This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Bump minimum required Bison version to 2.4
[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
e64a0c47 38our $Verbose;
3d7c117d 39BEGIN { require './regen/regen_lib.pl'; }
a9718e07 40
0de566d7
DM
41my $bison = 'bison';
42
43if (@ARGV >= 2 and $ARGV[0] eq '-b') {
44 shift;
45 $bison = shift;
46}
47
48my $y_file = shift || 'perly.y';
49
50usage unless @ARGV==0 && $y_file =~ /\.y$/;
51
52(my $h_file = $y_file) =~ s/\.y$/.h/;
53(my $act_file = $y_file) =~ s/\.y$/.act/;
54(my $tab_file = $y_file) =~ s/\.y$/.tab/;
55(my $tmpc_file = $y_file) =~ s/\.y$/tmp.c/;
56(my $tmph_file = $y_file) =~ s/\.y$/tmp.h/;
57
58# the yytranslate[] table generated by bison is ASCII/EBCDIC sensitive
59
60die "$0: must be run on an ASCII system\n" unless ord 'A' == 65;
61
62# check for correct version number. The constraints are:
63# * must be >= 1.24 to avoid licensing issues.
64# * it must generate the yystos[] table. Version 1.28 doesn't generate
65# this; 1.35+ does
66# * Must produce output which is extractable by the regexes below
67# * Must produce the right values.
8abc1060 68# These last two constraints may well be met by earlier versions, but
0de566d7
DM
69# I simply haven't tested them yet. If it works for you, then modify
70# the test below to allow that version too. DAPM Feb 04.
71
72my $version = `$bison -V`;
7b931d50
LB
73unless ($version) { die <<EOF; }
74Could not find a version of bison in your path. Please install bison.
75EOF
76
9c221ee4
NC
77# Don't change this to add new bison versions without testing that the generated
78# files actually work :-) Win32 in particular may not like them. :-(
0a2b3d88 79unless ($version =~ /\b(2\.[4567]|3\.[0-4])\b/) { die <<EOF; }
0de566d7 80
f39ff1f3 81You have the wrong version of bison in your path; currently versions
0a2b3d88
DIM
822.4-2.7 or 3.0-3.4 are known to work. Try installing
83 http://ftp.gnu.org/gnu/bison/bison-3.3.tar.gz
0de566d7
DM
84or similar. Your bison identifies itself as:
85
86$version
87EOF
88
9c221ee4
NC
89# bison's version number, not the entire string, is most useful later on.
90$version = $1;
91
0de566d7
DM
92# creates $tmpc_file and $tmph_file
93my_system("$bison -d -o $tmpc_file $y_file");
94
e8fb9efb 95open my $ctmp_fh, '<', $tmpc_file or die "Can't open $tmpc_file: $!\n";
0de566d7 96my $clines;
e8fb9efb 97{ local $/; $clines = <$ctmp_fh>; }
0de566d7
DM
98die "failed to read $tmpc_file: length mismatch\n"
99 unless length $clines == -s $tmpc_file;
e8fb9efb 100close $ctmp_fh;
0de566d7
DM
101
102my ($actlines, $tablines) = extract($clines);
103
6c7ae946 104our %tokens;
d5c6462e 105$tablines .= make_type_tab($y_file, $tablines);
0539ab63 106
cc49830d
NC
107my ($act_fh, $tab_fh, $h_fh) = map {
108 open_new($_, '>', { by => $0, from => $y_file });
109} $act_file, $tab_file, $h_file;
0de566d7 110
cc49830d 111print $act_fh $actlines;
e8fb9efb 112
cc49830d 113print $tab_fh $tablines;
0de566d7
DM
114
115unlink $tmpc_file;
116
117# Wrap PERL_CORE round the symbol definitions. Also, the
96f4e226
SH
118# C<#line 30 "perly.y"> confuses the Win32 resource compiler and the
119# C<#line 188 "perlytmp.h"> gets picked up by make depend, so remove them.
0de566d7 120
e8fb9efb 121open my $tmph_fh, '<', $tmph_file or die "Can't open $tmph_file: $!\n";
e8fb9efb 122
f39ff1f3
DM
123# add integer-encoded #def of the bison version
124
125{
126 $version =~ /^(\d+)\.(\d+)/
127 or die "Can't handle bison version format: '$version'";
128 my ($v1,$v2) = ($1,$2);
129 die "Unexpectedly large bison version '$v1'" if $v1 > 99;
130 die "Unexpectedly large bison subversion '$v2'" if $v2 > 9999;
131
132 printf $h_fh "#define PERL_BISON_VERSION %2d%04d\n\n", $v1, $v2;
133}
134
0de566d7 135my $endcore_done = 0;
0a2b3d88 136my $gather_tokens = 0;
ce1534ab 137my $tokens;
e8fb9efb 138while (<$tmph_fh>) {
04ff073f
JL
139 # bison 2.6 adds header guards, which break things because of where we
140 # insert #ifdef PERL_CORE, so strip them because they aren't important
141 next if /YY_PERLYTMP_H/;
142
e8fb9efb 143 print $h_fh "#ifdef PERL_CORE\n" if $. == 1;
0de566d7 144 if (!$endcore_done and /YYSTYPE_IS_DECLARED/) {
6c7ae946
FC
145 print $h_fh <<h;
146#ifdef PERL_IN_TOKE_C
147static bool
148S_is_opval_token(int type) {
149 switch (type) {
150h
151 print $h_fh <<i for sort grep $tokens{$_} eq 'opval', keys %tokens;
152 case $_:
153i
154 print $h_fh <<j;
155 return 1;
156 }
157 return 0;
158}
159#endif /* PERL_IN_TOKE_C */
160#endif /* PERL_CORE */
161j
0de566d7
DM
162 $endcore_done = 1;
163 }
96f4e226 164 next if /^#line \d+ ".*"/;
0a2b3d88 165 if (!$gather_tokens) {
ce1534ab
VP
166 $gather_tokens = 1 if /^\s* enum \s* yytokentype \s* \{/x;
167 }
0a2b3d88 168 else {
ce1534ab
VP
169 if (/^\# \s* endif/x) { # The #endif just after the end of the token enum
170 $gather_tokens = 0;
171 $_ .= "\n/* Tokens. */\n$tokens";
172 }
173 else {
174 my ($tok, $val) = /(\w+) \s* = \s* (\d+)/x;
175 $tokens .= "#define $tok $val\n" if $tok;
176 }
177 }
e8fb9efb 178 print $h_fh $_;
0de566d7 179}
e8fb9efb 180close $tmph_fh;
0de566d7
DM
181unlink $tmph_file;
182
c24c946d
NC
183foreach ($act_fh, $tab_fh, $h_fh) {
184 read_only_bottom_close_and_rename($_, ['regen_perly.pl', $y_file]);
185}
0de566d7
DM
186
187exit 0;
188
189
59966791
DM
190# extract the tables and actions from the generated .c file
191
0de566d7
DM
192sub extract {
193 my $clines = shift;
194 my $tablines;
195 my $actlines;
196
f39ff1f3 197 my $last_table = $version >= 3 ? 'yyr2' : 'yystos';
0de566d7
DM
198 $clines =~ m@
199 (?:
200 ^/* YYFINAL[^\n]+\n #optional comment
201 )?
202 \# \s* define \s* YYFINAL # first #define
203 .*? # other defines + most tables
f39ff1f3 204 $last_table\[\]\s*= # start of last table
0de566d7
DM
205 .*?
206 }\s*; # end of last table
207 @xms
208 or die "Can't extract tables from $tmpc_file\n";
209 $tablines = $&;
210
211
59966791
DM
212 # extract all the cases in the big action switch statement
213
0de566d7 214 $clines =~ m@
59966791
DM
215 switch \s* \( \s* yyn \s* \) \s* { \s*
216 ( .*? default: \s* break; \s* )
217 }
0de566d7
DM
218 @xms
219 or die "Can't extract actions from $tmpc_file\n";
220 $actlines = $1;
221
ce1534ab
VP
222 # Remove extraneous comments from bison 2.4
223 $actlines =~ s!\s* /\* \s* Line \s* \d+ \s* of \s* yacc\.c \s* \*/!!gx;
224
0d6f9730
DM
225 # C<#line 188 "perlytmp.c"> gets picked up by make depend, so remove them.
226 $actlines =~ s/^#line \d+ "\Q$tmpc_file\E".*$//gm;
227
1654d593
DM
228 # convert yyvsp[nnn] into ps[nnn].val
229
230 $actlines =~ s/yyvsp\[(.*?)\]/ps[$1].val/g
231 or die "Can't convert value stack name\n";
232
0de566d7
DM
233 return $actlines. "\n", $tablines. "\n";
234}
235
d5c6462e
DM
236# Generate a table, yy_type_tab[], that specifies for each token, what
237# type of value it holds.
0539ab63 238#
d5c6462e
DM
239# Read the .y file and extract a list of all the token names and
240# non-terminal names; then scan the string $tablines for the table yytname,
241# which gives the token index of each token/non-terminal; then use this to
242# create yy_type_tab.
0539ab63 243#
d5c6462e
DM
244# ie given (in perly.y),
245#
246# %token <opval> A
247# %token <ival> B
248# %type <pval> C
249# %type <opval> D
250#
251# and (in $tablines),
252#
253# yytname[] = { "A" "B", "C", "D", "E" };
0539ab63
DM
254#
255# then return
d5c6462e
DM
256#
257# typedef enum { toketype_ival, toketype_opval, toketype_pval } toketypes;
258#
259# static const toketypes yy_type_tab[]
260# = { toketype_opval, toketype_ival, toketype_pval,
261# toketype_opval, toketype_ival }
262#
263# where "E" has the default type. The default type is determined
264# by the __DEFAULT__ comment next to the appropriate union member in
265# perly.y
0539ab63 266
d5c6462e 267sub make_type_tab {
0539ab63 268 my ($y_file, $tablines) = @_;
6c7ae946 269 my %just_tokens;
0539ab63 270 my %tokens;
d5c6462e
DM
271 my %types;
272 my $default_token;
0539ab63
DM
273 open my $fh, '<', $y_file or die "Can't open $y_file: $!\n";
274 while (<$fh>) {
b5bbe64a 275 if (/(\$\d+)\s*=[^=]/) {
29522234
DM
276 warn "$y_file:$.: dangerous assignment to $1: $_";
277 }
278
d5c6462e
DM
279 if (/__DEFAULT__/) {
280 m{(\w+) \s* ; \s* /\* \s* __DEFAULT__}x
281 or die "$y_file: can't parse __DEFAULT__ line: $_";
282 die "$y_file: duplicate __DEFAULT__ line: $_"
283 if defined $default_token;
284 $default_token = $1;
285 next;
286 }
287
288 next unless /^%(token|type)/;
6c7ae946 289 s/^%((token)|type)\s+<(\w+)>\s+//
d5c6462e 290 or die "$y_file: unparseable token/type line: $_";
6c7ae946
FC
291 for (split ' ', $_) {
292 $tokens{$_} = $3;
293 if ($2) {
294 $just_tokens{$_} = $3;
295 }
296 }
297 $types{$3} = 1;
0539ab63 298 }
6c7ae946 299 *tokens = \%just_tokens; # perly.h needs this
d5c6462e
DM
300 die "$y_file: no __DEFAULT__ token defined\n" unless $default_token;
301 $types{$default_token} = 1;
0539ab63
DM
302
303 $tablines =~ /^\Qstatic const char *const yytname[] =\E\n
efcfdf1f 304 \{\n
0539ab63
DM
305 (.*?)
306 ^};
307 /xsm
308 or die "Can't extract yytname[] from table string\n";
309 my $fields = $1;
d5c6462e
DM
310 $fields =~ s{"([^"]+)"}
311 { "toketype_" .
312 (defined $tokens{$1} ? $tokens{$1} : $default_token)
313 }ge;
f39ff1f3 314 $fields =~ s/, \s* (?:0|YY_NULL|YY_NULLPTR) \s* $//x
d5c6462e
DM
315 or die "make_type_tab: couldn't delete trailing ',0'\n";
316
0539ab63 317 return
d5c6462e
DM
318 "\ntypedef enum {\n\t"
319 . join(", ", map "toketype_$_", sort keys %types)
320 . "\n} toketypes;\n\n"
321 . "/* type of each token/terminal */\n"
d5c6462e
DM
322 . "static const toketypes yy_type_tab[] =\n{\n"
323 . $fields
324 . "\n};\n";
0539ab63
DM
325}
326
327
0de566d7 328sub my_system {
95a1c520
DM
329 if ($Verbose) {
330 print "executing: @_\n";
331 }
0de566d7
DM
332 system(@_);
333 if ($? == -1) {
d5c6462e 334 die "failed to execute command '@_': $!\n";
0de566d7
DM
335 }
336 elsif ($? & 127) {
337 die sprintf "command '@_' died with signal %d\n",
338 ($? & 127);
339 }
340 elsif ($? >> 8) {
341 die sprintf "command '@_' exited with value %d\n", $? >> 8;
342 }
343}