tok.c

   1 #include <ctype.h>
   2 #include <unistd.h>
   3 #include <stdlib.h>
   4 #include <string.h>
   5 #include "gen.h"
   6 #include "ncc.h"
   7 #include "tok.h"
   8
   9 static char buf[BUFLEN];
  10 static int len;
  11 static int cur;
  12 static char name[NAMELEN];
  13 static int next = -1;
  14 static int pre;
  15
  16 static struct {
  17         char *name;
  18         unsigned id;
  19 } kwds[] = {
  20         {"void", TOK_VOID},
  21         {"static", TOK_STATIC},
  22         {"extern", TOK_EXTERN},
  23         {"return", TOK_RETURN},
  24         {"unsigned", TOK_UNSIGNED},
  25         {"signed", TOK_SIGNED},
  26         {"short", TOK_SHORT},
  27         {"long", TOK_LONG},
  28         {"int", TOK_INT},
  29         {"char", TOK_CHAR},
  30         {"struct", TOK_STRUCT},
  31         {"union", TOK_UNION},
  32         {"enum", TOK_ENUM},
  33         {"typedef", TOK_TYPEDEF},
  34         {"if", TOK_IF},
  35         {"else", TOK_ELSE},
  36         {"for", TOK_FOR},
  37         {"while", TOK_WHILE},
  38         {"do", TOK_DO},
  39         {"switch", TOK_SWITCH},
  40         {"case", TOK_CASE},
  41         {"sizeof", TOK_SIZEOF},
  42         {"break", TOK_BREAK},
  43         {"continue", TOK_CONTINUE},
  44         {"default", TOK_DEFAULT},
  45         {"goto", TOK_GOTO},
  46 };
  47
  48 static char *tok3[] = {
  49         "<<=", ">>=", "...", "<<", ">>", "++", "--", "+=", "-=", "*=", "/=",
  50         "%=", "|=", "&=", "^=", "&&", "||", "==", "!=", "<=", ">=", "->"
  51 };
  52
  53 static int get_tok3(int num)
  54 {
  55         int i;
  56         for (i = 0; i < LEN(tok3); i++)
  57                 if (num == TOK3(tok3[i]))
  58                         return num;
  59         return 0;
  60 }
  61
  62 static char *esc_code = "abefnrtv";
  63 static char *esc = "\a\b\e\f\n\r\t\v";
  64 static char *digs = "0123456789abcdef";
  65
  66 static int esc_char(int *c, char *s)
  67 {
  68         if (*s != '\\') {
  69                 *c = *s;
  70                 return 1;
  71         }
  72         if (strchr(esc_code, s[1])) {
  73                 *c = esc[strchr(esc_code, s[1]) - esc_code];
  74                 return 2;
  75         }
  76         if (isdigit(s[1]) || s[1] == 'x') {
  77                 int ret = 0;
  78                 int base = 8;
  79                 int i = 1;
  80                 char *d;
  81                 if (s[1] == 'x') {
  82                         base = 16;
  83                         i++;
  84                 }
  85                 while ((d = memchr(digs, s[i], base))) {
  86                         ret *= base;
  87                         ret += d - digs;
  88                         i++;
  89                 }
  90                 *c = ret;
  91                 return i;
  92         }
  93         *c = s[1];
  94         return 2;
  95 }
  96
  97 static long num;
  98 static int num_bt;
  99
 100 int tok_num(long *n)
 101 {
 102         *n = num;
 103         return num_bt;
 104 }
 105
 106 static void readnum(void)
 107 {
 108         int base = 10;
 109         num_bt = 4 | BT_SIGNED;
 110         if (buf[cur] == '0' && buf[cur + 1] == 'x') {
 111                 num_bt &= ~BT_SIGNED;
 112                 base = 16;
 113                 cur += 2;
 114         }
 115         if (strchr(digs, tolower(buf[cur]))) {
 116                 long result = 0;
 117                 char *c;
 118                 if (base == 10 && buf[cur] == '0')
 119                         base = 8;
 120                 while (cur < len && (c = strchr(digs, tolower(buf[cur])))) {
 121                         result *= base;
 122                         result += c - digs;
 123                         cur++;
 124                 }
 125                 num = result;
 126                 while (cur < len) {
 127                         int c = tolower(buf[cur]);
 128                         if (c != 'u' && c != 'l')
 129                                 break;
 130                         if (c == 'u')
 131                                 num_bt &= ~BT_SIGNED;
 132                         if (c == 'l')
 133                                 num_bt = (num_bt & BT_SIGNED) | LONGSZ;
 134                         cur++;
 135                 }
 136                 return;
 137         }
 138         if (buf[cur] == '\'') {
 139                 int ret;
 140                 cur += 2 + esc_char(&ret, buf + cur + 1);
 141                 num = ret;
 142                 return;
 143         }
 144         num = -1;
 145 }
 146
 147 static char str[STRLEN];
 148 static int str_len;
 149
 150 int tok_str(char *buf)
 151 {
 152         if (buf)
 153                 memcpy(buf, str, str_len);
 154         return str_len;
 155 }
 156
 157 static int readstr(char *out)
 158 {
 159         char *s = out;
 160         char *r = buf + cur;
 161         char *e = buf + len;
 162         r++;
 163         while (r < e && *r != '"') {
 164                 if (*r == '\\') {
 165                         int c;
 166                         r += esc_char(&c, r);
 167                         *s++ = c;
 168                 } else {
 169                         *s++ = *r++;
 170                 }
 171         }
 172         *s++ = '\0';
 173         cur = r - buf + 1;
 174         return s - out - 1;
 175 }
 176
 177 static int id_char(int c)
 178 {
 179         return isalnum(c) || c == '_';
 180 }
 181
 182 static int skipws(void)
 183 {
 184         while (1) {
 185                 if (cur == len) {
 186                         int r;
 187                         while (!(r = cpp_read(buf + cur)))
 188                                 ;
 189                         if (r == -1)
 190                                 return 1;
 191                         len += r;
 192                 }
 193                 while (cur < len && isspace(buf[cur]))
 194                         cur++;
 195                 if (cur == len)
 196                         continue;
 197                 if (buf[cur] == '\\' && buf[cur + 1] == '\n') {
 198                         cur += 2;
 199                         continue;
 200                 }
 201                 if (buf[cur] == '/' && buf[cur + 1] == '/') {
 202                         while (cur < len && buf[cur] != '\n')
 203                                 cur++;
 204                         continue;
 205                 }
 206                 if (buf[cur] == '/' && buf[cur + 1] == '*') {
 207                         while (++cur < len) {
 208                                 if (buf[cur] == '*' && buf[cur + 1] == '/') {
 209                                         cur += 2;
 210                                         break;
 211                                 }
 212                         }
 213                         continue;
 214                 }
 215                 break;
 216         }
 217         return 0;
 218 }
 219
 220 int tok_get(void)
 221 {
 222         int num;
 223         if (next != -1) {
 224                 int tok = next;
 225                 next = -1;
 226                 return tok;
 227         }
 228         pre = cur;
 229         if (skipws())
 230                 return TOK_EOF;
 231         if (buf[cur] == '"') {
 232                 str_len = 0;
 233                 while (buf[cur] == '"') {
 234                         str_len += readstr(str + str_len);
 235                         if (skipws())
 236                                 return TOK_EOF;
 237                 }
 238                 str_len++;
 239                 return TOK_STR;
 240         }
 241         if (isdigit(buf[cur]) || buf[cur] == '\'') {
 242                 readnum();
 243                 return TOK_NUM;
 244         }
 245         if (id_char(buf[cur])) {
 246                 char *s = name;
 247                 int i;
 248                 while (cur < len && id_char(buf[cur]))
 249                         *s++ = buf[cur++];
 250                 *s = '\0';
 251                 for (i = 0; i < LEN(kwds); i++)
 252                         if (!strcmp(kwds[i].name, name))
 253                                 return kwds[i].id;
 254                 return TOK_NAME;
 255         }
 256         if (cur + 3 <= len && (num = get_tok3(TOK3(buf + cur)))) {
 257                 cur += 3;
 258                 return num;
 259         }
 260         if ((num = get_tok3(TOK2(buf + cur)))) {
 261                 cur += 2;
 262                 return num;
 263         }
 264         if (strchr(";,{}()[]<>*&!=+-/%?:|^~.", buf[cur]))
 265                 return buf[cur++];
 266         return -1;
 267 }
 268
 269 int tok_see(void)
 270 {
 271         if (next == -1)
 272                 next = tok_get();
 273         return next;
 274 }
 275
 276 char *tok_id(void)
 277 {
 278         return name;
 279 }
 280
 281 long tok_addr(void)
 282 {
 283         return next == -1 ? cur : pre;
 284 }
 285
 286 void tok_jump(long addr)
 287 {
 288         cur = addr;
 289         pre = cur - 1;
 290         next = -1;
 291 }