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