preproc.c

汇编编译器的最新版本的源码.买了自己动手写操作系统这本书的人一定要下

		    }
		} else if (c == 'H' || c == 'h' || c == 'X' || c == 'x') {
		    is_hex = true;
		} else if (c == 'P' || c == 'p') {
		    is_float = true;
		    if (*p == '+' || *p == '-')
			p++;
		} else if (isnumchar(c) || c == '_')
		    ; /* just advance */
		else if (c == '.') {
		    /* we need to deal with consequences of the legacy
		       parser, like "1.nolist" being two tokens
		       (TOK_NUMBER, TOK_ID) here; at least give it
		       a shot for now.  In the future, we probably need
		       a flex-based scanner with proper pattern matching
		       to do it as well as it can be done.  Nothing in
		       the world is going to help the person who wants
		       0x123.p16 interpreted as two tokens, though. */
		    r = p;
		    while (*r == '_')
			r++;

		    if (nasm_isdigit(*r) || (is_hex && nasm_isxdigit(*r)) ||
			(!is_hex && (*r == 'e' || *r == 'E')) ||
			(*r == 'p' || *r == 'P')) {
			p = r;
			is_float = true;
		    } else
			break;	/* Terminate the token */
		} else
		    break;
	    }
	    p--;	/* Point to first character beyond number */

	    if (has_e && !is_hex) {
		/* 1e13 is floating-point, but 1e13h is not */
		is_float = true;
	    }

	    type = is_float ? TOK_FLOAT : TOK_NUMBER;
        } else if (nasm_isspace(*p)) {
            type = TOK_WHITESPACE;
            p++;
            while (*p && nasm_isspace(*p))
                p++;
            /*
             * Whitespace just before end-of-line is discarded by
             * pretending it's a comment; whitespace just before a
             * comment gets lumped into the comment.
             */
            if (!*p || *p == ';') {
                type = TOK_COMMENT;
                while (*p)
                    p++;
            }
        } else if (*p == ';') {
            type = TOK_COMMENT;
            while (*p)
                p++;
        } else {
            /*
             * Anything else is an operator of some kind. We check
             * for all the double-character operators (>>, <<, //,
             * %%, <=, >=, ==, !=, <>, &&, ||, ^^), but anything
             * else is a single-character operator.
             */
            type = TOK_OTHER;
            if ((p[0] == '>' && p[1] == '>') ||
                (p[0] == '<' && p[1] == '<') ||
                (p[0] == '/' && p[1] == '/') ||
                (p[0] == '<' && p[1] == '=') ||
                (p[0] == '>' && p[1] == '=') ||
                (p[0] == '=' && p[1] == '=') ||
                (p[0] == '!' && p[1] == '=') ||
                (p[0] == '<' && p[1] == '>') ||
                (p[0] == '&' && p[1] == '&') ||
                (p[0] == '|' && p[1] == '|') ||
                (p[0] == '^' && p[1] == '^')) {
                p++;
            }
            p++;
        }

        /* Handling unterminated string by UNV */
        /*if (type == -1)
           {
           *tail = t = new_Token(NULL, TOK_STRING, line, p-line+1);
           t->text[p-line] = *line;
           tail = &t->next;
           }
           else */
        if (type != TOK_COMMENT) {
            *tail = t = new_Token(NULL, type, line, p - line);
            tail = &t->next;
        }
        line = p;
    }
    return list;
}

/*
 * this function allocates a new managed block of memory and
 * returns a pointer to the block.  The managed blocks are
 * deleted only all at once by the delete_Blocks function.
 */
static void *new_Block(size_t size)
{
    Blocks *b = &blocks;

    /* first, get to the end of the linked list */
    while (b->next)
        b = b->next;
    /* now allocate the requested chunk */
    b->chunk = nasm_malloc(size);

    /* now allocate a new block for the next request */
    b->next = nasm_malloc(sizeof(Blocks));
    /* and initialize the contents of the new block */
    b->next->next = NULL;
    b->next->chunk = NULL;
    return b->chunk;
}

/*
 * this function deletes all managed blocks of memory
 */
static void delete_Blocks(void)
{
    Blocks *a, *b = &blocks;

    /*
     * keep in mind that the first block, pointed to by blocks
     * is a static and not dynamically allocated, so we don't
     * free it.
     */
    while (b) {
        if (b->chunk)
            nasm_free(b->chunk);
        a = b;
        b = b->next;
        if (a != &blocks)
            nasm_free(a);
    }
}

/*
 *  this function creates a new Token and passes a pointer to it
 *  back to the caller.  It sets the type and text elements, and
 *  also the a.mac and next elements to NULL.
 */
static Token *new_Token(Token * next, enum pp_token_type type,
			const char *text, int txtlen)
{
    Token *t;
    int i;

    if (freeTokens == NULL) {
        freeTokens = (Token *) new_Block(TOKEN_BLOCKSIZE * sizeof(Token));
        for (i = 0; i < TOKEN_BLOCKSIZE - 1; i++)
            freeTokens[i].next = &freeTokens[i + 1];
        freeTokens[i].next = NULL;
    }
    t = freeTokens;
    freeTokens = t->next;
    t->next = next;
    t->a.mac = NULL;
    t->type = type;
    if (type == TOK_WHITESPACE || text == NULL) {
        t->text = NULL;
    } else {
        if (txtlen == 0)
            txtlen = strlen(text);
        t->text = nasm_malloc(txtlen+1);
        memcpy(t->text, text, txtlen);
        t->text[txtlen] = '\0';
    }
    return t;
}

static Token *delete_Token(Token * t)
{
    Token *next = t->next;
    nasm_free(t->text);
    t->next = freeTokens;
    freeTokens = t;
    return next;
}

/*
 * Convert a line of tokens back into text.
 * If expand_locals is not zero, identifiers of the form "%$*xxx"
 * will be transformed into [e-mail]
 */
static char *detoken(Token * tlist, bool expand_locals)
{
    Token *t;
    int len;
    char *line, *p;
    const char *q;

    len = 0;
    for (t = tlist; t; t = t->next) {
        if (t->type == TOK_PREPROC_ID && t->text[1] == '!') {
            char *p = getenv(t->text + 2);
            nasm_free(t->text);
            if (p)
                t->text = nasm_strdup(p);
            else
                t->text = NULL;
        }
        /* Expand local macros here and not during preprocessing */
        if (expand_locals &&
            t->type == TOK_PREPROC_ID && t->text &&
            t->text[0] == '%' && t->text[1] == '$') {
            Context *ctx = get_ctx(t->text, false);
            if (ctx) {
                char buffer[40];
                char *p, *q = t->text + 2;

                q += strspn(q, "$");
                snprintf(buffer, sizeof(buffer), "..@%"PRIu32".", ctx->number);
                p = nasm_strcat(buffer, q);
                nasm_free(t->text);
                t->text = p;
            }
        }
        if (t->type == TOK_WHITESPACE) {
            len++;
        } else if (t->text) {
            len += strlen(t->text);
        }
    }
    p = line = nasm_malloc(len + 1);
    for (t = tlist; t; t = t->next) {
        if (t->type == TOK_WHITESPACE) {
            *p++ = ' ';
        } else if (t->text) {
	    q = t->text;
	    while (*q)
		*p++ = *q++;
        }
    }
    *p = '\0';
    return line;
}

/*
 * A scanner, suitable for use by the expression evaluator, which
 * operates on a line of Tokens. Expects a pointer to a pointer to
 * the first token in the line to be passed in as its private_data
 * field.
 *
 * FIX: This really needs to be unified with stdscan.
 */
static int ppscan(void *private_data, struct tokenval *tokval)
{
    Token **tlineptr = private_data;
    Token *tline;
    char ourcopy[MAX_KEYWORD+1], *p, *r, *s;

    do {
        tline = *tlineptr;
        *tlineptr = tline ? tline->next : NULL;
    }
    while (tline && (tline->type == TOK_WHITESPACE ||
                     tline->type == TOK_COMMENT));

    if (!tline)
        return tokval->t_type = TOKEN_EOS;

    tokval->t_charptr = tline->text;

    if (tline->text[0] == '$' && !tline->text[1])
        return tokval->t_type = TOKEN_HERE;
    if (tline->text[0] == '$' && tline->text[1] == '$' && !tline->text[2])
        return tokval->t_type = TOKEN_BASE;

    if (tline->type == TOK_ID) {
        p = tokval->t_charptr = tline->text;
        if (p[0] == '$') {
            tokval->t_charptr++;
            return tokval->t_type = TOKEN_ID;
        }

        for (r = p, s = ourcopy; *r; r++) {
	    if (r >= p+MAX_KEYWORD)
		return tokval->t_type = TOKEN_ID; /* Not a keyword */
            *s++ = nasm_tolower(*r);
	}
        *s = '\0';
        /* right, so we have an identifier sitting in temp storage. now,
         * is it actually a register or instruction name, or what? */
	return nasm_token_hash(ourcopy, tokval);
    }

    if (tline->type == TOK_NUMBER) {
	bool rn_error;
	tokval->t_integer = readnum(tline->text, &rn_error);
	tokval->t_charptr = tline->text;
	if (rn_error)
	    return tokval->t_type = TOKEN_ERRNUM;
	else
	    return tokval->t_type = TOKEN_NUM;
    }

    if (tline->type == TOK_FLOAT) {
	return tokval->t_type = TOKEN_FLOAT;
    }

    if (tline->type == TOK_STRING) {
	char bq, *ep;

	bq = tline->text[0];
        tokval->t_charptr = tline->text;
        tokval->t_inttwo = nasm_unquote(tline->text, &ep);

	if (ep[0] != bq || ep[1] != '\0')
	    return tokval->t_type = TOKEN_ERRSTR;
	else
	    return tokval->t_type = TOKEN_STR;
    }

    if (tline->type == TOK_OTHER) {
        if (!strcmp(tline->text, "<<"))
            return tokval->t_type = TOKEN_SHL;
        if (!strcmp(tline->text, ">>"))
            return tokval->t_type = TOKEN_SHR;
        if (!strcmp(tline->text, "//"))
            return tokval->t_type = TOKEN_SDIV;
        if (!strcmp(tline->text, "%%"))
            return tokval->t_type = TOKEN_SMOD;
        if (!strcmp(tline->text, "=="))
            return tokval->t_type = TOKEN_EQ;
        if (!strcmp(tline->text, "<>"))
            return tokval->t_type = TOKEN_NE;
        if (!strcmp(tline->text, "!="))
            return tokval->t_type = TOKEN_NE;
        if (!strcmp(tline->text, "<="))
            return tokval->t_type = TOKEN_LE;
        if (!strcmp(tline->text, ">="))
            return tokval->t_type = TOKEN_GE;
        if (!strcmp(tline->text, "&&"))
            return tokval->t_type = TOKEN_DBL_AND;
        if (!strcmp(tline->text, "^^"))
            return tokval->t_type = TOKEN_DBL_XOR;
        if (!strcmp(tline->text, "||"))
            return tokval->t_type = TOKEN_DBL_OR;
    }

    /*
     * We have no other options: just return the first character of
     * the token text.
     */
    return tokval->t_type = tline->text[0];
}

/*
 * Compare a string to the name of an existing macro; this is a
 * simple wrapper which calls either strcmp or nasm_stricmp
 * depending on the value of the `casesense' parameter.
 */
static int mstrcmp(const char *p, const char *q, bool casesense)
{
    return casesense ? strcmp(p, q) : nasm_stricmp(p, q);
}

/*
 * Compare a string to the name of an existing macro; this is a
 * simple wrapper which calls either strcmp or nasm_stricmp
 * depending on the value of the `casesense' parameter.
 */
static int mmemcmp(const char *p, const char *q, size_t l, bool casesense)
{
    return casesense ? memcmp(p, q, l) : nasm_memicmp(p, q, l);
}

/*
 * Return the Context structure associated with a %$ token. Return
 * NULL, having _already_ reported an error condition, if the
 * context stack isn't deep enough for the supplied number of $
 * signs.
 * If all_contexts == true, contexts that enclose current are
 * also scanned for such smacro, until it is found; if not -
 * only the context that directly results from the number of $'s
 * in variable's name.
 */
static Context *get_ctx(const char *name, bool all_contexts)
{
    Context *ctx;
    SMacro *m;
    int i;

    if (!name || name[0] != '%' || name[1] != '$')
        return NULL;

    if (!cstk) {
        error(ERR_NONFATAL, "`%s': context stack is empty", name);
        return NULL;
    }

    for (i = strspn(name + 2, "$"), ctx = cstk; (i > 0) && ctx; i--) {
        ctx = ctx->next;
/*        i--;  Lino - 02/25/02 */
    }
    if (!ctx) {
        error(ERR_NONFATAL, "`%s': context stack is only"
              " %d level%s deep", name, i - 1, (i == 2 ? "" : "s"));
        return NULL;
    }
    if (!all_contexts)
        return ctx;

    do {
        /* Search for this smacro in found context */
        m = hash_findix(&ctx->localmac, name);
        while (m) {
            if (!mstrcmp(m->name, name, m->casesense))
                return ctx;
            m = m->next;
        }
        ctx = ctx->next;
    }
    while (ctx);
    return NULL;
}

/*
 * Check to see if a file is already in a string list
 */
static bool in_list(const StrList *list, const char *str)
{
    while (list) {
	if (!strcmp(list->str, str))
	    return true;
	list = list->next;