token.c

   1 #include <ctype.h>
   2 #include <stdbool.h>
   3 #include <stdlib.h>
   4 #include <stdio.h>
   5 #include <string.h>
   6
   7 #include "token.h"
   8
   9 char *keywords[] =
  10 {
  11                  "boolean","char", "class", "constructor", "do", "else",
  12                  "false", "field", "function", "if", "int", "let", "method",
  13                  "null", "return", "static", "this", "true", "var", "void", "while"
  14 };
  15
  16 int has_more_tokens(char *pC)
  17 {
  18         if(pC == NULL) { return false; }
  19
  20         if(*(pC+1) == '\0')
  21         {
  22                 return false;
  23         } else {
  24                 return true;
  25         }
  26 }
  27
  28 char *advance(char *pC, char pT[])
  29 {
  30         char ch;
  31         int cont; /* continue */
  32         int in_quote = 0; /* inside quotes? */
  33         do {
  34                 cont = 0;
  35                 /* skip past C++ style comments */
  36                 if(*pC == '/' && *(pC+1) == '/')
  37                 {
  38                         pC = strchr(pC, '\n');
  39                 }
  40
  41                 /* skip past C style comments */
  42                 if(*pC == '/' && *(pC+1) == '*')
  43                 {
  44                         pC += 2;
  45                         do {
  46                                 pC++;
  47                         } while((*pC != '/') || (*(pC-1) != '*'));
  48                         pC++;
  49                 }
  50
  51                 /* advance past spaces and newline chars */
  52                 ch = *pC;
  53                 while((strchr(SPACES, ch)) != NULL)
  54                 {
  55                         pC++;
  56                         ch = *pC;
  57                 }
  58
  59                 /* determine if more advancement is needed */
  60                 if(*pC == '/' && *(pC+1) == '/') { cont++; }
  61                 if(*pC == '/' && *(pC+1) == '*') { cont++; }
  62
  63         } while (cont);
  64
  65         /* test for symbol - copy to buffer */
  66         ch = *pC;
  67         if(strchr(SYMBOLS, ch) != NULL)
  68         {
  69                 pT[0] = ch;
  70                 pT[1] = '\0';
  71                 pC++;
  72                 return pC;
  73         }
  74
  75         /* should be at beginning of token - copy to buffer */
  76         cont = 0; in_quote = 0;
  77         if(*pC == '"')
  78         {
  79                 in_quote++; /* inside quote */
  80                 pC++;           /* skip quote itself */
  81         }
  82         while(TRUE)
  83         {
  84                 if(in_quote)
  85                 {
  86                         if(*pC == '"') {  pC++; break; } /* found corresponding quote, stop */
  87                 } else {
  88                         if(((strchr(SPACES, ch)) != NULL) || ((strchr(SYMBOLS, ch)) != NULL))
  89                         {
  90                                 break;
  91                         }
  92                 }
  93                 pT[cont] = *pC;
  94                 pC++;
  95                 ch = *pC;
  96                 cont++;
  97         }
  98         pT[cont] = '\0';
  99         return pC;
 100 }
 101
 102 token token_type(char pT[])
 103 {
 104         token t = IDENTIFIER;
 105         int i = 0, j = 0;;
 106
 107         /* check token for symbols */
 108         if(strchr(SYMBOLS, pT[0]) != NULL) { t = SYMBOL; }
 109
 110         /* check token for integer constants */
 111         while(pT[i] != '\0')
 112         {
 113                 if(isdigit(pT[i]) == 0)
 114                 {
 115                         j++;
 116                         break;
 117                 }
 118                 i++;
 119         }
 120         if(j == 0) { t = INT_CONST; }
 121
 122         /* check token for keywords */
 123         i = 0; j = 0;
 124         do {
 125                 j = strcmp(keywords[i], pT);
 126                 i++;
 127         } while ((j != 0) && (i < KEYWORD_COUNT));
 128
 129         if(j == 0) { t = KEYWORD; }
 130         return t;
 131 }
 132
 133 ttype keyword(char pT[])
 134 {
 135          int i = 0;
 136          do {
 137                  i++;
 138          } while (strcmp(pT, keywords[i]) < 0 && i < KEYWORD_COUNT);
 139         return i;
 140 }
 141
 142 char symbol()
 143 {
 144         return 'z';
 145 }
 146
 147 char *identifier(char *str)
 148 {
 149         str = NULL;
 150         return NULL;
 151 }
 152
 153 int int_val(char pT[])
 154 {
 155         return atoi(pT);
 156 }
 157
 158 char *string_val()
 159 {
 160         return NULL;
 161 }
 162
 163 void token_print(char *s, ptoken print_spec)
 164 {
 165         int i = 0;
 166
 167         while(i < space_count)
 168         {
 169                 printf("  ");
 170                 i++;
 171         }
 172
 173         switch(print_spec)
 174         {
 175                 case OPEN:
 176                         printf("<%s>\r\n", s);
 177                         break;
 178                 case CLOSE:
 179                         printf("</%s>\r\n", s);
 180                         break;
 181                 case BOTH:
 182                         /* convert '<' and '>' to HTML */
 183                         if(*pT == '<')
 184                         {
 185                                 printf("<%s> &lt; </%s>\r\n", s, s);
 186                         } else if (*pT == '>')
 187                         {
 188                                 printf("<%s> &gt; </%s>\r\n", s, s);
 189                         } else if (*pT == '&')
 190                         {
 191                                 printf("<%s> &amp; </%s>\r\n", s, s);
 192                         } else {
 193                                 printf("<%s> %s </%s>\r\n", s, pT, s);
 194                         }
 195                         break;
 196         }
 197 }
 198