static std::pair <UnicodeChar, Point> decode (Point cur, Point end)
{
if ((*cur & 0x80) == 0)
{
UnicodeChar chr = *cur++;
return std::make_pair (chr, cur);
}
int octets;
UnicodeChar chr;
std::tie (octets, chr) = octet_count (*cur++);
if (octets > 5)
return std::make_pair (sBadChar(), cur);
Point eoc = cur + octets;
if (eoc > end)
return std::make_pair (sBadChar(), cur);
while (cur != eoc)
{
if ((*cur & 0xC0) != 0x80) // check continuation mark
return std::make_pair (sBadChar(), cur);
chr = (chr << 6) | UnicodeChar ((*cur++) & 0x3F);
}
return std::make_pair (chr, cur);
}
// Big state switch
int uslg_parser_exec(syslog_parser *parser, const syslog_parser_settings *settings, const char *data, size_t length) {
int d_index;
int error = 0;
char next_byte;
for (d_index = 0; d_index < length; d_index++) {
int action = pa_none;
next_byte = data[d_index];
#if DEBUG_OUTPUT
printf("Next byte: %c\n", next_byte);
#endif
// Token state is managed first
if (parser->token_state == ts_before) {
switch (next_byte) {
case ' ':
case '\t':
action = pa_advance;
break;
case '\r':
//case '\n': removed for issues with syslog messages with empty msg field
if (!(parser->flags & F_COUNT_OCTETS)) {
parser->error = SLERR_PREMATURE_MSG_END;
}
break;
default:
set_token_state(parser, ts_read);
action = pa_rehash;
}
} else {
// Parser state
switch (parser->state) {
case s_msg_start:
action = msg_start(parser, settings, next_byte);
break;
case s_octet_count:
action = octet_count(parser, next_byte);
break;
case s_priority_start:
action = priority_start(parser, next_byte);
break;
case s_priority:
action = priority(parser, next_byte);
break;
case s_version:
action = version(parser, next_byte);
break;
case s_timestamp:
action = parse_msg_head_part(parser, s_hostname, next_byte);
break;
case s_hostname:
action = parse_msg_head_part(parser, s_appname, next_byte);
break;
case s_appname:
action = parse_msg_head_part(parser, s_processid, next_byte);
break;
case s_processid:
action = parse_msg_head_part(parser, s_messageid, next_byte);
break;
case s_messageid:
action = parse_msg_head_part(parser, s_sd_start, next_byte);
break;
case s_sd_start:
action = sd_start(parser, settings, next_byte);
break;
case s_sd_element:
action = sd_element(parser, settings, next_byte);
break;
case s_sd_field_start:
action = sd_field_start(parser, next_byte);
break;
case s_sd_field:
action = sd_field(parser, settings, next_byte);
break;
case s_sd_value_start:
action = sd_value_start(parser, next_byte);
break;
case s_sd_value:
action = sd_value(parser, settings, next_byte);
break;
case s_message:
d_index += read_message(parser, settings, data + d_index, length - d_index);
action = pa_rehash;
break;
default:
parser->error = SLERR_BAD_STATE;
}
}
// Upon error, exit the read loop regardless of action
if (parser->error) {
error = parser->error;
uslg_parser_reset(parser);
break;
}
// What action should be taken for this byte
switch (action) {
case pa_advance:
if (parser->flags & F_COUNT_OCTETS) {
parser->octets_remaining--;
} else {
parser->message_length++;
}
break;
case pa_rehash:
d_index--;
break;
}
}
return error;
}
