/*
* Given a string "str", separate the numeric part into
* str, and the modifier into mod.
*/
static bool get_modifier(char *str, char *num, int num_len, char *mod, int mod_len)
{
int i, len, num_begin, num_end, mod_begin, mod_end;
strip_trailing_junk(str);
len = strlen(str);
for (i=0; i<len; i++) {
if (!B_ISSPACE(str[i])) {
break;
}
}
num_begin = i;
/* Walk through integer part */
for ( ; i<len; i++) {
if (!B_ISDIGIT(str[i]) && str[i] != '.') {
break;
}
}
num_end = i;
if (num_len > (num_end - num_begin + 1)) {
num_len = num_end - num_begin + 1;
}
if (num_len == 0) {
return false;
}
/* Eat any spaces in front of modifier */
for ( ; i<len; i++) {
if (!B_ISSPACE(str[i])) {
break;
}
}
mod_begin = i;
for ( ; i<len; i++) {
if (!B_ISALPHA(str[i])) {
break;
}
}
mod_end = i;
if (mod_len > (mod_end - mod_begin + 1)) {
mod_len = mod_end - mod_begin + 1;
}
Dmsg5(900, "str=%s: num_beg=%d num_end=%d mod_beg=%d mod_end=%d\n",
str, num_begin, num_end, mod_begin, mod_end);
bstrncpy(num, &str[num_begin], num_len);
bstrncpy(mod, &str[mod_begin], mod_len);
if (!is_a_number(num)) {
return false;
}
bstrncpy(str, &str[mod_end], len);
Dmsg2(900, "num=%s mod=%s\n", num, mod);
return true;
}
/* We assume ASCII input and don't worry about overflow */
uint64_t str_to_uint64(const char *str)
{
const char *p = str;
uint64_t value = 0;
if (!p) {
return 0;
}
while (B_ISSPACE(*p)) {
p++;
}
if (*p == '+') {
p++;
}
while (B_ISDIGIT(*p)) {
value = B_TIMES10(value) + *p - '0';
p++;
}
return value;
}
int64_t str_to_int64(const char *str)
{
const char *p = str;
int64_t value;
bool negative = false;
if (!p) {
return 0;
}
while (B_ISSPACE(*p)) {
p++;
}
if (*p == '+') {
p++;
} else if (*p == '-') {
negative = true;
p++;
}
value = str_to_uint64(p);
if (negative) {
value = -value;
}
return value;
}
/*
*
* Get the next token from the input
*
*/
int lex_get_token(LEX *lf, int expect)
{
int ch;
int token = T_NONE;
bool esc_next = false;
/* Unicode files, especially on Win32, may begin with a "Byte Order Mark"
to indicate which transmission format the file is in. The codepoint for
this mark is U+FEFF and is represented as the octets EF-BB-BF in UTF-8
and as FF-FE in UTF-16le(little endian) and FE-FF in UTF-16(big endian).
We use a distinct state for UTF-8 and UTF-16le, and use bom_bytes_seen
to tell which byte we are expecting. */
int bom_bytes_seen = 0;
Dmsg0(dbglvl, "enter lex_get_token\n");
while (token == T_NONE) {
ch = lex_get_char(lf);
switch (lf->state) {
case lex_none:
Dmsg2(dbglvl, "Lex state lex_none ch=%d,%x\n", ch, ch);
if (B_ISSPACE(ch))
break;
if (B_ISALPHA(ch)) {
if (lf->options & LOPT_NO_IDENT || lf->options & LOPT_STRING) {
lf->state = lex_string;
} else {
lf->state = lex_identifier;
}
begin_str(lf, ch);
break;
}
if (B_ISDIGIT(ch)) {
if (lf->options & LOPT_STRING) {
lf->state = lex_string;
} else {
lf->state = lex_number;
}
begin_str(lf, ch);
break;
}
Dmsg0(dbglvl, "Enter lex_none switch\n");
switch (ch) {
case L_EOF:
token = T_EOF;
Dmsg0(dbglvl, "got L_EOF set token=T_EOF\n");
break;
case '#':
lf->state = lex_comment;
break;
case '{':
token = T_BOB;
begin_str(lf, ch);
break;
case '}':
token = T_EOB;
begin_str(lf, ch);
break;
case '"':
lf->state = lex_quoted_string;
begin_str(lf, 0);
break;
case '=':
token = T_EQUALS;
begin_str(lf, ch);
break;
case ',':
token = T_COMMA;
begin_str(lf, ch);
break;
case ';':
if (expect != T_SKIP_EOL) {
token = T_EOL; /* treat ; like EOL */
}
break;
case L_EOL:
Dmsg0(dbglvl, "got L_EOL set token=T_EOL\n");
if (expect != T_SKIP_EOL) {
token = T_EOL;
}
break;
case '@':
/* In NO_EXTERN mode, @ is part of a string */
if (lf->options & LOPT_NO_EXTERN) {
lf->state = lex_string;
begin_str(lf, ch);
} else {
lf->state = lex_include;
begin_str(lf, 0);
}
break;
case 0xEF: /* probably a UTF-8 BOM */
case 0xFF: /* probably a UTF-16le BOM */
case 0xFE: /* probably a UTF-16be BOM (error)*/
if (lf->line_no != 1 || lf->col_no != 1)
{
lf->state = lex_string;
begin_str(lf, ch);
} else {
bom_bytes_seen = 1;
if (ch == 0xEF) {
lf->state = lex_utf8_bom;
} else if (ch == 0xFF) {
lf->state = lex_utf16_le_bom;
} else {
scan_err0(lf, _("This config file appears to be in an "
"unsupported Unicode format (UTF-16be). Please resave as UTF-8\n"));
return T_ERROR;
}
}
break;
default:
lf->state = lex_string;
begin_str(lf, ch);
break;
}
break;
case lex_comment:
Dmsg1(dbglvl, "Lex state lex_comment ch=%x\n", ch);
if (ch == L_EOL) {
lf->state = lex_none;
if (expect != T_SKIP_EOL) {
token = T_EOL;
}
} else if (ch == L_EOF) {
token = T_ERROR;
}
break;
case lex_number:
Dmsg2(dbglvl, "Lex state lex_number ch=%x %c\n", ch, ch);
if (ch == L_EOF) {
token = T_ERROR;
break;
}
/* Might want to allow trailing specifications here */
if (B_ISDIGIT(ch)) {
add_str(lf, ch);
break;
}
/* A valid number can be terminated by the following */
if (B_ISSPACE(ch) || ch == L_EOL || ch == ',' || ch == ';') {
token = T_NUMBER;
lf->state = lex_none;
} else {
lf->state = lex_string;
}
lex_unget_char(lf);
break;
case lex_ip_addr:
if (ch == L_EOF) {
token = T_ERROR;
break;
}
Dmsg1(dbglvl, "Lex state lex_ip_addr ch=%x\n", ch);
break;
case lex_string:
Dmsg1(dbglvl, "Lex state lex_string ch=%x\n", ch);
if (ch == L_EOF) {
token = T_ERROR;
break;
}
if (ch == '\n' || ch == L_EOL || ch == '=' || ch == '}' || ch == '{' ||
ch == '\r' || ch == ';' || ch == ',' || ch == '#' || (B_ISSPACE(ch)) ) {
lex_unget_char(lf);
token = T_UNQUOTED_STRING;
lf->state = lex_none;
break;
}
add_str(lf, ch);
break;
case lex_identifier:
Dmsg2(dbglvl, "Lex state lex_identifier ch=%x %c\n", ch, ch);
if (B_ISALPHA(ch)) {
add_str(lf, ch);
break;
} else if (B_ISSPACE(ch)) {
break;
} else if (ch == '\n' || ch == L_EOL || ch == '=' || ch == '}' || ch == '{' ||
ch == '\r' || ch == ';' || ch == ',' || ch == '"' || ch == '#') {
lex_unget_char(lf);
token = T_IDENTIFIER;
lf->state = lex_none;
break;
} else if (ch == L_EOF) {
token = T_ERROR;
lf->state = lex_none;
begin_str(lf, ch);
break;
}
/* Some non-alpha character => string */
lf->state = lex_string;
add_str(lf, ch);
break;
case lex_quoted_string:
Dmsg2(dbglvl, "Lex state lex_quoted_string ch=%x %c\n", ch, ch);
if (ch == L_EOF) {
token = T_ERROR;
break;
}
if (ch == L_EOL) {
esc_next = false;
break;
}
if (esc_next) {
add_str(lf, ch);
esc_next = false;
break;
}
if (ch == '\\') {
esc_next = true;
break;
}
if (ch == '"') {
token = T_QUOTED_STRING;
/*
* Since we may be scanning a quoted list of names,
* we get the next character (a comma indicates another
* one), then we put it back for rescanning.
*/
lex_get_char(lf);
lex_unget_char(lf);
lf->state = lex_none;
break;
}
add_str(lf, ch);
break;
case lex_include_quoted_string:
if (ch == L_EOF) {
token = T_ERROR;
break;
}
if (esc_next) {
add_str(lf, ch);
esc_next = false;
break;
}
if (ch == '\\') {
esc_next = true;
break;
}
if (ch == '"') {
/* Keep the original LEX so we can print an error if the included file can't be opened. */
LEX* lfori = lf;
/* Skip the double quote when restarting parsing */
lex_get_char(lf);
lf->state = lex_none;
lf = lex_open_file(lf, lf->str, lf->scan_error, lf->scan_warning);
if (lf == NULL) {
berrno be;
scan_err2(lfori, _("Cannot open included config file %s: %s\n"),
lfori->str, be.bstrerror());
return T_ERROR;
}
break;
}
add_str(lf, ch);
break;
case lex_include: /* scanning a filename */
if (ch == L_EOF) {
token = T_ERROR;
break;
}
if (ch == '"') {
lf->state = lex_include_quoted_string;
break;
}
if (B_ISSPACE(ch) || ch == '\n' || ch == L_EOL || ch == '}' || ch == '{' ||
ch == ';' || ch == ',' || ch == '"' || ch == '#') {
/* Keep the original LEX so we can print an error if the included file can't be opened. */
LEX* lfori = lf;
lf->state = lex_none;
lf = lex_open_file(lf, lf->str, lf->scan_error, lf->scan_warning);
if (lf == NULL) {
berrno be;
scan_err2(lfori, _("Cannot open included config file %s: %s\n"),
lfori->str, be.bstrerror());
return T_ERROR;
}
break;
}
add_str(lf, ch);
break;
case lex_utf8_bom:
/* we only end up in this state if we have read an 0xEF
as the first byte of the file, indicating we are probably
reading a UTF-8 file */
if (ch == 0xBB && bom_bytes_seen == 1) {
bom_bytes_seen++;
} else if (ch == 0xBF && bom_bytes_seen == 2) {
token = T_UTF8_BOM;
lf->state = lex_none;
} else {
token = T_ERROR;
}
break;
case lex_utf16_le_bom:
/* we only end up in this state if we have read an 0xFF
as the first byte of the file -- indicating that we are
probably dealing with an Intel based (little endian) UTF-16 file*/
if (ch == 0xFE) {
token = T_UTF16_BOM;
lf->state = lex_none;
} else {
token = T_ERROR;
}
break;
}
Dmsg4(dbglvl, "ch=%d state=%s token=%s %c\n", ch, lex_state_to_str(lf->state),
lex_tok_to_str(token), ch);
}
Dmsg2(dbglvl, "lex returning: line %d token: %s\n", lf->line_no, lex_tok_to_str(token));
lf->token = token;
/*
* Here is where we check to see if the user has set certain
* expectations (e.g. 32 bit integer). If so, we do type checking
* and possible additional scanning (e.g. for range).
*/
switch (expect) {
case T_PINT16:
lf->u.pint16_val = (scan_pint(lf, lf->str) & 0xffff);
lf->u2.pint16_val = lf->u.pint16_val;
token = T_PINT16;
break;
case T_PINT32:
lf->u.pint32_val = scan_pint(lf, lf->str);
lf->u2.pint32_val = lf->u.pint32_val;
token = T_PINT32;
break;
case T_PINT32_RANGE:
if (token == T_NUMBER) {
lf->u.pint32_val = scan_pint(lf, lf->str);
lf->u2.pint32_val = lf->u.pint32_val;
token = T_PINT32;
} else {
char *p = strchr(lf->str, '-');
if (!p) {
scan_err2(lf, _("expected an integer or a range, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
break;
}
*p++ = 0; /* terminate first half of range */
lf->u.pint32_val = scan_pint(lf, lf->str);
lf->u2.pint32_val = scan_pint(lf, p);
token = T_PINT32_RANGE;
}
break;
case T_INT16:
if (token != T_NUMBER || !is_a_number(lf->str)) {
scan_err2(lf, _("expected an integer number, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
break;
}
errno = 0;
lf->u.int16_val = (int16_t)str_to_int64(lf->str);
if (errno != 0) {
scan_err2(lf, _("expected an integer number, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
} else {
token = T_INT16;
}
break;
case T_INT32:
if (token != T_NUMBER || !is_a_number(lf->str)) {
scan_err2(lf, _("expected an integer number, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
break;
}
errno = 0;
lf->u.int32_val = (int32_t)str_to_int64(lf->str);
if (errno != 0) {
scan_err2(lf, _("expected an integer number, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
} else {
token = T_INT32;
}
break;
case T_INT64:
Dmsg2(dbglvl, "int64=:%s: %f\n", lf->str, strtod(lf->str, NULL));
if (token != T_NUMBER || !is_a_number(lf->str)) {
scan_err2(lf, _("expected an integer number, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
break;
}
errno = 0;
lf->u.int64_val = str_to_int64(lf->str);
if (errno != 0) {
scan_err2(lf, _("expected an integer number, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
} else {
token = T_INT64;
}
break;
case T_PINT64_RANGE:
if (token == T_NUMBER) {
lf->u.pint64_val = scan_pint64(lf, lf->str);
lf->u2.pint64_val = lf->u.pint64_val;
token = T_PINT64;
} else {
char *p = strchr(lf->str, '-');
if (!p) {
scan_err2(lf, _("expected an integer or a range, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
break;
}
*p++ = 0; /* terminate first half of range */
lf->u.pint64_val = scan_pint64(lf, lf->str);
lf->u2.pint64_val = scan_pint64(lf, p);
token = T_PINT64_RANGE;
}
break;
case T_NAME:
if (token != T_IDENTIFIER && token != T_UNQUOTED_STRING && token != T_QUOTED_STRING) {
scan_err2(lf, _("expected a name, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
} else if (lf->str_len > MAX_RES_NAME_LENGTH) {
scan_err3(lf, _("name %s length %d too long, max is %d\n"), lf->str,
lf->str_len, MAX_RES_NAME_LENGTH);
token = T_ERROR;
}
break;
case T_STRING:
if (token != T_IDENTIFIER && token != T_UNQUOTED_STRING && token != T_QUOTED_STRING) {
scan_err2(lf, _("expected a string, got %s: %s"),
lex_tok_to_str(token), lf->str);
token = T_ERROR;
} else {
token = T_STRING;
}
break;
default:
break; /* no expectation given */
}
lf->token = token; /* set possible new token */
return token;
}