4416730cbe321b621de40c4623f91f2739b983d1
[bison.git] / src / scan-gram.l
1 /* Bison Grammar Scanner -*- C -*-
2
3 Copyright (C) 2002, 2003 Free Software Foundation, Inc.
4
5 This file is part of Bison, the GNU Compiler Compiler.
6
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2 of the License, or
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
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
20 02111-1307 USA
21 */
22
23 %option debug nodefault nounput noyywrap never-interactive
24 %option prefix="gram_" outfile="lex.yy.c"
25
26 %{
27 #include "system.h"
28
29 #include <mbswidth.h>
30 #include <get-errno.h>
31 #include <quote.h>
32
33 #include "complain.h"
34 #include "files.h"
35 #include "getargs.h"
36 #include "gram.h"
37 #include "reader.h"
38 #include "uniqstr.h"
39
40 #define YY_USER_INIT \
41 do \
42 { \
43 scanner_cursor.file = current_file; \
44 scanner_cursor.line = 1; \
45 scanner_cursor.column = 1; \
46 code_start = scanner_cursor; \
47 } \
48 while (0)
49
50 /* Location of scanner cursor. */
51 boundary scanner_cursor;
52
53 static void adjust_location (location *, char const *, size_t);
54 #define YY_USER_ACTION adjust_location (loc, yytext, yyleng);
55
56 static size_t no_cr_read (FILE *, char *, size_t);
57 #define YY_INPUT(buf, result, size) ((result) = no_cr_read (yyin, buf, size))
58
59
60 /* OBSTACK_FOR_STRING -- Used to store all the characters that we need to
61 keep (to construct ID, STRINGS etc.). Use the following macros to
62 use it.
63
64 Use STRING_GROW to append what has just been matched, and
65 STRING_FINISH to end the string (it puts the ending 0).
66 STRING_FINISH also stores this string in LAST_STRING, which can be
67 used, and which is used by STRING_FREE to free the last string. */
68
69 static struct obstack obstack_for_string;
70
71 /* A string representing the most recently saved token. */
72 static char *last_string;
73
74
75 #define STRING_GROW \
76 obstack_grow (&obstack_for_string, yytext, yyleng)
77
78 #define STRING_FINISH \
79 do { \
80 obstack_1grow (&obstack_for_string, '\0'); \
81 last_string = obstack_finish (&obstack_for_string); \
82 } while (0)
83
84 #define STRING_FREE \
85 obstack_free (&obstack_for_string, last_string)
86
87 void
88 scanner_last_string_free (void)
89 {
90 STRING_FREE;
91 }
92
93 /* Within well-formed rules, RULE_LENGTH is the number of values in
94 the current rule so far, which says where to find `$0' with respect
95 to the top of the stack. It is not the same as the rule->length in
96 the case of mid rule actions.
97
98 Outside of well-formed rules, RULE_LENGTH has an undefined value. */
99 static int rule_length;
100
101 static void handle_dollar (int token_type, char *cp, location loc);
102 static void handle_at (int token_type, char *cp, location loc);
103 static void handle_syncline (char *args);
104 static int convert_ucn_to_byte (char const *hex_text);
105 static void unexpected_eof (boundary, char const *);
106
107 %}
108 %x SC_COMMENT SC_LINE_COMMENT SC_YACC_COMMENT
109 %x SC_STRING SC_CHARACTER
110 %x SC_AFTER_IDENTIFIER
111 %x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER
112 %x SC_PRE_CODE SC_BRACED_CODE SC_PROLOGUE SC_EPILOGUE
113
114 letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]
115 id {letter}({letter}|[0-9])*
116 directive %{letter}({letter}|[0-9]|-)*
117 int [0-9]+
118
119 /* POSIX says that a tag must be both an id and a C union member, but
120 historically almost any character is allowed in a tag. We disallow
121 NUL and newline, as this simplifies our implementation. */
122 tag [^\0\n>]+
123
124 /* Zero or more instances of backslash-newline. Following GCC, allow
125 white space between the backslash and the newline. */
126 splice (\\[ \f\t\v]*\n)*
127
128 %%
129 %{
130 /* Nesting level of the current code in braces. */
131 int braces_level IF_LINT (= 0);
132
133 /* Parent context state, when applicable. */
134 int context_state IF_LINT (= 0);
135
136 /* Token type to return, when applicable. */
137 int token_type IF_LINT (= 0);
138
139 /* Location of most recent identifier, when applicable. */
140 location id_loc IF_LINT (= empty_location);
141
142 /* Where containing code started, when applicable. Its initial
143 value is relevant only when yylex is invoked in the SC_EPILOGUE
144 start condition. */
145 boundary code_start = scanner_cursor;
146
147 /* Where containing comment or string or character literal started,
148 when applicable. */
149 boundary token_start IF_LINT (= scanner_cursor);
150 %}
151
152
153 /*-----------------------.
154 | Scanning white space. |
155 `-----------------------*/
156
157 <INITIAL,SC_AFTER_IDENTIFIER,SC_PRE_CODE>
158 {
159 [ \f\n\t\v] ;
160 "," warn_at (*loc, _("stray `,' treated as white space"));
161
162 /* Comments. */
163 "//".* ;
164 "/*" {
165 token_start = loc->start;
166 context_state = YY_START;
167 BEGIN SC_YACC_COMMENT;
168 }
169
170 /* #line directives are not documented, and may be withdrawn or
171 modified in future versions of Bison. */
172 ^"#line "{int}" \"".*"\"\n" {
173 handle_syncline (yytext + sizeof "#line " - 1);
174 }
175 }
176
177
178 /*----------------------------.
179 | Scanning Bison directives. |
180 `----------------------------*/
181 <INITIAL>
182 {
183 "%binary" return PERCENT_NONASSOC;
184 "%debug" return PERCENT_DEBUG;
185 "%define" return PERCENT_DEFINE;
186 "%defines" return PERCENT_DEFINES;
187 "%destructor" token_type = PERCENT_DESTRUCTOR; BEGIN SC_PRE_CODE;
188 "%dprec" return PERCENT_DPREC;
189 "%error"[-_]"verbose" return PERCENT_ERROR_VERBOSE;
190 "%expect" return PERCENT_EXPECT;
191 "%file-prefix" return PERCENT_FILE_PREFIX;
192 "%fixed"[-_]"output"[-_]"files" return PERCENT_YACC;
193 "%glr-parser" return PERCENT_GLR_PARSER;
194 "%left" return PERCENT_LEFT;
195 "%lex-param" token_type = PERCENT_LEX_PARAM; BEGIN SC_PRE_CODE;
196 "%locations" return PERCENT_LOCATIONS;
197 "%merge" return PERCENT_MERGE;
198 "%name"[-_]"prefix" return PERCENT_NAME_PREFIX;
199 "%no"[-_]"lines" return PERCENT_NO_LINES;
200 "%nonassoc" return PERCENT_NONASSOC;
201 "%nterm" return PERCENT_NTERM;
202 "%output" return PERCENT_OUTPUT;
203 "%parse-param" token_type = PERCENT_PARSE_PARAM; BEGIN SC_PRE_CODE;
204 "%prec" rule_length--; return PERCENT_PREC;
205 "%printer" token_type = PERCENT_PRINTER; BEGIN SC_PRE_CODE;
206 "%pure"[-_]"parser" return PERCENT_PURE_PARSER;
207 "%right" return PERCENT_RIGHT;
208 "%skeleton" return PERCENT_SKELETON;
209 "%start" return PERCENT_START;
210 "%term" return PERCENT_TOKEN;
211 "%token" return PERCENT_TOKEN;
212 "%token"[-_]"table" return PERCENT_TOKEN_TABLE;
213 "%type" return PERCENT_TYPE;
214 "%union" token_type = PERCENT_UNION; BEGIN SC_PRE_CODE;
215 "%verbose" return PERCENT_VERBOSE;
216 "%yacc" return PERCENT_YACC;
217
218 {directive} {
219 complain_at (*loc, _("invalid directive: %s"), quote (yytext));
220 }
221
222 "=" return EQUAL;
223 "|" rule_length = 0; return PIPE;
224 ";" return SEMICOLON;
225
226 {id} {
227 val->symbol = symbol_get (yytext, *loc);
228 id_loc = *loc;
229 rule_length++;
230 BEGIN SC_AFTER_IDENTIFIER;
231 }
232
233 {int} {
234 unsigned long num;
235 set_errno (0);
236 num = strtoul (yytext, 0, 10);
237 if (INT_MAX < num || get_errno ())
238 {
239 complain_at (*loc, _("integer out of range: %s"), quote (yytext));
240 num = INT_MAX;
241 }
242 val->integer = num;
243 return INT;
244 }
245
246 /* Characters. We don't check there is only one. */
247 "'" STRING_GROW; token_start = loc->start; BEGIN SC_ESCAPED_CHARACTER;
248
249 /* Strings. */
250 "\"" STRING_GROW; token_start = loc->start; BEGIN SC_ESCAPED_STRING;
251
252 /* Prologue. */
253 "%{" code_start = loc->start; BEGIN SC_PROLOGUE;
254
255 /* Code in between braces. */
256 "{" {
257 STRING_GROW;
258 token_type = BRACED_CODE;
259 braces_level = 0;
260 code_start = loc->start;
261 BEGIN SC_BRACED_CODE;
262 }
263
264 /* A type. */
265 "<"{tag}">" {
266 obstack_grow (&obstack_for_string, yytext + 1, yyleng - 2);
267 STRING_FINISH;
268 val->uniqstr = uniqstr_new (last_string);
269 STRING_FREE;
270 return TYPE;
271 }
272
273 "%%" {
274 static int percent_percent_count;
275 if (++percent_percent_count == 2)
276 BEGIN SC_EPILOGUE;
277 return PERCENT_PERCENT;
278 }
279
280 . {
281 complain_at (*loc, _("invalid character: %s"), quote (yytext));
282 }
283
284 <<EOF>> {
285 loc->start = loc->end = scanner_cursor;
286 yyterminate ();
287 }
288 }
289
290
291 /*-----------------------------------------------------------------.
292 | Scanning after an identifier, checking whether a colon is next. |
293 `-----------------------------------------------------------------*/
294
295 <SC_AFTER_IDENTIFIER>
296 {
297 ":" {
298 rule_length = 0;
299 *loc = id_loc;
300 BEGIN INITIAL;
301 return ID_COLON;
302 }
303 . {
304 scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0);
305 yyless (0);
306 *loc = id_loc;
307 BEGIN INITIAL;
308 return ID;
309 }
310 <<EOF>> {
311 *loc = id_loc;
312 BEGIN INITIAL;
313 return ID;
314 }
315 }
316
317
318 /*---------------------------------------------------------------.
319 | Scanning a Yacc comment. The initial `/ *' is already eaten. |
320 `---------------------------------------------------------------*/
321
322 <SC_YACC_COMMENT>
323 {
324 "*/" BEGIN context_state;
325 .|\n ;
326 <<EOF>> unexpected_eof (token_start, "*/"); BEGIN context_state;
327 }
328
329
330 /*------------------------------------------------------------.
331 | Scanning a C comment. The initial `/ *' is already eaten. |
332 `------------------------------------------------------------*/
333
334 <SC_COMMENT>
335 {
336 "*"{splice}"/" STRING_GROW; BEGIN context_state;
337 <<EOF>> unexpected_eof (token_start, "*/"); BEGIN context_state;
338 }
339
340
341 /*--------------------------------------------------------------.
342 | Scanning a line comment. The initial `//' is already eaten. |
343 `--------------------------------------------------------------*/
344
345 <SC_LINE_COMMENT>
346 {
347 "\n" STRING_GROW; BEGIN context_state;
348 {splice} STRING_GROW;
349 <<EOF>> BEGIN context_state;
350 }
351
352
353 /*----------------------------------------------------------------.
354 | Scanning a C string, including its escapes. The initial `"' is |
355 | already eaten. |
356 `----------------------------------------------------------------*/
357
358 <SC_ESCAPED_STRING>
359 {
360 "\"" {
361 STRING_GROW;
362 STRING_FINISH;
363 loc->start = token_start;
364 val->chars = last_string;
365 rule_length++;
366 BEGIN INITIAL;
367 return STRING;
368 }
369
370 .|\n STRING_GROW;
371 <<EOF>> unexpected_eof (token_start, "\""); BEGIN INITIAL;
372 }
373
374 /*---------------------------------------------------------------.
375 | Scanning a C character, decoding its escapes. The initial "'" |
376 | is already eaten. |
377 `---------------------------------------------------------------*/
378
379 <SC_ESCAPED_CHARACTER>
380 {
381 "'" {
382 unsigned char last_string_1;
383 STRING_GROW;
384 STRING_FINISH;
385 loc->start = token_start;
386 val->symbol = symbol_get (last_string, *loc);
387 symbol_class_set (val->symbol, token_sym, *loc);
388 last_string_1 = last_string[1];
389 symbol_user_token_number_set (val->symbol, last_string_1, *loc);
390 STRING_FREE;
391 rule_length++;
392 BEGIN INITIAL;
393 return ID;
394 }
395
396 .|\n STRING_GROW;
397 <<EOF>> unexpected_eof (token_start, "'"); BEGIN INITIAL;
398 }
399
400
401 /*----------------------------.
402 | Decode escaped characters. |
403 `----------------------------*/
404
405 <SC_ESCAPED_STRING,SC_ESCAPED_CHARACTER>
406 {
407 \\[0-7]{1,3} {
408 unsigned long c = strtoul (yytext + 1, 0, 8);
409 if (UCHAR_MAX < c)
410 complain_at (*loc, _("invalid escape sequence: %s"), quote (yytext));
411 else
412 obstack_1grow (&obstack_for_string, c);
413 }
414
415 \\x[0-9abcdefABCDEF]+ {
416 unsigned long c;
417 set_errno (0);
418 c = strtoul (yytext + 2, 0, 16);
419 if (UCHAR_MAX < c || get_errno ())
420 complain_at (*loc, _("invalid escape sequence: %s"), quote (yytext));
421 else
422 obstack_1grow (&obstack_for_string, c);
423 }
424
425 \\a obstack_1grow (&obstack_for_string, '\a');
426 \\b obstack_1grow (&obstack_for_string, '\b');
427 \\f obstack_1grow (&obstack_for_string, '\f');
428 \\n obstack_1grow (&obstack_for_string, '\n');
429 \\r obstack_1grow (&obstack_for_string, '\r');
430 \\t obstack_1grow (&obstack_for_string, '\t');
431 \\v obstack_1grow (&obstack_for_string, '\v');
432
433 /* \\[\"\'?\\] would be shorter, but it confuses xgettext. */
434 \\("\""|"'"|"?"|"\\") obstack_1grow (&obstack_for_string, yytext[1]);
435
436 \\(u|U[0-9abcdefABCDEF]{4})[0-9abcdefABCDEF]{4} {
437 int c = convert_ucn_to_byte (yytext);
438 if (c < 0)
439 complain_at (*loc, _("invalid escape sequence: %s"), quote (yytext));
440 else
441 obstack_1grow (&obstack_for_string, c);
442 }
443 \\(.|\n) {
444 complain_at (*loc, _("unrecognized escape sequence: %s"), quote (yytext));
445 STRING_GROW;
446 }
447 }
448
449
450 /*----------------------------------------------------------.
451 | Scanning a C character without decoding its escapes. The |
452 | initial "'" is already eaten. |
453 `----------------------------------------------------------*/
454
455 <SC_CHARACTER>
456 {
457 "'" STRING_GROW; BEGIN context_state;
458 \\{splice}[^$@\[\]] STRING_GROW;
459 <<EOF>> unexpected_eof (token_start, "'"); BEGIN context_state;
460 }
461
462
463 /*----------------------------------------------------------------.
464 | Scanning a C string, without decoding its escapes. The initial |
465 | `"' is already eaten. |
466 `----------------------------------------------------------------*/
467
468 <SC_STRING>
469 {
470 "\"" STRING_GROW; BEGIN context_state;
471 \\{splice}[^$@\[\]] STRING_GROW;
472 <<EOF>> {
473 unexpected_eof (token_start, "\"");
474 BEGIN context_state;
475 }
476 }
477
478
479 /*---------------------------------------------------.
480 | Strings, comments etc. can be found in user code. |
481 `---------------------------------------------------*/
482
483 <SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE>
484 {
485 "'" {
486 STRING_GROW;
487 context_state = YY_START;
488 token_start = loc->start;
489 BEGIN SC_CHARACTER;
490 }
491 "\"" {
492 STRING_GROW;
493 context_state = YY_START;
494 token_start = loc->start;
495 BEGIN SC_STRING;
496 }
497 "/"{splice}"*" {
498 STRING_GROW;
499 context_state = YY_START;
500 token_start = loc->start;
501 BEGIN SC_COMMENT;
502 }
503 "/"{splice}"/" {
504 STRING_GROW;
505 context_state = YY_START;
506 BEGIN SC_LINE_COMMENT;
507 }
508 }
509
510
511 /*---------------------------------------------------------------.
512 | Scanning after %union etc., possibly followed by white space. |
513 | For %union only, allow arbitrary C code to appear before the |
514 | following brace, as an extension to POSIX. |
515 `---------------------------------------------------------------*/
516
517 <SC_PRE_CODE>
518 {
519 . {
520 bool valid = yytext[0] == '{' || token_type == PERCENT_UNION;
521 scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0);
522 yyless (0);
523
524 if (valid)
525 {
526 braces_level = -1;
527 code_start = loc->start;
528 BEGIN SC_BRACED_CODE;
529 }
530 else
531 {
532 complain_at (*loc, _("missing `{' in `%s'"),
533 token_name (token_type));
534 obstack_sgrow (&obstack_for_string, "{}");
535 STRING_FINISH;
536 val->chars = last_string;
537 BEGIN INITIAL;
538 return token_type;
539 }
540 }
541
542 <<EOF>> unexpected_eof (scanner_cursor, "{}"); BEGIN INITIAL;
543 }
544
545
546 /*---------------------------------------------------------------.
547 | Scanning some code in braces (%union and actions). The initial |
548 | "{" is already eaten. |
549 `---------------------------------------------------------------*/
550
551 <SC_BRACED_CODE>
552 {
553 "{"|"<"{splice}"%" STRING_GROW; braces_level++;
554 "%"{splice}">" STRING_GROW; braces_level--;
555 "}" {
556 bool outer_brace = --braces_level < 0;
557
558 /* As an undocumented Bison extension, append `;' before the last
559 brace in braced code, so that the user code can omit trailing
560 `;'. But do not append `;' if emulating Yacc, since Yacc does
561 not append one.
562
563 FIXME: Bison should warn if a semicolon seems to be necessary
564 here, and should omit the semicolon if it seems unnecessary
565 (e.g., after ';', '{', or '}', each followed by comments or
566 white space). Such a warning shouldn't depend on --yacc; it
567 should depend on a new --pedantic option, which would cause
568 Bison to warn if it detects an extension to POSIX. --pedantic
569 should also diagnose other Bison extensions like %yacc.
570 Perhaps there should also be a GCC-style --pedantic-errors
571 option, so that such warnings are diagnosed as errors. */
572 if (outer_brace && token_type == BRACED_CODE && ! yacc_flag)
573 obstack_1grow (&obstack_for_string, ';');
574
575 obstack_1grow (&obstack_for_string, '}');
576
577 if (outer_brace)
578 {
579 STRING_FINISH;
580 rule_length++;
581 loc->start = code_start;
582 val->chars = last_string;
583 BEGIN INITIAL;
584 return token_type;
585 }
586 }
587
588 /* Tokenize `<<%' correctly (as `<<' `%') rather than incorrrectly
589 (as `<' `<%'). */
590 "<"{splice}"<" STRING_GROW;
591
592 "$"("<"{tag}">")?(-?[0-9]+|"$") handle_dollar (token_type, yytext, *loc);
593 "@"(-?[0-9]+|"$") handle_at (token_type, yytext, *loc);
594
595 <<EOF>> unexpected_eof (code_start, "}"); BEGIN INITIAL;
596 }
597
598
599 /*--------------------------------------------------------------.
600 | Scanning some prologue: from "%{" (already scanned) to "%}". |
601 `--------------------------------------------------------------*/
602
603 <SC_PROLOGUE>
604 {
605 "%}" {
606 STRING_FINISH;
607 loc->start = code_start;
608 val->chars = last_string;
609 BEGIN INITIAL;
610 return PROLOGUE;
611 }
612
613 <<EOF>> unexpected_eof (code_start, "%}"); BEGIN INITIAL;
614 }
615
616
617 /*---------------------------------------------------------------.
618 | Scanning the epilogue (everything after the second "%%", which |
619 | has already been eaten). |
620 `---------------------------------------------------------------*/
621
622 <SC_EPILOGUE>
623 {
624 <<EOF>> {
625 STRING_FINISH;
626 loc->start = code_start;
627 val->chars = last_string;
628 BEGIN INITIAL;
629 return EPILOGUE;
630 }
631 }
632
633
634 /*----------------------------------------------------------------.
635 | By default, grow the string obstack with the input, escaping M4 |
636 | quoting characters. |
637 `----------------------------------------------------------------*/
638
639 <SC_COMMENT,SC_LINE_COMMENT,SC_STRING,SC_CHARACTER,SC_BRACED_CODE,SC_PROLOGUE,SC_EPILOGUE>
640 {
641 \$ obstack_sgrow (&obstack_for_string, "$][");
642 \@ obstack_sgrow (&obstack_for_string, "@@");
643 \[ obstack_sgrow (&obstack_for_string, "@{");
644 \] obstack_sgrow (&obstack_for_string, "@}");
645 .|\n STRING_GROW;
646 }
647
648
649 %%
650
651 /* Set *LOC and adjust scanner cursor to account for token TOKEN of
652 size SIZE. */
653
654 static void
655 adjust_location (location *loc, char const *token, size_t size)
656 {
657 int line = scanner_cursor.line;
658 int column = scanner_cursor.column;
659 char const *p0 = token;
660 char const *p = token;
661 char const *lim = token + size;
662
663 loc->start = scanner_cursor;
664
665 for (p = token; p < lim; p++)
666 switch (*p)
667 {
668 case '\n':
669 line++;
670 column = 1;
671 p0 = p + 1;
672 break;
673
674 case '\t':
675 column += mbsnwidth (p0, p - p0, 0);
676 column += 8 - ((column - 1) & 7);
677 p0 = p + 1;
678 break;
679 }
680
681 scanner_cursor.line = line;
682 scanner_cursor.column = column + mbsnwidth (p0, p - p0, 0);
683
684 loc->end = scanner_cursor;
685 }
686
687
688 /* Read bytes from FP into buffer BUF of size SIZE. Return the
689 number of bytes read. Remove '\r' from input, treating \r\n
690 and isolated \r as \n. */
691
692 static size_t
693 no_cr_read (FILE *fp, char *buf, size_t size)
694 {
695 size_t bytes_read = fread (buf, 1, size, fp);
696 if (bytes_read)
697 {
698 char *w = memchr (buf, '\r', bytes_read);
699 if (w)
700 {
701 char const *r = ++w;
702 char const *lim = buf + bytes_read;
703
704 for (;;)
705 {
706 /* Found an '\r'. Treat it like '\n', but ignore any
707 '\n' that immediately follows. */
708 w[-1] = '\n';
709 if (r == lim)
710 {
711 int ch = getc (fp);
712 if (ch != '\n' && ungetc (ch, fp) != ch)
713 break;
714 }
715 else if (*r == '\n')
716 r++;
717
718 /* Copy until the next '\r'. */
719 do
720 {
721 if (r == lim)
722 return w - buf;
723 }
724 while ((*w++ = *r++) != '\r');
725 }
726
727 return w - buf;
728 }
729 }
730
731 return bytes_read;
732 }
733
734
735 /*------------------------------------------------------------------.
736 | TEXT is pointing to a wannabee semantic value (i.e., a `$'). |
737 | |
738 | Possible inputs: $[<TYPENAME>]($|integer) |
739 | |
740 | Output to OBSTACK_FOR_STRING a reference to this semantic value. |
741 `------------------------------------------------------------------*/
742
743 static inline bool
744 handle_action_dollar (char *text, location loc)
745 {
746 const char *type_name = NULL;
747 char *cp = text + 1;
748
749 if (! current_rule)
750 return false;
751
752 /* Get the type name if explicit. */
753 if (*cp == '<')
754 {
755 type_name = ++cp;
756 while (*cp != '>')
757 ++cp;
758 *cp = '\0';
759 ++cp;
760 }
761
762 if (*cp == '$')
763 {
764 if (!type_name)
765 type_name = symbol_list_n_type_name_get (current_rule, loc, 0);
766 if (!type_name && typed)
767 complain_at (loc, _("$$ of `%s' has no declared type"),
768 current_rule->sym->tag);
769 if (!type_name)
770 type_name = "";
771 obstack_fgrow1 (&obstack_for_string,
772 "]b4_lhs_value([%s])[", type_name);
773 }
774 else
775 {
776 long num;
777 set_errno (0);
778 num = strtol (cp, 0, 10);
779
780 if (INT_MIN <= num && num <= rule_length && ! get_errno ())
781 {
782 int n = num;
783 if (!type_name && n > 0)
784 type_name = symbol_list_n_type_name_get (current_rule, loc, n);
785 if (!type_name && typed)
786 complain_at (loc, _("$%d of `%s' has no declared type"),
787 n, current_rule->sym->tag);
788 if (!type_name)
789 type_name = "";
790 obstack_fgrow3 (&obstack_for_string,
791 "]b4_rhs_value([%d], [%d], [%s])[",
792 rule_length, n, type_name);
793 }
794 else
795 complain_at (loc, _("integer out of range: %s"), quote (text));
796 }
797
798 return true;
799 }
800
801
802 /*-----------------------------------------------------------------.
803 | Dispatch onto handle_action_dollar, or handle_destructor_dollar, |
804 | depending upon TOKEN_TYPE. |
805 `-----------------------------------------------------------------*/
806
807 static void
808 handle_dollar (int token_type, char *text, location loc)
809 {
810 switch (token_type)
811 {
812 case BRACED_CODE:
813 if (handle_action_dollar (text, loc))
814 return;
815 break;
816
817 case PERCENT_DESTRUCTOR:
818 case PERCENT_PRINTER:
819 if (text[1] == '$')
820 {
821 obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar[");
822 return;
823 }
824 break;
825
826 default:
827 break;
828 }
829
830 complain_at (loc, _("invalid value: %s"), quote (text));
831 }
832
833
834 /*------------------------------------------------------.
835 | TEXT is a location token (i.e., a `@...'). Output to |
836 | OBSTACK_FOR_STRING a reference to this location. |
837 `------------------------------------------------------*/
838
839 static inline bool
840 handle_action_at (char *text, location loc)
841 {
842 char *cp = text + 1;
843 locations_flag = 1;
844
845 if (! current_rule)
846 return false;
847
848 if (*cp == '$')
849 obstack_sgrow (&obstack_for_string, "]b4_lhs_location[");
850 else
851 {
852 long num;
853 set_errno (0);
854 num = strtol (cp, 0, 10);
855
856 if (INT_MIN <= num && num <= rule_length && ! get_errno ())
857 {
858 int n = num;
859 obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location([%d], [%d])[",
860 rule_length, n);
861 }
862 else
863 complain_at (loc, _("integer out of range: %s"), quote (text));
864 }
865
866 return true;
867 }
868
869
870 /*-------------------------------------------------------------------.
871 | Dispatch onto handle_action_at, or handle_destructor_at, depending |
872 | upon CODE_KIND. |
873 `-------------------------------------------------------------------*/
874
875 static void
876 handle_at (int token_type, char *text, location loc)
877 {
878 switch (token_type)
879 {
880 case BRACED_CODE:
881 handle_action_at (text, loc);
882 return;
883
884 case PERCENT_DESTRUCTOR:
885 case PERCENT_PRINTER:
886 if (text[1] == '$')
887 {
888 obstack_sgrow (&obstack_for_string, "]b4_at_dollar[");
889 return;
890 }
891 break;
892
893 default:
894 break;
895 }
896
897 complain_at (loc, _("invalid value: %s"), quote (text));
898 }
899
900
901 /*------------------------------------------------------------------.
902 | Convert universal character name UCN to a single-byte character, |
903 | and return that character. Return -1 if UCN does not correspond |
904 | to a single-byte character. |
905 `------------------------------------------------------------------*/
906
907 static int
908 convert_ucn_to_byte (char const *ucn)
909 {
910 unsigned long code = strtoul (ucn + 2, 0, 16);
911
912 /* FIXME: Currently we assume Unicode-compatible unibyte characters
913 on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes). On
914 non-ASCII hosts we support only the portable C character set.
915 These limitations should be removed once we add support for
916 multibyte characters. */
917
918 if (UCHAR_MAX < code)
919 return -1;
920
921 #if ! ('$' == 0x24 && '@' == 0x40 && '`' == 0x60 && '~' == 0x7e)
922 {
923 /* A non-ASCII host. Use CODE to index into a table of the C
924 basic execution character set, which is guaranteed to exist on
925 all Standard C platforms. This table also includes '$', '@',
926 and '`', which are not in the basic execution character set but
927 which are unibyte characters on all the platforms that we know
928 about. */
929 static signed char const table[] =
930 {
931 '\0', -1, -1, -1, -1, -1, -1, '\a',
932 '\b', '\t', '\n', '\v', '\f', '\r', -1, -1,
933 -1, -1, -1, -1, -1, -1, -1, -1,
934 -1, -1, -1, -1, -1, -1, -1, -1,
935 ' ', '!', '"', '#', '$', '%', '&', '\'',
936 '(', ')', '*', '+', ',', '-', '.', '/',
937 '0', '1', '2', '3', '4', '5', '6', '7',
938 '8', '9', ':', ';', '<', '=', '>', '?',
939 '@', 'A', 'B', 'C', 'D', 'E', 'F', 'G',
940 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O',
941 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
942 'X', 'Y', 'Z', '[', '\\', ']', '^', '_',
943 '`', 'a', 'b', 'c', 'd', 'e', 'f', 'g',
944 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o',
945 'p', 'q', 'r', 's', 't', 'u', 'v', 'w',
946 'x', 'y', 'z', '{', '|', '}', '~'
947 };
948
949 code = code < sizeof table ? table[code] : -1;
950 }
951 #endif
952
953 return code;
954 }
955
956
957 /*----------------------------------------------------------------.
958 | Handle `#line INT "FILE"'. ARGS has already skipped `#line '. |
959 `----------------------------------------------------------------*/
960
961 static void
962 handle_syncline (char *args)
963 {
964 int lineno = strtol (args, &args, 10);
965 const char *file = NULL;
966 file = strchr (args, '"') + 1;
967 *strchr (file, '"') = 0;
968 scanner_cursor.file = current_file = xstrdup (file);
969 scanner_cursor.line = lineno;
970 scanner_cursor.column = 1;
971 }
972
973
974 /*------------------------------------------------------------------------.
975 | Report an unexpected EOF in a token or comment starting at START. |
976 | An end of file was encountered and the expected TOKEN_END was missing. |
977 `------------------------------------------------------------------------*/
978
979 static void
980 unexpected_eof (boundary start, char const *token_end)
981 {
982 location loc;
983 loc.start = start;
984 loc.end = scanner_cursor;
985 complain_at (loc, _("missing `%s' at end of file"), token_end);
986 }
987
988
989 /*-------------------------.
990 | Initialize the scanner. |
991 `-------------------------*/
992
993 void
994 scanner_initialize (void)
995 {
996 obstack_init (&obstack_for_string);
997 }
998
999
1000 /*-----------------------------------------------.
1001 | Free all the memory allocated to the scanner. |
1002 `-----------------------------------------------*/
1003
1004 void
1005 scanner_free (void)
1006 {
1007 obstack_free (&obstack_for_string, 0);
1008 /* Reclaim Flex's buffers. */
1009 yy_delete_buffer (YY_CURRENT_BUFFER);
1010 }