
 372    print F 'static struct a_amv_var_defval const a_amv_var_i3vals[] = {', "\n";
 373    foreach(@skeys){
 374       my $e = $i3vals{$_};
 375       print F "${S}{", $e->{enum}, ', {0,}, ',
 376          (!$e->{bool} ? $e->{i3val} : "NULL"), "},\n"
 377    }
 378    print F "};\n";
 379
 380    #
 381    @skeys = sort keys %defvals;
 382
 383    print F "\n";
 384    print F '#define a_AMV_VAR_DEFVALS_CNT ', scalar @skeys, "\n";
 385    print F 'static struct a_amv_var_defval const a_amv_var_defvals[] = {', "\n";
 386    foreach(@skeys){
 387       my $e = $defvals{$_};
 388       print F "${S}{", $e->{enum}, ', {0,}, ',
 389          (!$e->{bool} ? $e->{defval} : "NULL"), "},\n"
 390    }
 391    print F "};\n";
 392
 393    print F "#endif /* __CREATE_OKEY_MAP_PL */\n";
 394
 395    # Special var backing [#@*?]|[1-9][0-9]*|0
 396    $i = 0;
 397    print F "\n";
 398    foreach my $e (@ENTS){
 399       if($e->{name} eq '--special-param'){
 400          print F "#define a_AMV_VAR_SPECIAL_PARAM_MAP_IDX ${i}u\n"
 401       }
 402       # The rest are only speedups
 403       elsif($e->{name} eq '?'){
 404          print F "#define a_AMV_VAR_QM_MAP_IDX ${i}u\n"
 405       }elsif($e->{name} eq '!'){
 406          print F "#define a_AMV_VAR_EM_MAP_IDX ${i}u\n"
 407       }
 408       ++$i
 409    }
 410
 411    print F "\n";
 412    die "$OUT: close: $^E" unless close F
 413 }
 414
 415 sub reverser{
 416    my $argv2 = $VERB ? ' verb' : '';
 417    system("c99 -I. -o $CTOOL_EXE $CTOOL");
 418    my $t = (@ENTS < 0xFF ? 'ui8_t' : (@ENTS < 0xFFFF ? 'ui16_t' : 'ui32_t'));
 419    `$CTOOL_EXE $t$argv2 >> $OUT`
 420 }
 421
 422 {package main; main_fun()}
 423
 424 # s-it-mode