void
memcache_parse_req(struct msg *r)
{
struct mbuf *b;
uint8_t *p, *m;
uint8_t ch;
enum {
SW_START,
SW_REQ_TYPE,
SW_SPACES_BEFORE_KEY,
SW_KEY,
SW_SPACES_BEFORE_KEYS,
SW_SPACES_BEFORE_FLAGS,
SW_FLAGS,
SW_SPACES_BEFORE_EXPIRY,
SW_EXPIRY,
SW_SPACES_BEFORE_VLEN,
SW_VLEN,
SW_SPACES_BEFORE_CAS,
SW_CAS,
SW_RUNTO_VAL,
SW_VAL,
SW_SPACES_BEFORE_NUM,
SW_NUM,
SW_RUNTO_CRLF,
SW_CRLF,
SW_NOREPLY,
SW_AFTER_NOREPLY,
SW_ALMOST_DONE,
SW_SENTINEL
} state;
state = r->state;
b = STAILQ_LAST(&r->mhdr, mbuf, next);
ASSERT(r->request);
ASSERT(state >= SW_START && state < SW_SENTINEL);
ASSERT(b != NULL);
ASSERT(b->pos <= b->last);
/* validate the parsing maker */
ASSERT(r->pos != NULL);
ASSERT(r->pos >= b->pos && r->pos <= b->last);
for (p = r->pos; p < b->last; p++) {
ch = *p;
switch (state) {
case SW_START:
if (ch == ' ') {
break;
}
if (!islower(ch)) {
goto error;
}
/* req_start <- p; type_start <- p */
r->token = p;
state = SW_REQ_TYPE;
break;
case SW_REQ_TYPE:
if (ch == ' ' || ch == CR) {
/* type_end = p - 1 */
m = r->token;
r->token = NULL;
r->type = MSG_UNKNOWN;
switch (p - m) {
case 3:
if (str4cmp(m, 'g', 'e', 't', ' ')) {
r->type = MSG_REQ_GET;
break;
}
if (str4cmp(m, 's', 'e', 't', ' ')) {
r->type = MSG_REQ_SET;
break;
}
if (str4cmp(m, 'a', 'd', 'd', ' ')) {
r->type = MSG_REQ_ADD;
break;
}
if (str4cmp(m, 'c', 'a', 's', ' ')) {
r->type = MSG_REQ_CAS;
break;
}
break;
case 4:
if (str4cmp(m, 'g', 'e', 't', 's')) {
r->type = MSG_REQ_GETS;
break;
}
if (str4cmp(m, 'i', 'n', 'c', 'r')) {
r->type = MSG_REQ_INCR;
break;
}
if (str4cmp(m, 'd', 'e', 'c', 'r')) {
r->type = MSG_REQ_DECR;
break;
}
if (str4cmp(m, 'q', 'u', 'i', 't')) {
r->type = MSG_REQ_QUIT;
r->quit = 1;
break;
}
break;
case 6:
if (str6cmp(m, 'a', 'p', 'p', 'e', 'n', 'd')) {
r->type = MSG_REQ_APPEND;
break;
}
if (str6cmp(m, 'd', 'e', 'l', 'e', 't', 'e')) {
r->type = MSG_REQ_DELETE;
break;
}
break;
case 7:
if (str7cmp(m, 'p', 'r', 'e', 'p', 'e', 'n', 'd')) {
r->type = MSG_REQ_PREPEND;
break;
}
if (str7cmp(m, 'r', 'e', 'p', 'l', 'a', 'c', 'e')) {
r->type = MSG_REQ_REPLACE;
break;
}
if (str7cmp(m, 'v', 'e', 'r', 's', 'i', 'o', 'n')) {
r->type = MSG_REQ_VERSION;
break;
}
break;
}
if (memcache_key(r)) {
if (ch == CR) {
goto error;
}
state = SW_SPACES_BEFORE_KEY;
} else if (memcache_quit(r) || memcache_version(r)) {
p = p - 1; /* go back by 1 byte */
state = SW_CRLF;
} else {
goto error;
}
} else if (!islower(ch)) {
goto error;
}
break;
case SW_SPACES_BEFORE_KEY:
if (ch != ' ') {
p = p - 1; /* go back by 1 byte */
state = SW_KEY;
}
break;
case SW_KEY:
if (r->token == NULL) {
r->token = p;
r->key_start = p;
}
if (ch == ' ' || ch == CR) {
if ((p - r->key_start) > MEMCACHE_MAX_KEY_LENGTH) {
log_error("parsed bad req %"PRIu64" of type %d with key "
"prefix '%.*s...' and length %d that exceeds "
"maximum key length", r->id, r->type, 16,
r->key_start, p - r->key_start);
goto error;
}
r->key_end = p;
r->token = NULL;
/* get next state */
if (memcache_storage(r)) {
state = SW_SPACES_BEFORE_FLAGS;
} else if (memcache_arithmetic(r)) {
state = SW_SPACES_BEFORE_NUM;
} else if (memcache_delete(r)) {
state = SW_RUNTO_CRLF;
} else if (memcache_retrieval(r)) {
state = SW_SPACES_BEFORE_KEYS;
} else {
state = SW_RUNTO_CRLF;
}
if (ch == CR) {
if (memcache_storage(r) || memcache_arithmetic(r)) {
goto error;
}
p = p - 1; /* go back by 1 byte */
}
}
break;
case SW_SPACES_BEFORE_KEYS:
ASSERT(memcache_retrieval(r));
switch (ch) {
case ' ':
break;
case CR:
state = SW_ALMOST_DONE;
break;
default:
r->token = p;
goto fragment;
}
break;
case SW_SPACES_BEFORE_FLAGS:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
p = p - 1; /* go back by 1 byte */
state = SW_FLAGS;
}
break;
case SW_FLAGS:
if (r->token == NULL) {
/* flags_start <- p */
r->token = p;
r->flags = 0;
}
if (isdigit(ch)) {
r->flags = r->flags * 10 + (uint32_t)(ch - '0');
} else if (ch == ' ') {
/* flags_end <- p - 1 */
r->token = NULL;
state = SW_SPACES_BEFORE_EXPIRY;
} else {
goto error;
}
break;
case SW_SPACES_BEFORE_EXPIRY:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
p = p - 1; /* go back by 1 byte */
state = SW_EXPIRY;
}
break;
case SW_EXPIRY:
if (r->token == NULL) {
/* expiry_start <- p */
r->token = p;
r->expiry = 0;
}
if (isdigit(ch)) {
r->expiry = r->expiry * 10 + (uint32_t)(ch - '0');
} else if (ch == ' ') {
/* expiry_end <- p - 1 */
r->token = NULL;
state = SW_SPACES_BEFORE_VLEN;
} else {
goto error;
}
break;
case SW_SPACES_BEFORE_VLEN:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
p = p - 1; /* go back by 1 byte */
state = SW_VLEN;
}
break;
case SW_VLEN:
if (r->token == NULL) {
/* vlen_start <- p */
r->token = p;
r->vlen = 0;
}
if (isdigit(ch)) {
r->vlen = r->vlen * 10 + (uint32_t)(ch - '0');
} else if (memcache_cas(r)) {
if (ch != ' ') {
goto error;
}
/* vlen_end <- p - 1 */
r->rvlen = r->vlen;
p = p - 1; /* go back by 1 byte */
r->token = NULL;
state = SW_SPACES_BEFORE_CAS;
} else if (ch == ' ' || ch == CR) {
/* vlen_end <- p - 1 */
r->rvlen = r->vlen;
p = p - 1; /* go back by 1 byte */
r->token = NULL;
state = SW_RUNTO_CRLF;
} else {
goto error;
}
break;
case SW_SPACES_BEFORE_CAS:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
p = p - 1; /* go back by 1 byte */
state = SW_CAS;
}
break;
case SW_CAS:
if (r->token == NULL) {
/* cas_start <- p */
r->token = p;
r->cas = 0;
}
if (isdigit(ch)) {
r->cas = r->cas * 10ULL + (uint64_t)(ch - '0');
} else if (ch == ' ' || ch == CR) {
/* cas_end <- p - 1 */
p = p - 1; /* go back by 1 byte */
r->token = NULL;
state = SW_RUNTO_CRLF;
} else {
goto error;
}
break;
case SW_RUNTO_VAL:
switch (ch) {
case LF:
/* val_start <- p + 1 */
state = SW_VAL;
break;
default:
goto error;
}
break;
case SW_VAL:
if (r->value == NULL) {
r->value = p;
}
m = p + r->rvlen;
if (m >= b->last) {
ASSERT(r->rvlen >= (uint32_t)(b->last - p));
r->rvlen -= (uint32_t)(b->last - p);
m = b->last - 1;
p = m; /* move forward by vlen bytes */
break;
}
switch (*m) {
case CR:
/* val_end <- p - 1 */
p = m; /* move forward by vlen bytes */
state = SW_ALMOST_DONE;
break;
default:
goto error;
}
break;
case SW_SPACES_BEFORE_NUM:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
p = p - 1; /* go back by 1 byte */
state = SW_NUM;
}
break;
case SW_NUM:
if (r->token == NULL) {
/* num_start <- p */
r->token = p;
r->num = 0;
}
if (isdigit(ch)) {
r->num = r->num * 10ULL + (uint64_t)(ch - '0');
} else if (ch == ' ' || ch == CR) {
r->token = NULL;
/* num_end <- p - 1 */
p = p - 1; /* go back by 1 byte */
state = SW_RUNTO_CRLF;
} else {
goto error;
}
break;
case SW_RUNTO_CRLF:
switch (ch) {
case ' ':
break;
case 'n':
if (memcache_storage(r) || memcache_arithmetic(r) || memcache_delete(r)) {
p = p - 1; /* go back by 1 byte */
state = SW_NOREPLY;
} else {
goto error;
}
break;
case CR:
if (memcache_storage(r)) {
state = SW_RUNTO_VAL;
} else {
state = SW_ALMOST_DONE;
}
break;
default:
goto error;
}
break;
case SW_NOREPLY:
if (r->token == NULL) {
/* noreply_start <- p */
r->token = p;
}
switch (ch) {
case ' ':
case CR:
m = r->token;
if (((p - m) == 7) && str7cmp(m, 'n', 'o', 'r', 'e', 'p', 'l', 'y')) {
ASSERT(memcache_storage(r) || memcache_arithmetic(r) || memcache_delete(r));
r->token = NULL;
/* noreply_end <- p - 1 */
r->noreply = 1;
state = SW_AFTER_NOREPLY;
p = p - 1; /* go back by 1 byte */
} else {
goto error;
}
}
break;
case SW_AFTER_NOREPLY:
switch (ch) {
case ' ':
break;
case CR:
if (memcache_storage(r)) {
state = SW_RUNTO_VAL;
} else {
state = SW_ALMOST_DONE;
}
break;
default:
goto error;
}
break;
case SW_CRLF:
switch (ch) {
case ' ':
break;
case CR:
state = SW_ALMOST_DONE;
break;
default:
goto error;
}
break;
case SW_ALMOST_DONE:
switch (ch) {
case LF:
/* req_end <- p */
goto done;
default:
goto error;
}
break;
case SW_SENTINEL:
default:
NOT_REACHED();
break;
}
}
/*
* At this point, buffer from b->pos to b->last has been parsed completely
* but we haven't been able to reach to any conclusion. Normally, this
* means that we have to parse again starting from the state we are in
* after more data has been read. The newly read data is either read into
* a new mbuf, if existing mbuf is full (b->last == b->end) or into the
* existing mbuf.
*
* The only exception to this is when the existing mbuf is full (b->last
* is at b->end) and token marker is set, which means that we have to
* copy the partial token into a new mbuf and parse again with more data
* read into new mbuf.
*/
ASSERT(p == b->last);
r->pos = p;
r->state = state;
if (b->last == b->end && r->token != NULL) {
r->pos = r->token;
r->token = NULL;
r->result = MSG_PARSE_REPAIR;
} else {
r->result = MSG_PARSE_AGAIN;
}
log_hexdump(LOG_VERB, b->pos, mbuf_length(b), "parsed req %"PRIu64" res %d "
"type %d state %d rpos %d of %d", r->id, r->result, r->type,
r->state, r->pos - b->pos, b->last - b->pos);
return;
fragment:
ASSERT(p != b->last);
ASSERT(r->token != NULL);
r->pos = r->token;
r->token = NULL;
r->state = state;
r->result = MSG_PARSE_FRAGMENT;
log_hexdump(LOG_VERB, b->pos, mbuf_length(b), "parsed req %"PRIu64" res %d "
"type %d state %d rpos %d of %d", r->id, r->result, r->type,
r->state, r->pos - b->pos, b->last - b->pos);
return;
done:
ASSERT(r->type > MSG_UNKNOWN && r->type < MSG_SENTINEL);
r->pos = p + 1;
ASSERT(r->pos <= b->last);
r->state = SW_START;
r->result = MSG_PARSE_OK;
log_hexdump(LOG_VERB, b->pos, mbuf_length(b), "parsed req %"PRIu64" res %d "
"type %d state %d rpos %d of %d", r->id, r->result, r->type,
r->state, r->pos - b->pos, b->last - b->pos);
return;
error:
r->result = MSG_PARSE_ERROR;
r->state = state;
errno = EINVAL;
log_hexdump(LOG_INFO, b->pos, mbuf_length(b), "parsed bad req %"PRIu64" "
"res %d type %d state %d", r->id, r->result, r->type,
r->state);
}
static ssize_t memcacheParseReq(struct memcacheProcData *r, struct slice *s)
{
char *m;
char ch;
size_t command_len;
size_t pos = 0;
uint64_t left = 0;
struct slice val_s;
enum reqStage state = r->stage;
r->token_pos = 0;
while (pos < s->len) {
ch = s->data[pos];
switch (state) {
case SW_START:
if (!islower(ch)) {
goto error;
}
r->token_pos = pos;
state = SW_REQ_TYPE;
break;
case SW_REQ_TYPE:
if (ch == ' ' || ch == CR) {
r->type = UNKNOWN;
if (wstrlen(r->command)) {
r->command = wstrCatLen(r->command, (char *)&s->data[r->token_pos], pos);
command_len = wstrlen(r->command);
m = (char*)&r->command;
} else {
command_len = pos - r->token_pos;
m = (char*)&s->data[r->token_pos];
}
switch (command_len) {
case 3:
if (str4icmp(m, 'g', 'e', 't', ' ')) {
r->type = REQ_MC_GET;
break;
}
if (str4icmp(m, 's', 'e', 't', ' ')) {
r->type = REQ_MC_SET;
break;
}
if (str4icmp(m, 'a', 'd', 'd', ' ')) {
r->type = REQ_MC_ADD;
break;
}
if (str4icmp(m, 'c', 'a', 's', ' ')) {
r->type = REQ_MC_CAS;
break;
}
break;
case 4:
if (str4icmp(m, 'g', 'e', 't', 's')) {
r->type = REQ_MC_GETS;
break;
}
if (str4icmp(m, 'i', 'n', 'c', 'r')) {
r->type = REQ_MC_INCR;
break;
}
if (str4icmp(m, 'd', 'e', 'c', 'r')) {
r->type = REQ_MC_DECR;
break;
}
if (str4icmp(m, 'q', 'u', 'i', 't')) {
r->type = REQ_MC_QUIT;
r->quit = 1;
break;
}
break;
case 6:
if (str6icmp(m, 'a', 'p', 'p', 'e', 'n', 'd')) {
r->type = REQ_MC_APPEND;
break;
}
if (str6icmp(m, 'd', 'e', 'l', 'e', 't', 'e')) {
r->type = REQ_MC_DELETE;
break;
}
break;
case 7:
if (str7icmp(m, 'p', 'r', 'e', 'p', 'e', 'n', 'd')) {
r->type = REQ_MC_PREPEND;
break;
}
if (str7icmp(m, 'r', 'e', 'p', 'l', 'a', 'c', 'e')) {
r->type = REQ_MC_REPLACE;
break;
}
break;
}
switch (r->type) {
case REQ_MC_GET:
case REQ_MC_GETS:
case REQ_MC_DELETE:
case REQ_MC_CAS:
case REQ_MC_SET:
case REQ_MC_ADD:
case REQ_MC_REPLACE:
case REQ_MC_APPEND:
case REQ_MC_PREPEND:
case REQ_MC_INCR:
case REQ_MC_DECR:
if (ch == CR) {
goto error;
}
state = SW_SPACES_BEFORE_KEY;
break;
case REQ_MC_QUIT:
pos--; /* go back by 1 byte */
state = SW_CRLF;
break;
case UNKNOWN:
goto error;
default:
ASSERT(0);
}
} else if (!islower(ch)) {
goto error;
}
break;
case SW_SPACES_BEFORE_KEY:
if (ch != ' ') {
pos--; /* go back by 1 byte */
r->token_pos = -1;
state = SW_KEY;
}
break;
case SW_KEY:
if (r->token_pos == -1) {
r->token_pos = pos;
}
if (ch == ' ' || ch == CR) {
if (wstrlen(r->key) + (pos - r->token_pos) > MEMCACHE_MAX_KEY_LENGTH) {
wheatLog(WHEAT_WARNING,
"parsed bad type %d with key prefix '%d' and length %d that exceeds "
"maximum key length", r->type, r->token_pos, pos);
goto error;
}
r->key = wstrCatLen(r->key, (char*)&s->data[r->token_pos], pos - r->token_pos);
arrayPush(r->keys, &r->key);
r->key = wstrNewLen(NULL, 64);
r->token_pos = -1;
/* get next state */
if (memcache_storage(r)) {
state = SW_SPACES_BEFORE_FLAGS;
} else if (memcache_arithmetic(r)) {
state = SW_SPACES_BEFORE_NUM;
} else if (memcache_delete(r)) {
// whether exist delay time
if (ch == ' ')
state = SW_SPACES_BEFORE_NUM;
else
state = SW_RUNTO_CRLF;
} else if (memcache_retrieval(r)) {
state = SW_SPACES_BEFORE_KEYS;
} else {
state = SW_RUNTO_CRLF;
}
if (ch == CR) {
if (memcache_storage(r) || memcache_arithmetic(r)) {
goto error;
}
pos--; /* go back by 1 byte */
}
}
break;
case SW_SPACES_BEFORE_KEYS:
ASSERT(memcache_retrieval(r));
switch (ch) {
case ' ':
break;
case CR:
state = SW_ALMOST_DONE;
break;
default:
r->token_pos = -1;
pos--; /* go back by 1 byte */
state = SW_KEY;
}
break;
case SW_SPACES_BEFORE_FLAGS:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
r->flag = ch - '0';
state = SW_FLAGS;
}
break;
case SW_FLAGS:
if (isdigit(ch)) {
r->flag = r->flag * 10 + (ch - '0');
} else if (ch == ' ') {
state = SW_SPACES_BEFORE_EXPIRY;
} else {
goto error;
}
break;
case SW_SPACES_BEFORE_EXPIRY:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
/* expiry_start <- p; expiry <- ch - '0' */
r->expire = ch - '0';
state = SW_EXPIRY;
}
break;
case SW_EXPIRY:
if (isdigit(ch)) {
r->expire = r->expire * 10 + (ch - '0');
} else if (ch == ' ') {
state = SW_SPACES_BEFORE_VLEN;
} else {
goto error;
}
break;
case SW_SPACES_BEFORE_VLEN:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
/* vlen_start <- p */
r->vlen = ch - '0';
state = SW_VLEN;
}
break;
case SW_VLEN:
if (isdigit(ch)) {
r->vlen = r->vlen * 10 + (uint32_t)(ch - '0');
} else if (memcache_cas(r)) {
if (ch != ' ') {
goto error;
}
pos--; /* go back by 1 byte */
state = SW_SPACES_BEFORE_CAS;
} else if (ch == ' ' || ch == CR) {
pos--; /* go back by 1 byte */
state = SW_RUNTO_CRLF;
} else {
goto error;
}
break;
case SW_SPACES_BEFORE_CAS:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
/* cas_start <- p; cas <- ch - '0' */
r->cas = ch - '0';
state = SW_CAS;
}
break;
case SW_CAS:
if (isdigit(ch)) {
r->cas = r->cas * 10 + (ch - '0');
} else if (ch == ' ' || ch == CR) {
pos--; /* go back by 1 byte */
state = SW_RUNTO_CRLF;
} else {
goto error;
}
break;
case SW_RUNTO_VAL:
switch (ch) {
case LF:
/* val_start <- p + 1 */
state = SW_VAL;
r->vlen_left = r->vlen;
break;
default:
goto error;
}
break;
case SW_VAL:
if (r->vlen_left) {
left = (s->len - pos) >= r->vlen ? (r->vlen) : (s->len-pos);
val_s.len = left;
val_s.data = &s->data[pos];
arrayPush(r->vals, &val_s);
r->vlen_left -= left;
pos += left;
}
switch (s->data[pos]) {
case CR:
state = SW_ALMOST_DONE;
break;
default:
goto error;
}
break;
case SW_SPACES_BEFORE_NUM:
if (ch != ' ') {
if (!isdigit(ch)) {
goto error;
}
r->num = ch - '0';
state = SW_NUM;
}
break;
case SW_NUM:
if (isdigit(ch)) {
r->num = r->num * 10 + ch - '0';
} else if (ch == ' ' || ch == CR) {
pos--; /* go back by 1 byte */
state = SW_RUNTO_CRLF;
} else {
goto error;
}
break;
case SW_RUNTO_CRLF:
switch (ch) {
case ' ':
break;
case 'n':
if (memcache_storage(r) || memcache_arithmetic(r) || memcache_delete(r)) {
state = SW_NOREPLY;
r->noreply_banner = wstrNewLen(NULL, 8);
r->token_pos = pos;
} else {
goto error;
}
break;
case CR:
if (memcache_storage(r)) {
state = SW_RUNTO_VAL;
} else {
state = SW_ALMOST_DONE;
}
break;
default:
goto error;
}
break;
case SW_NOREPLY:
switch (ch) {
case ' ':
case CR:
r->noreply_banner = wstrCatLen(r->noreply_banner, (char*)&s->data[r->token_pos], pos - r->token_pos);
if (str7icmp(r->noreply_banner, 'n', 'o', 'r', 'e', 'p', 'l', 'y')) {
ASSERT(memcache_storage(r) || memcache_arithmetic(r) || memcache_delete(r));
/* noreply_end <- p - 1 */
r->noreply = 1;
state = SW_AFTER_NOREPLY;
pos--; /* go back by 1 byte */
} else {
goto error;
}
}
break;
case SW_AFTER_NOREPLY:
switch (ch) {
case ' ':
break;
case CR:
if (memcache_storage(r)) {
state = SW_RUNTO_VAL;
} else {
state = SW_ALMOST_DONE;
}
break;
default:
goto error;
}
break;
case SW_CRLF:
switch (ch) {
case ' ':
break;
case CR:
state = SW_ALMOST_DONE;
break;
default:
goto error;
}
break;
case SW_ALMOST_DONE:
switch (ch) {
case LF:
/* req_end <- p */
pos++;
goto done;
default:
goto error;
}
break;
case SW_SENTINEL:
default:
ASSERT(0);
break;
}
pos++;
}
if (state == SW_REQ_TYPE) {
r->command = wstrCatLen(r->command, (char*)&s->data[r->token_pos], pos - r->token_pos);
} else if (state == SW_KEY) {
r->key = wstrCatLen(r->key, (char*)&s->data[r->token_pos], pos - r->token_pos);
} else if (state == SW_NOREPLY) {
r->noreply_banner = wstrCatLen(r->noreply_banner, (char*)&s->data[r->token_pos], pos - r->token_pos);
}
done:
r->stage = state;
wheatLog(WHEAT_DEBUG, "parsed successfully type %d state %d", r->type, r->stage);
return pos;
error:
wheatLog(WHEAT_DEBUG, "parsed failed type %d state %d: %s", r->type, r->stage,
s->data);
return -1;
}
