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')
663 /* Update the name in the top element of input_file_stack. */
664 if (input_file_stack
)
665 input_file_stack
->name
= input_filename
;
674 /* `1' after file name means entering new file.
675 `2' after file name means just left a file. */
677 if (token
== CONSTANT
678 && TREE_CODE (yylval
.ttype
) == INTEGER_CST
)
680 if (TREE_INT_CST_LOW (yylval
.ttype
) == 1)
682 /* Pushing to a new file. */
684 = (struct file_stack
*) xmalloc (sizeof (struct file_stack
));
685 input_file_stack
->line
= old_lineno
;
686 p
->next
= input_file_stack
;
687 p
->name
= input_filename
;
688 input_file_stack
= p
;
689 input_file_stack_tick
++;
690 #ifdef DWARF_DEBUGGING_INFO
691 if (debug_info_level
== DINFO_LEVEL_VERBOSE
692 && write_symbols
== DWARF_DEBUG
)
693 dwarfout_start_new_source_file (input_filename
);
694 #endif /* DWARF_DEBUGGING_INFO */
698 else if (TREE_INT_CST_LOW (yylval
.ttype
) == 2)
700 /* Popping out of a file. */
701 if (input_file_stack
->next
)
703 struct file_stack
*p
= input_file_stack
;
704 input_file_stack
= p
->next
;
706 input_file_stack_tick
++;
707 #ifdef DWARF_DEBUGGING_INFO
708 if (debug_info_level
== DINFO_LEVEL_VERBOSE
709 && write_symbols
== DWARF_DEBUG
)
710 dwarfout_resume_previous_source_file (input_file_stack
->line
);
711 #endif /* DWARF_DEBUGGING_INFO */
714 error ("#-lines for entering and leaving files don't match");
720 /* Now that we've pushed or popped the input stack,
721 update the name in the top element. */
722 if (input_file_stack
)
723 input_file_stack
->name
= input_filename
;
725 /* If we have handled a `1' or a `2',
726 see if there is another number to read. */
729 /* Is this the last nonwhite stuff on the line? */
731 while (c
== ' ' || c
== '\t')
741 /* `3' after file name means this is a system header file. */
743 if (token
== CONSTANT
744 && TREE_CODE (yylval
.ttype
) == INTEGER_CST
745 && TREE_INT_CST_LOW (yylval
.ttype
) == 3)
746 in_system_header
= 1;
749 error ("invalid #-line");
751 /* skip the rest of this line. */
755 while ((c
= getc (finput
)) != EOF
&& c
!= '\n');
759 #ifdef HANDLE_SYSV_PRAGMA
761 /* Handle a #pragma directive. INPUT is the current input stream,
762 and C is a character to reread. Processes the entire input line
763 and returns a character for the caller to reread: either \n or EOF. */
765 /* This function has to be in this file, in order to get at
769 handle_sysv_pragma (input
, c
)
775 while (c
== ' ' || c
== '\t')
777 if (c
== '\n' || c
== EOF
)
779 handle_pragma_token (0, 0);
789 handle_pragma_token (token_buffer
, yylval
.ttype
);
792 handle_pragma_token (token_buffer
, 0);
795 c
= nextchar
, nextchar
= -1;
801 #endif /* HANDLE_SYSV_PRAGMA */
803 #define isalnum(char) ((char >= 'a' && char <= 'z') || (char >= 'A' && char <= 'Z') || (char >= '0' && char <= '9'))
804 #define isdigit(char) (char >= '0' && char <= '9')
805 #define ENDFILE -1 /* token that represents end-of-file */
807 /* Read an escape sequence, returning its equivalent as a character,
808 or store 1 in *ignore_ptr if it is backslash-newline. */
811 readescape (ignore_ptr
)
814 register int c
= getc (finput
);
816 register unsigned count
;
823 if (warn_traditional
)
824 warning ("the meaning of `\\x' varies with -traditional");
826 if (flag_traditional
)
835 if (!(c
>= 'a' && c
<= 'f')
836 && !(c
>= 'A' && c
<= 'F')
837 && !(c
>= '0' && c
<= '9'))
843 if (c
>= 'a' && c
<= 'f')
844 code
+= c
- 'a' + 10;
845 if (c
>= 'A' && c
<= 'F')
846 code
+= c
- 'A' + 10;
847 if (c
>= '0' && c
<= '9')
849 if (code
!= 0 || count
!= 0)
858 error ("\\x used with no following hex digits");
860 /* Digits are all 0's. Ok. */
862 else if ((count
- 1) * 4 >= TYPE_PRECISION (integer_type_node
)
864 && ((1 << (TYPE_PRECISION (integer_type_node
) - (count
- 1) * 4))
866 pedwarn ("hex escape out of range");
869 case '0': case '1': case '2': case '3': case '4':
870 case '5': case '6': case '7':
873 while ((c
<= '7') && (c
>= '0') && (count
++ < 3))
875 code
= (code
* 8) + (c
- '0');
881 case '\\': case '\'': case '"':
890 return TARGET_NEWLINE
;
905 if (warn_traditional
)
906 warning ("the meaning of `\\a' varies with -traditional");
908 if (flag_traditional
)
913 #if 0 /* Vertical tab is present in common usage compilers. */
914 if (flag_traditional
)
922 pedwarn ("non-ANSI-standard escape sequence, `\\%c'", c
);
928 /* `\(', etc, are used at beginning of line to avoid confusing Emacs. */
933 pedwarn ("non-ANSI escape sequence `\\%c'", c
);
936 if (c
>= 040 && c
< 0177)
937 pedwarn ("unknown escape sequence `\\%c'", c
);
939 pedwarn ("unknown escape sequence: `\\' followed by char code 0x%x", c
);
949 strcpy (buf
, string
);
951 /* We can't print string and character constants well
952 because the token_buffer contains the result of processing escapes. */
954 strcat (buf
, " at end of input");
955 else if (token_buffer
[0] == 0)
956 strcat (buf
, " at null character");
957 else if (token_buffer
[0] == '"')
958 strcat (buf
, " before string constant");
959 else if (token_buffer
[0] == '\'')
960 strcat (buf
, " before character constant");
961 else if (token_buffer
[0] < 040 || (unsigned char) token_buffer
[0] >= 0177)
962 sprintf (buf
+ strlen (buf
), " before character 0%o",
963 (unsigned char) token_buffer
[0]);
965 strcat (buf
, " before `%s'");
967 error (buf
, token_buffer
);
980 struct try_type type_sequence
[] =
982 { &integer_type_node
, 0, 0, 0},
983 { &unsigned_type_node
, 1, 0, 0},
984 { &long_integer_type_node
, 0, 1, 0},
985 { &long_unsigned_type_node
, 1, 1, 0},
986 { &long_long_integer_type_node
, 0, 1, 1},
987 { &long_long_unsigned_type_node
, 1, 1, 1}
1000 c
= nextchar
, nextchar
= -1;
1004 /* Effectively do c = skip_white_space (c)
1005 but do it faster in the usual cases. */
1018 /* Call skip_white_space so we can warn if appropriate. */
1023 c
= skip_white_space (c
);
1025 goto found_nonwhite
;
1029 token_buffer
[0] = c
;
1030 token_buffer
[1] = 0;
1032 /* yylloc.first_line = lineno; */
1038 token_buffer
[0] = 0;
1043 if (dollars_in_ident
)
1048 /* Capital L may start a wide-string or wide-character constant. */
1050 register int c
= getc (finput
);
1059 goto string_constant
;
1066 if (!doing_objc_thang
)
1074 while (isalnum (c
) || c
== '_')
1076 if (p
>= token_buffer
+ maxtoken
)
1077 p
= extend_token_buffer (p
);
1085 value
= recognize_objc_keyword (token_buffer
+ 1);
1088 error ("invalid Objective C keyword `%s'", token_buffer
);
1089 /* Cause a syntax error--1 is not a valid token type. */
1093 case 'A': case 'B': case 'C': case 'D': case 'E':
1094 case 'F': case 'G': case 'H': case 'I': case 'J':
1095 case 'K': case 'M': case 'N': case 'O':
1096 case 'P': case 'Q': case 'R': case 'S': case 'T':
1097 case 'U': case 'V': case 'W': case 'X': case 'Y':
1099 case 'a': case 'b': case 'c': case 'd': case 'e':
1100 case 'f': case 'g': case 'h': case 'i': case 'j':
1101 case 'k': case 'l': case 'm': case 'n': case 'o':
1102 case 'p': case 'q': case 'r': case 's': case 't':
1103 case 'u': case 'v': case 'w': case 'x': case 'y':
1108 while (isalnum (c
) || c
== '_' || c
== '$' || c
== '@')
1110 /* Make sure this char really belongs in an identifier. */
1111 if (c
== '@' && ! doing_objc_thang
)
1113 if (c
== '$' && ! dollars_in_ident
)
1116 if (p
>= token_buffer
+ maxtoken
)
1117 p
= extend_token_buffer (p
);
1129 /* Try to recognize a keyword. Uses minimum-perfect hash function */
1132 register struct resword
*ptr
;
1134 if (ptr
= is_reserved_word (token_buffer
, p
- token_buffer
))
1137 yylval
.ttype
= ridpointers
[(int) ptr
->rid
];
1138 value
= (int) ptr
->token
;
1140 /* Even if we decided to recognize asm, still perhaps warn. */
1142 && (value
== ASM_KEYWORD
|| value
== TYPEOF
1143 || ptr
->rid
== RID_INLINE
)
1144 && token_buffer
[0] != '_')
1145 pedwarn ("ANSI does not permit the keyword `%s'",
1150 /* If we did not find a keyword, look for an identifier
1153 if (value
== IDENTIFIER
)
1155 yylval
.ttype
= get_identifier (token_buffer
);
1156 lastiddecl
= lookup_name (yylval
.ttype
);
1158 if (lastiddecl
!= 0 && TREE_CODE (lastiddecl
) == TYPE_DECL
)
1160 /* A user-invisible read-only initialized variable
1161 should be replaced by its value.
1162 We handle only strings since that's the only case used in C. */
1163 else if (lastiddecl
!= 0 && TREE_CODE (lastiddecl
) == VAR_DECL
1164 && DECL_IGNORED_P (lastiddecl
)
1165 && TREE_READONLY (lastiddecl
)
1166 && DECL_INITIAL (lastiddecl
) != 0
1167 && TREE_CODE (DECL_INITIAL (lastiddecl
)) == STRING_CST
)
1169 yylval
.ttype
= DECL_INITIAL (lastiddecl
);
1172 else if (doing_objc_thang
)
1174 tree objc_interface_decl
= lookup_interface (yylval
.ttype
);
1176 if (objc_interface_decl
)
1179 yylval
.ttype
= objc_interface_decl
;
1186 case '0': case '1': case '2': case '3': case '4':
1187 case '5': case '6': case '7': case '8': case '9':
1192 int largest_digit
= 0;
1194 /* for multi-precision arithmetic,
1195 we actually store only HOST_BITS_PER_CHAR bits in each part.
1196 The number of parts is chosen so as to be sufficient to hold
1197 the enough bits to fit into the two HOST_WIDE_INTs that contain
1198 the integer value (this is always at least as many bits as are
1199 in a target `long long' value, but may be wider). */
1200 #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2 + 2)
1201 int parts
[TOTAL_PARTS
];
1204 enum anon1
{ NOT_FLOAT
, AFTER_POINT
, TOO_MANY_POINTS
} floatflag
1207 for (count
= 0; count
< TOTAL_PARTS
; count
++)
1215 *p
++ = (c
= getc (finput
));
1216 if ((c
== 'x') || (c
== 'X'))
1219 *p
++ = (c
= getc (finput
));
1221 /* Leading 0 forces octal unless the 0 is the only digit. */
1222 else if (c
>= '0' && c
<= '9')
1231 /* Read all the digits-and-decimal-points. */
1234 || (isalnum (c
) && (c
!= 'l') && (c
!= 'L')
1235 && (c
!= 'u') && (c
!= 'U')
1236 && (floatflag
== NOT_FLOAT
|| ((c
!= 'f') && (c
!= 'F')))))
1241 error ("floating constant may not be in radix 16");
1242 if (floatflag
== AFTER_POINT
)
1244 error ("malformed floating constant");
1245 floatflag
= TOO_MANY_POINTS
;
1248 floatflag
= AFTER_POINT
;
1251 *p
++ = c
= getc (finput
);
1252 /* Accept '.' as the start of a floating-point number
1253 only when it is followed by a digit.
1254 Otherwise, unread the following non-digit
1255 and use the '.' as a structural token. */
1256 if (p
== token_buffer
+ 2 && !isdigit (c
))
1267 error ("parse error at `..'");
1270 token_buffer
[1] = 0;
1277 /* It is not a decimal point.
1278 It should be a digit (perhaps a hex digit). */
1284 else if (base
<= 10)
1286 if (c
== 'e' || c
== 'E')
1289 floatflag
= AFTER_POINT
;
1290 break; /* start of exponent */
1292 error ("nondigits in number and not hexadecimal");
1303 if (c
>= largest_digit
)
1307 for (count
= 0; count
< TOTAL_PARTS
; count
++)
1309 parts
[count
] *= base
;
1313 += (parts
[count
-1] >> HOST_BITS_PER_CHAR
);
1315 &= (1 << HOST_BITS_PER_CHAR
) - 1;
1321 /* If the extra highest-order part ever gets anything in it,
1322 the number is certainly too big. */
1323 if (parts
[TOTAL_PARTS
- 1] != 0)
1326 if (p
>= token_buffer
+ maxtoken
- 3)
1327 p
= extend_token_buffer (p
);
1328 *p
++ = (c
= getc (finput
));
1333 error ("numeric constant with no digits");
1335 if (largest_digit
>= base
)
1336 error ("numeric constant contains digits beyond the radix");
1338 /* Remove terminating char from the token buffer and delimit the string */
1341 if (floatflag
!= NOT_FLOAT
)
1343 tree type
= double_type_node
;
1344 int garbage_chars
= 0, exceeds_double
= 0;
1345 REAL_VALUE_TYPE value
;
1348 /* Read explicit exponent if any, and put it in tokenbuf. */
1350 if ((c
== 'e') || (c
== 'E'))
1352 if (p
>= token_buffer
+ maxtoken
- 3)
1353 p
= extend_token_buffer (p
);
1356 if ((c
== '+') || (c
== '-'))
1362 error ("floating constant exponent has no digits");
1365 if (p
>= token_buffer
+ maxtoken
- 3)
1366 p
= extend_token_buffer (p
);
1375 /* Convert string to a double, checking for overflow. */
1376 if (setjmp (handler
))
1378 error ("floating constant out of range");
1383 set_float_handler (handler
);
1384 value
= REAL_VALUE_ATOF (token_buffer
);
1385 set_float_handler (NULL_PTR
);
1388 if (errno
== ERANGE
&& !flag_traditional
&& pedantic
)
1390 /* ERANGE is also reported for underflow,
1391 so test the value to distinguish overflow from that. */
1392 if (REAL_VALUES_LESS (dconst1
, value
)
1393 || REAL_VALUES_LESS (value
, dconstm1
))
1395 pedwarn ("floating point number exceeds range of `double'");
1401 /* Read the suffixes to choose a data type. */
1405 type
= float_type_node
;
1406 value
= REAL_VALUE_TRUNCATE (TYPE_MODE (type
), value
);
1407 if (REAL_VALUE_ISINF (value
) && ! exceeds_double
&& pedantic
)
1408 pedwarn ("floating point number exceeds range of `float'");
1413 type
= long_double_type_node
;
1417 /* Note: garbage_chars is -1 if first char is *not* garbage. */
1418 while (isalnum (c
) || c
== '.' || c
== '_'
1419 || (!flag_traditional
&& (c
== '+' || c
== '-')
1420 && (p
[-1] == 'e' || p
[-1] == 'E')))
1422 if (p
>= token_buffer
+ maxtoken
- 3)
1423 p
= extend_token_buffer (p
);
1428 if (garbage_chars
> 0)
1429 error ("garbage at end of number");
1431 /* Create a node with determined type and value. */
1432 yylval
.ttype
= build_real (type
, value
);
1439 tree traditional_type
, ansi_type
, type
;
1440 HOST_WIDE_INT high
, low
;
1441 int spec_unsigned
= 0;
1443 int spec_long_long
= 0;
1448 if (c
== 'u' || c
== 'U')
1451 error ("two `u's in integer constant");
1454 else if (c
== 'l' || c
== 'L')
1459 error ("three `l's in integer constant");
1461 pedwarn ("ANSI C forbids long long integer constants");
1468 if (isalnum (c
) || c
== '.' || c
== '_'
1469 || (!flag_traditional
&& (c
== '+' || c
== '-')
1470 && (p
[-1] == 'e' || p
[-1] == 'E')))
1472 error ("garbage at end of number");
1473 while (isalnum (c
) || c
== '.' || c
== '_'
1474 || (!flag_traditional
&& (c
== '+' || c
== '-')
1475 && (p
[-1] == 'e' || p
[-1] == 'E')))
1477 if (p
>= token_buffer
+ maxtoken
- 3)
1478 p
= extend_token_buffer (p
);
1485 if (p
>= token_buffer
+ maxtoken
- 3)
1486 p
= extend_token_buffer (p
);
1493 /* If the constant is not long long and it won't fit in an
1494 unsigned long, or if the constant is long long and won't fit
1495 in an unsigned long long, then warn that the constant is out
1498 /* ??? This assumes that long long and long integer types are
1499 a multiple of 8 bits. This better than the original code
1500 though which assumed that long was exactly 32 bits and long
1501 long was exactly 64 bits. */
1504 bytes
= TYPE_PRECISION (long_long_integer_type_node
) / 8;
1506 bytes
= TYPE_PRECISION (long_integer_type_node
) / 8;
1509 for (i
= bytes
; i
< TOTAL_PARTS
; i
++)
1513 pedwarn ("integer constant out of range");
1515 /* This is simplified by the fact that our constant
1516 is always positive. */
1520 for (i
= 0; i
< HOST_BITS_PER_WIDE_INT
/ HOST_BITS_PER_CHAR
; i
++)
1522 high
|= ((HOST_WIDE_INT
) parts
[i
+ (HOST_BITS_PER_WIDE_INT
1523 / HOST_BITS_PER_CHAR
)]
1524 << (i
* HOST_BITS_PER_CHAR
));
1525 low
|= (HOST_WIDE_INT
) parts
[i
] << (i
* HOST_BITS_PER_CHAR
);
1528 yylval
.ttype
= build_int_2 (low
, high
);
1529 TREE_TYPE (yylval
.ttype
) = long_long_unsigned_type_node
;
1531 /* If warn_traditional, calculate both the ANSI type and the
1532 traditional type, then see if they disagree.
1533 Otherwise, calculate only the type for the dialect in use. */
1534 if (warn_traditional
|| flag_traditional
)
1536 /* Calculate the traditional type. */
1537 /* Traditionally, any constant is signed;
1538 but if unsigned is specified explicitly, obey that.
1539 Use the smallest size with the right number of bits,
1540 except for one special case with decimal constants. */
1541 if (! spec_long
&& base
!= 10
1542 && int_fits_type_p (yylval
.ttype
, unsigned_type_node
))
1543 traditional_type
= (spec_unsigned
? unsigned_type_node
1544 : integer_type_node
);
1545 /* A decimal constant must be long
1546 if it does not fit in type int.
1547 I think this is independent of whether
1548 the constant is signed. */
1549 else if (! spec_long
&& base
== 10
1550 && int_fits_type_p (yylval
.ttype
, integer_type_node
))
1551 traditional_type
= (spec_unsigned
? unsigned_type_node
1552 : integer_type_node
);
1553 else if (! spec_long_long
)
1554 traditional_type
= (spec_unsigned
? long_unsigned_type_node
1555 : long_integer_type_node
);
1557 traditional_type
= (spec_unsigned
1558 ? long_long_unsigned_type_node
1559 : long_long_integer_type_node
);
1561 if (warn_traditional
|| ! flag_traditional
)
1563 /* Calculate the ANSI type. */
1564 if (! spec_long
&& ! spec_unsigned
1565 && int_fits_type_p (yylval
.ttype
, integer_type_node
))
1566 ansi_type
= integer_type_node
;
1567 else if (! spec_long
&& (base
!= 10 || spec_unsigned
)
1568 && int_fits_type_p (yylval
.ttype
, unsigned_type_node
))
1569 ansi_type
= unsigned_type_node
;
1570 else if (! spec_unsigned
&& !spec_long_long
1571 && int_fits_type_p (yylval
.ttype
, long_integer_type_node
))
1572 ansi_type
= long_integer_type_node
;
1573 else if (! spec_long_long
)
1574 ansi_type
= long_unsigned_type_node
;
1575 else if (! spec_unsigned
1576 /* Verify value does not overflow into sign bit. */
1577 && TREE_INT_CST_HIGH (yylval
.ttype
) >= 0
1578 && int_fits_type_p (yylval
.ttype
,
1579 long_long_integer_type_node
))
1580 ansi_type
= long_long_integer_type_node
;
1582 ansi_type
= long_long_unsigned_type_node
;
1585 type
= flag_traditional
? traditional_type
: ansi_type
;
1587 if (warn_traditional
&& traditional_type
!= ansi_type
)
1589 if (TYPE_PRECISION (traditional_type
)
1590 != TYPE_PRECISION (ansi_type
))
1591 warning ("width of integer constant changes with -traditional");
1592 else if (TREE_UNSIGNED (traditional_type
)
1593 != TREE_UNSIGNED (ansi_type
))
1594 warning ("integer constant is unsigned in ANSI C, signed with -traditional");
1596 warning ("width of integer constant may change on other systems with -traditional");
1599 if (!flag_traditional
&& !int_fits_type_p (yylval
.ttype
, type
)
1601 pedwarn ("integer constant out of range");
1603 if (base
== 10 && ! spec_unsigned
&& TREE_UNSIGNED (type
))
1604 warning ("decimal constant is so large that it is unsigned");
1606 if (flag_traditional
&& !int_fits_type_p (yylval
.ttype
, type
))
1607 /* The traditional constant 0x80000000 is signed
1608 but doesn't fit in the range of int.
1609 This will change it to -0x80000000, which does fit. */
1611 TREE_TYPE (yylval
.ttype
) = unsigned_type (type
);
1612 yylval
.ttype
= convert (type
, yylval
.ttype
);
1615 TREE_TYPE (yylval
.ttype
) = type
;
1620 value
= CONSTANT
; break;
1626 register int result
= 0;
1627 register int num_chars
= 0;
1628 unsigned width
= TYPE_PRECISION (char_type_node
);
1633 width
= WCHAR_TYPE_SIZE
;
1634 #ifdef MULTIBYTE_CHARS
1635 max_chars
= MB_CUR_MAX
;
1641 max_chars
= TYPE_PRECISION (integer_type_node
) / width
;
1649 if (c
== '\'' || c
== EOF
)
1655 c
= readescape (&ignore
);
1658 if (width
< HOST_BITS_PER_INT
1659 && (unsigned) c
>= (1 << width
))
1660 pedwarn ("escape sequence out of range for character");
1665 pedwarn ("ANSI C forbids newline in character constant");
1670 if (num_chars
> maxtoken
- 4)
1671 extend_token_buffer (token_buffer
);
1673 token_buffer
[num_chars
] = c
;
1675 /* Merge character into result; ignore excess chars. */
1676 if (num_chars
< max_chars
+ 1)
1678 if (width
< HOST_BITS_PER_INT
)
1679 result
= (result
<< width
) | (c
& ((1 << width
) - 1));
1685 token_buffer
[num_chars
+ 1] = '\'';
1686 token_buffer
[num_chars
+ 2] = 0;
1689 error ("malformatted character constant");
1690 else if (num_chars
== 0)
1691 error ("empty character constant");
1692 else if (num_chars
> max_chars
)
1694 num_chars
= max_chars
;
1695 error ("character constant too long");
1697 else if (num_chars
!= 1 && ! flag_traditional
)
1698 warning ("multi-character character constant");
1700 /* If char type is signed, sign-extend the constant. */
1703 int num_bits
= num_chars
* width
;
1704 if (TREE_UNSIGNED (char_type_node
)
1705 || ((result
>> (num_bits
- 1)) & 1) == 0)
1707 = build_int_2 (result
& ((unsigned HOST_WIDE_INT
) ~0
1708 >> (HOST_BITS_PER_WIDE_INT
- num_bits
)),
1712 = build_int_2 (result
| ~((unsigned HOST_WIDE_INT
) ~0
1713 >> (HOST_BITS_PER_WIDE_INT
- num_bits
)),
1715 TREE_TYPE (yylval
.ttype
) = integer_type_node
;
1719 #ifdef MULTIBYTE_CHARS
1720 /* Set the initial shift state and convert the next sequence. */
1722 /* In all locales L'\0' is zero and mbtowc will return zero,
1725 || (num_chars
== 1 && token_buffer
[1] != '\0'))
1728 (void) mbtowc (NULL_PTR
, NULL_PTR
, 0);
1729 if (mbtowc (& wc
, token_buffer
+ 1, num_chars
) == num_chars
)
1732 warning ("Ignoring invalid multibyte character");
1735 yylval
.ttype
= build_int_2 (result
, 0);
1736 TREE_TYPE (yylval
.ttype
) = wchar_type_node
;
1747 p
= token_buffer
+ 1;
1749 while (c
!= '"' && c
>= 0)
1751 /* ignore_escape_flag is set for reading the filename in #line. */
1752 if (!ignore_escape_flag
&& c
== '\\')
1755 c
= readescape (&ignore
);
1759 && TYPE_PRECISION (char_type_node
) < HOST_BITS_PER_INT
1760 && c
>= (1 << TYPE_PRECISION (char_type_node
)))
1761 pedwarn ("escape sequence out of range for character");
1766 pedwarn ("ANSI C forbids newline in string constant");
1770 if (p
== token_buffer
+ maxtoken
)
1771 p
= extend_token_buffer (p
);
1779 /* We have read the entire constant.
1780 Construct a STRING_CST for the result. */
1784 /* If this is a L"..." wide-string, convert the multibyte string
1785 to a wide character string. */
1786 char *widep
= (char *) alloca ((p
- token_buffer
) * WCHAR_BYTES
);
1789 #ifdef MULTIBYTE_CHARS
1790 len
= mbstowcs ((wchar_t *) widep
, token_buffer
+ 1, p
- token_buffer
);
1791 if ((unsigned) len
>= (p
- token_buffer
))
1793 warning ("Ignoring invalid multibyte string");
1796 bzero (widep
+ (len
* WCHAR_BYTES
), WCHAR_BYTES
);
1799 union { long l
; char c
[sizeof (long)]; } u
;
1803 /* Determine whether host is little or big endian. */
1805 big_endian
= u
.c
[sizeof (long) - 1];
1806 wp
= widep
+ (big_endian
? WCHAR_BYTES
- 1 : 0);
1808 bzero (widep
, (p
- token_buffer
) * WCHAR_BYTES
);
1809 for (cp
= token_buffer
+ 1; cp
< p
; cp
++)
1810 *wp
= *cp
, wp
+= WCHAR_BYTES
;
1811 len
= p
- token_buffer
- 1;
1814 yylval
.ttype
= build_string ((len
+ 1) * WCHAR_BYTES
, widep
);
1815 TREE_TYPE (yylval
.ttype
) = wchar_array_type_node
;
1819 yylval
.ttype
= build_string (p
- token_buffer
, token_buffer
+ 1);
1820 TREE_TYPE (yylval
.ttype
) = char_array_type_node
;
1826 value
= STRING
; break;
1849 yylval
.code
= PLUS_EXPR
; break;
1851 yylval
.code
= MINUS_EXPR
; break;
1853 yylval
.code
= BIT_AND_EXPR
; break;
1855 yylval
.code
= BIT_IOR_EXPR
; break;
1857 yylval
.code
= MULT_EXPR
; break;
1859 yylval
.code
= TRUNC_DIV_EXPR
; break;
1861 yylval
.code
= TRUNC_MOD_EXPR
; break;
1863 yylval
.code
= BIT_XOR_EXPR
; break;
1865 yylval
.code
= LSHIFT_EXPR
; break;
1867 yylval
.code
= RSHIFT_EXPR
; break;
1869 yylval
.code
= LT_EXPR
; break;
1871 yylval
.code
= GT_EXPR
; break;
1874 token_buffer
[1] = c1
= getc (finput
);
1875 token_buffer
[2] = 0;
1882 value
= ARITHCOMPARE
; yylval
.code
= LE_EXPR
; goto done
;
1884 value
= ARITHCOMPARE
; yylval
.code
= GE_EXPR
; goto done
;
1886 value
= EQCOMPARE
; yylval
.code
= NE_EXPR
; goto done
;
1888 value
= EQCOMPARE
; yylval
.code
= EQ_EXPR
; goto done
;
1890 value
= ASSIGN
; goto done
;
1896 value
= PLUSPLUS
; goto done
;
1898 value
= MINUSMINUS
; goto done
;
1900 value
= ANDAND
; goto done
;
1902 value
= OROR
; goto done
;
1910 else if ((c
== '-') && (c1
== '>'))
1911 { value
= POINTSAT
; goto done
; }
1912 ungetc (c1
, finput
);
1913 token_buffer
[1] = 0;
1915 if ((c
== '<') || (c
== '>'))
1916 value
= ARITHCOMPARE
;
1922 /* Don't make yyparse think this is eof. */
1931 /* yylloc.last_line = lineno; */
1936 /* Sets the value of the 'yydebug' variable to VALUE.
1937 This is a function so we don't have to have YYDEBUG defined
1938 in order to build the compiler. */
1947 warning ("YYDEBUG not defined.");