make-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 gen-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 $MAILX = 'LC_ALL=C s-nail -#:/';
  14 my $OUT = 'gen-okeys.h';
  15
  16 ##  --  >8  --  8<  --  ##
  17
  18 use diagnostics -verbose;
  19 use strict;
  20 use warnings;
  21
  22 use FileHandle;
  23 use IPC::Open2;
  24
  25 use sigtrap qw(handler cleanup normal-signals);
  26
  27 my ($S, @ENTS, $CTOOL, $CTOOL_EXE) = ($VERB ? '   ' : '');
  28
  29 sub main_fun{
  30    if(@ARGV) {$VERB = 0; $S = ''}
  31
  32    parse_nail_h();
  33
  34    create_c_tool();
  35
  36    hash_em();
  37
  38    dump_map();
  39
  40    reverser();
  41
  42    cleanup(undef);
  43    exit 0
  44 }
  45
  46 sub cleanup{
  47    die "$CTOOL_EXE: couldn't unlink: $^E"
  48       if $CTOOL_EXE && -f $CTOOL_EXE && 1 != unlink $CTOOL_EXE;
  49    die "$CTOOL: couldn't unlink: $^E"
  50       if $CTOOL && -f $CTOOL && 1 != unlink $CTOOL;
  51    die "Terminating due to signal $_[0]" if $_[0]
  52 };
  53
  54 sub parse_nail_h{
  55    die "nail.h: open: $^E" unless open F, '<', 'nail.h';
  56    my ($init) = (0);
  57    while(<F>){
  58       # Only want the enum okeys content
  59       if(/^enum okeys/) {$init = 1; next}
  60       if(/^};/) {if($init) {$init = 2; last}; next}
  61       $init || next;
  62
  63       # Ignore empty and comment lines
  64       /^$/ && next;
  65       /^\s*\/\*/ && next;
  66
  67       # An entry may have a comment with special directives
  68       /^\s*(\w+),?\s*(?:\/\*\s*(?:{(.*)})\s*\*\/\s*)?$/;
  69       next unless $1;
  70       my ($k, $x) = ($1, $2);
  71       my %vals;
  72       $vals{enum} = $k;
  73       $vals{bool} = ($k =~ /^ok_b/ ? 1 : 0);
  74       $k = $1 if $k =~ /^ok_[bv]_(.+)$/;
  75       $k =~ s/_/-/g;
  76       $vals{name} = $k;
  77       if($x){
  78          # {\}: overlong entry, placed on follow line
  79          if($x =~ /\s*\\\s*$/){
  80             $_ = <F>;
  81             die 'nail.h: missing continuation line' unless $_;
  82             /^\s*\/\*\s*{(.*)}\s*\*\/\s*$/;
  83             $x = $1;
  84             die 'nail.h: invalid continuation line' unless $x
  85          }
  86
  87          while($x && $x =~ /^([^,]+?)(?:,(.*))?$/){
  88             $x = $2;
  89             $1 =~ /([^=]+)=(.+)/;
  90             die "Unsupported special directive: $1"
  91                if($1 ne 'name' &&
  92                   $1 ne 'virt' && $1 ne 'nolopts' &&
  93                   $1 ne 'rdonly' && $1 ne 'nodel' && $1 ne 'notempty' &&
  94                      $1 ne 'nocntrls' &&
  95                   $1 ne 'num' && $1 ne 'posnum' && $1 ne 'lower' &&
  96                   $1 ne 'vip' && $1 ne 'import' && $1 ne 'env' &&
  97                   $1 ne 'i3val' && $1 ne 'defval');
  98             $vals{$1} = $2
  99          }
 100       }
 101       push @ENTS, \%vals
 102    }
 103    if($init != 2) {die 'nail.h does not have the expected content'}
 104    close F
 105 }
 106
 107 sub create_c_tool{
 108    $CTOOL = './tmp-okey-tool-' . $$ . '.c';
 109    $CTOOL_EXE = $CTOOL . '.exe';
 110
 111    die "$CTOOL: open: $^E" unless open F, '>', $CTOOL;
 112    print F '#define MAX_DISTANCE_PENALTY ', $MAXDISTANCE_PENALTY, "\n";
 113 # >>>>>>>>>>>>>>>>>>>
 114    print F <<'_EOT';
 115 #define __CREATE_OKEY_MAP_PL
 116 #include <stdint.h>
 117 #include <stdlib.h>
 118 #include <stdio.h>
 119 #include <string.h>
 120
 121 #define n_NELEM(A) (sizeof(A) / sizeof(A[0]))
 122
 123 #define ui32_t uint32_t
 124 #define ui16_t uint16_t
 125 #define ui8_t uint8_t
 126
 127 enum a_amv_var_flags{
 128    a_AMV_VF_NONE = 0,
 129    a_AMV_VF_BOOL = 1u<<0,     /* ok_b_* */
 130    a_AMV_VF_VIRT = 1u<<1,     /* "Stateless" automatic variable */
 131    a_AMV_VF_NOLOPTS = 1u<<2,  /* May not be tracked by `localopts' */
 132    a_AMV_VF_RDONLY = 1u<<3,   /* May not be set by user */
 133    a_AMV_VF_NODEL = 1u<<4,    /* May not be deleted */
 134    a_AMV_VF_NOTEMPTY = 1u<<5, /* May not be assigned an empty value */
 135    a_AMV_VF_NOCNTRLS = 1u<<6, /* Value may not contain control characters */
 136    a_AMV_VF_NUM = 1u<<7,      /* Value must be a 32-bit number */
 137    a_AMV_VF_POSNUM = 1u<<8,   /* Value must be positive 32-bit number */
 138    a_AMV_VF_LOWER = 1u<<9,    /* Values will be stored in a lowercase version */
 139    a_AMV_VF_VIP = 1u<<10,     /* Wants _var_check_vips() evaluation */
 140    a_AMV_VF_IMPORT = 1u<<11,  /* Import ONLY from environ (pre n_PSO_STARTED) */
 141    a_AMV_VF_ENV = 1u<<12,     /* Update environment on change */
 142    a_AMV_VF_I3VAL = 1u<<13,   /* Has an initial value */
 143    a_AMV_VF_DEFVAL = 1u<<14,  /* Has a default value */
 144    a_AMV_VF_LINKED = 1u<<15,  /* `environ' linked */
 145    a_AMV_VF__MASK = (1u<<(15+1)) - 1
 146 };
 147
 148 struct a_amv_var_map{
 149    ui32_t avm_hash;
 150    ui16_t avm_keyoff;
 151    ui16_t avm_flags;    /* enum a_amv_var_flags */
 152 };
 153
 154 #define n_CTA(A,S)
 155 #include "gen-okeys.h"
 156
 157 static ui8_t seen_wraparound;
 158 static size_t longest_distance;
 159
 160 static size_t
 161 next_prime(size_t no){ /* blush (brute force) */
 162 jredo:
 163    ++no;
 164    for(size_t i = 3; i < no; i += 2)
 165       if(no % i == 0)
 166          goto jredo;
 167    return no;
 168 }
 169
 170 static size_t *
 171 reversy(size_t size){
 172    struct a_amv_var_map const *vmp = a_amv_var_map,
 173       *vmaxp = vmp + n_NELEM(a_amv_var_map);
 174    size_t ldist = 0, *arr;
 175
 176    arr = malloc(sizeof *arr * size);
 177    for(size_t i = 0; i < size; ++i)
 178       arr[i] = n_NELEM(a_amv_var_map);
 179
 180    seen_wraparound = 0;
 181    longest_distance = 0;
 182
 183    while(vmp < vmaxp){
 184       ui32_t hash = vmp->avm_hash, i = hash % size, l;
 185
 186       for(l = 0; arr[i] != n_NELEM(a_amv_var_map); ++l)
 187          if(++i == size){
 188             seen_wraparound = 1;
 189             i = 0;
 190          }
 191       if(l > longest_distance)
 192          longest_distance = l;
 193       arr[i] = (size_t)(vmp++ - a_amv_var_map);
 194    }
 195    return arr;
 196 }
 197
 198 int
 199 main(int argc, char **argv){
 200    size_t *arr, size = n_NELEM(a_amv_var_map);
 201
 202    fprintf(stderr, "Starting reversy, okeys=%zu\n", size);
 203    for(;;){
 204       arr = reversy(size = next_prime(size));
 205       fprintf(stderr, " - size=%zu longest_distance=%zu seen_wraparound=%d\n",
 206          size, longest_distance, seen_wraparound);
 207       if(longest_distance <= MAX_DISTANCE_PENALTY)
 208          break;
 209       free(arr);
 210    }
 211
 212    printf(
 213       "#define a_AMV_VAR_REV_ILL %zuu\n"
 214       "#define a_AMV_VAR_REV_PRIME %zuu\n"
 215       "#define a_AMV_VAR_REV_LONGEST %zuu\n"
 216       "#define a_AMV_VAR_REV_WRAPAROUND %d\n"
 217       "static %s const a_amv_var_revmap[a_AMV_VAR_REV_PRIME] = {\n%s",
 218       n_NELEM(a_amv_var_map), size, longest_distance, seen_wraparound,
 219       argv[1], (argc > 2 ? "  " : ""));
 220    for(size_t i = 0; i < size; ++i)
 221       printf("%s%zuu", (i == 0 ? ""
 222          : (i % 10 == 0 ? (argc > 2 ? ",\n  " : ",\n")
 223             : (argc > 2 ? ", " : ","))),
 224          arr[i]);
 225    printf("\n};\n");
 226    return 0;
 227 }
 228 _EOT
 229 # <<<<<<<<<<<<<<<<<<<
 230    close F
 231 }
 232
 233 sub hash_em{
 234    die "hash_em: open: $^E"
 235       unless my $pid = open2 *RFD, *WFD, $MAILX;
 236    foreach my $e (@ENTS){
 237       print WFD "vexpr hash $e->{name}\n";
 238       my $h = <RFD>;
 239       chomp $h;
 240       $e->{hash} = $h
 241    }
 242    print WFD "x\n";
 243    waitpid $pid, 0;
 244 }
 245
 246 sub dump_map{
 247    die "$OUT: open: $^E" unless open F, '>', $OUT;
 248    print F "/*@ $OUT, generated by $0.\n",
 249       " *@ See accmacvar.c for more */\n\n";
 250
 251    print F 'static char const a_amv_var_names[] = {', "\n";
 252    my ($i, $alen) = (0, 0);
 253    my (%virts, %defvals, %i3vals);
 254    foreach my $e (@ENTS){
 255       $e->{keyoff} = $alen;
 256       my $k = $e->{name};
 257       my $l = length $k;
 258       my $a = join '\',\'', split(//, $k);
 259       my (@fa);
 260       if($e->{bool}) {push @fa, 'a_AMV_VF_BOOL'}
 261       if($e->{virt}){
 262          # Virtuals are implicitly rdonly and nodel
 263          $e->{rdonly} = $e->{nodel} = 1;
 264          $virts{$k} = $e;
 265          push @fa, 'a_AMV_VF_VIRT'
 266       }
 267       if(defined $e->{i3val}){
 268          $i3vals{$k} = $e;
 269          push @fa, 'a_AMV_VF_I3VAL'
 270       }
 271       if($e->{defval}){
 272          $defvals{$k} = $e;
 273          push @fa, 'a_AMV_VF_DEFVAL'
 274       }
 275       if($e->{import}){
 276          $e->{env} = 1;
 277          push @fa, 'a_AMV_VF_IMPORT'
 278       }
 279       if($e->{nolopts}) {push @fa, 'a_AMV_VF_NOLOPTS'}
 280       if($e->{rdonly}) {push @fa, 'a_AMV_VF_RDONLY'}
 281       if($e->{nodel}) {push @fa, 'a_AMV_VF_NODEL'}
 282       if($e->{notempty}) {push @fa, 'a_AMV_VF_NOTEMPTY'}
 283       if($e->{nocntrls}) {push @fa, 'a_AMV_VF_NOCNTRLS'}
 284       if($e->{num}) {push @fa, 'a_AMV_VF_NUM'}
 285       if($e->{posnum}) {push @fa, 'a_AMV_VF_POSNUM'}
 286       if($e->{lower}) {push @fa, 'a_AMV_VF_LOWER'}
 287       if($e->{vip}) {push @fa, 'a_AMV_VF_VIP'}
 288       if($e->{env}) {push @fa, 'a_AMV_VF_ENV'}
 289       $e->{flags} = \@fa;
 290       my $f = join('|', @fa);
 291       $f = ', ' . $f if length $f;
 292       print F "${S}/* $i. [$alen]+$l $k$f */\n" if $VERB;
 293       print F "${S}'$a','\\0',\n";
 294       ++$i;
 295       $alen += $l + 1
 296    }
 297    print F '};', "\n\n";
 298
 299    print F 'n_CTA(a_AMV_VF_NONE == 0, "Value not 0 as expected");', "\n";
 300    print F 'static struct a_amv_var_map const a_amv_var_map[] = {', "\n";
 301    foreach my $e (@ENTS){
 302       my $f = $VERB ? 'a_AMV_VF_NONE' : '0';
 303       my $fa = join '|', @{$e->{flags}};
 304       $f .= '|' . $fa if length $fa;
 305       print F "${S}{$e->{hash}u, $e->{keyoff}u, $f},";
 306       if($VERB) {print F "${S}/* $e->{name} */\n"}
 307       else {print F "\n"}
 308    }
 309    print F '};', "\n\n";
 310
 311    # We have at least version stuff in here
 312    # The problem is that struct var uses a variable sized character buffer
 313    # which cannot be initialized in a conforming way :(
 314    print F <<_EOT;
 315 #ifndef __CREATE_OKEY_MAP_PL
 316 # ifdef HAVE_PUTENV
 317 #  define a_X(X) X
 318 # else
 319 #  define a_X(X)
 320 # endif
 321
 322 /* Unfortunately init of varsized buffer won't work: define "subclass"es */
 323 _EOT
 324    my @skeys = sort keys %virts;
 325
 326    foreach(@skeys){
 327       my $e = $virts{$_};
 328       $e->{vname} = $1 if $e->{enum} =~ /ok_._(.*)/;
 329       $e->{vstruct} = "var_virt_$e->{vname}";
 330       print F "static char const a_amv_$e->{vstruct}_val[] = {$e->{virt}};\n";
 331       print F "static struct{\n";
 332       print F "${S}struct a_amv_var *av_link;\n";
 333       print F "${S}char const *av_value;\n";
 334       print F "${S}a_X(char *av_env;)\n";
 335       print F "${S}ui16_t av_flags;\n";
 336       print F "${S}char const av_name[", length($e->{name}), " +1];\n";
 337       my $f = $VERB ? 'a_AMV_VF_NONE' : '0';
 338       my $fa = join '|', @{$e->{flags}};
 339       $f .= '|' . $fa if length $fa;
 340       print F "} const a_amv_$e->{vstruct} = ",
 341          "{NULL, a_amv_$e->{vstruct}_val, a_X(0 COMMA) $f, ",
 342          "\"$e->{name}\"};\n\n"
 343    }
 344    print F "# undef a_X\n";
 345
 346    print F "\n";
 347    print F '#define a_AMV_VAR_VIRTS_CNT ', scalar @skeys, "\n";
 348    print F 'static struct a_amv_var_virt const a_amv_var_virts[] = {', "\n";
 349    foreach(@skeys){
 350       my $e = $virts{$_};
 351       my $n = $1 if $e->{enum} =~ /ok_._(.*)/;
 352       print F "${S}{$e->{enum}, {0,}, (void const*)&a_amv_$e->{vstruct}},\n";
 353    }
 354    print F "};\n";
 355
 356    #
 357    @skeys = sort keys %i3vals;
 358
 359    print F "\n";
 360    print F '#define a_AMV_VAR_I3VALS_CNT ', scalar @skeys, "\n";
 361    print F 'static struct a_amv_var_defval const a_amv_var_i3vals[] = {', "\n";
 362    foreach(@skeys){
 363       my $e = $i3vals{$_};
 364       print F "${S}{", $e->{enum}, ', {0,}, ',
 365          (!$e->{bool} ? $e->{i3val} : "NULL"), "},\n"
 366    }
 367    print F "};\n";
 368
 369    #
 370    @skeys = sort keys %defvals;
 371
 372    print F "\n";
 373    print F '#define a_AMV_VAR_DEFVALS_CNT ', scalar @skeys, "\n";
 374    print F 'static struct a_amv_var_defval const a_amv_var_defvals[] = {', "\n";
 375    foreach(@skeys){
 376       my $e = $defvals{$_};
 377       print F "${S}{", $e->{enum}, ', {0,}, ',
 378          (!$e->{bool} ? $e->{defval} : "NULL"), "},\n"
 379    }
 380    print F "};\n";
 381
 382    print F "#endif /* __CREATE_OKEY_MAP_PL */\n";
 383
 384    # Special var backing [#@*?]|[1-9][0-9]*|0
 385    $i = 0;
 386    print F "\n";
 387    foreach my $e (@ENTS){
 388       if($e->{name} eq '--special-param'){
 389          print F "#define a_AMV_VAR__SPECIAL_PARAM_MAP_IDX ${i}u\n"
 390       }
 391       # The rest are only speedups
 392       elsif($e->{name} eq '?'){
 393          print F "#define a_AMV_VAR__QM_MAP_IDX ${i}u\n"
 394       }elsif($e->{name} eq '!'){
 395          print F "#define a_AMV_VAR__EM_MAP_IDX ${i}u\n"
 396       }
 397       ++$i
 398    }
 399
 400    print F "\n";
 401    die "$OUT: close: $^E" unless close F
 402 }
 403
 404 sub reverser{
 405    my $argv2 = $VERB ? ' verb' : '';
 406    system("c99 -I. -o $CTOOL_EXE $CTOOL");
 407    my $t = (@ENTS < 0xFF ? 'ui8_t' : (@ENTS < 0xFFFF ? 'ui16_t' : 'ui32_t'));
 408    `$CTOOL_EXE $t$argv2 >> $OUT`
 409 }
 410
 411 {package main; main_fun()}
 412
 413 # s-it-mode