]>
Commit | Line | Data |
---|---|---|
e9071366 AD |
1 | /* Bison Action Scanner -*- C -*- |
2 | ||
34136e65 | 3 | Copyright (C) 2006-2012 Free Software Foundation, Inc. |
e9071366 AD |
4 | |
5 | This file is part of Bison, the GNU Compiler Compiler. | |
6 | ||
f16b0819 | 7 | This program is free software: you can redistribute it and/or modify |
e9071366 | 8 | it under the terms of the GNU General Public License as published by |
f16b0819 | 9 | the Free Software Foundation, either version 3 of the License, or |
e9071366 AD |
10 | (at your option) any later version. |
11 | ||
12 | This program is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
f16b0819 | 18 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
e9071366 | 19 | |
8d90395d | 20 | %option debug nodefault noinput nounput noyywrap never-interactive |
e9071366 AD |
21 | %option prefix="code_" outfile="lex.yy.c" |
22 | ||
23 | %{ | |
24 | /* Work around a bug in flex 2.5.31. See Debian bug 333231 | |
25 | <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=333231>. */ | |
26 | #undef code_wrap | |
27 | #define code_wrap() 1 | |
28 | ||
29 | #define FLEX_PREFIX(Id) code_ ## Id | |
0305d25e | 30 | #include <src/flex-scanner.h> |
f9bfc42a | 31 | |
0305d25e AD |
32 | #include <src/complain.h> |
33 | #include <src/reader.h> | |
34 | #include <src/getargs.h> | |
00f5d575 | 35 | #include <src/muscle-tab.h> |
0305d25e AD |
36 | #include <src/scan-code.h> |
37 | #include <src/symlist.h> | |
e9071366 | 38 | |
872b52bc AR |
39 | #include <c-ctype.h> |
40 | #include <get-errno.h> | |
41 | #include <quote.h> | |
42 | ||
e9071366 AD |
43 | /* The current calling start condition: SC_RULE_ACTION or |
44 | SC_SYMBOL_ACTION. */ | |
28e52c0d | 45 | # define YY_DECL static char *code_lex (code_props *self, int sc_context) |
e9071366 AD |
46 | YY_DECL; |
47 | ||
48 | #define YY_USER_ACTION location_compute (loc, &loc->end, yytext, yyleng); | |
49 | ||
65a99eca AD |
50 | static char *fetch_type_name (char *cp, char const **type_name, |
51 | location dollar_loc); | |
52 | ||
676997e5 | 53 | static void handle_action_dollar (symbol_list *rule, char *cp, |
e9690142 | 54 | location dollar_loc); |
676997e5 JD |
55 | static void handle_action_at (symbol_list *rule, char *cp, location at_loc); |
56 | ||
57 | /* A string to be pushed to obstack after dollar/at has been handled. */ | |
58 | static char *ref_tail_fields; | |
b9f1d9a4 | 59 | |
e9071366 AD |
60 | static location the_location; |
61 | static location *loc = &the_location; | |
ddc8ede1 | 62 | |
28e52c0d JD |
63 | /* A string representing the most recent translation. */ |
64 | static char *last_string; | |
65 | ||
ddc8ede1 PE |
66 | /* True if an untyped $$ or $n was seen. */ |
67 | static bool untyped_var_seen; | |
b9f1d9a4 | 68 | |
e9071366 AD |
69 | %} |
70 | /* C and C++ comments in code. */ | |
71 | %x SC_COMMENT SC_LINE_COMMENT | |
72 | /* Strings and characters in code. */ | |
73 | %x SC_STRING SC_CHARACTER | |
74 | /* Whether in a rule or symbol action. Specifies the translation | |
75 | of $ and @. */ | |
76 | %x SC_RULE_ACTION SC_SYMBOL_ACTION | |
77 | ||
78 | ||
79 | /* POSIX says that a tag must be both an id and a C union member, but | |
80 | historically almost any character is allowed in a tag. We disallow | |
81 | NUL and newline, as this simplifies our implementation. */ | |
e9690142 | 82 | tag [^\0\n>]+ |
e9071366 AD |
83 | |
84 | /* Zero or more instances of backslash-newline. Following GCC, allow | |
85 | white space between the backslash and the newline. */ | |
e9690142 | 86 | splice (\\[ \f\t\v]*\n)* |
e9071366 | 87 | |
676997e5 JD |
88 | /* C style identifier. Must start with letter. Will be used for |
89 | named symbol references. Shall be kept synchronized with | |
90 | scan-gram.l "letter" and "id". */ | |
e9690142 JD |
91 | letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_] |
92 | id {letter}({letter}|[-0-9])* | |
676997e5 | 93 | ref -?[0-9]+|{id}|"["{id}"]"|"$" |
b9f1d9a4 | 94 | |
e9071366 AD |
95 | %% |
96 | ||
97 | %{ | |
2346344a | 98 | /* Nesting level of the current code in braces. */ |
4f82b42a | 99 | int braces_level = 0; |
2346344a | 100 | |
e8cd1ad6 | 101 | /* Whether a semicolon is probably needed. |
4d24ffb7 AD |
102 | |
103 | The heuristic is that a semicolon is not needed after '{', '}', | |
104 | ';', or a C preprocessor directive, and that whitespaces and | |
105 | comments do not affect this flag. Note that '{' does not need a | |
106 | semicolon because of '{}'. A semicolon may be needed before a | |
26313726 AD |
107 | cpp directive, but don't bother. |
108 | ||
109 | While it is maintained in several start-conditions (factoring | |
110 | opportunities), it is meaningful only for SC_RULE_ACTION. */ | |
e8cd1ad6 DJ |
111 | bool need_semicolon = false; |
112 | ||
113 | /* Whether in a C preprocessor directive. Don't use a start condition | |
114 | for this because, at the end of strings and comments, we still need | |
115 | to know whether we're in a directive. */ | |
116 | bool in_cpp = false; | |
117 | ||
e9071366 AD |
118 | /* This scanner is special: it is invoked only once, henceforth |
119 | is expected to return only once. This initialization is | |
120 | therefore done once per action to translate. */ | |
4f82b42a | 121 | aver (sc_context == SC_SYMBOL_ACTION |
e9690142 JD |
122 | || sc_context == SC_RULE_ACTION |
123 | || sc_context == INITIAL); | |
e9071366 AD |
124 | BEGIN sc_context; |
125 | %} | |
126 | ||
127 | /*------------------------------------------------------------. | |
ae93e4e4 | 128 | | Scanning a C comment. The initial '/ *' is already eaten. | |
e9071366 AD |
129 | `------------------------------------------------------------*/ |
130 | ||
131 | <SC_COMMENT> | |
132 | { | |
133 | "*"{splice}"/" STRING_GROW; BEGIN sc_context; | |
134 | } | |
135 | ||
136 | ||
137 | /*--------------------------------------------------------------. | |
ae93e4e4 | 138 | | Scanning a line comment. The initial '//' is already eaten. | |
e9071366 AD |
139 | `--------------------------------------------------------------*/ |
140 | ||
141 | <SC_LINE_COMMENT> | |
142 | { | |
e9690142 JD |
143 | "\n" STRING_GROW; BEGIN sc_context; |
144 | {splice} STRING_GROW; | |
e9071366 AD |
145 | } |
146 | ||
147 | ||
148 | /*--------------------------------------------. | |
149 | | Scanning user-code characters and strings. | | |
150 | `--------------------------------------------*/ | |
151 | ||
152 | <SC_CHARACTER,SC_STRING> | |
153 | { | |
e9690142 | 154 | {splice}|\\{splice}. STRING_GROW; |
e9071366 AD |
155 | } |
156 | ||
157 | <SC_CHARACTER> | |
158 | { | |
e9690142 | 159 | "'" STRING_GROW; BEGIN sc_context; |
e9071366 AD |
160 | } |
161 | ||
162 | <SC_STRING> | |
163 | { | |
e9690142 | 164 | "\"" STRING_GROW; BEGIN sc_context; |
e9071366 AD |
165 | } |
166 | ||
167 | ||
4d24ffb7 AD |
168 | <SC_RULE_ACTION,SC_SYMBOL_ACTION> |
169 | { | |
e9071366 AD |
170 | "'" { |
171 | STRING_GROW; | |
172 | BEGIN SC_CHARACTER; | |
e8cd1ad6 | 173 | need_semicolon = true; |
e9071366 AD |
174 | } |
175 | "\"" { | |
176 | STRING_GROW; | |
177 | BEGIN SC_STRING; | |
e8cd1ad6 | 178 | need_semicolon = true; |
e9071366 AD |
179 | } |
180 | "/"{splice}"*" { | |
181 | STRING_GROW; | |
182 | BEGIN SC_COMMENT; | |
183 | } | |
184 | "/"{splice}"/" { | |
185 | STRING_GROW; | |
186 | BEGIN SC_LINE_COMMENT; | |
187 | } | |
26313726 | 188 | [$@] { |
4323e0da | 189 | complain_at (*loc, Wother, _("stray '%s'"), yytext); |
26313726 AD |
190 | obstack_escape (&obstack_for_string, yytext); |
191 | need_semicolon = true; | |
192 | } | |
193 | [\[\]] { | |
194 | obstack_escape (&obstack_for_string, yytext); | |
195 | need_semicolon = true; | |
196 | } | |
e9071366 AD |
197 | } |
198 | ||
199 | <SC_RULE_ACTION> | |
200 | { | |
b9f1d9a4 | 201 | "$"("<"{tag}">")?{ref} { |
9b858541 | 202 | ref_tail_fields = NULL; |
676997e5 | 203 | handle_action_dollar (self->rule, yytext, *loc); |
9b858541 | 204 | if (ref_tail_fields) |
676997e5 | 205 | obstack_sgrow (&obstack_for_string, ref_tail_fields); |
e8cd1ad6 | 206 | need_semicolon = true; |
28e52c0d | 207 | } |
b9f1d9a4 | 208 | "@"{ref} { |
9b858541 | 209 | ref_tail_fields = NULL; |
676997e5 | 210 | handle_action_at (self->rule, yytext, *loc); |
9b858541 | 211 | if (ref_tail_fields) |
676997e5 | 212 | obstack_sgrow (&obstack_for_string, ref_tail_fields); |
e8cd1ad6 | 213 | need_semicolon = true; |
e9071366 | 214 | } |
2346344a | 215 | |
e8cd1ad6 DJ |
216 | ";" STRING_GROW; need_semicolon = false; |
217 | "{" STRING_GROW; ++braces_level; need_semicolon = false; | |
2346344a | 218 | "}" { |
58bd33b7 | 219 | bool outer_brace = --braces_level == 0; |
2346344a | 220 | |
ae93e4e4 | 221 | /* As an undocumented Bison extension, append ';' before the last |
2346344a | 222 | brace in braced code, so that the user code can omit trailing |
ae93e4e4 | 223 | ';'. But do not append ';' if emulating Yacc, since Yacc does |
e8cd1ad6 | 224 | not append one. */ |
d9a9b96b | 225 | if (outer_brace && !yacc_flag && language_prio == default_prio |
e8cd1ad6 DJ |
226 | && skeleton_prio == default_prio && need_semicolon && ! in_cpp) |
227 | { | |
6fb8b256 VS |
228 | complain_at (*loc, Wother, |
229 | _("a ';' might be needed at the end of action code")); | |
230 | complain_at (*loc, Wother, | |
231 | _("future versions of Bison will not add the ';'")); | |
e9690142 | 232 | obstack_1grow (&obstack_for_string, ';'); |
e8cd1ad6 | 233 | } |
2346344a AD |
234 | |
235 | STRING_GROW; | |
e8cd1ad6 | 236 | need_semicolon = false; |
2346344a | 237 | } |
e8cd1ad6 DJ |
238 | |
239 | /* Preprocessing directives should only be recognized at the beginning | |
240 | of lines, allowing whitespace including comments, but in C/C++, | |
ae93e4e4 JM |
241 | '#' can only be the start of preprocessor directives or within |
242 | '#define' directives anyway, so don't bother with begin of line. */ | |
e8cd1ad6 DJ |
243 | "#" STRING_GROW; in_cpp = true; |
244 | ||
245 | {splice} STRING_GROW; | |
bc0f5737 | 246 | [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; |
e8cd1ad6 | 247 | [ \t\f] STRING_GROW; |
47db7ed1 | 248 | . STRING_GROW; need_semicolon = true; |
e9071366 AD |
249 | } |
250 | ||
251 | <SC_SYMBOL_ACTION> | |
252 | { | |
4982f078 AD |
253 | "$"("<"{tag}">")?"$" { |
254 | const char *type_name = NULL; | |
255 | fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0; | |
256 | obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar("); | |
257 | obstack_quote (&obstack_for_string, type_name); | |
258 | obstack_sgrow (&obstack_for_string, ")["); | |
28e52c0d JD |
259 | self->is_value_used = true; |
260 | } | |
e785ccf7 JD |
261 | "@$" { |
262 | obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); | |
bc0f5737 | 263 | muscle_percent_define_ensure("locations", the_location, true); |
e785ccf7 | 264 | } |
e9071366 AD |
265 | } |
266 | ||
267 | ||
e9071366 AD |
268 | <*> |
269 | { | |
26313726 AD |
270 | /* Escape M4 quoting characters in C code. */ |
271 | [$@\[\]] obstack_escape (&obstack_for_string, yytext); | |
e9071366 | 272 | |
26313726 AD |
273 | /* By default, grow the string obstack with the input. */ |
274 | .|\n STRING_GROW; | |
e9071366 AD |
275 | |
276 | /* End of processing. */ | |
26313726 AD |
277 | <<EOF>> STRING_FINISH; return last_string; |
278 | } | |
e9071366 AD |
279 | |
280 | %% | |
281 | ||
b9f1d9a4 | 282 | static inline bool |
872b52bc | 283 | is_dot_or_dash (char ch) |
b9f1d9a4 AR |
284 | { |
285 | return ch == '.' || ch == '-'; | |
286 | } | |
287 | ||
288 | static inline bool | |
872b52bc | 289 | contains_dot_or_dash (const char* p) |
b9f1d9a4 | 290 | { |
872b52bc AR |
291 | for (; *p; ++p) |
292 | if (is_dot_or_dash (*p)) | |
293 | return true; | |
294 | return false; | |
b9f1d9a4 AR |
295 | } |
296 | ||
872b52bc | 297 | /* Defines a variant of a symbolic name resolution. */ |
b9f1d9a4 AR |
298 | typedef struct |
299 | { | |
300 | /* Index in symbol list. */ | |
ce268795 | 301 | unsigned symbol_index; |
b9f1d9a4 AR |
302 | |
303 | /* Matched symbol id and loc. */ | |
304 | uniqstr id; | |
305 | location loc; | |
306 | ||
872b52bc | 307 | /* Hiding named reference. */ |
b9f1d9a4 AR |
308 | named_ref* hidden_by; |
309 | ||
872b52bc AR |
310 | /* Error flags. May contain zero (no errors) or |
311 | a combination of VARIANT_* values. */ | |
b9f1d9a4 AR |
312 | unsigned err; |
313 | } variant; | |
314 | ||
872b52bc AR |
315 | /* Set when the variant refers to a symbol hidden |
316 | by an explicit symbol reference. */ | |
317 | #define VARIANT_HIDDEN (1 << 0) | |
318 | ||
676997e5 JD |
319 | /* Set when the variant refers to a symbol containing |
320 | dots or dashes. Will require explicit bracketing. */ | |
321 | #define VARIANT_BAD_BRACKETING (1 << 1) | |
322 | ||
872b52bc AR |
323 | /* Set when the variant refers to a symbol which is |
324 | not visible from current midrule. */ | |
676997e5 | 325 | #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) |
872b52bc | 326 | |
9b858541 | 327 | static variant *variant_table = NULL; |
b9f1d9a4 AR |
328 | static unsigned variant_table_size = 0; |
329 | static unsigned variant_count = 0; | |
330 | ||
331 | static variant * | |
ba4184ec | 332 | variant_table_grow (void) |
b9f1d9a4 AR |
333 | { |
334 | ++variant_count; | |
335 | if (variant_count > variant_table_size) | |
336 | { | |
337 | while (variant_count > variant_table_size) | |
e9690142 | 338 | variant_table_size = 2 * variant_table_size + 3; |
b9f1d9a4 | 339 | variant_table = xnrealloc (variant_table, variant_table_size, |
e9690142 | 340 | sizeof *variant_table); |
b9f1d9a4 AR |
341 | } |
342 | return &variant_table[variant_count - 1]; | |
343 | } | |
344 | ||
13cdf208 | 345 | static void |
ba4184ec | 346 | variant_table_free (void) |
13cdf208 | 347 | { |
ba4184ec | 348 | free (variant_table); |
9b858541 | 349 | variant_table = NULL; |
13cdf208 AR |
350 | variant_table_size = variant_count = 0; |
351 | } | |
352 | ||
676997e5 JD |
353 | static char * |
354 | find_prefix_end (const char *prefix, char *begin, char *end) | |
b9f1d9a4 | 355 | { |
676997e5 JD |
356 | char *ptr = begin; |
357 | ||
358 | for (; *prefix && ptr != end; ++prefix, ++ptr) | |
359 | if (*prefix != *ptr) | |
360 | return 0; | |
361 | ||
362 | if (*prefix) | |
363 | return 0; | |
364 | ||
365 | return ptr; | |
b9f1d9a4 AR |
366 | } |
367 | ||
368 | static variant * | |
ba4184ec | 369 | variant_add (uniqstr id, location id_loc, unsigned symbol_index, |
e9690142 | 370 | char *cp, char *cp_end, bool explicit_bracketing) |
b9f1d9a4 | 371 | { |
676997e5 JD |
372 | char *prefix_end; |
373 | ||
374 | prefix_end = find_prefix_end (id, cp, cp_end); | |
375 | if (prefix_end && | |
376 | (prefix_end == cp_end || | |
377 | (!explicit_bracketing && is_dot_or_dash (*prefix_end)))) | |
b9f1d9a4 | 378 | { |
872b52bc | 379 | variant *r = variant_table_grow (); |
ce268795 | 380 | r->symbol_index = symbol_index; |
b9f1d9a4 | 381 | r->id = id; |
ba4184ec | 382 | r->loc = id_loc; |
b9f1d9a4 AR |
383 | r->hidden_by = NULL; |
384 | r->err = 0; | |
385 | return r; | |
386 | } | |
387 | else | |
388 | return NULL; | |
389 | } | |
390 | ||
872b52bc | 391 | static const char * |
ba4184ec | 392 | get_at_spec(unsigned symbol_index) |
872b52bc AR |
393 | { |
394 | static char at_buf[20]; | |
ba4184ec | 395 | if (symbol_index == 0) |
872b52bc AR |
396 | strcpy (at_buf, "$$"); |
397 | else | |
ba4184ec | 398 | snprintf (at_buf, sizeof at_buf, "$%u", symbol_index); |
872b52bc AR |
399 | return at_buf; |
400 | } | |
401 | ||
ce268795 | 402 | static void |
676997e5 JD |
403 | show_sub_messages (const char* cp, bool explicit_bracketing, |
404 | int midrule_rhs_index, char dollar_or_at, | |
66381412 | 405 | bool is_warning, unsigned indent) |
ce268795 AR |
406 | { |
407 | unsigned i; | |
408 | ||
409 | for (i = 0; i < variant_count; ++i) | |
410 | { | |
411 | const variant *var = &variant_table[i]; | |
412 | const char *at_spec = get_at_spec (var->symbol_index); | |
413 | ||
414 | if (var->err == 0) | |
415 | { | |
416 | if (is_warning) | |
6fb8b256 VS |
417 | complain_at_indent (var->loc, Wother, &indent, |
418 | _("refers to: %c%s at %s"), dollar_or_at, | |
419 | var->id, at_spec); | |
ce268795 | 420 | else |
6fb8b256 VS |
421 | complain_at_indent (var->loc, complaint, &indent, |
422 | _("refers to: %c%s at %s"), dollar_or_at, | |
423 | var->id, at_spec); | |
ce268795 AR |
424 | } |
425 | else | |
e9690142 JD |
426 | { |
427 | static struct obstack msg_buf; | |
428 | const char *tail = explicit_bracketing ? "" : | |
429 | cp + strlen (var->id); | |
430 | const char *id = var->hidden_by ? var->hidden_by->id : | |
431 | var->id; | |
432 | location id_loc = var->hidden_by ? var->hidden_by->loc : | |
433 | var->loc; | |
434 | ||
435 | /* Create the explanation message. */ | |
436 | obstack_init (&msg_buf); | |
437 | ||
cb9ec4fa | 438 | obstack_printf (&msg_buf, _("possibly meant: %c"), dollar_or_at); |
e9690142 | 439 | if (contains_dot_or_dash (id)) |
cb9ec4fa | 440 | obstack_printf (&msg_buf, "[%s]", id); |
e9690142 JD |
441 | else |
442 | obstack_sgrow (&msg_buf, id); | |
443 | obstack_sgrow (&msg_buf, tail); | |
444 | ||
445 | if (var->err & VARIANT_HIDDEN) | |
446 | { | |
cb9ec4fa | 447 | obstack_printf (&msg_buf, _(", hiding %c"), dollar_or_at); |
e9690142 | 448 | if (contains_dot_or_dash (var->id)) |
cb9ec4fa | 449 | obstack_printf (&msg_buf, "[%s]", var->id); |
e9690142 JD |
450 | else |
451 | obstack_sgrow (&msg_buf, var->id); | |
452 | obstack_sgrow (&msg_buf, tail); | |
453 | } | |
454 | ||
cb9ec4fa | 455 | obstack_printf (&msg_buf, _(" at %s"), at_spec); |
e9690142 JD |
456 | |
457 | if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) | |
66381412 AR |
458 | { |
459 | const char *format = | |
460 | _(", cannot be accessed from mid-rule action at $%d"); | |
aaf63e45 | 461 | obstack_printf (&msg_buf, format, midrule_rhs_index); |
66381412 | 462 | } |
ce268795 | 463 | |
e9690142 | 464 | obstack_1grow (&msg_buf, '\0'); |
ce268795 | 465 | if (is_warning) |
6fb8b256 VS |
466 | complain_at_indent (id_loc, Wother, &indent, "%s", |
467 | (char *) obstack_finish (&msg_buf)); | |
ce268795 | 468 | else |
6fb8b256 | 469 | complain_at_indent (id_loc, complaint, &indent, "%s", |
66381412 | 470 | (char *) obstack_finish (&msg_buf)); |
e9690142 JD |
471 | obstack_free (&msg_buf, 0); |
472 | } | |
ce268795 AR |
473 | } |
474 | } | |
475 | ||
872b52bc AR |
476 | /* Returned from "parse_ref" when the reference |
477 | is inappropriate. */ | |
b9f1d9a4 | 478 | #define INVALID_REF (INT_MIN) |
872b52bc AR |
479 | |
480 | /* Returned from "parse_ref" when the reference | |
ce268795 | 481 | points to LHS ($$) of the current rule or midrule. */ |
b9f1d9a4 AR |
482 | #define LHS_REF (INT_MIN + 1) |
483 | ||
66381412 AR |
484 | /* Sub-messages indent. */ |
485 | #define SUB_INDENT (4) | |
486 | ||
676997e5 JD |
487 | /* Parse named or positional reference. In case of positional |
488 | references, can return negative values for $-n "deep" stack | |
489 | accesses. */ | |
b9f1d9a4 | 490 | static long int |
676997e5 | 491 | parse_ref (char *cp, symbol_list *rule, int rule_length, |
e9690142 JD |
492 | int midrule_rhs_index, char *text, location text_loc, |
493 | char dollar_or_at) | |
b9f1d9a4 | 494 | { |
676997e5 JD |
495 | symbol_list *l; |
496 | char *cp_end; | |
61bc57e5 | 497 | bool explicit_bracketing; |
872b52bc | 498 | unsigned i; |
ce268795 AR |
499 | unsigned valid_variants = 0; |
500 | unsigned valid_variant_index = 0; | |
b9f1d9a4 AR |
501 | |
502 | if ('$' == *cp) | |
503 | return LHS_REF; | |
504 | ||
676997e5 | 505 | if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1)))) |
b9f1d9a4 | 506 | { |
676997e5 | 507 | long int num = strtol (cp, &cp, 10); |
b9f1d9a4 | 508 | if (1 - INT_MAX + rule_length <= num && num <= rule_length) |
e9690142 | 509 | return num; |
b9f1d9a4 | 510 | else |
e9690142 | 511 | { |
6fb8b256 | 512 | complain_at (text_loc, complaint, _("integer out of range: %s"), |
ba4184ec | 513 | quote (text)); |
e9690142 JD |
514 | return INVALID_REF; |
515 | } | |
b9f1d9a4 AR |
516 | } |
517 | ||
676997e5 JD |
518 | if ('[' == *cp) |
519 | { | |
520 | /* Ignore the brackets. */ | |
521 | char *p; | |
522 | for (p = ++cp; *p != ']'; ++p) | |
e9690142 | 523 | continue; |
676997e5 JD |
524 | cp_end = p; |
525 | ||
526 | explicit_bracketing = true; | |
527 | } | |
528 | else | |
529 | { | |
530 | /* Take all characters of the name. */ | |
531 | char* p; | |
532 | for (p = cp; *p; ++p) | |
e9690142 JD |
533 | if (is_dot_or_dash (*p)) |
534 | { | |
535 | ref_tail_fields = p; | |
536 | break; | |
537 | } | |
676997e5 | 538 | for (p = cp; *p; ++p) |
e9690142 | 539 | continue; |
676997e5 JD |
540 | cp_end = p; |
541 | ||
542 | explicit_bracketing = false; | |
543 | } | |
b9f1d9a4 AR |
544 | |
545 | /* Add all relevant variants. */ | |
872b52bc | 546 | { |
ba4184ec | 547 | unsigned symbol_index; |
872b52bc | 548 | variant_count = 0; |
ba4184ec AR |
549 | for (symbol_index = 0, l = rule; !symbol_list_null (l); |
550 | ++symbol_index, l = l->next) | |
872b52bc | 551 | { |
e9690142 JD |
552 | variant *var; |
553 | if (l->content_type != SYMLIST_SYMBOL) | |
554 | continue; | |
b9f1d9a4 | 555 | |
e9690142 | 556 | var = variant_add (l->content.sym->tag, l->sym_loc, |
676997e5 | 557 | symbol_index, cp, cp_end, explicit_bracketing); |
e9690142 JD |
558 | if (var && l->named_ref) |
559 | var->hidden_by = l->named_ref; | |
b9f1d9a4 | 560 | |
e9690142 JD |
561 | if (l->named_ref) |
562 | variant_add (l->named_ref->id, l->named_ref->loc, | |
676997e5 | 563 | symbol_index, cp, cp_end, explicit_bracketing); |
872b52bc AR |
564 | } |
565 | } | |
b9f1d9a4 AR |
566 | |
567 | /* Check errors. */ | |
b9f1d9a4 AR |
568 | for (i = 0; i < variant_count; ++i) |
569 | { | |
ba4184ec | 570 | variant *var = &variant_table[i]; |
ce268795 | 571 | unsigned symbol_index = var->symbol_index; |
b9f1d9a4 AR |
572 | |
573 | /* Check visibility from mid-rule actions. */ | |
872b52bc | 574 | if (midrule_rhs_index != 0 |
e9690142 | 575 | && (symbol_index == 0 || midrule_rhs_index < symbol_index)) |
ce268795 | 576 | var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE; |
b9f1d9a4 | 577 | |
676997e5 JD |
578 | /* Check correct bracketing. */ |
579 | if (!explicit_bracketing && contains_dot_or_dash (var->id)) | |
580 | var->err |= VARIANT_BAD_BRACKETING; | |
581 | ||
b9f1d9a4 | 582 | /* Check using of hidden symbols. */ |
ba4184ec | 583 | if (var->hidden_by) |
ce268795 | 584 | var->err |= VARIANT_HIDDEN; |
b9f1d9a4 | 585 | |
ba4184ec | 586 | if (!var->err) |
ce268795 AR |
587 | { |
588 | valid_variant_index = i; | |
589 | ++valid_variants; | |
590 | } | |
b9f1d9a4 AR |
591 | } |
592 | ||
ce268795 | 593 | switch (valid_variants) |
b9f1d9a4 | 594 | { |
ce268795 | 595 | case 0: |
f64e406b | 596 | { |
676997e5 JD |
597 | unsigned len = (explicit_bracketing || !ref_tail_fields) ? |
598 | cp_end - cp : ref_tail_fields - cp; | |
66381412 | 599 | unsigned indent = 0; |
f64e406b | 600 | |
6fb8b256 VS |
601 | complain_at_indent (text_loc, complaint, &indent, |
602 | _("invalid reference: %s"), quote (text)); | |
66381412 | 603 | indent += SUB_INDENT; |
5c9efc75 JD |
604 | if (len == 0) |
605 | { | |
606 | location sym_loc = text_loc; | |
607 | sym_loc.start.column += 1; | |
608 | sym_loc.end = sym_loc.start; | |
609 | const char *format = | |
ae93e4e4 JM |
610 | _("syntax error after '%c', expecting integer, letter," |
611 | " '_', '[', or '$'"); | |
6fb8b256 VS |
612 | complain_at_indent (sym_loc, complaint, &indent, format, |
613 | dollar_or_at); | |
5c9efc75 JD |
614 | } |
615 | else if (midrule_rhs_index) | |
66381412 AR |
616 | { |
617 | const char *format = | |
618 | _("symbol not found in production before $%d: %.*s"); | |
6fb8b256 | 619 | complain_at_indent (rule->location, complaint, &indent, format, |
66381412 AR |
620 | midrule_rhs_index, len, cp); |
621 | } | |
f64e406b | 622 | else |
66381412 AR |
623 | { |
624 | const char *format = | |
625 | _("symbol not found in production: %.*s"); | |
6fb8b256 | 626 | complain_at_indent (rule->location, complaint, &indent, format, |
66381412 AR |
627 | len, cp); |
628 | } | |
f64e406b AR |
629 | |
630 | if (variant_count > 0) | |
676997e5 JD |
631 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
632 | dollar_or_at, false, indent); | |
f64e406b AR |
633 | return INVALID_REF; |
634 | } | |
ce268795 AR |
635 | case 1: |
636 | { | |
66381412 | 637 | unsigned indent = 0; |
ce268795 AR |
638 | if (variant_count > 1) |
639 | { | |
6fb8b256 VS |
640 | complain_at_indent (text_loc, Wother, &indent, |
641 | _("misleading reference: %s"), quote (text)); | |
676997e5 JD |
642 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
643 | dollar_or_at, true, indent + SUB_INDENT); | |
ce268795 AR |
644 | } |
645 | { | |
646 | unsigned symbol_index = | |
647 | variant_table[valid_variant_index].symbol_index; | |
648 | return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index; | |
649 | } | |
650 | } | |
651 | case 2: | |
652 | default: | |
66381412 AR |
653 | { |
654 | unsigned indent = 0; | |
6fb8b256 VS |
655 | complain_at_indent (text_loc, complaint, &indent, |
656 | _("ambiguous reference: %s"), quote (text)); | |
676997e5 JD |
657 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
658 | dollar_or_at, false, indent + SUB_INDENT); | |
66381412 AR |
659 | return INVALID_REF; |
660 | } | |
b9f1d9a4 AR |
661 | } |
662 | ||
ce268795 | 663 | /* Not reachable. */ |
b9f1d9a4 AR |
664 | return INVALID_REF; |
665 | } | |
666 | ||
e9071366 AD |
667 | /* Keeps track of the maximum number of semantic values to the left of |
668 | a handle (those referenced by $0, $-1, etc.) are required by the | |
669 | semantic actions of this grammar. */ | |
670 | int max_left_semantic_context = 0; | |
671 | ||
672 | ||
65a99eca AD |
673 | /* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its |
674 | beginning (i.e., after the opening "<", and return the pointer | |
675 | immediately after it. */ | |
676 | ||
677 | static | |
678 | char * | |
679 | fetch_type_name (char *cp, char const **type_name, | |
680 | location dollar_loc) | |
681 | { | |
682 | if (*cp == '<') | |
683 | { | |
684 | *type_name = ++cp; | |
685 | while (*cp != '>') | |
4323e0da | 686 | ++cp; |
65a99eca AD |
687 | |
688 | /* The '>' symbol will be later replaced by '\0'. Original | |
4323e0da | 689 | 'text' is needed for error messages. */ |
65a99eca AD |
690 | ++cp; |
691 | if (untyped_var_seen) | |
4323e0da AD |
692 | complain_at (dollar_loc, complaint, |
693 | _("explicit type given in untyped grammar")); | |
65a99eca AD |
694 | tag_seen = true; |
695 | } | |
696 | return cp; | |
697 | } | |
698 | ||
e9071366 | 699 | /*------------------------------------------------------------------. |
ae93e4e4 | 700 | | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | |
e9071366 | 701 | | | |
676997e5 | 702 | | Possible inputs: $[<TYPENAME>]($|integer) | |
e9071366 AD |
703 | | | |
704 | | Output to OBSTACK_FOR_STRING a reference to this semantic value. | | |
705 | `------------------------------------------------------------------*/ | |
706 | ||
707 | static void | |
676997e5 | 708 | handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) |
e9071366 | 709 | { |
28e52c0d | 710 | char const *type_name = NULL; |
e9071366 | 711 | char *cp = text + 1; |
ffa4ba3a | 712 | symbol_list *effective_rule; |
872b52bc AR |
713 | int effective_rule_length; |
714 | int n; | |
ffa4ba3a JD |
715 | |
716 | if (rule->midrule_parent_rule) | |
717 | { | |
718 | effective_rule = rule->midrule_parent_rule; | |
719 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
720 | } | |
721 | else | |
722 | { | |
723 | effective_rule = rule; | |
724 | effective_rule_length = symbol_list_length (rule->next); | |
725 | } | |
e9071366 AD |
726 | |
727 | /* Get the type name if explicit. */ | |
65a99eca | 728 | cp = fetch_type_name (cp, &type_name, dollar_loc); |
e9071366 | 729 | |
872b52bc | 730 | n = parse_ref (cp, effective_rule, effective_rule_length, |
e9690142 | 731 | rule->midrule_parent_rhs_index, text, dollar_loc, '$'); |
b9f1d9a4 | 732 | |
65a99eca AD |
733 | /* End type_name. */ |
734 | if (type_name) | |
735 | cp[-1] = '\0'; | |
b9f1d9a4 AR |
736 | |
737 | switch (n) | |
e9071366 | 738 | { |
b9f1d9a4 AR |
739 | case INVALID_REF: |
740 | break; | |
741 | ||
742 | case LHS_REF: | |
e9071366 | 743 | if (!type_name) |
e9690142 | 744 | type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0); |
ddc8ede1 PE |
745 | |
746 | if (!type_name) | |
e9690142 JD |
747 | { |
748 | if (union_seen | tag_seen) | |
749 | { | |
750 | if (rule->midrule_parent_rule) | |
6fb8b256 | 751 | complain_at (dollar_loc, complaint, |
4a9cd8f2 | 752 | _("$$ for the midrule at $%d of %s" |
e9690142 JD |
753 | " has no declared type"), |
754 | rule->midrule_parent_rhs_index, | |
4a9cd8f2 | 755 | quote (effective_rule->content.sym->tag)); |
e9690142 | 756 | else |
6fb8b256 VS |
757 | complain_at (dollar_loc, complaint, |
758 | _("$$ of %s has no declared type"), | |
4a9cd8f2 | 759 | quote (rule->content.sym->tag)); |
e9690142 JD |
760 | } |
761 | else | |
762 | untyped_var_seen = true; | |
e9690142 | 763 | } |
ddc8ede1 | 764 | |
8617d87e AD |
765 | obstack_sgrow (&obstack_for_string, "]b4_lhs_value("); |
766 | obstack_quote (&obstack_for_string, type_name); | |
767 | obstack_sgrow (&obstack_for_string, ")["); | |
f6857bbf | 768 | rule->action_props.is_value_used = true; |
b9f1d9a4 AR |
769 | break; |
770 | ||
771 | default: | |
772 | if (max_left_semantic_context < 1 - n) | |
e9690142 | 773 | max_left_semantic_context = 1 - n; |
b9f1d9a4 | 774 | if (!type_name && 0 < n) |
e9690142 JD |
775 | type_name = |
776 | symbol_list_n_type_name_get (effective_rule, dollar_loc, n); | |
b9f1d9a4 | 777 | if (!type_name) |
e9690142 JD |
778 | { |
779 | if (union_seen | tag_seen) | |
6fb8b256 VS |
780 | complain_at (dollar_loc, complaint, |
781 | _("$%s of %s has no declared type"), cp, | |
782 | quote (effective_rule->content.sym->tag)); | |
e9690142 JD |
783 | else |
784 | untyped_var_seen = true; | |
e9690142 | 785 | } |
b9f1d9a4 | 786 | |
aaf63e45 | 787 | obstack_printf (&obstack_for_string, |
4323e0da | 788 | "]b4_rhs_value(%d, %d, ", effective_rule_length, n); |
8617d87e AD |
789 | obstack_quote (&obstack_for_string, type_name); |
790 | obstack_sgrow (&obstack_for_string, ")["); | |
b9f1d9a4 | 791 | if (n > 0) |
e9690142 JD |
792 | symbol_list_n_get (effective_rule, n)->action_props.is_value_used = |
793 | true; | |
b9f1d9a4 | 794 | break; |
e9071366 AD |
795 | } |
796 | } | |
797 | ||
798 | ||
799 | /*------------------------------------------------------. | |
ae93e4e4 | 800 | | TEXT is a location token (i.e., a '@...'). Output to | |
e9071366 AD |
801 | | OBSTACK_FOR_STRING a reference to this location. | |
802 | `------------------------------------------------------*/ | |
803 | ||
804 | static void | |
676997e5 | 805 | handle_action_at (symbol_list *rule, char *text, location at_loc) |
e9071366 | 806 | { |
676997e5 JD |
807 | char *cp = text + 1; |
808 | symbol_list *effective_rule; | |
872b52bc AR |
809 | int effective_rule_length; |
810 | int n; | |
b9f1d9a4 AR |
811 | |
812 | if (rule->midrule_parent_rule) | |
813 | { | |
814 | effective_rule = rule->midrule_parent_rule; | |
815 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
816 | } | |
817 | else | |
818 | { | |
819 | effective_rule = rule; | |
820 | effective_rule_length = symbol_list_length (rule->next); | |
821 | } | |
ffa4ba3a | 822 | |
bc0f5737 | 823 | muscle_percent_define_ensure("locations", at_loc, true); |
e9071366 | 824 | |
872b52bc | 825 | n = parse_ref (cp, effective_rule, effective_rule_length, |
e9690142 | 826 | rule->midrule_parent_rhs_index, text, at_loc, '@'); |
b9f1d9a4 | 827 | switch (n) |
e9071366 | 828 | { |
b9f1d9a4 AR |
829 | case INVALID_REF: |
830 | break; | |
e9071366 | 831 | |
b9f1d9a4 AR |
832 | case LHS_REF: |
833 | obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); | |
834 | break; | |
835 | ||
836 | default: | |
aaf63e45 | 837 | obstack_printf (&obstack_for_string, "]b4_rhs_location(%d, %d)[", |
e9690142 | 838 | effective_rule_length, n); |
b9f1d9a4 | 839 | break; |
e9071366 AD |
840 | } |
841 | } | |
842 | ||
843 | ||
844 | /*-------------------------. | |
845 | | Initialize the scanner. | | |
846 | `-------------------------*/ | |
847 | ||
28e52c0d | 848 | /* Translate the dollars and ats in \a self, in the context \a sc_context |
ddc8ede1 | 849 | (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */ |
e9071366 | 850 | |
eb095650 | 851 | static char const * |
28e52c0d | 852 | translate_action (code_props *self, int sc_context) |
e9071366 | 853 | { |
2ce4ed68 | 854 | char *res; |
e9071366 AD |
855 | static bool initialized = false; |
856 | if (!initialized) | |
857 | { | |
858 | obstack_init (&obstack_for_string); | |
e9071366 AD |
859 | yy_flex_debug = 0; |
860 | initialized = true; | |
861 | } | |
862 | ||
28e52c0d JD |
863 | loc->start = loc->end = self->location.start; |
864 | yy_switch_to_buffer (yy_scan_string (self->code)); | |
865 | res = code_lex (self, sc_context); | |
e9071366 AD |
866 | yy_delete_buffer (YY_CURRENT_BUFFER); |
867 | ||
868 | return res; | |
869 | } | |
870 | ||
28e52c0d JD |
871 | /*------------------------------------------------------------------------. |
872 | | Implementation of the public interface as documented in "scan-code.h". | | |
873 | `------------------------------------------------------------------------*/ | |
874 | ||
875 | void | |
876 | code_props_none_init (code_props *self) | |
76290368 | 877 | { |
28e52c0d | 878 | *self = code_props_none; |
76290368 JD |
879 | } |
880 | ||
70946cff | 881 | code_props code_props_none = CODE_PROPS_NONE_INIT; |
28e52c0d JD |
882 | |
883 | void | |
872b52bc | 884 | code_props_plain_init (code_props *self, char const *code, |
e9690142 | 885 | location code_loc) |
76290368 | 886 | { |
ea9a35c6 | 887 | code_props_none_init (self); |
28e52c0d JD |
888 | self->kind = CODE_PROPS_PLAIN; |
889 | self->code = code; | |
890 | self->location = code_loc; | |
76290368 JD |
891 | } |
892 | ||
28e52c0d JD |
893 | void |
894 | code_props_symbol_action_init (code_props *self, char const *code, | |
895 | location code_loc) | |
e9071366 | 896 | { |
ea9a35c6 | 897 | code_props_none_init (self); |
28e52c0d JD |
898 | self->kind = CODE_PROPS_SYMBOL_ACTION; |
899 | self->code = code; | |
900 | self->location = code_loc; | |
e9071366 AD |
901 | } |
902 | ||
28e52c0d JD |
903 | void |
904 | code_props_rule_action_init (code_props *self, char const *code, | |
b9f1d9a4 | 905 | location code_loc, symbol_list *rule, |
e9690142 | 906 | named_ref *name, bool is_predicate) |
28e52c0d | 907 | { |
ea9a35c6 | 908 | code_props_none_init (self); |
28e52c0d JD |
909 | self->kind = CODE_PROPS_RULE_ACTION; |
910 | self->code = code; | |
911 | self->location = code_loc; | |
28e52c0d | 912 | self->rule = rule; |
ba4184ec | 913 | self->named_ref = name; |
ca2a6d15 | 914 | self->is_predicate = is_predicate; |
28e52c0d JD |
915 | } |
916 | ||
917 | void | |
918 | code_props_translate_code (code_props *self) | |
919 | { | |
920 | switch (self->kind) | |
921 | { | |
922 | case CODE_PROPS_NONE: | |
923 | break; | |
924 | case CODE_PROPS_PLAIN: | |
925 | self->code = translate_action (self, INITIAL); | |
926 | break; | |
927 | case CODE_PROPS_SYMBOL_ACTION: | |
928 | self->code = translate_action (self, SC_SYMBOL_ACTION); | |
929 | break; | |
930 | case CODE_PROPS_RULE_ACTION: | |
931 | self->code = translate_action (self, SC_RULE_ACTION); | |
932 | break; | |
933 | } | |
934 | } | |
935 | ||
936 | void | |
937 | code_scanner_last_string_free (void) | |
938 | { | |
939 | STRING_FREE; | |
940 | } | |
e9071366 AD |
941 | |
942 | void | |
943 | code_scanner_free (void) | |
944 | { | |
945 | obstack_free (&obstack_for_string, 0); | |
13cdf208 AR |
946 | variant_table_free (); |
947 | ||
e9071366 | 948 | /* Reclaim Flex's buffers. */ |
580b8926 | 949 | yylex_destroy (); |
e9071366 | 950 | } |