]>
Commit | Line | Data |
---|---|---|
e9071366 AD |
1 | /* Bison Action Scanner -*- C -*- |
2 | ||
68f91d58 | 3 | Copyright (C) 2006-2013 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 |
d4e985d5 | 223 | not append one. This is deprecated since release 2.4.1. */ |
7ed73f82 | 224 | if (outer_brace && !yacc_flag && language_prio == default_prio |
42f4393a DJ |
225 | && skeleton_prio == default_prio && need_semicolon && ! in_cpp) |
226 | { | |
d4e985d5 TR |
227 | unsigned int indent = 0; |
228 | warn_at_indent (*loc, &indent, | |
229 | _("a ';' might be needed at the end of action code")); | |
230 | indent += SUB_INDENT; | |
231 | warn_at_indent (*loc, &indent, | |
232 | _("future versions of Bison will not add the ';'")); | |
9874f80b | 233 | obstack_1grow (&obstack_for_string, ';'); |
42f4393a | 234 | } |
2346344a AD |
235 | |
236 | STRING_GROW; | |
42f4393a | 237 | need_semicolon = false; |
2346344a | 238 | } |
42f4393a DJ |
239 | |
240 | /* Preprocessing directives should only be recognized at the beginning | |
241 | of lines, allowing whitespace including comments, but in C/C++, | |
9874f80b JM |
242 | '#' can only be the start of preprocessor directives or within |
243 | '#define' directives anyway, so don't bother with begin of line. */ | |
42f4393a DJ |
244 | "#" STRING_GROW; in_cpp = true; |
245 | ||
246 | {splice} STRING_GROW; | |
e6070f22 | 247 | [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; |
42f4393a | 248 | [ \t\f] STRING_GROW; |
62efdd2a JD |
249 | |
250 | /* YYFAIL is undocumented and was formally deprecated in Bison | |
251 | 2.4.2. */ | |
252 | YYFAIL { | |
253 | STRING_GROW; need_semicolon = true; | |
254 | warn_at (*loc, _("use of YYFAIL, which is deprecated and will be" | |
255 | " removed")); | |
256 | } | |
257 | ||
258 | /* The sole purpose of this is to make sure identifiers that merely | |
259 | contain YYFAIL don't produce the above warning. */ | |
260 | [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true; | |
261 | ||
262 | . STRING_GROW; need_semicolon = true; | |
e9071366 AD |
263 | } |
264 | ||
265 | <SC_SYMBOL_ACTION> | |
266 | { | |
4982f078 AD |
267 | "$"("<"{tag}">")?"$" { |
268 | const char *type_name = NULL; | |
269 | fetch_type_name (yytext + 1, &type_name, *loc)[-1] = 0; | |
270 | obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar("); | |
271 | obstack_quote (&obstack_for_string, type_name); | |
272 | obstack_sgrow (&obstack_for_string, ")["); | |
28e52c0d JD |
273 | self->is_value_used = true; |
274 | } | |
e785ccf7 JD |
275 | "@$" { |
276 | obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); | |
277 | locations_flag = true; | |
278 | } | |
e9071366 AD |
279 | } |
280 | ||
281 | ||
e9071366 AD |
282 | <*> |
283 | { | |
26313726 AD |
284 | /* Escape M4 quoting characters in C code. */ |
285 | [$@\[\]] obstack_escape (&obstack_for_string, yytext); | |
e9071366 | 286 | |
26313726 AD |
287 | /* By default, grow the string obstack with the input. */ |
288 | .|\n STRING_GROW; | |
e9071366 | 289 | |
26313726 AD |
290 | /* End of processing. */ |
291 | <<EOF>> STRING_FINISH; return last_string; | |
292 | } | |
e9071366 AD |
293 | |
294 | %% | |
295 | ||
7685e2f7 | 296 | static inline bool |
d5e8574b | 297 | is_dot_or_dash (char ch) |
7685e2f7 AR |
298 | { |
299 | return ch == '.' || ch == '-'; | |
300 | } | |
301 | ||
302 | static inline bool | |
d5e8574b | 303 | contains_dot_or_dash (const char* p) |
7685e2f7 | 304 | { |
d5e8574b AR |
305 | for (; *p; ++p) |
306 | if (is_dot_or_dash (*p)) | |
307 | return true; | |
308 | return false; | |
7685e2f7 AR |
309 | } |
310 | ||
d5e8574b | 311 | /* Defines a variant of a symbolic name resolution. */ |
7685e2f7 AR |
312 | typedef struct |
313 | { | |
314 | /* Index in symbol list. */ | |
1a323c2f | 315 | unsigned symbol_index; |
7685e2f7 AR |
316 | |
317 | /* Matched symbol id and loc. */ | |
318 | uniqstr id; | |
319 | location loc; | |
320 | ||
d5e8574b | 321 | /* Hiding named reference. */ |
7685e2f7 AR |
322 | named_ref* hidden_by; |
323 | ||
d5e8574b AR |
324 | /* Error flags. May contain zero (no errors) or |
325 | a combination of VARIANT_* values. */ | |
7685e2f7 AR |
326 | unsigned err; |
327 | } variant; | |
328 | ||
d5e8574b AR |
329 | /* Set when the variant refers to a symbol hidden |
330 | by an explicit symbol reference. */ | |
331 | #define VARIANT_HIDDEN (1 << 0) | |
332 | ||
333 | /* Set when the variant refers to a symbol containing | |
334 | dots or dashes. Will require explicit bracketing. */ | |
335 | #define VARIANT_BAD_BRACKETING (1 << 1) | |
336 | ||
337 | /* Set when the variant refers to a symbol which is | |
338 | not visible from current midrule. */ | |
339 | #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) | |
340 | ||
9b858541 | 341 | static variant *variant_table = NULL; |
7685e2f7 AR |
342 | static unsigned variant_table_size = 0; |
343 | static unsigned variant_count = 0; | |
344 | ||
345 | static variant * | |
992e874a | 346 | variant_table_grow (void) |
7685e2f7 AR |
347 | { |
348 | ++variant_count; | |
349 | if (variant_count > variant_table_size) | |
350 | { | |
351 | while (variant_count > variant_table_size) | |
352 | variant_table_size = 2 * variant_table_size + 3; | |
353 | variant_table = xnrealloc (variant_table, variant_table_size, | |
354 | sizeof *variant_table); | |
355 | } | |
356 | return &variant_table[variant_count - 1]; | |
357 | } | |
358 | ||
1e20ad11 | 359 | static void |
992e874a | 360 | variant_table_free (void) |
1e20ad11 | 361 | { |
992e874a | 362 | free (variant_table); |
9b858541 | 363 | variant_table = NULL; |
1e20ad11 AR |
364 | variant_table_size = variant_count = 0; |
365 | } | |
366 | ||
7685e2f7 | 367 | static char * |
d5e8574b | 368 | find_prefix_end (const char *prefix, char *begin, char *end) |
7685e2f7 AR |
369 | { |
370 | char *ptr = begin; | |
371 | ||
d5e8574b AR |
372 | for (; *prefix && ptr != end; ++prefix, ++ptr) |
373 | if (*prefix != *ptr) | |
374 | return 0; | |
7685e2f7 AR |
375 | |
376 | if (*prefix) | |
377 | return 0; | |
378 | ||
379 | return ptr; | |
380 | } | |
381 | ||
382 | static variant * | |
992e874a | 383 | variant_add (uniqstr id, location id_loc, unsigned symbol_index, |
14e568d7 | 384 | char *cp, char *cp_end, bool explicit_bracketing) |
7685e2f7 AR |
385 | { |
386 | char *prefix_end; | |
387 | ||
d5e8574b | 388 | prefix_end = find_prefix_end (id, cp, cp_end); |
7685e2f7 AR |
389 | if (prefix_end && |
390 | (prefix_end == cp_end || | |
14e568d7 | 391 | (!explicit_bracketing && is_dot_or_dash (*prefix_end)))) |
7685e2f7 | 392 | { |
d5e8574b | 393 | variant *r = variant_table_grow (); |
1a323c2f | 394 | r->symbol_index = symbol_index; |
7685e2f7 | 395 | r->id = id; |
992e874a | 396 | r->loc = id_loc; |
7685e2f7 AR |
397 | r->hidden_by = NULL; |
398 | r->err = 0; | |
399 | return r; | |
400 | } | |
401 | else | |
402 | return NULL; | |
403 | } | |
404 | ||
d5e8574b | 405 | static const char * |
9b858541 | 406 | get_at_spec (unsigned symbol_index) |
d5e8574b AR |
407 | { |
408 | static char at_buf[20]; | |
992e874a | 409 | if (symbol_index == 0) |
d5e8574b AR |
410 | strcpy (at_buf, "$$"); |
411 | else | |
992e874a | 412 | snprintf (at_buf, sizeof at_buf, "$%u", symbol_index); |
d5e8574b AR |
413 | return at_buf; |
414 | } | |
415 | ||
1a323c2f | 416 | static void |
14e568d7 | 417 | show_sub_messages (const char* cp, bool explicit_bracketing, |
348f5608 AR |
418 | int midrule_rhs_index, char dollar_or_at, |
419 | bool is_warning, unsigned indent) | |
1a323c2f AR |
420 | { |
421 | unsigned i; | |
422 | ||
423 | for (i = 0; i < variant_count; ++i) | |
424 | { | |
425 | const variant *var = &variant_table[i]; | |
426 | const char *at_spec = get_at_spec (var->symbol_index); | |
427 | ||
428 | if (var->err == 0) | |
429 | { | |
430 | if (is_warning) | |
348f5608 AR |
431 | warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
432 | dollar_or_at, var->id, at_spec); | |
1a323c2f | 433 | else |
348f5608 AR |
434 | complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
435 | dollar_or_at, var->id, at_spec); | |
1a323c2f AR |
436 | } |
437 | else | |
438 | { | |
439 | static struct obstack msg_buf; | |
14e568d7 | 440 | const char *tail = explicit_bracketing ? "" : |
1a323c2f AR |
441 | cp + strlen (var->id); |
442 | const char *id = var->hidden_by ? var->hidden_by->id : | |
443 | var->id; | |
444 | location id_loc = var->hidden_by ? var->hidden_by->loc : | |
445 | var->loc; | |
446 | ||
447 | /* Create the explanation message. */ | |
448 | obstack_init (&msg_buf); | |
449 | ||
aaf63e45 | 450 | obstack_printf (&msg_buf, _("possibly meant: %c"), dollar_or_at); |
1a323c2f | 451 | if (contains_dot_or_dash (id)) |
aaf63e45 | 452 | obstack_printf (&msg_buf, "[%s]", id); |
1a323c2f AR |
453 | else |
454 | obstack_sgrow (&msg_buf, id); | |
455 | obstack_sgrow (&msg_buf, tail); | |
456 | ||
457 | if (var->err & VARIANT_HIDDEN) | |
458 | { | |
aaf63e45 | 459 | obstack_printf (&msg_buf, _(", hiding %c"), dollar_or_at); |
1a323c2f | 460 | if (contains_dot_or_dash (var->id)) |
aaf63e45 | 461 | obstack_printf (&msg_buf, "[%s]", var->id); |
1a323c2f AR |
462 | else |
463 | obstack_sgrow (&msg_buf, var->id); | |
464 | obstack_sgrow (&msg_buf, tail); | |
465 | } | |
466 | ||
aaf63e45 | 467 | obstack_printf (&msg_buf, _(" at %s"), at_spec); |
1a323c2f AR |
468 | |
469 | if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) | |
348f5608 AR |
470 | { |
471 | const char *format = | |
472 | _(", cannot be accessed from mid-rule action at $%d"); | |
aaf63e45 | 473 | obstack_printf (&msg_buf, format, midrule_rhs_index); |
348f5608 | 474 | } |
1a323c2f AR |
475 | |
476 | obstack_1grow (&msg_buf, '\0'); | |
477 | if (is_warning) | |
348f5608 AR |
478 | warn_at_indent (id_loc, &indent, "%s", |
479 | (char *) obstack_finish (&msg_buf)); | |
1a323c2f | 480 | else |
348f5608 AR |
481 | complain_at_indent (id_loc, &indent, "%s", |
482 | (char *) obstack_finish (&msg_buf)); | |
1a323c2f AR |
483 | obstack_free (&msg_buf, 0); |
484 | } | |
485 | } | |
486 | } | |
487 | ||
d5e8574b AR |
488 | /* Returned from "parse_ref" when the reference |
489 | is inappropriate. */ | |
7685e2f7 | 490 | #define INVALID_REF (INT_MIN) |
d5e8574b AR |
491 | |
492 | /* Returned from "parse_ref" when the reference | |
1a323c2f | 493 | points to LHS ($$) of the current rule or midrule. */ |
7685e2f7 AR |
494 | #define LHS_REF (INT_MIN + 1) |
495 | ||
d5e8574b AR |
496 | /* Parse named or positional reference. In case of positional |
497 | references, can return negative values for $-n "deep" stack | |
498 | accesses. */ | |
7685e2f7 | 499 | static long int |
d5e8574b | 500 | parse_ref (char *cp, symbol_list *rule, int rule_length, |
992e874a | 501 | int midrule_rhs_index, char *text, location text_loc, |
d5e8574b | 502 | char dollar_or_at) |
7685e2f7 AR |
503 | { |
504 | symbol_list *l; | |
505 | char *cp_end; | |
14e568d7 | 506 | bool explicit_bracketing; |
d5e8574b | 507 | unsigned i; |
1a323c2f AR |
508 | unsigned valid_variants = 0; |
509 | unsigned valid_variant_index = 0; | |
7685e2f7 AR |
510 | |
511 | if ('$' == *cp) | |
512 | return LHS_REF; | |
513 | ||
d5e8574b | 514 | if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1)))) |
7685e2f7 AR |
515 | { |
516 | long int num = strtol (cp, &cp, 10); | |
517 | if (1 - INT_MAX + rule_length <= num && num <= rule_length) | |
518 | return num; | |
519 | else | |
520 | { | |
992e874a AR |
521 | complain_at (text_loc, _("integer out of range: %s"), |
522 | quote (text)); | |
7685e2f7 AR |
523 | return INVALID_REF; |
524 | } | |
525 | } | |
526 | ||
527 | if ('[' == *cp) | |
528 | { | |
7685e2f7 | 529 | /* Ignore the brackets. */ |
d5e8574b AR |
530 | char *p; |
531 | for (p = ++cp; *p != ']'; ++p) | |
532 | continue; | |
7685e2f7 | 533 | cp_end = p; |
d5e8574b | 534 | |
14e568d7 | 535 | explicit_bracketing = true; |
7685e2f7 AR |
536 | } |
537 | else | |
538 | { | |
7685e2f7 | 539 | /* Take all characters of the name. */ |
d5e8574b | 540 | char* p; |
7685e2f7 | 541 | for (p = cp; *p; ++p) |
d5e8574b | 542 | if (is_dot_or_dash (*p)) |
7685e2f7 AR |
543 | { |
544 | ref_tail_fields = p; | |
545 | break; | |
546 | } | |
d5e8574b AR |
547 | for (p = cp; *p; ++p) |
548 | continue; | |
7685e2f7 | 549 | cp_end = p; |
d5e8574b | 550 | |
14e568d7 | 551 | explicit_bracketing = false; |
7685e2f7 AR |
552 | } |
553 | ||
554 | /* Add all relevant variants. */ | |
d5e8574b | 555 | { |
992e874a | 556 | unsigned symbol_index; |
d5e8574b | 557 | variant_count = 0; |
992e874a AR |
558 | for (symbol_index = 0, l = rule; !symbol_list_null (l); |
559 | ++symbol_index, l = l->next) | |
d5e8574b | 560 | { |
992e874a | 561 | variant *var; |
d5e8574b AR |
562 | if (l->content_type != SYMLIST_SYMBOL) |
563 | continue; | |
7685e2f7 | 564 | |
992e874a | 565 | var = variant_add (l->content.sym->tag, l->sym_loc, |
14e568d7 | 566 | symbol_index, cp, cp_end, explicit_bracketing); |
992e874a AR |
567 | if (var && l->named_ref) |
568 | var->hidden_by = l->named_ref; | |
7685e2f7 | 569 | |
d5e8574b | 570 | if (l->named_ref) |
992e874a | 571 | variant_add (l->named_ref->id, l->named_ref->loc, |
14e568d7 | 572 | symbol_index, cp, cp_end, explicit_bracketing); |
d5e8574b AR |
573 | } |
574 | } | |
7685e2f7 AR |
575 | |
576 | /* Check errors. */ | |
7685e2f7 AR |
577 | for (i = 0; i < variant_count; ++i) |
578 | { | |
992e874a | 579 | variant *var = &variant_table[i]; |
1a323c2f | 580 | unsigned symbol_index = var->symbol_index; |
7685e2f7 AR |
581 | |
582 | /* Check visibility from mid-rule actions. */ | |
d5e8574b | 583 | if (midrule_rhs_index != 0 |
992e874a | 584 | && (symbol_index == 0 || midrule_rhs_index < symbol_index)) |
1a323c2f | 585 | var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE; |
7685e2f7 AR |
586 | |
587 | /* Check correct bracketing. */ | |
14e568d7 | 588 | if (!explicit_bracketing && contains_dot_or_dash (var->id)) |
1a323c2f | 589 | var->err |= VARIANT_BAD_BRACKETING; |
7685e2f7 AR |
590 | |
591 | /* Check using of hidden symbols. */ | |
992e874a | 592 | if (var->hidden_by) |
1a323c2f | 593 | var->err |= VARIANT_HIDDEN; |
7685e2f7 | 594 | |
992e874a | 595 | if (!var->err) |
1a323c2f AR |
596 | { |
597 | valid_variant_index = i; | |
598 | ++valid_variants; | |
599 | } | |
7685e2f7 AR |
600 | } |
601 | ||
1a323c2f | 602 | switch (valid_variants) |
7685e2f7 | 603 | { |
1a323c2f | 604 | case 0: |
4c56b131 AR |
605 | { |
606 | unsigned len = (explicit_bracketing || !ref_tail_fields) ? | |
607 | cp_end - cp : ref_tail_fields - cp; | |
348f5608 | 608 | unsigned indent = 0; |
4c56b131 | 609 | |
348f5608 AR |
610 | complain_at_indent (text_loc, &indent, _("invalid reference: %s"), |
611 | quote (text)); | |
612 | indent += SUB_INDENT; | |
8cbbf178 JD |
613 | if (len == 0) |
614 | { | |
615 | location sym_loc = text_loc; | |
616 | sym_loc.start.column += 1; | |
617 | sym_loc.end = sym_loc.start; | |
618 | const char *format = | |
9874f80b JM |
619 | _("syntax error after '%c', expecting integer, letter," |
620 | " '_', '[', or '$'"); | |
8cbbf178 JD |
621 | complain_at_indent (sym_loc, &indent, format, dollar_or_at); |
622 | } | |
623 | else if (midrule_rhs_index) | |
348f5608 AR |
624 | { |
625 | const char *format = | |
626 | _("symbol not found in production before $%d: %.*s"); | |
627 | complain_at_indent (rule->location, &indent, format, | |
628 | midrule_rhs_index, len, cp); | |
629 | } | |
4c56b131 | 630 | else |
348f5608 AR |
631 | { |
632 | const char *format = | |
633 | _("symbol not found in production: %.*s"); | |
634 | complain_at_indent (rule->location, &indent, format, | |
635 | len, cp); | |
636 | } | |
4c56b131 AR |
637 | |
638 | if (variant_count > 0) | |
14e568d7 | 639 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
348f5608 | 640 | dollar_or_at, false, indent); |
4c56b131 AR |
641 | return INVALID_REF; |
642 | } | |
1a323c2f AR |
643 | case 1: |
644 | { | |
348f5608 | 645 | unsigned indent = 0; |
1a323c2f AR |
646 | if (variant_count > 1) |
647 | { | |
348f5608 AR |
648 | warn_at_indent (text_loc, &indent, _("misleading reference: %s"), |
649 | quote (text)); | |
14e568d7 | 650 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
348f5608 | 651 | dollar_or_at, true, indent + SUB_INDENT); |
1a323c2f AR |
652 | } |
653 | { | |
654 | unsigned symbol_index = | |
655 | variant_table[valid_variant_index].symbol_index; | |
656 | return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index; | |
657 | } | |
658 | } | |
659 | case 2: | |
660 | default: | |
348f5608 AR |
661 | { |
662 | unsigned indent = 0; | |
663 | complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"), | |
664 | quote (text)); | |
665 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, | |
666 | dollar_or_at, false, indent + SUB_INDENT); | |
667 | return INVALID_REF; | |
668 | } | |
7685e2f7 AR |
669 | } |
670 | ||
1a323c2f | 671 | /* Not reachable. */ |
7685e2f7 AR |
672 | return INVALID_REF; |
673 | } | |
674 | ||
e9071366 AD |
675 | /* Keeps track of the maximum number of semantic values to the left of |
676 | a handle (those referenced by $0, $-1, etc.) are required by the | |
677 | semantic actions of this grammar. */ | |
678 | int max_left_semantic_context = 0; | |
679 | ||
680 | ||
65a99eca AD |
681 | /* If CP points to a typename (i.e., <.*?>), set TYPE_NAME to its |
682 | beginning (i.e., after the opening "<", and return the pointer | |
683 | immediately after it. */ | |
684 | ||
685 | static | |
686 | char * | |
687 | fetch_type_name (char *cp, char const **type_name, | |
688 | location dollar_loc) | |
689 | { | |
690 | if (*cp == '<') | |
691 | { | |
692 | *type_name = ++cp; | |
693 | while (*cp != '>') | |
694 | ++cp; | |
695 | ||
696 | /* The '>' symbol will be later replaced by '\0'. Original | |
697 | 'text' is needed for error messages. */ | |
698 | ++cp; | |
699 | if (untyped_var_seen) | |
700 | complain_at (dollar_loc, _("explicit type given in untyped grammar")); | |
701 | tag_seen = true; | |
702 | } | |
703 | return cp; | |
704 | } | |
705 | ||
e9071366 | 706 | /*------------------------------------------------------------------. |
9874f80b | 707 | | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | |
e9071366 AD |
708 | | | |
709 | | Possible inputs: $[<TYPENAME>]($|integer) | | |
710 | | | | |
711 | | Output to OBSTACK_FOR_STRING a reference to this semantic value. | | |
712 | `------------------------------------------------------------------*/ | |
713 | ||
714 | static void | |
4210cd0b | 715 | handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) |
e9071366 | 716 | { |
28e52c0d | 717 | char const *type_name = NULL; |
e9071366 | 718 | char *cp = text + 1; |
ffa4ba3a | 719 | symbol_list *effective_rule; |
d5e8574b AR |
720 | int effective_rule_length; |
721 | int n; | |
ffa4ba3a JD |
722 | |
723 | if (rule->midrule_parent_rule) | |
724 | { | |
725 | effective_rule = rule->midrule_parent_rule; | |
726 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
727 | } | |
728 | else | |
729 | { | |
730 | effective_rule = rule; | |
731 | effective_rule_length = symbol_list_length (rule->next); | |
732 | } | |
e9071366 AD |
733 | |
734 | /* Get the type name if explicit. */ | |
65a99eca | 735 | cp = fetch_type_name (cp, &type_name, dollar_loc); |
e9071366 | 736 | |
d5e8574b AR |
737 | n = parse_ref (cp, effective_rule, effective_rule_length, |
738 | rule->midrule_parent_rhs_index, text, dollar_loc, '$'); | |
7685e2f7 | 739 | |
65a99eca AD |
740 | /* End type_name. */ |
741 | if (type_name) | |
742 | cp[-1] = '\0'; | |
7685e2f7 AR |
743 | |
744 | switch (n) | |
e9071366 | 745 | { |
7685e2f7 AR |
746 | case INVALID_REF: |
747 | break; | |
748 | ||
749 | case LHS_REF: | |
e9071366 | 750 | if (!type_name) |
4210cd0b | 751 | type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0); |
ddc8ede1 PE |
752 | |
753 | if (!type_name) | |
9874f80b JM |
754 | { |
755 | if (union_seen | tag_seen) | |
756 | { | |
757 | if (rule->midrule_parent_rule) | |
758 | complain_at (dollar_loc, | |
4c787a31 | 759 | _("$$ for the midrule at $%d of %s" |
9874f80b JM |
760 | " has no declared type"), |
761 | rule->midrule_parent_rhs_index, | |
4c787a31 | 762 | quote (effective_rule->content.sym->tag)); |
9874f80b | 763 | else |
4c787a31 AD |
764 | complain_at (dollar_loc, _("$$ of %s has no declared type"), |
765 | quote (rule->content.sym->tag)); | |
9874f80b JM |
766 | } |
767 | else | |
768 | untyped_var_seen = true; | |
9874f80b | 769 | } |
ddc8ede1 | 770 | |
8617d87e AD |
771 | obstack_sgrow (&obstack_for_string, "]b4_lhs_value("); |
772 | obstack_quote (&obstack_for_string, type_name); | |
773 | obstack_sgrow (&obstack_for_string, ")["); | |
f6857bbf | 774 | rule->action_props.is_value_used = true; |
7685e2f7 AR |
775 | break; |
776 | ||
777 | default: | |
778 | if (max_left_semantic_context < 1 - n) | |
779 | max_left_semantic_context = 1 - n; | |
780 | if (!type_name && 0 < n) | |
781 | type_name = | |
782 | symbol_list_n_type_name_get (effective_rule, dollar_loc, n); | |
783 | if (!type_name) | |
9874f80b JM |
784 | { |
785 | if (union_seen | tag_seen) | |
4c787a31 AD |
786 | complain_at (dollar_loc, _("$%s of %s has no declared type"), |
787 | cp, quote (effective_rule->content.sym->tag)); | |
9874f80b JM |
788 | else |
789 | untyped_var_seen = true; | |
9874f80b | 790 | } |
7685e2f7 | 791 | |
aaf63e45 | 792 | obstack_printf (&obstack_for_string, |
8617d87e AD |
793 | "]b4_rhs_value(%d, %d, ", effective_rule_length, n); |
794 | obstack_quote (&obstack_for_string, type_name); | |
795 | obstack_sgrow (&obstack_for_string, ")["); | |
7685e2f7 AR |
796 | if (n > 0) |
797 | symbol_list_n_get (effective_rule, n)->action_props.is_value_used = | |
798 | true; | |
799 | break; | |
e9071366 AD |
800 | } |
801 | } | |
802 | ||
803 | ||
804 | /*------------------------------------------------------. | |
9874f80b | 805 | | TEXT is a location token (i.e., a '@...'). Output to | |
e9071366 AD |
806 | | OBSTACK_FOR_STRING a reference to this location. | |
807 | `------------------------------------------------------*/ | |
808 | ||
809 | static void | |
4210cd0b | 810 | handle_action_at (symbol_list *rule, char *text, location at_loc) |
e9071366 AD |
811 | { |
812 | char *cp = text + 1; | |
7685e2f7 | 813 | symbol_list *effective_rule; |
d5e8574b AR |
814 | int effective_rule_length; |
815 | int n; | |
7685e2f7 AR |
816 | |
817 | if (rule->midrule_parent_rule) | |
818 | { | |
819 | effective_rule = rule->midrule_parent_rule; | |
820 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
821 | } | |
822 | else | |
823 | { | |
824 | effective_rule = rule; | |
825 | effective_rule_length = symbol_list_length (rule->next); | |
826 | } | |
ffa4ba3a | 827 | |
4f82b42a | 828 | locations_flag = true; |
e9071366 | 829 | |
d5e8574b | 830 | n = parse_ref (cp, effective_rule, effective_rule_length, |
9b858541 | 831 | rule->midrule_parent_rhs_index, text, at_loc, '@'); |
7685e2f7 | 832 | switch (n) |
e9071366 | 833 | { |
7685e2f7 AR |
834 | case INVALID_REF: |
835 | break; | |
e9071366 | 836 | |
7685e2f7 AR |
837 | case LHS_REF: |
838 | obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); | |
839 | break; | |
840 | ||
841 | default: | |
aaf63e45 | 842 | obstack_printf (&obstack_for_string, "]b4_rhs_location(%d, %d)[", |
7685e2f7 AR |
843 | effective_rule_length, n); |
844 | break; | |
e9071366 AD |
845 | } |
846 | } | |
847 | ||
848 | ||
849 | /*-------------------------. | |
850 | | Initialize the scanner. | | |
851 | `-------------------------*/ | |
852 | ||
28e52c0d | 853 | /* Translate the dollars and ats in \a self, in the context \a sc_context |
ddc8ede1 | 854 | (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */ |
e9071366 | 855 | |
eb095650 | 856 | static char const * |
28e52c0d | 857 | translate_action (code_props *self, int sc_context) |
e9071366 | 858 | { |
2ce4ed68 | 859 | char *res; |
e9071366 AD |
860 | static bool initialized = false; |
861 | if (!initialized) | |
862 | { | |
863 | obstack_init (&obstack_for_string); | |
e9071366 AD |
864 | yy_flex_debug = 0; |
865 | initialized = true; | |
866 | } | |
867 | ||
28e52c0d JD |
868 | loc->start = loc->end = self->location.start; |
869 | yy_switch_to_buffer (yy_scan_string (self->code)); | |
870 | res = code_lex (self, sc_context); | |
e9071366 AD |
871 | yy_delete_buffer (YY_CURRENT_BUFFER); |
872 | ||
873 | return res; | |
874 | } | |
875 | ||
28e52c0d JD |
876 | /*------------------------------------------------------------------------. |
877 | | Implementation of the public interface as documented in "scan-code.h". | | |
878 | `------------------------------------------------------------------------*/ | |
879 | ||
880 | void | |
881 | code_props_none_init (code_props *self) | |
76290368 | 882 | { |
28e52c0d | 883 | *self = code_props_none; |
76290368 JD |
884 | } |
885 | ||
28e52c0d JD |
886 | code_props const code_props_none = CODE_PROPS_NONE_INIT; |
887 | ||
888 | void | |
d5e8574b AR |
889 | code_props_plain_init (code_props *self, char const *code, |
890 | location code_loc) | |
76290368 | 891 | { |
28e52c0d JD |
892 | self->kind = CODE_PROPS_PLAIN; |
893 | self->code = code; | |
894 | self->location = code_loc; | |
895 | self->is_value_used = false; | |
896 | self->rule = NULL; | |
7685e2f7 | 897 | self->named_ref = NULL; |
76290368 JD |
898 | } |
899 | ||
28e52c0d JD |
900 | void |
901 | code_props_symbol_action_init (code_props *self, char const *code, | |
902 | location code_loc) | |
e9071366 | 903 | { |
28e52c0d JD |
904 | self->kind = CODE_PROPS_SYMBOL_ACTION; |
905 | self->code = code; | |
906 | self->location = code_loc; | |
907 | self->is_value_used = false; | |
908 | self->rule = NULL; | |
7685e2f7 | 909 | self->named_ref = NULL; |
e9071366 AD |
910 | } |
911 | ||
28e52c0d JD |
912 | void |
913 | code_props_rule_action_init (code_props *self, char const *code, | |
7685e2f7 | 914 | location code_loc, symbol_list *rule, |
992e874a | 915 | named_ref *name) |
28e52c0d JD |
916 | { |
917 | self->kind = CODE_PROPS_RULE_ACTION; | |
918 | self->code = code; | |
919 | self->location = code_loc; | |
920 | self->is_value_used = false; | |
921 | self->rule = rule; | |
992e874a | 922 | self->named_ref = name; |
28e52c0d JD |
923 | } |
924 | ||
925 | void | |
926 | code_props_translate_code (code_props *self) | |
927 | { | |
928 | switch (self->kind) | |
929 | { | |
930 | case CODE_PROPS_NONE: | |
931 | break; | |
932 | case CODE_PROPS_PLAIN: | |
933 | self->code = translate_action (self, INITIAL); | |
934 | break; | |
935 | case CODE_PROPS_SYMBOL_ACTION: | |
936 | self->code = translate_action (self, SC_SYMBOL_ACTION); | |
937 | break; | |
938 | case CODE_PROPS_RULE_ACTION: | |
939 | self->code = translate_action (self, SC_RULE_ACTION); | |
940 | break; | |
941 | } | |
942 | } | |
943 | ||
944 | void | |
945 | code_scanner_last_string_free (void) | |
946 | { | |
947 | STRING_FREE; | |
948 | } | |
e9071366 AD |
949 | |
950 | void | |
951 | code_scanner_free (void) | |
952 | { | |
953 | obstack_free (&obstack_for_string, 0); | |
1e20ad11 AR |
954 | variant_table_free (); |
955 | ||
e9071366 | 956 | /* Reclaim Flex's buffers. */ |
580b8926 | 957 | yylex_destroy (); |
e9071366 | 958 | } |