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 };
  42
  43 static char *tok3[] = {
  44         "<<", ">>", "++", "--", "<<=", ">>=", "...", "+=", "-=", "*=", "/=",
  45         "%=", "|=", "&=", "^=", "&&", "||", "==", "!=", "<=", ">=", "->", "/*"
  46 };
  47
  48 static int get_tok3(int num)
  49 {
  50         int i;
  51         for (i = 0; i < ARRAY_SIZE(tok3); i++)
  52                 if (num == TOK3(tok3[i]))
  53                         return num;
  54         return 0;
  55 }
  56
  57 static char *esc_code = "abefnrtv";
  58 static char *esc = "\a\b\e\f\n\r\t\v";
  59
  60 static int esc_char(int *c, char *s)
  61 {
  62         if (*s != '\\') {
  63                 *c = *s;
  64                 return 1;
  65         }
  66         if (strchr(esc_code, s[1])) {
  67                 *c = esc[strchr(esc_code, s[1]) - esc_code];
  68                 return 2;
  69         }
  70         *c = s[1];
  71         return 2;
  72 }
  73
  74 long tok_num(void)
  75 {
  76         if (buf[cur] == '0' && buf[cur + 1] == 'x') {
  77                 long result = 0;
  78                 cur += 2;
  79                 while (isalnum(buf[cur])) {
  80                         int c = buf[cur];
  81                         result <<= 4;
  82                         if (c >= '0' && c <= '9')
  83                                 result |= c - '0';
  84                         else
  85                                 result |= 10 + tolower(c) - 'a';
  86                         cur++;
  87                 }
  88                 return result;
  89         }
  90         if (isdigit(buf[cur])) {
  91                 long result = 0;
  92                 while (isdigit(buf[cur])) {
  93                         result *= 10;
  94                         result += buf[cur++] - '0';
  95                 }
  96                 return result;
  97         }
  98         if (buf[cur] == '\'') {
  99                 int ret;
 100                 cur += 2 + esc_char(&ret, buf + cur + 1);
 101                 return ret;
 102         }
 103         return -1;
 104 }
 105
 106 int tok_str(char *out)
 107 {
 108         char *s = out;
 109         char *r = buf + cur;
 110         char *e = buf + len;
 111         r++;
 112         while (r < e && *r != '"') {
 113                 if (*r == '\\') {
 114                         int c;
 115                         r += esc_char(&c, r);
 116                         *s++ = c;
 117                 } else {
 118                         *s++ = *r++;
 119                 }
 120         }
 121         *s++ = '\0';
 122         cur = r - buf + 1;
 123         return s - out;
 124 }
 125
 126 static int id_char(int c)
 127 {
 128         return isalnum(c) || c == '_';
 129 }
 130
 131 static int skipws(void)
 132 {
 133         while (1) {
 134                 while (cur < len && isspace(buf[cur]))
 135                         cur++;
 136                 if (cur == len)
 137                         return 1;
 138                 if (TOK2(buf + cur) != TOK2("/*"))
 139                         return 0;
 140                 while (++cur < len) {
 141                         if (buf[cur] == '*' && buf[cur + 1] == '/') {
 142                                 cur += 2;
 143                                 break;
 144                         }
 145                 }
 146         }
 147         return 0;
 148 }
 149
 150 int tok_get(void)
 151 {
 152         int num;
 153         if (next != -1) {
 154                 int tok = next;
 155                 next = -1;
 156                 return tok;
 157         }
 158         if (skipws())
 159                 return TOK_EOF;
 160         if (buf[cur] == '"')
 161                 return TOK_STR;
 162         if (isdigit(buf[cur]) || buf[cur] == '\'')
 163                 return TOK_NUM;
 164         if (id_char(buf[cur])) {
 165                 char *s = name;
 166                 int i;
 167                 while (cur < len && id_char(buf[cur]))
 168                         *s++ = buf[cur++];
 169                 *s = '\0';
 170                 for (i = 0; i < ARRAY_SIZE(kwds); i++)
 171                         if (!strcmp(kwds[i].name, name))
 172                                 return kwds[i].id;
 173                 return TOK_NAME;
 174         }
 175         if ((num = get_tok3(TOK3(buf + cur)))) {
 176                 cur += 3;
 177                 return num;
 178         }
 179         if ((num = get_tok3(TOK2(buf + cur)))) {
 180                 cur += 2;
 181                 return num;
 182         }
 183         if (strchr(";,{}()[]<>*&!=+-/%?:|^~.", buf[cur]))
 184                 return buf[cur++];
 185         return -1;
 186 }
 187
 188 int tok_see(void)
 189 {
 190         if (next == -1)
 191                 next = tok_get();
 192         return next;
 193 }
 194
 195 void tok_init(int fd)
 196 {
 197         int n = 0;
 198         while ((n = read(fd, buf + len, sizeof(buf) - len)) > 0)
 199                 len += n;
 200         next = -1;
 201 }
 202
 203 char *tok_id(void)
 204 {
 205         return name;
 206 }