This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Sort magic and magic vtable names in files generated by mg_vtable.pl
[perl5.git] / regen / regcomp.pl
CommitLineData
916e4025 1#!/usr/bin/perl -w
6294c161
DM
2#
3# Regenerate (overwriting only if changed):
4#
5# regnodes.h
6#
7# from information stored in
8#
9# regcomp.sym
10# regexp.h
11#
12# Accepts the standard regen_lib -q and -v args.
13#
14# This script is normally invoked from regen.pl.
15
36bb303b
NC
16BEGIN {
17 # Get function prototypes
af001346 18 require 'regen/regen_lib.pl';
36bb303b 19}
03363afd 20use strict;
03363afd 21
d09b2d29 22open DESC, 'regcomp.sym';
d09b2d29 23
03363afd 24my $ind = 0;
f9ef50a7 25my (@name,@rest,@type,@code,@args,@flags,@longj);
03363afd 26my ($desc,$lastregop);
d09b2d29 27while (<DESC>) {
03363afd
YO
28 s/#.*$//;
29 next if /^\s*$/;
916e4025
NC
30 chomp; # No \z in 5.004
31 s/\s*$//;
03363afd
YO
32 if (/^-+\s*$/) {
33 $lastregop= $ind;
34 next;
35 }
36 unless ($lastregop) {
f8abb37e 37 ($name[$ind], $desc, $rest[$ind]) = /^(\S+)\s+([^\t]+)\s*;\s*(.*)/;
f9ef50a7
NC
38 ($type[$ind], $code[$ind], $args[$ind], $flags[$ind], $longj[$ind])
39 = split /[,\s]\s*/, $desc;
93882df0 40 ++$ind;
03363afd 41 } else {
f8abb37e 42 my ($type,@lists)=split /\s+/, $_;
03363afd
YO
43 die "No list? $type" if !@lists;
44 foreach my $list (@lists) {
45 my ($names,$special)=split /:/, $list , 2;
46 $special ||= "";
47 foreach my $name (split /,/,$names) {
48 my $real= $name eq 'resume'
49 ? "resume_$type"
50 : "${type}_$name";
51 my @suffix;
52 if (!$special) {
53 @suffix=("");
54 } elsif ($special=~/\d/) {
55 @suffix=(1..$special);
56 } elsif ($special eq 'FAIL') {
57 @suffix=("","_fail");
58 } else {
59 die "unknown :type ':$special'";
60 }
61 foreach my $suffix (@suffix) {
03363afd
YO
62 $name[$ind]="$real$suffix";
63 $type[$ind]=$type;
24b23f37 64 $rest[$ind]="state for $type";
93882df0 65 ++$ind;
03363afd
YO
66 }
67 }
68 }
69
70 }
71}
5d458dd8
YO
72# use fixed width to keep the diffs between regcomp.pl recompiles
73# as small as possible.
74my ($width,$rwidth,$twidth)=(22,12,9);
03363afd
YO
75$lastregop ||= $ind;
76my $tot = $ind;
d09b2d29 77close DESC;
03363afd
YO
78die "Too many regexp/state opcodes! Maximum is 256, but there are $lastregop in file!"
79 if $lastregop>256;
d09b2d29 80
f9ef50a7
NC
81sub process_flags {
82 my ($flag, $varname, $comment) = @_;
83 $comment = '' unless defined $comment;
84
85 $ind = 0;
86 my @selected;
ded4dd2a 87 my $bitmap = '';
93882df0 88 do {
ded4dd2a
NC
89 my $set = $flags[$ind] && $flags[$ind] eq $flag ? 1 : 0;
90 # Whilst I could do this with vec, I'd prefer to do longhand the arithmetic
91 # ops in the C code.
92 my $current = do {
916e4025 93 local $^W;
ded4dd2a 94 ord do {
ded4dd2a
NC
95 substr $bitmap, ($ind >> 3);
96 }
97 };
916e4025 98 substr($bitmap, ($ind >> 3), 1) = chr($current | ($set << ($ind & 7)));
ded4dd2a
NC
99
100 push @selected, $name[$ind] if $set;
93882df0 101 } while (++$ind < $lastregop);
f9ef50a7
NC
102 my $out_string = join ', ', @selected, 0;
103 $out_string =~ s/(.{1,70},) /$1\n /g;
ded4dd2a
NC
104
105 my $out_mask = join ', ', map {sprintf "0x%02X", ord $_} split '', $bitmap;
106
f9ef50a7 107 return $comment . <<"EOP";
ded4dd2a 108#define REGNODE_\U$varname\E(node) (PL_${varname}_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 109
f9ef50a7 110#ifndef DOINIT
916e4025 111EXTCONST U8 PL_${varname}\[] __attribute__deprecated__;
f9ef50a7 112#else
916e4025 113EXTCONST U8 PL_${varname}\[] __attribute__deprecated__ = {
f9ef50a7
NC
114 $out_string
115};
116#endif /* DOINIT */
117
ded4dd2a
NC
118#ifndef DOINIT
119EXTCONST U8 PL_${varname}_bitmask[];
120#else
121EXTCONST U8 PL_${varname}_bitmask[] = {
122 $out_mask
123};
124#endif /* DOINIT */
f9ef50a7
NC
125EOP
126}
127
cc49830d
NC
128my $out = open_new('regnodes.h', '>',
129 { by => 'regen/regcomp.pl', from => 'regcomp.sym' });
424a4936 130printf $out <<EOP,
6bda09f9
YO
131/* Regops and State definitions */
132
03363afd
YO
133#define %*s\t%d
134#define %*s\t%d
135
d09b2d29 136EOP
f9f4320a
YO
137 -$width, REGNODE_MAX => $lastregop - 1,
138 -$width, REGMATCH_STATE_MAX => $tot - 1
139;
d09b2d29 140
24b23f37 141
93882df0 142for ($ind=0; $ind < $lastregop ; ++$ind) {
424a4936 143 printf $out "#define\t%*s\t%d\t/* %#04x %s */\n",
93882df0 144 -$width, $name[$ind], $ind, $ind, $rest[$ind];
24b23f37 145}
424a4936 146print $out "\t/* ------------ States ------------- */\n";
93882df0 147for ( ; $ind < $tot ; $ind++) {
424a4936 148 printf $out "#define\t%*s\t(REGNODE_MAX + %d)\t/* %s */\n",
93882df0 149 -$width, $name[$ind], $ind - $lastregop + 1, $rest[$ind];
d09b2d29
IZ
150}
151
424a4936 152print $out <<EOP;
03363afd 153
6bda09f9 154/* PL_regkind[] What type of regop or state is this. */
d09b2d29
IZ
155
156#ifndef DOINIT
22c35a8c 157EXTCONST U8 PL_regkind[];
d09b2d29 158#else
22c35a8c 159EXTCONST U8 PL_regkind[] = {
d09b2d29
IZ
160EOP
161
162$ind = 0;
93882df0 163do {
424a4936 164 printf $out "\t%*s\t/* %*s */\n",
03363afd 165 -1-$twidth, "$type[$ind],", -$width, $name[$ind];
424a4936 166 print $out "\t/* ------------ States ------------- */\n"
93882df0
NC
167 if $ind + 1 == $lastregop and $lastregop != $tot;
168} while (++$ind < $tot);
d09b2d29 169
424a4936 170print $out <<EOP;
d09b2d29
IZ
171};
172#endif
173
6bda09f9 174/* regarglen[] - How large is the argument part of the node (in regnodes) */
d09b2d29
IZ
175
176#ifdef REG_COMP_C
29de9391 177static const U8 regarglen[] = {
d09b2d29
IZ
178EOP
179
180$ind = 0;
93882df0 181do {
03363afd 182 my $size = 0;
d09b2d29
IZ
183 $size = "EXTRA_SIZE(struct regnode_$args[$ind])" if $args[$ind];
184
424a4936 185 printf $out "\t%*s\t/* %*s */\n",
03363afd 186 -37, "$size,",-$rwidth,$name[$ind];
93882df0 187} while (++$ind < $lastregop);
d09b2d29 188
424a4936 189print $out <<EOP;
d09b2d29
IZ
190};
191
6bda09f9
YO
192/* reg_off_by_arg[] - Which argument holds the offset to the next node */
193
29de9391 194static const char reg_off_by_arg[] = {
d09b2d29
IZ
195EOP
196
197$ind = 0;
93882df0 198do {
03363afd 199 my $size = $longj[$ind] || 0;
9b155405 200
424a4936 201 printf $out "\t%d,\t/* %*s */\n",
03363afd 202 $size, -$rwidth, $name[$ind]
93882df0 203} while (++$ind < $lastregop);
d09b2d29 204
424a4936 205print $out <<EOP;
d09b2d29 206};
9b155405 207
13d6edb4
NC
208#endif /* REG_COMP_C */
209
6bda09f9
YO
210/* reg_name[] - Opcode/state names in string form, for debugging */
211
22429478 212#ifndef DOINIT
13d6edb4 213EXTCONST char * PL_reg_name[];
22429478 214#else
4764e399 215EXTCONST char * const PL_reg_name[] = {
9b155405
IZ
216EOP
217
218$ind = 0;
93882df0 219my $ofs = 0;
24b23f37 220my $sym = "";
93882df0 221do {
03363afd 222 my $size = $longj[$ind] || 0;
9b155405 223
424a4936 224 printf $out "\t%*s\t/* $sym%#04x */\n",
24b23f37 225 -3-$width,qq("$name[$ind]",), $ind - $ofs;
93882df0 226 if ($ind + 1 == $lastregop and $lastregop != $tot) {
424a4936 227 print $out "\t/* ------------ States ------------- */\n";
93882df0 228 $ofs = $lastregop - 1;
24b23f37
YO
229 $sym = 'REGNODE_MAX +';
230 }
231
93882df0 232} while (++$ind < $tot);
9b155405 233
424a4936 234print $out <<EOP;
9b155405 235};
22429478 236#endif /* DOINIT */
d09b2d29 237
f7819f85
A
238/* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
239
240#ifndef DOINIT
241EXTCONST char * PL_reg_extflags_name[];
242#else
243EXTCONST char * const PL_reg_extflags_name[] = {
d09b2d29
IZ
244EOP
245
f7819f85 246my %rxfv;
6a080ccd 247my %definitions; # Remember what the symbol definitions are
c8e4cf8b
NC
248my $val = 0;
249my %reverse;
1850c8f9 250foreach my $file ("op_reg_common.h", "regexp.h") {
916e4025
NC
251 open FH,"<$file" or die "Can't read $file: $!";
252 while (<FH>) {
1850c8f9
KW
253
254 # optional leading '_'. Return symbol in $1, and strip it from
255 # rest of line
256 if (s/ \#define \s+ ( _? RXf_ \w+ ) \s+ //xi) {
257 chomp;
258 my $define = $1;
259 s: / \s* \* .*? \* \s* / : :x; # Replace comments by a blank
260
261 # Replace any prior defined symbols by their values
262 foreach my $key (keys %definitions) {
263 s/\b$key\b/$definitions{$key}/g;
264 }
265 my $newval = eval $_; # Get numeric definition
6a080ccd 266
1850c8f9 267 $definitions{$define} = $newval;
6a080ccd 268
1850c8f9
KW
269 next unless $_ =~ /<</; # Bit defines use left shift
270 if($val & $newval) {
271 die sprintf "Both $define and $reverse{$newval} use %08X", $newval;
272 }
273 $val|=$newval;
274 $rxfv{$define}= $newval;
275 $reverse{$newval} = $define;
6a080ccd 276 }
f7819f85 277 }
1850c8f9 278}
f7819f85 279my %vrxf=reverse %rxfv;
916e4025 280printf $out "\t/* Bits in extflags defined: %s */\n", unpack 'B*', pack 'N', $val;
f7819f85 281for (0..31) {
5458d9a0
KW
282 my $power_of_2 = 2**$_;
283 my $n=$vrxf{$power_of_2};
284 if (! $n) {
285
286 # Here, there was no name that matched exactly the bit. It could be
287 # either that it is unused, or the name matches multiple bits.
288 if (! ($val & $power_of_2)) {
289 $n = "UNUSED_BIT_$_";
290 }
291 else {
292
293 # Here, must be because it matches multiple bits. Look through
294 # all possibilities until find one that matches this one. Use
295 # that name, and all the bits it matches
296 foreach my $name (keys %rxfv) {
297 if ($rxfv{$name} & $power_of_2) {
298 $n = $name;
299 $power_of_2 = $rxfv{$name};
300 last;
301 }
302 }
303 }
304 }
f7819f85 305 $n=~s/^RXf_(PMf_)?//;
424a4936 306 printf $out qq(\t%-20s/* 0x%08x */\n),
5458d9a0 307 qq("$n",),$power_of_2;
f7819f85
A
308}
309
424a4936 310print $out <<EOP;
f7819f85
A
311};
312#endif /* DOINIT */
313
f9ef50a7
NC
314EOP
315
316print $out process_flags('V', 'varies', <<'EOC');
317/* The following have no fixed length. U8 so we can do strchr() on it. */
318EOC
319
320print $out process_flags('S', 'simple', <<'EOC');
ce716c52 321
f9ef50a7
NC
322/* The following always have a length of 1. U8 we can do strchr() on it. */
323/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
324EOC
325
ce716c52 326read_only_bottom_close_and_rename($out);