3 # regen_perly.pl, DAPM 12-Feb-04
5 # Copyright (c) 2004, 2005, 2006, 2009, 2010, 2011 Larry Wall
7 # Given an input file perly.y, run bison on it and produce
8 # the following output files:
10 # perly.h standard bison header file with minor doctoring of
11 # #line directives plus adding a #ifdef PERL_CORE
13 # perly.tab the parser table C definitions extracted from the bison output
14 # plus an extra table generated by this script.
16 # perly.act the action case statements extracted from the bison output
18 # Note that perly.c is *not* regenerated - this is now a static file which
19 # is not dependent on perly.y any more.
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
25 # Note that temporary files of the form perlytmp.h and perlytmp.c are
26 # created and then deleted during this process
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
33 sub usage { die "usage: $0 [ -b bison_executable ] [ file.y ]\n" }
39 BEGIN { require './regen/regen_lib.pl'; }
43 if (@ARGV >= 2 and $ARGV[0] eq '-b') {
48 my $y_file = shift || 'perly.y';
50 usage unless @ARGV==0 && $y_file =~ /\.y$/;
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/;
58 # the yytranslate[] table generated by bison is ASCII/EBCDIC sensitive
60 die "$0: must be run on an ASCII system\n" unless ord 'A' == 65;
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
66 # * Must produce output which is extractable by the regexes below
67 # * Must produce the right values.
68 # These last two constraints may well be met by earlier versions, but
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.
72 my $version = `$bison -V`;
73 unless ($version) { die <<EOF; }
74 Could not find a version of bison in your path. Please install bison.
77 # Don't change this to add new bison versions without testing that the generated
78 # files actually work :-) Win32 in particular may not like them. :-(
79 unless ($version =~ /\b(2\.[4567]|3\.[0-7])\b/) { die <<EOF; }
81 You have the wrong version of bison in your path; currently versions
82 2.4-2.7 or 3.0-3.7 are known to work. Try installing
83 http://ftp.gnu.org/gnu/bison/bison-3.3.tar.gz
84 or similar. Your bison identifies itself as:
89 # bison's version number, not the entire string, is most useful later on.
92 # creates $tmpc_file and $tmph_file
93 my_system("$bison -d -o $tmpc_file $y_file");
95 open my $ctmp_fh, '<', $tmpc_file or die "Can't open $tmpc_file: $!\n";
97 { local $/; $clines = <$ctmp_fh>; }
98 die "failed to read $tmpc_file: length mismatch\n"
99 unless length $clines == -s $tmpc_file;
102 my ($actlines, $tablines) = extract($clines);
105 $tablines .= make_type_tab($y_file, $tablines);
107 my ($act_fh, $tab_fh, $h_fh) = map {
108 open_new($_, '>', { by => $0, from => $y_file });
109 } $act_file, $tab_file, $h_file;
111 print $act_fh $actlines;
113 print $tab_fh $tablines;
117 # Wrap PERL_CORE round the symbol definitions. Also, the
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.
121 open my $tmph_fh, '<', $tmph_file or die "Can't open $tmph_file: $!\n";
123 # add integer-encoded #def of the bison version
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;
132 printf $h_fh "#define PERL_BISON_VERSION %2d%04d\n\n", $v1, $v2;
135 my $endcore_done = 0;
136 my $gather_tokens = 0;
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/;
143 print $h_fh "#ifdef PERL_CORE\n" if $. == 1;
144 if (!$endcore_done and /YYSTYPE_IS_DECLARED/) {
146 #ifdef PERL_IN_TOKE_C
148 S_is_opval_token(int type) {
151 print $h_fh <<i for sort grep $tokens{$_} eq 'opval', keys %tokens;
159 #endif /* PERL_IN_TOKE_C */
160 #endif /* PERL_CORE */
164 next if /^#line \d+ ".*"/;
165 if (!$gather_tokens) {
166 $gather_tokens = 1 if /^\s* enum \s* yytokentype \s* \{/x;
169 if (/^\# \s* endif/x) { # The #endif just after the end of the token enum
171 $_ .= "\n/* Tokens. */\n$tokens";
174 my ($tok, $val) = /(\w+) \s* = \s* (\d+)/x;
175 $tokens .= "#define $tok $val\n" if $tok;
183 foreach ($act_fh, $tab_fh, $h_fh) {
184 read_only_bottom_close_and_rename($_, ['regen_perly.pl', $y_file]);
190 # extract the symbol kinds, tables and actions from the generated .c file
197 # extract the symbol kind table if it exists
200 ^/\* \s* Symbol \s+ kind\. \s* \*/\n
202 enum \s+ yysymbol_kind_t \s* \{
205 typedef \s+ enum \s+ \w+ \s+ \w+ ; \n+
209 my $last_table = $version >= 3 ? 'yyr2' : 'yystos';
212 ^/* YYFINAL[^\n]+\n #optional comment
214 \# \s* define \s* YYFINAL # first #define
215 .*? # other defines + most tables
216 $last_table\[\]\s*= # start of last table
218 }\s*; # end of last table
220 or die "Can't extract tables from $tmpc_file\n";
224 # extract all the cases in the big action switch statement
227 switch \s* \( \s* yyn \s* \) \s* { \s*
228 ( .*? default: \s* break; \s* )
231 or die "Can't extract actions from $tmpc_file\n";
234 # Remove extraneous comments from bison 2.4
235 $actlines =~ s!\s* /\* \s* Line \s* \d+ \s* of \s* yacc\.c \s* \*/!!gx;
237 # Remove extraneous comments from bison 3.x
238 $actlines =~ s!\s* /\* \s* yacc\.c : \d+ \s* \*/!!gx;
240 # C<#line 188 "perlytmp.c"> gets picked up by make depend, so remove them.
241 $actlines =~ s/^#line \d+ "\Q$tmpc_file\E".*$//gm;
243 # convert yyvsp[nnn] into ps[nnn].val
245 $actlines =~ s/yyvsp\[(.*?)\]/ps[$1].val/g
246 or die "Can't convert value stack name\n";
248 return $actlines. "\n", $tablines. "\n";
251 # Generate a table, yy_type_tab[], that specifies for each token, what
252 # type of value it holds.
254 # Read the .y file and extract a list of all the token names and
255 # non-terminal names; then scan the string $tablines for the table yytname,
256 # which gives the token index of each token/non-terminal; then use this to
257 # create yy_type_tab.
259 # ie given (in perly.y),
266 # and (in $tablines),
268 # yytname[] = { "A" "B", "C", "D", "E" };
272 # typedef enum { toketype_ival, toketype_opval, toketype_pval } toketypes;
274 # static const toketypes yy_type_tab[]
275 # = { toketype_opval, toketype_ival, toketype_pval,
276 # toketype_opval, toketype_ival }
278 # where "E" has the default type. The default type is determined
279 # by the __DEFAULT__ comment next to the appropriate union member in
283 my ($y_file, $tablines) = @_;
288 open my $fh, '<', $y_file or die "Can't open $y_file: $!\n";
290 if (/(\$\d+)\s*=[^=]/) {
291 warn "$y_file:$.: dangerous assignment to $1: $_";
295 m{(\w+) \s* ; \s* /\* \s* __DEFAULT__}x
296 or die "$y_file: can't parse __DEFAULT__ line: $_";
297 die "$y_file: duplicate __DEFAULT__ line: $_"
298 if defined $default_token;
303 next unless /^%(token|type)/;
304 s/^%((token)|type)\s+<(\w+)>\s+//
305 or die "$y_file: unparseable token/type line: $_";
306 for (split ' ', $_) {
309 $just_tokens{$_} = $3;
314 *tokens = \%just_tokens; # perly.h needs this
315 die "$y_file: no __DEFAULT__ token defined\n" unless $default_token;
316 $types{$default_token} = 1;
318 $tablines =~ /^\Qstatic const char *const yytname[] =\E\n
323 or die "Can't extract yytname[] from table string\n";
325 $fields =~ s{"((?:[^"\\]|\\.)+)"}
327 (defined $tokens{$1} ? $tokens{$1} : $default_token)
329 $fields =~ s/, \s* (?:0|YY_NULL|YY_NULLPTR) \s* $//x
330 or die "make_type_tab: couldn't delete trailing ',0'\n";
333 "\ntypedef enum {\n\t"
334 . join(", ", map "toketype_$_", sort keys %types)
335 . "\n} toketypes;\n\n"
336 . "/* type of each token/terminal */\n"
337 . "static const toketypes yy_type_tab[] =\n{\n"
345 print "executing: @_\n";
349 die "failed to execute command '@_': $!\n";
352 die sprintf "command '@_' died with signal %d\n",
356 die sprintf "command '@_' exited with value %d\n", $? >> 8;