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                 num_bt &= ~BT_SIGNED;
 113                 base = 16;
 114                 cur += 2;
 115         }
 116         if (strchr(digs, tolower(buf[cur]))) {
 117                 long result = 0;
 118                 char *c;
 119                 if (base == 10 && buf[cur] == '0')
 120                         base = 8;
 121                 while (cur < len && (c = strchr(digs, tolower(buf[cur])))) {
 122                         result *= base;
 123                         result += c - digs;
 124                         cur++;
 125                 }
 126                 num = result;
 127                 while (cur < len) {
 128                         int c = tolower(buf[cur]);
 129                         if (c != 'u' && c != 'l')
 130                                 break;
 131                         if (c == 'u')
 132                                 num_bt &= ~BT_SIGNED;
 133                         if (c == 'l')
 134                                 num_bt = (num_bt & BT_SIGNED) | LONGSZ;
 135                         cur++;
 136                 }
 137                 return;
 138         }
 139         if (buf[cur] == '\'') {
 140                 int ret;
 141                 cur += 2 + esc_char(&ret, buf + cur + 1);
 142                 num = ret;
 143                 return;
 144         }
 145         num = -1;
 146 }
 147
 148 static char str[BUFSIZE];
 149 static int str_len;
 150
 151 int tok_str(char *buf)
 152 {
 153         if (buf)
 154                 memcpy(buf, str, str_len);
 155         return str_len;
 156 }
 157
 158 static int readstr(char *out)
 159 {
 160         char *s = out;
 161         char *r = buf + cur;
 162         char *e = buf + len;
 163         r++;
 164         while (r < e && *r != '"') {
 165                 if (*r == '\\') {
 166                         int c;
 167                         r += esc_char(&c, r);
 168                         *s++ = c;
 169                 } else {
 170                         *s++ = *r++;
 171                 }
 172         }
 173         *s++ = '\0';
 174         cur = r - buf + 1;
 175         return s - out - 1;
 176 }
 177
 178 static int id_char(int c)
 179 {
 180         return isalnum(c) || c == '_';
 181 }
 182
 183 static int skipws(void)
 184 {
 185         while (1) {
 186                 if (cur == len) {
 187                         int r;
 188                         while (!(r = cpp_read(buf + cur)))
 189                                 ;
 190                         if (r == -1)
 191                                 return 1;
 192                         len += r;
 193                 }
 194                 while (cur < len && isspace(buf[cur]))
 195                         cur++;
 196                 if (cur == len)
 197                         continue;
 198                 if (TOK2(buf + cur) != TOK2("/*"))
 199                         return 0;
 200                 while (++cur < len) {
 201                         if (buf[cur] == '*' && buf[cur + 1] == '/') {
 202                                 cur += 2;
 203                                 break;
 204                         }
 205                 }
 206         }
 207         return 0;
 208 }
 209
 210 int tok_get(void)
 211 {
 212         int num;
 213         if (next != -1) {
 214                 int tok = next;
 215                 next = -1;
 216                 return tok;
 217         }
 218         pre = cur;
 219         if (skipws())
 220                 return TOK_EOF;
 221         if (buf[cur] == '"') {
 222                 str_len = 0;
 223                 while (buf[cur] == '"') {
 224                         str_len += readstr(str + str_len);
 225                         if (skipws())
 226                                 return TOK_EOF;
 227                 }
 228                 str_len++;
 229                 return TOK_STR;
 230         }
 231         if (isdigit(buf[cur]) || buf[cur] == '\'') {
 232                 readnum();
 233                 return TOK_NUM;
 234         }
 235         if (id_char(buf[cur])) {
 236                 char *s = name;
 237                 int i;
 238                 while (cur < len && id_char(buf[cur]))
 239                         *s++ = buf[cur++];
 240                 *s = '\0';
 241                 for (i = 0; i < ARRAY_SIZE(kwds); i++)
 242                         if (!strcmp(kwds[i].name, name))
 243                                 return kwds[i].id;
 244                 return TOK_NAME;
 245         }
 246         if (cur + 3 <= len && (num = get_tok3(TOK3(buf + cur)))) {
 247                 cur += 3;
 248                 return num;
 249         }
 250         if ((num = get_tok3(TOK2(buf + cur)))) {
 251                 cur += 2;
 252                 return num;
 253         }
 254         if (strchr(";,{}()[]<>*&!=+-/%?:|^~.", buf[cur]))
 255                 return buf[cur++];
 256         return -1;
 257 }
 258
 259 int tok_see(void)
 260 {
 261         if (next == -1)
 262                 next = tok_get();
 263         return next;
 264 }
 265
 266 char *tok_id(void)
 267 {
 268         return name;
 269 }
 270
 271 long tok_addr(void)
 272 {
 273         return next == -1 ? cur : pre;
 274 }
 275
 276 void tok_jump(long addr)
 277 {
 278         cur = addr;
 279         pre = cur - 1;
 280         next = -1;
 281 }