This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
die if multiple macros in regexp.h attempt to use the same flag bit.
[perl5.git] / regcomp.pl
CommitLineData
36bb303b
NC
1BEGIN {
2 # Get function prototypes
9ad884cb 3 require 'regen_lib.pl';
36bb303b 4}
d09b2d29 5#use Fatal qw(open close rename chmod unlink);
03363afd
YO
6use strict;
7use warnings;
8
d09b2d29 9open DESC, 'regcomp.sym';
d09b2d29 10
03363afd
YO
11my $ind = 0;
12my (@name,@rest,@type,@code,@args,@longj);
13my ($desc,$lastregop);
d09b2d29 14while (<DESC>) {
03363afd
YO
15 s/#.*$//;
16 next if /^\s*$/;
17 s/\s*\z//;
18 if (/^-+\s*$/) {
19 $lastregop= $ind;
20 next;
21 }
22 unless ($lastregop) {
23 $ind++;
24 ($name[$ind], $desc, $rest[$ind]) = split /\t+/, $_, 3;
25 ($type[$ind], $code[$ind], $args[$ind], $longj[$ind])
26 = split /[,\s]\s*/, $desc, 4;
27 } else {
28 my ($type,@lists)=split /\s*\t+\s*/, $_;
29 die "No list? $type" if !@lists;
30 foreach my $list (@lists) {
31 my ($names,$special)=split /:/, $list , 2;
32 $special ||= "";
33 foreach my $name (split /,/,$names) {
34 my $real= $name eq 'resume'
35 ? "resume_$type"
36 : "${type}_$name";
37 my @suffix;
38 if (!$special) {
39 @suffix=("");
40 } elsif ($special=~/\d/) {
41 @suffix=(1..$special);
42 } elsif ($special eq 'FAIL') {
43 @suffix=("","_fail");
44 } else {
45 die "unknown :type ':$special'";
46 }
47 foreach my $suffix (@suffix) {
48 $ind++;
49 $name[$ind]="$real$suffix";
50 $type[$ind]=$type;
24b23f37 51 $rest[$ind]="state for $type";
03363afd
YO
52 }
53 }
54 }
55
56 }
57}
5d458dd8
YO
58# use fixed width to keep the diffs between regcomp.pl recompiles
59# as small as possible.
60my ($width,$rwidth,$twidth)=(22,12,9);
03363afd
YO
61$lastregop ||= $ind;
62my $tot = $ind;
d09b2d29 63close DESC;
03363afd
YO
64die "Too many regexp/state opcodes! Maximum is 256, but there are $lastregop in file!"
65 if $lastregop>256;
d09b2d29 66
03363afd 67my $tmp_h = 'tmp_reg.h';
d09b2d29
IZ
68
69unlink $tmp_h if -f $tmp_h;
70
71open OUT, ">$tmp_h";
03363afd 72#*OUT=\*STDOUT;
dfb1454f 73binmode OUT;
d09b2d29 74
03363afd 75printf OUT <<EOP,
37442d52
RGS
76/* -*- buffer-read-only: t -*-
77 !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
9b155405 78 This file is built by regcomp.pl from regcomp.sym.
d09b2d29
IZ
79 Any changes made here will be lost!
80*/
81
6bda09f9
YO
82/* Regops and State definitions */
83
03363afd
YO
84#define %*s\t%d
85#define %*s\t%d
86
d09b2d29 87EOP
f9f4320a
YO
88 -$width, REGNODE_MAX => $lastregop - 1,
89 -$width, REGMATCH_STATE_MAX => $tot - 1
90;
d09b2d29 91
24b23f37
YO
92
93for ($ind=1; $ind <= $lastregop ; $ind++) {
03363afd 94 my $oind = $ind - 1;
f9f4320a 95 printf OUT "#define\t%*s\t%d\t/* %#04x %s */\n",
03363afd 96 -$width, $name[$ind], $ind-1, $ind-1, $rest[$ind];
24b23f37
YO
97}
98print OUT "\t/* ------------ States ------------- */\n";
99for ( ; $ind <= $tot ; $ind++) {
100 printf OUT "#define\t%*s\t(REGNODE_MAX + %d)\t/* %s */\n",
101 -$width, $name[$ind], $ind - $lastregop, $rest[$ind];
d09b2d29
IZ
102}
103
104print OUT <<EOP;
03363afd 105
6bda09f9 106/* PL_regkind[] What type of regop or state is this. */
d09b2d29
IZ
107
108#ifndef DOINIT
22c35a8c 109EXTCONST U8 PL_regkind[];
d09b2d29 110#else
22c35a8c 111EXTCONST U8 PL_regkind[] = {
d09b2d29
IZ
112EOP
113
114$ind = 0;
115while (++$ind <= $tot) {
03363afd
YO
116 printf OUT "\t%*s\t/* %*s */\n",
117 -1-$twidth, "$type[$ind],", -$width, $name[$ind];
118 print OUT "\t/* ------------ States ------------- */\n"
119 if $ind == $lastregop and $lastregop != $tot;
d09b2d29
IZ
120}
121
122print OUT <<EOP;
123};
124#endif
125
6bda09f9 126/* regarglen[] - How large is the argument part of the node (in regnodes) */
d09b2d29
IZ
127
128#ifdef REG_COMP_C
29de9391 129static const U8 regarglen[] = {
d09b2d29
IZ
130EOP
131
132$ind = 0;
03363afd
YO
133while (++$ind <= $lastregop) {
134 my $size = 0;
d09b2d29
IZ
135 $size = "EXTRA_SIZE(struct regnode_$args[$ind])" if $args[$ind];
136
03363afd
YO
137 printf OUT "\t%*s\t/* %*s */\n",
138 -37, "$size,",-$rwidth,$name[$ind];
d09b2d29
IZ
139}
140
141print OUT <<EOP;
142};
143
6bda09f9
YO
144/* reg_off_by_arg[] - Which argument holds the offset to the next node */
145
29de9391 146static const char reg_off_by_arg[] = {
d09b2d29
IZ
147EOP
148
149$ind = 0;
03363afd
YO
150while (++$ind <= $lastregop) {
151 my $size = $longj[$ind] || 0;
9b155405 152
03363afd
YO
153 printf OUT "\t%d,\t/* %*s */\n",
154 $size, -$rwidth, $name[$ind]
d09b2d29
IZ
155}
156
157print OUT <<EOP;
158};
9b155405 159
13d6edb4
NC
160#endif /* REG_COMP_C */
161
6bda09f9
YO
162/* reg_name[] - Opcode/state names in string form, for debugging */
163
22429478 164#ifndef DOINIT
13d6edb4 165EXTCONST char * PL_reg_name[];
22429478 166#else
4764e399 167EXTCONST char * const PL_reg_name[] = {
9b155405
IZ
168EOP
169
170$ind = 0;
24b23f37
YO
171my $ofs = 1;
172my $sym = "";
9b155405 173while (++$ind <= $tot) {
03363afd 174 my $size = $longj[$ind] || 0;
9b155405 175
24b23f37
YO
176 printf OUT "\t%*s\t/* $sym%#04x */\n",
177 -3-$width,qq("$name[$ind]",), $ind - $ofs;
178 if ($ind == $lastregop and $lastregop != $tot) {
179 print OUT "\t/* ------------ States ------------- */\n";
180 $ofs = $lastregop;
181 $sym = 'REGNODE_MAX +';
182 }
183
9b155405
IZ
184}
185
186print OUT <<EOP;
187};
22429478 188#endif /* DOINIT */
d09b2d29 189
f7819f85
A
190/* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
191
192#ifndef DOINIT
193EXTCONST char * PL_reg_extflags_name[];
194#else
195EXTCONST char * const PL_reg_extflags_name[] = {
d09b2d29
IZ
196EOP
197
f7819f85
A
198open my $fh,"<","regexp.h" or die "Can't read regexp.h: $!";
199my %rxfv;
c8e4cf8b
NC
200my $val = 0;
201my %reverse;
f7819f85
A
202while (<$fh>) {
203 if (/#define\s+(RXf_\w+)\s+(0x[A-F\d]+)/i) {
c8e4cf8b
NC
204 my $newval = eval $2;
205 if($val & $newval) {
206 die sprintf "Both $1 and $reverse{$newval} use %08X", $newval;
207 }
208 $val|=$newval;
209 $rxfv{$1}= $newval;
210 $reverse{$newval} = $1;
f7819f85
A
211 }
212}
213my %vrxf=reverse %rxfv;
214printf OUT "\t/* Bits in extflags defined: %032b */\n",$val;
215for (0..31) {
216 my $n=$vrxf{2**$_}||"UNUSED_BIT_$_";
217 $n=~s/^RXf_(PMf_)?//;
218 printf OUT qq(\t%-20s/* 0x%08x */\n),
219 qq("$n",),2**$_;
220}
221
222print OUT <<EOP;
223};
224#endif /* DOINIT */
225
226/* ex: set ro: */
227EOP
36bb303b 228close OUT or die "close $tmp_h: $!";
d09b2d29 229
36bb303b 230safer_rename $tmp_h, 'regnodes.h';