stdscan.c

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