This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
l1_char_class_tab.h: Remove multi-char fold targets
[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
f038801a 128my $out = safer_open('regnodes.h-new', 'regnodes.h');
d09b2d29 129
78102347
NC
130print $out read_only_top(lang => 'C', by => 'regen/regcomp.pl',
131 from => 'regcomp.sym');
424a4936 132printf $out <<EOP,
6bda09f9
YO
133/* Regops and State definitions */
134
03363afd
YO
135#define %*s\t%d
136#define %*s\t%d
137
d09b2d29 138EOP
f9f4320a
YO
139 -$width, REGNODE_MAX => $lastregop - 1,
140 -$width, REGMATCH_STATE_MAX => $tot - 1
141;
d09b2d29 142
24b23f37 143
93882df0 144for ($ind=0; $ind < $lastregop ; ++$ind) {
424a4936 145 printf $out "#define\t%*s\t%d\t/* %#04x %s */\n",
93882df0 146 -$width, $name[$ind], $ind, $ind, $rest[$ind];
24b23f37 147}
424a4936 148print $out "\t/* ------------ States ------------- */\n";
93882df0 149for ( ; $ind < $tot ; $ind++) {
424a4936 150 printf $out "#define\t%*s\t(REGNODE_MAX + %d)\t/* %s */\n",
93882df0 151 -$width, $name[$ind], $ind - $lastregop + 1, $rest[$ind];
d09b2d29
IZ
152}
153
424a4936 154print $out <<EOP;
03363afd 155
6bda09f9 156/* PL_regkind[] What type of regop or state is this. */
d09b2d29
IZ
157
158#ifndef DOINIT
22c35a8c 159EXTCONST U8 PL_regkind[];
d09b2d29 160#else
22c35a8c 161EXTCONST U8 PL_regkind[] = {
d09b2d29
IZ
162EOP
163
164$ind = 0;
93882df0 165do {
424a4936 166 printf $out "\t%*s\t/* %*s */\n",
03363afd 167 -1-$twidth, "$type[$ind],", -$width, $name[$ind];
424a4936 168 print $out "\t/* ------------ States ------------- */\n"
93882df0
NC
169 if $ind + 1 == $lastregop and $lastregop != $tot;
170} while (++$ind < $tot);
d09b2d29 171
424a4936 172print $out <<EOP;
d09b2d29
IZ
173};
174#endif
175
6bda09f9 176/* regarglen[] - How large is the argument part of the node (in regnodes) */
d09b2d29
IZ
177
178#ifdef REG_COMP_C
29de9391 179static const U8 regarglen[] = {
d09b2d29
IZ
180EOP
181
182$ind = 0;
93882df0 183do {
03363afd 184 my $size = 0;
d09b2d29
IZ
185 $size = "EXTRA_SIZE(struct regnode_$args[$ind])" if $args[$ind];
186
424a4936 187 printf $out "\t%*s\t/* %*s */\n",
03363afd 188 -37, "$size,",-$rwidth,$name[$ind];
93882df0 189} while (++$ind < $lastregop);
d09b2d29 190
424a4936 191print $out <<EOP;
d09b2d29
IZ
192};
193
6bda09f9
YO
194/* reg_off_by_arg[] - Which argument holds the offset to the next node */
195
29de9391 196static const char reg_off_by_arg[] = {
d09b2d29
IZ
197EOP
198
199$ind = 0;
93882df0 200do {
03363afd 201 my $size = $longj[$ind] || 0;
9b155405 202
424a4936 203 printf $out "\t%d,\t/* %*s */\n",
03363afd 204 $size, -$rwidth, $name[$ind]
93882df0 205} while (++$ind < $lastregop);
d09b2d29 206
424a4936 207print $out <<EOP;
d09b2d29 208};
9b155405 209
13d6edb4
NC
210#endif /* REG_COMP_C */
211
6bda09f9
YO
212/* reg_name[] - Opcode/state names in string form, for debugging */
213
22429478 214#ifndef DOINIT
13d6edb4 215EXTCONST char * PL_reg_name[];
22429478 216#else
4764e399 217EXTCONST char * const PL_reg_name[] = {
9b155405
IZ
218EOP
219
220$ind = 0;
93882df0 221my $ofs = 0;
24b23f37 222my $sym = "";
93882df0 223do {
03363afd 224 my $size = $longj[$ind] || 0;
9b155405 225
424a4936 226 printf $out "\t%*s\t/* $sym%#04x */\n",
24b23f37 227 -3-$width,qq("$name[$ind]",), $ind - $ofs;
93882df0 228 if ($ind + 1 == $lastregop and $lastregop != $tot) {
424a4936 229 print $out "\t/* ------------ States ------------- */\n";
93882df0 230 $ofs = $lastregop - 1;
24b23f37
YO
231 $sym = 'REGNODE_MAX +';
232 }
233
93882df0 234} while (++$ind < $tot);
9b155405 235
424a4936 236print $out <<EOP;
9b155405 237};
22429478 238#endif /* DOINIT */
d09b2d29 239
f7819f85
A
240/* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
241
242#ifndef DOINIT
243EXTCONST char * PL_reg_extflags_name[];
244#else
245EXTCONST char * const PL_reg_extflags_name[] = {
d09b2d29
IZ
246EOP
247
f7819f85 248my %rxfv;
6a080ccd 249my %definitions; # Remember what the symbol definitions are
c8e4cf8b
NC
250my $val = 0;
251my %reverse;
1850c8f9 252foreach my $file ("op_reg_common.h", "regexp.h") {
916e4025
NC
253 open FH,"<$file" or die "Can't read $file: $!";
254 while (<FH>) {
1850c8f9
KW
255
256 # optional leading '_'. Return symbol in $1, and strip it from
257 # rest of line
258 if (s/ \#define \s+ ( _? RXf_ \w+ ) \s+ //xi) {
259 chomp;
260 my $define = $1;
261 s: / \s* \* .*? \* \s* / : :x; # Replace comments by a blank
262
263 # Replace any prior defined symbols by their values
264 foreach my $key (keys %definitions) {
265 s/\b$key\b/$definitions{$key}/g;
266 }
267 my $newval = eval $_; # Get numeric definition
6a080ccd 268
1850c8f9 269 $definitions{$define} = $newval;
6a080ccd 270
1850c8f9
KW
271 next unless $_ =~ /<</; # Bit defines use left shift
272 if($val & $newval) {
273 die sprintf "Both $define and $reverse{$newval} use %08X", $newval;
274 }
275 $val|=$newval;
276 $rxfv{$define}= $newval;
277 $reverse{$newval} = $define;
6a080ccd 278 }
f7819f85 279 }
1850c8f9 280}
f7819f85 281my %vrxf=reverse %rxfv;
916e4025 282printf $out "\t/* Bits in extflags defined: %s */\n", unpack 'B*', pack 'N', $val;
f7819f85 283for (0..31) {
5458d9a0
KW
284 my $power_of_2 = 2**$_;
285 my $n=$vrxf{$power_of_2};
286 if (! $n) {
287
288 # Here, there was no name that matched exactly the bit. It could be
289 # either that it is unused, or the name matches multiple bits.
290 if (! ($val & $power_of_2)) {
291 $n = "UNUSED_BIT_$_";
292 }
293 else {
294
295 # Here, must be because it matches multiple bits. Look through
296 # all possibilities until find one that matches this one. Use
297 # that name, and all the bits it matches
298 foreach my $name (keys %rxfv) {
299 if ($rxfv{$name} & $power_of_2) {
300 $n = $name;
301 $power_of_2 = $rxfv{$name};
302 last;
303 }
304 }
305 }
306 }
f7819f85 307 $n=~s/^RXf_(PMf_)?//;
424a4936 308 printf $out qq(\t%-20s/* 0x%08x */\n),
5458d9a0 309 qq("$n",),$power_of_2;
f7819f85
A
310}
311
424a4936 312print $out <<EOP;
f7819f85
A
313};
314#endif /* DOINIT */
315
f9ef50a7
NC
316EOP
317
318print $out process_flags('V', 'varies', <<'EOC');
319/* The following have no fixed length. U8 so we can do strchr() on it. */
320EOC
321
322print $out process_flags('S', 'simple', <<'EOC');
ce716c52 323
f9ef50a7
NC
324/* The following always have a length of 1. U8 we can do strchr() on it. */
325/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
326EOC
327
ce716c52 328read_only_bottom_close_and_rename($out);