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 extern int cpp_read(char *s);
   9
  10 static char buf[BUFSIZE];
  11 static int len;
  12 static int cur;
  13 static char name[NAMELEN];
  14 static int next = -1;
  15 static int pre;
  16
  17 static struct {
  18         char *name;
  19         unsigned id;
  20 } kwds[] = {
  21         {"void", TOK_VOID},
  22         {"static", TOK_STATIC},
  23         {"extern", TOK_EXTERN},
  24         {"return", TOK_RETURN},
  25         {"unsigned", TOK_UNSIGNED},
  26         {"signed", TOK_SIGNED},
  27         {"short", TOK_SHORT},
  28         {"long", TOK_LONG},
  29         {"int", TOK_INT},
  30         {"char", TOK_CHAR},
  31         {"struct", TOK_STRUCT},
  32         {"union", TOK_UNION},
  33         {"enum", TOK_ENUM},
  34         {"typedef", TOK_TYPEDEF},
  35         {"if", TOK_IF},
  36         {"else", TOK_ELSE},
  37         {"for", TOK_FOR},
  38         {"while", TOK_WHILE},
  39         {"do", TOK_DO},
  40         {"switch", TOK_SWITCH},
  41         {"case", TOK_CASE},
  42         {"sizeof", TOK_SIZEOF},
  43         {"break", TOK_BREAK},
  44         {"continue", TOK_CONTINUE},
  45         {"default", TOK_DEFAULT},
  46         {"goto", TOK_GOTO},
  47 };
  48
  49 static char *tok3[] = {
  50         "<<", ">>", "++", "--", "<<=", ">>=", "...", "+=", "-=", "*=", "/=",
  51         "%=", "|=", "&=", "^=", "&&", "||", "==", "!=", "<=", ">=", "->", "/*"
  52 };
  53
  54 static int get_tok3(int num)
  55 {
  56         int i;
  57         for (i = 0; i < ARRAY_SIZE(tok3); i++)
  58                 if (num == TOK3(tok3[i]))
  59                         return num;
  60         return 0;
  61 }
  62
  63 static char *esc_code = "abefnrtv";
  64 static char *esc = "\a\b\e\f\n\r\t\v";
  65 static char *digs = "0123456789abcdef";
  66
  67 static int esc_char(int *c, char *s)
  68 {
  69         if (*s != '\\') {
  70                 *c = *s;
  71                 return 1;
  72         }
  73         if (strchr(esc_code, s[1])) {
  74                 *c = esc[strchr(esc_code, s[1]) - esc_code];
  75                 return 2;
  76         }
  77         if (isdigit(s[1]) || s[1] == 'x') {
  78                 int ret = 0;
  79                 int base = 8;
  80                 int i = 1;
  81                 char *d;
  82                 if (s[1] == 'x') {
  83                         base = 16;
  84                         i++;
  85                 }
  86                 while ((d = strchr(digs, s[i]))) {
  87                         ret *= base;
  88                         ret += d - digs;
  89                         i++;
  90                 }
  91                 *c = ret;
  92                 return i;
  93         }
  94         *c = s[1];
  95         return 2;
  96 }
  97
  98 static long num;
  99 static int num_bt;
 100
 101 int tok_num(long *n)
 102 {
 103         *n = num;
 104         return num_bt;
 105 }
 106
 107 static void readnum(void)
 108 {
 109         int base = 10;
 110         num_bt = 4 | BT_SIGNED;
 111         if (buf[cur] == '0' && buf[cur + 1] == 'x') {
 112                 base = 16;
 113                 cur += 2;
 114         }
 115         if (strchr(digs, 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) | 8;
 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[BUFSIZE];
 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 (TOK2(buf + cur) != TOK2("/*"))
 198                         return 0;
 199                 while (++cur < len) {
 200                         if (buf[cur] == '*' && buf[cur + 1] == '/') {
 201                                 cur += 2;
 202                                 break;
 203                         }
 204                 }
 205         }
 206         return 0;
 207 }
 208
 209 int tok_get(void)
 210 {
 211         int num;
 212         if (next != -1) {
 213                 int tok = next;
 214                 next = -1;
 215                 return tok;
 216         }
 217         pre = cur;
 218         if (skipws())
 219                 return TOK_EOF;
 220         if (buf[cur] == '"') {
 221                 str_len = 0;
 222                 while (buf[cur] == '"') {
 223                         str_len += readstr(str + str_len);
 224                         if (skipws())
 225                                 return TOK_EOF;
 226                 }
 227                 str_len++;
 228                 return TOK_STR;
 229         }
 230         if (isdigit(buf[cur]) || buf[cur] == '\'') {
 231                 readnum();
 232                 return TOK_NUM;
 233         }
 234         if (id_char(buf[cur])) {
 235                 char *s = name;
 236                 int i;
 237                 while (cur < len && id_char(buf[cur]))
 238                         *s++ = buf[cur++];
 239                 *s = '\0';
 240                 for (i = 0; i < ARRAY_SIZE(kwds); i++)
 241                         if (!strcmp(kwds[i].name, name))
 242                                 return kwds[i].id;
 243                 return TOK_NAME;
 244         }
 245         if ((num = get_tok3(TOK3(buf + cur)))) {
 246                 cur += 3;
 247                 return num;
 248         }
 249         if ((num = get_tok3(TOK2(buf + cur)))) {
 250                 cur += 2;
 251                 return num;
 252         }
 253         if (strchr(";,{}()[]<>*&!=+-/%?:|^~.", buf[cur]))
 254                 return buf[cur++];
 255         return -1;
 256 }
 257
 258 int tok_see(void)
 259 {
 260         if (next == -1)
 261                 next = tok_get();
 262         return next;
 263 }
 264
 265 char *tok_id(void)
 266 {
 267         return name;
 268 }
 269
 270 long tok_addr(void)
 271 {
 272         return next == -1 ? cur : pre;
 273 }
 274
 275 void tok_jump(long addr)
 276 {
 277         cur = addr;
 278         pre = cur - 1;
 279         next = -1;
 280 }