2 /* $NetBSD: scan.l,v 1.8 1995/10/23 13:38:51 jpo Exp $ */
5 * Copyright (c) 1994, 1995 Jochen Pohl
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by Jochen Pohl for
20 * 4. The name of the author may not be used to endorse or promote products
21 * derived from this software without specific prior written permission.
23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
24 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
25 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
26 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
27 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
28 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
29 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
30 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
31 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
32 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
34 * $NetBSD: scan.l,v 1.8 1995/10/23 13:38:51 jpo Exp $
49 #define CHAR_MASK (~(~0 << CHAR_BIT))
51 /* Current position (its also updated when an included file is parsed) */
52 pos_t curr_pos = { 1, "" };
55 * Current position in C source (not updated when an included file is
58 pos_t csrc_pos = { 1, "" };
60 static void incline __P((void));
61 static void badchar __P((int));
62 static sbuf_t *allocsb __P((void));
63 static void freesb __P((sbuf_t *));
64 static int inpc __P((void));
65 static int hash __P((const char *));
66 static sym_t *search __P((sbuf_t *));
67 static int name __P((void));
68 static int keyw __P((sym_t *));
69 static int icon __P((int));
70 static int fcon __P((void));
71 static int operator __P((int, op_t));
72 static int ccon __P((void));
73 static int wccon __P((void));
74 static int getescc __P((int));
75 static void directive __P((void));
76 static void comment __P((void));
77 static int string __P((void));
78 static int wcstrg __P((void));
91 {L}({L}|{D})* return (name());
92 0{OD}*[lLuU]* return (icon(8));
93 {NZD}{D}*[lLuU]* return (icon(10));
94 0[xX]{HD}+[lLuU]* return (icon(16));
95 {D}+\.{D}*{EX}?[fFlL]? |
97 \.{D}+{EX}?[fFlL]? return (fcon());
98 "=" return (operator(T_ASSIGN, ASSIGN));
99 "*=" return (operator(T_OPASS, MULASS));
100 "/=" return (operator(T_OPASS, DIVASS));
101 "%=" return (operator(T_OPASS, MODASS));
102 "+=" return (operator(T_OPASS, ADDASS));
103 "-=" return (operator(T_OPASS, SUBASS));
104 "<<=" return (operator(T_OPASS, SHLASS));
105 ">>=" return (operator(T_OPASS, SHRASS));
106 "&=" return (operator(T_OPASS, ANDASS));
107 "^=" return (operator(T_OPASS, XORASS));
108 "|=" return (operator(T_OPASS, ORASS));
109 "||" return (operator(T_LOGOR, LOGOR));
110 "&&" return (operator(T_LOGAND, LOGAND));
111 "|" return (operator(T_OR, OR));
112 "&" return (operator(T_AND, AND));
113 "^" return (operator(T_XOR, XOR));
114 "==" return (operator(T_EQOP, EQ));
115 "!=" return (operator(T_EQOP, NE));
116 "<" return (operator(T_RELOP, LT));
117 ">" return (operator(T_RELOP, GT));
118 "<=" return (operator(T_RELOP, LE));
119 ">=" return (operator(T_RELOP, GE));
120 "<<" return (operator(T_SHFTOP, SHL));
121 ">>" return (operator(T_SHFTOP, SHR));
122 "++" return (operator(T_INCDEC, INC));
123 "--" return (operator(T_INCDEC, DEC));
124 "->" return (operator(T_STROP, ARROW));
125 "." return (operator(T_STROP, POINT));
126 "+" return (operator(T_ADDOP, PLUS));
127 "-" return (operator(T_ADDOP, MINUS));
128 "*" return (operator(T_MULT, MULT));
129 "/" return (operator(T_DIVOP, DIV));
130 "%" return (operator(T_DIVOP, MOD));
131 "!" return (operator(T_UNOP, NOT));
132 "~" return (operator(T_UNOP, COMPL));
133 "\"" return (string());
134 "L\"" return (wcstrg());
136 "{" return (T_LBRACE);
137 "}" return (T_RBRACE);
138 "," return (T_COMMA);
139 ":" return (T_COLON);
140 "?" return (T_QUEST);
141 "[" return (T_LBRACK);
142 "]" return (T_RBRACK);
143 "(" return (T_LPARN);
144 ")" return (T_RPARN);
145 "..." return (T_ELLIPSE);
147 "L'" return (wccon());
152 . badchar(yytext[0]);
160 if (curr_pos.p_file == csrc_pos.p_file)
168 /* unknown character \%o */
174 * During initialisation they are written to the symbol table.
176 static struct kwtab {
177 const char *kw_name; /* keyword */
178 int kw_token; /* token returned by yylex() */
179 scl_t kw_scl; /* storage class if kw_token T_SCLASS */
180 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */
181 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
182 u_int kw_stdc : 1; /* STDC keyword */
183 u_int kw_gcc : 1; /* GCC keyword */
185 { "asm", T_ASM, 0, 0, 0, 0, 1 },
186 { "__asm", T_ASM, 0, 0, 0, 0, 0 },
187 { "__asm__", T_ASM, 0, 0, 0, 0, 0 },
188 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 },
189 { "break", T_BREAK, 0, 0, 0, 0, 0 },
190 { "case", T_CASE, 0, 0, 0, 0, 0 },
191 { "char", T_TYPE, 0, CHAR, 0, 0, 0 },
192 { "const", T_QUAL, 0, 0, CONST, 1, 0 },
193 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 },
194 { "__const", T_QUAL, 0, 0, CONST, 0, 0 },
195 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 },
196 { "default", T_DEFAULT, 0, 0, 0, 0, 0 },
197 { "do", T_DO, 0, 0, 0, 0, 0 },
198 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 },
199 { "else", T_ELSE, 0, 0, 0, 0, 0 },
200 { "enum", T_ENUM, 0, 0, 0, 0, 0 },
201 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 },
202 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 },
203 { "for", T_FOR, 0, 0, 0, 0, 0 },
204 { "goto", T_GOTO, 0, 0, 0, 0, 0 },
205 { "if", T_IF, 0, 0, 0, 0, 0 },
206 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 },
207 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 },
208 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 },
209 { "int", T_TYPE, 0, INT, 0, 0, 0 },
210 { "long", T_TYPE, 0, LONG, 0, 0, 0 },
211 { "register", T_SCLASS, REG, 0, 0, 0, 0 },
212 { "return", T_RETURN, 0, 0, 0, 0, 0 },
213 { "short", T_TYPE, 0, SHORT, 0, 0, 0 },
214 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 },
215 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 },
216 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 },
217 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 },
218 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 },
219 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 },
220 { "switch", T_SWITCH, 0, 0, 0, 0, 0 },
221 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 },
222 { "union", T_SOU, 0, UNION, 0, 0, 0 },
223 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 },
224 { "void", T_TYPE, 0, VOID, 0, 0, 0 },
225 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 },
226 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 },
227 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 },
228 { "while", T_WHILE, 0, 0, 0, 0, 0 },
229 { NULL, 0, 0, 0, 0, 0, 0 }
233 static sym_t *symtab[HSHSIZ1];
235 /* bit i of the entry with index i is set */
236 u_quad_t qbmasks[sizeof(u_quad_t) * CHAR_BIT];
238 /* least significant i bits are set in the entry with index i */
239 u_quad_t qlmasks[sizeof(u_quad_t) * CHAR_BIT + 1];
241 /* least significant i bits are not set in the entry with index i */
242 u_quad_t qumasks[sizeof(u_quad_t) * CHAR_BIT + 1];
244 /* free list for sbuf structures */
245 static sbuf_t *sbfrlst;
247 /* Typ of next expected symbol */
252 * All keywords are written to the symbol table. This saves us looking
253 * in a extra table for each name we found.
263 for (kw = kwtab; kw->kw_name != NULL; kw++) {
264 if (kw->kw_stdc && tflag)
266 if (kw->kw_gcc && !gflag)
268 sym = getblk(sizeof (sym_t));
269 sym->s_name = kw->kw_name;
271 sym->s_value.v_quad = kw->kw_token;
272 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
273 sym->s_tspec = kw->kw_tspec;
274 } else if (kw->kw_token == T_SCLASS) {
275 sym->s_scl = kw->kw_scl;
276 } else if (kw->kw_token == T_QUAL) {
277 sym->s_tqual = kw->kw_tqual;
279 h = hash(sym->s_name);
280 if ((sym->s_link = symtab[h]) != NULL)
281 symtab[h]->s_rlink = &sym->s_link;
282 (symtab[h] = sym)->s_rlink = &symtab[h];
285 /* initialize bit-masks for quads */
286 for (i = 0; i < sizeof (u_quad_t) * CHAR_BIT; i++) {
287 qbmasks[i] = (u_quad_t)1 << i;
288 uq = ~(u_quad_t)0 << i;
293 qlmasks[i] = ~(u_quad_t)0;
297 * Get a free sbuf structure, if possible from the free list
304 if ((sb = sbfrlst) != NULL) {
305 sbfrlst = sb->sb_nxt;
307 sb = xmalloc(sizeof (sbuf_t));
309 (void)memset(sb, 0, sizeof (sb));
314 * Put a sbuf structure to the free list
320 sb->sb_nxt = sbfrlst;
325 * Read a character and ensure that it is positive (except EOF).
326 * Increment line count(s) if necessary.
333 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
346 for (us = (const u_char *)s; *us != '\0'; us++) {
347 v = (v << sizeof (v)) + *us;
348 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
350 return (v % HSHSIZ1);
354 * Lex has found a letter followed by zero or more letters or digits.
355 * It looks for a symbol in the symbol table with the same name. This
356 * symbol must either be a keyword or a symbol of the type required by
357 * symtyp (label, member, tag, ...).
359 * If it is a keyword, the token is returned. In some cases it is described
360 * more deeply by data written to yylval.
362 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
363 * is stored in yylval. This struct contains the name of the symbol, it's
364 * length and hash value. If there is already a symbol of the same name
365 * and type in the symbol table, the sbuf struct also contains a pointer
366 * to the symbol table entry.
377 sb->sb_name = yytext;
379 sb->sb_hash = hash(yytext);
381 if ((sym = search(sb)) != NULL && sym->s_keyw) {
389 if (blklev < sym->s_blklev)
391 sb->sb_name = sym->s_name;
392 sb->sb_len = strlen(sym->s_name);
393 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
395 s = getblk(yyleng + 1);
396 (void)memcpy(s, yytext, yyleng + 1);
412 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
413 if (strcmp(sym->s_name, sb->sb_name) == 0) {
414 if (sym->s_keyw || sym->s_kind == symtyp)
428 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
429 yylval.y_scl = sym->s_scl;
430 } else if (t == T_TYPE || t == T_SOU) {
431 yylval.y_tspec = sym->s_tspec;
432 } else if (t == T_QUAL) {
433 yylval.y_tqual = sym->s_tqual;
439 * Convert a string representing an integer into internal representation.
440 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
446 int l_suffix, u_suffix;
454 static tspec_t contypes[2][3] = {
456 { UINT, ULONG, UQUAD }
469 l_suffix = u_suffix = 0;
471 if ((c = cp[len - 1]) == 'l' || c == 'L') {
473 } else if (c == 'u' || c == 'U') {
480 if (l_suffix > 2 || u_suffix > 1) {
481 /* malformed integer constant */
488 if (tflag && u_suffix != 0) {
489 /* suffix U is illegal in traditional C */
492 typ = contypes[u_suffix][l_suffix];
496 ul = strtoul(cp, &eptr, base);
498 uq = strtouq(cp, &eptr, base);
500 if (eptr != cp + len)
503 /* integer constant out of range */
507 * If the value is to big for the current type, we must choose
515 } else if (ul <= (unsigned)UINT_MAX && base != 10) {
517 } else if (ul <= LONG_MAX) {
522 if (typ == UINT || typ == ULONG) {
527 * Remember that the constant is unsigned
535 if (ul > (u_int)UINT_MAX)
539 if (ul > LONG_MAX && !tflag) {
546 if (uq > QUAD_MAX && !tflag) {
552 /* LINTED (enumeration values not handled in switch) */
556 if (typ != QUAD && typ != UQUAD) {
560 uq = (quad_t)(long)ul;
564 uq = (u_quad_t)xsign((quad_t)uq, typ, -1);
566 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
567 yylval.y_val->v_ansiu = ansiu;
568 yylval.y_val->v_quad = (quad_t)uq;
574 * Returns 1 if t is a signed type and the value is negative.
576 * len is the number of significant bits. If len is -1, len is set
577 * to the width of type t.
585 if (t == PTR || isutyp(t))
587 return (msb(q, t, len));
598 return ((q & qbmasks[len - 1]) != 0);
602 * Extends the sign of q.
613 if (t == PTR || isutyp(t) || !sign(q, t, len)) {
622 * Convert a string representing a floating point value into its interal
623 * representation. Type and value are returned in yylval. fcon()
624 * (and yylex()) returns T_CON.
625 * XXX Currently it is not possible to convert constants of type
626 * long double which are greater then DBL_MAX.
641 if ((c = cp[len - 1]) == 'f' || c == 'F') {
644 } else if (c == 'l' || c == 'L') {
651 if (tflag && typ != DOUBLE) {
652 /* suffixes F and L are illegal in traditional C */
657 d = strtod(cp, &eptr);
658 if (eptr != cp + len)
661 /* floating-point constant out of range */
667 /* floating-point constant out of range */
669 f = f > 0 ? FLT_MAX : -FLT_MAX;
673 (yylval.y_val = xcalloc(1, sizeof (val_t)))->v_tspec = typ;
675 yylval.y_val->v_ldbl = f;
677 yylval.y_val->v_ldbl = d;
693 * Called if lex found a leading \'.
703 while ((c = getescc('\'')) >= 0) {
704 val = (val << CHAR_BIT) + c;
708 /* unterminated character constant */
711 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
712 /* too many characters in character constant */
715 /* multi-character character constant */
718 /* empty character constant */
727 yylval.y_val = xcalloc(1, sizeof (val_t));
728 yylval.y_val->v_tspec = INT;
729 yylval.y_val->v_quad = val;
735 * Called if lex found a leading L\'
740 static char buf[MB_LEN_MAX + 1];
745 while ((c = getescc('\'')) >= 0) {
754 /* unterminated character constant */
757 /* empty character constant */
760 if (i > MB_CUR_MAX) {
762 /* too many characters in character constant */
766 (void)mbtowc(NULL, NULL, 0);
767 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
768 /* invalid multibyte character */
773 yylval.y_val = xcalloc(1, sizeof (val_t));
774 yylval.y_val->v_tspec = WCHAR;
775 yylval.y_val->v_quad = wc;
781 * Read a character which is part of a character constant or of a string
782 * and handle escapes.
784 * The Argument is the character which delimits the character constant or
787 * Returns -1 if the end of the character constant or string is reached,
788 * -2 if the EOF is reached, and the charachter otherwise.
807 /* newline in string or char constant */
813 switch (c = inpc()) {
815 if (tflag && d == '\'')
816 /* \" inside character constant undef. ... */
823 /* \? undefined in traditional C */
830 /* \a undefined in traditional C */
849 /* \v undefined in traditional C */
857 /* bad octal digit %c */
860 case '0': case '1': case '2': case '3':
861 case '4': case '5': case '6': case '7':
865 v = (v << 3) + (c - '0');
867 } while (--n && isdigit(c) && (tflag || c <= '7'));
868 if (tflag && n > 0 && isdigit(c))
869 /* bad octal digit %c */
873 /* character escape does not fit in char. */
880 /* \x undefined in traditional C */
884 while ((c = inpc()) >= 0 && isxdigit(c)) {
886 c - '0' : toupper(c) - 'A' + 10;
889 if ((v & ~CHAR_MASK) != 0) {
890 /* overflow in hex escape */
900 /* no hex digits follow \x */
912 /* dubious escape \%c */
915 /* dubious escape \%o */
924 * Called for preprocessor directives. Currently implemented are:
926 * # lineno "filename"
935 static int first = 1;
937 /* Go to first non-whitespace after # */
938 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++) ;
942 /* undefined or invalid # directive */
946 ln = strtol(--cp, &eptr, 10);
949 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
951 while ((c = *cp++) == ' ' || c == '\t') ;
956 while ((c = *cp) != '"' && c != '\0')
960 if ((fnl = cp++ - fn) > PATH_MAX)
962 while ((c = *cp++) == ' ' || c == '\t') ;
965 warning("extra character(s) after directive");
967 curr_pos.p_file = fnnalloc(fn, fnl);
969 * If this is the first directive, the name is the name
970 * of the C source file as specified at the command line.
971 * It is written to the output file.
974 csrc_pos.p_file = curr_pos.p_file;
975 outsrc(curr_pos.p_file);
979 curr_pos.p_line = (int)ln - 1;
980 if (curr_pos.p_file == csrc_pos.p_file)
981 csrc_pos.p_line = (int)ln - 1;
985 * Handle lint comments. Following comments are currently understood:
987 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
988 * FALLTHRU FALLTHROUGH
997 * If one of this comments is recognized, the arguments, if any, are
998 * parsed and a function which handles this comment is called.
1007 void (*func) __P((int));
1009 { "ARGSUSED", 1, argsused },
1010 { "CONSTCOND", 0, constcond },
1011 { "CONSTANTCOND", 0, constcond },
1012 { "CONSTANTCONDITION", 0, constcond },
1013 { "FALLTHRU", 0, fallthru },
1014 { "FALLTHROUGH", 0, fallthru },
1015 { "LINTLIBRARY", 0, lintlib },
1016 { "LINTED", 0, linted },
1017 { "LONGLONG", 0, longlong },
1018 { "NOSTRICT", 0, linted },
1019 { "NOTREACHED", 0, notreach },
1020 { "PRINTFLIKE", 1, printflike },
1021 { "PROTOLIB", 1, protolib },
1022 { "SCANFLIKE", 1, scanflike },
1023 { "VARARGS", 1, varargs },
1032 /* Skip white spaces after the start of the comment */
1033 while ((c = inpc()) != EOF && isspace(c)) ;
1035 /* Read the potential keyword to keywd */
1037 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1038 keywd[l++] = (char)c;
1043 /* look for the keyword */
1044 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1045 if (strcmp(keywtab[i].keywd, keywd) == 0)
1048 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1051 /* skip white spaces after the keyword */
1052 while (c != EOF && isspace(c))
1055 /* read the argument, if the keyword accepts one and there is one */
1057 if (keywtab[i].arg) {
1058 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1064 a = l != 0 ? atoi(arg) : -1;
1066 /* skip white spaces after the argument */
1067 while (c != EOF && isspace(c))
1070 if (c != '*' || (c = inpc()) != '/') {
1071 if (keywtab[i].func != linted)
1072 /* extra characters in lint comment */
1076 * remember that we have already found the end of the
1082 if (keywtab[i].func != NULL)
1083 (*keywtab[i].func)(a);
1088 if ((c = inpc()) == EOF) {
1089 /* unterminated comment */
1093 if (lc == '*' && c == '/')
1099 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1100 * clrwflgs() is called after function definitions and global and
1101 * local declarations and definitions. It is also called between
1102 * the controlling expression and the body of control statements
1103 * (if, switch, for, while).
1114 * Strings are stored in a dynamically alloceted buffer and passed
1115 * in yylval.y_xstrg to the parser. The parser or the routines called
1116 * by the parser are responsible for freeing this buffer.
1126 s = xmalloc(max = 64);
1129 while ((c = getescc('"')) >= 0) {
1130 /* +1 to reserve space for a trailing NUL character */
1132 s = xrealloc(s, max *= 2);
1137 /* unterminated string constant */
1140 strg = xcalloc(1, sizeof (strg_t));
1141 strg->st_tspec = CHAR;
1145 yylval.y_strg = strg;
1154 size_t len, max, wlen;
1158 s = xmalloc(max = 64);
1160 while ((c = getescc('"')) >= 0) {
1161 /* +1 to save space for a trailing NUL character */
1163 s = xrealloc(s, max *= 2);
1168 /* unterminated string constant */
1171 /* get length of wide character string */
1172 (void)mblen(NULL, 0);
1173 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1174 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1175 /* invalid multibyte character */
1183 ws = xmalloc((wlen + 1) * sizeof (wchar_t));
1185 /* convert from multibyte to wide char */
1186 (void)mbtowc(NULL, NULL, 0);
1187 for (i = 0, wi = 0; i < len; i += n, wi++) {
1188 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1196 strg = xcalloc(1, sizeof (strg_t));
1197 strg->st_tspec = WCHAR;
1198 strg->st_len = wlen;
1201 yylval.y_strg = strg;
1206 * As noted above the scanner does not create new symbol table entries
1207 * for symbols it cannot find in the symbol table. This is to avoid
1208 * putting undeclared symbols into the symbol table if a syntax error
1211 * getsym() is called as soon as it is probably ok to put the symbol to
1212 * the symbol table. This does not mean that it is not possible that
1213 * symbols are put to the symbol table which are than not completely
1214 * declared due to syntax errors. To avoid too many problems in this
1215 * case symbols get type int in getsym().
1217 * XXX calls to getsym() should be delayed until decl1*() is called
1230 * During member declaration it is possible that name() looked
1231 * for symbols of type FVFT, although it should have looked for
1232 * symbols of type FTAG. Same can happen for labels. Both cases
1233 * are compensated here.
1235 if (symtyp == FMOS || symtyp == FLAB) {
1236 if (sym == NULL || sym->s_kind == FVFT)
1241 if (sym->s_kind != symtyp)
1242 lerror("storesym() 1");
1248 /* create a new symbol table entry */
1250 /* labels must always be allocated at level 1 (outhermost block) */
1251 if (symtyp == FLAB) {
1252 sym = getlblk(1, sizeof (sym_t));
1253 s = getlblk(1, sb->sb_len + 1);
1254 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1258 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1260 if (di->d_ctx != AUTO)
1261 lerror("storesym() 2");
1263 sym = getblk(sizeof (sym_t));
1264 sym->s_name = sb->sb_name;
1265 sym->s_blklev = blklev;
1269 STRUCT_ASSIGN(sym->s_dpos, curr_pos);
1270 if ((sym->s_kind = symtyp) != FLAB)
1271 sym->s_type = gettyp(INT);
1275 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1276 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1277 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash];
1279 *di->d_ldlsym = sym;
1280 di->d_ldlsym = &sym->s_dlnxt;
1287 * Remove a symbol forever from the symbol table. s_blklev
1288 * is set to -1 to avoid that the symbol will later be put
1289 * back to the symbol table.
1295 if ((*sym->s_rlink = sym->s_link) != NULL)
1296 sym->s_link->s_rlink = sym->s_rlink;
1302 * Remove a list of symbols declared at one level from the symbol
1311 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1312 if (sym->s_blklev != -1) {
1313 if ((*sym->s_rlink = sym->s_link) != NULL)
1314 sym->s_link->s_rlink = sym->s_rlink;
1316 sym->s_rlink = NULL;
1322 * Put a symbol into the symbol table
1331 h = hash(sym->s_name);
1332 if ((sym->s_link = symtab[h]) != NULL)
1333 symtab[h]->s_rlink = &sym->s_link;
1334 (symtab[h] = sym)->s_rlink = &symtab[h];
1336 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1341 * Called at level 0 after syntax errors
1342 * Removes all symbols which are not declared at level 0 from the
1343 * symbol table. Also frees all memory which is not associated with
1352 for (i = 0; i < HSHSIZ1; i++) {
1353 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1355 if (sym->s_blklev >= 1) {
1356 if ((*sym->s_rlink = nsym) != NULL)
1357 nsym->s_rlink = sym->s_rlink;
1362 for (i = mblklev; i > 0; i--)
1367 * Create a new symbol with the name of an existing symbol.
1376 h = hash(sym->s_name);
1377 nsym = getblk(sizeof (sym_t));
1378 if (sym->s_blklev > blklev)
1379 lerror("pushdown()");
1380 nsym->s_name = sym->s_name;
1381 STRUCT_ASSIGN(nsym->s_dpos, curr_pos);
1382 nsym->s_kind = sym->s_kind;
1383 nsym->s_blklev = blklev;
1385 if ((nsym->s_link = symtab[h]) != NULL)
1386 symtab[h]->s_rlink = &nsym->s_link;
1387 (symtab[h] = nsym)->s_rlink = &symtab[h];
1389 *dcs->d_ldlsym = nsym;
1390 dcs->d_ldlsym = &nsym->s_dlnxt;
1396 * Free any dynamically allocated memory referenced by
1397 * the value stack or yylval.
1398 * The type of information in yylval is described by tok.
1405 if (tok == T_NAME || tok == T_TYPENAME) {
1406 sbuf_t *sb = *(sbuf_t **)sp;
1408 } else if (tok == T_CON) {
1409 val_t *val = *(val_t **)sp;
1411 } else if (tok == T_STRING) {
1412 strg_t *strg = *(strg_t **)sp;
1413 if (strg->st_tspec == CHAR) {
1415 } else if (strg->st_tspec == WCHAR) {
1418 lerror("fryylv() 1");