]>
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 | "$" { |
ae93e4e4 | 201 | warn_at (*loc, _("stray '$'")); |
e9071366 | 202 | obstack_sgrow (&obstack_for_string, "$]["); |
e8cd1ad6 | 203 | need_semicolon = true; |
e9071366 AD |
204 | } |
205 | "@" { | |
ae93e4e4 | 206 | warn_at (*loc, _("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 | { | |
ae93e4e4 JM |
231 | warn_at (*loc, _("a ';' might be needed at the end of action code")); |
232 | warn_at (*loc, _("future versions of Bison will not add the ';'")); | |
e9690142 | 233 | obstack_1grow (&obstack_for_string, ';'); |
e8cd1ad6 | 234 | } |
2346344a AD |
235 | |
236 | STRING_GROW; | |
e8cd1ad6 | 237 | need_semicolon = false; |
2346344a | 238 | } |
e8cd1ad6 DJ |
239 | |
240 | /* Preprocessing directives should only be recognized at the beginning | |
241 | of lines, allowing whitespace including comments, but in C/C++, | |
ae93e4e4 JM |
242 | '#' can only be the start of preprocessor directives or within |
243 | '#define' directives anyway, so don't bother with begin of line. */ | |
e8cd1ad6 DJ |
244 | "#" STRING_GROW; in_cpp = true; |
245 | ||
246 | {splice} STRING_GROW; | |
bc0f5737 | 247 | [\n\r] STRING_GROW; if (in_cpp) in_cpp = need_semicolon = false; |
e8cd1ad6 | 248 | [ \t\f] STRING_GROW; |
4395a9ff 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 | { | |
28e52c0d JD |
267 | "$$" { |
268 | obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar["); | |
269 | self->is_value_used = true; | |
270 | } | |
e785ccf7 JD |
271 | "@$" { |
272 | obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); | |
bc0f5737 | 273 | muscle_percent_define_ensure("locations", the_location, true); |
e785ccf7 | 274 | } |
e9071366 AD |
275 | } |
276 | ||
277 | ||
278 | /*-----------------------------------------. | |
279 | | Escape M4 quoting characters in C code. | | |
280 | `-----------------------------------------*/ | |
281 | ||
282 | <*> | |
283 | { | |
e9690142 JD |
284 | \$ obstack_sgrow (&obstack_for_string, "$]["); |
285 | \@ obstack_sgrow (&obstack_for_string, "@@"); | |
286 | \[ obstack_sgrow (&obstack_for_string, "@{"); | |
287 | \] obstack_sgrow (&obstack_for_string, "@}"); | |
e9071366 AD |
288 | } |
289 | ||
290 | /*-----------------------------------------------------. | |
291 | | By default, grow the string obstack with the input. | | |
292 | `-----------------------------------------------------*/ | |
293 | ||
e9690142 | 294 | <*>.|\n STRING_GROW; |
e9071366 AD |
295 | |
296 | /* End of processing. */ | |
e9690142 | 297 | <*><<EOF>> { |
28e52c0d JD |
298 | STRING_FINISH; |
299 | return last_string; | |
e9071366 AD |
300 | } |
301 | ||
302 | %% | |
303 | ||
b9f1d9a4 | 304 | static inline bool |
872b52bc | 305 | is_dot_or_dash (char ch) |
b9f1d9a4 AR |
306 | { |
307 | return ch == '.' || ch == '-'; | |
308 | } | |
309 | ||
310 | static inline bool | |
872b52bc | 311 | contains_dot_or_dash (const char* p) |
b9f1d9a4 | 312 | { |
872b52bc AR |
313 | for (; *p; ++p) |
314 | if (is_dot_or_dash (*p)) | |
315 | return true; | |
316 | return false; | |
b9f1d9a4 AR |
317 | } |
318 | ||
872b52bc | 319 | /* Defines a variant of a symbolic name resolution. */ |
b9f1d9a4 AR |
320 | typedef struct |
321 | { | |
322 | /* Index in symbol list. */ | |
ce268795 | 323 | unsigned symbol_index; |
b9f1d9a4 AR |
324 | |
325 | /* Matched symbol id and loc. */ | |
326 | uniqstr id; | |
327 | location loc; | |
328 | ||
872b52bc | 329 | /* Hiding named reference. */ |
b9f1d9a4 AR |
330 | named_ref* hidden_by; |
331 | ||
872b52bc AR |
332 | /* Error flags. May contain zero (no errors) or |
333 | a combination of VARIANT_* values. */ | |
b9f1d9a4 AR |
334 | unsigned err; |
335 | } variant; | |
336 | ||
872b52bc AR |
337 | /* Set when the variant refers to a symbol hidden |
338 | by an explicit symbol reference. */ | |
339 | #define VARIANT_HIDDEN (1 << 0) | |
340 | ||
676997e5 JD |
341 | /* Set when the variant refers to a symbol containing |
342 | dots or dashes. Will require explicit bracketing. */ | |
343 | #define VARIANT_BAD_BRACKETING (1 << 1) | |
344 | ||
872b52bc AR |
345 | /* Set when the variant refers to a symbol which is |
346 | not visible from current midrule. */ | |
676997e5 | 347 | #define VARIANT_NOT_VISIBLE_FROM_MIDRULE (1 << 2) |
872b52bc | 348 | |
b9f1d9a4 AR |
349 | static variant *variant_table = 0; |
350 | static unsigned variant_table_size = 0; | |
351 | static unsigned variant_count = 0; | |
352 | ||
353 | static variant * | |
ba4184ec | 354 | variant_table_grow (void) |
b9f1d9a4 AR |
355 | { |
356 | ++variant_count; | |
357 | if (variant_count > variant_table_size) | |
358 | { | |
359 | while (variant_count > variant_table_size) | |
e9690142 | 360 | variant_table_size = 2 * variant_table_size + 3; |
b9f1d9a4 | 361 | variant_table = xnrealloc (variant_table, variant_table_size, |
e9690142 | 362 | sizeof *variant_table); |
b9f1d9a4 AR |
363 | } |
364 | return &variant_table[variant_count - 1]; | |
365 | } | |
366 | ||
13cdf208 | 367 | static void |
ba4184ec | 368 | variant_table_free (void) |
13cdf208 | 369 | { |
ba4184ec | 370 | free (variant_table); |
13cdf208 AR |
371 | variant_table = 0; |
372 | variant_table_size = variant_count = 0; | |
373 | } | |
374 | ||
676997e5 JD |
375 | static char * |
376 | find_prefix_end (const char *prefix, char *begin, char *end) | |
b9f1d9a4 | 377 | { |
676997e5 JD |
378 | char *ptr = begin; |
379 | ||
380 | for (; *prefix && ptr != end; ++prefix, ++ptr) | |
381 | if (*prefix != *ptr) | |
382 | return 0; | |
383 | ||
384 | if (*prefix) | |
385 | return 0; | |
386 | ||
387 | return ptr; | |
b9f1d9a4 AR |
388 | } |
389 | ||
390 | static variant * | |
ba4184ec | 391 | variant_add (uniqstr id, location id_loc, unsigned symbol_index, |
e9690142 | 392 | char *cp, char *cp_end, bool explicit_bracketing) |
b9f1d9a4 | 393 | { |
676997e5 JD |
394 | char *prefix_end; |
395 | ||
396 | prefix_end = find_prefix_end (id, cp, cp_end); | |
397 | if (prefix_end && | |
398 | (prefix_end == cp_end || | |
399 | (!explicit_bracketing && is_dot_or_dash (*prefix_end)))) | |
b9f1d9a4 | 400 | { |
872b52bc | 401 | variant *r = variant_table_grow (); |
ce268795 | 402 | r->symbol_index = symbol_index; |
b9f1d9a4 | 403 | r->id = id; |
ba4184ec | 404 | r->loc = id_loc; |
b9f1d9a4 AR |
405 | r->hidden_by = NULL; |
406 | r->err = 0; | |
407 | return r; | |
408 | } | |
409 | else | |
410 | return NULL; | |
411 | } | |
412 | ||
872b52bc | 413 | static const char * |
ba4184ec | 414 | get_at_spec(unsigned symbol_index) |
872b52bc AR |
415 | { |
416 | static char at_buf[20]; | |
ba4184ec | 417 | if (symbol_index == 0) |
872b52bc AR |
418 | strcpy (at_buf, "$$"); |
419 | else | |
ba4184ec | 420 | snprintf (at_buf, sizeof at_buf, "$%u", symbol_index); |
872b52bc AR |
421 | return at_buf; |
422 | } | |
423 | ||
ce268795 | 424 | static void |
676997e5 JD |
425 | show_sub_messages (const char* cp, bool explicit_bracketing, |
426 | int midrule_rhs_index, char dollar_or_at, | |
66381412 | 427 | bool is_warning, unsigned indent) |
ce268795 AR |
428 | { |
429 | unsigned i; | |
430 | ||
431 | for (i = 0; i < variant_count; ++i) | |
432 | { | |
433 | const variant *var = &variant_table[i]; | |
434 | const char *at_spec = get_at_spec (var->symbol_index); | |
435 | ||
436 | if (var->err == 0) | |
437 | { | |
438 | if (is_warning) | |
66381412 AR |
439 | warn_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
440 | dollar_or_at, var->id, at_spec); | |
ce268795 | 441 | else |
66381412 AR |
442 | complain_at_indent (var->loc, &indent, _("refers to: %c%s at %s"), |
443 | dollar_or_at, var->id, at_spec); | |
ce268795 AR |
444 | } |
445 | else | |
e9690142 JD |
446 | { |
447 | static struct obstack msg_buf; | |
448 | const char *tail = explicit_bracketing ? "" : | |
449 | cp + strlen (var->id); | |
450 | const char *id = var->hidden_by ? var->hidden_by->id : | |
451 | var->id; | |
452 | location id_loc = var->hidden_by ? var->hidden_by->loc : | |
453 | var->loc; | |
454 | ||
455 | /* Create the explanation message. */ | |
456 | obstack_init (&msg_buf); | |
457 | ||
458 | obstack_fgrow1 (&msg_buf, _("possibly meant: %c"), dollar_or_at); | |
459 | if (contains_dot_or_dash (id)) | |
460 | obstack_fgrow1 (&msg_buf, "[%s]", id); | |
461 | else | |
462 | obstack_sgrow (&msg_buf, id); | |
463 | obstack_sgrow (&msg_buf, tail); | |
464 | ||
465 | if (var->err & VARIANT_HIDDEN) | |
466 | { | |
467 | obstack_fgrow1 (&msg_buf, _(", hiding %c"), dollar_or_at); | |
468 | if (contains_dot_or_dash (var->id)) | |
469 | obstack_fgrow1 (&msg_buf, "[%s]", var->id); | |
470 | else | |
471 | obstack_sgrow (&msg_buf, var->id); | |
472 | obstack_sgrow (&msg_buf, tail); | |
473 | } | |
474 | ||
475 | obstack_fgrow1 (&msg_buf, _(" at %s"), at_spec); | |
476 | ||
477 | if (var->err & VARIANT_NOT_VISIBLE_FROM_MIDRULE) | |
66381412 AR |
478 | { |
479 | const char *format = | |
480 | _(", cannot be accessed from mid-rule action at $%d"); | |
481 | obstack_fgrow1 (&msg_buf, format, midrule_rhs_index); | |
482 | } | |
ce268795 | 483 | |
e9690142 | 484 | obstack_1grow (&msg_buf, '\0'); |
ce268795 | 485 | if (is_warning) |
66381412 AR |
486 | warn_at_indent (id_loc, &indent, "%s", |
487 | (char *) obstack_finish (&msg_buf)); | |
ce268795 | 488 | else |
66381412 AR |
489 | complain_at_indent (id_loc, &indent, "%s", |
490 | (char *) obstack_finish (&msg_buf)); | |
e9690142 JD |
491 | obstack_free (&msg_buf, 0); |
492 | } | |
ce268795 AR |
493 | } |
494 | } | |
495 | ||
872b52bc AR |
496 | /* Returned from "parse_ref" when the reference |
497 | is inappropriate. */ | |
b9f1d9a4 | 498 | #define INVALID_REF (INT_MIN) |
872b52bc AR |
499 | |
500 | /* Returned from "parse_ref" when the reference | |
ce268795 | 501 | points to LHS ($$) of the current rule or midrule. */ |
b9f1d9a4 AR |
502 | #define LHS_REF (INT_MIN + 1) |
503 | ||
66381412 AR |
504 | /* Sub-messages indent. */ |
505 | #define SUB_INDENT (4) | |
506 | ||
676997e5 JD |
507 | /* Parse named or positional reference. In case of positional |
508 | references, can return negative values for $-n "deep" stack | |
509 | accesses. */ | |
b9f1d9a4 | 510 | static long int |
676997e5 | 511 | parse_ref (char *cp, symbol_list *rule, int rule_length, |
e9690142 JD |
512 | int midrule_rhs_index, char *text, location text_loc, |
513 | char dollar_or_at) | |
b9f1d9a4 | 514 | { |
676997e5 JD |
515 | symbol_list *l; |
516 | char *cp_end; | |
61bc57e5 | 517 | bool explicit_bracketing; |
872b52bc | 518 | unsigned i; |
ce268795 AR |
519 | unsigned valid_variants = 0; |
520 | unsigned valid_variant_index = 0; | |
b9f1d9a4 AR |
521 | |
522 | if ('$' == *cp) | |
523 | return LHS_REF; | |
524 | ||
676997e5 | 525 | if (c_isdigit (*cp) || (*cp == '-' && c_isdigit (* (cp + 1)))) |
b9f1d9a4 | 526 | { |
676997e5 | 527 | long int num = strtol (cp, &cp, 10); |
b9f1d9a4 | 528 | if (1 - INT_MAX + rule_length <= num && num <= rule_length) |
e9690142 | 529 | return num; |
b9f1d9a4 | 530 | else |
e9690142 JD |
531 | { |
532 | complain_at (text_loc, _("integer out of range: %s"), | |
ba4184ec | 533 | quote (text)); |
e9690142 JD |
534 | return INVALID_REF; |
535 | } | |
b9f1d9a4 AR |
536 | } |
537 | ||
676997e5 JD |
538 | if ('[' == *cp) |
539 | { | |
540 | /* Ignore the brackets. */ | |
541 | char *p; | |
542 | for (p = ++cp; *p != ']'; ++p) | |
e9690142 | 543 | continue; |
676997e5 JD |
544 | cp_end = p; |
545 | ||
546 | explicit_bracketing = true; | |
547 | } | |
548 | else | |
549 | { | |
550 | /* Take all characters of the name. */ | |
551 | char* p; | |
552 | for (p = cp; *p; ++p) | |
e9690142 JD |
553 | if (is_dot_or_dash (*p)) |
554 | { | |
555 | ref_tail_fields = p; | |
556 | break; | |
557 | } | |
676997e5 | 558 | for (p = cp; *p; ++p) |
e9690142 | 559 | continue; |
676997e5 JD |
560 | cp_end = p; |
561 | ||
562 | explicit_bracketing = false; | |
563 | } | |
b9f1d9a4 AR |
564 | |
565 | /* Add all relevant variants. */ | |
872b52bc | 566 | { |
ba4184ec | 567 | unsigned symbol_index; |
872b52bc | 568 | variant_count = 0; |
ba4184ec AR |
569 | for (symbol_index = 0, l = rule; !symbol_list_null (l); |
570 | ++symbol_index, l = l->next) | |
872b52bc | 571 | { |
e9690142 JD |
572 | variant *var; |
573 | if (l->content_type != SYMLIST_SYMBOL) | |
574 | continue; | |
b9f1d9a4 | 575 | |
e9690142 | 576 | var = variant_add (l->content.sym->tag, l->sym_loc, |
676997e5 | 577 | symbol_index, cp, cp_end, explicit_bracketing); |
e9690142 JD |
578 | if (var && l->named_ref) |
579 | var->hidden_by = l->named_ref; | |
b9f1d9a4 | 580 | |
e9690142 JD |
581 | if (l->named_ref) |
582 | variant_add (l->named_ref->id, l->named_ref->loc, | |
676997e5 | 583 | symbol_index, cp, cp_end, explicit_bracketing); |
872b52bc AR |
584 | } |
585 | } | |
b9f1d9a4 AR |
586 | |
587 | /* Check errors. */ | |
b9f1d9a4 AR |
588 | for (i = 0; i < variant_count; ++i) |
589 | { | |
ba4184ec | 590 | variant *var = &variant_table[i]; |
ce268795 | 591 | unsigned symbol_index = var->symbol_index; |
b9f1d9a4 AR |
592 | |
593 | /* Check visibility from mid-rule actions. */ | |
872b52bc | 594 | if (midrule_rhs_index != 0 |
e9690142 | 595 | && (symbol_index == 0 || midrule_rhs_index < symbol_index)) |
ce268795 | 596 | var->err |= VARIANT_NOT_VISIBLE_FROM_MIDRULE; |
b9f1d9a4 | 597 | |
676997e5 JD |
598 | /* Check correct bracketing. */ |
599 | if (!explicit_bracketing && contains_dot_or_dash (var->id)) | |
600 | var->err |= VARIANT_BAD_BRACKETING; | |
601 | ||
b9f1d9a4 | 602 | /* Check using of hidden symbols. */ |
ba4184ec | 603 | if (var->hidden_by) |
ce268795 | 604 | var->err |= VARIANT_HIDDEN; |
b9f1d9a4 | 605 | |
ba4184ec | 606 | if (!var->err) |
ce268795 AR |
607 | { |
608 | valid_variant_index = i; | |
609 | ++valid_variants; | |
610 | } | |
b9f1d9a4 AR |
611 | } |
612 | ||
ce268795 | 613 | switch (valid_variants) |
b9f1d9a4 | 614 | { |
ce268795 | 615 | case 0: |
f64e406b | 616 | { |
676997e5 JD |
617 | unsigned len = (explicit_bracketing || !ref_tail_fields) ? |
618 | cp_end - cp : ref_tail_fields - cp; | |
66381412 | 619 | unsigned indent = 0; |
f64e406b | 620 | |
66381412 AR |
621 | complain_at_indent (text_loc, &indent, _("invalid reference: %s"), |
622 | quote (text)); | |
623 | indent += SUB_INDENT; | |
5c9efc75 JD |
624 | if (len == 0) |
625 | { | |
626 | location sym_loc = text_loc; | |
627 | sym_loc.start.column += 1; | |
628 | sym_loc.end = sym_loc.start; | |
629 | const char *format = | |
ae93e4e4 JM |
630 | _("syntax error after '%c', expecting integer, letter," |
631 | " '_', '[', or '$'"); | |
676997e5 | 632 | complain_at_indent (sym_loc, &indent, format, dollar_or_at); |
5c9efc75 JD |
633 | } |
634 | else if (midrule_rhs_index) | |
66381412 AR |
635 | { |
636 | const char *format = | |
637 | _("symbol not found in production before $%d: %.*s"); | |
638 | complain_at_indent (rule->location, &indent, format, | |
639 | midrule_rhs_index, len, cp); | |
640 | } | |
f64e406b | 641 | else |
66381412 AR |
642 | { |
643 | const char *format = | |
644 | _("symbol not found in production: %.*s"); | |
645 | complain_at_indent (rule->location, &indent, format, | |
646 | len, cp); | |
647 | } | |
f64e406b AR |
648 | |
649 | if (variant_count > 0) | |
676997e5 JD |
650 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
651 | dollar_or_at, false, indent); | |
f64e406b AR |
652 | return INVALID_REF; |
653 | } | |
ce268795 AR |
654 | case 1: |
655 | { | |
66381412 | 656 | unsigned indent = 0; |
ce268795 AR |
657 | if (variant_count > 1) |
658 | { | |
66381412 AR |
659 | warn_at_indent (text_loc, &indent, _("misleading reference: %s"), |
660 | quote (text)); | |
676997e5 JD |
661 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
662 | dollar_or_at, true, indent + SUB_INDENT); | |
ce268795 AR |
663 | } |
664 | { | |
665 | unsigned symbol_index = | |
666 | variant_table[valid_variant_index].symbol_index; | |
667 | return (symbol_index == midrule_rhs_index) ? LHS_REF : symbol_index; | |
668 | } | |
669 | } | |
670 | case 2: | |
671 | default: | |
66381412 AR |
672 | { |
673 | unsigned indent = 0; | |
674 | complain_at_indent (text_loc, &indent, _("ambiguous reference: %s"), | |
675 | quote (text)); | |
676997e5 JD |
676 | show_sub_messages (cp, explicit_bracketing, midrule_rhs_index, |
677 | dollar_or_at, false, indent + SUB_INDENT); | |
66381412 AR |
678 | return INVALID_REF; |
679 | } | |
b9f1d9a4 AR |
680 | } |
681 | ||
ce268795 | 682 | /* Not reachable. */ |
b9f1d9a4 AR |
683 | return INVALID_REF; |
684 | } | |
685 | ||
e9071366 AD |
686 | /* Keeps track of the maximum number of semantic values to the left of |
687 | a handle (those referenced by $0, $-1, etc.) are required by the | |
688 | semantic actions of this grammar. */ | |
689 | int max_left_semantic_context = 0; | |
690 | ||
691 | ||
692 | /*------------------------------------------------------------------. | |
ae93e4e4 | 693 | | TEXT is pointing to a wannabee semantic value (i.e., a '$'). | |
e9071366 | 694 | | | |
676997e5 | 695 | | Possible inputs: $[<TYPENAME>]($|integer) | |
e9071366 AD |
696 | | | |
697 | | Output to OBSTACK_FOR_STRING a reference to this semantic value. | | |
698 | `------------------------------------------------------------------*/ | |
699 | ||
700 | static void | |
676997e5 | 701 | handle_action_dollar (symbol_list *rule, char *text, location dollar_loc) |
e9071366 | 702 | { |
28e52c0d | 703 | char const *type_name = NULL; |
e9071366 | 704 | char *cp = text + 1; |
b9f1d9a4 | 705 | char *gt_ptr = 0; |
ffa4ba3a | 706 | symbol_list *effective_rule; |
872b52bc AR |
707 | int effective_rule_length; |
708 | int n; | |
ffa4ba3a JD |
709 | |
710 | if (rule->midrule_parent_rule) | |
711 | { | |
712 | effective_rule = rule->midrule_parent_rule; | |
713 | effective_rule_length = rule->midrule_parent_rhs_index - 1; | |
714 | } | |
715 | else | |
716 | { | |
717 | effective_rule = rule; | |
718 | effective_rule_length = symbol_list_length (rule->next); | |
719 | } | |
e9071366 AD |
720 | |
721 | /* Get the type name if explicit. */ | |
722 | if (*cp == '<') | |
723 | { | |
724 | type_name = ++cp; | |
725 | while (*cp != '>') | |
e9690142 | 726 | ++cp; |
b9f1d9a4 AR |
727 | |
728 | /* The '>' symbol will be later replaced by '\0'. Original | |
e9690142 | 729 | 'text' is needed for error messages. */ |
b9f1d9a4 | 730 | gt_ptr = cp; |
e9071366 | 731 | ++cp; |
ddc8ede1 | 732 | if (untyped_var_seen) |
e9690142 | 733 | complain_at (dollar_loc, _("explicit type given in untyped grammar")); |
ddc8ede1 | 734 | tag_seen = true; |
e9071366 AD |
735 | } |
736 | ||
872b52bc | 737 | n = parse_ref (cp, effective_rule, effective_rule_length, |
e9690142 | 738 | rule->midrule_parent_rhs_index, text, dollar_loc, '$'); |
b9f1d9a4 AR |
739 | |
740 | if (gt_ptr) | |
741 | *gt_ptr = '\0'; | |
742 | ||
743 | switch (n) | |
e9071366 | 744 | { |
b9f1d9a4 AR |
745 | case INVALID_REF: |
746 | break; | |
747 | ||
748 | case LHS_REF: | |
e9071366 | 749 | if (!type_name) |
e9690142 | 750 | type_name = symbol_list_n_type_name_get (rule, dollar_loc, 0); |
ddc8ede1 PE |
751 | |
752 | if (!type_name) | |
e9690142 JD |
753 | { |
754 | if (union_seen | tag_seen) | |
755 | { | |
756 | if (rule->midrule_parent_rule) | |
757 | complain_at (dollar_loc, | |
4a9cd8f2 | 758 | _("$$ for the midrule at $%d of %s" |
e9690142 JD |
759 | " has no declared type"), |
760 | rule->midrule_parent_rhs_index, | |
4a9cd8f2 | 761 | quote (effective_rule->content.sym->tag)); |
e9690142 | 762 | else |
4a9cd8f2 AD |
763 | complain_at (dollar_loc, _("$$ of %s has no declared type"), |
764 | quote (rule->content.sym->tag)); | |
e9690142 JD |
765 | } |
766 | else | |
767 | untyped_var_seen = true; | |
768 | type_name = ""; | |
769 | } | |
ddc8ede1 | 770 | |
e9071366 | 771 | obstack_fgrow1 (&obstack_for_string, |
e9690142 | 772 | "]b4_lhs_value([%s])[", type_name); |
f6857bbf | 773 | rule->action_props.is_value_used = true; |
b9f1d9a4 AR |
774 | break; |
775 | ||
776 | default: | |
777 | if (max_left_semantic_context < 1 - n) | |
e9690142 | 778 | max_left_semantic_context = 1 - n; |
b9f1d9a4 | 779 | if (!type_name && 0 < n) |
e9690142 JD |
780 | type_name = |
781 | symbol_list_n_type_name_get (effective_rule, dollar_loc, n); | |
b9f1d9a4 | 782 | if (!type_name) |
e9690142 JD |
783 | { |
784 | if (union_seen | tag_seen) | |
4a9cd8f2 AD |
785 | complain_at (dollar_loc, _("$%s of %s has no declared type"), |
786 | cp, quote (effective_rule->content.sym->tag)); | |
e9690142 JD |
787 | else |
788 | untyped_var_seen = true; | |
789 | type_name = ""; | |
790 | } | |
b9f1d9a4 AR |
791 | |
792 | obstack_fgrow3 (&obstack_for_string, | |
e9690142 JD |
793 | "]b4_rhs_value(%d, %d, [%s])[", |
794 | effective_rule_length, n, type_name); | |
b9f1d9a4 | 795 | if (n > 0) |
e9690142 JD |
796 | symbol_list_n_get (effective_rule, n)->action_props.is_value_used = |
797 | true; | |
b9f1d9a4 | 798 | break; |
e9071366 AD |
799 | } |
800 | } | |
801 | ||
802 | ||
803 | /*------------------------------------------------------. | |
ae93e4e4 | 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 | |
676997e5 | 809 | handle_action_at (symbol_list *rule, char *text, location at_loc) |
e9071366 | 810 | { |
676997e5 JD |
811 | char *cp = text + 1; |
812 | symbol_list *effective_rule; | |
872b52bc AR |
813 | int effective_rule_length; |
814 | int n; | |
b9f1d9a4 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 | |
bc0f5737 | 827 | muscle_percent_define_ensure("locations", at_loc, true); |
e9071366 | 828 | |
872b52bc | 829 | n = parse_ref (cp, effective_rule, effective_rule_length, |
e9690142 | 830 | rule->midrule_parent_rhs_index, text, at_loc, '@'); |
b9f1d9a4 | 831 | switch (n) |
e9071366 | 832 | { |
b9f1d9a4 AR |
833 | case INVALID_REF: |
834 | break; | |
e9071366 | 835 | |
b9f1d9a4 AR |
836 | case LHS_REF: |
837 | obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); | |
838 | break; | |
839 | ||
840 | default: | |
841 | obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location(%d, %d)[", | |
e9690142 | 842 | effective_rule_length, n); |
b9f1d9a4 | 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 | |
872b52bc | 888 | code_props_plain_init (code_props *self, char const *code, |
e9690142 | 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; | |
b9f1d9a4 | 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; | |
b9f1d9a4 | 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, | |
b9f1d9a4 | 913 | location code_loc, symbol_list *rule, |
e9690142 | 914 | named_ref *name, bool is_predicate) |
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; | |
ba4184ec | 921 | self->named_ref = name; |
ca2a6d15 | 922 | self->is_predicate = is_predicate; |
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); | |
13cdf208 AR |
954 | variant_table_free (); |
955 | ||
e9071366 | 956 | /* Reclaim Flex's buffers. */ |
580b8926 | 957 | yylex_destroy (); |
e9071366 | 958 | } |