1 /* Lexical analyzer for C and Objective C.
2 Copyright (C) 1987, 1988, 1989, 1992 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, 675 Mass Ave, Cambridge, MA 02139, USA. */
34 #ifdef MULTIBYTE_CHARS
43 /* The elements of `ridpointers' are identifier nodes
44 for the reserved type names and storage classes.
45 It is indexed by a RID_... value. */
46 tree ridpointers
[(int) RID_MAX
];
48 /* Cause the `yydebug' variable to be defined. */
51 /* the declaration found for the last IDENTIFIER token read in.
52 yylex must look this up to detect typedefs, which get token type TYPENAME,
53 so it is left around in case the identifier is not a typedef but is
54 used in a context which makes it a reference to a variable. */
57 /* Nonzero enables objc features. */
61 extern tree
lookup_interface ();
65 /* File used for outputting assembler code. */
66 extern FILE *asm_out_file
;
68 #ifndef WCHAR_TYPE_SIZE
70 #define WCHAR_TYPE_SIZE INT_TYPE_SIZE
72 #define WCHAR_TYPE_SIZE BITS_PER_WORD
76 /* Number of bytes in a wide character. */
77 #define WCHAR_BYTES (WCHAR_TYPE_SIZE / BITS_PER_UNIT)
79 static int maxtoken
; /* Current nominal length of token buffer. */
80 char *token_buffer
; /* Pointer to token buffer.
81 Actual allocated length is maxtoken + 2.
82 This is not static because objc-parse.y uses it. */
84 /* Nonzero if end-of-file has been seen on input. */
85 static int end_of_file
;
87 /* Buffered-back input character; faster than using ungetc. */
88 static int nextchar
= -1;
92 /* Nonzero tells yylex to ignore \ in string constants. */
93 static int ignore_escape_flag
= 0;
95 /* C code produced by gperf version 2.5 (GNU C++ version) */
96 /* Command-line: gperf -p -j1 -i 1 -g -o -t -N is_reserved_word -k1,3,$ c-parse.gperf */
97 struct resword
{ char *name
; short token
; enum rid rid
; };
99 #define TOTAL_KEYWORDS 53
100 #define MIN_WORD_LENGTH 2
101 #define MAX_WORD_LENGTH 13
102 #define MIN_HASH_VALUE 7
103 #define MAX_HASH_VALUE 102
104 /* maximum key range = 96, duplicates = 0 */
112 register int unsigned len
;
114 static unsigned char asso_values
[] =
116 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
117 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
118 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
119 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
120 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
121 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
122 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
123 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
124 103, 103, 103, 103, 103, 103, 103, 103, 103, 103,
125 103, 103, 103, 103, 103, 1, 103, 2, 1, 24,
126 1, 5, 19, 39, 16, 13, 103, 1, 25, 1,
127 34, 34, 24, 103, 13, 12, 1, 45, 24, 7,
128 103, 103, 2, 103, 103, 103, 103, 103,
130 register int hval
= len
;
136 hval
+= asso_values
[str
[2]];
139 hval
+= asso_values
[str
[0]];
141 return hval
+ asso_values
[str
[len
- 1]];
148 is_reserved_word (str
, len
)
150 register unsigned int len
;
152 static struct resword wordlist
[] =
154 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
155 {"asm", ASM_KEYWORD
, NORID
},
157 {"__asm", ASM_KEYWORD
, NORID
},
159 {"__asm__", ASM_KEYWORD
, NORID
},
160 {"break", BREAK
, NORID
},
161 {"__typeof__", TYPEOF
, NORID
},
163 {"__alignof__", ALIGNOF
, NORID
},
165 {"__attribute__", ATTRIBUTE
, NORID
},
166 {"int", TYPESPEC
, RID_INT
},
167 {"__attribute", ATTRIBUTE
, NORID
},
168 {"__extension__", EXTENSION
, NORID
},
170 {"__signed", TYPESPEC
, RID_SIGNED
},
172 {"__signed__", TYPESPEC
, RID_SIGNED
},
173 {"__inline__", SCSPEC
, RID_INLINE
},
174 {"else", ELSE
, NORID
},
175 {"__inline", SCSPEC
, RID_INLINE
},
176 {"default", DEFAULT
, NORID
},
177 {"__typeof", TYPEOF
, NORID
},
178 {"while", WHILE
, NORID
},
179 {"__alignof", ALIGNOF
, NORID
},
180 {"struct", STRUCT
, NORID
},
181 {"__const", TYPE_QUAL
, RID_CONST
},
183 {"__const__", TYPE_QUAL
, RID_CONST
},
184 {"__label__", LABEL
, NORID
},
186 {"__volatile__", TYPE_QUAL
, RID_VOLATILE
},
187 {"sizeof", SIZEOF
, NORID
},
188 {"__volatile", TYPE_QUAL
, RID_VOLATILE
},
189 {"auto", SCSPEC
, RID_AUTO
},
190 {"void", TYPESPEC
, RID_VOID
},
191 {"char", TYPESPEC
, RID_CHAR
},
192 {"static", SCSPEC
, RID_STATIC
},
193 {"case", CASE
, NORID
},
194 {"extern", SCSPEC
, RID_EXTERN
},
195 {"switch", SWITCH
, NORID
},
197 {"inline", SCSPEC
, RID_INLINE
},
198 {"typeof", TYPEOF
, NORID
},
199 {"typedef", SCSPEC
, RID_TYPEDEF
},
200 {"short", TYPESPEC
, RID_SHORT
},
202 {"return", RETURN
, NORID
},
203 {"enum", ENUM
, NORID
},
205 {"double", TYPESPEC
, RID_DOUBLE
},
206 {"signed", TYPESPEC
, RID_SIGNED
},
207 {"float", TYPESPEC
, RID_FLOAT
},
209 {"volatile", TYPE_QUAL
, RID_VOLATILE
},
211 {"const", TYPE_QUAL
, RID_CONST
},
213 {"unsigned", TYPESPEC
, RID_UNSIGNED
},
214 {"",}, {"",}, {"",}, {"",},
215 {"continue", CONTINUE
, NORID
},
217 {"register", SCSPEC
, RID_REGISTER
},
218 {"",}, {"",}, {"",}, {"",},
219 {"goto", GOTO
, NORID
},
220 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
221 {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",}, {"",},
223 {"union", UNION
, NORID
},
224 {"",}, {"",}, {"",}, {"",},
225 {"long", TYPESPEC
, RID_LONG
},
228 if (len
<= MAX_WORD_LENGTH
&& len
>= MIN_WORD_LENGTH
)
230 register int key
= hash (str
, len
);
232 if (key
<= MAX_HASH_VALUE
&& key
>= 0)
234 register char *s
= wordlist
[key
].name
;
236 if (*s
== *str
&& !strcmp (str
+ 1, s
+ 1))
237 return &wordlist
[key
];
243 /* Return something to represent absolute declarators containing a *.
244 TARGET is the absolute declarator that the * contains.
245 TYPE_QUALS is a list of modifiers such as const or volatile
246 to apply to the pointer type, represented as identifiers.
248 We return an INDIRECT_REF whose "contents" are TARGET
249 and whose type is the modifier list. */
252 make_pointer_declarator (type_quals
, target
)
253 tree type_quals
, target
;
255 return build1 (INDIRECT_REF
, type_quals
, target
);
261 /* Make identifier nodes long enough for the language-specific slots. */
262 set_identifier_size (sizeof (struct lang_identifier
));
264 /* Start it at 0, because check_newline is called at the very beginning
265 and will increment it to 1. */
268 #ifdef MULTIBYTE_CHARS
269 /* Change to the native locale for multibyte conversions. */
270 setlocale (LC_CTYPE
, "");
274 token_buffer
= (char *) xmalloc (maxtoken
+ 2);
276 ridpointers
[(int) RID_INT
] = get_identifier ("int");
277 ridpointers
[(int) RID_CHAR
] = get_identifier ("char");
278 ridpointers
[(int) RID_VOID
] = get_identifier ("void");
279 ridpointers
[(int) RID_FLOAT
] = get_identifier ("float");
280 ridpointers
[(int) RID_DOUBLE
] = get_identifier ("double");
281 ridpointers
[(int) RID_SHORT
] = get_identifier ("short");
282 ridpointers
[(int) RID_LONG
] = get_identifier ("long");
283 ridpointers
[(int) RID_UNSIGNED
] = get_identifier ("unsigned");
284 ridpointers
[(int) RID_SIGNED
] = get_identifier ("signed");
285 ridpointers
[(int) RID_INLINE
] = get_identifier ("inline");
286 ridpointers
[(int) RID_CONST
] = get_identifier ("const");
287 ridpointers
[(int) RID_VOLATILE
] = get_identifier ("volatile");
288 ridpointers
[(int) RID_AUTO
] = get_identifier ("auto");
289 ridpointers
[(int) RID_STATIC
] = get_identifier ("static");
290 ridpointers
[(int) RID_EXTERN
] = get_identifier ("extern");
291 ridpointers
[(int) RID_TYPEDEF
] = get_identifier ("typedef");
292 ridpointers
[(int) RID_REGISTER
] = get_identifier ("register");
294 /* Some options inhibit certain reserved words.
295 Clear those words out of the hash table so they won't be recognized. */
296 #define UNSET_RESERVED_WORD(STRING) \
297 do { struct resword *s = is_reserved_word (STRING, sizeof (STRING) - 1); \
298 if (s) s->name = ""; } while (0)
300 if (flag_traditional
)
302 UNSET_RESERVED_WORD ("const");
303 UNSET_RESERVED_WORD ("volatile");
304 UNSET_RESERVED_WORD ("typeof");
305 UNSET_RESERVED_WORD ("signed");
306 UNSET_RESERVED_WORD ("inline");
310 UNSET_RESERVED_WORD ("asm");
311 UNSET_RESERVED_WORD ("typeof");
312 UNSET_RESERVED_WORD ("inline");
317 reinit_parse_for_function ()
321 /* Function used when yydebug is set, to print a token in more detail. */
324 yyprint (file
, yychar
, yylval
)
335 if (IDENTIFIER_POINTER (t
))
336 fprintf (file
, " `%s'", IDENTIFIER_POINTER (t
));
341 if (TREE_CODE (t
) == INTEGER_CST
)
343 #if HOST_BITS_PER_WIDE_INT == 64
344 #if HOST_BITS_PER_WIDE_INT != HOST_BITS_PER_INT
350 #if HOST_BITS_PER_WIDE_INT != HOST_BITS_PER_INT
356 TREE_INT_CST_HIGH (t
), TREE_INT_CST_LOW (t
));
362 /* If C is not whitespace, return C.
363 Otherwise skip whitespace and return first nonwhite char read. */
369 static int newline_warning
= 0;
375 /* We don't recognize comments here, because
376 cpp output can include / and * consecutively as operators.
377 Also, there's no need, since cpp removes all comments. */
380 c
= check_newline ();
392 /* ANSI C says the effects of a carriage return in a source file
394 if (pedantic
&& !newline_warning
)
396 warning ("carriage return in source file");
397 warning ("(we only warn about the first carriage return)");
408 error ("stray '\\' in program");
418 /* Skips all of the white space at the current location in the input file.
419 Must use and reset nextchar if it has the next character. */
422 position_after_white_space ()
427 c
= nextchar
, nextchar
= -1;
431 ungetc (skip_white_space (c
), finput
);
434 /* Make the token buffer longer, preserving the data in it.
435 P should point to just beyond the last valid character in the old buffer.
436 The value we return is a pointer to the new buffer
437 at a place corresponding to P. */
440 extend_token_buffer (p
)
443 int offset
= p
- token_buffer
;
445 maxtoken
= maxtoken
* 2 + 10;
446 token_buffer
= (char *) xrealloc (token_buffer
, maxtoken
+ 2);
448 return token_buffer
+ offset
;
451 /* At the beginning of a line, increment the line number
452 and process any #-directive on this line.
453 If the line is a #-directive, read the entire line and return a newline.
454 Otherwise, return the line's first non-whitespace character. */
464 /* Read first nonwhite char on the line. */
467 while (c
== ' ' || c
== '\t')
472 /* If not #, return it so caller will use it. */
476 /* Read first nonwhite char after the `#'. */
479 while (c
== ' ' || c
== '\t')
482 /* If a letter follows, then if the word here is `line', skip
483 it and ignore it; otherwise, ignore the line, with an error
484 if the word isn't `pragma', `ident', `define', or `undef'. */
486 if ((c
>= 'a' && c
<= 'z') || (c
>= 'A' && c
<= 'Z'))
490 if (getc (finput
) == 'r'
491 && getc (finput
) == 'a'
492 && getc (finput
) == 'g'
493 && getc (finput
) == 'm'
494 && getc (finput
) == 'a'
495 && ((c
= getc (finput
)) == ' ' || c
== '\t' || c
== '\n'))
497 #ifdef HANDLE_SYSV_PRAGMA
498 return handle_sysv_pragma (finput
, c
);
499 #endif /* HANDLE_SYSV_PRAGMA */
501 HANDLE_PRAGMA (finput
);
502 #endif /* HANDLE_PRAGMA */
509 if (getc (finput
) == 'e'
510 && getc (finput
) == 'f'
511 && getc (finput
) == 'i'
512 && getc (finput
) == 'n'
513 && getc (finput
) == 'e'
514 && ((c
= getc (finput
)) == ' ' || c
== '\t' || c
== '\n'))
516 #ifdef DWARF_DEBUGGING_INFO
517 if ((debug_info_level
== DINFO_LEVEL_VERBOSE
)
518 && (write_symbols
== DWARF_DEBUG
))
519 dwarfout_define (lineno
, get_directive_line (finput
));
520 #endif /* DWARF_DEBUGGING_INFO */
526 if (getc (finput
) == 'n'
527 && getc (finput
) == 'd'
528 && getc (finput
) == 'e'
529 && getc (finput
) == 'f'
530 && ((c
= getc (finput
)) == ' ' || c
== '\t' || c
== '\n'))
532 #ifdef DWARF_DEBUGGING_INFO
533 if ((debug_info_level
== DINFO_LEVEL_VERBOSE
)
534 && (write_symbols
== DWARF_DEBUG
))
535 dwarfout_undef (lineno
, get_directive_line (finput
));
536 #endif /* DWARF_DEBUGGING_INFO */
542 if (getc (finput
) == 'i'
543 && getc (finput
) == 'n'
544 && getc (finput
) == 'e'
545 && ((c
= getc (finput
)) == ' ' || c
== '\t'))
550 if (getc (finput
) == 'd'
551 && getc (finput
) == 'e'
552 && getc (finput
) == 'n'
553 && getc (finput
) == 't'
554 && ((c
= getc (finput
)) == ' ' || c
== '\t'))
556 /* #ident. The pedantic warning is now in cccp.c. */
558 /* Here we have just seen `#ident '.
559 A string constant should follow. */
561 while (c
== ' ' || c
== '\t')
564 /* If no argument, ignore the line. */
571 || TREE_CODE (yylval
.ttype
) != STRING_CST
)
573 error ("invalid #ident");
579 #ifdef ASM_OUTPUT_IDENT
580 ASM_OUTPUT_IDENT (asm_out_file
, TREE_STRING_POINTER (yylval
.ttype
));
584 /* Skip the rest of this line. */
589 error ("undefined or invalid # directive");
594 /* Here we have either `#line' or `# <nonletter>'.
595 In either case, it should be a line number; a digit should follow. */
597 while (c
== ' ' || c
== '\t')
600 /* If the # is the only nonwhite char on the line,
601 just ignore it. Check the new newline. */
605 /* Something follows the #; read a token. */
610 if (token
== CONSTANT
611 && TREE_CODE (yylval
.ttype
) == INTEGER_CST
)
613 int old_lineno
= lineno
;
615 /* subtract one, because it is the following line that
616 gets the specified number */
618 int l
= TREE_INT_CST_LOW (yylval
.ttype
) - 1;
620 /* Is this the last nonwhite stuff on the line? */
622 while (c
== ' ' || c
== '\t')
626 /* No more: store the line number and check following line. */
632 /* More follows: it must be a string constant (filename). */
634 /* Read the string constant, but don't treat \ as special. */
635 ignore_escape_flag
= 1;
637 ignore_escape_flag
= 0;
639 if (token
!= STRING
|| TREE_CODE (yylval
.ttype
) != STRING_CST
)
641 error ("invalid #line");
646 = (char *) permalloc (TREE_STRING_LENGTH (yylval
.ttype
) + 1);
647 strcpy (input_filename
, TREE_STRING_POINTER (yylval
.ttype
));
650 /* Each change of file name
651 reinitializes whether we are now in a system header. */
652 in_system_header
= 0;
654 if (main_input_filename
== 0)
655 main_input_filename
= input_filename
;
657 /* Is this the last nonwhite stuff on the line? */
659 while (c
== ' ' || c
== '\t')
668 /* `1' after file name means entering new file.
669 `2' after file name means just left a file. */
671 if (token
== CONSTANT
672 && TREE_CODE (yylval
.ttype
) == INTEGER_CST
)
674 if (TREE_INT_CST_LOW (yylval
.ttype
) == 1)
676 /* Pushing to a new file. */
678 = (struct file_stack
*) xmalloc (sizeof (struct file_stack
));
679 input_file_stack
->line
= old_lineno
;
680 p
->next
= input_file_stack
;
681 p
->name
= input_filename
;
682 input_file_stack
= p
;
683 input_file_stack_tick
++;
684 #ifdef DWARF_DEBUGGING_INFO
685 if (debug_info_level
== DINFO_LEVEL_VERBOSE
686 && write_symbols
== DWARF_DEBUG
)
687 dwarfout_start_new_source_file (input_filename
);
688 #endif /* DWARF_DEBUGGING_INFO */
692 else if (TREE_INT_CST_LOW (yylval
.ttype
) == 2)
694 /* Popping out of a file. */
695 if (input_file_stack
->next
)
697 struct file_stack
*p
= input_file_stack
;
698 input_file_stack
= p
->next
;
700 input_file_stack_tick
++;
701 #ifdef DWARF_DEBUGGING_INFO
702 if (debug_info_level
== DINFO_LEVEL_VERBOSE
703 && write_symbols
== DWARF_DEBUG
)
704 dwarfout_resume_previous_source_file (input_file_stack
->line
);
705 #endif /* DWARF_DEBUGGING_INFO */
708 error ("#-lines for entering and leaving files don't match");
714 /* If we have handled a `1' or a `2',
715 see if there is another number to read. */
718 /* Is this the last nonwhite stuff on the line? */
720 while (c
== ' ' || c
== '\t')
730 /* `3' after file name means this is a system header file. */
732 if (token
== CONSTANT
733 && TREE_CODE (yylval
.ttype
) == INTEGER_CST
734 && TREE_INT_CST_LOW (yylval
.ttype
) == 3)
735 in_system_header
= 1;
738 error ("invalid #-line");
740 /* skip the rest of this line. */
744 while ((c
= getc (finput
)) != EOF
&& c
!= '\n');
748 #ifdef HANDLE_SYSV_PRAGMA
750 /* Handle a #pragma directive. INPUT is the current input stream,
751 and C is a character to reread. Processes the entire input line
752 and returns a character for the caller to reread: either \n or EOF. */
754 /* This function has to be in this file, in order to get at
758 handle_sysv_pragma (input
, c
)
764 while (c
== ' ' || c
== '\t')
766 if (c
== '\n' || c
== EOF
)
768 handle_pragma_token (0, 0);
778 handle_pragma_token (token_buffer
, yylval
.ttype
);
781 handle_pragma_token (token_buffer
, 0);
784 c
= nextchar
, nextchar
= -1;
790 #endif /* HANDLE_SYSV_PRAGMA */
792 #define isalnum(char) ((char >= 'a' && char <= 'z') || (char >= 'A' && char <= 'Z') || (char >= '0' && char <= '9'))
793 #define isdigit(char) (char >= '0' && char <= '9')
794 #define ENDFILE -1 /* token that represents end-of-file */
796 /* Read an escape sequence, returning its equivalent as a character,
797 or -1 if it is backslash-newline. */
802 register int c
= getc (finput
);
804 register unsigned count
;
810 if (warn_traditional
)
811 warning ("the meaning of `\\x' varies with -traditional");
813 if (flag_traditional
)
821 if (!(c
>= 'a' && c
<= 'f')
822 && !(c
>= 'A' && c
<= 'F')
823 && !(c
>= '0' && c
<= '9'))
829 if (c
>= 'a' && c
<= 'f')
830 code
+= c
- 'a' + 10;
831 if (c
>= 'A' && c
<= 'F')
832 code
+= c
- 'A' + 10;
833 if (c
>= '0' && c
<= '9')
840 error ("\\x used with no following hex digits");
841 else if ((count
- 1) * 4 >= TYPE_PRECISION (integer_type_node
)
843 && ((1 << (TYPE_PRECISION (integer_type_node
) - (count
- 1) * 4))
845 pedwarn ("hex escape out of range");
848 case '0': case '1': case '2': case '3': case '4':
849 case '5': case '6': case '7':
852 while ((c
<= '7') && (c
>= '0') && (count
++ < 3))
854 code
= (code
* 8) + (c
- '0');
860 case '\\': case '\'': case '"':
868 return TARGET_NEWLINE
;
883 if (warn_traditional
)
884 warning ("the meaning of `\\a' varies with -traditional");
886 if (flag_traditional
)
891 #if 0 /* Vertical tab is present in common usage compilers. */
892 if (flag_traditional
)
898 pedwarn ("non-ANSI-standard escape sequence, `\\E'");
904 /* `\(', etc, are used at beginning of line to avoid confusing Emacs. */
909 pedwarn ("non-ANSI escape sequence `\\%c'", c
);
912 if (c
>= 040 && c
< 0177)
913 pedwarn ("unknown escape sequence `\\%c'", c
);
915 pedwarn ("unknown escape sequence: `\\' followed by char code 0x%x", c
);
925 strcpy (buf
, string
);
927 /* We can't print string and character constants well
928 because the token_buffer contains the result of processing escapes. */
930 strcat (buf
, " at end of input");
931 else if (token_buffer
[0] == 0)
932 strcat (buf
, " at null character");
933 else if (token_buffer
[0] == '"')
934 strcat (buf
, " before string constant");
935 else if (token_buffer
[0] == '\'')
936 strcat (buf
, " before character constant");
937 else if (token_buffer
[0] < 040 || (unsigned char) token_buffer
[0] >= 0177)
938 sprintf (buf
+ strlen (buf
), " before character 0%o",
939 (unsigned char) token_buffer
[0]);
941 strcat (buf
, " before `%s'");
943 error (buf
, token_buffer
);
956 struct try_type type_sequence
[] =
958 { &integer_type_node
, 0, 0, 0},
959 { &unsigned_type_node
, 1, 0, 0},
960 { &long_integer_type_node
, 0, 1, 0},
961 { &long_unsigned_type_node
, 1, 1, 0},
962 { &long_long_integer_type_node
, 0, 1, 1},
963 { &long_long_unsigned_type_node
, 1, 1, 1}
976 c
= nextchar
, nextchar
= -1;
980 /* Effectively do c = skip_white_space (c)
981 but do it faster in the usual cases. */
994 /* Call skip_white_space so we can warn if appropriate. */
999 c
= skip_white_space (c
);
1001 goto found_nonwhite
;
1005 token_buffer
[0] = c
;
1006 token_buffer
[1] = 0;
1008 /* yylloc.first_line = lineno; */
1014 token_buffer
[0] = 0;
1019 if (dollars_in_ident
)
1024 /* Capital L may start a wide-string or wide-character constant. */
1026 register int c
= getc (finput
);
1035 goto string_constant
;
1042 if (!doing_objc_thang
)
1050 while (isalnum (c
) || c
== '_')
1052 if (p
>= token_buffer
+ maxtoken
)
1053 p
= extend_token_buffer (p
);
1061 value
= recognize_objc_keyword (token_buffer
+ 1);
1064 error ("invalid Objective C keyword `%s'", token_buffer
);
1065 /* Cause a syntax error--1 is not a valid token type. */
1069 case 'A': case 'B': case 'C': case 'D': case 'E':
1070 case 'F': case 'G': case 'H': case 'I': case 'J':
1071 case 'K': case 'M': case 'N': case 'O':
1072 case 'P': case 'Q': case 'R': case 'S': case 'T':
1073 case 'U': case 'V': case 'W': case 'X': case 'Y':
1075 case 'a': case 'b': case 'c': case 'd': case 'e':
1076 case 'f': case 'g': case 'h': case 'i': case 'j':
1077 case 'k': case 'l': case 'm': case 'n': case 'o':
1078 case 'p': case 'q': case 'r': case 's': case 't':
1079 case 'u': case 'v': case 'w': case 'x': case 'y':
1084 while (isalnum (c
) || c
== '_' || c
== '$' || c
== '@')
1086 if (p
>= token_buffer
+ maxtoken
)
1087 p
= extend_token_buffer (p
);
1088 if (c
== '$' && ! dollars_in_ident
)
1101 /* Try to recognize a keyword. Uses minimum-perfect hash function */
1104 register struct resword
*ptr
;
1106 if (ptr
= is_reserved_word (token_buffer
, p
- token_buffer
))
1109 yylval
.ttype
= ridpointers
[(int) ptr
->rid
];
1110 value
= (int) ptr
->token
;
1112 /* Even if we decided to recognize asm, still perhaps warn. */
1114 && (value
== ASM_KEYWORD
|| value
== TYPEOF
1115 || ptr
->rid
== RID_INLINE
)
1116 && token_buffer
[0] != '_')
1117 pedwarn ("ANSI does not permit the keyword `%s'",
1122 /* If we did not find a keyword, look for an identifier
1125 if (value
== IDENTIFIER
)
1127 yylval
.ttype
= get_identifier (token_buffer
);
1128 lastiddecl
= lookup_name (yylval
.ttype
);
1130 if (lastiddecl
!= 0 && TREE_CODE (lastiddecl
) == TYPE_DECL
)
1132 /* A user-invisible read-only initialized variable
1133 should be replaced by its value.
1134 We handle only strings since that's the only case used in C. */
1135 else if (lastiddecl
!= 0 && TREE_CODE (lastiddecl
) == VAR_DECL
1136 && DECL_IGNORED_P (lastiddecl
)
1137 && TREE_READONLY (lastiddecl
)
1138 && DECL_INITIAL (lastiddecl
) != 0
1139 && TREE_CODE (DECL_INITIAL (lastiddecl
)) == STRING_CST
)
1141 yylval
.ttype
= DECL_INITIAL (lastiddecl
);
1144 else if (doing_objc_thang
)
1146 tree objc_interface_decl
= lookup_interface (yylval
.ttype
);
1148 if (objc_interface_decl
)
1151 yylval
.ttype
= objc_interface_decl
;
1158 case '0': case '1': case '2': case '3': case '4':
1159 case '5': case '6': case '7': case '8': case '9':
1164 int largest_digit
= 0;
1166 /* for multi-precision arithmetic,
1167 we actually store only HOST_BITS_PER_CHAR bits in each part.
1168 The number of parts is chosen so as to be sufficient to hold
1169 the enough bits to fit into the two HOST_WIDE_INTs that contain
1170 the integer value (this is always at least as many bits as are
1171 in a target `long long' value, but may be wider). */
1172 #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2 + 2)
1173 int parts
[TOTAL_PARTS
];
1176 enum anon1
{ NOT_FLOAT
, AFTER_POINT
, TOO_MANY_POINTS
} floatflag
1179 for (count
= 0; count
< TOTAL_PARTS
; count
++)
1187 *p
++ = (c
= getc (finput
));
1188 if ((c
== 'x') || (c
== 'X'))
1191 *p
++ = (c
= getc (finput
));
1193 /* Leading 0 forces octal unless the 0 is the only digit. */
1194 else if (c
>= '0' && c
<= '9')
1203 /* Read all the digits-and-decimal-points. */
1206 || (isalnum (c
) && (c
!= 'l') && (c
!= 'L')
1207 && (c
!= 'u') && (c
!= 'U')
1208 && (floatflag
== NOT_FLOAT
|| ((c
!= 'f') && (c
!= 'F')))))
1213 error ("floating constant may not be in radix 16");
1214 if (floatflag
== AFTER_POINT
)
1216 error ("malformed floating constant");
1217 floatflag
= TOO_MANY_POINTS
;
1220 floatflag
= AFTER_POINT
;
1223 *p
++ = c
= getc (finput
);
1224 /* Accept '.' as the start of a floating-point number
1225 only when it is followed by a digit.
1226 Otherwise, unread the following non-digit
1227 and use the '.' as a structural token. */
1228 if (p
== token_buffer
+ 2 && !isdigit (c
))
1239 error ("parse error at `..'");
1242 token_buffer
[1] = 0;
1249 /* It is not a decimal point.
1250 It should be a digit (perhaps a hex digit). */
1256 else if (base
<= 10)
1258 if ((c
&~040) == 'E')
1261 floatflag
= AFTER_POINT
;
1262 break; /* start of exponent */
1264 error ("nondigits in number and not hexadecimal");
1275 if (c
>= largest_digit
)
1279 for (count
= 0; count
< TOTAL_PARTS
; count
++)
1281 parts
[count
] *= base
;
1285 += (parts
[count
-1] >> HOST_BITS_PER_CHAR
);
1287 &= (1 << HOST_BITS_PER_CHAR
) - 1;
1293 /* If the extra highest-order part ever gets anything in it,
1294 the number is certainly too big. */
1295 if (parts
[TOTAL_PARTS
- 1] != 0)
1298 if (p
>= token_buffer
+ maxtoken
- 3)
1299 p
= extend_token_buffer (p
);
1300 *p
++ = (c
= getc (finput
));
1305 error ("numeric constant with no digits");
1307 if (largest_digit
>= base
)
1308 error ("numeric constant contains digits beyond the radix");
1310 /* Remove terminating char from the token buffer and delimit the string */
1313 if (floatflag
!= NOT_FLOAT
)
1315 tree type
= double_type_node
;
1316 int garbage_chars
= 0, exceeds_double
= 0;
1317 REAL_VALUE_TYPE value
;
1320 /* Read explicit exponent if any, and put it in tokenbuf. */
1322 if ((c
== 'e') || (c
== 'E'))
1324 if (p
>= token_buffer
+ maxtoken
- 3)
1325 p
= extend_token_buffer (p
);
1328 if ((c
== '+') || (c
== '-'))
1334 error ("floating constant exponent has no digits");
1337 if (p
>= token_buffer
+ maxtoken
- 3)
1338 p
= extend_token_buffer (p
);
1347 /* Convert string to a double, checking for overflow. */
1348 if (setjmp (handler
))
1350 error ("floating constant out of range");
1355 set_float_handler (handler
);
1356 value
= REAL_VALUE_ATOF (token_buffer
);
1357 set_float_handler (NULL_PTR
);
1360 if (errno
== ERANGE
&& !flag_traditional
&& pedantic
)
1362 char *p1
= token_buffer
;
1363 /* Check for "0.0" and variants;
1364 SunOS 4 spuriously returns ERANGE for them. */
1365 while (*p1
== '0') p1
++;
1369 while (*p1
== '0') p1
++;
1371 if (*p1
== 'e' || *p1
== 'E')
1373 /* with significand==0, ignore the exponent */
1375 while (*p1
!= 0) p1
++;
1377 /* ERANGE is also reported for underflow,
1378 so test the value to distinguish overflow from that. */
1379 if (*p1
!= 0 && (value
> 1.0 || value
< -1.0))
1381 pedwarn ("floating point number exceeds range of `double'");
1387 /* Read the suffixes to choose a data type. */
1391 type
= float_type_node
;
1392 value
= REAL_VALUE_TRUNCATE (TYPE_MODE (type
), value
);
1393 if (REAL_VALUE_ISINF (value
) && ! exceeds_double
&& pedantic
)
1394 pedwarn ("floating point number exceeds range of `float'");
1399 type
= long_double_type_node
;
1403 /* Note: garbage_chars is -1 if first char is *not* garbage. */
1406 if (p
>= token_buffer
+ maxtoken
- 3)
1407 p
= extend_token_buffer (p
);
1412 if (garbage_chars
> 0)
1413 error ("garbage at end of number");
1415 /* Create a node with determined type and value. */
1416 yylval
.ttype
= build_real (type
, value
);
1423 tree traditional_type
, ansi_type
, type
;
1424 HOST_WIDE_INT high
, low
;
1425 int spec_unsigned
= 0;
1427 int spec_long_long
= 0;
1432 if (c
== 'u' || c
== 'U')
1435 error ("two `u's in integer constant");
1438 else if (c
== 'l' || c
== 'L')
1443 error ("three `l's in integer constant");
1445 pedwarn ("ANSI C forbids long long integer constants");
1454 error ("garbage at end of number");
1457 if (p
>= token_buffer
+ maxtoken
- 3)
1458 p
= extend_token_buffer (p
);
1465 if (p
>= token_buffer
+ maxtoken
- 3)
1466 p
= extend_token_buffer (p
);
1473 /* If the constant is not long long and it won't fit in an
1474 unsigned long, or if the constant is long long and won't fit
1475 in an unsigned long long, then warn that the constant is out
1478 /* ??? This assumes that long long and long integer types are
1479 a multiple of 8 bits. This better than the original code
1480 though which assumed that long was exactly 32 bits and long
1481 long was exactly 64 bits. */
1484 bytes
= TYPE_PRECISION (long_long_integer_type_node
) / 8;
1486 bytes
= TYPE_PRECISION (long_integer_type_node
) / 8;
1489 for (i
= bytes
; i
< TOTAL_PARTS
; i
++)
1493 pedwarn ("integer constant out of range");
1495 /* This is simplified by the fact that our constant
1496 is always positive. */
1500 for (i
= 0; i
< HOST_BITS_PER_WIDE_INT
/ HOST_BITS_PER_CHAR
; i
++)
1502 high
|= ((HOST_WIDE_INT
) parts
[i
+ (HOST_BITS_PER_WIDE_INT
1503 / HOST_BITS_PER_CHAR
)]
1504 << (i
* HOST_BITS_PER_CHAR
));
1505 low
|= (HOST_WIDE_INT
) parts
[i
] << (i
* HOST_BITS_PER_CHAR
);
1508 yylval
.ttype
= build_int_2 (low
, high
);
1510 /* If warn_traditional, calculate both the ANSI type and the
1511 traditional type, then see if they disagree.
1512 Otherwise, calculate only the type for the dialect in use. */
1513 if (warn_traditional
|| flag_traditional
)
1515 /* Calculate the traditional type. */
1516 /* Traditionally, any constant is signed;
1517 but if unsigned is specified explicitly, obey that.
1518 Use the smallest size with the right number of bits,
1519 except for one special case with decimal constants. */
1520 if (! spec_long
&& base
!= 10
1521 && int_fits_type_p (yylval
.ttype
, unsigned_type_node
))
1522 traditional_type
= (spec_unsigned
? unsigned_type_node
1523 : integer_type_node
);
1524 /* A decimal constant must be long
1525 if it does not fit in type int.
1526 I think this is independent of whether
1527 the constant is signed. */
1528 else if (! spec_long
&& base
== 10
1529 && int_fits_type_p (yylval
.ttype
, integer_type_node
))
1530 traditional_type
= (spec_unsigned
? unsigned_type_node
1531 : integer_type_node
);
1532 else if (! spec_long_long
)
1533 traditional_type
= (spec_unsigned
? long_unsigned_type_node
1534 : long_integer_type_node
);
1536 traditional_type
= (spec_unsigned
1537 ? long_long_unsigned_type_node
1538 : long_long_integer_type_node
);
1540 if (warn_traditional
|| ! flag_traditional
)
1542 /* Calculate the ANSI type. */
1543 if (! spec_long
&& ! spec_unsigned
1544 && int_fits_type_p (yylval
.ttype
, integer_type_node
))
1545 ansi_type
= integer_type_node
;
1546 else if (! spec_long
&& (base
!= 10 || spec_unsigned
)
1547 && int_fits_type_p (yylval
.ttype
, unsigned_type_node
))
1548 ansi_type
= unsigned_type_node
;
1549 else if (! spec_unsigned
&& !spec_long_long
1550 && int_fits_type_p (yylval
.ttype
, long_integer_type_node
))
1551 ansi_type
= long_integer_type_node
;
1552 else if (! spec_long_long
)
1553 ansi_type
= long_unsigned_type_node
;
1554 else if (! spec_unsigned
1555 && int_fits_type_p (yylval
.ttype
,
1556 long_long_integer_type_node
))
1557 ansi_type
= long_long_integer_type_node
;
1559 ansi_type
= long_long_unsigned_type_node
;
1562 type
= flag_traditional
? traditional_type
: ansi_type
;
1564 if (warn_traditional
&& traditional_type
!= ansi_type
)
1566 if (TYPE_PRECISION (traditional_type
)
1567 != TYPE_PRECISION (ansi_type
))
1568 warning ("width of integer constant changes with -traditional");
1569 else if (TREE_UNSIGNED (traditional_type
)
1570 != TREE_UNSIGNED (ansi_type
))
1571 warning ("integer constant is unsigned in ANSI C, signed with -traditional");
1573 warning ("width of integer constant may change on other systems with -traditional");
1576 if (!flag_traditional
&& !int_fits_type_p (yylval
.ttype
, type
)
1578 pedwarn ("integer constant out of range");
1580 if (flag_traditional
&& !int_fits_type_p (yylval
.ttype
, type
))
1581 /* The traditional constant 0x80000000 is signed
1582 but doesn't fit in the range of int.
1583 This will change it to -0x80000000, which does fit. */
1585 TREE_TYPE (yylval
.ttype
) = unsigned_type (type
);
1586 yylval
.ttype
= convert (type
, yylval
.ttype
);
1589 TREE_TYPE (yylval
.ttype
) = type
;
1594 value
= CONSTANT
; break;
1600 register int result
= 0;
1601 register int num_chars
= 0;
1602 unsigned width
= TYPE_PRECISION (char_type_node
);
1607 width
= WCHAR_TYPE_SIZE
;
1608 #ifdef MULTIBYTE_CHARS
1609 max_chars
= MB_CUR_MAX
;
1615 max_chars
= TYPE_PRECISION (integer_type_node
) / width
;
1623 if (c
== '\'' || c
== EOF
)
1631 if (width
< HOST_BITS_PER_INT
1632 && (unsigned) c
>= (1 << width
))
1633 pedwarn ("escape sequence out of range for character");
1638 pedwarn ("ANSI C forbids newline in character constant");
1643 if (num_chars
> maxtoken
- 4)
1644 extend_token_buffer (token_buffer
);
1646 token_buffer
[num_chars
] = c
;
1648 /* Merge character into result; ignore excess chars. */
1649 if (num_chars
< max_chars
+ 1)
1651 if (width
< HOST_BITS_PER_INT
)
1652 result
= (result
<< width
) | (c
& ((1 << width
) - 1));
1658 token_buffer
[num_chars
+ 1] = '\'';
1659 token_buffer
[num_chars
+ 2] = 0;
1662 error ("malformatted character constant");
1663 else if (num_chars
== 0)
1664 error ("empty character constant");
1665 else if (num_chars
> max_chars
)
1667 num_chars
= max_chars
;
1668 error ("character constant too long");
1670 else if (num_chars
!= 1 && ! flag_traditional
)
1671 warning ("multi-character character constant");
1673 /* If char type is signed, sign-extend the constant. */
1676 int num_bits
= num_chars
* width
;
1677 if (TREE_UNSIGNED (char_type_node
)
1678 || ((result
>> (num_bits
- 1)) & 1) == 0)
1680 = build_int_2 (result
& ((unsigned HOST_WIDE_INT
) ~0
1681 >> (HOST_BITS_PER_WIDE_INT
- num_bits
)),
1685 = build_int_2 (result
| ~((unsigned HOST_WIDE_INT
) ~0
1686 >> (HOST_BITS_PER_WIDE_INT
- num_bits
)),
1691 #ifdef MULTIBYTE_CHARS
1692 /* Set the initial shift state and convert the next sequence. */
1694 /* In all locales L'\0' is zero and mbtowc will return zero,
1697 || (num_chars
== 1 && token_buffer
[1] != '\0'))
1700 (void) mbtowc (NULL_PTR
, NULL_PTR
, 0);
1701 if (mbtowc (& wc
, token_buffer
+ 1, num_chars
) == num_chars
)
1704 warning ("Ignoring invalid multibyte character");
1707 yylval
.ttype
= build_int_2 (result
, 0);
1710 TREE_TYPE (yylval
.ttype
) = integer_type_node
;
1719 p
= token_buffer
+ 1;
1721 while (c
!= '"' && c
>= 0)
1723 /* ignore_escape_flag is set for reading the filename in #line. */
1724 if (!ignore_escape_flag
&& c
== '\\')
1730 && TYPE_PRECISION (char_type_node
) < HOST_BITS_PER_INT
1731 && c
>= (1 << TYPE_PRECISION (char_type_node
)))
1732 pedwarn ("escape sequence out of range for character");
1737 pedwarn ("ANSI C forbids newline in string constant");
1741 if (p
== token_buffer
+ maxtoken
)
1742 p
= extend_token_buffer (p
);
1750 /* We have read the entire constant.
1751 Construct a STRING_CST for the result. */
1755 /* If this is a L"..." wide-string, convert the multibyte string
1756 to a wide character string. */
1757 char *widep
= (char *) alloca ((p
- token_buffer
) * WCHAR_BYTES
);
1760 #ifdef MULTIBYTE_CHARS
1761 len
= mbstowcs ((wchar_t *) widep
, token_buffer
+ 1, p
- token_buffer
);
1762 if ((unsigned) len
>= (p
- token_buffer
))
1764 warning ("Ignoring invalid multibyte string");
1767 bzero (widep
+ (len
* WCHAR_BYTES
), WCHAR_BYTES
);
1770 union { long l
; char c
[sizeof (long)]; } u
;
1774 /* Determine whether host is little or big endian. */
1776 big_endian
= u
.c
[sizeof (long) - 1];
1777 wp
= widep
+ (big_endian
? WCHAR_BYTES
- 1 : 0);
1779 bzero (widep
, (p
- token_buffer
) * WCHAR_BYTES
);
1780 for (cp
= token_buffer
+ 1; cp
< p
; cp
++)
1781 *wp
= *cp
, wp
+= WCHAR_BYTES
;
1782 len
= p
- token_buffer
- 1;
1785 yylval
.ttype
= build_string ((len
+ 1) * WCHAR_BYTES
, widep
);
1786 TREE_TYPE (yylval
.ttype
) = wchar_array_type_node
;
1790 yylval
.ttype
= build_string (p
- token_buffer
, token_buffer
+ 1);
1791 TREE_TYPE (yylval
.ttype
) = char_array_type_node
;
1797 value
= STRING
; break;
1820 yylval
.code
= PLUS_EXPR
; break;
1822 yylval
.code
= MINUS_EXPR
; break;
1824 yylval
.code
= BIT_AND_EXPR
; break;
1826 yylval
.code
= BIT_IOR_EXPR
; break;
1828 yylval
.code
= MULT_EXPR
; break;
1830 yylval
.code
= TRUNC_DIV_EXPR
; break;
1832 yylval
.code
= TRUNC_MOD_EXPR
; break;
1834 yylval
.code
= BIT_XOR_EXPR
; break;
1836 yylval
.code
= LSHIFT_EXPR
; break;
1838 yylval
.code
= RSHIFT_EXPR
; break;
1840 yylval
.code
= LT_EXPR
; break;
1842 yylval
.code
= GT_EXPR
; break;
1845 token_buffer
[1] = c1
= getc (finput
);
1846 token_buffer
[2] = 0;
1853 value
= ARITHCOMPARE
; yylval
.code
= LE_EXPR
; goto done
;
1855 value
= ARITHCOMPARE
; yylval
.code
= GE_EXPR
; goto done
;
1857 value
= EQCOMPARE
; yylval
.code
= NE_EXPR
; goto done
;
1859 value
= EQCOMPARE
; yylval
.code
= EQ_EXPR
; goto done
;
1861 value
= ASSIGN
; goto done
;
1867 value
= PLUSPLUS
; goto done
;
1869 value
= MINUSMINUS
; goto done
;
1871 value
= ANDAND
; goto done
;
1873 value
= OROR
; goto done
;
1881 else if ((c
== '-') && (c1
== '>'))
1882 { value
= POINTSAT
; goto done
; }
1883 ungetc (c1
, finput
);
1884 token_buffer
[1] = 0;
1886 if ((c
== '<') || (c
== '>'))
1887 value
= ARITHCOMPARE
;
1893 /* Don't make yyparse think this is eof. */
1902 /* yylloc.last_line = lineno; */
1907 /* Sets the value of the 'yydebug' variable to VALUE.
1908 This is a function so we don't have to have YYDEBUG defined
1909 in order to build the compiler. */
1918 warning ("YYDEBUG not defined.");