1#!/usr/bin/perl -w 2# 3# Regenerate (overwriting only if changed): 4# 5# mg_names.inc 6# mg_raw.h 7# mg_vtable.h 8# pod/perlguts.pod 9# 10# from information stored in this file. pod/perlguts.pod is not completely 11# regenerated. Only the magic table is replaced; the other parts remain 12# untouched. 13# 14# Accepts the standard regen_lib -q and -v args. 15# 16# This script is normally invoked from regen.pl. 17# 18# Its output files contain: 19# 20# mg_names.inc 21# included by dump.c - the textual representation for each magic type. 22# Contains a list of 23# 24# { PERL_MAGIC_foo, "foo(f)" } 25# 26# pairs. 27# 28# 29# mg_raw.h 30# processed by generate_uudmap.c into mg_data.h which eventually 31# populates PL_magic_vtables[]. 32# 33# Contains a list of: 34# 35# { 'f', "want_vtbl_foo | FLAGS", "description for comments" } 36# 37# triplets. FLAGS can be: 38# PERL_MAGIC_READONLY_ACCEPTABLE 39# ok set this type of magic on an SvREADONLY() SV 40# PERL_MAGIC_VALUE_MAGIC 41# this is value magic (pos, taint etc) 42# rather than container magic (%ENV, $1 etc) 43# 44# 45# mg_vtable.h 46# This contains five kinds of entries: 47# 48# #define PERL_MAGIC_arylen '#' 49# .... 50# 51# enum { /* pass one of these to get_vtbl */ 52# want_vtbl_arylen, 53# ... 54# } 55# 56# PL_magic_vtable_names[] = { 57# "arylen", 58# ... 59# } 60# 61# PL_magic_vtables[] = { 62# /* per-magic sets of vtable function pointers */ 63# { get, set, len, clear, free, copy, dup, local }, 64# } 65# 66# define PL_vtbl_arylen PL_magic_vtables[want_vtbl_arylen] 67# .... 68# 69# 70# 71# pod/perlguts.pod 72# updates the list of magic types between 73# =for mg_vtable.pl begin 74# ... 75# =for mg_vtable.pl end 76 77 78use strict; 79require 5.004; 80 81BEGIN { 82 # Get function prototypes 83 require './regen/regen_lib.pl'; 84} 85 86# ===================================================================== 87# 88# START OF CONFIGURATION DATA 89 90 91# %mg 92# 93# This hash is mainly concerned with populating all the other stuff 94# ancillary to the vtable. 95# 96# The key is the name, e.g. 'regdata' for PERL_MAGIC_regdata 97# 98# The keys of the value hash are: 99# char 100# the magic's char identifier 101# 102# desc 103# a description which appears in code comments in generated files 104# 105# readonly_acceptable 106# If true, set PERL_MAGIC_READONLY_ACCEPTABLE flag; 107# SvREADONLY() svs are allowed to have this magic added to them 108# 109# unknown_to_sv_magic 110# if true, this isn't one of the standard magic types which 111# Perl_sv_magic() knows how to deal with 112# 113# value_magic 114# If true, set PERL_MAGIC_VALUE_MAGIC flag; 115# this kind of magic is value (pos, taint etc) rather than 116# container magic (%ENV, $1 etc) 117# 118# vtable 119# name of the vtable lookup enum, e.g. 'foo' creates want_vtbl_foo 120 121my %mg = 122 ( 123 sv => { char => "\0", vtable => 'sv', readonly_acceptable => 1, 124 desc => 'Special scalar variable' }, 125 # overload, or type "A" magic, used to be here. Hence overloaded is 126 # often called AMAGIC internally, even though it does not use "A" 127 # magic any more. 128 overload_table => { char => 'c', vtable => 'ovrld', 129 desc => 'Holds overload table (AMT) on stash' }, 130 bm => { char => 'B', vtable => 'regexp', value_magic => 1, 131 readonly_acceptable => 1, 132 desc => 'Boyer-Moore (fast string search)' }, 133 regdata => { char => 'D', vtable => 'regdata', 134 desc => "Regex match position data\n(\@+ and \@- vars)" }, 135 regdatum => { char => 'd', vtable => 'regdatum', 136 desc => 'Regex match position data element' }, 137 env => { char => 'E', vtable => 'env', desc => '%ENV hash' }, 138 envelem => { char => 'e', vtable => 'envelem', 139 desc => '%ENV hash element' }, 140 fm => { char => 'f', vtable => 'regexp', value_magic => 1, 141 readonly_acceptable => 1, desc => "Formline ('compiled' format)" }, 142 regex_global => { char => 'g', vtable => 'mglob', value_magic => 1, 143 readonly_acceptable => 1, desc => 'm//g target' }, 144 hints => { char => 'H', vtable => 'hints', desc => '%^H hash' }, 145 hintselem => { char => 'h', vtable => 'hintselem', 146 desc => '%^H hash element' }, 147 isa => { char => 'I', vtable => 'isa', desc => '@ISA array' }, 148 isaelem => { char => 'i', vtable => 'isaelem', 149 desc => '@ISA array element' }, 150 nkeys => { char => 'k', vtable => 'nkeys', value_magic => 1, 151 desc => 'scalar(keys()) lvalue' }, 152 dbfile => { char => 'L', 153 desc => 'Debugger %_<filename' }, 154 dbline => { char => 'l', vtable => 'dbline', 155 desc => 'Debugger %_<filename element' }, 156 shared => { char => 'N', desc => 'Shared between threads', 157 unknown_to_sv_magic => 1 }, 158 shared_scalar => { char => 'n', desc => 'Shared between threads', 159 unknown_to_sv_magic => 1 }, 160 collxfrm => { char => 'o', vtable => 'collxfrm', value_magic => 1, 161 desc => 'Locale transformation' }, 162 tied => { char => 'P', vtable => 'pack', 163 value_magic => 1, # treat as value, so 'local @tied' isn't tied 164 desc => 'Tied array or hash' }, 165 tiedelem => { char => 'p', vtable => 'packelem', 166 desc => 'Tied array or hash element' }, 167 tiedscalar => { char => 'q', vtable => 'packelem', 168 desc => 'Tied scalar or handle' }, 169 qr => { char => 'r', vtable => 'regexp', value_magic => 1, 170 readonly_acceptable => 1, desc => 'Precompiled qr// regex' }, 171 172 hook => { char => 'Z', 173 vtable => 'hook', desc => '%{^HOOK} hash' }, 174 hookelem => { char => 'z', 175 vtable => 'hookelem', desc => '%{^HOOK} hash element' }, 176 177 sig => { char => 'S', vtable => 'sig', 178 desc => '%SIG hash' }, 179 sigelem => { char => 's', vtable => 'sigelem', 180 desc => '%SIG hash element' }, 181 182 taint => { char => 't', vtable => 'taint', value_magic => 1, 183 desc => 'Taintedness' }, 184 uvar => { char => 'U', vtable => 'uvar', 185 desc => 'Available for use by extensions' }, 186 uvar_elem => { char => 'u', desc => 'Reserved for use by extensions', 187 unknown_to_sv_magic => 1 }, 188 vec => { char => 'v', vtable => 'vec', value_magic => 1, 189 desc => 'vec() lvalue' }, 190 vstring => { char => 'V', value_magic => 1, 191 desc => 'SV was vstring literal' }, 192 utf8 => { char => 'w', vtable => 'utf8', value_magic => 1, 193 desc => 'Cached UTF-8 information' }, 194 substr => { char => 'x', vtable => 'substr', value_magic => 1, 195 desc => 'substr() lvalue' }, 196 defelem => { char => 'y', vtable => 'defelem', value_magic => 1, 197 desc => "Shadow \"foreach\" iterator variable /\nsmart parameter vivification" }, 198 nonelem => { char => 'Y', vtable => 'nonelem', value_magic => 1, 199 desc => "Array element that does not exist" }, 200 arylen => { char => '#', vtable => 'arylen', value_magic => 1, 201 desc => 'Array length ($#ary)' }, 202 pos => { char => '.', vtable => 'pos', value_magic => 1, 203 desc => 'pos() lvalue' }, 204 backref => { char => '<', vtable => 'backref', value_magic => 1, 205 readonly_acceptable => 1, desc => 'For weak ref data' }, 206 symtab => { char => ':', value_magic => 1, 207 desc => 'Extra data for symbol tables' }, 208 rhash => { char => '%', value_magic => 1, 209 desc => 'Extra data for restricted hashes' }, 210 arylen_p => { char => '@', value_magic => 1, 211 desc => 'To move arylen out of XPVAV' }, 212 ext => { char => '~', desc => 'Variable magic available for use by extensions', 213 readonly_acceptable => 1 }, 214 extvalue => { char => '^', desc => 'Value magic available for use by extensions', 215 readonly_acceptable => 1, value_magic => 1 }, 216 checkcall => { char => ']', value_magic => 1, vtable => 'checkcall', 217 desc => 'Inlining/mutation of call to this CV'}, 218 debugvar => { char => '*', desc => '$DB::single, signal, trace vars', 219 vtable => 'debugvar' }, 220 lvref => { char => '\\', vtable => 'lvref', 221 desc => "Lvalue reference constructor" }, 222 destruct => { 223 char => "X", 224 vtable => 'destruct', 225 desc => "destruct callback", 226 value_magic => 1, 227 }, 228); 229 230 231# %vtable_conf 232# 233# This hash is mainly concerned with populating the vtable. 234# 235# These have a subtly different "namespace" from the magic types. 236# 237# The key is the name, e.g. 'regdata' for PERL_MAGIC_regdata 238# The keys of the value hash are: 239# 240# alias 241# for each entry in the anon array, add 242# add "#define want_vtbl_$_ want_vtbl_$name" 243# 244# cond 245# prefix the vtable with the specified entry (e.g. '#ifdef FOO') 246# and suffix it with '#else { 0, 0, 0, 0, 0, 0, 0, 0 } #endif' 247# 248# const 249# special-case cast a 'get' function whose signature expects 250# a pointer to constant magic, so that it can be added to a vtable 251# which expects pointers to functions without the 'const'. 252# 253# get 254# set 255# len 256# clear 257# free 258# copy 259# dup 260# local 261# For each specified method, add a vtable function pointer 262# of the form "Perl_magic_$vtable_conf{foo}{get}" etc 263 264my %vtable_conf = 265 ( 266 'sv' => {get => 'get', set => 'set'}, 267 'env' => {set => 'set_all_env', clear => 'clear_all_env'}, 268 'envelem' => {set => 'setenv', clear => 'clearenv'}, 269 'sig' => { set => 'setsigall' }, 270 'sigelem' => {get => 'getsig', set => 'setsig', clear => 'clearsig'}, 271 272 'hook' => { set => 'sethookall', clear => 'clearhookall' }, 273 'hookelem' => {set => 'sethook', clear => 'clearhook'}, 274 275 'pack' => {len => 'sizepack', clear => 'wipepack'}, 276 'packelem' => {get => 'getpack', set => 'setpack', clear => 'clearpack'}, 277 'dbline' => {set => 'setdbline'}, 278 'isa' => {set => 'setisa', clear => 'clearisa'}, 279 'isaelem' => {set => 'setisa'}, 280 'arylen' => {get => 'getarylen', set => 'setarylen', const => 1}, 281 'arylen_p' => {clear => 'cleararylen_p', free => 'freearylen_p'}, 282 'mglob' => {set => 'setmglob', 283 free => 'freemglob' }, 284 'nkeys' => {get => 'getnkeys', set => 'setnkeys'}, 285 'taint' => {get => 'gettaint', set => 'settaint'}, 286 'substr' => {get => 'getsubstr', set => 'setsubstr'}, 287 'vec' => {get => 'getvec', set => 'setvec'}, 288 'pos' => {get => 'getpos', set => 'setpos'}, 289 'uvar' => {get => 'getuvar', set => 'setuvar'}, 290 'defelem' => {get => 'getdefelem', set => 'setdefelem'}, 291 'nonelem' => {set => 'setnonelem'}, 292 'regexp' => {set => 'setregexp', alias => [qw(bm fm)]}, 293 'regdata' => {len => 'regdata_cnt'}, 294 'regdatum' => {get => 'regdatum_get', set => 'regdatum_set'}, 295 'backref' => {free => 'killbackrefs'}, 296 'ovrld' => {free => 'freeovrld'}, 297 'utf8' => {set => 'setutf8', 298 free => 'freeutf8' }, 299 'collxfrm' => {set => 'setcollxfrm', 300 free => 'freecollxfrm', 301 cond => '#ifdef USE_LOCALE_COLLATE'}, 302 'hintselem' => {set => 'sethint', clear => 'clearhint'}, 303 'hints' => {clear => 'clearhints'}, 304 'checkcall' => {copy => 'copycallchecker'}, 305 'debugvar' => { set => 'setdebugvar', get => 'getdebugvar' }, 306 'lvref' => {set => 'setlvref'}, 307 'destruct' => {free => 'freedestruct'}, 308); 309 310 311# END OF CONFIGURATION DATA 312# 313# ===================================================================== 314 315 316 317my ($vt, $raw, $names) = map { 318 open_new($_, '>', 319 { by => 'regen/mg_vtable.pl', file => $_, style => '*' }); 320} 'mg_vtable.h', 'mg_raw.h', 'mg_names.inc'; 321my $guts = open_new("pod/perlguts.pod", ">"); 322 323print $vt <<'EOH'; 324/* These constants should be used in preference to raw characters 325 * when using magic. Note that some perl guts still assume 326 * certain character properties of these constants, namely that 327 * isUPPER() and toLOWER() may do useful mappings. 328 */ 329 330EOH 331 332# Of course, it would be *much* easier if we could output this table directly 333# here and now. However, for our sins, we try to support EBCDIC, which wouldn't 334# be *so* bad, except that there are (at least) 3 EBCDIC charset variants, and 335# they don't agree on the code point for '~'. Which we use. Great. 336# So we have to get the local build runtime to sort our table in character order 337# (And of course, just to be helpful, in POSIX BC '~' is \xFF, so we can't even 338# simplify the C code by assuming that the last element of the array is 339# predictable) 340 341# Process %mg 342 343{ 344 my $longest = 0; 345 foreach (keys %mg) { 346 $longest = length $_ if length $_ > $longest; 347 } 348 349 my $longest_p1 = $longest + 1; 350 351 my %mg_order; 352 while (my ($name, $data) = each %mg) { 353 my $byte = $data->{char}; 354 if ($byte =~ /[[:print:]]/) { 355 $data->{r_char} = $byte; # readable char 356 ($data->{c_char} = $byte) =~ s/([\\"])/\\$1/g; # for C strings 357 } 358 else { 359 $data->{c_char} = $data->{r_char} = '\\'.ord $byte; 360 } 361 $mg_order{(uc $byte) . $byte} = $name; 362 } 363 364 my @rows; 365 my @names; 366 foreach (sort keys %mg_order) { 367 my $name = $mg_order{$_}; 368 push @names, $name; 369 my $data = $mg{$name}; 370 my $i = ord $data->{char}; 371 372 # add entry to mg_raw.h 373 374 unless ($data->{unknown_to_sv_magic}) { 375 my $value = $data->{vtable} 376 ? "want_vtbl_$data->{vtable}" : 'magic_vtable_max'; 377 $value .= ' | PERL_MAGIC_READONLY_ACCEPTABLE' 378 if $data->{readonly_acceptable}; 379 $value .= ' | PERL_MAGIC_VALUE_MAGIC' if $data->{value_magic}; 380 my $comment = "/* $name '$data->{r_char}' $data->{desc} */"; 381 $comment =~ s/([\\"])/\\$1/g; 382 $comment =~ tr/\n/ /; 383 print $raw qq{ { '$data->{c_char}', "$value",\n "$comment" },\n}; 384 } 385 386 # add #define PERL_MAGIC_foo entry to vt_table.h 387 388 my $comment = $data->{desc}; 389 my $leader = ' ' x ($longest + 27); 390 $comment =~ s/\n/\n$leader/s; 391 printf $vt "#define PERL_MAGIC_%-${longest}s '%s' /* %s */\n", 392 $name, $data->{c_char}, $comment; 393 394 # add entry to mg_names.inc 395 396 my $char = $data->{r_char}; 397 $char =~ s/([\\"])/\\$1/g; 398 printf $names qq[\t{ PERL_MAGIC_%-${longest_p1}s "%s(%s)" },\n], 399 "$name,", $name, $char; 400 401 # construct perlguts.pod entry 402 403 push @rows, [(sprintf "%-2s PERL_MAGIC_%s", $data->{r_char},$name), 404 $data->{vtable} ? "vtbl_$data->{vtable}" : '(none)', 405 $data->{desc}]; 406 } 407 408 # output @rows to perlguts.pod 409 410 select +(select($guts), do { 411 my @header = ('(old-style char and macro)', 'MGVTBL', 'Type of magic'); 412 my @widths = (0, 0); 413 foreach my $row (@rows) { 414 for (0, 1) { 415 $widths[$_] = length $row->[$_] 416 if length $row->[$_] > $widths[$_]; 417 } 418 } 419 my $indent = ' '; 420 my $format 421 = sprintf "$indent%%-%ds%%-%ds%%s\n", $widths[0] + 1, $widths[1] + 1; 422 my $desc_wrap = 423 79 - 7 - (length $indent) - $widths[0] - $widths[1] - 2; 424 425 open my $oldguts, "<", "pod/perlguts.pod" 426 or die "$0 cannot open pod/perlguts.pod for reading: $!"; 427 while (<$oldguts>) { 428 print; 429 last if /^=for mg_vtable.pl begin/ 430 } 431 432 print "\n", $indent . "mg_type\n"; 433 printf $format, @header; 434 printf $format, map {'-' x length $_} @header; 435 foreach (@rows) { 436 my ($type, $vtbl, $desc) = @$_; 437 $desc =~ tr/\n/ /; 438 my @cont; 439 if (length $desc > $desc_wrap) { 440 # If it's too long, first split on '(', if there. 441 # [Which, if there, is always short enough, currently. 442 # Make this more robust if that changes] 443 ($desc, @cont) = split /(?=\()/, $desc; 444 if (!@cont) { 445 ($desc, @cont) = $desc =~ /(.{1,$desc_wrap})(?: |\z)/g 446 } 447 } 448 s/\s+\z// for $desc, @cont; 449 printf $format, $type, $vtbl, $desc; 450 printf $format, '', '', $_ foreach @cont; 451 } 452 print "\n\n"; 453 454 my $first = 1; 455 for my $magic (sort @names) { 456 if ($first) { 457 $first = 0; 458 print "=for apidoc_section \$magic\n"; 459 print "=for apidoc AmnhU||PERL_MAGIC_$magic\n"; 460 } 461 else { 462 print "=for apidoc_item ||PERL_MAGIC_$magic\n"; 463 } 464 } 465 print "\n"; 466 467 while (<$oldguts>) { 468 last if /^=for mg_vtable.pl end/; 469 } 470 do { print } while <$oldguts>; 471 })[0]; 472} 473 474 475# Process %vtable_conf - everything goes to mg_vtable.h 476 477my @names = sort keys %vtable_conf; 478{ 479 my $want = join ",\n ", (map {"want_vtbl_$_"} @names), 'magic_vtable_max'; 480 my $names = join qq{",\n "}, @names; 481 482 print $vt <<"EOH"; 483 484enum { /* pass one of these to get_vtbl */ 485 $want 486}; 487 488#ifdef DOINIT 489EXTCONST char * const PL_magic_vtable_names[magic_vtable_max] = { 490 "$names" 491}; 492#else 493EXTCONST char * const PL_magic_vtable_names[magic_vtable_max]; 494#endif 495 496EOH 497} 498 499print $vt <<'EOH'; 500/* These all need to be 0, not NULL, as NULL can be (void*)0, which is a 501 * pointer to data, whereas we're assigning pointers to functions, which are 502 * not the same beast. ANSI doesn't allow the assignment from one to the other. 503 * (although most, but not all, compilers are prepared to do it) 504 */ 505 506/* order is: 507 get 508 set 509 len 510 clear 511 free 512 copy 513 dup 514 local 515*/ 516 517#ifdef DOINIT 518EXT_MGVTBL PL_magic_vtables[magic_vtable_max] = { 519EOH 520 521my @vtable_names; 522my @aliases; 523 524while (my $name = shift @names) { 525 my $data = $vtable_conf{$name}; 526 push @vtable_names, $name; 527 my @funcs = map { 528 $data->{$_} ? "Perl_magic_$data->{$_}" : 0; 529 } qw(get set len clear free copy dup local); 530 531 $funcs[0] = "(int (*)(pTHX_ SV *, MAGIC *))" . $funcs[0] if $data->{const}; 532 my $funcs = join ", ", @funcs; 533 534 # Because we can't have a , after the last {...} 535 my $comma = @names ? ',' : ''; 536 537 print $vt "$data->{cond}\n" if $data->{cond}; 538 print $vt " { $funcs }$comma\n"; 539 print $vt <<"EOH" if $data->{cond}; 540#else 541 { 0, 0, 0, 0, 0, 0, 0, 0 }$comma 542#endif 543EOH 544 foreach(@{$data->{alias}}) { 545 push @aliases, "#define want_vtbl_$_ want_vtbl_$name\n"; 546 push @vtable_names, $_; 547 } 548} 549 550print $vt <<'EOH'; 551}; 552#else 553EXT_MGVTBL PL_magic_vtables[magic_vtable_max]; 554#endif 555 556EOH 557 558print $vt (sort @aliases), "\n"; 559 560print $vt "#define PL_vtbl_$_ PL_magic_vtables[want_vtbl_$_]\n" 561 foreach sort @vtable_names; 562 563# 63, not 64, As we rely on the last possible value to mean "NULL vtable" 564die "Too many vtable names" if @vtable_names > 63; 565 566read_only_bottom_close_and_rename($_) foreach $vt, $raw, $names; 567 close_and_rename($guts); 568