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 void cpp_init(int fd);
   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;
  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
 100 long tok_num(void)
 101 {
 102         return num;
 103 }
 104
 105 static void readnum(void)
 106 {
 107         int base = 10;
 108         if (buf[cur] == '0' && buf[cur + 1] == 'x') {
 109                 base = 16;
 110                 cur += 2;
 111         }
 112         if (strchr(digs, buf[cur])) {
 113                 long result = 0;
 114                 char *c;
 115                 if (base == 10 && buf[cur] == '0')
 116                         base = 8;
 117                 while (cur < len && (c = strchr(digs, buf[cur]))) {
 118                         result *= base;
 119                         result += c - digs;
 120                         cur++;
 121                 }
 122                 num = result;
 123                 while (cur < len && tolower(buf[cur]) == 'u' ||
 124                                 tolower(buf[cur]) == 'l')
 125                         cur++;
 126                 return;
 127         }
 128         if (buf[cur] == '\'') {
 129                 int ret;
 130                 cur += 2 + esc_char(&ret, buf + cur + 1);
 131                 num = ret;
 132                 return;
 133         }
 134         num = -1;
 135 }
 136
 137 static char str[BUFSIZE];
 138 static int str_len;
 139
 140 int tok_str(char *buf)
 141 {
 142         if (buf)
 143                 memcpy(buf, str, str_len);
 144         return str_len;
 145 }
 146
 147 static int readstr(char *out)
 148 {
 149         char *s = out;
 150         char *r = buf + cur;
 151         char *e = buf + len;
 152         r++;
 153         while (r < e && *r != '"') {
 154                 if (*r == '\\') {
 155                         int c;
 156                         r += esc_char(&c, r);
 157                         *s++ = c;
 158                 } else {
 159                         *s++ = *r++;
 160                 }
 161         }
 162         *s++ = '\0';
 163         cur = r - buf + 1;
 164         return s - out - 1;
 165 }
 166
 167 static int id_char(int c)
 168 {
 169         return isalnum(c) || c == '_';
 170 }
 171
 172 static int skipws(void)
 173 {
 174         while (1) {
 175                 if (cur == len) {
 176                         int r;
 177                         while (!(r = cpp_read(buf + cur)))
 178                                 ;
 179                         if (r == -1)
 180                                 return 1;
 181                         len += r;
 182                 }
 183                 while (cur < len && isspace(buf[cur]))
 184                         cur++;
 185                 if (cur == len)
 186                         continue;
 187                 if (TOK2(buf + cur) != TOK2("/*"))
 188                         return 0;
 189                 while (++cur < len) {
 190                         if (buf[cur] == '*' && buf[cur + 1] == '/') {
 191                                 cur += 2;
 192                                 break;
 193                         }
 194                 }
 195         }
 196         return 0;
 197 }
 198
 199 int tok_get(void)
 200 {
 201         int num;
 202         if (next != -1) {
 203                 int tok = next;
 204                 next = -1;
 205                 return tok;
 206         }
 207         pre = cur;
 208         if (skipws())
 209                 return TOK_EOF;
 210         if (buf[cur] == '"') {
 211                 str_len = 0;
 212                 while (buf[cur] == '"') {
 213                         str_len += readstr(str + str_len);
 214                         if (skipws())
 215                                 return TOK_EOF;
 216                 }
 217                 str_len++;
 218                 return TOK_STR;
 219         }
 220         if (isdigit(buf[cur]) || buf[cur] == '\'') {
 221                 readnum();
 222                 return TOK_NUM;
 223         }
 224         if (id_char(buf[cur])) {
 225                 char *s = name;
 226                 int i;
 227                 while (cur < len && id_char(buf[cur]))
 228                         *s++ = buf[cur++];
 229                 *s = '\0';
 230                 for (i = 0; i < ARRAY_SIZE(kwds); i++)
 231                         if (!strcmp(kwds[i].name, name))
 232                                 return kwds[i].id;
 233                 return TOK_NAME;
 234         }
 235         if ((num = get_tok3(TOK3(buf + cur)))) {
 236                 cur += 3;
 237                 return num;
 238         }
 239         if ((num = get_tok3(TOK2(buf + cur)))) {
 240                 cur += 2;
 241                 return num;
 242         }
 243         if (strchr(";,{}()[]<>*&!=+-/%?:|^~.", buf[cur]))
 244                 return buf[cur++];
 245         return -1;
 246 }
 247
 248 int tok_see(void)
 249 {
 250         if (next == -1)
 251                 next = tok_get();
 252         return next;
 253 }
 254
 255 void tok_init(int fd)
 256 {
 257         next = -1;
 258         cpp_init(fd);
 259 }
 260
 261 char *tok_id(void)
 262 {
 263         return name;
 264 }
 265
 266 long tok_addr(void)
 267 {
 268         return next == -1 ? cur : pre;
 269 }
 270
 271 void tok_jump(long addr)
 272 {
 273         cur = addr;
 274         pre = cur - 1;
 275         next = -1;
 276 }