1 /* Lexical analyzer for C and Objective C.
2 Copyright (C) 1987, 88, 89, 92, 94-97, 1998 Free Software Foundation, Inc.
4 This file is part of GNU CC.
6 GNU CC is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2, or (at your option)
11 GNU CC is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with GNU CC; see the file COPYING. If not, write to
18 the Free Software Foundation, 59 Temple Place - Suite 330,
19 Boston, MA 02111-1307, USA. */
36 /* MULTIBYTE_CHARS support only works for native compilers.
37 ??? Ideally what we want is to model widechar support after
38 the current floating point support. */
40 #undef MULTIBYTE_CHARS
43 #ifdef MULTIBYTE_CHARS
49 extern cpp_reader parse_in;
50 extern cpp_options parse_options;
52 /* Stream for reading from the input file. */
56 extern void yyprint PROTO((FILE *, int, YYSTYPE));
58 /* The elements of `ridpointers' are identifier nodes
59 for the reserved type names and storage classes.
60 It is indexed by a RID_... value. */
61 tree ridpointers[(int) RID_MAX];
63 /* Cause the `yydebug' variable to be defined. */
67 extern unsigned char *yy_cur, *yy_lim;
69 extern int yy_get_token ();
71 #define GETC() (yy_cur < yy_lim ? *yy_cur++ : yy_get_token ())
72 #define UNGETC(c) ((c), yy_cur--)
74 #define GETC() getc (finput)
75 #define UNGETC(c) ungetc (c, finput)
78 /* the declaration found for the last IDENTIFIER token read in.
79 yylex must look this up to detect typedefs, which get token type TYPENAME,
80 so it is left around in case the identifier is not a typedef but is
81 used in a context which makes it a reference to a variable. */
84 /* Nonzero enables objc features. */
90 /* File used for outputting assembler code. */
91 extern FILE *asm_out_file;
93 #ifndef WCHAR_TYPE_SIZE
95 #define WCHAR_TYPE_SIZE INT_TYPE_SIZE
97 #define WCHAR_TYPE_SIZE BITS_PER_WORD
101 /* Number of bytes in a wide character. */
102 #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT)
104 static int maxtoken; /* Current nominal length of token buffer. */
105 char *token_buffer; /* Pointer to token buffer.
106 Actual allocated length is maxtoken + 2.
107 This is not static because objc-parse.y uses it. */
109 static int indent_level = 0; /* Number of { minus number of }. */
111 /* Nonzero if end-of-file has been seen on input. */
112 static int end_of_file;
115 /* Buffered-back input character; faster than using ungetc. */
116 static int nextchar = -1;
119 #ifdef HANDLE_SYSV_PRAGMA
120 static int handle_sysv_pragma PROTO((int));
121 #endif /* HANDLE_SYSV_PRAGMA */
122 static int whitespace_cr PROTO((int));
123 static int skip_white_space PROTO((int));
124 static int skip_white_space_on_line PROTO((void));
125 static char *extend_token_buffer PROTO((char *));
126 static int readescape PROTO((int *));
128 /* Do not insert generated code into the source, instead, include it.
129 This allows us to build gcc automatically even for targets that
130 need to add or modify the reserved keyword lists. */
133 /* Return something to represent absolute declarators containing a *.
134 TARGET is the absolute declarator that the * contains.
135 TYPE_QUALS is a list of modifiers such as const or volatile
136 to apply to the pointer type, represented as identifiers.
138 We return an INDIRECT_REF whose "contents" are TARGET
139 and whose type is the modifier list. */
142 make_pointer_declarator (type_quals, target)
143 tree type_quals, target;
145 return build1 (INDIRECT_REF, type_quals, target);
149 forget_protocol_qualifiers ()
151 int i, n = sizeof wordlist / sizeof (struct resword);
153 for (i = 0; i < n; i++)
154 if ((int) wordlist[i].rid >= (int) RID_IN
155 && (int) wordlist[i].rid <= (int) RID_ONEWAY)
156 wordlist[i].name = "";
160 remember_protocol_qualifiers ()
162 int i, n = sizeof wordlist / sizeof (struct resword);
164 for (i = 0; i < n; i++)
165 if (wordlist[i].rid == RID_IN)
166 wordlist[i].name = "in";
167 else if (wordlist[i].rid == RID_OUT)
168 wordlist[i].name = "out";
169 else if (wordlist[i].rid == RID_INOUT)
170 wordlist[i].name = "inout";
171 else if (wordlist[i].rid == RID_BYCOPY)
172 wordlist[i].name = "bycopy";
173 else if (wordlist[i].rid == RID_ONEWAY)
174 wordlist[i].name = "oneway";
178 init_parse (filename)
182 /* Open input file. */
183 if (filename == 0 || !strcmp (filename, "-"))
189 finput = fopen (filename, "r");
191 pfatal_with_name (filename);
193 #ifdef IO_BUFFER_SIZE
194 setvbuf (finput, (char *) xmalloc (IO_BUFFER_SIZE), _IOFBF, IO_BUFFER_SIZE);
196 #endif /* !USE_CPPLIB */
204 parse_in.show_column = 1;
205 if (! cpp_start_read (&parse_in, filename))
216 cpp_finish (&parse_in);
225 /* Make identifier nodes long enough for the language-specific slots. */
226 set_identifier_size (sizeof (struct lang_identifier));
228 /* Start it at 0, because check_newline is called at the very beginning
229 and will increment it to 1. */
232 #ifdef MULTIBYTE_CHARS
233 /* Change to the native locale for multibyte conversions. */
234 setlocale (LC_CTYPE, "");
238 token_buffer = (char *) xmalloc (maxtoken + 2);
240 ridpointers[(int) RID_INT] = get_identifier ("int");
241 ridpointers[(int) RID_CHAR] = get_identifier ("char");
242 ridpointers[(int) RID_VOID] = get_identifier ("void");
243 ridpointers[(int) RID_FLOAT] = get_identifier ("float");
244 ridpointers[(int) RID_DOUBLE] = get_identifier ("double");
245 ridpointers[(int) RID_SHORT] = get_identifier ("short");
246 ridpointers[(int) RID_LONG] = get_identifier ("long");
247 ridpointers[(int) RID_UNSIGNED] = get_identifier ("unsigned");
248 ridpointers[(int) RID_SIGNED] = get_identifier ("signed");
249 ridpointers[(int) RID_INLINE] = get_identifier ("inline");
250 ridpointers[(int) RID_CONST] = get_identifier ("const");
251 ridpointers[(int) RID_VOLATILE] = get_identifier ("volatile");
252 ridpointers[(int) RID_AUTO] = get_identifier ("auto");
253 ridpointers[(int) RID_STATIC] = get_identifier ("static");
254 ridpointers[(int) RID_EXTERN] = get_identifier ("extern");
255 ridpointers[(int) RID_TYPEDEF] = get_identifier ("typedef");
256 ridpointers[(int) RID_REGISTER] = get_identifier ("register");
257 ridpointers[(int) RID_ITERATOR] = get_identifier ("iterator");
258 ridpointers[(int) RID_COMPLEX] = get_identifier ("complex");
259 ridpointers[(int) RID_ID] = get_identifier ("id");
260 ridpointers[(int) RID_IN] = get_identifier ("in");
261 ridpointers[(int) RID_OUT] = get_identifier ("out");
262 ridpointers[(int) RID_INOUT] = get_identifier ("inout");
263 ridpointers[(int) RID_BYCOPY] = get_identifier ("bycopy");
264 ridpointers[(int) RID_ONEWAY] = get_identifier ("oneway");
265 forget_protocol_qualifiers();
267 /* Some options inhibit certain reserved words.
268 Clear those words out of the hash table so they won't be recognized. */
269 #define UNSET_RESERVED_WORD(STRING) \
270 do { struct resword *s = is_reserved_word (STRING, sizeof (STRING) - 1); \
271 if (s) s->name = ""; } while (0)
273 if (! doing_objc_thang)
274 UNSET_RESERVED_WORD ("id");
276 if (flag_traditional)
278 UNSET_RESERVED_WORD ("const");
279 UNSET_RESERVED_WORD ("volatile");
280 UNSET_RESERVED_WORD ("typeof");
281 UNSET_RESERVED_WORD ("signed");
282 UNSET_RESERVED_WORD ("inline");
283 UNSET_RESERVED_WORD ("iterator");
284 UNSET_RESERVED_WORD ("complex");
288 UNSET_RESERVED_WORD ("asm");
289 UNSET_RESERVED_WORD ("typeof");
290 UNSET_RESERVED_WORD ("inline");
291 UNSET_RESERVED_WORD ("iterator");
292 UNSET_RESERVED_WORD ("complex");
297 reinit_parse_for_function ()
301 /* Function used when yydebug is set, to print a token in more detail. */
304 yyprint (file, yychar, yylval)
316 if (IDENTIFIER_POINTER (t))
317 fprintf (file, " `%s'", IDENTIFIER_POINTER (t));
322 if (TREE_CODE (t) == INTEGER_CST)
324 #if HOST_BITS_PER_WIDE_INT == 64
325 #if HOST_BITS_PER_WIDE_INT == HOST_BITS_PER_INT
328 #if HOST_BITS_PER_WIDE_INT == HOST_BITS_PER_LONG
335 #if HOST_BITS_PER_WIDE_INT != HOST_BITS_PER_INT
341 TREE_INT_CST_HIGH (t), TREE_INT_CST_LOW (t));
346 /* Iff C is a carriage return, warn about it - if appropriate -
347 and return nonzero. */
352 static int newline_warning = 0;
356 /* ANSI C says the effects of a carriage return in a source file
358 if (pedantic && !newline_warning)
360 warning ("carriage return in source file");
361 warning ("(we only warn about the first carriage return)");
369 /* If C is not whitespace, return C.
370 Otherwise skip whitespace and return first nonwhite char read. */
380 /* We don't recognize comments here, because
381 cpp output can include / and * consecutively as operators.
382 Also, there's no need, since cpp removes all comments. */
385 c = check_newline ();
406 error ("stray '\\' in program");
416 /* Skips all of the white space at the current location in the input file.
417 Must use and reset nextchar if it has the next character. */
420 position_after_white_space ()
426 c = nextchar, nextchar = -1;
431 UNGETC (skip_white_space (c));
434 /* Like skip_white_space, but don't advance beyond the end of line.
435 Moreover, we don't get passed a character to start with. */
437 skip_white_space_on_line ()
466 /* Make the token buffer longer, preserving the data in it.
467 P should point to just beyond the last valid character in the old buffer.
468 The value we return is a pointer to the new buffer
469 at a place corresponding to P. */
472 extend_token_buffer (p)
475 int offset = p - token_buffer;
477 maxtoken = maxtoken * 2 + 10;
478 token_buffer = (char *) xrealloc (token_buffer, maxtoken + 2);
480 return token_buffer + offset;
483 /* At the beginning of a line, increment the line number
484 and process any #-directive on this line.
485 If the line is a #-directive, read the entire line and return a newline.
486 Otherwise, return the line's first non-whitespace character. */
496 /* Read first nonwhite char on the line. */
499 while (c == ' ' || c == '\t')
504 /* If not #, return it so caller will use it. */
508 /* Read first nonwhite char after the `#'. */
511 while (c == ' ' || c == '\t')
514 /* If a letter follows, then if the word here is `line', skip
515 it and ignore it; otherwise, ignore the line, with an error
516 if the word isn't `pragma', `ident', `define', or `undef'. */
518 if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'))
527 && ((c = GETC()) == ' ' || c == '\t' || c == '\n'
528 || whitespace_cr (c) ))
530 while (c == ' ' || c == '\t' || whitespace_cr (c))
534 #ifdef HANDLE_SYSV_PRAGMA
537 if (token != IDENTIFIER)
539 return handle_sysv_pragma (token);
540 #else /* !HANDLE_SYSV_PRAGMA */
545 if (token != IDENTIFIER)
547 if (HANDLE_PRAGMA (finput, yylval.ttype))
553 ??? do not know what to do ???;
554 #endif /* !USE_CPPLIB */
555 #endif /* HANDLE_PRAGMA */
556 #endif /* !HANDLE_SYSV_PRAGMA */
568 && ((c = GETC()) == ' ' || c == '\t' || c == '\n'))
571 debug_define (lineno, GET_DIRECTIVE_LINE ());
581 && ((c = GETC()) == ' ' || c == '\t' || c == '\n'))
584 debug_undef (lineno, GET_DIRECTIVE_LINE ());
593 && ((c = GETC()) == ' ' || c == '\t'))
602 && ((c = GETC()) == ' ' || c == '\t'))
604 /* #ident. The pedantic warning is now in cccp.c. */
606 /* Here we have just seen `#ident '.
607 A string constant should follow. */
609 c = skip_white_space_on_line ();
611 /* If no argument, ignore the line. */
618 || TREE_CODE (yylval.ttype) != STRING_CST)
620 error ("invalid #ident");
626 #ifdef ASM_OUTPUT_IDENT
627 ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (yylval.ttype));
631 /* Skip the rest of this line. */
636 error ("undefined or invalid # directive");
641 /* Here we have either `#line' or `# <nonletter>'.
642 In either case, it should be a line number; a digit should follow. */
644 /* Can't use skip_white_space here, but must handle all whitespace
645 that is not '\n', lest we get a recursion for '\r' '\n' when
648 c = skip_white_space_on_line ();
650 /* If the # is the only nonwhite char on the line,
651 just ignore it. Check the new newline. */
655 /* Something follows the #; read a token. */
660 if (token == CONSTANT
661 && TREE_CODE (yylval.ttype) == INTEGER_CST)
663 int old_lineno = lineno;
665 /* subtract one, because it is the following line that
666 gets the specified number */
668 int l = TREE_INT_CST_LOW (yylval.ttype) - 1;
670 /* Is this the last nonwhite stuff on the line? */
671 c = skip_white_space_on_line ();
674 /* No more: store the line number and check following line. */
680 /* More follows: it must be a string constant (filename). */
682 /* Read the string constant. */
685 if (token != STRING || TREE_CODE (yylval.ttype) != STRING_CST)
687 error ("invalid #line");
692 = (char *) permalloc (TREE_STRING_LENGTH (yylval.ttype) + 1);
693 strcpy (input_filename, TREE_STRING_POINTER (yylval.ttype));
696 /* Each change of file name
697 reinitializes whether we are now in a system header. */
698 in_system_header = 0;
700 if (main_input_filename == 0)
701 main_input_filename = input_filename;
703 /* Is this the last nonwhite stuff on the line? */
704 c = skip_white_space_on_line ();
707 /* Update the name in the top element of input_file_stack. */
708 if (input_file_stack)
709 input_file_stack->name = input_filename;
718 /* `1' after file name means entering new file.
719 `2' after file name means just left a file. */
721 if (token == CONSTANT
722 && TREE_CODE (yylval.ttype) == INTEGER_CST)
724 if (TREE_INT_CST_LOW (yylval.ttype) == 1)
726 /* Pushing to a new file. */
728 = (struct file_stack *) xmalloc (sizeof (struct file_stack));
729 input_file_stack->line = old_lineno;
730 p->next = input_file_stack;
731 p->name = input_filename;
732 p->indent_level = indent_level;
733 input_file_stack = p;
734 input_file_stack_tick++;
735 debug_start_source_file (input_filename);
738 else if (TREE_INT_CST_LOW (yylval.ttype) == 2)
740 /* Popping out of a file. */
741 if (input_file_stack->next)
743 struct file_stack *p = input_file_stack;
744 if (indent_level != p->indent_level)
746 warning_with_file_and_line
747 (p->name, old_lineno,
748 "This file contains more `%c's than `%c's.",
749 indent_level > p->indent_level ? '{' : '}',
750 indent_level > p->indent_level ? '}' : '{');
752 input_file_stack = p->next;
754 input_file_stack_tick++;
755 debug_end_source_file (input_file_stack->line);
758 error ("#-lines for entering and leaving files don't match");
764 /* Now that we've pushed or popped the input stack,
765 update the name in the top element. */
766 if (input_file_stack)
767 input_file_stack->name = input_filename;
769 /* If we have handled a `1' or a `2',
770 see if there is another number to read. */
773 /* Is this the last nonwhite stuff on the line? */
774 c = skip_white_space_on_line ();
783 /* `3' after file name means this is a system header file. */
785 if (token == CONSTANT
786 && TREE_CODE (yylval.ttype) == INTEGER_CST
787 && TREE_INT_CST_LOW (yylval.ttype) == 3)
788 in_system_header = 1, used_up = 1;
792 /* Is this the last nonwhite stuff on the line? */
793 c = skip_white_space_on_line ();
799 warning ("unrecognized text at end of #line");
802 error ("invalid #-line");
804 /* skip the rest of this line. */
807 if (c != '\n' && c != EOF && nextchar >= 0)
808 c = nextchar, nextchar = -1;
810 while (c != '\n' && c != EOF)
815 #ifdef HANDLE_SYSV_PRAGMA
817 /* Handle a #pragma directive.
818 TOKEN is the token we read after `#pragma'. Processes the entire input
819 line and returns a character for the caller to reread: either \n or EOF. */
821 /* This function has to be in this file, in order to get at
825 handle_sysv_pragma (token)
838 handle_pragma_token (token_buffer, yylval.ttype);
841 handle_pragma_token (token_buffer, 0);
845 c = nextchar, nextchar = -1;
850 while (c == ' ' || c == '\t')
852 if (c == '\n' || c == EOF)
854 handle_pragma_token (0, 0);
862 #endif /* HANDLE_SYSV_PRAGMA */
864 #define ENDFILE -1 /* token that represents end-of-file */
866 /* Read an escape sequence, returning its equivalent as a character,
867 or store 1 in *ignore_ptr if it is backslash-newline. */
870 readescape (ignore_ptr)
873 register int c = GETC();
875 register unsigned count;
876 unsigned firstdig = 0;
882 if (warn_traditional)
883 warning ("the meaning of `\\x' varies with -traditional");
885 if (flag_traditional)
894 if (!(c >= 'a' && c <= 'f')
895 && !(c >= 'A' && c <= 'F')
896 && !(c >= '0' && c <= '9'))
902 if (c >= 'a' && c <= 'f')
903 code += c - 'a' + 10;
904 if (c >= 'A' && c <= 'F')
905 code += c - 'A' + 10;
906 if (c >= '0' && c <= '9')
908 if (code != 0 || count != 0)
917 error ("\\x used with no following hex digits");
919 /* Digits are all 0's. Ok. */
921 else if ((count - 1) * 4 >= TYPE_PRECISION (integer_type_node)
923 && ((1 << (TYPE_PRECISION (integer_type_node) - (count - 1) * 4))
925 pedwarn ("hex escape out of range");
928 case '0': case '1': case '2': case '3': case '4':
929 case '5': case '6': case '7':
932 while ((c <= '7') && (c >= '0') && (count++ < 3))
934 code = (code * 8) + (c - '0');
940 case '\\': case '\'': case '"':
949 return TARGET_NEWLINE;
964 if (warn_traditional)
965 warning ("the meaning of `\\a' varies with -traditional");
967 if (flag_traditional)
972 #if 0 /* Vertical tab is present in common usage compilers. */
973 if (flag_traditional)
981 pedwarn ("non-ANSI-standard escape sequence, `\\%c'", c);
987 /* `\(', etc, are used at beginning of line to avoid confusing Emacs. */
991 /* `\%' is used to prevent SCCS from getting confused. */
994 pedwarn ("non-ANSI escape sequence `\\%c'", c);
997 if (c >= 040 && c < 0177)
998 pedwarn ("unknown escape sequence `\\%c'", c);
1000 pedwarn ("unknown escape sequence: `\\' followed by char code 0x%x", c);
1010 strcpy (buf, string);
1012 /* We can't print string and character constants well
1013 because the token_buffer contains the result of processing escapes. */
1015 strcat (buf, " at end of input");
1016 else if (token_buffer[0] == 0)
1017 strcat (buf, " at null character");
1018 else if (token_buffer[0] == '"')
1019 strcat (buf, " before string constant");
1020 else if (token_buffer[0] == '\'')
1021 strcat (buf, " before character constant");
1022 else if (token_buffer[0] < 040 || (unsigned char) token_buffer[0] >= 0177)
1023 sprintf (buf + strlen (buf), " before character 0%o",
1024 (unsigned char) token_buffer[0]);
1026 strcat (buf, " before `%s'");
1028 error (buf, token_buffer);
1038 char long_long_flag;
1041 struct try_type type_sequence[] =
1043 { &integer_type_node, 0, 0, 0},
1044 { &unsigned_type_node, 1, 0, 0},
1045 { &long_integer_type_node, 0, 1, 0},
1046 { &long_unsigned_type_node, 1, 1, 0},
1047 { &long_long_integer_type_node, 0, 1, 1},
1048 { &long_long_unsigned_type_node, 1, 1, 1}
1063 c = nextchar, nextchar = -1;
1068 /* Effectively do c = skip_white_space (c)
1069 but do it faster in the usual cases. */
1082 /* Call skip_white_space so we can warn if appropriate. */
1087 c = skip_white_space (c);
1089 goto found_nonwhite;
1093 token_buffer[0] = c;
1094 token_buffer[1] = 0;
1096 /* yylloc.first_line = lineno; */
1102 token_buffer[0] = 0;
1107 /* Capital L may start a wide-string or wide-character constant. */
1109 register int c = GETC();
1118 goto string_constant;
1125 if (!doing_objc_thang)
1132 /* '@' may start a constant string object. */
1133 register int c = GETC ();
1137 goto string_constant;
1140 /* Fall through to treat '@' as the start of an identifier. */
1143 case 'A': case 'B': case 'C': case 'D': case 'E':
1144 case 'F': case 'G': case 'H': case 'I': case 'J':
1145 case 'K': case 'M': case 'N': case 'O':
1146 case 'P': case 'Q': case 'R': case 'S': case 'T':
1147 case 'U': case 'V': case 'W': case 'X': case 'Y':
1149 case 'a': case 'b': case 'c': case 'd': case 'e':
1150 case 'f': case 'g': case 'h': case 'i': case 'j':
1151 case 'k': case 'l': case 'm': case 'n': case 'o':
1152 case 'p': case 'q': case 'r': case 's': case 't':
1153 case 'u': case 'v': case 'w': case 'x': case 'y':
1159 while (ISALNUM (c) || c == '_' || c == '$' || c == '@')
1161 /* Make sure this char really belongs in an identifier. */
1162 if (c == '@' && ! doing_objc_thang)
1166 if (! dollars_in_ident)
1167 error ("`$' in identifier");
1169 pedwarn ("`$' in identifier");
1172 if (p >= token_buffer + maxtoken)
1173 p = extend_token_buffer (p);
1189 /* Try to recognize a keyword. Uses minimum-perfect hash function */
1192 register struct resword *ptr;
1194 if ((ptr = is_reserved_word (token_buffer, p - token_buffer)))
1197 yylval.ttype = ridpointers[(int) ptr->rid];
1198 value = (int) ptr->token;
1200 /* Only return OBJECTNAME if it is a typedef. */
1201 if (doing_objc_thang && value == OBJECTNAME)
1203 lastiddecl = lookup_name(yylval.ttype);
1205 if (lastiddecl == NULL_TREE
1206 || TREE_CODE (lastiddecl) != TYPE_DECL)
1210 /* Even if we decided to recognize asm, still perhaps warn. */
1212 && (value == ASM_KEYWORD || value == TYPEOF
1213 || ptr->rid == RID_INLINE)
1214 && token_buffer[0] != '_')
1215 pedwarn ("ANSI does not permit the keyword `%s'",
1220 /* If we did not find a keyword, look for an identifier
1223 if (value == IDENTIFIER)
1225 if (token_buffer[0] == '@')
1226 error("invalid identifier `%s'", token_buffer);
1228 yylval.ttype = get_identifier (token_buffer);
1229 lastiddecl = lookup_name (yylval.ttype);
1231 if (lastiddecl != 0 && TREE_CODE (lastiddecl) == TYPE_DECL)
1233 /* A user-invisible read-only initialized variable
1234 should be replaced by its value.
1235 We handle only strings since that's the only case used in C. */
1236 else if (lastiddecl != 0 && TREE_CODE (lastiddecl) == VAR_DECL
1237 && DECL_IGNORED_P (lastiddecl)
1238 && TREE_READONLY (lastiddecl)
1239 && DECL_INITIAL (lastiddecl) != 0
1240 && TREE_CODE (DECL_INITIAL (lastiddecl)) == STRING_CST)
1242 tree stringval = DECL_INITIAL (lastiddecl);
1244 /* Copy the string value so that we won't clobber anything
1245 if we put something in the TREE_CHAIN of this one. */
1246 yylval.ttype = build_string (TREE_STRING_LENGTH (stringval),
1247 TREE_STRING_POINTER (stringval));
1250 else if (doing_objc_thang)
1252 tree objc_interface_decl = is_class_name (yylval.ttype);
1254 if (objc_interface_decl)
1257 yylval.ttype = objc_interface_decl;
1267 /* Check first for common special case: single-digit 0 or 1. */
1270 UNGETC (next_c); /* Always undo this lookahead. */
1271 if (!ISALNUM (next_c) && next_c != '.')
1273 token_buffer[0] = (char)c, token_buffer[1] = '\0';
1274 yylval.ttype = (c == '0') ? integer_zero_node : integer_one_node;
1280 case '2': case '3': case '4':
1281 case '5': case '6': case '7': case '8': case '9':
1286 int largest_digit = 0;
1288 /* for multi-precision arithmetic,
1289 we actually store only HOST_BITS_PER_CHAR bits in each part.
1290 The number of parts is chosen so as to be sufficient to hold
1291 the enough bits to fit into the two HOST_WIDE_INTs that contain
1292 the integer value (this is always at least as many bits as are
1293 in a target `long long' value, but may be wider). */
1294 #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2 + 2)
1295 int parts[TOTAL_PARTS];
1298 enum anon1 { NOT_FLOAT, AFTER_POINT, TOO_MANY_POINTS} floatflag
1301 for (count = 0; count < TOTAL_PARTS; count++)
1309 *p++ = (c = GETC());
1310 if ((c == 'x') || (c == 'X'))
1313 *p++ = (c = GETC());
1315 /* Leading 0 forces octal unless the 0 is the only digit. */
1316 else if (c >= '0' && c <= '9')
1325 /* Read all the digits-and-decimal-points. */
1328 || (ISALNUM (c) && c != 'l' && c != 'L'
1329 && c != 'u' && c != 'U'
1330 && c != 'i' && c != 'I' && c != 'j' && c != 'J'
1331 && (floatflag == NOT_FLOAT || ((c != 'f') && (c != 'F')))))
1336 error ("floating constant may not be in radix 16");
1337 if (floatflag == TOO_MANY_POINTS)
1338 /* We have already emitted an error. Don't need another. */
1340 else if (floatflag == AFTER_POINT)
1342 error ("malformed floating constant");
1343 floatflag = TOO_MANY_POINTS;
1344 /* Avoid another error from atof by forcing all characters
1345 from here on to be ignored. */
1349 floatflag = AFTER_POINT;
1353 /* Accept '.' as the start of a floating-point number
1354 only when it is followed by a digit.
1355 Otherwise, unread the following non-digit
1356 and use the '.' as a structural token. */
1357 if (p == token_buffer + 2 && !ISDIGIT (c))
1368 error ("parse error at `..'");
1371 token_buffer[1] = 0;
1378 /* It is not a decimal point.
1379 It should be a digit (perhaps a hex digit). */
1385 else if (base <= 10)
1387 if (c == 'e' || c == 'E')
1390 floatflag = AFTER_POINT;
1391 break; /* start of exponent */
1393 error ("nondigits in number and not hexadecimal");
1404 if (c >= largest_digit)
1408 for (count = 0; count < TOTAL_PARTS; count++)
1410 parts[count] *= base;
1414 += (parts[count-1] >> HOST_BITS_PER_CHAR);
1416 &= (1 << HOST_BITS_PER_CHAR) - 1;
1422 /* If the extra highest-order part ever gets anything in it,
1423 the number is certainly too big. */
1424 if (parts[TOTAL_PARTS - 1] != 0)
1427 if (p >= token_buffer + maxtoken - 3)
1428 p = extend_token_buffer (p);
1429 *p++ = (c = GETC());
1434 error ("numeric constant with no digits");
1436 if (largest_digit >= base)
1437 error ("numeric constant contains digits beyond the radix");
1439 /* Remove terminating char from the token buffer and delimit the string */
1442 if (floatflag != NOT_FLOAT)
1444 tree type = double_type_node;
1446 int conversion_errno = 0;
1447 REAL_VALUE_TYPE value;
1450 /* Read explicit exponent if any, and put it in tokenbuf. */
1452 if ((c == 'e') || (c == 'E'))
1454 if (p >= token_buffer + maxtoken - 3)
1455 p = extend_token_buffer (p);
1458 if ((c == '+') || (c == '-'))
1464 error ("floating constant exponent has no digits");
1467 if (p >= token_buffer + maxtoken - 3)
1468 p = extend_token_buffer (p);
1476 /* Convert string to a double, checking for overflow. */
1477 if (setjmp (handler))
1479 error ("floating constant out of range");
1484 int fflag = 0, lflag = 0;
1485 /* Copy token_buffer now, while it has just the number
1486 and not the suffixes; once we add `f' or `i',
1487 REAL_VALUE_ATOF may not work any more. */
1488 char *copy = (char *) alloca (p - token_buffer + 1);
1489 bcopy (token_buffer, copy, p - token_buffer + 1);
1491 set_float_handler (handler);
1497 /* Read the suffixes to choose a data type. */
1502 error ("more than one `f' in numeric constant");
1508 error ("more than one `l' in numeric constant");
1514 error ("more than one `i' or `j' in numeric constant");
1516 pedwarn ("ANSI C forbids imaginary numeric constants");
1527 if (p >= token_buffer + maxtoken - 3)
1528 p = extend_token_buffer (p);
1534 /* The second argument, machine_mode, of REAL_VALUE_ATOF
1535 tells the desired precision of the binary result
1536 of decimal-to-binary conversion. */
1541 error ("both `f' and `l' in floating constant");
1543 type = float_type_node;
1545 value = REAL_VALUE_ATOF (copy, TYPE_MODE (type));
1546 conversion_errno = errno;
1547 /* A diagnostic is required here by some ANSI C testsuites.
1548 This is not pedwarn, become some people don't want
1549 an error for this. */
1550 if (REAL_VALUE_ISINF (value) && pedantic)
1551 warning ("floating point number exceeds range of `float'");
1555 type = long_double_type_node;
1557 value = REAL_VALUE_ATOF (copy, TYPE_MODE (type));
1558 conversion_errno = errno;
1559 if (REAL_VALUE_ISINF (value) && pedantic)
1560 warning ("floating point number exceeds range of `long double'");
1565 value = REAL_VALUE_ATOF (copy, TYPE_MODE (type));
1566 conversion_errno = errno;
1567 if (REAL_VALUE_ISINF (value) && pedantic)
1568 warning ("floating point number exceeds range of `double'");
1571 set_float_handler (NULL_PTR);
1574 /* ERANGE is also reported for underflow,
1575 so test the value to distinguish overflow from that. */
1576 if (conversion_errno == ERANGE && !flag_traditional && pedantic
1577 && (REAL_VALUES_LESS (dconst1, value)
1578 || REAL_VALUES_LESS (value, dconstm1)))
1579 warning ("floating point number exceeds range of `double'");
1582 /* If the result is not a number, assume it must have been
1583 due to some error message above, so silently convert
1585 if (REAL_VALUE_ISNAN (value))
1588 /* Create a node with determined type and value. */
1590 yylval.ttype = build_complex (NULL_TREE,
1591 convert (type, integer_zero_node),
1592 build_real (type, value));
1594 yylval.ttype = build_real (type, value);
1598 tree traditional_type, ansi_type, type;
1599 HOST_WIDE_INT high, low;
1600 int spec_unsigned = 0;
1602 int spec_long_long = 0;
1606 traditional_type = ansi_type = type = NULL_TREE;
1609 if (c == 'u' || c == 'U')
1612 error ("two `u's in integer constant");
1615 else if (c == 'l' || c == 'L')
1620 error ("three `l's in integer constant");
1622 pedwarn ("ANSI C forbids long long integer constants");
1627 else if (c == 'i' || c == 'j' || c == 'I' || c == 'J')
1630 error ("more than one `i' or `j' in numeric constant");
1632 pedwarn ("ANSI C forbids imaginary numeric constants");
1637 if (p >= token_buffer + maxtoken - 3)
1638 p = extend_token_buffer (p);
1643 /* If the constant won't fit in an unsigned long long,
1644 then warn that the constant is out of range. */
1646 /* ??? This assumes that long long and long integer types are
1647 a multiple of 8 bits. This better than the original code
1648 though which assumed that long was exactly 32 bits and long
1649 long was exactly 64 bits. */
1651 bytes = TYPE_PRECISION (long_long_integer_type_node) / 8;
1654 for (i = bytes; i < TOTAL_PARTS; i++)
1658 pedwarn ("integer constant out of range");
1660 /* This is simplified by the fact that our constant
1661 is always positive. */
1665 for (i = 0; i < HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; i++)
1667 high |= ((HOST_WIDE_INT) parts[i + (HOST_BITS_PER_WIDE_INT
1668 / HOST_BITS_PER_CHAR)]
1669 << (i * HOST_BITS_PER_CHAR));
1670 low |= (HOST_WIDE_INT) parts[i] << (i * HOST_BITS_PER_CHAR);
1673 yylval.ttype = build_int_2 (low, high);
1674 TREE_TYPE (yylval.ttype) = long_long_unsigned_type_node;
1676 /* If warn_traditional, calculate both the ANSI type and the
1677 traditional type, then see if they disagree.
1678 Otherwise, calculate only the type for the dialect in use. */
1679 if (warn_traditional || flag_traditional)
1681 /* Calculate the traditional type. */
1682 /* Traditionally, any constant is signed;
1683 but if unsigned is specified explicitly, obey that.
1684 Use the smallest size with the right number of bits,
1685 except for one special case with decimal constants. */
1686 if (! spec_long && base != 10
1687 && int_fits_type_p (yylval.ttype, unsigned_type_node))
1688 traditional_type = (spec_unsigned ? unsigned_type_node
1689 : integer_type_node);
1690 /* A decimal constant must be long
1691 if it does not fit in type int.
1692 I think this is independent of whether
1693 the constant is signed. */
1694 else if (! spec_long && base == 10
1695 && int_fits_type_p (yylval.ttype, integer_type_node))
1696 traditional_type = (spec_unsigned ? unsigned_type_node
1697 : integer_type_node);
1698 else if (! spec_long_long)
1699 traditional_type = (spec_unsigned ? long_unsigned_type_node
1700 : long_integer_type_node);
1702 traditional_type = (spec_unsigned
1703 ? long_long_unsigned_type_node
1704 : long_long_integer_type_node);
1706 if (warn_traditional || ! flag_traditional)
1708 /* Calculate the ANSI type. */
1709 if (! spec_long && ! spec_unsigned
1710 && int_fits_type_p (yylval.ttype, integer_type_node))
1711 ansi_type = integer_type_node;
1712 else if (! spec_long && (base != 10 || spec_unsigned)
1713 && int_fits_type_p (yylval.ttype, unsigned_type_node))
1714 ansi_type = unsigned_type_node;
1715 else if (! spec_unsigned && !spec_long_long
1716 && int_fits_type_p (yylval.ttype, long_integer_type_node))
1717 ansi_type = long_integer_type_node;
1718 else if (! spec_long_long
1719 && int_fits_type_p (yylval.ttype,
1720 long_unsigned_type_node))
1721 ansi_type = long_unsigned_type_node;
1722 else if (! spec_unsigned
1723 && int_fits_type_p (yylval.ttype,
1724 long_long_integer_type_node))
1725 ansi_type = long_long_integer_type_node;
1727 ansi_type = long_long_unsigned_type_node;
1730 type = flag_traditional ? traditional_type : ansi_type;
1732 if (warn_traditional && traditional_type != ansi_type)
1734 if (TYPE_PRECISION (traditional_type)
1735 != TYPE_PRECISION (ansi_type))
1736 warning ("width of integer constant changes with -traditional");
1737 else if (TREE_UNSIGNED (traditional_type)
1738 != TREE_UNSIGNED (ansi_type))
1739 warning ("integer constant is unsigned in ANSI C, signed with -traditional");
1741 warning ("width of integer constant may change on other systems with -traditional");
1744 if (pedantic && !flag_traditional && !spec_long_long && !warn
1745 && (TYPE_PRECISION (long_integer_type_node)
1746 < TYPE_PRECISION (type)))
1747 pedwarn ("integer constant out of range");
1749 if (base == 10 && ! spec_unsigned && TREE_UNSIGNED (type))
1750 warning ("decimal constant is so large that it is unsigned");
1754 if (TYPE_PRECISION (type)
1755 <= TYPE_PRECISION (integer_type_node))
1757 = build_complex (NULL_TREE, integer_zero_node,
1758 convert (integer_type_node,
1761 error ("complex integer constant is too wide for `complex int'");
1763 else if (flag_traditional && !int_fits_type_p (yylval.ttype, type))
1764 /* The traditional constant 0x80000000 is signed
1765 but doesn't fit in the range of int.
1766 This will change it to -0x80000000, which does fit. */
1768 TREE_TYPE (yylval.ttype) = unsigned_type (type);
1769 yylval.ttype = convert (type, yylval.ttype);
1770 TREE_OVERFLOW (yylval.ttype)
1771 = TREE_CONSTANT_OVERFLOW (yylval.ttype) = 0;
1774 TREE_TYPE (yylval.ttype) = type;
1780 if (ISALNUM (c) || c == '.' || c == '_' || c == '$'
1781 || (!flag_traditional && (c == '-' || c == '+')
1782 && (p[-1] == 'e' || p[-1] == 'E')))
1783 error ("missing white space after number `%s'", token_buffer);
1785 value = CONSTANT; break;
1791 register int result = 0;
1792 register int num_chars = 0;
1793 unsigned width = TYPE_PRECISION (char_type_node);
1798 width = WCHAR_TYPE_SIZE;
1799 #ifdef MULTIBYTE_CHARS
1800 max_chars = MB_CUR_MAX;
1806 max_chars = TYPE_PRECISION (integer_type_node) / width;
1814 if (c == '\'' || c == EOF)
1820 c = readescape (&ignore);
1823 if (width < HOST_BITS_PER_INT
1824 && (unsigned) c >= (1 << width))
1825 pedwarn ("escape sequence out of range for character");
1826 #ifdef MAP_CHARACTER
1828 c = MAP_CHARACTER (c);
1834 pedwarn ("ANSI C forbids newline in character constant");
1837 #ifdef MAP_CHARACTER
1839 c = MAP_CHARACTER (c);
1843 if (num_chars > maxtoken - 4)
1844 extend_token_buffer (token_buffer);
1846 token_buffer[num_chars] = c;
1848 /* Merge character into result; ignore excess chars. */
1849 if (num_chars < max_chars + 1)
1851 if (width < HOST_BITS_PER_INT)
1852 result = (result << width) | (c & ((1 << width) - 1));
1858 token_buffer[num_chars + 1] = '\'';
1859 token_buffer[num_chars + 2] = 0;
1862 error ("malformatted character constant");
1863 else if (num_chars == 0)
1864 error ("empty character constant");
1865 else if (num_chars > max_chars)
1867 num_chars = max_chars;
1868 error ("character constant too long");
1870 else if (num_chars != 1 && ! flag_traditional)
1871 warning ("multi-character character constant");
1873 /* If char type is signed, sign-extend the constant. */
1876 int num_bits = num_chars * width;
1878 /* We already got an error; avoid invalid shift. */
1879 yylval.ttype = build_int_2 (0, 0);
1880 else if (TREE_UNSIGNED (char_type_node)
1881 || ((result >> (num_bits - 1)) & 1) == 0)
1883 = build_int_2 (result & (~(unsigned HOST_WIDE_INT) 0
1884 >> (HOST_BITS_PER_WIDE_INT - num_bits)),
1888 = build_int_2 (result | ~(~(unsigned HOST_WIDE_INT) 0
1889 >> (HOST_BITS_PER_WIDE_INT - num_bits)),
1891 TREE_TYPE (yylval.ttype) = integer_type_node;
1895 #ifdef MULTIBYTE_CHARS
1896 /* Set the initial shift state and convert the next sequence. */
1898 /* In all locales L'\0' is zero and mbtowc will return zero,
1901 || (num_chars == 1 && token_buffer[1] != '\0'))
1904 (void) mbtowc (NULL_PTR, NULL_PTR, 0);
1905 if (mbtowc (& wc, token_buffer + 1, num_chars) == num_chars)
1908 warning ("Ignoring invalid multibyte character");
1911 yylval.ttype = build_int_2 (result, 0);
1912 TREE_TYPE (yylval.ttype) = wchar_type_node;
1923 p = token_buffer + 1;
1925 while (c != '"' && c >= 0)
1930 c = readescape (&ignore);
1934 && TYPE_PRECISION (char_type_node) < HOST_BITS_PER_INT
1935 && c >= (1 << TYPE_PRECISION (char_type_node)))
1936 pedwarn ("escape sequence out of range for character");
1941 pedwarn ("ANSI C forbids newline in string constant");
1945 if (p == token_buffer + maxtoken)
1946 p = extend_token_buffer (p);
1955 error ("Unterminated string constant");
1957 /* We have read the entire constant.
1958 Construct a STRING_CST for the result. */
1962 /* If this is a L"..." wide-string, convert the multibyte string
1963 to a wide character string. */
1964 char *widep = (char *) alloca ((p - token_buffer) * WCHAR_BYTES);
1967 #ifdef MULTIBYTE_CHARS
1968 len = mbstowcs ((wchar_t *) widep, token_buffer + 1, p - token_buffer);
1969 if (len < 0 || len >= (p - token_buffer))
1971 warning ("Ignoring invalid multibyte string");
1974 bzero (widep + (len * WCHAR_BYTES), WCHAR_BYTES);
1979 wp = widep + (BYTES_BIG_ENDIAN ? WCHAR_BYTES - 1 : 0);
1980 bzero (widep, (p - token_buffer) * WCHAR_BYTES);
1981 for (cp = token_buffer + 1; cp < p; cp++)
1982 *wp = *cp, wp += WCHAR_BYTES;
1983 len = p - token_buffer - 1;
1986 yylval.ttype = build_string ((len + 1) * WCHAR_BYTES, widep);
1987 TREE_TYPE (yylval.ttype) = wchar_array_type_node;
1992 /* Return an Objective-C @"..." constant string object. */
1993 yylval.ttype = build_objc_string (p - token_buffer,
1995 TREE_TYPE (yylval.ttype) = char_array_type_node;
1996 value = OBJC_STRING;
2000 yylval.ttype = build_string (p - token_buffer, token_buffer + 1);
2001 TREE_TYPE (yylval.ttype) = char_array_type_node;
2032 yylval.code = PLUS_EXPR; break;
2034 yylval.code = MINUS_EXPR; break;
2036 yylval.code = BIT_AND_EXPR; break;
2038 yylval.code = BIT_IOR_EXPR; break;
2040 yylval.code = MULT_EXPR; break;
2042 yylval.code = TRUNC_DIV_EXPR; break;
2044 yylval.code = TRUNC_MOD_EXPR; break;
2046 yylval.code = BIT_XOR_EXPR; break;
2048 yylval.code = LSHIFT_EXPR; break;
2050 yylval.code = RSHIFT_EXPR; break;
2052 yylval.code = LT_EXPR; break;
2054 yylval.code = GT_EXPR; break;
2057 token_buffer[1] = c1 = GETC();
2058 token_buffer[2] = 0;
2065 value = ARITHCOMPARE; yylval.code = LE_EXPR; goto done;
2067 value = ARITHCOMPARE; yylval.code = GE_EXPR; goto done;
2069 value = EQCOMPARE; yylval.code = NE_EXPR; goto done;
2071 value = EQCOMPARE; yylval.code = EQ_EXPR; goto done;
2073 value = ASSIGN; goto done;
2079 value = PLUSPLUS; goto done;
2081 value = MINUSMINUS; goto done;
2083 value = ANDAND; goto done;
2085 value = OROR; goto done;
2098 { value = POINTSAT; goto done; }
2102 { value = ']'; goto done; }
2106 { value = '{'; indent_level++; goto done; }
2108 { value = '['; goto done; }
2112 { value = '}'; indent_level--; goto done; }
2116 token_buffer[1] = 0;
2118 if ((c == '<') || (c == '>'))
2119 value = ARITHCOMPARE;
2125 /* Don't make yyparse think this is eof. */
2144 /* yylloc.last_line = lineno; */
2149 /* Sets the value of the 'yydebug' variable to VALUE.
2150 This is a function so we don't have to have YYDEBUG defined
2151 in order to build the compiler. */
2160 warning ("YYDEBUG not defined.");