tok.c

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