stdscan.c

   1 #include "compiler.h"
   2
   3 #include <stdio.h>
   4 #include <stdlib.h>
   5 #include <string.h>
   6 #include <ctype.h>
   7 #include <inttypes.h>
   8
   9 #include "nasm.h"
  10 #include "nasmlib.h"
  11 #include "quote.h"
  12 #include "stdscan.h"
  13 #include "insns.h"
  14
  15 /*
  16  * Standard scanner routine used by parser.c and some output
  17  * formats. It keeps a succession of temporary-storage strings in
  18  * stdscan_tempstorage, which can be cleared using stdscan_reset.
  19  */
  20 static char **stdscan_tempstorage = NULL;
  21 static int stdscan_tempsize = 0, stdscan_templen = 0;
  22 #define STDSCAN_TEMP_DELTA 256
  23
  24 static void stdscan_pop(void)
  25 {
  26     nasm_free(stdscan_tempstorage[--stdscan_templen]);
  27 }
  28
  29 void stdscan_reset(void)
  30 {
  31     while (stdscan_templen > 0)
  32         stdscan_pop();
  33 }
  34
  35 /*
  36  * Unimportant cleanup is done to avoid confusing people who are trying
  37  * to debug real memory leaks
  38  */
  39 void stdscan_cleanup(void)
  40 {
  41     stdscan_reset();
  42     nasm_free(stdscan_tempstorage);
  43 }
  44
  45 static char *stdscan_copy(char *p, int len)
  46 {
  47     char *text;
  48
  49     text = nasm_malloc(len + 1);
  50     memcpy(text, p, len);
  51     text[len] = '\0';
  52
  53     if (stdscan_templen >= stdscan_tempsize) {
  54         stdscan_tempsize += STDSCAN_TEMP_DELTA;
  55         stdscan_tempstorage = nasm_realloc(stdscan_tempstorage,
  56                                            stdscan_tempsize *
  57                                            sizeof(char *));
  58     }
  59     stdscan_tempstorage[stdscan_templen++] = text;
  60
  61     return text;
  62 }
  63
  64 char *stdscan_bufptr = NULL;
  65 int stdscan(void *private_data, struct tokenval *tv)
  66 {
  67     char ourcopy[MAX_KEYWORD + 1], *r, *s;
  68
  69     (void)private_data;         /* Don't warn that this parameter is unused */
  70
  71     while (nasm_isspace(*stdscan_bufptr))
  72         stdscan_bufptr++;
  73     if (!*stdscan_bufptr)
  74         return tv->t_type = 0;
  75
  76     /* we have a token; either an id, a number or a char */
  77     if (isidstart(*stdscan_bufptr) ||
  78         (*stdscan_bufptr == '$' && isidstart(stdscan_bufptr[1]))) {
  79         /* now we've got an identifier */
  80         bool is_sym = false;
  81
  82         if (*stdscan_bufptr == '$') {
  83             is_sym = true;
  84             stdscan_bufptr++;
  85         }
  86
  87         r = stdscan_bufptr++;
  88         /* read the entire buffer to advance the buffer pointer but... */
  89         while (isidchar(*stdscan_bufptr))
  90             stdscan_bufptr++;
  91
  92         /* ... copy only up to IDLEN_MAX-1 characters */
  93         tv->t_charptr = stdscan_copy(r, stdscan_bufptr - r < IDLEN_MAX ?
  94                                      stdscan_bufptr - r : IDLEN_MAX - 1);
  95
  96         if (is_sym || stdscan_bufptr - r > MAX_KEYWORD)
  97             return tv->t_type = TOKEN_ID;       /* bypass all other checks */
  98
  99         for (s = tv->t_charptr, r = ourcopy; *s; s++)
 100             *r++ = nasm_tolower(*s);
 101         *r = '\0';
 102         /* right, so we have an identifier sitting in temp storage. now,
 103          * is it actually a register or instruction name, or what? */
 104         return nasm_token_hash(ourcopy, tv);
 105     } else if (*stdscan_bufptr == '$' && !isnumchar(stdscan_bufptr[1])) {
 106         /*
 107          * It's a $ sign with no following hex number; this must
 108          * mean it's a Here token ($), evaluating to the current
 109          * assembly location, or a Base token ($$), evaluating to
 110          * the base of the current segment.
 111          */
 112         stdscan_bufptr++;
 113         if (*stdscan_bufptr == '$') {
 114             stdscan_bufptr++;
 115             return tv->t_type = TOKEN_BASE;
 116         }
 117         return tv->t_type = TOKEN_HERE;
 118     } else if (isnumstart(*stdscan_bufptr)) {   /* now we've got a number */
 119         bool rn_error;
 120         bool is_hex = false;
 121         bool is_float = false;
 122         bool has_e = false;
 123         char c;
 124
 125         r = stdscan_bufptr;
 126
 127         if (*stdscan_bufptr == '$') {
 128             stdscan_bufptr++;
 129             is_hex = true;
 130         }
 131
 132         for (;;) {
 133             c = *stdscan_bufptr++;
 134
 135             if (!is_hex && (c == 'e' || c == 'E')) {
 136                 has_e = true;
 137                 if (*stdscan_bufptr == '+' || *stdscan_bufptr == '-') {
 138                     /* e can only be followed by +/- if it is either a
 139                        prefixed hex number or a floating-point number */
 140                     is_float = true;
 141                     stdscan_bufptr++;
 142                 }
 143             } else if (c == 'H' || c == 'h' || c == 'X' || c == 'x') {
 144                 is_hex = true;
 145             } else if (c == 'P' || c == 'p') {
 146                 is_float = true;
 147                 if (*stdscan_bufptr == '+' || *stdscan_bufptr == '-')
 148                     stdscan_bufptr++;
 149             } else if (isnumchar(c) || c == '_')
 150                 ; /* just advance */
 151             else if (c == '.')
 152                 is_float = true;
 153             else
 154                 break;
 155         }
 156         stdscan_bufptr--;       /* Point to first character beyond number */
 157
 158         if (has_e && !is_hex) {
 159             /* 1e13 is floating-point, but 1e13h is not */
 160             is_float = true;
 161         }
 162
 163         if (is_float) {
 164             tv->t_charptr = stdscan_copy(r, stdscan_bufptr - r);
 165             return tv->t_type = TOKEN_FLOAT;
 166         } else {
 167             r = stdscan_copy(r, stdscan_bufptr - r);
 168             tv->t_integer = readnum(r, &rn_error);
 169             stdscan_pop();
 170             if (rn_error) {
 171                 /* some malformation occurred */
 172                 return tv->t_type = TOKEN_ERRNUM;
 173             }
 174             tv->t_charptr = NULL;
 175             return tv->t_type = TOKEN_NUM;
 176         }
 177     } else if (*stdscan_bufptr == '\'' || *stdscan_bufptr == '"' ||
 178                *stdscan_bufptr == '`') {
 179         /* a quoted string */
 180         char start_quote = *stdscan_bufptr;
 181         tv->t_charptr = stdscan_bufptr;
 182         tv->t_inttwo = nasm_unquote(tv->t_charptr, &stdscan_bufptr);
 183         if (*stdscan_bufptr != start_quote)
 184             return tv->t_type = TOKEN_ERRSTR;
 185         stdscan_bufptr++;       /* Skip final quote */
 186         return tv->t_type = TOKEN_STR;
 187     } else if (*stdscan_bufptr == ';') {
 188         /* a comment has happened - stay */
 189         return tv->t_type = 0;
 190     } else if (stdscan_bufptr[0] == '>' && stdscan_bufptr[1] == '>') {
 191         stdscan_bufptr += 2;
 192         return tv->t_type = TOKEN_SHR;
 193     } else if (stdscan_bufptr[0] == '<' && stdscan_bufptr[1] == '<') {
 194         stdscan_bufptr += 2;
 195         return tv->t_type = TOKEN_SHL;
 196     } else if (stdscan_bufptr[0] == '/' && stdscan_bufptr[1] == '/') {
 197         stdscan_bufptr += 2;
 198         return tv->t_type = TOKEN_SDIV;
 199     } else if (stdscan_bufptr[0] == '%' && stdscan_bufptr[1] == '%') {
 200         stdscan_bufptr += 2;
 201         return tv->t_type = TOKEN_SMOD;
 202     } else if (stdscan_bufptr[0] == '=' && stdscan_bufptr[1] == '=') {
 203         stdscan_bufptr += 2;
 204         return tv->t_type = TOKEN_EQ;
 205     } else if (stdscan_bufptr[0] == '<' && stdscan_bufptr[1] == '>') {
 206         stdscan_bufptr += 2;
 207         return tv->t_type = TOKEN_NE;
 208     } else if (stdscan_bufptr[0] == '!' && stdscan_bufptr[1] == '=') {
 209         stdscan_bufptr += 2;
 210         return tv->t_type = TOKEN_NE;
 211     } else if (stdscan_bufptr[0] == '<' && stdscan_bufptr[1] == '=') {
 212         stdscan_bufptr += 2;
 213         return tv->t_type = TOKEN_LE;
 214     } else if (stdscan_bufptr[0] == '>' && stdscan_bufptr[1] == '=') {
 215         stdscan_bufptr += 2;
 216         return tv->t_type = TOKEN_GE;
 217     } else if (stdscan_bufptr[0] == '&' && stdscan_bufptr[1] == '&') {
 218         stdscan_bufptr += 2;
 219         return tv->t_type = TOKEN_DBL_AND;
 220     } else if (stdscan_bufptr[0] == '^' && stdscan_bufptr[1] == '^') {
 221         stdscan_bufptr += 2;
 222         return tv->t_type = TOKEN_DBL_XOR;
 223     } else if (stdscan_bufptr[0] == '|' && stdscan_bufptr[1] == '|') {
 224         stdscan_bufptr += 2;
 225         return tv->t_type = TOKEN_DBL_OR;
 226     } else                      /* just an ordinary char */
 227         return tv->t_type = (uint8_t)(*stdscan_bufptr++);
 228 }