This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
toke.c: Remove TOKENTYPE_GVVAL
[perl5.git] / regen_perly.pl
CommitLineData
0de566d7
DM
1#!/usr/bin/perl
2#
3# regen_perly.pl, DAPM 12-Feb-04
4#
2eee27d7 5# Copyright (c) 2004, 2005, 2006, 2009, 2010, 2011 Larry Wall
0de566d7
DM
6#
7# Given an input file perly.y, run bison on it and produce
8# the following output files:
9#
10# perly.h standard bison header file with minor doctoring of
11# #line directives plus adding a #ifdef PERL_CORE
12#
13# perly.tab the parser table C definitions extracted from the bison output
0539ab63 14# plus an extra table generated by this script.
0de566d7
DM
15#
16# perly.act the action case statements extracted from the bison output
17#
18# Note that perly.c is *not* regenerated - this is now a static file which
19# is not dependent on perly.y any more.
20#
21# If a filename of the form foo.y is given on the command line, then
22# this is used instead as the basename for all the files mentioned
23# above.
24#
25# Note that temporary files of the form perlytmp.h and perlytmp.c are
26# created and then deleted during this process
27#
28# Note also that this script is intended to be run on a UNIX system;
29# it may work elsewhere but no specific attempt has been made to make it
30# portable.
31
e8fb9efb 32use 5.006;
0de566d7
DM
33sub usage { die "usage: $0 [ -b bison_executable ] [ file.y ]\n" }
34
35use warnings;
36use strict;
37
2bd4012d 38BEGIN { require 'regen/regen_lib.pl'; }
a9718e07 39
0de566d7
DM
40my $bison = 'bison';
41
42if (@ARGV >= 2 and $ARGV[0] eq '-b') {
43 shift;
44 $bison = shift;
45}
46
47my $y_file = shift || 'perly.y';
48
49usage unless @ARGV==0 && $y_file =~ /\.y$/;
50
51(my $h_file = $y_file) =~ s/\.y$/.h/;
52(my $act_file = $y_file) =~ s/\.y$/.act/;
53(my $tab_file = $y_file) =~ s/\.y$/.tab/;
54(my $tmpc_file = $y_file) =~ s/\.y$/tmp.c/;
55(my $tmph_file = $y_file) =~ s/\.y$/tmp.h/;
56
57# the yytranslate[] table generated by bison is ASCII/EBCDIC sensitive
58
59die "$0: must be run on an ASCII system\n" unless ord 'A' == 65;
60
61# check for correct version number. The constraints are:
62# * must be >= 1.24 to avoid licensing issues.
63# * it must generate the yystos[] table. Version 1.28 doesn't generate
64# this; 1.35+ does
65# * Must produce output which is extractable by the regexes below
66# * Must produce the right values.
8abc1060 67# These last two constraints may well be met by earlier versions, but
0de566d7
DM
68# I simply haven't tested them yet. If it works for you, then modify
69# the test below to allow that version too. DAPM Feb 04.
70
71my $version = `$bison -V`;
7b931d50
LB
72unless ($version) { die <<EOF; }
73Could not find a version of bison in your path. Please install bison.
74EOF
75
2434f628 76unless ($version =~ /\b(1\.875[a-z]?|2\.[01345])\b/) { die <<EOF; }
0de566d7 77
2080282f 78You have the wrong version of bison in your path; currently 1.875
2434f628
RU
792.0, 2.1, 2.3, 2.4 or 2.5 is required. Try installing
80 http://ftp.gnu.org/gnu/bison/bison-2.5.1.tar.gz
0de566d7
DM
81or similar. Your bison identifies itself as:
82
83$version
84EOF
85
86# creates $tmpc_file and $tmph_file
87my_system("$bison -d -o $tmpc_file $y_file");
88
e8fb9efb 89open my $ctmp_fh, '<', $tmpc_file or die "Can't open $tmpc_file: $!\n";
0de566d7 90my $clines;
e8fb9efb 91{ local $/; $clines = <$ctmp_fh>; }
0de566d7
DM
92die "failed to read $tmpc_file: length mismatch\n"
93 unless length $clines == -s $tmpc_file;
e8fb9efb 94close $ctmp_fh;
0de566d7
DM
95
96my ($actlines, $tablines) = extract($clines);
97
d5c6462e 98$tablines .= make_type_tab($y_file, $tablines);
0539ab63 99
cc49830d
NC
100my ($act_fh, $tab_fh, $h_fh) = map {
101 open_new($_, '>', { by => $0, from => $y_file });
102} $act_file, $tab_file, $h_file;
0de566d7 103
cc49830d 104print $act_fh $actlines;
e8fb9efb 105
cc49830d 106print $tab_fh $tablines;
0de566d7
DM
107
108unlink $tmpc_file;
109
110# Wrap PERL_CORE round the symbol definitions. Also, the
96f4e226
SH
111# C<#line 30 "perly.y"> confuses the Win32 resource compiler and the
112# C<#line 188 "perlytmp.h"> gets picked up by make depend, so remove them.
0de566d7 113
e8fb9efb 114open my $tmph_fh, '<', $tmph_file or die "Can't open $tmph_file: $!\n";
e8fb9efb 115
0de566d7 116my $endcore_done = 0;
2434f628
RU
117# Token macros need to be generated manually from bison 2.4 on
118my $gather_tokens = ($version =~ /\b2\.[45]\b/ ? undef : 0);
ce1534ab 119my $tokens;
e8fb9efb
NC
120while (<$tmph_fh>) {
121 print $h_fh "#ifdef PERL_CORE\n" if $. == 1;
0de566d7 122 if (!$endcore_done and /YYSTYPE_IS_DECLARED/) {
e8fb9efb 123 print $h_fh "#endif /* PERL_CORE */\n";
0de566d7
DM
124 $endcore_done = 1;
125 }
96f4e226 126 next if /^#line \d+ ".*"/;
ce1534ab
VP
127 if (not defined $gather_tokens) {
128 $gather_tokens = 1 if /^\s* enum \s* yytokentype \s* \{/x;
129 }
130 elsif ($gather_tokens) {
131 if (/^\# \s* endif/x) { # The #endif just after the end of the token enum
132 $gather_tokens = 0;
133 $_ .= "\n/* Tokens. */\n$tokens";
134 }
135 else {
136 my ($tok, $val) = /(\w+) \s* = \s* (\d+)/x;
137 $tokens .= "#define $tok $val\n" if $tok;
138 }
139 }
e8fb9efb 140 print $h_fh $_;
0de566d7 141}
e8fb9efb 142close $tmph_fh;
0de566d7
DM
143unlink $tmph_file;
144
c24c946d
NC
145foreach ($act_fh, $tab_fh, $h_fh) {
146 read_only_bottom_close_and_rename($_, ['regen_perly.pl', $y_file]);
147}
0de566d7
DM
148
149exit 0;
150
151
152sub extract {
153 my $clines = shift;
154 my $tablines;
155 my $actlines;
156
157 $clines =~ m@
158 (?:
159 ^/* YYFINAL[^\n]+\n #optional comment
160 )?
161 \# \s* define \s* YYFINAL # first #define
162 .*? # other defines + most tables
163 yystos\[\]\s*= # start of last table
164 .*?
165 }\s*; # end of last table
166 @xms
167 or die "Can't extract tables from $tmpc_file\n";
168 $tablines = $&;
169
170
171 $clines =~ m@
172 switch \s* \( \s* \w+ \s* \) \s* { \s*
173 (
ce1534ab
VP
174 case \s* \d+ \s* :
175 \s*
176 (?: \s* /\* .*? \*/ \s* )* # optional C-comments
177 \s*
2ade6388 178 \#line [^\n]+"\Q$y_file\E"
0de566d7
DM
179 .*?
180 )
181 }
182 \s*
ce1534ab 183 (?: \s* /\* .*? \*/ \s* )* # optional C-comments
0de566d7
DM
184 \s*
185 (
186 \#line[^\n]+\.c"
187 |
188 \#line[^\n]+\.simple"
3797f23d
DM
189 |
190 YY_SYMBOL_PRINT
0de566d7
DM
191 )
192 @xms
193 or die "Can't extract actions from $tmpc_file\n";
194 $actlines = $1;
195
ce1534ab
VP
196 # Remove extraneous comments from bison 2.4
197 $actlines =~ s!\s* /\* \s* Line \s* \d+ \s* of \s* yacc\.c \s* \*/!!gx;
198
0d6f9730
DM
199 # C<#line 188 "perlytmp.c"> gets picked up by make depend, so remove them.
200 $actlines =~ s/^#line \d+ "\Q$tmpc_file\E".*$//gm;
201
1654d593
DM
202 # convert yyvsp[nnn] into ps[nnn].val
203
204 $actlines =~ s/yyvsp\[(.*?)\]/ps[$1].val/g
205 or die "Can't convert value stack name\n";
206
0de566d7
DM
207 return $actlines. "\n", $tablines. "\n";
208}
209
d5c6462e
DM
210# Generate a table, yy_type_tab[], that specifies for each token, what
211# type of value it holds.
0539ab63 212#
d5c6462e
DM
213# Read the .y file and extract a list of all the token names and
214# non-terminal names; then scan the string $tablines for the table yytname,
215# which gives the token index of each token/non-terminal; then use this to
216# create yy_type_tab.
0539ab63 217#
d5c6462e
DM
218# ie given (in perly.y),
219#
220# %token <opval> A
221# %token <ival> B
222# %type <pval> C
223# %type <opval> D
224#
225# and (in $tablines),
226#
227# yytname[] = { "A" "B", "C", "D", "E" };
0539ab63
DM
228#
229# then return
d5c6462e
DM
230#
231# typedef enum { toketype_ival, toketype_opval, toketype_pval } toketypes;
232#
233# static const toketypes yy_type_tab[]
234# = { toketype_opval, toketype_ival, toketype_pval,
235# toketype_opval, toketype_ival }
236#
237# where "E" has the default type. The default type is determined
238# by the __DEFAULT__ comment next to the appropriate union member in
239# perly.y
0539ab63 240
d5c6462e 241sub make_type_tab {
0539ab63
DM
242 my ($y_file, $tablines) = @_;
243 my %tokens;
d5c6462e
DM
244 my %types;
245 my $default_token;
0539ab63
DM
246 open my $fh, '<', $y_file or die "Can't open $y_file: $!\n";
247 while (<$fh>) {
29522234
DM
248 if (/(\$\d+)\s*=/) {
249 warn "$y_file:$.: dangerous assignment to $1: $_";
250 }
251
d5c6462e
DM
252 if (/__DEFAULT__/) {
253 m{(\w+) \s* ; \s* /\* \s* __DEFAULT__}x
254 or die "$y_file: can't parse __DEFAULT__ line: $_";
255 die "$y_file: duplicate __DEFAULT__ line: $_"
256 if defined $default_token;
257 $default_token = $1;
258 next;
259 }
260
261 next unless /^%(token|type)/;
262 s/^%(token|type)\s+<(\w+)>\s+//
263 or die "$y_file: unparseable token/type line: $_";
264 $tokens{$_} = $2 for (split ' ', $_);
265 $types{$2} = 1;
0539ab63 266 }
d5c6462e
DM
267 die "$y_file: no __DEFAULT__ token defined\n" unless $default_token;
268 $types{$default_token} = 1;
0539ab63
DM
269
270 $tablines =~ /^\Qstatic const char *const yytname[] =\E\n
efcfdf1f 271 \{\n
0539ab63
DM
272 (.*?)
273 ^};
274 /xsm
275 or die "Can't extract yytname[] from table string\n";
276 my $fields = $1;
d5c6462e
DM
277 $fields =~ s{"([^"]+)"}
278 { "toketype_" .
279 (defined $tokens{$1} ? $tokens{$1} : $default_token)
280 }ge;
281 $fields =~ s/, \s* 0 \s* $//x
282 or die "make_type_tab: couldn't delete trailing ',0'\n";
283
0539ab63 284 return
d5c6462e
DM
285 "\ntypedef enum {\n\t"
286 . join(", ", map "toketype_$_", sort keys %types)
287 . "\n} toketypes;\n\n"
288 . "/* type of each token/terminal */\n"
d5c6462e
DM
289 . "static const toketypes yy_type_tab[] =\n{\n"
290 . $fields
291 . "\n};\n";
0539ab63
DM
292}
293
294
0de566d7
DM
295sub my_system {
296 system(@_);
297 if ($? == -1) {
d5c6462e 298 die "failed to execute command '@_': $!\n";
0de566d7
DM
299 }
300 elsif ($? & 127) {
301 die sprintf "command '@_' died with signal %d\n",
302 ($? & 127);
303 }
304 elsif ($? >> 8) {
305 die sprintf "command '@_' exited with value %d\n", $? >> 8;
306 }
307}