]>
Commit | Line | Data |
---|---|---|
e9071366 AD |
1 | /* Bison Action Scanner -*- C -*- |
2 | ||
34136e65 | 3 | Copyright (C) 2006-2012 Free Software Foundation, Inc. |
e9071366 AD |
4 | |
5 | This file is part of Bison, the GNU Compiler Compiler. | |
6 | ||
f16b0819 | 7 | This program is free software: you can redistribute it and/or modify |
e9071366 | 8 | it under the terms of the GNU General Public License as published by |
f16b0819 | 9 | the Free Software Foundation, either version 3 of the License, or |
e9071366 AD |
10 | (at your option) any later version. |
11 | ||
12 | This program is distributed in the hope that it will be useful, | |
13 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | GNU General Public License for more details. | |
16 | ||
17 | You should have received a copy of the GNU General Public License | |
f16b0819 | 18 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
e9071366 | 19 | |
8d90395d | 20 | %option debug nodefault noinput nounput noyywrap never-interactive |
e9071366 AD |
21 | %option prefix="code_" outfile="lex.yy.c" |
22 | ||
23 | %{ | |
24 | /* Work around a bug in flex 2.5.31. See Debian bug 333231 | |
25 | <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=333231>. */ | |
26 | #undef code_wrap | |
27 | #define code_wrap() 1 | |
28 | ||
29 | #define FLEX_PREFIX(Id) code_ ## Id | |
0305d25e | 30 | #include <src/flex-scanner.h> |
f9bfc42a | 31 | |
0305d25e AD |
32 | #include <src/complain.h> |
33 | #include <src/reader.h> | |
34 | #include <src/getargs.h> | |
00f5d575 | 35 | #include <src/muscle-tab.h> |
0305d25e AD |
36 | #include <src/scan-code.h> |
37 | #include <src/symlist.h> | |
e9071366 | 38 | |
872b52bc AR |
39 | #include <c-ctype.h> |
40 | #include <get-errno.h> | |
41 | #include <quote.h> | |
42 | ||
e9071366 AD |
43 | /* The current calling start condition: SC_RULE_ACTION or |
44 | SC_SYMBOL_ACTION. */ | |
28e52c0d | 45 | # define YY_DECL static char *code_lex (code_props *self, int sc_context) |
e9071366 AD |
46 | YY_DECL; |
47 | ||
48 | #define YY_USER_ACTION location_compute (loc, &loc->end, yytext, yyleng); | |
49 | ||
676997e5 | 50 | static void handle_action_dollar (symbol_list *rule, char *cp, |
e9690142 | 51 | location dollar_loc); |
676997e5 JD |
52 | static void handle_action_at (symbol_list *rule, char *cp, location at_loc); |
53 | ||
54 | /* A string to be pushed to obstack after dollar/at has been handled. */ | |
55 | static char *ref_tail_fields; | |
b9f1d9a4 | 56 | |
e9071366 AD |
57 | static location the_location; |
58 | static location *loc = &the_location; | |
ddc8ede1 | 59 | |
28e52c0d JD |
60 | /* A string representing the most recent translation. */ |
61 | static char *last_string; | |
62 | ||
ddc8ede1 PE |
63 | /* True if an untyped $$ or $n was seen. */ |
64 | static bool untyped_var_seen; | |
b9f1d9a4 | 65 | |
e9071366 AD |
66 | %} |
67 | /* C and C++ comments in code. */ | |
68 | %x SC_COMMENT SC_LINE_COMMENT | |
69 | /* Strings and characters in code. */ | |
70 | %x SC_STRING SC_CHARACTER | |
71 | /* Whether in a rule or symbol action. Specifies the translation | |
72 | of $ and @. */ | |
73 | %x SC_RULE_ACTION SC_SYMBOL_ACTION | |
74 | ||
75 | ||
76 | /* POSIX says that a tag must be both an id and a C union member, but | |
77 | historically almost any character is allowed in a tag. We disallow | |
78 | NUL and newline, as this simplifies our implementation. */ | |
e9690142 | 79 | tag [^\0\n>]+ |
e9071366 AD |
80 | |
81 | /* Zero or more instances of backslash-newline. Following GCC, allow | |
82 | white space between the backslash and the newline. */ | |
e9690142 | 83 | splice (\\[ \f\t\v]*\n)* |
e9071366 | 84 | |
676997e5 JD |
85 | /* C style identifier. Must start with letter. Will be used for |
86 | named symbol references. Shall be kept synchronized with | |
87 | scan-gram.l "letter" and "id". */ | |
e9690142 JD |
88 | letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_] |
89 | id {letter}({letter}|[-0-9])* | |
676997e5 | 90 | ref -?[0-9]+|{id}|"["{id}"]"|"$" |
b9f1d9a4 | 91 | |
e9071366 AD |
92 | %% |
93 | ||
94 | %{ | |
2346344a | 95 | /* Nesting level of the current code in braces. */ |
4f82b42a | 96 | int braces_level = 0; |
2346344a | 97 | |
e8cd1ad6 | 98 | /* Whether a semicolon is probably needed. |
ae93e4e4 | 99 | The heuristic is that a semicolon is not needed after '{', '}', ';', |
e8cd1ad6 DJ |
100 | or a C preprocessor directive, and that whitespaces and comments |
101 | do not affect this flag. | |
ae93e4e4 | 102 | Note that '{' does not need a semicolon because of '{}'. |
e8cd1ad6 DJ |
103 | A semicolon may be needed before a cpp direcive, but don't bother. */ |
104 | bool need_semicolon = false; | |
105 | ||
106 | /* Whether in a C preprocessor directive. Don't use a start condition | |
107 | for this because, at the end of strings and comments, we still need | |
108 | to know whether we're in a directive. */ | |
109 | bool in_cpp = false; | |
110 | ||
e9071366 AD |
111 | /* This scanner is special: it is invoked only once, henceforth |
112 | is expected to return only once. This initialization is | |
113 | therefore done once per action to translate. */ | |
4f82b42a | 114 | aver (sc_context == SC_SYMBOL_ACTION |
e9690142 JD |
115 | || sc_context == SC_RULE_ACTION |
116 | || sc_context == INITIAL); | |
e9071366 AD |
117 | BEGIN sc_context; |
118 | %} | |
119 | ||
120 | /*------------------------------------------------------------. | |
ae93e4e4 | 121 | | Scanning a C comment. The initial '/ *' is already eaten. | |
e9071366 AD |
122 | `------------------------------------------------------------*/ |
123 | ||
124 | <SC_COMMENT> | |
125 | { | |
126 | "*"{splice}"/" STRING_GROW; BEGIN sc_context; | |
127 | } | |
128 | ||
129 | ||
130 | /*--------------------------------------------------------------. | |
ae93e4e4 | 131 | | Scanning a line comment. The initial '//' is already eaten. | |
e9071366 AD |
132 | `--------------------------------------------------------------*/ |
133 | ||
134 | <SC_LINE_COMMENT> | |
135 | { | |
e9690142 JD |
136 | "\n" STRING_GROW; BEGIN sc_context; |
137 | {splice} STRING_GROW; | |
e9071366 AD |
138 | } |
139 | ||
140 | ||
141 | /*--------------------------------------------. | |
142 | | Scanning user-code characters and strings. | | |
143 | `--------------------------------------------*/ | |
144 | ||
145 | <SC_CHARACTER,SC_STRING> | |
146 | { | |
e9690142 | 147 | {splice}|\\{splice}. STRING_GROW; |
e9071366 AD |
148 | } |
149 | ||
150 | <SC_CHARACTER> | |
151 | { | |
e9690142 | 152 | "'" STRING_GROW; BEGIN sc_context; |
e9071366 AD |
153 | } |
154 | ||
155 | <SC_STRING> | |
156 | { | |
e9690142 | 157 | "\"" STRING_GROW; BEGIN sc_context; |
e9071366 AD |
158 | } |
159 | ||
160 | ||
161 | <SC_RULE_ACTION,SC_SYMBOL_ACTION>{ | |
162 | "'" { | |
163 | STRING_GROW; | |
164 | BEGIN SC_CHARACTER; | |
e8cd1ad6 | 165 | need_semicolon = true; |
e9071366 AD |
166 | } |
167 | "\"" { | |
168 | STRING_GROW; | |
169 | BEGIN SC_STRING; | |
e8cd1ad6 | 170 | need_semicolon = true; |
e9071366 AD |
171 | } |
172 | "/"{splice}"*" { | |
173 | STRING_GROW; | |
174 | BEGIN SC_COMMENT; | |
175 | } | |
176 | "/"{splice}"/" { | |
177 | STRING_GROW; | |
178 | BEGIN SC_LINE_COMMENT; | |
179 | } | |
180 | } | |
181 | ||
182 | <SC_RULE_ACTION> | |
183 | { | |
b9f1d9a4 | 184 | "$"("<"{tag}">")?{ref} { |
676997e5 JD |
185 | ref_tail_fields = 0; |
186 | handle_action_dollar (self->rule, yytext, *loc); | |
187 | if (ref_tail_fields) { | |
188 | obstack_sgrow (&obstack_for_string, ref_tail_fields); | |
189 | } | |
e8cd1ad6 | 190 | need_semicolon = true; |
28e52c0d | 191 | } |
b9f1d9a4 | 192 | "@"{ref} { |
676997e5 JD |
193 | ref_tail_fields = 0; |
194 | handle_action_at (self->rule, yytext, *loc); | |
195 | if (ref_tail_fields) { | |
196 | obstack_sgrow (&obstack_for_string, ref_tail_fields); | |
197 | } | |
e8cd1ad6 | 198 | need_semicolon = true; |
28e52c0d | 199 | } |
e9071366 | 200 | "$" { |
6fb8b256 | 201 | complain_at (*loc, Wother, _("stray '$'")); |
e9071366 | 202 | obstack_sgrow (&obstack_for_string, "$]["); |
e8cd1ad6 | 203 | need_semicolon = true; |
e9071366 AD |
204 | } |
205 | "@" { | |
6fb8b256 | 206 | complain_at (*loc, Wother, _("stray '@'")); |
e9071366 | 207 | obstack_sgrow (&obstack_for_string, "@@"); |
e8cd1ad6 DJ |
208 | need_semicolon = true; |
209 | } | |
210 | "[" { | |
211 | obstack_sgrow (&obstack_for_string, "@{"); | |
212 | need_semicolon = true; | |
213 | } | |
214 | "]" { | |
215 | obstack_sgrow (&obstack_for_string, "@}"); | |
216 | need_semicolon = true; | |
e9071366 | 217 | } |
2346344a | 218 | |
e8cd1ad6 DJ |
219 | ";" STRING_GROW; need_semicolon = false; |
220 | "{" STRING_GROW; ++braces_level; need_semicolon = false; | |
2346344a | 221 | "}" { |
58bd33b7 | 222 | bool outer_brace = --braces_level == 0; |
2346344a | 223 | |
ae93e4e4 | 224 | /* As an undocumented Bison extension, append ';' before the last |
2346344a | 225 | brace in braced code, so that the user code can omit trailing |
ae93e4e4 | 226 | ';'. But do not append ';' if emulating Yacc, since Yacc does |
e8cd1ad6 | 227 | not append one. */ |
d9a9b96b | 228 | if (outer_brace && !yacc_flag && language_prio == default_prio |
e8cd1ad6 DJ |
229 | && skeleton_prio == default_prio && need_semicolon && ! in_cpp) |
230 | { | |
6fb8b256 VS |
231 | complain_at (*loc, Wother, |
232 | _("a ';' might be needed at the end of action code")); | |
233 | complain_at (*loc, Wother, | |
234 | _("future versions of Bison will not add the ';'")); | |
e9690142 | 235 | obstack_1grow (&obstack_for_string, ';'); |
e8cd1ad6 | 236 | } |
2346344a AD |
237 | |
238 | STRING_GROW; | |
e8cd1ad6 | 239 | need_semicolon = false; |
2346344a | 240 | } |
e8cd1ad6 DJ |
241 | |
242 | /* Preprocessing directives should only be recognized at the beginning | |
243 | of lines, allowing whitespace including comments, but in C/C++, | |
ae93e4e4 JM |
244 | '#' can only be the start of preprocessor directives or within |
245 | '#define' directives anyway, so don't bother with begin of line. */ | |
e8cd1ad6 DJ |
246 | "#" STRING_GROW; in_cpp = true; |
247 | ||
248 | {splice} STRING_GROW; | |
bc0f5737 | 249 | [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; |
e8cd1ad6 | 250 | [ \t\f] STRING_GROW; |
4395a9ff JD |
251 | |
252 | /* YYFAIL is undocumented and was formally deprecated in Bison | |
253 | 2.4.2. */ | |
254 | YYFAIL { | |
255 | STRING_GROW; need_semicolon = true; | |
6fb8b256 VS |
256 | complain_at (*loc, Wother, |
257 | _("use of YYFAIL, which is deprecated and will be removed")); | |
4395a9ff JD |
258 | } |
259 | ||
260 | /* The sole purpose of this is to make sure identifiers that merely | |
261 | contain YYFAIL don't produce the above warning. */ | |
262 | [A-Za-z_][0-9A-Za-z_]* STRING_GROW; need_semicolon = true; | |
263 | ||
264 | . STRING_GROW; need_semicolon = true; | |
e9071366 AD |
265 | } |
266 | ||
267 | <SC_SYMBOL_ACTION> | |
268 | { | |
28e52c0d JD |
269 | "$$" { |
270 | obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar["); | |
271 | self->is_value_used = true; | |
272 | } | |
e785ccf7 JD |
273 | "@$" { |
274 | obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); | |
bc0f5737 | 275 | muscle_percent_define_ensure("locations", the_location, true); |
e785ccf7 | 276 | } |
e9071366 AD |
277 | } |
278 | ||
279 | ||
280 | /*-----------------------------------------. | |
281 | | Escape M4 quoting characters in C code. | | |
282 | `-----------------------------------------*/ | |
283 | ||
284 | <*> | |
285 | { | |
e9690142 JD |
286 | \$ obstack_sgrow (&obstack_for_string, "$]["); |
287 | \@ obstack_sgrow (&obstack_for_string, "@@"); | |
288 | \[ obstack_sgrow (&obstack_for_string, "@{"); | |
289 | \] obstack_sgrow (&obstack_for_string, "@}"); | |
e9071366 AD |
290 | } |
291 | ||
292 | /*-----------------------------------------------------. | |
293 | | By default, grow the string obstack with the input. | | |
294 | `-----------------------------------------------------*/ | |
295 | ||
e9690142 | 296 | <*>.|\n STRING_GROW; |
e9071366 AD |
297 | |
298 | /* End of processing. */ | |
e9690142 | 299 | <*><<EOF>> { |
28e52c0d JD |
300 | STRING_FINISH; |
301 | return last_string; | |
e9071366 AD |
302 | } |
303 | ||
304 | %% | |
305 | ||
b9f1d9a4 | 306 | static inline bool |
872b52bc | 307 | is_dot_or_dash (char ch) |
b9f1d9a4 AR |
308 | { |
309 | return ch == '.' || ch == '-'; | |
310 | } | |
311 | ||
312 | static inline bool | |
872b52bc | 313 | contains_dot_or_dash (const char* p) |
b9f1d9a4 | 314 | { |
872b52bc AR |
315 | for (; *p; ++p) |
316 | if (is_dot_or_dash (*p)) | |
317 | return true; | |
318 | return false; | |
b9f1d9a4 AR |
319 | } |
320 | ||
872b52bc | 321 | /* Defines a variant of a symbolic name resolution. */ |
b9f1d9a4 AR |
322 | typedef struct |
323 | { | |
324 | /* Index in symbol list. */ | |
ce268795 | 325 | unsigned symbol_index; |
b9f1d9a4 AR |
326 | |
327 | /* Matched symbol id and loc. */ | |
328 | uniqstr id; | |
329 | location loc; | |
330 | ||
872b52bc | 331 | /* Hiding named reference. */ |
b9f1d9a4 AR |
332 | named_ref* hidden_by; |
333 | ||
872b52bc AR |
334 | /* Error flags. May contain zero (no errors) or |
335 | a combination of VARIANT_* values. */ | |
b9f1d9a4 AR |
336 | unsigned err; |
337 | } variant; | |
338 | ||
872b52bc AR |
339 | /* Set when the variant refers to a symbol hidden |
340 | by an explicit symbol reference. */ | |
341 | #define VARIANT_HIDDEN (1 << 0) | |
342 | ||
676997e5 JD |
343 | /* Set when the variant refers to a symbol containing |
344 | dots or dashes. Will require explicit bracketing. */ | |
345 | #define VARIANT_BAD_BRACKETING (1 << 1) | |
346 | ||
872b52bc AR |
347 | /* Set when the variant refers to a symbol which is |
348 | not visible from current midrule. */ | |
676997e5 | 349 | #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) |
872b52bc | 350 | |
b9f1d9a4 AR |
351 | static variant *variant_table = 0; |
352 | static unsigned variant_table_size = 0; | |
353 | static unsigned variant_count = 0; | |
354 | ||
355 | static variant * | |
ba4184ec | 356 | variant_table_grow (void) |
b9f1d9a4 AR |
357 | { |
358 | ++variant_count; | |
359 | if (variant_count > variant_table_size) | |
360 | { | |
361 | while (variant_count > variant_table_size) | |
e9690142 | 362 | variant_table_size = 2 * variant_table_size + 3; |
b9f1d9a4 | 363 | variant_table = xnrealloc (variant_table, variant_table_size, |
e9690142 | 364 | sizeof *variant_table); |
b9f1d9a4 AR |
365 | } |
366 | return &variant_table[variant_count - 1]; | |
367 | } | |
368 | ||
13cdf208 | 369 | static void |
ba4184ec | 370 | variant_table_free (void) |
13cdf208 | 371 | { |
ba4184ec | 372 | free (variant_table); |
13cdf208 AR |
373 | variant_table = 0; |
374 | variant_table_size = variant_count = 0; | |
375 | } | |
376 | ||
676997e5 JD |
377 | static char * |
378 | find_prefix_end (const char *prefix, char *begin, char *end) | |
b9f1d9a4 | 379 | { |
676997e5 JD |
380 | char *ptr = begin; |
381 | ||
382 | for (; *prefix && ptr != end; ++prefix, ++ptr) | |
383 | if (*prefix != *ptr) | |
384 | return 0; | |
385 | ||
386 | if (*prefix) | |
387 | return 0; | |
388 | ||
389 | return ptr; | |
b9f1d9a4 AR |
390 | } |
391 | ||
392 | static variant * | |
ba4184ec | 393 | variant_add (uniqstr id, location id_loc, unsigned symbol_index, |
e9690142 | 394 | char *cp, char *cp_end, bool explicit_bracketing) |
b9f1d9a4 | 395 | { |
676997e5 JD |
396 | char *prefix_end; |
397 | ||
398 | prefix_end = find_prefix_end (id, cp, cp_end); | |
399 | if (prefix_end && | |
400 | (prefix_end == cp_end || | |
401 | (!explicit_bracketing && is_dot_or_dash (*prefix_end)))) | |
b9f1d9a4 | 402 | { |
872b52bc | 403 | variant *r = variant_table_grow (); |
ce268795 | 404 | r->symbol_index = symbol_index; |
b9f1d9a4 | 405 | r->id = id; |
ba4184ec | 406 | r->loc = id_loc; |
b9f1d9a4 AR |
407 | r->hidden_by = NULL; |
408 | r->err = 0; | |
409 | return r; | |
410 | } | |
411 | else | |
412 | return NULL; | |
413 | } | |
414 | ||
872b52bc | 415 | static const char * |
ba4184ec | 416 | get_at_spec(unsigned symbol_index) |
872b52bc AR |
417 | { |
418 | static char at_buf[20]; | |
ba4184ec | 419 | if (symbol_index == 0) |
872b52bc AR |
420 | strcpy (at_buf, "$$"); |
421 | else | |
ba4184ec | 422 | snprintf (at_buf, sizeof at_buf, "$%u", symbol_index); |
872b52bc AR |
423 | return at_buf; |
424 | } | |
425 | ||
ce268795 | 426 | static void |
676997e5 JD |
427 | show_sub_messages (const char* cp, bool explicit_bracketing, |
428 | int midrule_rhs_index, char dollar_or_at, | |
66381412 | 429 | bool is_warning, unsigned indent) |
ce268795 AR |
430 | { |
431 | unsigned i; | |
432 | ||
433 | for (i = 0; i < variant_count; ++i) | |
434 | { | |
435 | const variant *var = &variant_table[i]; | |
436 | const char *at_spec = get_at_spec (var->symbol_index); | |
437 | ||
438 | if (var->err == 0) | |
439 | { | |
440 | if (is_warning) | |
6fb8b256 VS |
441 | complain_at_indent (var->loc, Wother, &indent, |
442 | _("refers to: %c%s at %s"), dollar_or_at, | |
443 | var->id, at_spec); | |
ce268795 | 444 | else |
6fb8b256 VS |
445 | complain_at_indent (var->loc, complaint, &indent, |
446 | _("refers to: %c%s at %s"), dollar_or_at, | |
447 | var->id, at_spec); | |
ce268795 AR |
448 | } |
449 | else | |
e9690142 JD |
450 | { |
451 | static struct obstack msg_buf; | |
452 | const char *tail = explicit_bracketing ? "" : | |
453 | cp + strlen (var->id); | |
454 | const char *id = var->hidden_by ? var->hidden_by->id : | |
455 | var->id; | |
456 | location id_loc = var->hidden_by ? var->hidden_by->loc : | |
457 | var->loc; | |
458 | ||
459 | /* Create the explanation message. */ | |
460 | obstack_init (&msg_buf); | |
461 | ||
462 | obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at); | |
463 | if (contains_dot_or_dash (id)) | |
464 | obstack_fgrow1 (&msg_buf, "[%s]", id); | |
465 | else | |
466 | obstack_sgrow (&msg_buf, id); | |
467 | obstack_sgrow (&msg_buf, tail); | |
468 | ||
469 | if (var->err & VARIANT_HIDDEN) | |
470 | { | |
471 | obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at); | |
472 | if (contains_dot_or_dash (var->id)) | |
473 | obstack_fgrow1 (&msg_buf, "[%s]", var->id); | |
474 | else | |
475 | obstack_sgrow (&msg_buf, var->id); | |
476 | obstack_sgrow (&msg_buf, tail); | |
477 | } | |
478 | ||
479 | obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec); | |
480 | ||
481 | if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) | |
66381412 AR |
482 | { |
483 | const char *format = | |
484 | _(", cannot be accessed from mid-rule action at $%d"); | |
485 | obstack_fgrow1 (&msg_buf, format, midrule_rhs_index); | |
486 | } | |
ce268795 | 487 | |
e9690142 | 488 | obstack_1grow (&msg_buf, '\0'); |
ce268795 | 489 | if (is_warning) |
6fb8b256 VS |
490 | complain_at_indent (id_loc, Wother, &indent, "%s", |
491 | (char *) obstack_finish (&msg_buf)); | |
ce268795 | 492 | else |
6fb8b256 | 493 | complain_at_indent (id_loc, complaint, &indent, "%s", |
66381412 | 494 | (char *) obstack_finish (&msg_buf)); |
e9690142 JD |
495 | obstack_free (&msg_buf, 0); |
496 | } | |
ce268795 AR |
497 | } |
498 | } | |
499 | ||
872b52bc AR |
500 | /* Returned from "parse_ref" when the reference |
501 | is inappropriate. */ | |
b9f1d9a4 | 502 | #define INVALID_REF (INT_MIN) |
872b52bc AR |
503 | |
504 | /* Returned from "parse_ref" when the reference | |
ce268795 | 505 | points to LHS ($$) of the current rule or midrule. */ |
b9f1d9a4 AR |
506 | #define LHS_REF (INT_MIN + 1) |
507 | ||
66381412 AR |
508 | /* Sub-messages indent. */ |
509 | #define SUB_INDENT (4) | |
510 | ||
676997e5 JD |
511 | /* Parse named or positional reference. In case of positional |
512 | references, can return negative values for $-n "deep" stack | |
513 | accesses. */ | |
b9f1d9a4 | 514 | static long int |
676997e5 | 515 | parse_ref (char *cp, symbol_list *rule, int rule_length, |
e9690142 JD |
516 | int midrule_rhs_index, char *text, location text_loc, |
517 | char dollar_or_at) | |
b9f1d9a4 | 518 | { |
676997e5 JD |
519 | symbol_list *l; |
520 | char *cp_end; | |
61bc57e5 | 521 | bool explicit_bracketing; |
872b52bc | 522 | unsigned i; |
ce268795 AR |
523 | unsigned valid_variants = 0; |
524 | unsigned valid_variant_index = 0; | |
b9f1d9a4 AR |
525 | |
526 | if ('$' == *cp) | |
527 | return LHS_REF; | |
528 | ||
676997e5 | 529 | if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1)))) |
b9f1d9a4 | 530 | { |
676997e5 | 531 | long int num = strtol (cp, &cp, 10); |
b9f1d9a4 | 532 | if (1 - INT_MAX + rule_length <= num && num <= rule_length) |
e9690142 | 533 | return num; |
b9f1d9a4 | 534 | else |
e9690142 | 535 | { |
6fb8b256 | 536 | complain_at (text_loc, complaint, _("integer out of range: %s"), |
ba4184ec | 537 | quote (text)); |
e9690142 JD |
538 | return INVALID_REF; |
539 | } | |
b9f1d9a4 AR |
540 | } |
541 | ||
676997e5 JD |
542 | if ('[' == *cp) |
543 | { | |
544 | /* Ignore the brackets. */ | |
545 | char *p; | |
546 | for (p = ++cp; *p != ']'; ++p) | |
e9690142 | 547 | continue; |
676997e5 JD |
548 | cp_end = p; |
549 | ||
550 | explicit_bracketing = true; | |
551 | } | |
552 | else | |
553 | { | |
554 | /* Take all characters of the name. */ | |
555 | char* p; | |
556 | for (p = cp; *p; ++p) | |
e9690142 JD |
557 | if (is_dot_or_dash (*p)) |
558 | { | |
559 | ref_tail_fields = p; | |
560 | break; | |
561 | } | |
676997e5 | 562 | for (p = cp; *p; ++p) |
e9690142 | 563 | continue; |
676997e5 JD |
564 | cp_end = p; |
565 | ||
566 | explicit_bracketing = false; | |
567 | } | |
b9f1d9a4 AR |
568 | |
569 | /* Add all relevant variants. */ | |
872b52bc | 570 | { |
ba4184ec | 571 | unsigned symbol_index; |
872b52bc | 572 | variant_count = 0; |
ba4184ec AR |
573 | for (symbol_index = 0, l = rule; !symbol_list_null (l); |
574 | ++symbol_index, l = l->next) | |
872b52bc | 575 | { |
e9690142 JD |
576 | variant *var; |
577 | if (l->content_type != SYMLIST_SYMBOL) | |
578 | continue; | |
b9f1d9a4 | 579 | |
e9690142 | 580 | var = variant_add (l->content.sym->tag, l->sym_loc, |
676997e5 | 581 | symbol_index, cp, cp_end, explicit_bracketing); |
e9690142 JD |
582 | if (var && l->named_ref) |
583 | var->hidden_by = l->named_ref; | |
b9f1d9a4 | 584 | |
e9690142 JD |
585 | if (l->named_ref) |
586 | variant_add (l->named_ref->id, l->named_ref->loc, | |
676997e5 | 587 | symbol_index, cp, cp_end, explicit_bracketing); |
872b52bc AR |
588 | } |
589 | } | |
b9f1d9a4 AR |
590 | |
591 | /* Check errors. */ | |
b9f1d9a4 AR |
592 | for (i = 0; i < variant_count; ++i) |
593 | { | |
ba4184ec | 594 | variant *var = &variant_table[i]; |
ce268795 | 595 | unsigned symbol_index = var->symbol_index; |
b9f1d9a4 AR |
596 | |
597 | /* Check visibility from mid-rule actions. */ | |
872b52bc | 598 | if (midrule_rhs_index != 0 |
e9690142 | 599 | && (symbol_index == 0 || midrule_rhs_index < symbol_index)) |
ce268795 | 600 | var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE; |
b9f1d9a4 | 601 | |
676997e5 JD |
602 | /* Check correct bracketing. */ |
603 | if (!explicit_bracketing && contains_dot_or_dash (var->id)) | |
604 | var->err |= VARIANT_BAD_BRACKETING; | |
605 | ||
b9f1d9a4 | 606 | /* Check using of hidden symbols. */ |
ba4184ec | 607 | if (var->hidden_by) |
ce268795 | 608 | var->err |= VARIANT_HIDDEN; |
b9f1d9a4 | 609 | |
ba4184ec | 610 | if (!var->err) |
ce268795 AR |
611 | { |
612 | valid_variant_index = i; | |
613 | ++valid_variants; | |
614 | } | |
b9f1d9a4 AR |
615 | } |
616 | ||
ce268795 | 617 | switch (valid_variants) |
b9f1d9a4 | 618 | { |
ce268795 | 619 | case 0: |
f64e406b | 620 | { |
676997e5 JD |
621 | unsigned len = (explicit_bracketing || !ref_tail_fields) ? |
622 | cp_end - cp : ref_tail_fields - cp; | |
66381412 | 623 | unsigned indent = 0; |
f64e406b | 624 | |
6fb8b256 VS |
625 | complain_at_indent (text_loc, complaint, &indent, |
626 | _("invalid reference: %s"), quote (text)); | |
66381412 | 627 | indent += SUB_INDENT; |
5c9efc75 JD |
628 | if (len == 0) |
629 | { | |
630 | location sym_loc = text_loc; | |
631 | sym_loc.start.column += 1; | |
632 | sym_loc.end = sym_loc.start; | |
633 | const char *format = | |
ae93e4e4 JM |
634 | _("syntax error after '%c', expecting integer, letter," |
635 | " '_', '[', or '$'"); | |
6fb8b256 VS |
636 | complain_at_indent (sym_loc, complaint, &indent, format, |
637 | dollar_or_at); | |
5c9efc75 JD |
638 | } |
639 | else if (midrule_rhs_index) | |
66381412 AR |
640 | { |
641 | const char *format = | |
642 | _("symbol not found in production before $%d: %.*s"); | |
6fb8b256 | 643 | complain_at_indent (rule->location, complaint, &indent, format, |
66381412 AR |
644 | midrule_rhs_index, len, cp); |
645 | } | |
f64e406b | 646 | else |
66381412 AR |
647 | { |
648 | const char *format = | |
649 | _("symbol not found in production: %.*s"); | |
6fb8b256 | 650 | complain_at_indent (rule->location, complaint, &indent, format, |
66381412 AR |
651 | len, cp); |
652 | } | |
f64e406b AR |
653 | |
654 | if (variant_count > 0) | |
676997e5 JD |
655 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
656 | dollar_or_at, false, indent); | |
f64e406b AR |
657 | return INVALID_REF; |
658 | } | |
ce268795 AR |
659 | case 1: |
660 | { | |
66381412 | 661 | unsigned indent = 0; |
ce268795 AR |
662 | if (variant_count > 1) |
663 | { | |
6fb8b256 VS |
664 | complain_at_indent (text_loc, Wother, &indent, |
665 | _("misleading reference: %s"), quote (text)); | |
676997e5 JD |
666 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
667 | dollar_or_at, true, indent + SUB_INDENT); | |
ce268795 AR |
668 | } |
669 | { | |
670 | unsigned symbol_index = | |
671 | variant_table[valid_variant_index].symbol_index; | |
672 | return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index; | |
673 | } | |
674 | } | |
675 | case 2: | |
676 | default: | |
66381412 AR |
677 | { |
678 | unsigned indent = 0; | |
6fb8b256 VS |
679 | complain_at_indent (text_loc, complaint, &indent, |
680 | _("ambiguous reference: %s"), quote (text)); | |
676997e5 JD |
681 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
682 | dollar_or_at, false, indent + SUB_INDENT); | |
66381412 AR |
683 | return INVALID_REF; |
684 | } | |
b9f1d9a4 AR |
685 | } |
686 | ||
ce268795 | 687 | /* Not reachable. */ |
b9f1d9a4 AR |
688 | return INVALID_REF; |
689 | } | |
690 | ||
e9071366 AD |
691 | /* Keeps track of the maximum number of semantic values to the left of |
692 | a handle (those referenced by $0, $-1, etc.) are required by the | |
693 | semantic actions of this grammar. */ | |
694 | int max_left_semantic_context = 0; | |
695 | ||
696 | ||
697 | /*------------------------------------------------------------------. | |
ae93e4e4 | 698 | | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | |
e9071366 | 699 | | | |
676997e5 | 700 | | Possible inputs: $[<TYPENAME>]($|integer) | |
e9071366 AD |
701 | | | |
702 | | Output to OBSTACK_FOR_STRING a reference to this semantic value. | | |
703 | `------------------------------------------------------------------*/ | |
704 | ||
705 | static void | |
676997e5 | 706 | handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) |
e9071366 | 707 | { |
28e52c0d | 708 | char const *type_name = NULL; |
e9071366 | 709 | char *cp = text + 1; |
b9f1d9a4 | 710 | char *gt_ptr = 0; |
ffa4ba3a | 711 | symbol_list *effective_rule; |
872b52bc AR |
712 | int effective_rule_length; |
713 | int n; | |
ffa4ba3a JD |
714 | |
715 | if (rule->midrule_parent_rule) | |
716 | { | |
717 | effective_rule = rule->midrule_parent_rule; | |
718 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
719 | } | |
720 | else | |
721 | { | |
722 | effective_rule = rule; | |
723 | effective_rule_length = symbol_list_length (rule->next); | |
724 | } | |
e9071366 AD |
725 | |
726 | /* Get the type name if explicit. */ | |
727 | if (*cp == '<') | |
728 | { | |
729 | type_name = ++cp; | |
730 | while (*cp != '>') | |
e9690142 | 731 | ++cp; |
b9f1d9a4 AR |
732 | |
733 | /* The '>' symbol will be later replaced by '\0'. Original | |
e9690142 | 734 | 'text' is needed for error messages. */ |
b9f1d9a4 | 735 | gt_ptr = cp; |
e9071366 | 736 | ++cp; |
ddc8ede1 | 737 | if (untyped_var_seen) |
6fb8b256 VS |
738 | complain_at (dollar_loc, complaint, |
739 | _("explicit type given in untyped grammar")); | |
ddc8ede1 | 740 | tag_seen = true; |
e9071366 AD |
741 | } |
742 | ||
872b52bc | 743 | n = parse_ref (cp, effective_rule, effective_rule_length, |
e9690142 | 744 | rule->midrule_parent_rhs_index, text, dollar_loc, '$'); |
b9f1d9a4 AR |
745 | |
746 | if (gt_ptr) | |
747 | *gt_ptr = '\0'; | |
748 | ||
749 | switch (n) | |
e9071366 | 750 | { |
b9f1d9a4 AR |
751 | case INVALID_REF: |
752 | break; | |
753 | ||
754 | case LHS_REF: | |
e9071366 | 755 | if (!type_name) |
e9690142 | 756 | type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0); |
ddc8ede1 PE |
757 | |
758 | if (!type_name) | |
e9690142 JD |
759 | { |
760 | if (union_seen | tag_seen) | |
761 | { | |
762 | if (rule->midrule_parent_rule) | |
6fb8b256 | 763 | complain_at (dollar_loc, complaint, |
4a9cd8f2 | 764 | _("$$ for the midrule at $%d of %s" |
e9690142 JD |
765 | " has no declared type"), |
766 | rule->midrule_parent_rhs_index, | |
4a9cd8f2 | 767 | quote (effective_rule->content.sym->tag)); |
e9690142 | 768 | else |
6fb8b256 VS |
769 | complain_at (dollar_loc, complaint, |
770 | _("$$ of %s has no declared type"), | |
4a9cd8f2 | 771 | quote (rule->content.sym->tag)); |
e9690142 JD |
772 | } |
773 | else | |
774 | untyped_var_seen = true; | |
775 | type_name = ""; | |
776 | } | |
ddc8ede1 | 777 | |
e9071366 | 778 | obstack_fgrow1 (&obstack_for_string, |
e9690142 | 779 | "]b4_lhs_value([%s])[", type_name); |
f6857bbf | 780 | rule->action_props.is_value_used = true; |
b9f1d9a4 AR |
781 | break; |
782 | ||
783 | default: | |
784 | if (max_left_semantic_context < 1 - n) | |
e9690142 | 785 | max_left_semantic_context = 1 - n; |
b9f1d9a4 | 786 | if (!type_name && 0 < n) |
e9690142 JD |
787 | type_name = |
788 | symbol_list_n_type_name_get (effective_rule, dollar_loc, n); | |
b9f1d9a4 | 789 | if (!type_name) |
e9690142 JD |
790 | { |
791 | if (union_seen | tag_seen) | |
6fb8b256 VS |
792 | complain_at (dollar_loc, complaint, |
793 | _("$%s of %s has no declared type"), cp, | |
794 | quote (effective_rule->content.sym->tag)); | |
e9690142 JD |
795 | else |
796 | untyped_var_seen = true; | |
797 | type_name = ""; | |
798 | } | |
b9f1d9a4 AR |
799 | |
800 | obstack_fgrow3 (&obstack_for_string, | |
e9690142 JD |
801 | "]b4_rhs_value(%d, %d, [%s])[", |
802 | effective_rule_length, n, type_name); | |
b9f1d9a4 | 803 | if (n > 0) |
e9690142 JD |
804 | symbol_list_n_get (effective_rule, n)->action_props.is_value_used = |
805 | true; | |
b9f1d9a4 | 806 | break; |
e9071366 AD |
807 | } |
808 | } | |
809 | ||
810 | ||
811 | /*------------------------------------------------------. | |
ae93e4e4 | 812 | | TEXT is a location token (i.e., a '@...'). Output to | |
e9071366 AD |
813 | | OBSTACK_FOR_STRING a reference to this location. | |
814 | `------------------------------------------------------*/ | |
815 | ||
816 | static void | |
676997e5 | 817 | handle_action_at (symbol_list *rule, char *text, location at_loc) |
e9071366 | 818 | { |
676997e5 JD |
819 | char *cp = text + 1; |
820 | symbol_list *effective_rule; | |
872b52bc AR |
821 | int effective_rule_length; |
822 | int n; | |
b9f1d9a4 AR |
823 | |
824 | if (rule->midrule_parent_rule) | |
825 | { | |
826 | effective_rule = rule->midrule_parent_rule; | |
827 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
828 | } | |
829 | else | |
830 | { | |
831 | effective_rule = rule; | |
832 | effective_rule_length = symbol_list_length (rule->next); | |
833 | } | |
ffa4ba3a | 834 | |
bc0f5737 | 835 | muscle_percent_define_ensure("locations", at_loc, true); |
e9071366 | 836 | |
872b52bc | 837 | n = parse_ref (cp, effective_rule, effective_rule_length, |
e9690142 | 838 | rule->midrule_parent_rhs_index, text, at_loc, '@'); |
b9f1d9a4 | 839 | switch (n) |
e9071366 | 840 | { |
b9f1d9a4 AR |
841 | case INVALID_REF: |
842 | break; | |
e9071366 | 843 | |
b9f1d9a4 AR |
844 | case LHS_REF: |
845 | obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); | |
846 | break; | |
847 | ||
848 | default: | |
849 | obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[", | |
e9690142 | 850 | effective_rule_length, n); |
b9f1d9a4 | 851 | break; |
e9071366 AD |
852 | } |
853 | } | |
854 | ||
855 | ||
856 | /*-------------------------. | |
857 | | Initialize the scanner. | | |
858 | `-------------------------*/ | |
859 | ||
28e52c0d | 860 | /* Translate the dollars and ats in \a self, in the context \a sc_context |
ddc8ede1 | 861 | (SC_RULE_ACTION, SC_SYMBOL_ACTION, INITIAL). */ |
e9071366 | 862 | |
eb095650 | 863 | static char const * |
28e52c0d | 864 | translate_action (code_props *self, int sc_context) |
e9071366 | 865 | { |
2ce4ed68 | 866 | char *res; |
e9071366 AD |
867 | static bool initialized = false; |
868 | if (!initialized) | |
869 | { | |
870 | obstack_init (&obstack_for_string); | |
e9071366 AD |
871 | yy_flex_debug = 0; |
872 | initialized = true; | |
873 | } | |
874 | ||
28e52c0d JD |
875 | loc->start = loc->end = self->location.start; |
876 | yy_switch_to_buffer (yy_scan_string (self->code)); | |
877 | res = code_lex (self, sc_context); | |
e9071366 AD |
878 | yy_delete_buffer (YY_CURRENT_BUFFER); |
879 | ||
880 | return res; | |
881 | } | |
882 | ||
28e52c0d JD |
883 | /*------------------------------------------------------------------------. |
884 | | Implementation of the public interface as documented in "scan-code.h". | | |
885 | `------------------------------------------------------------------------*/ | |
886 | ||
887 | void | |
888 | code_props_none_init (code_props *self) | |
76290368 | 889 | { |
28e52c0d | 890 | *self = code_props_none; |
76290368 JD |
891 | } |
892 | ||
28e52c0d JD |
893 | code_props const code_props_none = CODE_PROPS_NONE_INIT; |
894 | ||
895 | void | |
872b52bc | 896 | code_props_plain_init (code_props *self, char const *code, |
e9690142 | 897 | location code_loc) |
76290368 | 898 | { |
ea9a35c6 | 899 | code_props_none_init (self); |
28e52c0d JD |
900 | self->kind = CODE_PROPS_PLAIN; |
901 | self->code = code; | |
902 | self->location = code_loc; | |
76290368 JD |
903 | } |
904 | ||
28e52c0d JD |
905 | void |
906 | code_props_symbol_action_init (code_props *self, char const *code, | |
907 | location code_loc) | |
e9071366 | 908 | { |
ea9a35c6 | 909 | code_props_none_init (self); |
28e52c0d JD |
910 | self->kind = CODE_PROPS_SYMBOL_ACTION; |
911 | self->code = code; | |
912 | self->location = code_loc; | |
e9071366 AD |
913 | } |
914 | ||
28e52c0d JD |
915 | void |
916 | code_props_rule_action_init (code_props *self, char const *code, | |
b9f1d9a4 | 917 | location code_loc, symbol_list *rule, |
e9690142 | 918 | named_ref *name, bool is_predicate) |
28e52c0d | 919 | { |
ea9a35c6 | 920 | code_props_none_init (self); |
28e52c0d JD |
921 | self->kind = CODE_PROPS_RULE_ACTION; |
922 | self->code = code; | |
923 | self->location = code_loc; | |
28e52c0d | 924 | self->rule = rule; |
ba4184ec | 925 | self->named_ref = name; |
ca2a6d15 | 926 | self->is_predicate = is_predicate; |
28e52c0d JD |
927 | } |
928 | ||
929 | void | |
930 | code_props_translate_code (code_props *self) | |
931 | { | |
932 | switch (self->kind) | |
933 | { | |
934 | case CODE_PROPS_NONE: | |
935 | break; | |
936 | case CODE_PROPS_PLAIN: | |
937 | self->code = translate_action (self, INITIAL); | |
938 | break; | |
939 | case CODE_PROPS_SYMBOL_ACTION: | |
940 | self->code = translate_action (self, SC_SYMBOL_ACTION); | |
941 | break; | |
942 | case CODE_PROPS_RULE_ACTION: | |
943 | self->code = translate_action (self, SC_RULE_ACTION); | |
944 | break; | |
945 | } | |
946 | } | |
947 | ||
948 | void | |
949 | code_scanner_last_string_free (void) | |
950 | { | |
951 | STRING_FREE; | |
952 | } | |
e9071366 AD |
953 | |
954 | void | |
955 | code_scanner_free (void) | |
956 | { | |
957 | obstack_free (&obstack_for_string, 0); | |
13cdf208 AR |
958 | variant_table_free (); |
959 | ||
e9071366 | 960 | /* Reclaim Flex's buffers. */ |
580b8926 | 961 | yylex_destroy (); |
e9071366 | 962 | } |