Пример #1
0
VALUE
rb_enc_associate_index(VALUE obj, int idx)
{
    rb_encoding *enc;
    int oldidx, oldtermlen, termlen;

/*    enc_check_capable(obj);*/
    rb_check_frozen(obj);
    oldidx = rb_enc_get_index(obj);
    if (oldidx == idx)
	return obj;
    if (SPECIAL_CONST_P(obj)) {
	rb_raise(rb_eArgError, "cannot set encoding");
    }
    enc = must_encindex(idx);
    if (!ENC_CODERANGE_ASCIIONLY(obj) ||
	!rb_enc_asciicompat(enc)) {
	ENC_CODERANGE_CLEAR(obj);
    }
    termlen = rb_enc_mbminlen(enc);
    oldtermlen = rb_enc_mbminlen(rb_enc_from_index(oldidx));
    if (oldtermlen < termlen && RB_TYPE_P(obj, T_STRING)) {
	rb_str_fill_terminator(obj, termlen);
    }
    enc_set_index(obj, idx);
    return obj;
}
Пример #2
0
/*
 * Read a string and then force the encoding to UTF 8 if running ruby 1.9
 */
VALUE des_read_string(AMF_DESERIALIZER *des, unsigned long len) {
    DES_BOUNDS_CHECK(des, len);
    VALUE str = rb_str_new(des->stream + des->pos, len);
#ifdef HAVE_RB_STR_ENCODE
    rb_encoding *utf8 = rb_utf8_encoding();
    rb_enc_associate(str, utf8);
    ENC_CODERANGE_CLEAR(str);
#endif
    des->pos += len;
    return str;
}
Пример #3
0
VALUE
rb_enc_associate_index(VALUE obj, int idx)
{
/*    enc_check_capable(obj);*/
    if (rb_enc_get_index(obj) == idx)
	return obj;
    if (SPECIAL_CONST_P(obj)) {
	rb_raise(rb_eArgError, "cannot set encoding");
    }
    if (!ENC_CODERANGE_ASCIIONLY(obj) ||
	!rb_enc_asciicompat(rb_enc_from_index(idx))) {
	ENC_CODERANGE_CLEAR(obj);
    }
    rb_enc_set_index(obj, idx);
    return obj;
}
Пример #4
0
VALUE
rb_file_expand_path_internal(VALUE fname, VALUE dname, int abs_mode, int long_name, VALUE result)
{
    size_t size = 0, wpath_len = 0, wdir_len = 0, whome_len = 0;
    size_t buffer_len = 0;
    char *fullpath = NULL;
    wchar_t *wfullpath = NULL, *wpath = NULL, *wpath_pos = NULL;
    wchar_t *wdir = NULL, *wdir_pos = NULL;
    wchar_t *whome = NULL, *buffer = NULL, *buffer_pos = NULL;
    UINT path_cp, cp;
    VALUE path = fname, dir = dname;
    wchar_t wfullpath_buffer[PATH_BUFFER_SIZE];
    wchar_t path_drive = L'\0', dir_drive = L'\0';
    int ignore_dir = 0;
    rb_encoding *path_encoding;
    int tainted = 0;

    /* tainted if path is tainted */
    tainted = OBJ_TAINTED(path);

    /* get path encoding */
    if (NIL_P(dir)) {
	path_encoding = rb_enc_get(path);
    }
    else {
	path_encoding = rb_enc_check(path, dir);
    }

    cp = path_cp = code_page(path_encoding);

    /* workaround invalid codepage */
    if (path_cp == INVALID_CODE_PAGE) {
	cp = CP_UTF8;
	if (!NIL_P(path)) {
	    path = fix_string_encoding(path, path_encoding);
	}
    }

    /* convert char * to wchar_t */
    convert_mb_to_wchar(path, &wpath, &wpath_pos, &wpath_len, cp);

    /* determine if we need the user's home directory */
    /* expand '~' only if NOT rb_file_absolute_path() where `abs_mode` is 1 */
    if (abs_mode == 0 && wpath_len > 0 && wpath_pos[0] == L'~' &&
	(wpath_len == 1 || IS_DIR_SEPARATOR_P(wpath_pos[1]))) {
	/* tainted if expanding '~' */
	tainted = 1;

	whome = home_dir();
	if (whome == NULL) {
	    xfree(wpath);
	    rb_raise(rb_eArgError, "couldn't find HOME environment -- expanding `~'");
	}
	whome_len = wcslen(whome);

	if (PathIsRelativeW(whome) && !(whome_len >= 2 && IS_DIR_UNC_P(whome))) {
	    xfree(wpath);
	    rb_raise(rb_eArgError, "non-absolute home");
	}

	/* use filesystem encoding if expanding home dir */
	path_encoding = rb_filesystem_encoding();
	cp = path_cp = system_code_page();

	/* ignores dir since we are expading home */
	ignore_dir = 1;

	/* exclude ~ from the result */
	wpath_pos++;
	wpath_len--;

	/* exclude separator if present */
	if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
	    wpath_pos++;
	    wpath_len--;
	}
    }
    else if (wpath_len >= 2 && wpath_pos[1] == L':') {
	if (wpath_len >= 3 && IS_DIR_SEPARATOR_P(wpath_pos[2])) {
	    /* ignore dir since path contains a drive letter and a root slash */
	    ignore_dir = 1;
	}
	else {
	    /* determine if we ignore dir or not later */
	    path_drive = wpath_pos[0];
	}
    }
    else if (abs_mode == 0 && wpath_len >= 2 && wpath_pos[0] == L'~') {
	result = get_user_from_path(&wpath_pos, 1, cp, path_cp, path_encoding);

	if (wpath)
	    xfree(wpath);

	rb_raise(rb_eArgError, "can't find user %s", StringValuePtr(result));
    }

    /* convert dir */
    if (!ignore_dir && !NIL_P(dir)) {
	/* fix string encoding */
	if (path_cp == INVALID_CODE_PAGE) {
	    dir = fix_string_encoding(dir, path_encoding);
	}

	/* convert char * to wchar_t */
	convert_mb_to_wchar(dir, &wdir, &wdir_pos, &wdir_len, cp);

	if (abs_mode == 0 && wdir_len > 0 && wdir_pos[0] == L'~' &&
	    (wdir_len == 1 || IS_DIR_SEPARATOR_P(wdir_pos[1]))) {
	    /* tainted if expanding '~' */
	    tainted = 1;

	    whome = home_dir();
	    if (whome == NULL) {
		xfree(wpath);
		xfree(wdir);
		rb_raise(rb_eArgError, "couldn't find HOME environment -- expanding `~'");
	    }
	    whome_len = wcslen(whome);

	    if (PathIsRelativeW(whome) && !(whome_len >= 2 && IS_DIR_UNC_P(whome))) {
		xfree(wpath);
		xfree(wdir);
		rb_raise(rb_eArgError, "non-absolute home");
	    }

	    /* exclude ~ from the result */
	    wdir_pos++;
	    wdir_len--;

	    /* exclude separator if present */
	    if (wdir_len && IS_DIR_SEPARATOR_P(wdir_pos[0])) {
		wdir_pos++;
		wdir_len--;
	    }
	}
	else if (wdir_len >= 2 && wdir[1] == L':') {
	    dir_drive = wdir[0];
	    if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
		wdir_len = 2;
	    }
	}
	else if (wdir_len >= 2 && IS_DIR_UNC_P(wdir)) {
	    /* UNC path */
	    if (wpath_len && IS_DIR_SEPARATOR_P(wpath_pos[0])) {
		/* cut the UNC path tail to '//host/share' */
		size_t separators = 0;
		size_t pos = 2;
		while (pos < wdir_len && separators < 2) {
		    if (IS_DIR_SEPARATOR_P(wdir[pos])) {
			separators++;
		    }
		    pos++;
		}
		if (separators == 2)
		    wdir_len = pos - 1;
	    }
	}
	else if (abs_mode == 0 && wdir_len >= 2 && wdir_pos[0] == L'~') {
	    result = get_user_from_path(&wdir_pos, 1, cp, path_cp, path_encoding);
	    if (wpath)
		xfree(wpath);

	    if (wdir)
		xfree(wdir);

	    rb_raise(rb_eArgError, "can't find user %s", StringValuePtr(result));
	}
    }

    /* determine if we ignore dir or not */
    if (!ignore_dir && path_drive && dir_drive) {
	if (towupper(path_drive) == towupper(dir_drive)) {
	    /* exclude path drive letter to use dir */
	    wpath_pos += 2;
	    wpath_len -= 2;
	}
	else {
	    /* ignore dir since path drive is different from dir drive */
	    ignore_dir = 1;
	    wdir_len = 0;
	}
    }

    if (!ignore_dir && wpath_len >= 2 && IS_DIR_UNC_P(wpath)) {
	/* ignore dir since path has UNC root */
	ignore_dir = 1;
	wdir_len = 0;
    }
    else if (!ignore_dir && wpath_len >= 1 && IS_DIR_SEPARATOR_P(wpath[0]) &&
	     !dir_drive && !(wdir_len >= 2 && IS_DIR_UNC_P(wdir))) {
	/* ignore dir since path has root slash and dir doesn't have drive or UNC root */
	ignore_dir = 1;
	wdir_len = 0;
    }

    buffer_len = wpath_len + 1 + wdir_len + 1 + whome_len + 1;

    buffer = buffer_pos = (wchar_t *)xmalloc((buffer_len + 1) * sizeof(wchar_t));

    /* add home */
    if (whome_len) {
	wcsncpy(buffer_pos, whome, whome_len);
	buffer_pos += whome_len;
    }

    /* Add separator if required */
    if (whome_len && wcsrchr(L"\\/:", buffer_pos[-1]) == NULL) {
	buffer_pos[0] = L'\\';
	buffer_pos++;
    }

    if (wdir_len) {
	/* tainted if dir is used and dir is tainted */
	if (!tainted && OBJ_TAINTED(dir))
	    tainted = 1;

	wcsncpy(buffer_pos, wdir_pos, wdir_len);
	buffer_pos += wdir_len;
    }

    /* add separator if required */
    if (wdir_len && wcsrchr(L"\\/:", buffer_pos[-1]) == NULL) {
	buffer_pos[0] = L'\\';
	buffer_pos++;
    }

    /* now deal with path */
    if (wpath_len) {
	wcsncpy(buffer_pos, wpath_pos, wpath_len);
	buffer_pos += wpath_len;
    }

    /* GetFullPathNameW requires at least "." to determine current directory */
    if (wpath_len == 0) {
	buffer_pos[0] = L'.';
	buffer_pos++;
    }

    /* Ensure buffer is NULL terminated */
    buffer_pos[0] = L'\0';

    /* tainted if path is relative */
    if (!tainted && PathIsRelativeW(buffer) && !(buffer_len >= 2 && IS_DIR_UNC_P(buffer)))
	tainted = 1;

    /* FIXME: Make this more robust */
    /* Determine require buffer size */
    size = GetFullPathNameW(buffer, PATH_BUFFER_SIZE, wfullpath_buffer, NULL);
    if (size > PATH_BUFFER_SIZE) {
	/* allocate more memory than alloted originally by PATH_BUFFER_SIZE */
	wfullpath = (wchar_t *)xmalloc(size * sizeof(wchar_t));
	size = GetFullPathNameW(buffer, size, wfullpath, NULL);
    }
    else {
	wfullpath = wfullpath_buffer;
    }

    /* Remove any trailing slashes */
    if (IS_DIR_SEPARATOR_P(wfullpath[size - 1]) &&
	wfullpath[size - 2] != L':' &&
	!(size == 2 && IS_DIR_UNC_P(wfullpath))) {
	size -= 1;
	wfullpath[size] = L'\0';
    }

    /* Remove any trailing dot */
    if (wfullpath[size - 1] == L'.') {
	size -= 1;
	wfullpath[size] = L'\0';
    }

    /* removes trailing invalid ':$DATA' */
    size = remove_invalid_alternative_data(wfullpath, size);

    /* Replace the trailing path to long name */
    if (long_name)
	size = replace_to_long_name(&wfullpath, size, (wfullpath != wfullpath_buffer));

    /* sanitize backslashes with forwardslashes */
    replace_wchar(wfullpath, L'\\', L'/');

    /* convert to char * */
    size = WideCharToMultiByte(cp, 0, wfullpath, size, NULL, 0, NULL, NULL);
    if (size > (size_t)RSTRING_LEN(result)) {
	rb_str_modify(result);
	rb_str_resize(result, size);
    }

    WideCharToMultiByte(cp, 0, wfullpath, size, RSTRING_PTR(result), size, NULL, NULL);
    rb_str_set_len(result, size);

    /* convert to VALUE and set the path encoding */
    if (path_cp == INVALID_CODE_PAGE) {
	VALUE tmp;
	size_t len;

	rb_enc_associate(result, rb_utf8_encoding());
	ENC_CODERANGE_CLEAR(result);
	tmp = rb_str_encode(result, rb_enc_from_encoding(path_encoding), 0, Qnil);
	len = RSTRING_LEN(tmp);
	rb_str_modify(result);
	rb_str_resize(result, len);
	memcpy(RSTRING_PTR(result), RSTRING_PTR(tmp), len);
	rb_str_resize(tmp, 0);
    }
    rb_enc_associate(result, path_encoding);
    ENC_CODERANGE_CLEAR(result);

    /* makes the result object tainted if expanding tainted strings or returning modified path */
    if (tainted)
	OBJ_TAINT(result);

    /* TODO: better cleanup */
    if (buffer)
	xfree(buffer);

    if (wpath)
	xfree(wpath);

    if (wdir)
	xfree(wdir);

    if (whome)
	xfree(whome);

    if (wfullpath && wfullpath != wfullpath_buffer)
	xfree(wfullpath);

    if (fullpath)
	xfree(fullpath);

    return result;
}