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