]>
Commit | Line | Data |
---|---|---|
e9071366 AD |
1 | /* Bison Action Scanner -*- C -*- |
2 | ||
c932d613 | 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 | |
4521fcdf | 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 | |
30 | #include "flex-scanner.h" | |
f9bfc42a JD |
31 | |
32 | #include "complain.h" | |
e9071366 AD |
33 | #include "reader.h" |
34 | #include "getargs.h" | |
e9071366 | 35 | #include "scan-code.h" |
28e52c0d | 36 | #include "symlist.h" |
e9071366 | 37 | |
d5e8574b AR |
38 | #include <c-ctype.h> |
39 | #include <get-errno.h> | |
40 | #include <quote.h> | |
41 | ||
e9071366 AD |
42 | /* The current calling start condition: SC_RULE_ACTION or |
43 | SC_SYMBOL_ACTION. */ | |
28e52c0d | 44 | # define YY_DECL static char *code_lex (code_props *self, int sc_context) |
e9071366 AD |
45 | YY_DECL; |
46 | ||
47 | #define YY_USER_ACTION location_compute (loc, &loc->end, yytext, yyleng); | |
48 | ||
4210cd0b JD |
49 | static void handle_action_dollar (symbol_list *rule, char *cp, |
50 | location dollar_loc); | |
51 | static void handle_action_at (symbol_list *rule, char *cp, location at_loc); | |
7685e2f7 | 52 | |
d5e8574b | 53 | /* A string to be pushed to obstack after dollar/at has been handled. */ |
7685e2f7 AR |
54 | static char *ref_tail_fields; |
55 | ||
e9071366 AD |
56 | static location the_location; |
57 | static location *loc = &the_location; | |
ddc8ede1 | 58 | |
28e52c0d JD |
59 | /* A string representing the most recent translation. */ |
60 | static char *last_string; | |
61 | ||
ddc8ede1 PE |
62 | /* True if an untyped $$ or $n was seen. */ |
63 | static bool untyped_var_seen; | |
7685e2f7 | 64 | |
e9071366 AD |
65 | %} |
66 | /* C and C++ comments in code. */ | |
67 | %x SC_COMMENT SC_LINE_COMMENT | |
68 | /* Strings and characters in code. */ | |
69 | %x SC_STRING SC_CHARACTER | |
70 | /* Whether in a rule or symbol action. Specifies the translation | |
71 | of $ and @. */ | |
72 | %x SC_RULE_ACTION SC_SYMBOL_ACTION | |
73 | ||
74 | ||
75 | /* POSIX says that a tag must be both an id and a C union member, but | |
76 | historically almost any character is allowed in a tag. We disallow | |
77 | NUL and newline, as this simplifies our implementation. */ | |
78 | tag [^\0\n>]+ | |
79 | ||
80 | /* Zero or more instances of backslash-newline. Following GCC, allow | |
81 | white space between the backslash and the newline. */ | |
82 | splice (\\[ \f\t\v]*\n)* | |
83 | ||
7685e2f7 | 84 | /* C style identifier. Must start with letter. Will be used for |
d5e8574b AR |
85 | named symbol references. Shall be kept synchronized with |
86 | scan-gram.l "letter" and "id". */ | |
d236ad94 | 87 | letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_] |
eb8c66bb | 88 | id {letter}({letter}|[-0-9])* |
7685e2f7 AR |
89 | ref -?[0-9]+|{id}|"["{id}"]"|"$" |
90 | ||
e9071366 AD |
91 | %% |
92 | ||
93 | %{ | |
2346344a | 94 | /* Nesting level of the current code in braces. */ |
4f82b42a | 95 | int braces_level = 0; |
2346344a | 96 | |
42f4393a | 97 | /* Whether a semicolon is probably needed. |
9874f80b | 98 | The heuristic is that a semicolon is not needed after '{', '}', ';', |
42f4393a DJ |
99 | or a C preprocessor directive, and that whitespaces and comments |
100 | do not affect this flag. | |
9874f80b | 101 | Note that '{' does not need a semicolon because of '{}'. |
42f4393a DJ |
102 | A semicolon may be needed before a cpp direcive, but don't bother. */ |
103 | bool need_semicolon = false; | |
104 | ||
105 | /* Whether in a C preprocessor directive. Don't use a start condition | |
106 | for this because, at the end of strings and comments, we still need | |
107 | to know whether we're in a directive. */ | |
108 | bool in_cpp = false; | |
109 | ||
e9071366 AD |
110 | /* This scanner is special: it is invoked only once, henceforth |
111 | is expected to return only once. This initialization is | |
112 | therefore done once per action to translate. */ | |
4f82b42a PE |
113 | aver (sc_context == SC_SYMBOL_ACTION |
114 | || sc_context == SC_RULE_ACTION | |
115 | || sc_context == INITIAL); | |
e9071366 AD |
116 | BEGIN sc_context; |
117 | %} | |
118 | ||
119 | /*------------------------------------------------------------. | |
9874f80b | 120 | | Scanning a C comment. The initial '/ *' is already eaten. | |
e9071366 AD |
121 | `------------------------------------------------------------*/ |
122 | ||
123 | <SC_COMMENT> | |
124 | { | |
125 | "*"{splice}"/" STRING_GROW; BEGIN sc_context; | |
126 | } | |
127 | ||
128 | ||
129 | /*--------------------------------------------------------------. | |
9874f80b | 130 | | Scanning a line comment. The initial '//' is already eaten. | |
e9071366 AD |
131 | `--------------------------------------------------------------*/ |
132 | ||
133 | <SC_LINE_COMMENT> | |
134 | { | |
135 | "\n" STRING_GROW; BEGIN sc_context; | |
136 | {splice} STRING_GROW; | |
137 | } | |
138 | ||
139 | ||
140 | /*--------------------------------------------. | |
141 | | Scanning user-code characters and strings. | | |
142 | `--------------------------------------------*/ | |
143 | ||
144 | <SC_CHARACTER,SC_STRING> | |
145 | { | |
146 | {splice}|\\{splice}. STRING_GROW; | |
147 | } | |
148 | ||
149 | <SC_CHARACTER> | |
150 | { | |
151 | "'" STRING_GROW; BEGIN sc_context; | |
152 | } | |
153 | ||
154 | <SC_STRING> | |
155 | { | |
156 | "\"" STRING_GROW; BEGIN sc_context; | |
157 | } | |
158 | ||
159 | ||
160 | <SC_RULE_ACTION,SC_SYMBOL_ACTION>{ | |
161 | "'" { | |
162 | STRING_GROW; | |
163 | BEGIN SC_CHARACTER; | |
42f4393a | 164 | need_semicolon = true; |
e9071366 AD |
165 | } |
166 | "\"" { | |
167 | STRING_GROW; | |
168 | BEGIN SC_STRING; | |
42f4393a | 169 | need_semicolon = true; |
e9071366 AD |
170 | } |
171 | "/"{splice}"*" { | |
172 | STRING_GROW; | |
173 | BEGIN SC_COMMENT; | |
174 | } | |
175 | "/"{splice}"/" { | |
176 | STRING_GROW; | |
177 | BEGIN SC_LINE_COMMENT; | |
178 | } | |
179 | } | |
180 | ||
181 | <SC_RULE_ACTION> | |
182 | { | |
7685e2f7 AR |
183 | "$"("<"{tag}">")?{ref} { |
184 | ref_tail_fields = 0; | |
28e52c0d | 185 | handle_action_dollar (self->rule, yytext, *loc); |
d5e8574b | 186 | if (ref_tail_fields) { |
7685e2f7 AR |
187 | obstack_sgrow (&obstack_for_string, ref_tail_fields); |
188 | } | |
42f4393a | 189 | need_semicolon = true; |
28e52c0d | 190 | } |
7685e2f7 AR |
191 | "@"{ref} { |
192 | ref_tail_fields = 0; | |
28e52c0d | 193 | handle_action_at (self->rule, yytext, *loc); |
d5e8574b | 194 | if (ref_tail_fields) { |
7685e2f7 AR |
195 | obstack_sgrow (&obstack_for_string, ref_tail_fields); |
196 | } | |
42f4393a | 197 | need_semicolon = true; |
28e52c0d | 198 | } |
e9071366 | 199 | "$" { |
9874f80b | 200 | warn_at (*loc, _("stray '$'")); |
e9071366 | 201 | obstack_sgrow (&obstack_for_string, "$]["); |
42f4393a | 202 | need_semicolon = true; |
e9071366 AD |
203 | } |
204 | "@" { | |
9874f80b | 205 | warn_at (*loc, _("stray '@'")); |
e9071366 | 206 | obstack_sgrow (&obstack_for_string, "@@"); |
42f4393a DJ |
207 | need_semicolon = true; |
208 | } | |
209 | "[" { | |
210 | obstack_sgrow (&obstack_for_string, "@{"); | |
211 | need_semicolon = true; | |
212 | } | |
213 | "]" { | |
214 | obstack_sgrow (&obstack_for_string, "@}"); | |
215 | need_semicolon = true; | |
e9071366 | 216 | } |
2346344a | 217 | |
42f4393a DJ |
218 | ";" STRING_GROW; need_semicolon = false; |
219 | "{" STRING_GROW; ++braces_level; need_semicolon = false; | |
2346344a | 220 | "}" { |
14da0cdd | 221 | bool outer_brace = --braces_level == 0; |
2346344a | 222 | |
9874f80b | 223 | /* As an undocumented Bison extension, append ';' before the last |
2346344a | 224 | brace in braced code, so that the user code can omit trailing |
9874f80b | 225 | ';'. But do not append ';' if emulating Yacc, since Yacc does |
42f4393a | 226 | not append one. */ |
7ed73f82 | 227 | if (outer_brace && !yacc_flag && language_prio == default_prio |
42f4393a DJ |
228 | && skeleton_prio == default_prio && need_semicolon && ! in_cpp) |
229 | { | |
9874f80b JM |
230 | warn_at (*loc, _("a ';' might be needed at the end of action code")); |
231 | warn_at (*loc, _("future versions of Bison will not add the ';'")); | |
232 | obstack_1grow (&obstack_for_string, ';'); | |
42f4393a | 233 | } |
2346344a AD |
234 | |
235 | STRING_GROW; | |
42f4393a | 236 | need_semicolon = false; |
2346344a | 237 | } |
42f4393a DJ |
238 | |
239 | /* Preprocessing directives should only be recognized at the beginning | |
240 | of lines, allowing whitespace including comments, but in C/C++, | |
9874f80b JM |
241 | '#' can only be the start of preprocessor directives or within |
242 | '#define' directives anyway, so don't bother with begin of line. */ | |
42f4393a DJ |
243 | "#" STRING_GROW; in_cpp = true; |
244 | ||
245 | {splice} STRING_GROW; | |
e6070f22 | 246 | [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; |
42f4393a | 247 | [ \t\f] STRING_GROW; |
62efdd2a JD |
248 | |
249 | /* YYFAIL is undocumented and was formally deprecated in Bison | |
250 | 2.4.2. */ | |
251 | YYFAIL { | |
252 | STRING_GROW; need_semicolon = true; | |
253 | warn_at (*loc, _("use of YYFAIL, which is deprecated and will be" | |
254 | " removed")); | |
255 | } | |
256 | ||
257 | /* The sole purpose of this is to make sure identifiers that merely | |
258 | contain YYFAIL don't produce the above warning. */ | |
259 | [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true; | |
260 | ||
261 | . STRING_GROW; need_semicolon = true; | |
e9071366 AD |
262 | } |
263 | ||
264 | <SC_SYMBOL_ACTION> | |
265 | { | |
28e52c0d JD |
266 | "$$" { |
267 | obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar["); | |
268 | self->is_value_used = true; | |
269 | } | |
e785ccf7 JD |
270 | "@$" { |
271 | obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); | |
272 | locations_flag = true; | |
273 | } | |
e9071366 AD |
274 | } |
275 | ||
276 | ||
277 | /*-----------------------------------------. | |
278 | | Escape M4 quoting characters in C code. | | |
279 | `-----------------------------------------*/ | |
280 | ||
281 | <*> | |
282 | { | |
283 | \$ obstack_sgrow (&obstack_for_string, "$]["); | |
284 | \@ obstack_sgrow (&obstack_for_string, "@@"); | |
285 | \[ obstack_sgrow (&obstack_for_string, "@{"); | |
286 | \] obstack_sgrow (&obstack_for_string, "@}"); | |
287 | } | |
288 | ||
289 | /*-----------------------------------------------------. | |
290 | | By default, grow the string obstack with the input. | | |
291 | `-----------------------------------------------------*/ | |
292 | ||
293 | <*>.|\n STRING_GROW; | |
294 | ||
295 | /* End of processing. */ | |
296 | <*><<EOF>> { | |
28e52c0d JD |
297 | STRING_FINISH; |
298 | return last_string; | |
e9071366 AD |
299 | } |
300 | ||
301 | %% | |
302 | ||
7685e2f7 | 303 | static inline bool |
d5e8574b | 304 | is_dot_or_dash (char ch) |
7685e2f7 AR |
305 | { |
306 | return ch == '.' || ch == '-'; | |
307 | } | |
308 | ||
309 | static inline bool | |
d5e8574b | 310 | contains_dot_or_dash (const char* p) |
7685e2f7 | 311 | { |
d5e8574b AR |
312 | for (; *p; ++p) |
313 | if (is_dot_or_dash (*p)) | |
314 | return true; | |
315 | return false; | |
7685e2f7 AR |
316 | } |
317 | ||
d5e8574b | 318 | /* Defines a variant of a symbolic name resolution. */ |
7685e2f7 AR |
319 | typedef struct |
320 | { | |
321 | /* Index in symbol list. */ | |
1a323c2f | 322 | unsigned symbol_index; |
7685e2f7 AR |
323 | |
324 | /* Matched symbol id and loc. */ | |
325 | uniqstr id; | |
326 | location loc; | |
327 | ||
d5e8574b | 328 | /* Hiding named reference. */ |
7685e2f7 AR |
329 | named_ref* hidden_by; |
330 | ||
d5e8574b AR |
331 | /* Error flags. May contain zero (no errors) or |
332 | a combination of VARIANT_* values. */ | |
7685e2f7 AR |
333 | unsigned err; |
334 | } variant; | |
335 | ||
d5e8574b AR |
336 | /* Set when the variant refers to a symbol hidden |
337 | by an explicit symbol reference. */ | |
338 | #define VARIANT_HIDDEN (1 << 0) | |
339 | ||
340 | /* Set when the variant refers to a symbol containing | |
341 | dots or dashes. Will require explicit bracketing. */ | |
342 | #define VARIANT_BAD_BRACKETING (1 << 1) | |
343 | ||
344 | /* Set when the variant refers to a symbol which is | |
345 | not visible from current midrule. */ | |
346 | #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) | |
347 | ||
7685e2f7 AR |
348 | static variant *variant_table = 0; |
349 | static unsigned variant_table_size = 0; | |
350 | static unsigned variant_count = 0; | |
351 | ||
352 | static variant * | |
992e874a | 353 | variant_table_grow (void) |
7685e2f7 AR |
354 | { |
355 | ++variant_count; | |
356 | if (variant_count > variant_table_size) | |
357 | { | |
358 | while (variant_count > variant_table_size) | |
359 | variant_table_size = 2 * variant_table_size + 3; | |
360 | variant_table = xnrealloc (variant_table, variant_table_size, | |
361 | sizeof *variant_table); | |
362 | } | |
363 | return &variant_table[variant_count - 1]; | |
364 | } | |
365 | ||
1e20ad11 | 366 | static void |
992e874a | 367 | variant_table_free (void) |
1e20ad11 | 368 | { |
992e874a | 369 | free (variant_table); |
1e20ad11 AR |
370 | variant_table = 0; |
371 | variant_table_size = variant_count = 0; | |
372 | } | |
373 | ||
7685e2f7 | 374 | static char * |
d5e8574b | 375 | find_prefix_end (const char *prefix, char *begin, char *end) |
7685e2f7 AR |
376 | { |
377 | char *ptr = begin; | |
378 | ||
d5e8574b AR |
379 | for (; *prefix && ptr != end; ++prefix, ++ptr) |
380 | if (*prefix != *ptr) | |
381 | return 0; | |
7685e2f7 AR |
382 | |
383 | if (*prefix) | |
384 | return 0; | |
385 | ||
386 | return ptr; | |
387 | } | |
388 | ||
389 | static variant * | |
992e874a | 390 | variant_add (uniqstr id, location id_loc, unsigned symbol_index, |
14e568d7 | 391 | char *cp, char *cp_end, bool explicit_bracketing) |
7685e2f7 AR |
392 | { |
393 | char *prefix_end; | |
394 | ||
d5e8574b | 395 | prefix_end = find_prefix_end (id, cp, cp_end); |
7685e2f7 AR |
396 | if (prefix_end && |
397 | (prefix_end == cp_end || | |
14e568d7 | 398 | (!explicit_bracketing && is_dot_or_dash (*prefix_end)))) |
7685e2f7 | 399 | { |
d5e8574b | 400 | variant *r = variant_table_grow (); |
1a323c2f | 401 | r->symbol_index = symbol_index; |
7685e2f7 | 402 | r->id = id; |
992e874a | 403 | r->loc = id_loc; |
7685e2f7 AR |
404 | r->hidden_by = NULL; |
405 | r->err = 0; | |
406 | return r; | |
407 | } | |
408 | else | |
409 | return NULL; | |
410 | } | |
411 | ||
d5e8574b | 412 | static const char * |
992e874a | 413 | get_at_spec(unsigned symbol_index) |
d5e8574b AR |
414 | { |
415 | static char at_buf[20]; | |
992e874a | 416 | if (symbol_index == 0) |
d5e8574b AR |
417 | strcpy (at_buf, "$$"); |
418 | else | |
992e874a | 419 | snprintf (at_buf, sizeof at_buf, "$%u", symbol_index); |
d5e8574b AR |
420 | return at_buf; |
421 | } | |
422 | ||
1a323c2f | 423 | static void |
14e568d7 | 424 | show_sub_messages (const char* cp, bool explicit_bracketing, |
348f5608 AR |
425 | int midrule_rhs_index, char dollar_or_at, |
426 | bool is_warning, unsigned indent) | |
1a323c2f AR |
427 | { |
428 | unsigned i; | |
429 | ||
430 | for (i = 0; i < variant_count; ++i) | |
431 | { | |
432 | const variant *var = &variant_table[i]; | |
433 | const char *at_spec = get_at_spec (var->symbol_index); | |
434 | ||
435 | if (var->err == 0) | |
436 | { | |
437 | if (is_warning) | |
348f5608 AR |
438 | warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
439 | dollar_or_at, var->id, at_spec); | |
1a323c2f | 440 | else |
348f5608 AR |
441 | complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
442 | dollar_or_at, var->id, at_spec); | |
1a323c2f AR |
443 | } |
444 | else | |
445 | { | |
446 | static struct obstack msg_buf; | |
14e568d7 | 447 | const char *tail = explicit_bracketing ? "" : |
1a323c2f AR |
448 | cp + strlen (var->id); |
449 | const char *id = var->hidden_by ? var->hidden_by->id : | |
450 | var->id; | |
451 | location id_loc = var->hidden_by ? var->hidden_by->loc : | |
452 | var->loc; | |
453 | ||
454 | /* Create the explanation message. */ | |
455 | obstack_init (&msg_buf); | |
456 | ||
348f5608 | 457 | obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at); |
1a323c2f AR |
458 | if (contains_dot_or_dash (id)) |
459 | obstack_fgrow1 (&msg_buf, "[%s]", id); | |
460 | else | |
461 | obstack_sgrow (&msg_buf, id); | |
462 | obstack_sgrow (&msg_buf, tail); | |
463 | ||
464 | if (var->err & VARIANT_HIDDEN) | |
465 | { | |
348f5608 | 466 | obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at); |
1a323c2f AR |
467 | if (contains_dot_or_dash (var->id)) |
468 | obstack_fgrow1 (&msg_buf, "[%s]", var->id); | |
469 | else | |
470 | obstack_sgrow (&msg_buf, var->id); | |
471 | obstack_sgrow (&msg_buf, tail); | |
472 | } | |
473 | ||
348f5608 | 474 | obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec); |
1a323c2f AR |
475 | |
476 | if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) | |
348f5608 AR |
477 | { |
478 | const char *format = | |
479 | _(", cannot be accessed from mid-rule action at $%d"); | |
480 | obstack_fgrow1 (&msg_buf, format, midrule_rhs_index); | |
481 | } | |
1a323c2f AR |
482 | |
483 | obstack_1grow (&msg_buf, '\0'); | |
484 | if (is_warning) | |
348f5608 AR |
485 | warn_at_indent (id_loc, &indent, "%s", |
486 | (char *) obstack_finish (&msg_buf)); | |
1a323c2f | 487 | else |
348f5608 AR |
488 | complain_at_indent (id_loc, &indent, "%s", |
489 | (char *) obstack_finish (&msg_buf)); | |
1a323c2f AR |
490 | obstack_free (&msg_buf, 0); |
491 | } | |
492 | } | |
493 | } | |
494 | ||
d5e8574b AR |
495 | /* Returned from "parse_ref" when the reference |
496 | is inappropriate. */ | |
7685e2f7 | 497 | #define INVALID_REF (INT_MIN) |
d5e8574b AR |
498 | |
499 | /* Returned from "parse_ref" when the reference | |
1a323c2f | 500 | points to LHS ($$) of the current rule or midrule. */ |
7685e2f7 AR |
501 | #define LHS_REF (INT_MIN + 1) |
502 | ||
348f5608 AR |
503 | /* Sub-messages indent. */ |
504 | #define SUB_INDENT (4) | |
505 | ||
d5e8574b AR |
506 | /* Parse named or positional reference. In case of positional |
507 | references, can return negative values for $-n "deep" stack | |
508 | accesses. */ | |
7685e2f7 | 509 | static long int |
d5e8574b | 510 | parse_ref (char *cp, symbol_list *rule, int rule_length, |
992e874a | 511 | int midrule_rhs_index, char *text, location text_loc, |
d5e8574b | 512 | char dollar_or_at) |
7685e2f7 AR |
513 | { |
514 | symbol_list *l; | |
515 | char *cp_end; | |
14e568d7 | 516 | bool explicit_bracketing; |
d5e8574b | 517 | unsigned i; |
1a323c2f AR |
518 | unsigned valid_variants = 0; |
519 | unsigned valid_variant_index = 0; | |
7685e2f7 AR |
520 | |
521 | if ('$' == *cp) | |
522 | return LHS_REF; | |
523 | ||
d5e8574b | 524 | if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1)))) |
7685e2f7 AR |
525 | { |
526 | long int num = strtol (cp, &cp, 10); | |
527 | if (1 - INT_MAX + rule_length <= num && num <= rule_length) | |
528 | return num; | |
529 | else | |
530 | { | |
992e874a AR |
531 | complain_at (text_loc, _("integer out of range: %s"), |
532 | quote (text)); | |
7685e2f7 AR |
533 | return INVALID_REF; |
534 | } | |
535 | } | |
536 | ||
537 | if ('[' == *cp) | |
538 | { | |
7685e2f7 | 539 | /* Ignore the brackets. */ |
d5e8574b AR |
540 | char *p; |
541 | for (p = ++cp; *p != ']'; ++p) | |
542 | continue; | |
7685e2f7 | 543 | cp_end = p; |
d5e8574b | 544 | |
14e568d7 | 545 | explicit_bracketing = true; |
7685e2f7 AR |
546 | } |
547 | else | |
548 | { | |
7685e2f7 | 549 | /* Take all characters of the name. */ |
d5e8574b | 550 | char* p; |
7685e2f7 | 551 | for (p = cp; *p; ++p) |
d5e8574b | 552 | if (is_dot_or_dash (*p)) |
7685e2f7 AR |
553 | { |
554 | ref_tail_fields = p; | |
555 | break; | |
556 | } | |
d5e8574b AR |
557 | for (p = cp; *p; ++p) |
558 | continue; | |
7685e2f7 | 559 | cp_end = p; |
d5e8574b | 560 | |
14e568d7 | 561 | explicit_bracketing = false; |
7685e2f7 AR |
562 | } |
563 | ||
564 | /* Add all relevant variants. */ | |
d5e8574b | 565 | { |
992e874a | 566 | unsigned symbol_index; |
d5e8574b | 567 | variant_count = 0; |
992e874a AR |
568 | for (symbol_index = 0, l = rule; !symbol_list_null (l); |
569 | ++symbol_index, l = l->next) | |
d5e8574b | 570 | { |
992e874a | 571 | variant *var; |
d5e8574b AR |
572 | if (l->content_type != SYMLIST_SYMBOL) |
573 | continue; | |
7685e2f7 | 574 | |
992e874a | 575 | var = variant_add (l->content.sym->tag, l->sym_loc, |
14e568d7 | 576 | symbol_index, cp, cp_end, explicit_bracketing); |
992e874a AR |
577 | if (var && l->named_ref) |
578 | var->hidden_by = l->named_ref; | |
7685e2f7 | 579 | |
d5e8574b | 580 | if (l->named_ref) |
992e874a | 581 | variant_add (l->named_ref->id, l->named_ref->loc, |
14e568d7 | 582 | symbol_index, cp, cp_end, explicit_bracketing); |
d5e8574b AR |
583 | } |
584 | } | |
7685e2f7 AR |
585 | |
586 | /* Check errors. */ | |
7685e2f7 AR |
587 | for (i = 0; i < variant_count; ++i) |
588 | { | |
992e874a | 589 | variant *var = &variant_table[i]; |
1a323c2f | 590 | unsigned symbol_index = var->symbol_index; |
7685e2f7 AR |
591 | |
592 | /* Check visibility from mid-rule actions. */ | |
d5e8574b | 593 | if (midrule_rhs_index != 0 |
992e874a | 594 | && (symbol_index == 0 || midrule_rhs_index < symbol_index)) |
1a323c2f | 595 | var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE; |
7685e2f7 AR |
596 | |
597 | /* Check correct bracketing. */ | |
14e568d7 | 598 | if (!explicit_bracketing && contains_dot_or_dash (var->id)) |
1a323c2f | 599 | var->err |= VARIANT_BAD_BRACKETING; |
7685e2f7 AR |
600 | |
601 | /* Check using of hidden symbols. */ | |
992e874a | 602 | if (var->hidden_by) |
1a323c2f | 603 | var->err |= VARIANT_HIDDEN; |
7685e2f7 | 604 | |
992e874a | 605 | if (!var->err) |
1a323c2f AR |
606 | { |
607 | valid_variant_index = i; | |
608 | ++valid_variants; | |
609 | } | |
7685e2f7 AR |
610 | } |
611 | ||
1a323c2f | 612 | switch (valid_variants) |
7685e2f7 | 613 | { |
1a323c2f | 614 | case 0: |
4c56b131 AR |
615 | { |
616 | unsigned len = (explicit_bracketing || !ref_tail_fields) ? | |
617 | cp_end - cp : ref_tail_fields - cp; | |
348f5608 | 618 | unsigned indent = 0; |
4c56b131 | 619 | |
348f5608 AR |
620 | complain_at_indent (text_loc, &indent, _("invalid reference: %s"), |
621 | quote (text)); | |
622 | indent += SUB_INDENT; | |
8cbbf178 JD |
623 | if (len == 0) |
624 | { | |
625 | location sym_loc = text_loc; | |
626 | sym_loc.start.column += 1; | |
627 | sym_loc.end = sym_loc.start; | |
628 | const char *format = | |
9874f80b JM |
629 | _("syntax error after '%c', expecting integer, letter," |
630 | " '_', '[', or '$'"); | |
8cbbf178 JD |
631 | complain_at_indent (sym_loc, &indent, format, dollar_or_at); |
632 | } | |
633 | else if (midrule_rhs_index) | |
348f5608 AR |
634 | { |
635 | const char *format = | |
636 | _("symbol not found in production before $%d: %.*s"); | |
637 | complain_at_indent (rule->location, &indent, format, | |
638 | midrule_rhs_index, len, cp); | |
639 | } | |
4c56b131 | 640 | else |
348f5608 AR |
641 | { |
642 | const char *format = | |
643 | _("symbol not found in production: %.*s"); | |
644 | complain_at_indent (rule->location, &indent, format, | |
645 | len, cp); | |
646 | } | |
4c56b131 AR |
647 | |
648 | if (variant_count > 0) | |
14e568d7 | 649 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
348f5608 | 650 | dollar_or_at, false, indent); |
4c56b131 AR |
651 | return INVALID_REF; |
652 | } | |
1a323c2f AR |
653 | case 1: |
654 | { | |
348f5608 | 655 | unsigned indent = 0; |
1a323c2f AR |
656 | if (variant_count > 1) |
657 | { | |
348f5608 AR |
658 | warn_at_indent (text_loc, &indent, _("misleading reference: %s"), |
659 | quote (text)); | |
14e568d7 | 660 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
348f5608 | 661 | dollar_or_at, true, indent + SUB_INDENT); |
1a323c2f AR |
662 | } |
663 | { | |
664 | unsigned symbol_index = | |
665 | variant_table[valid_variant_index].symbol_index; | |
666 | return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index; | |
667 | } | |
668 | } | |
669 | case 2: | |
670 | default: | |
348f5608 AR |
671 | { |
672 | unsigned indent = 0; | |
673 | complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"), | |
674 | quote (text)); | |
675 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, | |
676 | dollar_or_at, false, indent + SUB_INDENT); | |
677 | return INVALID_REF; | |
678 | } | |
7685e2f7 AR |
679 | } |
680 | ||
1a323c2f | 681 | /* Not reachable. */ |
7685e2f7 AR |
682 | return INVALID_REF; |
683 | } | |
684 | ||
e9071366 AD |
685 | /* Keeps track of the maximum number of semantic values to the left of |
686 | a handle (those referenced by $0, $-1, etc.) are required by the | |
687 | semantic actions of this grammar. */ | |
688 | int max_left_semantic_context = 0; | |
689 | ||
690 | ||
691 | /*------------------------------------------------------------------. | |
9874f80b | 692 | | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | |
e9071366 AD |
693 | | | |
694 | | Possible inputs: $[<TYPENAME>]($|integer) | | |
695 | | | | |
696 | | Output to OBSTACK_FOR_STRING a reference to this semantic value. | | |
697 | `------------------------------------------------------------------*/ | |
698 | ||
699 | static void | |
4210cd0b | 700 | handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) |
e9071366 | 701 | { |
28e52c0d | 702 | char const *type_name = NULL; |
e9071366 | 703 | char *cp = text + 1; |
7685e2f7 | 704 | char *gt_ptr = 0; |
ffa4ba3a | 705 | symbol_list *effective_rule; |
d5e8574b AR |
706 | int effective_rule_length; |
707 | int n; | |
ffa4ba3a JD |
708 | |
709 | if (rule->midrule_parent_rule) | |
710 | { | |
711 | effective_rule = rule->midrule_parent_rule; | |
712 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
713 | } | |
714 | else | |
715 | { | |
716 | effective_rule = rule; | |
717 | effective_rule_length = symbol_list_length (rule->next); | |
718 | } | |
e9071366 AD |
719 | |
720 | /* Get the type name if explicit. */ | |
721 | if (*cp == '<') | |
722 | { | |
723 | type_name = ++cp; | |
724 | while (*cp != '>') | |
725 | ++cp; | |
7685e2f7 AR |
726 | |
727 | /* The '>' symbol will be later replaced by '\0'. Original | |
728 | 'text' is needed for error messages. */ | |
729 | gt_ptr = cp; | |
e9071366 | 730 | ++cp; |
ddc8ede1 PE |
731 | if (untyped_var_seen) |
732 | complain_at (dollar_loc, _("explicit type given in untyped grammar")); | |
733 | tag_seen = true; | |
e9071366 AD |
734 | } |
735 | ||
d5e8574b AR |
736 | n = parse_ref (cp, effective_rule, effective_rule_length, |
737 | rule->midrule_parent_rhs_index, text, dollar_loc, '$'); | |
7685e2f7 AR |
738 | |
739 | if (gt_ptr) | |
740 | *gt_ptr = '\0'; | |
741 | ||
742 | switch (n) | |
e9071366 | 743 | { |
7685e2f7 AR |
744 | case INVALID_REF: |
745 | break; | |
746 | ||
747 | case LHS_REF: | |
e9071366 | 748 | if (!type_name) |
4210cd0b | 749 | type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0); |
ddc8ede1 PE |
750 | |
751 | if (!type_name) | |
9874f80b JM |
752 | { |
753 | if (union_seen | tag_seen) | |
754 | { | |
755 | if (rule->midrule_parent_rule) | |
756 | complain_at (dollar_loc, | |
4c787a31 | 757 | _("$$ for the midrule at $%d of %s" |
9874f80b JM |
758 | " has no declared type"), |
759 | rule->midrule_parent_rhs_index, | |
4c787a31 | 760 | quote (effective_rule->content.sym->tag)); |
9874f80b | 761 | else |
4c787a31 AD |
762 | complain_at (dollar_loc, _("$$ of %s has no declared type"), |
763 | quote (rule->content.sym->tag)); | |
9874f80b JM |
764 | } |
765 | else | |
766 | untyped_var_seen = true; | |
767 | type_name = ""; | |
768 | } | |
ddc8ede1 | 769 | |
e9071366 AD |
770 | obstack_fgrow1 (&obstack_for_string, |
771 | "]b4_lhs_value([%s])[", type_name); | |
f6857bbf | 772 | rule->action_props.is_value_used = true; |
7685e2f7 AR |
773 | break; |
774 | ||
775 | default: | |
776 | if (max_left_semantic_context < 1 - n) | |
777 | max_left_semantic_context = 1 - n; | |
778 | if (!type_name && 0 < n) | |
779 | type_name = | |
780 | symbol_list_n_type_name_get (effective_rule, dollar_loc, n); | |
781 | if (!type_name) | |
9874f80b JM |
782 | { |
783 | if (union_seen | tag_seen) | |
4c787a31 AD |
784 | complain_at (dollar_loc, _("$%s of %s has no declared type"), |
785 | cp, quote (effective_rule->content.sym->tag)); | |
9874f80b JM |
786 | else |
787 | untyped_var_seen = true; | |
788 | type_name = ""; | |
789 | } | |
7685e2f7 AR |
790 | |
791 | obstack_fgrow3 (&obstack_for_string, | |
792 | "]b4_rhs_value(%d, %d, [%s])[", | |
793 | effective_rule_length, n, type_name); | |
794 | if (n > 0) | |
795 | symbol_list_n_get (effective_rule, n)->action_props.is_value_used = | |
796 | true; | |
797 | break; | |
e9071366 AD |
798 | } |
799 | } | |
800 | ||
801 | ||
802 | /*------------------------------------------------------. | |
9874f80b | 803 | | TEXT is a location token (i.e., a '@...'). Output to | |
e9071366 AD |
804 | | OBSTACK_FOR_STRING a reference to this location. | |
805 | `------------------------------------------------------*/ | |
806 | ||
807 | static void | |
4210cd0b | 808 | handle_action_at (symbol_list *rule, char *text, location at_loc) |
e9071366 AD |
809 | { |
810 | char *cp = text + 1; | |
7685e2f7 | 811 | symbol_list *effective_rule; |
d5e8574b AR |
812 | int effective_rule_length; |
813 | int n; | |
7685e2f7 AR |
814 | |
815 | if (rule->midrule_parent_rule) | |
816 | { | |
817 | effective_rule = rule->midrule_parent_rule; | |
818 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
819 | } | |
820 | else | |
821 | { | |
822 | effective_rule = rule; | |
823 | effective_rule_length = symbol_list_length (rule->next); | |
824 | } | |
ffa4ba3a | 825 | |
4f82b42a | 826 | locations_flag = true; |
e9071366 | 827 | |
d5e8574b | 828 | n = parse_ref (cp, effective_rule, effective_rule_length, |
7685e2f7 AR |
829 | rule->midrule_parent_rhs_index, text, at_loc, '@'); |
830 | switch (n) | |
e9071366 | 831 | { |
7685e2f7 AR |
832 | case INVALID_REF: |
833 | break; | |
e9071366 | 834 | |
7685e2f7 AR |
835 | case LHS_REF: |
836 | obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); | |
837 | break; | |
838 | ||
839 | default: | |
840 | obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[", | |
841 | effective_rule_length, n); | |
842 | break; | |
e9071366 AD |
843 | } |
844 | } | |
845 | ||
846 | ||
847 | /*-------------------------. | |
848 | | Initialize the scanner. | | |
849 | `-------------------------*/ | |
850 | ||
28e52c0d | 851 | /* Translate the dollars and ats in \a self, in the context \a sc_context |
ddc8ede1 | 852 | (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */ |
e9071366 | 853 | |
eb095650 | 854 | static char const * |
28e52c0d | 855 | translate_action (code_props *self, int sc_context) |
e9071366 | 856 | { |
2ce4ed68 | 857 | char *res; |
e9071366 AD |
858 | static bool initialized = false; |
859 | if (!initialized) | |
860 | { | |
861 | obstack_init (&obstack_for_string); | |
e9071366 AD |
862 | yy_flex_debug = 0; |
863 | initialized = true; | |
864 | } | |
865 | ||
28e52c0d JD |
866 | loc->start = loc->end = self->location.start; |
867 | yy_switch_to_buffer (yy_scan_string (self->code)); | |
868 | res = code_lex (self, sc_context); | |
e9071366 AD |
869 | yy_delete_buffer (YY_CURRENT_BUFFER); |
870 | ||
871 | return res; | |
872 | } | |
873 | ||
28e52c0d JD |
874 | /*------------------------------------------------------------------------. |
875 | | Implementation of the public interface as documented in "scan-code.h". | | |
876 | `------------------------------------------------------------------------*/ | |
877 | ||
878 | void | |
879 | code_props_none_init (code_props *self) | |
76290368 | 880 | { |
28e52c0d | 881 | *self = code_props_none; |
76290368 JD |
882 | } |
883 | ||
28e52c0d JD |
884 | code_props const code_props_none = CODE_PROPS_NONE_INIT; |
885 | ||
886 | void | |
d5e8574b AR |
887 | code_props_plain_init (code_props *self, char const *code, |
888 | location code_loc) | |
76290368 | 889 | { |
28e52c0d JD |
890 | self->kind = CODE_PROPS_PLAIN; |
891 | self->code = code; | |
892 | self->location = code_loc; | |
893 | self->is_value_used = false; | |
894 | self->rule = NULL; | |
7685e2f7 | 895 | self->named_ref = NULL; |
76290368 JD |
896 | } |
897 | ||
28e52c0d JD |
898 | void |
899 | code_props_symbol_action_init (code_props *self, char const *code, | |
900 | location code_loc) | |
e9071366 | 901 | { |
28e52c0d JD |
902 | self->kind = CODE_PROPS_SYMBOL_ACTION; |
903 | self->code = code; | |
904 | self->location = code_loc; | |
905 | self->is_value_used = false; | |
906 | self->rule = NULL; | |
7685e2f7 | 907 | self->named_ref = NULL; |
e9071366 AD |
908 | } |
909 | ||
28e52c0d JD |
910 | void |
911 | code_props_rule_action_init (code_props *self, char const *code, | |
7685e2f7 | 912 | location code_loc, symbol_list *rule, |
992e874a | 913 | named_ref *name) |
28e52c0d JD |
914 | { |
915 | self->kind = CODE_PROPS_RULE_ACTION; | |
916 | self->code = code; | |
917 | self->location = code_loc; | |
918 | self->is_value_used = false; | |
919 | self->rule = rule; | |
992e874a | 920 | self->named_ref = name; |
28e52c0d JD |
921 | } |
922 | ||
923 | void | |
924 | code_props_translate_code (code_props *self) | |
925 | { | |
926 | switch (self->kind) | |
927 | { | |
928 | case CODE_PROPS_NONE: | |
929 | break; | |
930 | case CODE_PROPS_PLAIN: | |
931 | self->code = translate_action (self, INITIAL); | |
932 | break; | |
933 | case CODE_PROPS_SYMBOL_ACTION: | |
934 | self->code = translate_action (self, SC_SYMBOL_ACTION); | |
935 | break; | |
936 | case CODE_PROPS_RULE_ACTION: | |
937 | self->code = translate_action (self, SC_RULE_ACTION); | |
938 | break; | |
939 | } | |
940 | } | |
941 | ||
942 | void | |
943 | code_scanner_last_string_free (void) | |
944 | { | |
945 | STRING_FREE; | |
946 | } | |
e9071366 AD |
947 | |
948 | void | |
949 | code_scanner_free (void) | |
950 | { | |
951 | obstack_free (&obstack_for_string, 0); | |
1e20ad11 AR |
952 | variant_table_free (); |
953 | ||
e9071366 | 954 | /* Reclaim Flex's buffers. */ |
580b8926 | 955 | yylex_destroy (); |
e9071366 | 956 | } |