tok.c

   1 #include <ctype.h>
   2 #include <unistd.h>
   3 #include <stdlib.h>
   4 #include <string.h>
   5 #include "tok.h"
   6
   7 extern int cpp_read(char *s);
   8
   9 static char buf[BUFSIZE];
  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 < ARRAY_SIZE(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 = strchr(digs, s[i]))) {
  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
  99 long tok_num(void)
 100 {
 101         return num;
 102 }
 103
 104 static void readnum(void)
 105 {
 106         int base = 10;
 107         if (buf[cur] == '0' && buf[cur + 1] == 'x') {
 108                 base = 16;
 109                 cur += 2;
 110         }
 111         if (strchr(digs, buf[cur])) {
 112                 long result = 0;
 113                 char *c;
 114                 if (base == 10 && buf[cur] == '0')
 115                         base = 8;
 116                 while (cur < len && (c = strchr(digs, tolower(buf[cur])))) {
 117                         result *= base;
 118                         result += c - digs;
 119                         cur++;
 120                 }
 121                 num = result;
 122                 while (cur < len && tolower(buf[cur]) == 'u' ||
 123                                 tolower(buf[cur]) == 'l')
 124                         cur++;
 125                 return;
 126         }
 127         if (buf[cur] == '\'') {
 128                 int ret;
 129                 cur += 2 + esc_char(&ret, buf + cur + 1);
 130                 num = ret;
 131                 return;
 132         }
 133         num = -1;
 134 }
 135
 136 static char str[BUFSIZE];
 137 static int str_len;
 138
 139 int tok_str(char *buf)
 140 {
 141         if (buf)
 142                 memcpy(buf, str, str_len);
 143         return str_len;
 144 }
 145
 146 static int readstr(char *out)
 147 {
 148         char *s = out;
 149         char *r = buf + cur;
 150         char *e = buf + len;
 151         r++;
 152         while (r < e && *r != '"') {
 153                 if (*r == '\\') {
 154                         int c;
 155                         r += esc_char(&c, r);
 156                         *s++ = c;
 157                 } else {
 158                         *s++ = *r++;
 159                 }
 160         }
 161         *s++ = '\0';
 162         cur = r - buf + 1;
 163         return s - out - 1;
 164 }
 165
 166 static int id_char(int c)
 167 {
 168         return isalnum(c) || c == '_';
 169 }
 170
 171 static int skipws(void)
 172 {
 173         while (1) {
 174                 if (cur == len) {
 175                         int r;
 176                         while (!(r = cpp_read(buf + cur)))
 177                                 ;
 178                         if (r == -1)
 179                                 return 1;
 180                         len += r;
 181                 }
 182                 while (cur < len && isspace(buf[cur]))
 183                         cur++;
 184                 if (cur == len)
 185                         continue;
 186                 if (TOK2(buf + cur) != TOK2("/*"))
 187                         return 0;
 188                 while (++cur < len) {
 189                         if (buf[cur] == '*' && buf[cur + 1] == '/') {
 190                                 cur += 2;
 191                                 break;
 192                         }
 193                 }
 194         }
 195         return 0;
 196 }
 197
 198 int tok_get(void)
 199 {
 200         int num;
 201         if (next != -1) {
 202                 int tok = next;
 203                 next = -1;
 204                 return tok;
 205         }
 206         pre = cur;
 207         if (skipws())
 208                 return TOK_EOF;
 209         if (buf[cur] == '"') {
 210                 str_len = 0;
 211                 while (buf[cur] == '"') {
 212                         str_len += readstr(str + str_len);
 213                         if (skipws())
 214                                 return TOK_EOF;
 215                 }
 216                 str_len++;
 217                 return TOK_STR;
 218         }
 219         if (isdigit(buf[cur]) || buf[cur] == '\'') {
 220                 readnum();
 221                 return TOK_NUM;
 222         }
 223         if (id_char(buf[cur])) {
 224                 char *s = name;
 225                 int i;
 226                 while (cur < len && id_char(buf[cur]))
 227                         *s++ = buf[cur++];
 228                 *s = '\0';
 229                 for (i = 0; i < ARRAY_SIZE(kwds); i++)
 230                         if (!strcmp(kwds[i].name, name))
 231                                 return kwds[i].id;
 232                 return TOK_NAME;
 233         }
 234         if ((num = get_tok3(TOK3(buf + cur)))) {
 235                 cur += 3;
 236                 return num;
 237         }
 238         if ((num = get_tok3(TOK2(buf + cur)))) {
 239                 cur += 2;
 240                 return num;
 241         }
 242         if (strchr(";,{}()[]<>*&!=+-/%?:|^~.", buf[cur]))
 243                 return buf[cur++];
 244         return -1;
 245 }
 246
 247 int tok_see(void)
 248 {
 249         if (next == -1)
 250                 next = tok_get();
 251         return next;
 252 }
 253
 254 char *tok_id(void)
 255 {
 256         return name;
 257 }
 258
 259 long tok_addr(void)
 260 {
 261         return next == -1 ? cur : pre;
 262 }
 263
 264 void tok_jump(long addr)
 265 {
 266         cur = addr;
 267         pre = cur - 1;
 268         next = -1;
 269 }