2 /* $NetBSD: scan.l,v 1.26 2002/01/31 22:30:21 tv Exp $ */
5 * Copyright (c) 1996 Christopher G. Demetriou. All Rights Reserved.
6 * Copyright (c) 1994, 1995 Jochen Pohl
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. All advertising materials mentioning features or use of this software
18 * must display the following acknowledgement:
19 * This product includes software developed by Jochen Pohl for
21 * 4. The name of the author may not be used to endorse or promote products
22 * derived from this software without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
25 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
26 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
27 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
28 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
29 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
30 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
31 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
32 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
33 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
36 #include <sys/cdefs.h>
37 #if defined(__RCSID) && !defined(lint)
38 __RCSID("$NetBSD: scan.l,v 1.26 2002/01/31 22:30:21 tv Exp $");
40 __FBSDID("$FreeBSD$");
54 #define CHAR_MASK (~(~0 << CHAR_BIT))
57 /* Current position (its also updated when an included file is parsed) */
58 pos_t curr_pos = { 1, "", 0 };
61 * Current position in C source (not updated when an included file is
64 pos_t csrc_pos = { 1, "", 0 };
66 static void incline(void);
67 static void badchar(int);
68 static sbuf_t *allocsb(void);
69 static void freesb(sbuf_t *);
70 static int inpc(void);
71 static int hash(const char *);
72 static sym_t *search(sbuf_t *);
73 static int name(void);
74 static int keyw(sym_t *);
76 static int fcon(void);
77 static int operator(int, op_t);
78 static int ccon(void);
79 static int wccon(void);
80 static int getescc(int);
81 static void directive(void);
82 static void comment(void);
83 static void slashslashcomment(void);
84 static int string(void);
85 static int wcstrg(void);
98 {L}({L}|{D})* return (name());
99 0{OD}*[lLuU]* return (icon(8));
100 {NZD}{D}*[lLuU]* return (icon(10));
101 0[xX]{HD}+[lLuU]* return (icon(16));
102 {D}+\.{D}*{EX}?[fFlL]? |
104 \.{D}+{EX}?[fFlL]? return (fcon());
105 "=" return (operator(T_ASSIGN, ASSIGN));
106 "*=" return (operator(T_OPASS, MULASS));
107 "/=" return (operator(T_OPASS, DIVASS));
108 "%=" return (operator(T_OPASS, MODASS));
109 "+=" return (operator(T_OPASS, ADDASS));
110 "-=" return (operator(T_OPASS, SUBASS));
111 "<<=" return (operator(T_OPASS, SHLASS));
112 ">>=" return (operator(T_OPASS, SHRASS));
113 "&=" return (operator(T_OPASS, ANDASS));
114 "^=" return (operator(T_OPASS, XORASS));
115 "|=" return (operator(T_OPASS, ORASS));
116 "||" return (operator(T_LOGOR, LOGOR));
117 "&&" return (operator(T_LOGAND, LOGAND));
118 "|" return (operator(T_OR, OR));
119 "&" return (operator(T_AND, AND));
120 "^" return (operator(T_XOR, XOR));
121 "==" return (operator(T_EQOP, EQ));
122 "!=" return (operator(T_EQOP, NE));
123 "<" return (operator(T_RELOP, LT));
124 ">" return (operator(T_RELOP, GT));
125 "<=" return (operator(T_RELOP, LE));
126 ">=" return (operator(T_RELOP, GE));
127 "<<" return (operator(T_SHFTOP, SHL));
128 ">>" return (operator(T_SHFTOP, SHR));
129 "++" return (operator(T_INCDEC, INC));
130 "--" return (operator(T_INCDEC, DEC));
131 "->" return (operator(T_STROP, ARROW));
132 "." return (operator(T_STROP, POINT));
133 "+" return (operator(T_ADDOP, PLUS));
134 "-" return (operator(T_ADDOP, MINUS));
135 "*" return (operator(T_MULT, MULT));
136 "/" return (operator(T_DIVOP, DIV));
137 "%" return (operator(T_DIVOP, MOD));
138 "!" return (operator(T_UNOP, NOT));
139 "~" return (operator(T_UNOP, COMPL));
140 "\"" return (string());
141 "L\"" return (wcstrg());
143 "{" return (T_LBRACE);
144 "}" return (T_RBRACE);
145 "," return (T_COMMA);
146 ":" return (T_COLON);
147 "?" return (T_QUEST);
148 "[" return (T_LBRACK);
149 "]" return (T_RBRACK);
150 "(" return (T_LPARN);
151 ")" return (T_RPARN);
152 "..." return (T_ELLIPSE);
154 "L'" return (wccon());
159 "//" slashslashcomment();
160 . badchar(yytext[0]);
169 if (curr_pos.p_file == csrc_pos.p_file) {
179 /* unknown character \%o */
185 * During initialisation they are written to the symbol table.
187 static struct kwtab {
188 const char *kw_name; /* keyword */
189 int kw_token; /* token returned by yylex() */
190 scl_t kw_scl; /* storage class if kw_token T_SCLASS */
191 tspec_t kw_tspec; /* type spec. if kw_token T_TYPE or T_SOU */
192 tqual_t kw_tqual; /* type qual. fi kw_token T_QUAL */
193 u_int kw_stdc : 1; /* STDC keyword */
194 u_int kw_gcc : 1; /* GCC keyword */
196 { "asm", T_ASM, 0, 0, 0, 0, 1 },
197 { "__asm", T_ASM, 0, 0, 0, 0, 0 },
198 { "__asm__", T_ASM, 0, 0, 0, 0, 0 },
199 { "auto", T_SCLASS, AUTO, 0, 0, 0, 0 },
200 { "break", T_BREAK, 0, 0, 0, 0, 0 },
201 { "case", T_CASE, 0, 0, 0, 0, 0 },
202 { "char", T_TYPE, 0, CHAR, 0, 0, 0 },
203 { "const", T_QUAL, 0, 0, CONST, 1, 0 },
204 { "__const__", T_QUAL, 0, 0, CONST, 0, 0 },
205 { "__const", T_QUAL, 0, 0, CONST, 0, 0 },
206 { "continue", T_CONTINUE, 0, 0, 0, 0, 0 },
207 { "default", T_DEFAULT, 0, 0, 0, 0, 0 },
208 { "do", T_DO, 0, 0, 0, 0, 0 },
209 { "double", T_TYPE, 0, DOUBLE, 0, 0, 0 },
210 { "else", T_ELSE, 0, 0, 0, 0, 0 },
211 { "enum", T_ENUM, 0, 0, 0, 0, 0 },
212 { "extern", T_SCLASS, EXTERN, 0, 0, 0, 0 },
213 { "float", T_TYPE, 0, FLOAT, 0, 0, 0 },
214 { "for", T_FOR, 0, 0, 0, 0, 0 },
215 { "goto", T_GOTO, 0, 0, 0, 0, 0 },
216 { "if", T_IF, 0, 0, 0, 0, 0 },
217 { "inline", T_SCLASS, INLINE, 0, 0, 0, 1 },
218 { "__inline__", T_SCLASS, INLINE, 0, 0, 0, 0 },
219 { "__inline", T_SCLASS, INLINE, 0, 0, 0, 0 },
220 { "int", T_TYPE, 0, INT, 0, 0, 0 },
221 { "__symbolrename", T_SYMBOLRENAME, 0, 0, 0, 0, 0 },
222 { "long", T_TYPE, 0, LONG, 0, 0, 0 },
223 { "register", T_SCLASS, REG, 0, 0, 0, 0 },
224 { "return", T_RETURN, 0, 0, 0, 0, 0 },
225 { "short", T_TYPE, 0, SHORT, 0, 0, 0 },
226 { "signed", T_TYPE, 0, SIGNED, 0, 1, 0 },
227 { "__signed__", T_TYPE, 0, SIGNED, 0, 0, 0 },
228 { "__signed", T_TYPE, 0, SIGNED, 0, 0, 0 },
229 { "sizeof", T_SIZEOF, 0, 0, 0, 0, 0 },
230 { "static", T_SCLASS, STATIC, 0, 0, 0, 0 },
231 { "struct", T_SOU, 0, STRUCT, 0, 0, 0 },
232 { "switch", T_SWITCH, 0, 0, 0, 0, 0 },
233 { "typedef", T_SCLASS, TYPEDEF, 0, 0, 0, 0 },
234 { "union", T_SOU, 0, UNION, 0, 0, 0 },
235 { "unsigned", T_TYPE, 0, UNSIGN, 0, 0, 0 },
236 { "void", T_TYPE, 0, VOID, 0, 0, 0 },
237 { "volatile", T_QUAL, 0, 0, VOLATILE, 1, 0 },
238 { "__volatile__", T_QUAL, 0, 0, VOLATILE, 0, 0 },
239 { "__volatile", T_QUAL, 0, 0, VOLATILE, 0, 0 },
240 { "while", T_WHILE, 0, 0, 0, 0, 0 },
241 { NULL, 0, 0, 0, 0, 0, 0 }
245 static sym_t *symtab[HSHSIZ1];
247 /* bit i of the entry with index i is set */
248 uint64_t qbmasks[sizeof(uint64_t) * CHAR_BIT];
250 /* least significant i bits are set in the entry with index i */
251 uint64_t qlmasks[sizeof(uint64_t) * CHAR_BIT + 1];
253 /* least significant i bits are not set in the entry with index i */
254 uint64_t qumasks[sizeof(uint64_t) * CHAR_BIT + 1];
256 /* free list for sbuf structures */
257 static sbuf_t *sbfrlst;
259 /* Typ of next expected symbol */
264 * All keywords are written to the symbol table. This saves us looking
265 * in an extra table for each name we found.
275 for (kw = kwtab; kw->kw_name != NULL; kw++) {
276 if (kw->kw_stdc && tflag)
278 if (kw->kw_gcc && !gflag)
280 sym = getblk(sizeof (sym_t));
281 sym->s_name = kw->kw_name;
283 sym->s_value.v_quad = kw->kw_token;
284 if (kw->kw_token == T_TYPE || kw->kw_token == T_SOU) {
285 sym->s_tspec = kw->kw_tspec;
286 } else if (kw->kw_token == T_SCLASS) {
287 sym->s_scl = kw->kw_scl;
288 } else if (kw->kw_token == T_QUAL) {
289 sym->s_tqual = kw->kw_tqual;
291 h = hash(sym->s_name);
292 if ((sym->s_link = symtab[h]) != NULL)
293 symtab[h]->s_rlink = &sym->s_link;
294 (symtab[h] = sym)->s_rlink = &symtab[h];
297 /* initialize bit-masks for quads */
298 for (i = 0; i < sizeof (uint64_t) * CHAR_BIT; i++) {
299 qbmasks[i] = (uint64_t)1 << i;
300 uq = ~(uint64_t)0 << i;
305 qlmasks[i] = ~(uint64_t)0;
309 * Get a free sbuf structure, if possible from the free list
316 if ((sb = sbfrlst) != NULL) {
317 sbfrlst = sb->sb_nxt;
319 if ((sb = malloc(sizeof (sbuf_t))) == NULL)
322 (void)memset(sb, 0, sizeof (*sb));
327 * Put a sbuf structure to the free list
333 sb->sb_nxt = sbfrlst;
338 * Read a character and ensure that it is positive (except EOF).
339 * Increment line count(s) if necessary.
346 if ((c = input()) != EOF && (c &= CHAR_MASK) == '\n')
358 for (us = (const u_char *)s; *us != '\0'; us++) {
359 v = (v << sizeof (v)) + *us;
360 v ^= v >> (sizeof (v) * CHAR_BIT - sizeof (v));
362 return (v % HSHSIZ1);
366 * Lex has found a letter followed by zero or more letters or digits.
367 * It looks for a symbol in the symbol table with the same name. This
368 * symbol must either be a keyword or a symbol of the type required by
369 * symtyp (label, member, tag, ...).
371 * If it is a keyword, the token is returned. In some cases it is described
372 * more deeply by data written to yylval.
374 * If it is a symbol, T_NAME is returned and the pointer to a sbuf struct
375 * is stored in yylval. This struct contains the name of the symbol, it's
376 * length and hash value. If there is already a symbol of the same name
377 * and type in the symbol table, the sbuf struct also contains a pointer
378 * to the symbol table entry.
389 sb->sb_name = yytext;
391 sb->sb_hash = hash(yytext);
393 if ((sym = search(sb)) != NULL && sym->s_keyw) {
401 if (blklev < sym->s_blklev)
403 sb->sb_name = sym->s_name;
404 sb->sb_len = strlen(sym->s_name);
405 tok = sym->s_scl == TYPEDEF ? T_TYPENAME : T_NAME;
407 s = getblk(yyleng + 1);
408 (void)memcpy(s, yytext, yyleng + 1);
423 for (sym = symtab[sb->sb_hash]; sym != NULL; sym = sym->s_link) {
424 if (strcmp(sym->s_name, sb->sb_name) == 0) {
425 if (sym->s_keyw || sym->s_kind == symtyp)
438 if ((t = (int)sym->s_value.v_quad) == T_SCLASS) {
439 yylval.y_scl = sym->s_scl;
440 } else if (t == T_TYPE || t == T_SOU) {
441 yylval.y_tspec = sym->s_tspec;
442 } else if (t == T_QUAL) {
443 yylval.y_tqual = sym->s_tqual;
449 * Convert a string representing an integer into internal representation.
450 * The value is returned in yylval. icon() (and yylex()) returns T_CON.
455 int l_suffix, u_suffix;
463 static tspec_t contypes[2][3] = {
465 { UINT, ULONG, UQUAD }
478 l_suffix = u_suffix = 0;
480 if ((c = cp[len - 1]) == 'l' || c == 'L') {
482 } else if (c == 'u' || c == 'U') {
489 if (l_suffix > 2 || u_suffix > 1) {
490 /* malformed integer constant */
497 if (tflag && u_suffix != 0) {
498 /* suffix U is illegal in traditional C */
501 typ = contypes[u_suffix][l_suffix];
505 ul = strtoul(cp, &eptr, base);
507 uq = strtouq(cp, &eptr, base);
509 if (eptr != cp + len)
512 /* integer constant out of range */
516 * If the value is to big for the current type, we must choose
524 } else if (ul <= (unsigned)UINT_MAX && base != 10) {
526 } else if (ul <= LONG_MAX) {
531 if (typ == UINT || typ == ULONG) {
536 * Remember that the constant is unsigned
544 if (ul > (u_int)UINT_MAX)
548 if (ul > LONG_MAX && !tflag) {
555 if (uq > QUAD_MAX && !tflag) {
561 /* LINTED (enumeration values not handled in switch) */
585 if (typ != QUAD && typ != UQUAD) {
589 uq = (int64_t)(long)ul;
593 uq = (uint64_t)xsign((int64_t)uq, typ, -1);
595 if ((yylval.y_val = calloc(1, sizeof(val_t))) == NULL)
597 yylval.y_val->v_tspec = typ;
598 yylval.y_val->v_ansiu = ansiu;
599 yylval.y_val->v_quad = (int64_t)uq;
605 * Returns 1 if t is a signed type and the value is negative.
607 * len is the number of significant bits. If len is -1, len is set
608 * to the width of type t.
611 sign(int64_t q, tspec_t t, int len)
614 if (t == PTR || isutyp(t))
616 return (msb(q, t, len));
620 msb(int64_t q, tspec_t t, int len)
625 return ((q & qbmasks[len - 1]) != 0);
629 * Extends the sign of q.
632 xsign(int64_t q, tspec_t t, int len)
638 if (t == PTR || isutyp(t) || !sign(q, t, len)) {
647 * Convert a string representing a floating point value into its interal
648 * representation. Type and value are returned in yylval. fcon()
649 * (and yylex()) returns T_CON.
650 * XXX Currently it is not possible to convert constants of type
651 * long double which are greater than DBL_MAX.
666 if ((c = cp[len - 1]) == 'f' || c == 'F') {
669 } else if (c == 'l' || c == 'L') {
676 if (tflag && typ != DOUBLE) {
677 /* suffixes F and L are illegal in traditional C */
682 d = strtod(cp, &eptr);
683 if (eptr != cp + len)
686 /* floating-point constant out of range */
692 /* floating-point constant out of range */
694 f = f > 0 ? FLT_MAX : -FLT_MAX;
698 if ((yylval.y_val = calloc(1, sizeof (val_t))) == NULL)
700 yylval.y_val->v_tspec = typ;
702 yylval.y_val->v_ldbl = f;
704 yylval.y_val->v_ldbl = d;
711 operator(int t, op_t o)
719 * Called if lex found a leading \'.
729 while ((c = getescc('\'')) >= 0) {
730 val = (val << CHAR_BIT) + c;
734 /* unterminated character constant */
737 if (n > sizeof (int) || (n > 1 && (pflag || hflag))) {
738 /* too many characters in character constant */
741 /* multi-character character constant */
744 /* empty character constant */
753 yylval.y_val = xcalloc(1, sizeof (val_t));
754 yylval.y_val->v_tspec = INT;
755 yylval.y_val->v_quad = val;
761 * Called if lex found a leading L\'
766 static char buf[MB_LEN_MAX + 1];
771 while ((c = getescc('\'')) >= 0) {
780 /* unterminated character constant */
783 /* empty character constant */
786 if (i > MB_CUR_MAX) {
788 /* too many characters in character constant */
792 (void)mbtowc(NULL, NULL, 0);
793 if (mbtowc(&wc, buf, MB_CUR_MAX) < 0)
794 /* invalid multibyte character */
799 if ((yylval.y_val = calloc(1, sizeof (val_t))) == NULL)
801 yylval.y_val->v_tspec = WCHAR;
802 yylval.y_val->v_quad = wc;
808 * Read a character which is part of a character constant or of a string
809 * and handle escapes.
811 * The Argument is the character which delimits the character constant or
814 * Returns -1 if the end of the character constant or string is reached,
815 * -2 if the EOF is reached, and the character otherwise.
834 /* newline in string or char constant */
842 switch (c = inpc()) {
844 if (tflag && d == '\'')
845 /* \" inside character constant undef. ... */
852 /* \? undefined in traditional C */
859 /* \a undefined in traditional C */
874 /* \v undefined in traditional C */
878 /* bad octal digit %c */
881 case '0': case '1': case '2': case '3':
882 case '4': case '5': case '6': case '7':
886 v = (v << 3) + (c - '0');
888 } while (--n && isdigit(c) && (tflag || c <= '7'));
889 if (tflag && n > 0 && isdigit(c))
890 /* bad octal digit %c */
894 /* character escape does not fit in char. */
901 /* \x undefined in traditional C */
905 while ((c = inpc()) >= 0 && isxdigit(c)) {
907 c - '0' : toupper(c) - 'A' + 10;
910 if ((v & ~CHAR_MASK) != 0) {
911 /* overflow in hex escape */
921 /* no hex digits follow \x */
933 /* dubious escape \%c */
936 /* dubious escape \%o */
945 * Called for preprocessor directives. Currently implemented are:
947 * # lineno "filename"
956 static int first = 1;
958 /* Go to first non-whitespace after # */
959 for (cp = yytext + 1; (c = *cp) == ' ' || c == '\t'; cp++)
962 if (!isdigit((unsigned char)c)) {
964 /* undefined or invalid # directive */
968 ln = strtol(--cp, &eptr, 10);
971 if ((c = *(cp = eptr)) != ' ' && c != '\t' && c != '\0')
973 while ((c = *cp++) == ' ' || c == '\t')
979 while ((c = *cp) != '"' && c != '\0')
983 if ((fnl = cp++ - fn) > PATH_MAX)
985 while ((c = *cp++) == ' ' || c == '\t')
989 warning("extra character(s) after directive");
992 /* empty string means stdin */
994 fn = "{standard input}";
995 fnl = 16; /* strlen (fn) */
997 curr_pos.p_file = fnnalloc(fn, fnl);
999 * If this is the first directive, the name is the name
1000 * of the C source file as specified at the command line.
1001 * It is written to the output file.
1004 csrc_pos.p_file = curr_pos.p_file;
1005 outsrc(curr_pos.p_file);
1009 curr_pos.p_line = (int)ln - 1;
1010 curr_pos.p_uniq = 0;
1011 if (curr_pos.p_file == csrc_pos.p_file) {
1012 csrc_pos.p_line = (int)ln - 1;
1013 csrc_pos.p_uniq = 0;
1018 * Handle lint comments. Following comments are currently understood:
1021 * CONSTCOND CONSTANTCOND CONSTANTCONDITION
1022 * FALLTHRU FALLTHROUGH
1031 * If one of this comments is recognized, the arguments, if any, are
1032 * parsed and a function which handles this comment is called.
1043 { "ARGSUSED", 1, argsused },
1044 { "BITFIELDTYPE", 0, bitfieldtype },
1045 { "CONSTCOND", 0, constcond },
1046 { "CONSTANTCOND", 0, constcond },
1047 { "CONSTANTCONDITION", 0, constcond },
1048 { "FALLTHRU", 0, fallthru },
1049 { "FALLTHROUGH", 0, fallthru },
1050 { "LINTLIBRARY", 0, lintlib },
1051 { "LINTED", 0, linted },
1052 { "LONGLONG", 0, longlong },
1053 { "NOSTRICT", 0, linted },
1054 { "NOTREACHED", 0, notreach },
1055 { "PRINTFLIKE", 1, printflike },
1056 { "PROTOLIB", 1, protolib },
1057 { "SCANFLIKE", 1, scanflike },
1058 { "VARARGS", 1, varargs },
1067 /* Skip white spaces after the start of the comment */
1068 while ((c = inpc()) != EOF && isspace(c))
1071 /* Read the potential keyword to keywd */
1073 while (c != EOF && isupper(c) && l < sizeof (keywd) - 1) {
1074 keywd[l++] = (char)c;
1079 /* look for the keyword */
1080 for (i = 0; i < sizeof (keywtab) / sizeof (keywtab[0]); i++) {
1081 if (strcmp(keywtab[i].keywd, keywd) == 0)
1084 if (i == sizeof (keywtab) / sizeof (keywtab[0]))
1087 /* skip white spaces after the keyword */
1088 while (c != EOF && isspace(c))
1091 /* read the argument, if the keyword accepts one and there is one */
1093 if (keywtab[i].arg) {
1094 while (c != EOF && isdigit(c) && l < sizeof (arg) - 1) {
1100 a = l != 0 ? atoi(arg) : -1;
1102 /* skip white spaces after the argument */
1103 while (c != EOF && isspace(c))
1106 if (c != '*' || (c = inpc()) != '/') {
1107 if (keywtab[i].func != linted)
1108 /* extra characters in lint comment */
1112 * remember that we have already found the end of the
1118 if (keywtab[i].func != NULL)
1119 (*keywtab[i].func)(a);
1124 if ((c = inpc()) == EOF) {
1125 /* unterminated comment */
1129 if (lc == '*' && c == '/')
1135 * Handle // style comments
1138 slashslashcomment(void)
1142 if (sflag < 2 && !gflag)
1143 /* // comments only supported in C99 */
1144 (void)gnuism(312, tflag ? "traditional" : "ANSI");
1146 while ((c = inpc()) != EOF && c != '\n')
1151 * Clear flags for lint comments LINTED, LONGLONG and CONSTCOND.
1152 * clrwflgs() is called after function definitions and global and
1153 * local declarations and definitions. It is also called between
1154 * the controlling expression and the body of control statements
1155 * (if, switch, for, while).
1167 * Strings are stored in a dynamically alloceted buffer and passed
1168 * in yylval.y_xstrg to the parser. The parser or the routines called
1169 * by the parser are responsible for freeing this buffer.
1179 if ((s = malloc(max = 64)) == NULL)
1183 while ((c = getescc('"')) >= 0) {
1184 /* +1 to reserve space for a trailing NUL character */
1186 if ((s = realloc(s, max *= 2)) == NULL)
1192 /* unterminated string constant */
1195 if ((strg = calloc(1, sizeof (strg_t))) == NULL)
1197 strg->st_tspec = CHAR;
1201 yylval.y_strg = strg;
1210 size_t len, max, wlen;
1214 if ((s = malloc(max = 64)) == NULL)
1217 while ((c = getescc('"')) >= 0) {
1218 /* +1 to save space for a trailing NUL character */
1220 if ((s = realloc(s, max *= 2)) == NULL)
1226 /* unterminated string constant */
1229 /* get length of wide character string */
1230 (void)mblen(NULL, 0);
1231 for (i = 0, wlen = 0; i < len; i += n, wlen++) {
1232 if ((n = mblen(&s[i], MB_CUR_MAX)) == -1) {
1233 /* invalid multibyte character */
1241 if ((ws = malloc((wlen + 1) * sizeof (wchar_t))) == NULL)
1244 /* convert from multibyte to wide char */
1245 (void)mbtowc(NULL, NULL, 0);
1246 for (i = 0, wi = 0; i < len; i += n, wi++) {
1247 if ((n = mbtowc(&ws[wi], &s[i], MB_CUR_MAX)) == -1)
1255 if ((strg = calloc(1, sizeof (strg_t))) == NULL)
1257 strg->st_tspec = WCHAR;
1258 strg->st_len = wlen;
1261 yylval.y_strg = strg;
1266 * As noted above the scanner does not create new symbol table entries
1267 * for symbols it cannot find in the symbol table. This is to avoid
1268 * putting undeclared symbols into the symbol table if a syntax error
1271 * getsym() is called as soon as it is probably ok to put the symbol to
1272 * the symbol table. This does not mean that it is not possible that
1273 * symbols are put to the symbol table which are than not completely
1274 * declared due to syntax errors. To avoid too many problems in this
1275 * case symbols get type int in getsym().
1277 * XXX calls to getsym() should be delayed until decl1*() is called
1289 * During member declaration it is possible that name() looked
1290 * for symbols of type FVFT, although it should have looked for
1291 * symbols of type FTAG. Same can happen for labels. Both cases
1292 * are compensated here.
1294 if (symtyp == FMOS || symtyp == FLAB) {
1295 if (sym == NULL || sym->s_kind == FVFT)
1300 if (sym->s_kind != symtyp)
1301 lerror("storesym() 1");
1307 /* create a new symbol table entry */
1309 /* labels must always be allocated at level 1 (outhermost block) */
1310 if (symtyp == FLAB) {
1311 sym = getlblk(1, sizeof (sym_t));
1312 s = getlblk(1, sb->sb_len + 1);
1313 (void)memcpy(s, sb->sb_name, sb->sb_len + 1);
1317 while (di->d_nxt != NULL && di->d_nxt->d_nxt != NULL)
1319 if (di->d_ctx != AUTO)
1320 lerror("storesym() 2");
1322 sym = getblk(sizeof (sym_t));
1323 sym->s_name = sb->sb_name;
1324 sym->s_blklev = blklev;
1328 UNIQUE_CURR_POS(sym->s_dpos);
1329 if ((sym->s_kind = symtyp) != FLAB)
1330 sym->s_type = gettyp(INT);
1334 if ((sym->s_link = symtab[sb->sb_hash]) != NULL)
1335 symtab[sb->sb_hash]->s_rlink = &sym->s_link;
1336 (symtab[sb->sb_hash] = sym)->s_rlink = &symtab[sb->sb_hash];
1338 *di->d_ldlsym = sym;
1339 di->d_ldlsym = &sym->s_dlnxt;
1346 * Remove a symbol forever from the symbol table. s_blklev
1347 * is set to -1 to avoid that the symbol will later be put
1348 * back to the symbol table.
1354 if ((*sym->s_rlink = sym->s_link) != NULL)
1355 sym->s_link->s_rlink = sym->s_rlink;
1361 * Remove a list of symbols declared at one level from the symbol
1369 for (sym = syms; sym != NULL; sym = sym->s_dlnxt) {
1370 if (sym->s_blklev != -1) {
1371 if ((*sym->s_rlink = sym->s_link) != NULL)
1372 sym->s_link->s_rlink = sym->s_rlink;
1374 sym->s_rlink = NULL;
1380 * Put a symbol into the symbol table
1383 inssym(int bl, sym_t *sym)
1387 h = hash(sym->s_name);
1388 if ((sym->s_link = symtab[h]) != NULL)
1389 symtab[h]->s_rlink = &sym->s_link;
1390 (symtab[h] = sym)->s_rlink = &symtab[h];
1392 if (sym->s_link != NULL && sym->s_blklev < sym->s_link->s_blklev)
1397 * Called at level 0 after syntax errors
1398 * Removes all symbols which are not declared at level 0 from the
1399 * symbol table. Also frees all memory which is not associated with
1408 for (i = 0; i < HSHSIZ1; i++) {
1409 for (sym = symtab[i]; sym != NULL; sym = nsym) {
1411 if (sym->s_blklev >= 1) {
1412 if ((*sym->s_rlink = nsym) != NULL)
1413 nsym->s_rlink = sym->s_rlink;
1418 for (i = mblklev; i > 0; i--)
1423 * Create a new symbol with the name of an existing symbol.
1426 pushdown(sym_t *sym)
1431 h = hash(sym->s_name);
1432 nsym = getblk(sizeof (sym_t));
1433 if (sym->s_blklev > blklev)
1434 lerror("pushdown()");
1435 nsym->s_name = sym->s_name;
1436 UNIQUE_CURR_POS(nsym->s_dpos);
1437 nsym->s_kind = sym->s_kind;
1438 nsym->s_blklev = blklev;
1440 if ((nsym->s_link = symtab[h]) != NULL)
1441 symtab[h]->s_rlink = &nsym->s_link;
1442 (symtab[h] = nsym)->s_rlink = &symtab[h];
1444 *dcs->d_ldlsym = nsym;
1445 dcs->d_ldlsym = &nsym->s_dlnxt;
1451 * Free any dynamically allocated memory referenced by
1452 * the value stack or yylval.
1453 * The type of information in yylval is described by tok.
1456 freeyyv(void *sp, int tok)
1458 if (tok == T_NAME || tok == T_TYPENAME) {
1459 sbuf_t *sb = *(sbuf_t **)sp;
1461 } else if (tok == T_CON) {
1462 val_t *val = *(val_t **)sp;
1464 } else if (tok == T_STRING) {
1465 strg_t *strg = *(strg_t **)sp;
1466 if (strg->st_tspec == CHAR) {
1468 } else if (strg->st_tspec == WCHAR) {
1471 lerror("fryylv() 1");