]>
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 | ||
65a99eca AD |
49 | static char *fetch_type_name (char *cp, char const **type_name, |
50 | location dollar_loc); | |
51 | ||
4210cd0b JD |
52 | static void handle_action_dollar (symbol_list *rule, char *cp, |
53 | location dollar_loc); | |
54 | static void handle_action_at (symbol_list *rule, char *cp, location at_loc); | |
7685e2f7 | 55 | |
d5e8574b | 56 | /* A string to be pushed to obstack after dollar/at has been handled. */ |
7685e2f7 AR |
57 | static char *ref_tail_fields; |
58 | ||
e9071366 AD |
59 | static location the_location; |
60 | static location *loc = &the_location; | |
ddc8ede1 | 61 | |
28e52c0d JD |
62 | /* A string representing the most recent translation. */ |
63 | static char *last_string; | |
64 | ||
ddc8ede1 PE |
65 | /* True if an untyped $$ or $n was seen. */ |
66 | static bool untyped_var_seen; | |
7685e2f7 | 67 | |
e9071366 AD |
68 | %} |
69 | /* C and C++ comments in code. */ | |
70 | %x SC_COMMENT SC_LINE_COMMENT | |
71 | /* Strings and characters in code. */ | |
72 | %x SC_STRING SC_CHARACTER | |
73 | /* Whether in a rule or symbol action. Specifies the translation | |
74 | of $ and @. */ | |
75 | %x SC_RULE_ACTION SC_SYMBOL_ACTION | |
76 | ||
77 | ||
78 | /* POSIX says that a tag must be both an id and a C union member, but | |
79 | historically almost any character is allowed in a tag. We disallow | |
80 | NUL and newline, as this simplifies our implementation. */ | |
81 | tag [^\0\n>]+ | |
82 | ||
83 | /* Zero or more instances of backslash-newline. Following GCC, allow | |
84 | white space between the backslash and the newline. */ | |
85 | splice (\\[ \f\t\v]*\n)* | |
86 | ||
7685e2f7 | 87 | /* C style identifier. Must start with letter. Will be used for |
d5e8574b AR |
88 | named symbol references. Shall be kept synchronized with |
89 | scan-gram.l "letter" and "id". */ | |
d236ad94 | 90 | letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_] |
eb8c66bb | 91 | id {letter}({letter}|[-0-9])* |
7685e2f7 AR |
92 | ref -?[0-9]+|{id}|"["{id}"]"|"$" |
93 | ||
e9071366 AD |
94 | %% |
95 | ||
96 | %{ | |
2346344a | 97 | /* Nesting level of the current code in braces. */ |
4f82b42a | 98 | int braces_level = 0; |
2346344a | 99 | |
42f4393a | 100 | /* Whether a semicolon is probably needed. |
4d24ffb7 AD |
101 | |
102 | The heuristic is that a semicolon is not needed after '{', '}', | |
103 | ';', or a C preprocessor directive, and that whitespaces and | |
104 | comments do not affect this flag. Note that '{' does not need a | |
105 | semicolon because of '{}'. A semicolon may be needed before a | |
26313726 AD |
106 | cpp directive, but don't bother. |
107 | ||
108 | While it is maintained in several start-conditions (factoring | |
109 | opportunities), it is meaningful only for SC_RULE_ACTION. */ | |
42f4393a DJ |
110 | bool need_semicolon = false; |
111 | ||
112 | /* Whether in a C preprocessor directive. Don't use a start condition | |
113 | for this because, at the end of strings and comments, we still need | |
114 | to know whether we're in a directive. */ | |
115 | bool in_cpp = false; | |
116 | ||
e9071366 AD |
117 | /* This scanner is special: it is invoked only once, henceforth |
118 | is expected to return only once. This initialization is | |
119 | therefore done once per action to translate. */ | |
4f82b42a PE |
120 | aver (sc_context == SC_SYMBOL_ACTION |
121 | || sc_context == SC_RULE_ACTION | |
122 | || sc_context == INITIAL); | |
e9071366 AD |
123 | BEGIN sc_context; |
124 | %} | |
125 | ||
126 | /*------------------------------------------------------------. | |
9874f80b | 127 | | Scanning a C comment. The initial '/ *' is already eaten. | |
e9071366 AD |
128 | `------------------------------------------------------------*/ |
129 | ||
130 | <SC_COMMENT> | |
131 | { | |
132 | "*"{splice}"/" STRING_GROW; BEGIN sc_context; | |
133 | } | |
134 | ||
135 | ||
136 | /*--------------------------------------------------------------. | |
9874f80b | 137 | | Scanning a line comment. The initial '//' is already eaten. | |
e9071366 AD |
138 | `--------------------------------------------------------------*/ |
139 | ||
140 | <SC_LINE_COMMENT> | |
141 | { | |
142 | "\n" STRING_GROW; BEGIN sc_context; | |
143 | {splice} STRING_GROW; | |
144 | } | |
145 | ||
146 | ||
147 | /*--------------------------------------------. | |
148 | | Scanning user-code characters and strings. | | |
149 | `--------------------------------------------*/ | |
150 | ||
151 | <SC_CHARACTER,SC_STRING> | |
152 | { | |
153 | {splice}|\\{splice}. STRING_GROW; | |
154 | } | |
155 | ||
156 | <SC_CHARACTER> | |
157 | { | |
158 | "'" STRING_GROW; BEGIN sc_context; | |
159 | } | |
160 | ||
161 | <SC_STRING> | |
162 | { | |
163 | "\"" STRING_GROW; BEGIN sc_context; | |
164 | } | |
165 | ||
166 | ||
4d24ffb7 AD |
167 | <SC_RULE_ACTION,SC_SYMBOL_ACTION> |
168 | { | |
e9071366 AD |
169 | "'" { |
170 | STRING_GROW; | |
171 | BEGIN SC_CHARACTER; | |
42f4393a | 172 | need_semicolon = true; |
e9071366 AD |
173 | } |
174 | "\"" { | |
175 | STRING_GROW; | |
176 | BEGIN SC_STRING; | |
42f4393a | 177 | need_semicolon = true; |
e9071366 AD |
178 | } |
179 | "/"{splice}"*" { | |
180 | STRING_GROW; | |
181 | BEGIN SC_COMMENT; | |
182 | } | |
183 | "/"{splice}"/" { | |
184 | STRING_GROW; | |
185 | BEGIN SC_LINE_COMMENT; | |
186 | } | |
26313726 AD |
187 | [$@] { |
188 | warn_at (*loc, _("stray '%s'"), yytext); | |
189 | obstack_escape (&obstack_for_string, yytext); | |
190 | need_semicolon = true; | |
191 | } | |
192 | [\[\]] { | |
193 | obstack_escape (&obstack_for_string, yytext); | |
194 | need_semicolon = true; | |
195 | } | |
e9071366 AD |
196 | } |
197 | ||
198 | <SC_RULE_ACTION> | |
199 | { | |
7685e2f7 | 200 | "$"("<"{tag}">")?{ref} { |
9b858541 | 201 | ref_tail_fields = NULL; |
28e52c0d | 202 | handle_action_dollar (self->rule, yytext, *loc); |
9b858541 | 203 | if (ref_tail_fields) |
7685e2f7 | 204 | obstack_sgrow (&obstack_for_string, ref_tail_fields); |
42f4393a | 205 | need_semicolon = true; |
28e52c0d | 206 | } |
7685e2f7 | 207 | "@"{ref} { |
9b858541 | 208 | ref_tail_fields = NULL; |
28e52c0d | 209 | handle_action_at (self->rule, yytext, *loc); |
9b858541 | 210 | if (ref_tail_fields) |
7685e2f7 | 211 | obstack_sgrow (&obstack_for_string, ref_tail_fields); |
42f4393a | 212 | need_semicolon = true; |
28e52c0d | 213 | } |
2346344a | 214 | |
42f4393a DJ |
215 | ";" STRING_GROW; need_semicolon = false; |
216 | "{" STRING_GROW; ++braces_level; need_semicolon = false; | |
2346344a | 217 | "}" { |
14da0cdd | 218 | bool outer_brace = --braces_level == 0; |
2346344a | 219 | |
9874f80b | 220 | /* As an undocumented Bison extension, append ';' before the last |
2346344a | 221 | brace in braced code, so that the user code can omit trailing |
9874f80b | 222 | ';'. But do not append ';' if emulating Yacc, since Yacc does |
42f4393a | 223 | not append one. */ |
7ed73f82 | 224 | if (outer_brace && !yacc_flag && language_prio == default_prio |
42f4393a DJ |
225 | && skeleton_prio == default_prio && need_semicolon && ! in_cpp) |
226 | { | |
9874f80b JM |
227 | warn_at (*loc, _("a ';' might be needed at the end of action code")); |
228 | warn_at (*loc, _("future versions of Bison will not add the ';'")); | |
229 | obstack_1grow (&obstack_for_string, ';'); | |
42f4393a | 230 | } |
2346344a AD |
231 | |
232 | STRING_GROW; | |
42f4393a | 233 | need_semicolon = false; |
2346344a | 234 | } |
42f4393a DJ |
235 | |
236 | /* Preprocessing directives should only be recognized at the beginning | |
237 | of lines, allowing whitespace including comments, but in C/C++, | |
9874f80b JM |
238 | '#' can only be the start of preprocessor directives or within |
239 | '#define' directives anyway, so don't bother with begin of line. */ | |
42f4393a DJ |
240 | "#" STRING_GROW; in_cpp = true; |
241 | ||
242 | {splice} STRING_GROW; | |
e6070f22 | 243 | [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; |
42f4393a | 244 | [ \t\f] STRING_GROW; |
62efdd2a JD |
245 | |
246 | /* YYFAIL is undocumented and was formally deprecated in Bison | |
247 | 2.4.2. */ | |
248 | YYFAIL { | |
249 | STRING_GROW; need_semicolon = true; | |
250 | warn_at (*loc, _("use of YYFAIL, which is deprecated and will be" | |
251 | " removed")); | |
252 | } | |
253 | ||
254 | /* The sole purpose of this is to make sure identifiers that merely | |
255 | contain YYFAIL don't produce the above warning. */ | |
256 | [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true; | |
257 | ||
258 | . STRING_GROW; need_semicolon = true; | |
e9071366 AD |
259 | } |
260 | ||
261 | <SC_SYMBOL_ACTION> | |
262 | { | |
4982f078 AD |
263 | "$"("<"{tag}">")?"$" { |
264 | const char *type_name = NULL; | |
265 | fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0; | |
266 | obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar("); | |
267 | obstack_quote (&obstack_for_string, type_name); | |
268 | obstack_sgrow (&obstack_for_string, ")["); | |
28e52c0d JD |
269 | self->is_value_used = true; |
270 | } | |
e785ccf7 JD |
271 | "@$" { |
272 | obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); | |
273 | locations_flag = true; | |
274 | } | |
e9071366 AD |
275 | } |
276 | ||
277 | ||
e9071366 AD |
278 | <*> |
279 | { | |
26313726 AD |
280 | /* Escape M4 quoting characters in C code. */ |
281 | [$@\[\]] obstack_escape (&obstack_for_string, yytext); | |
e9071366 | 282 | |
26313726 AD |
283 | /* By default, grow the string obstack with the input. */ |
284 | .|\n STRING_GROW; | |
e9071366 | 285 | |
26313726 AD |
286 | /* End of processing. */ |
287 | <<EOF>> STRING_FINISH; return last_string; | |
288 | } | |
e9071366 AD |
289 | |
290 | %% | |
291 | ||
7685e2f7 | 292 | static inline bool |
d5e8574b | 293 | is_dot_or_dash (char ch) |
7685e2f7 AR |
294 | { |
295 | return ch == '.' || ch == '-'; | |
296 | } | |
297 | ||
298 | static inline bool | |
d5e8574b | 299 | contains_dot_or_dash (const char* p) |
7685e2f7 | 300 | { |
d5e8574b AR |
301 | for (; *p; ++p) |
302 | if (is_dot_or_dash (*p)) | |
303 | return true; | |
304 | return false; | |
7685e2f7 AR |
305 | } |
306 | ||
d5e8574b | 307 | /* Defines a variant of a symbolic name resolution. */ |
7685e2f7 AR |
308 | typedef struct |
309 | { | |
310 | /* Index in symbol list. */ | |
1a323c2f | 311 | unsigned symbol_index; |
7685e2f7 AR |
312 | |
313 | /* Matched symbol id and loc. */ | |
314 | uniqstr id; | |
315 | location loc; | |
316 | ||
d5e8574b | 317 | /* Hiding named reference. */ |
7685e2f7 AR |
318 | named_ref* hidden_by; |
319 | ||
d5e8574b AR |
320 | /* Error flags. May contain zero (no errors) or |
321 | a combination of VARIANT_* values. */ | |
7685e2f7 AR |
322 | unsigned err; |
323 | } variant; | |
324 | ||
d5e8574b AR |
325 | /* Set when the variant refers to a symbol hidden |
326 | by an explicit symbol reference. */ | |
327 | #define VARIANT_HIDDEN (1 << 0) | |
328 | ||
329 | /* Set when the variant refers to a symbol containing | |
330 | dots or dashes. Will require explicit bracketing. */ | |
331 | #define VARIANT_BAD_BRACKETING (1 << 1) | |
332 | ||
333 | /* Set when the variant refers to a symbol which is | |
334 | not visible from current midrule. */ | |
335 | #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) | |
336 | ||
9b858541 | 337 | static variant *variant_table = NULL; |
7685e2f7 AR |
338 | static unsigned variant_table_size = 0; |
339 | static unsigned variant_count = 0; | |
340 | ||
341 | static variant * | |
992e874a | 342 | variant_table_grow (void) |
7685e2f7 AR |
343 | { |
344 | ++variant_count; | |
345 | if (variant_count > variant_table_size) | |
346 | { | |
347 | while (variant_count > variant_table_size) | |
348 | variant_table_size = 2 * variant_table_size + 3; | |
349 | variant_table = xnrealloc (variant_table, variant_table_size, | |
350 | sizeof *variant_table); | |
351 | } | |
352 | return &variant_table[variant_count - 1]; | |
353 | } | |
354 | ||
1e20ad11 | 355 | static void |
992e874a | 356 | variant_table_free (void) |
1e20ad11 | 357 | { |
992e874a | 358 | free (variant_table); |
9b858541 | 359 | variant_table = NULL; |
1e20ad11 AR |
360 | variant_table_size = variant_count = 0; |
361 | } | |
362 | ||
7685e2f7 | 363 | static char * |
d5e8574b | 364 | find_prefix_end (const char *prefix, char *begin, char *end) |
7685e2f7 AR |
365 | { |
366 | char *ptr = begin; | |
367 | ||
d5e8574b AR |
368 | for (; *prefix && ptr != end; ++prefix, ++ptr) |
369 | if (*prefix != *ptr) | |
370 | return 0; | |
7685e2f7 AR |
371 | |
372 | if (*prefix) | |
373 | return 0; | |
374 | ||
375 | return ptr; | |
376 | } | |
377 | ||
378 | static variant * | |
992e874a | 379 | variant_add (uniqstr id, location id_loc, unsigned symbol_index, |
14e568d7 | 380 | char *cp, char *cp_end, bool explicit_bracketing) |
7685e2f7 AR |
381 | { |
382 | char *prefix_end; | |
383 | ||
d5e8574b | 384 | prefix_end = find_prefix_end (id, cp, cp_end); |
7685e2f7 AR |
385 | if (prefix_end && |
386 | (prefix_end == cp_end || | |
14e568d7 | 387 | (!explicit_bracketing && is_dot_or_dash (*prefix_end)))) |
7685e2f7 | 388 | { |
d5e8574b | 389 | variant *r = variant_table_grow (); |
1a323c2f | 390 | r->symbol_index = symbol_index; |
7685e2f7 | 391 | r->id = id; |
992e874a | 392 | r->loc = id_loc; |
7685e2f7 AR |
393 | r->hidden_by = NULL; |
394 | r->err = 0; | |
395 | return r; | |
396 | } | |
397 | else | |
398 | return NULL; | |
399 | } | |
400 | ||
d5e8574b | 401 | static const char * |
9b858541 | 402 | get_at_spec (unsigned symbol_index) |
d5e8574b AR |
403 | { |
404 | static char at_buf[20]; | |
992e874a | 405 | if (symbol_index == 0) |
d5e8574b AR |
406 | strcpy (at_buf, "$$"); |
407 | else | |
992e874a | 408 | snprintf (at_buf, sizeof at_buf, "$%u", symbol_index); |
d5e8574b AR |
409 | return at_buf; |
410 | } | |
411 | ||
1a323c2f | 412 | static void |
14e568d7 | 413 | show_sub_messages (const char* cp, bool explicit_bracketing, |
348f5608 AR |
414 | int midrule_rhs_index, char dollar_or_at, |
415 | bool is_warning, unsigned indent) | |
1a323c2f AR |
416 | { |
417 | unsigned i; | |
418 | ||
419 | for (i = 0; i < variant_count; ++i) | |
420 | { | |
421 | const variant *var = &variant_table[i]; | |
422 | const char *at_spec = get_at_spec (var->symbol_index); | |
423 | ||
424 | if (var->err == 0) | |
425 | { | |
426 | if (is_warning) | |
348f5608 AR |
427 | warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
428 | dollar_or_at, var->id, at_spec); | |
1a323c2f | 429 | else |
348f5608 AR |
430 | complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
431 | dollar_or_at, var->id, at_spec); | |
1a323c2f AR |
432 | } |
433 | else | |
434 | { | |
435 | static struct obstack msg_buf; | |
14e568d7 | 436 | const char *tail = explicit_bracketing ? "" : |
1a323c2f AR |
437 | cp + strlen (var->id); |
438 | const char *id = var->hidden_by ? var->hidden_by->id : | |
439 | var->id; | |
440 | location id_loc = var->hidden_by ? var->hidden_by->loc : | |
441 | var->loc; | |
442 | ||
443 | /* Create the explanation message. */ | |
444 | obstack_init (&msg_buf); | |
445 | ||
aaf63e45 | 446 | obstack_printf (&msg_buf, _("possibly meant: %c"), dollar_or_at); |
1a323c2f | 447 | if (contains_dot_or_dash (id)) |
aaf63e45 | 448 | obstack_printf (&msg_buf, "[%s]", id); |
1a323c2f AR |
449 | else |
450 | obstack_sgrow (&msg_buf, id); | |
451 | obstack_sgrow (&msg_buf, tail); | |
452 | ||
453 | if (var->err & VARIANT_HIDDEN) | |
454 | { | |
aaf63e45 | 455 | obstack_printf (&msg_buf, _(", hiding %c"), dollar_or_at); |
1a323c2f | 456 | if (contains_dot_or_dash (var->id)) |
aaf63e45 | 457 | obstack_printf (&msg_buf, "[%s]", var->id); |
1a323c2f AR |
458 | else |
459 | obstack_sgrow (&msg_buf, var->id); | |
460 | obstack_sgrow (&msg_buf, tail); | |
461 | } | |
462 | ||
aaf63e45 | 463 | obstack_printf (&msg_buf, _(" at %s"), at_spec); |
1a323c2f AR |
464 | |
465 | if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) | |
348f5608 AR |
466 | { |
467 | const char *format = | |
468 | _(", cannot be accessed from mid-rule action at $%d"); | |
aaf63e45 | 469 | obstack_printf (&msg_buf, format, midrule_rhs_index); |
348f5608 | 470 | } |
1a323c2f AR |
471 | |
472 | obstack_1grow (&msg_buf, '\0'); | |
473 | if (is_warning) | |
348f5608 AR |
474 | warn_at_indent (id_loc, &indent, "%s", |
475 | (char *) obstack_finish (&msg_buf)); | |
1a323c2f | 476 | else |
348f5608 AR |
477 | complain_at_indent (id_loc, &indent, "%s", |
478 | (char *) obstack_finish (&msg_buf)); | |
1a323c2f AR |
479 | obstack_free (&msg_buf, 0); |
480 | } | |
481 | } | |
482 | } | |
483 | ||
d5e8574b AR |
484 | /* Returned from "parse_ref" when the reference |
485 | is inappropriate. */ | |
7685e2f7 | 486 | #define INVALID_REF (INT_MIN) |
d5e8574b AR |
487 | |
488 | /* Returned from "parse_ref" when the reference | |
1a323c2f | 489 | points to LHS ($$) of the current rule or midrule. */ |
7685e2f7 AR |
490 | #define LHS_REF (INT_MIN + 1) |
491 | ||
348f5608 AR |
492 | /* Sub-messages indent. */ |
493 | #define SUB_INDENT (4) | |
494 | ||
d5e8574b AR |
495 | /* Parse named or positional reference. In case of positional |
496 | references, can return negative values for $-n "deep" stack | |
497 | accesses. */ | |
7685e2f7 | 498 | static long int |
d5e8574b | 499 | parse_ref (char *cp, symbol_list *rule, int rule_length, |
992e874a | 500 | int midrule_rhs_index, char *text, location text_loc, |
d5e8574b | 501 | char dollar_or_at) |
7685e2f7 AR |
502 | { |
503 | symbol_list *l; | |
504 | char *cp_end; | |
14e568d7 | 505 | bool explicit_bracketing; |
d5e8574b | 506 | unsigned i; |
1a323c2f AR |
507 | unsigned valid_variants = 0; |
508 | unsigned valid_variant_index = 0; | |
7685e2f7 AR |
509 | |
510 | if ('$' == *cp) | |
511 | return LHS_REF; | |
512 | ||
d5e8574b | 513 | if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1)))) |
7685e2f7 AR |
514 | { |
515 | long int num = strtol (cp, &cp, 10); | |
516 | if (1 - INT_MAX + rule_length <= num && num <= rule_length) | |
517 | return num; | |
518 | else | |
519 | { | |
992e874a AR |
520 | complain_at (text_loc, _("integer out of range: %s"), |
521 | quote (text)); | |
7685e2f7 AR |
522 | return INVALID_REF; |
523 | } | |
524 | } | |
525 | ||
526 | if ('[' == *cp) | |
527 | { | |
7685e2f7 | 528 | /* Ignore the brackets. */ |
d5e8574b AR |
529 | char *p; |
530 | for (p = ++cp; *p != ']'; ++p) | |
531 | continue; | |
7685e2f7 | 532 | cp_end = p; |
d5e8574b | 533 | |
14e568d7 | 534 | explicit_bracketing = true; |
7685e2f7 AR |
535 | } |
536 | else | |
537 | { | |
7685e2f7 | 538 | /* Take all characters of the name. */ |
d5e8574b | 539 | char* p; |
7685e2f7 | 540 | for (p = cp; *p; ++p) |
d5e8574b | 541 | if (is_dot_or_dash (*p)) |
7685e2f7 AR |
542 | { |
543 | ref_tail_fields = p; | |
544 | break; | |
545 | } | |
d5e8574b AR |
546 | for (p = cp; *p; ++p) |
547 | continue; | |
7685e2f7 | 548 | cp_end = p; |
d5e8574b | 549 | |
14e568d7 | 550 | explicit_bracketing = false; |
7685e2f7 AR |
551 | } |
552 | ||
553 | /* Add all relevant variants. */ | |
d5e8574b | 554 | { |
992e874a | 555 | unsigned symbol_index; |
d5e8574b | 556 | variant_count = 0; |
992e874a AR |
557 | for (symbol_index = 0, l = rule; !symbol_list_null (l); |
558 | ++symbol_index, l = l->next) | |
d5e8574b | 559 | { |
992e874a | 560 | variant *var; |
d5e8574b AR |
561 | if (l->content_type != SYMLIST_SYMBOL) |
562 | continue; | |
7685e2f7 | 563 | |
992e874a | 564 | var = variant_add (l->content.sym->tag, l->sym_loc, |
14e568d7 | 565 | symbol_index, cp, cp_end, explicit_bracketing); |
992e874a AR |
566 | if (var && l->named_ref) |
567 | var->hidden_by = l->named_ref; | |
7685e2f7 | 568 | |
d5e8574b | 569 | if (l->named_ref) |
992e874a | 570 | variant_add (l->named_ref->id, l->named_ref->loc, |
14e568d7 | 571 | symbol_index, cp, cp_end, explicit_bracketing); |
d5e8574b AR |
572 | } |
573 | } | |
7685e2f7 AR |
574 | |
575 | /* Check errors. */ | |
7685e2f7 AR |
576 | for (i = 0; i < variant_count; ++i) |
577 | { | |
992e874a | 578 | variant *var = &variant_table[i]; |
1a323c2f | 579 | unsigned symbol_index = var->symbol_index; |
7685e2f7 AR |
580 | |
581 | /* Check visibility from mid-rule actions. */ | |
d5e8574b | 582 | if (midrule_rhs_index != 0 |
992e874a | 583 | && (symbol_index == 0 || midrule_rhs_index < symbol_index)) |
1a323c2f | 584 | var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE; |
7685e2f7 AR |
585 | |
586 | /* Check correct bracketing. */ | |
14e568d7 | 587 | if (!explicit_bracketing && contains_dot_or_dash (var->id)) |
1a323c2f | 588 | var->err |= VARIANT_BAD_BRACKETING; |
7685e2f7 AR |
589 | |
590 | /* Check using of hidden symbols. */ | |
992e874a | 591 | if (var->hidden_by) |
1a323c2f | 592 | var->err |= VARIANT_HIDDEN; |
7685e2f7 | 593 | |
992e874a | 594 | if (!var->err) |
1a323c2f AR |
595 | { |
596 | valid_variant_index = i; | |
597 | ++valid_variants; | |
598 | } | |
7685e2f7 AR |
599 | } |
600 | ||
1a323c2f | 601 | switch (valid_variants) |
7685e2f7 | 602 | { |
1a323c2f | 603 | case 0: |
4c56b131 AR |
604 | { |
605 | unsigned len = (explicit_bracketing || !ref_tail_fields) ? | |
606 | cp_end - cp : ref_tail_fields - cp; | |
348f5608 | 607 | unsigned indent = 0; |
4c56b131 | 608 | |
348f5608 AR |
609 | complain_at_indent (text_loc, &indent, _("invalid reference: %s"), |
610 | quote (text)); | |
611 | indent += SUB_INDENT; | |
8cbbf178 JD |
612 | if (len == 0) |
613 | { | |
614 | location sym_loc = text_loc; | |
615 | sym_loc.start.column += 1; | |
616 | sym_loc.end = sym_loc.start; | |
617 | const char *format = | |
9874f80b JM |
618 | _("syntax error after '%c', expecting integer, letter," |
619 | " '_', '[', or '$'"); | |
8cbbf178 JD |
620 | complain_at_indent (sym_loc, &indent, format, dollar_or_at); |
621 | } | |
622 | else if (midrule_rhs_index) | |
348f5608 AR |
623 | { |
624 | const char *format = | |
625 | _("symbol not found in production before $%d: %.*s"); | |
626 | complain_at_indent (rule->location, &indent, format, | |
627 | midrule_rhs_index, len, cp); | |
628 | } | |
4c56b131 | 629 | else |
348f5608 AR |
630 | { |
631 | const char *format = | |
632 | _("symbol not found in production: %.*s"); | |
633 | complain_at_indent (rule->location, &indent, format, | |
634 | len, cp); | |
635 | } | |
4c56b131 AR |
636 | |
637 | if (variant_count > 0) | |
14e568d7 | 638 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
348f5608 | 639 | dollar_or_at, false, indent); |
4c56b131 AR |
640 | return INVALID_REF; |
641 | } | |
1a323c2f AR |
642 | case 1: |
643 | { | |
348f5608 | 644 | unsigned indent = 0; |
1a323c2f AR |
645 | if (variant_count > 1) |
646 | { | |
348f5608 AR |
647 | warn_at_indent (text_loc, &indent, _("misleading reference: %s"), |
648 | quote (text)); | |
14e568d7 | 649 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
348f5608 | 650 | dollar_or_at, true, indent + SUB_INDENT); |
1a323c2f AR |
651 | } |
652 | { | |
653 | unsigned symbol_index = | |
654 | variant_table[valid_variant_index].symbol_index; | |
655 | return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index; | |
656 | } | |
657 | } | |
658 | case 2: | |
659 | default: | |
348f5608 AR |
660 | { |
661 | unsigned indent = 0; | |
662 | complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"), | |
663 | quote (text)); | |
664 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, | |
665 | dollar_or_at, false, indent + SUB_INDENT); | |
666 | return INVALID_REF; | |
667 | } | |
7685e2f7 AR |
668 | } |
669 | ||
1a323c2f | 670 | /* Not reachable. */ |
7685e2f7 AR |
671 | return INVALID_REF; |
672 | } | |
673 | ||
e9071366 AD |
674 | /* Keeps track of the maximum number of semantic values to the left of |
675 | a handle (those referenced by $0, $-1, etc.) are required by the | |
676 | semantic actions of this grammar. */ | |
677 | int max_left_semantic_context = 0; | |
678 | ||
679 | ||
65a99eca AD |
680 | /* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its |
681 | beginning (i.e., after the opening "<", and return the pointer | |
682 | immediately after it. */ | |
683 | ||
684 | static | |
685 | char * | |
686 | fetch_type_name (char *cp, char const **type_name, | |
687 | location dollar_loc) | |
688 | { | |
689 | if (*cp == '<') | |
690 | { | |
691 | *type_name = ++cp; | |
692 | while (*cp != '>') | |
693 | ++cp; | |
694 | ||
695 | /* The '>' symbol will be later replaced by '\0'. Original | |
696 | 'text' is needed for error messages. */ | |
697 | ++cp; | |
698 | if (untyped_var_seen) | |
699 | complain_at (dollar_loc, _("explicit type given in untyped grammar")); | |
700 | tag_seen = true; | |
701 | } | |
702 | return cp; | |
703 | } | |
704 | ||
e9071366 | 705 | /*------------------------------------------------------------------. |
9874f80b | 706 | | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | |
e9071366 AD |
707 | | | |
708 | | Possible inputs: $[<TYPENAME>]($|integer) | | |
709 | | | | |
710 | | Output to OBSTACK_FOR_STRING a reference to this semantic value. | | |
711 | `------------------------------------------------------------------*/ | |
712 | ||
713 | static void | |
4210cd0b | 714 | handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) |
e9071366 | 715 | { |
28e52c0d | 716 | char const *type_name = NULL; |
e9071366 | 717 | char *cp = text + 1; |
ffa4ba3a | 718 | symbol_list *effective_rule; |
d5e8574b AR |
719 | int effective_rule_length; |
720 | int n; | |
ffa4ba3a JD |
721 | |
722 | if (rule->midrule_parent_rule) | |
723 | { | |
724 | effective_rule = rule->midrule_parent_rule; | |
725 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
726 | } | |
727 | else | |
728 | { | |
729 | effective_rule = rule; | |
730 | effective_rule_length = symbol_list_length (rule->next); | |
731 | } | |
e9071366 AD |
732 | |
733 | /* Get the type name if explicit. */ | |
65a99eca | 734 | cp = fetch_type_name (cp, &type_name, dollar_loc); |
e9071366 | 735 | |
d5e8574b AR |
736 | n = parse_ref (cp, effective_rule, effective_rule_length, |
737 | rule->midrule_parent_rhs_index, text, dollar_loc, '$'); | |
7685e2f7 | 738 | |
65a99eca AD |
739 | /* End type_name. */ |
740 | if (type_name) | |
741 | cp[-1] = '\0'; | |
7685e2f7 AR |
742 | |
743 | switch (n) | |
e9071366 | 744 | { |
7685e2f7 AR |
745 | case INVALID_REF: |
746 | break; | |
747 | ||
748 | case LHS_REF: | |
e9071366 | 749 | if (!type_name) |
4210cd0b | 750 | type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0); |
ddc8ede1 PE |
751 | |
752 | if (!type_name) | |
9874f80b JM |
753 | { |
754 | if (union_seen | tag_seen) | |
755 | { | |
756 | if (rule->midrule_parent_rule) | |
757 | complain_at (dollar_loc, | |
4c787a31 | 758 | _("$$ for the midrule at $%d of %s" |
9874f80b JM |
759 | " has no declared type"), |
760 | rule->midrule_parent_rhs_index, | |
4c787a31 | 761 | quote (effective_rule->content.sym->tag)); |
9874f80b | 762 | else |
4c787a31 AD |
763 | complain_at (dollar_loc, _("$$ of %s has no declared type"), |
764 | quote (rule->content.sym->tag)); | |
9874f80b JM |
765 | } |
766 | else | |
767 | untyped_var_seen = true; | |
9874f80b | 768 | } |
ddc8ede1 | 769 | |
8617d87e AD |
770 | obstack_sgrow (&obstack_for_string, "]b4_lhs_value("); |
771 | obstack_quote (&obstack_for_string, type_name); | |
772 | obstack_sgrow (&obstack_for_string, ")["); | |
f6857bbf | 773 | rule->action_props.is_value_used = true; |
7685e2f7 AR |
774 | break; |
775 | ||
776 | default: | |
777 | if (max_left_semantic_context < 1 - n) | |
778 | max_left_semantic_context = 1 - n; | |
779 | if (!type_name && 0 < n) | |
780 | type_name = | |
781 | symbol_list_n_type_name_get (effective_rule, dollar_loc, n); | |
782 | if (!type_name) | |
9874f80b JM |
783 | { |
784 | if (union_seen | tag_seen) | |
4c787a31 AD |
785 | complain_at (dollar_loc, _("$%s of %s has no declared type"), |
786 | cp, quote (effective_rule->content.sym->tag)); | |
9874f80b JM |
787 | else |
788 | untyped_var_seen = true; | |
9874f80b | 789 | } |
7685e2f7 | 790 | |
aaf63e45 | 791 | obstack_printf (&obstack_for_string, |
8617d87e AD |
792 | "]b4_rhs_value(%d, %d, ", effective_rule_length, n); |
793 | obstack_quote (&obstack_for_string, type_name); | |
794 | obstack_sgrow (&obstack_for_string, ")["); | |
7685e2f7 AR |
795 | if (n > 0) |
796 | symbol_list_n_get (effective_rule, n)->action_props.is_value_used = | |
797 | true; | |
798 | break; | |
e9071366 AD |
799 | } |
800 | } | |
801 | ||
802 | ||
803 | /*------------------------------------------------------. | |
9874f80b | 804 | | TEXT is a location token (i.e., a '@...'). Output to | |
e9071366 AD |
805 | | OBSTACK_FOR_STRING a reference to this location. | |
806 | `------------------------------------------------------*/ | |
807 | ||
808 | static void | |
4210cd0b | 809 | handle_action_at (symbol_list *rule, char *text, location at_loc) |
e9071366 AD |
810 | { |
811 | char *cp = text + 1; | |
7685e2f7 | 812 | symbol_list *effective_rule; |
d5e8574b AR |
813 | int effective_rule_length; |
814 | int n; | |
7685e2f7 AR |
815 | |
816 | if (rule->midrule_parent_rule) | |
817 | { | |
818 | effective_rule = rule->midrule_parent_rule; | |
819 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
820 | } | |
821 | else | |
822 | { | |
823 | effective_rule = rule; | |
824 | effective_rule_length = symbol_list_length (rule->next); | |
825 | } | |
ffa4ba3a | 826 | |
4f82b42a | 827 | locations_flag = true; |
e9071366 | 828 | |
d5e8574b | 829 | n = parse_ref (cp, effective_rule, effective_rule_length, |
9b858541 | 830 | rule->midrule_parent_rhs_index, text, at_loc, '@'); |
7685e2f7 | 831 | switch (n) |
e9071366 | 832 | { |
7685e2f7 AR |
833 | case INVALID_REF: |
834 | break; | |
e9071366 | 835 | |
7685e2f7 AR |
836 | case LHS_REF: |
837 | obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); | |
838 | break; | |
839 | ||
840 | default: | |
aaf63e45 | 841 | obstack_printf (&obstack_for_string, "]b4_rhs_location(%d, %d)[", |
7685e2f7 AR |
842 | effective_rule_length, n); |
843 | break; | |
e9071366 AD |
844 | } |
845 | } | |
846 | ||
847 | ||
848 | /*-------------------------. | |
849 | | Initialize the scanner. | | |
850 | `-------------------------*/ | |
851 | ||
28e52c0d | 852 | /* Translate the dollars and ats in \a self, in the context \a sc_context |
ddc8ede1 | 853 | (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */ |
e9071366 | 854 | |
eb095650 | 855 | static char const * |
28e52c0d | 856 | translate_action (code_props *self, int sc_context) |
e9071366 | 857 | { |
2ce4ed68 | 858 | char *res; |
e9071366 AD |
859 | static bool initialized = false; |
860 | if (!initialized) | |
861 | { | |
862 | obstack_init (&obstack_for_string); | |
e9071366 AD |
863 | yy_flex_debug = 0; |
864 | initialized = true; | |
865 | } | |
866 | ||
28e52c0d JD |
867 | loc->start = loc->end = self->location.start; |
868 | yy_switch_to_buffer (yy_scan_string (self->code)); | |
869 | res = code_lex (self, sc_context); | |
e9071366 AD |
870 | yy_delete_buffer (YY_CURRENT_BUFFER); |
871 | ||
872 | return res; | |
873 | } | |
874 | ||
28e52c0d JD |
875 | /*------------------------------------------------------------------------. |
876 | | Implementation of the public interface as documented in "scan-code.h". | | |
877 | `------------------------------------------------------------------------*/ | |
878 | ||
879 | void | |
880 | code_props_none_init (code_props *self) | |
76290368 | 881 | { |
28e52c0d | 882 | *self = code_props_none; |
76290368 JD |
883 | } |
884 | ||
28e52c0d JD |
885 | code_props const code_props_none = CODE_PROPS_NONE_INIT; |
886 | ||
887 | void | |
d5e8574b AR |
888 | code_props_plain_init (code_props *self, char const *code, |
889 | location code_loc) | |
76290368 | 890 | { |
28e52c0d JD |
891 | self->kind = CODE_PROPS_PLAIN; |
892 | self->code = code; | |
893 | self->location = code_loc; | |
894 | self->is_value_used = false; | |
895 | self->rule = NULL; | |
7685e2f7 | 896 | self->named_ref = NULL; |
76290368 JD |
897 | } |
898 | ||
28e52c0d JD |
899 | void |
900 | code_props_symbol_action_init (code_props *self, char const *code, | |
901 | location code_loc) | |
e9071366 | 902 | { |
28e52c0d JD |
903 | self->kind = CODE_PROPS_SYMBOL_ACTION; |
904 | self->code = code; | |
905 | self->location = code_loc; | |
906 | self->is_value_used = false; | |
907 | self->rule = NULL; | |
7685e2f7 | 908 | self->named_ref = NULL; |
e9071366 AD |
909 | } |
910 | ||
28e52c0d JD |
911 | void |
912 | code_props_rule_action_init (code_props *self, char const *code, | |
7685e2f7 | 913 | location code_loc, symbol_list *rule, |
992e874a | 914 | named_ref *name) |
28e52c0d JD |
915 | { |
916 | self->kind = CODE_PROPS_RULE_ACTION; | |
917 | self->code = code; | |
918 | self->location = code_loc; | |
919 | self->is_value_used = false; | |
920 | self->rule = rule; | |
992e874a | 921 | self->named_ref = name; |
28e52c0d JD |
922 | } |
923 | ||
924 | void | |
925 | code_props_translate_code (code_props *self) | |
926 | { | |
927 | switch (self->kind) | |
928 | { | |
929 | case CODE_PROPS_NONE: | |
930 | break; | |
931 | case CODE_PROPS_PLAIN: | |
932 | self->code = translate_action (self, INITIAL); | |
933 | break; | |
934 | case CODE_PROPS_SYMBOL_ACTION: | |
935 | self->code = translate_action (self, SC_SYMBOL_ACTION); | |
936 | break; | |
937 | case CODE_PROPS_RULE_ACTION: | |
938 | self->code = translate_action (self, SC_RULE_ACTION); | |
939 | break; | |
940 | } | |
941 | } | |
942 | ||
943 | void | |
944 | code_scanner_last_string_free (void) | |
945 | { | |
946 | STRING_FREE; | |
947 | } | |
e9071366 AD |
948 | |
949 | void | |
950 | code_scanner_free (void) | |
951 | { | |
952 | obstack_free (&obstack_for_string, 0); | |
1e20ad11 AR |
953 | variant_table_free (); |
954 | ||
e9071366 | 955 | /* Reclaim Flex's buffers. */ |
580b8926 | 956 | yylex_destroy (); |
e9071366 | 957 | } |