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");
307 UNSET_RESERVED_WORD ("iterator");
311 UNSET_RESERVED_WORD ("asm");
312 UNSET_RESERVED_WORD ("typeof");
313 UNSET_RESERVED_WORD ("inline");
314 UNSET_RESERVED_WORD ("iterator");
319 reinit_parse_for_function ()
323 /* Function used when yydebug is set, to print a token in more detail. */
326 yyprint (file, yychar, yylval)
337 if (IDENTIFIER_POINTER (t))
338 fprintf (file, " `%s'", IDENTIFIER_POINTER (t));
343 if (TREE_CODE (t) == INTEGER_CST)
345 #if HOST_BITS_PER_WIDE_INT == 64
346 #if HOST_BITS_PER_WIDE_INT != HOST_BITS_PER_INT
352 #if HOST_BITS_PER_WIDE_INT != HOST_BITS_PER_INT
358 TREE_INT_CST_HIGH (t), TREE_INT_CST_LOW (t));
364 /* If C is not whitespace, return C.
365 Otherwise skip whitespace and return first nonwhite char read. */
371 static int newline_warning = 0;
377 /* We don't recognize comments here, because
378 cpp output can include / and * consecutively as operators.
379 Also, there's no need, since cpp removes all comments. */
382 c = check_newline ();
394 /* ANSI C says the effects of a carriage return in a source file
396 if (pedantic && !newline_warning)
398 warning ("carriage return in source file");
399 warning ("(we only warn about the first carriage return)");
410 error ("stray '\\' in program");
420 /* Skips all of the white space at the current location in the input file.
421 Must use and reset nextchar if it has the next character. */
424 position_after_white_space ()
429 c = nextchar, nextchar = -1;
433 ungetc (skip_white_space (c), finput);
436 /* Make the token buffer longer, preserving the data in it.
437 P should point to just beyond the last valid character in the old buffer.
438 The value we return is a pointer to the new buffer
439 at a place corresponding to P. */
442 extend_token_buffer (p)
445 int offset = p - token_buffer;
447 maxtoken = maxtoken * 2 + 10;
448 token_buffer = (char *) xrealloc (token_buffer, maxtoken + 2);
450 return token_buffer + offset;
453 /* At the beginning of a line, increment the line number
454 and process any #-directive on this line.
455 If the line is a #-directive, read the entire line and return a newline.
456 Otherwise, return the line's first non-whitespace character. */
466 /* Read first nonwhite char on the line. */
469 while (c == ' ' || c == '\t')
474 /* If not #, return it so caller will use it. */
478 /* Read first nonwhite char after the `#'. */
481 while (c == ' ' || c == '\t')
484 /* If a letter follows, then if the word here is `line', skip
485 it and ignore it; otherwise, ignore the line, with an error
486 if the word isn't `pragma', `ident', `define', or `undef'. */
488 if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'))
492 if (getc (finput) == 'r'
493 && getc (finput) == 'a'
494 && getc (finput) == 'g'
495 && getc (finput) == 'm'
496 && getc (finput) == 'a'
497 && ((c = getc (finput)) == ' ' || c == '\t' || c == '\n'))
499 #ifdef HANDLE_SYSV_PRAGMA
500 return handle_sysv_pragma (finput, c);
501 #endif /* HANDLE_SYSV_PRAGMA */
503 HANDLE_PRAGMA (finput);
504 #endif /* HANDLE_PRAGMA */
511 if (getc (finput) == 'e'
512 && getc (finput) == 'f'
513 && getc (finput) == 'i'
514 && getc (finput) == 'n'
515 && getc (finput) == 'e'
516 && ((c = getc (finput)) == ' ' || c == '\t' || c == '\n'))
518 #ifdef DWARF_DEBUGGING_INFO
519 if ((debug_info_level == DINFO_LEVEL_VERBOSE)
520 && (write_symbols == DWARF_DEBUG))
521 dwarfout_define (lineno, get_directive_line (finput));
522 #endif /* DWARF_DEBUGGING_INFO */
528 if (getc (finput) == 'n'
529 && getc (finput) == 'd'
530 && getc (finput) == 'e'
531 && getc (finput) == 'f'
532 && ((c = getc (finput)) == ' ' || c == '\t' || c == '\n'))
534 #ifdef DWARF_DEBUGGING_INFO
535 if ((debug_info_level == DINFO_LEVEL_VERBOSE)
536 && (write_symbols == DWARF_DEBUG))
537 dwarfout_undef (lineno, get_directive_line (finput));
538 #endif /* DWARF_DEBUGGING_INFO */
544 if (getc (finput) == 'i'
545 && getc (finput) == 'n'
546 && getc (finput) == 'e'
547 && ((c = getc (finput)) == ' ' || c == '\t'))
552 if (getc (finput) == 'd'
553 && getc (finput) == 'e'
554 && getc (finput) == 'n'
555 && getc (finput) == 't'
556 && ((c = getc (finput)) == ' ' || c == '\t'))
558 /* #ident. The pedantic warning is now in cccp.c. */
560 /* Here we have just seen `#ident '.
561 A string constant should follow. */
563 while (c == ' ' || c == '\t')
566 /* If no argument, ignore the line. */
573 || TREE_CODE (yylval.ttype) != STRING_CST)
575 error ("invalid #ident");
581 #ifdef ASM_OUTPUT_IDENT
582 ASM_OUTPUT_IDENT (asm_out_file, TREE_STRING_POINTER (yylval.ttype));
586 /* Skip the rest of this line. */
591 error ("undefined or invalid # directive");
596 /* Here we have either `#line' or `# <nonletter>'.
597 In either case, it should be a line number; a digit should follow. */
599 while (c == ' ' || c == '\t')
602 /* If the # is the only nonwhite char on the line,
603 just ignore it. Check the new newline. */
607 /* Something follows the #; read a token. */
612 if (token == CONSTANT
613 && TREE_CODE (yylval.ttype) == INTEGER_CST)
615 int old_lineno = lineno;
617 /* subtract one, because it is the following line that
618 gets the specified number */
620 int l = TREE_INT_CST_LOW (yylval.ttype) - 1;
622 /* Is this the last nonwhite stuff on the line? */
624 while (c == ' ' || c == '\t')
628 /* No more: store the line number and check following line. */
634 /* More follows: it must be a string constant (filename). */
636 /* Read the string constant, but don't treat \ as special. */
637 ignore_escape_flag = 1;
639 ignore_escape_flag = 0;
641 if (token != STRING || TREE_CODE (yylval.ttype) != STRING_CST)
643 error ("invalid #line");
648 = (char *) permalloc (TREE_STRING_LENGTH (yylval.ttype) + 1);
649 strcpy (input_filename, TREE_STRING_POINTER (yylval.ttype));
652 /* Each change of file name
653 reinitializes whether we are now in a system header. */
654 in_system_header = 0;
656 if (main_input_filename == 0)
657 main_input_filename = input_filename;
659 /* Is this the last nonwhite stuff on the line? */
661 while (c == ' ' || c == '\t')
665 /* Update the name in the top element of input_file_stack. */
666 if (input_file_stack)
667 input_file_stack->name = input_filename;
676 /* `1' after file name means entering new file.
677 `2' after file name means just left a file. */
679 if (token == CONSTANT
680 && TREE_CODE (yylval.ttype) == INTEGER_CST)
682 if (TREE_INT_CST_LOW (yylval.ttype) == 1)
684 /* Pushing to a new file. */
686 = (struct file_stack *) xmalloc (sizeof (struct file_stack));
687 input_file_stack->line = old_lineno;
688 p->next = input_file_stack;
689 p->name = input_filename;
690 input_file_stack = p;
691 input_file_stack_tick++;
692 #ifdef DWARF_DEBUGGING_INFO
693 if (debug_info_level == DINFO_LEVEL_VERBOSE
694 && write_symbols == DWARF_DEBUG)
695 dwarfout_start_new_source_file (input_filename);
696 #endif /* DWARF_DEBUGGING_INFO */
700 else if (TREE_INT_CST_LOW (yylval.ttype) == 2)
702 /* Popping out of a file. */
703 if (input_file_stack->next)
705 struct file_stack *p = input_file_stack;
706 input_file_stack = p->next;
708 input_file_stack_tick++;
709 #ifdef DWARF_DEBUGGING_INFO
710 if (debug_info_level == DINFO_LEVEL_VERBOSE
711 && write_symbols == DWARF_DEBUG)
712 dwarfout_resume_previous_source_file (input_file_stack->line);
713 #endif /* DWARF_DEBUGGING_INFO */
716 error ("#-lines for entering and leaving files don't match");
722 /* Now that we've pushed or popped the input stack,
723 update the name in the top element. */
724 if (input_file_stack)
725 input_file_stack->name = input_filename;
727 /* If we have handled a `1' or a `2',
728 see if there is another number to read. */
731 /* Is this the last nonwhite stuff on the line? */
733 while (c == ' ' || c == '\t')
743 /* `3' after file name means this is a system header file. */
745 if (token == CONSTANT
746 && TREE_CODE (yylval.ttype) == INTEGER_CST
747 && TREE_INT_CST_LOW (yylval.ttype) == 3)
748 in_system_header = 1;
751 error ("invalid #-line");
753 /* skip the rest of this line. */
757 while ((c = getc (finput)) != EOF && c != '\n');
761 #ifdef HANDLE_SYSV_PRAGMA
763 /* Handle a #pragma directive. INPUT is the current input stream,
764 and C is a character to reread. Processes the entire input line
765 and returns a character for the caller to reread: either \n or EOF. */
767 /* This function has to be in this file, in order to get at
771 handle_sysv_pragma (input, c)
777 while (c == ' ' || c == '\t')
779 if (c == '\n' || c == EOF)
781 handle_pragma_token (0, 0);
791 handle_pragma_token (token_buffer, yylval.ttype);
794 handle_pragma_token (token_buffer, 0);
797 c = nextchar, nextchar = -1;
803 #endif /* HANDLE_SYSV_PRAGMA */
805 #define isalnum(char) ((char >= 'a' && char <= 'z') || (char >= 'A' && char <= 'Z') || (char >= '0' && char <= '9'))
806 #define isdigit(char) (char >= '0' && char <= '9')
807 #define ENDFILE -1 /* token that represents end-of-file */
809 /* Read an escape sequence, returning its equivalent as a character,
810 or store 1 in *ignore_ptr if it is backslash-newline. */
813 readescape (ignore_ptr)
816 register int c = getc (finput);
818 register unsigned count;
825 if (warn_traditional)
826 warning ("the meaning of `\\x' varies with -traditional");
828 if (flag_traditional)
837 if (!(c >= 'a' && c <= 'f')
838 && !(c >= 'A' && c <= 'F')
839 && !(c >= '0' && c <= '9'))
845 if (c >= 'a' && c <= 'f')
846 code += c - 'a' + 10;
847 if (c >= 'A' && c <= 'F')
848 code += c - 'A' + 10;
849 if (c >= '0' && c <= '9')
851 if (code != 0 || count != 0)
860 error ("\\x used with no following hex digits");
862 /* Digits are all 0's. Ok. */
864 else if ((count - 1) * 4 >= TYPE_PRECISION (integer_type_node)
866 && ((1 << (TYPE_PRECISION (integer_type_node) - (count - 1) * 4))
868 pedwarn ("hex escape out of range");
871 case '0': case '1': case '2': case '3': case '4':
872 case '5': case '6': case '7':
875 while ((c <= '7') && (c >= '0') && (count++ < 3))
877 code = (code * 8) + (c - '0');
883 case '\\': case '\'': case '"':
892 return TARGET_NEWLINE;
907 if (warn_traditional)
908 warning ("the meaning of `\\a' varies with -traditional");
910 if (flag_traditional)
915 #if 0 /* Vertical tab is present in common usage compilers. */
916 if (flag_traditional)
924 pedwarn ("non-ANSI-standard escape sequence, `\\%c'", c);
930 /* `\(', etc, are used at beginning of line to avoid confusing Emacs. */
935 pedwarn ("non-ANSI escape sequence `\\%c'", c);
938 if (c >= 040 && c < 0177)
939 pedwarn ("unknown escape sequence `\\%c'", c);
941 pedwarn ("unknown escape sequence: `\\' followed by char code 0x%x", c);
951 strcpy (buf, string);
953 /* We can't print string and character constants well
954 because the token_buffer contains the result of processing escapes. */
956 strcat (buf, " at end of input");
957 else if (token_buffer[0] == 0)
958 strcat (buf, " at null character");
959 else if (token_buffer[0] == '"')
960 strcat (buf, " before string constant");
961 else if (token_buffer[0] == '\'')
962 strcat (buf, " before character constant");
963 else if (token_buffer[0] < 040 || (unsigned char) token_buffer[0] >= 0177)
964 sprintf (buf + strlen (buf), " before character 0%o",
965 (unsigned char) token_buffer[0]);
967 strcat (buf, " before `%s'");
969 error (buf, token_buffer);
982 struct try_type type_sequence[] =
984 { &integer_type_node, 0, 0, 0},
985 { &unsigned_type_node, 1, 0, 0},
986 { &long_integer_type_node, 0, 1, 0},
987 { &long_unsigned_type_node, 1, 1, 0},
988 { &long_long_integer_type_node, 0, 1, 1},
989 { &long_long_unsigned_type_node, 1, 1, 1}
1002 c = nextchar, nextchar = -1;
1006 /* Effectively do c = skip_white_space (c)
1007 but do it faster in the usual cases. */
1020 /* Call skip_white_space so we can warn if appropriate. */
1025 c = skip_white_space (c);
1027 goto found_nonwhite;
1031 token_buffer[0] = c;
1032 token_buffer[1] = 0;
1034 /* yylloc.first_line = lineno; */
1040 token_buffer[0] = 0;
1045 if (dollars_in_ident)
1050 /* Capital L may start a wide-string or wide-character constant. */
1052 register int c = getc (finput);
1061 goto string_constant;
1068 if (!doing_objc_thang)
1076 while (isalnum (c) || c == '_')
1078 if (p >= token_buffer + maxtoken)
1079 p = extend_token_buffer (p);
1087 value = recognize_objc_keyword (token_buffer + 1);
1090 error ("invalid Objective C keyword `%s'", token_buffer);
1091 /* Cause a syntax error--1 is not a valid token type. */
1095 case 'A': case 'B': case 'C': case 'D': case 'E':
1096 case 'F': case 'G': case 'H': case 'I': case 'J':
1097 case 'K': case 'M': case 'N': case 'O':
1098 case 'P': case 'Q': case 'R': case 'S': case 'T':
1099 case 'U': case 'V': case 'W': case 'X': case 'Y':
1101 case 'a': case 'b': case 'c': case 'd': case 'e':
1102 case 'f': case 'g': case 'h': case 'i': case 'j':
1103 case 'k': case 'l': case 'm': case 'n': case 'o':
1104 case 'p': case 'q': case 'r': case 's': case 't':
1105 case 'u': case 'v': case 'w': case 'x': case 'y':
1110 while (isalnum (c) || c == '_' || c == '$' || c == '@')
1112 /* Make sure this char really belongs in an identifier. */
1113 if (c == '@' && ! doing_objc_thang)
1115 if (c == '$' && ! dollars_in_ident)
1118 if (p >= token_buffer + maxtoken)
1119 p = extend_token_buffer (p);
1131 /* Try to recognize a keyword. Uses minimum-perfect hash function */
1134 register struct resword *ptr;
1136 if (ptr = is_reserved_word (token_buffer, p - token_buffer))
1139 yylval.ttype = ridpointers[(int) ptr->rid];
1140 value = (int) ptr->token;
1142 /* Even if we decided to recognize asm, still perhaps warn. */
1144 && (value == ASM_KEYWORD || value == TYPEOF
1145 || ptr->rid == RID_INLINE)
1146 && token_buffer[0] != '_')
1147 pedwarn ("ANSI does not permit the keyword `%s'",
1152 /* If we did not find a keyword, look for an identifier
1155 if (value == IDENTIFIER)
1157 yylval.ttype = get_identifier (token_buffer);
1158 lastiddecl = lookup_name (yylval.ttype);
1160 if (lastiddecl != 0 && TREE_CODE (lastiddecl) == TYPE_DECL)
1162 /* A user-invisible read-only initialized variable
1163 should be replaced by its value.
1164 We handle only strings since that's the only case used in C. */
1165 else if (lastiddecl != 0 && TREE_CODE (lastiddecl) == VAR_DECL
1166 && DECL_IGNORED_P (lastiddecl)
1167 && TREE_READONLY (lastiddecl)
1168 && DECL_INITIAL (lastiddecl) != 0
1169 && TREE_CODE (DECL_INITIAL (lastiddecl)) == STRING_CST)
1171 yylval.ttype = DECL_INITIAL (lastiddecl);
1174 else if (doing_objc_thang)
1176 tree objc_interface_decl = lookup_interface (yylval.ttype);
1178 if (objc_interface_decl)
1181 yylval.ttype = objc_interface_decl;
1188 case '0': case '1': case '2': case '3': case '4':
1189 case '5': case '6': case '7': case '8': case '9':
1194 int largest_digit = 0;
1196 /* for multi-precision arithmetic,
1197 we actually store only HOST_BITS_PER_CHAR bits in each part.
1198 The number of parts is chosen so as to be sufficient to hold
1199 the enough bits to fit into the two HOST_WIDE_INTs that contain
1200 the integer value (this is always at least as many bits as are
1201 in a target `long long' value, but may be wider). */
1202 #define TOTAL_PARTS ((HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR) * 2 + 2)
1203 int parts[TOTAL_PARTS];
1206 enum anon1 { NOT_FLOAT, AFTER_POINT, TOO_MANY_POINTS} floatflag
1209 for (count = 0; count < TOTAL_PARTS; count++)
1217 *p++ = (c = getc (finput));
1218 if ((c == 'x') || (c == 'X'))
1221 *p++ = (c = getc (finput));
1223 /* Leading 0 forces octal unless the 0 is the only digit. */
1224 else if (c >= '0' && c <= '9')
1233 /* Read all the digits-and-decimal-points. */
1236 || (isalnum (c) && (c != 'l') && (c != 'L')
1237 && (c != 'u') && (c != 'U')
1238 && (floatflag == NOT_FLOAT || ((c != 'f') && (c != 'F')))))
1243 error ("floating constant may not be in radix 16");
1244 if (floatflag == AFTER_POINT)
1246 error ("malformed floating constant");
1247 floatflag = TOO_MANY_POINTS;
1250 floatflag = AFTER_POINT;
1253 *p++ = c = getc (finput);
1254 /* Accept '.' as the start of a floating-point number
1255 only when it is followed by a digit.
1256 Otherwise, unread the following non-digit
1257 and use the '.' as a structural token. */
1258 if (p == token_buffer + 2 && !isdigit (c))
1269 error ("parse error at `..'");
1272 token_buffer[1] = 0;
1279 /* It is not a decimal point.
1280 It should be a digit (perhaps a hex digit). */
1286 else if (base <= 10)
1288 if (c == 'e' || c == 'E')
1291 floatflag = AFTER_POINT;
1292 break; /* start of exponent */
1294 error ("nondigits in number and not hexadecimal");
1305 if (c >= largest_digit)
1309 for (count = 0; count < TOTAL_PARTS; count++)
1311 parts[count] *= base;
1315 += (parts[count-1] >> HOST_BITS_PER_CHAR);
1317 &= (1 << HOST_BITS_PER_CHAR) - 1;
1323 /* If the extra highest-order part ever gets anything in it,
1324 the number is certainly too big. */
1325 if (parts[TOTAL_PARTS - 1] != 0)
1328 if (p >= token_buffer + maxtoken - 3)
1329 p = extend_token_buffer (p);
1330 *p++ = (c = getc (finput));
1335 error ("numeric constant with no digits");
1337 if (largest_digit >= base)
1338 error ("numeric constant contains digits beyond the radix");
1340 /* Remove terminating char from the token buffer and delimit the string */
1343 if (floatflag != NOT_FLOAT)
1345 tree type = double_type_node;
1346 int garbage_chars = 0, exceeds_double = 0;
1347 REAL_VALUE_TYPE value;
1350 /* Read explicit exponent if any, and put it in tokenbuf. */
1352 if ((c == 'e') || (c == 'E'))
1354 if (p >= token_buffer + maxtoken - 3)
1355 p = extend_token_buffer (p);
1358 if ((c == '+') || (c == '-'))
1364 error ("floating constant exponent has no digits");
1367 if (p >= token_buffer + maxtoken - 3)
1368 p = extend_token_buffer (p);
1377 /* Convert string to a double, checking for overflow. */
1378 if (setjmp (handler))
1380 error ("floating constant out of range");
1385 set_float_handler (handler);
1386 value = REAL_VALUE_ATOF (token_buffer);
1387 set_float_handler (NULL_PTR);
1390 if (errno == ERANGE && !flag_traditional && pedantic)
1392 /* ERANGE is also reported for underflow,
1393 so test the value to distinguish overflow from that. */
1394 if (REAL_VALUES_LESS (dconst1, value)
1395 || REAL_VALUES_LESS (value, dconstm1))
1397 pedwarn ("floating point number exceeds range of `double'");
1403 /* Read the suffixes to choose a data type. */
1407 type = float_type_node;
1408 value = REAL_VALUE_TRUNCATE (TYPE_MODE (type), value);
1409 if (REAL_VALUE_ISINF (value) && ! exceeds_double && pedantic)
1410 pedwarn ("floating point number exceeds range of `float'");
1415 type = long_double_type_node;
1419 /* Note: garbage_chars is -1 if first char is *not* garbage. */
1420 while (isalnum (c) || c == '.' || c == '_'
1421 || (!flag_traditional && (c == '+' || c == '-')
1422 && (p[-1] == 'e' || p[-1] == 'E')))
1424 if (p >= token_buffer + maxtoken - 3)
1425 p = extend_token_buffer (p);
1430 if (garbage_chars > 0)
1431 error ("garbage at end of number");
1433 /* Create a node with determined type and value. */
1434 yylval.ttype = build_real (type, value);
1441 tree traditional_type, ansi_type, type;
1442 HOST_WIDE_INT high, low;
1443 int spec_unsigned = 0;
1445 int spec_long_long = 0;
1450 if (c == 'u' || c == 'U')
1453 error ("two `u's in integer constant");
1456 else if (c == 'l' || c == 'L')
1461 error ("three `l's in integer constant");
1463 pedwarn ("ANSI C forbids long long integer constants");
1470 if (isalnum (c) || c == '.' || c == '_'
1471 || (!flag_traditional && (c == '+' || c == '-')
1472 && (p[-1] == 'e' || p[-1] == 'E')))
1474 error ("garbage at end of number");
1475 while (isalnum (c) || c == '.' || c == '_'
1476 || (!flag_traditional && (c == '+' || c == '-')
1477 && (p[-1] == 'e' || p[-1] == 'E')))
1479 if (p >= token_buffer + maxtoken - 3)
1480 p = extend_token_buffer (p);
1487 if (p >= token_buffer + maxtoken - 3)
1488 p = extend_token_buffer (p);
1495 /* If the constant is not long long and it won't fit in an
1496 unsigned long, or if the constant is long long and won't fit
1497 in an unsigned long long, then warn that the constant is out
1500 /* ??? This assumes that long long and long integer types are
1501 a multiple of 8 bits. This better than the original code
1502 though which assumed that long was exactly 32 bits and long
1503 long was exactly 64 bits. */
1506 bytes = TYPE_PRECISION (long_long_integer_type_node) / 8;
1508 bytes = TYPE_PRECISION (long_integer_type_node) / 8;
1511 for (i = bytes; i < TOTAL_PARTS; i++)
1515 pedwarn ("integer constant out of range");
1517 /* This is simplified by the fact that our constant
1518 is always positive. */
1522 for (i = 0; i < HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; i++)
1524 high |= ((HOST_WIDE_INT) parts[i + (HOST_BITS_PER_WIDE_INT
1525 / HOST_BITS_PER_CHAR)]
1526 << (i * HOST_BITS_PER_CHAR));
1527 low |= (HOST_WIDE_INT) parts[i] << (i * HOST_BITS_PER_CHAR);
1530 yylval.ttype = build_int_2 (low, high);
1531 TREE_TYPE (yylval.ttype) = long_long_unsigned_type_node;
1533 /* If warn_traditional, calculate both the ANSI type and the
1534 traditional type, then see if they disagree.
1535 Otherwise, calculate only the type for the dialect in use. */
1536 if (warn_traditional || flag_traditional)
1538 /* Calculate the traditional type. */
1539 /* Traditionally, any constant is signed;
1540 but if unsigned is specified explicitly, obey that.
1541 Use the smallest size with the right number of bits,
1542 except for one special case with decimal constants. */
1543 if (! spec_long && base != 10
1544 && int_fits_type_p (yylval.ttype, unsigned_type_node))
1545 traditional_type = (spec_unsigned ? unsigned_type_node
1546 : integer_type_node);
1547 /* A decimal constant must be long
1548 if it does not fit in type int.
1549 I think this is independent of whether
1550 the constant is signed. */
1551 else if (! spec_long && base == 10
1552 && int_fits_type_p (yylval.ttype, integer_type_node))
1553 traditional_type = (spec_unsigned ? unsigned_type_node
1554 : integer_type_node);
1555 else if (! spec_long_long)
1556 traditional_type = (spec_unsigned ? long_unsigned_type_node
1557 : long_integer_type_node);
1559 traditional_type = (spec_unsigned
1560 ? long_long_unsigned_type_node
1561 : long_long_integer_type_node);
1563 if (warn_traditional || ! flag_traditional)
1565 /* Calculate the ANSI type. */
1566 if (! spec_long && ! spec_unsigned
1567 && int_fits_type_p (yylval.ttype, integer_type_node))
1568 ansi_type = integer_type_node;
1569 else if (! spec_long && (base != 10 || spec_unsigned)
1570 && int_fits_type_p (yylval.ttype, unsigned_type_node))
1571 ansi_type = unsigned_type_node;
1572 else if (! spec_unsigned && !spec_long_long
1573 && int_fits_type_p (yylval.ttype, long_integer_type_node))
1574 ansi_type = long_integer_type_node;
1575 else if (! spec_long_long)
1576 ansi_type = long_unsigned_type_node;
1577 else if (! spec_unsigned
1578 /* Verify value does not overflow into sign bit. */
1579 && TREE_INT_CST_HIGH (yylval.ttype) >= 0
1580 && int_fits_type_p (yylval.ttype,
1581 long_long_integer_type_node))
1582 ansi_type = long_long_integer_type_node;
1584 ansi_type = long_long_unsigned_type_node;
1587 type = flag_traditional ? traditional_type : ansi_type;
1589 if (warn_traditional && traditional_type != ansi_type)
1591 if (TYPE_PRECISION (traditional_type)
1592 != TYPE_PRECISION (ansi_type))
1593 warning ("width of integer constant changes with -traditional");
1594 else if (TREE_UNSIGNED (traditional_type)
1595 != TREE_UNSIGNED (ansi_type))
1596 warning ("integer constant is unsigned in ANSI C, signed with -traditional");
1598 warning ("width of integer constant may change on other systems with -traditional");
1601 if (!flag_traditional && !int_fits_type_p (yylval.ttype, type)
1603 pedwarn ("integer constant out of range");
1605 if (base == 10 && ! spec_unsigned && TREE_UNSIGNED (type))
1606 warning ("decimal constant is so large that it is unsigned");
1608 if (flag_traditional && !int_fits_type_p (yylval.ttype, type))
1609 /* The traditional constant 0x80000000 is signed
1610 but doesn't fit in the range of int.
1611 This will change it to -0x80000000, which does fit. */
1613 TREE_TYPE (yylval.ttype) = unsigned_type (type);
1614 yylval.ttype = convert (type, yylval.ttype);
1617 TREE_TYPE (yylval.ttype) = type;
1622 value = CONSTANT; break;
1628 register int result = 0;
1629 register int num_chars = 0;
1630 unsigned width = TYPE_PRECISION (char_type_node);
1635 width = WCHAR_TYPE_SIZE;
1636 #ifdef MULTIBYTE_CHARS
1637 max_chars = MB_CUR_MAX;
1643 max_chars = TYPE_PRECISION (integer_type_node) / width;
1651 if (c == '\'' || c == EOF)
1657 c = readescape (&ignore);
1660 if (width < HOST_BITS_PER_INT
1661 && (unsigned) c >= (1 << width))
1662 pedwarn ("escape sequence out of range for character");
1667 pedwarn ("ANSI C forbids newline in character constant");
1672 if (num_chars > maxtoken - 4)
1673 extend_token_buffer (token_buffer);
1675 token_buffer[num_chars] = c;
1677 /* Merge character into result; ignore excess chars. */
1678 if (num_chars < max_chars + 1)
1680 if (width < HOST_BITS_PER_INT)
1681 result = (result << width) | (c & ((1 << width) - 1));
1687 token_buffer[num_chars + 1] = '\'';
1688 token_buffer[num_chars + 2] = 0;
1691 error ("malformatted character constant");
1692 else if (num_chars == 0)
1693 error ("empty character constant");
1694 else if (num_chars > max_chars)
1696 num_chars = max_chars;
1697 error ("character constant too long");
1699 else if (num_chars != 1 && ! flag_traditional)
1700 warning ("multi-character character constant");
1702 /* If char type is signed, sign-extend the constant. */
1705 int num_bits = num_chars * width;
1706 if (TREE_UNSIGNED (char_type_node)
1707 || ((result >> (num_bits - 1)) & 1) == 0)
1709 = build_int_2 (result & ((unsigned HOST_WIDE_INT) ~0
1710 >> (HOST_BITS_PER_WIDE_INT - num_bits)),
1714 = build_int_2 (result | ~((unsigned HOST_WIDE_INT) ~0
1715 >> (HOST_BITS_PER_WIDE_INT - num_bits)),
1717 TREE_TYPE (yylval.ttype) = integer_type_node;
1721 #ifdef MULTIBYTE_CHARS
1722 /* Set the initial shift state and convert the next sequence. */
1724 /* In all locales L'\0' is zero and mbtowc will return zero,
1727 || (num_chars == 1 && token_buffer[1] != '\0'))
1730 (void) mbtowc (NULL_PTR, NULL_PTR, 0);
1731 if (mbtowc (& wc, token_buffer + 1, num_chars) == num_chars)
1734 warning ("Ignoring invalid multibyte character");
1737 yylval.ttype = build_int_2 (result, 0);
1738 TREE_TYPE (yylval.ttype) = wchar_type_node;
1749 p = token_buffer + 1;
1751 while (c != '"' && c >= 0)
1753 /* ignore_escape_flag is set for reading the filename in #line. */
1754 if (!ignore_escape_flag && c == '\\')
1757 c = readescape (&ignore);
1761 && TYPE_PRECISION (char_type_node) < HOST_BITS_PER_INT
1762 && c >= (1 << TYPE_PRECISION (char_type_node)))
1763 pedwarn ("escape sequence out of range for character");
1768 pedwarn ("ANSI C forbids newline in string constant");
1772 if (p == token_buffer + maxtoken)
1773 p = extend_token_buffer (p);
1781 /* We have read the entire constant.
1782 Construct a STRING_CST for the result. */
1786 /* If this is a L"..." wide-string, convert the multibyte string
1787 to a wide character string. */
1788 char *widep = (char *) alloca ((p - token_buffer) * WCHAR_BYTES);
1791 #ifdef MULTIBYTE_CHARS
1792 len = mbstowcs ((wchar_t *) widep, token_buffer + 1, p - token_buffer);
1793 if ((unsigned) len >= (p - token_buffer))
1795 warning ("Ignoring invalid multibyte string");
1798 bzero (widep + (len * WCHAR_BYTES), WCHAR_BYTES);
1801 union { long l; char c[sizeof (long)]; } u;
1805 /* Determine whether host is little or big endian. */
1807 big_endian = u.c[sizeof (long) - 1];
1808 wp = widep + (big_endian ? WCHAR_BYTES - 1 : 0);
1810 bzero (widep, (p - token_buffer) * WCHAR_BYTES);
1811 for (cp = token_buffer + 1; cp < p; cp++)
1812 *wp = *cp, wp += WCHAR_BYTES;
1813 len = p - token_buffer - 1;
1816 yylval.ttype = build_string ((len + 1) * WCHAR_BYTES, widep);
1817 TREE_TYPE (yylval.ttype) = wchar_array_type_node;
1821 yylval.ttype = build_string (p - token_buffer, token_buffer + 1);
1822 TREE_TYPE (yylval.ttype) = char_array_type_node;
1828 value = STRING; break;
1851 yylval.code = PLUS_EXPR; break;
1853 yylval.code = MINUS_EXPR; break;
1855 yylval.code = BIT_AND_EXPR; break;
1857 yylval.code = BIT_IOR_EXPR; break;
1859 yylval.code = MULT_EXPR; break;
1861 yylval.code = TRUNC_DIV_EXPR; break;
1863 yylval.code = TRUNC_MOD_EXPR; break;
1865 yylval.code = BIT_XOR_EXPR; break;
1867 yylval.code = LSHIFT_EXPR; break;
1869 yylval.code = RSHIFT_EXPR; break;
1871 yylval.code = LT_EXPR; break;
1873 yylval.code = GT_EXPR; break;
1876 token_buffer[1] = c1 = getc (finput);
1877 token_buffer[2] = 0;
1884 value = ARITHCOMPARE; yylval.code = LE_EXPR; goto done;
1886 value = ARITHCOMPARE; yylval.code = GE_EXPR; goto done;
1888 value = EQCOMPARE; yylval.code = NE_EXPR; goto done;
1890 value = EQCOMPARE; yylval.code = EQ_EXPR; goto done;
1892 value = ASSIGN; goto done;
1898 value = PLUSPLUS; goto done;
1900 value = MINUSMINUS; goto done;
1902 value = ANDAND; goto done;
1904 value = OROR; goto done;
1912 else if ((c == '-') && (c1 == '>'))
1913 { value = POINTSAT; goto done; }
1914 ungetc (c1, finput);
1915 token_buffer[1] = 0;
1917 if ((c == '<') || (c == '>'))
1918 value = ARITHCOMPARE;
1924 /* Don't make yyparse think this is eof. */
1933 /* yylloc.last_line = lineno; */
1938 /* Sets the value of the 'yydebug' variable to VALUE.
1939 This is a function so we don't have to have YYDEBUG defined
1940 in order to build the compiler. */
1949 warning ("YYDEBUG not defined.");