tok.c

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