static int request_check_hostname(server * srv, connection * con, buffer * host)
{
enum { DOMAINLABEL, TOPLABEL } stage = TOPLABEL;
size_t i;
int label_len = 0;
size_t host_len;
char *colon;
int is_ip = -1; /* -1 don't know yet, 0 no, 1 yes */
int level = 0;
UNUSED(srv);
UNUSED(con);
/*
* hostport = host [ ":" port ]
* host = hostname | IPv4address | IPv6address
* hostname = *( domainlabel "." ) toplabel [ "." ]
* domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
* toplabel = alpha | alpha *( alphanum | "-" ) alphanum
* IPv4address = 1*digit "." 1*digit "." 1*digit "." 1*digit
* IPv6address = "[" ... "]" //IPv6地址用[]包围.
* port = *digit
*/
/*
* no Host:
*/
if (!host || host->used == 0)
return 0;
host_len = host->used - 1;
/*
* IPv6 adress
*/
if (host->ptr[0] == '[')
{
char *c = host->ptr + 1;
int colon_cnt = 0;
/*
* check portnumber
*/
for (; *c && *c != ']'; c++)
{
if (*c == ':')
{
//至多7个分号。
if (++colon_cnt > 7)
{
return -1;
}
}
else if (!light_isxdigit(*c))
{
return -1;
}
}
/*
* missing ]
*/
if (!*c)
{
return -1;
}
/*
* check port
*/
if (*(c + 1) == ':')
{
for (c += 2; *c; c++)
{
if (!light_isdigit(*c))
{
return -1;
}
}
}
return 0;
}
if (NULL != (colon = memchr(host->ptr, ':', host_len)))
{
char *c = colon + 1;
/*
* check portnumber
*/
for (; *c; c++)
{
if (!light_isdigit(*c))
return -1;
}
/*
* remove the port from the host-len
*/
host_len = colon - host->ptr;
}
/*
* Host is empty
*/
if (host_len == 0)
return -1;
/*
* if the hostname ends in a "." strip it
*/
if (host->ptr[host_len - 1] == '.')
host_len -= 1;
/*
* scan from the right and skip the \0
*/
for (i = host_len - 1; i + 1 > 0; i--)
{
const char c = host->ptr[i];
switch (stage)
{
case TOPLABEL:
if (c == '.')
{
/*/**
* header lines中的value，可以是以","分割的多个value。
* 这个函数将v中value，按照","分割成多个值，存放在vals中。
*/
int http_request_split_value(array * vals, buffer * b)
{
char *s;
size_t i;
int state = 0;
/*
* parse
* val1, val2, val3, val4
* into a array (more or less a explode() incl. striping of whitespaces
*/
if (b->used == 0)
return 0;
s = b->ptr;
for (i = 0; i < b->used - 1;)
{
char *start = NULL, *end = NULL;
data_string *ds;
switch (state)
{
case 0: /* ws */
/*
* skip ws
*/
for (; (*s == ' ' || *s == '\t') && i < b->used - 1; i++, s++);
state = 1;
break;
case 1: /* value */
start = s;
for (; *s != ',' && i < b->used - 1; i++, s++);
end = s - 1;
//去掉空格
for (; (*end == ' ' || *end == '\t') && end > start; end--);
if (NULL == (ds = (data_string *) array_get_unused_element(vals, TYPE_STRING)))
{
ds = data_string_init();
}
buffer_copy_string_len(ds->value, start, end - start + 1);
array_insert_unique(vals, (data_unset *) ds);
if (*s == ',')
{
state = 0;
i++;
s++;
}
else
{
/*
* end of string
*/
state = 2;
}
break;
default:
i++;
break;
}
}
return 0;
}
* only switch stage, if this is not the last character
*/
if (i != host_len - 1)
{
if (label_len == 0)
{
return -1;
}
/*
* check the first character at right of the dot
*/
if (is_ip == 0)
{
if (!light_isalpha(host->ptr[i + 1]))
{
return -1;
}
}
else if (!light_isdigit(host->ptr[i + 1]))
{
is_ip = 0;
}
else if ('-' == host->ptr[i + 1])
{
return -1;
}
else
{
/*
* just digits
*/
is_ip = 1;
}
stage = DOMAINLABEL;
label_len = 0;
level++;
}
else if (i == 0)
{
/*
* just a dot and nothing else is evil
*/
return -1;
}
}
else if (i == 0)
static handler_t proxy_http_parse_chunked_stream(server *srv, proxy_session *sess, chunkqueue *in, chunkqueue *out) {
protocol_state_data *data = (protocol_state_data *)sess->proxy_con->protocol_data;
char *err = NULL;
off_t we_have = 0, we_want = 0;
off_t chunk_len = 0;
off_t offset = 0;
buffer *b;
chunk *c;
char ch = '\0';
int finished = 0;
UNUSED(srv);
for (c = in->first; c && !finished;) {
if(c->mem->used == 0) {
c = c->next;
continue;
}
switch(data->chunk_parse_state) {
case HTTP_CHUNK_LEN:
/* parse chunk len. */
for(offset = c->offset; (size_t)(offset) < (c->mem->used - 1) ; offset++) {
ch = c->mem->ptr[offset];
if(!light_isxdigit(ch)) break;
}
if(offset > c->offset) {
buffer_append_string_len(data->buf, (c->mem->ptr + c->offset), offset - c->offset);
in->bytes_out += (offset - c->offset);
c->offset = offset;
}
if (!(ch == ' ' || ch == '\r' || ch == ';')) {
if (ch == '\0') {
/* get next chunk from queue */
break;
}
/* protocol error. bad http-chunk len */
return HANDLER_ERROR;
}
data->chunk_len = strtol(BUF_STR(data->buf), &err, 16);
data->chunk_offset = 0;
buffer_reset(data->buf);
data->chunk_parse_state = HTTP_CHUNK_EXTENSION;
case HTTP_CHUNK_EXTENSION:
/* find CRLF. discard chunk-extension */
for(ch = 0; (size_t)(c->offset) < (c->mem->used - 1) && ch != '\n' ;) {
ch = c->mem->ptr[c->offset];
c->offset++;
in->bytes_out++;
}
if(ch != '\n') {
/* get next chunk from queue */
break;
}
if(data->chunk_len > 0) {
data->chunk_parse_state = HTTP_CHUNK_DATA;
} else {
data->chunk_parse_state = HTTP_CHUNK_END;
}
case HTTP_CHUNK_DATA:
chunk_len = data->chunk_len - data->chunk_offset;
/* copy chunk_len bytes from in queue to out queue. */
we_have = c->mem->used - c->offset - 1;
we_want = chunk_len > we_have ? we_have : chunk_len;
if (c->offset == 0 && we_want == we_have) {
/* we are copying the whole buffer, just steal it */
chunkqueue_steal_chunk(out, c);
/* c is an empty chunk now */
} else {
b = chunkqueue_get_append_buffer(out);
buffer_copy_string_len(b, c->mem->ptr + c->offset, we_want);
c->offset += we_want;
}
chunk_len -= we_want;
out->bytes_in += we_want;
in->bytes_out += we_want;
data->chunk_offset += we_want;
if(chunk_len > 0) {
/* get next chunk from queue */
break;
}
data->chunk_offset = 0;
data->chunk_parse_state = HTTP_CHUNK_END;
case HTTP_CHUNK_END:
/* discard CRLF.*/
for(ch = 0; c->mem->used > 0 && (size_t)(c->offset) < (c->mem->used - 1) && ch != '\n' ;) {
ch = c->mem->ptr[c->offset];
c->offset++;
in->bytes_out++;
}
if(ch != '\n') {
/* get next chunk from queue */
break;
}
/* final chunk */
if(data->chunk_len == 0) {
finished = 1;
}
/* finished http-chunk. reset and parse next chunk. */
protocol_state_data_reset(data);
break;
}
if((size_t)(c->offset) == c->mem->used - 1) {
c = c->next;
}
}
chunkqueue_remove_finished_chunks(in);
if (finished) {
sess->is_request_finished = 1;
return HANDLER_FINISHED;
}
/* ran out of data. */
return HANDLER_GO_ON;
}
static int request_check_hostname(buffer *host) {
enum { DOMAINLABEL, TOPLABEL } stage = TOPLABEL;
size_t i;
int label_len = 0;
size_t host_len, hostport_len;
char *colon;
int is_ip = -1; /* -1 don't know yet, 0 no, 1 yes */
int level = 0;
/*
* hostport = host [ ":" port ]
* host = hostname | IPv4address | IPv6address
* hostname = *( domainlabel "." ) toplabel [ "." ]
* domainlabel = alphanum | alphanum *( alphanum | "-" ) alphanum
* toplabel = alpha | alpha *( alphanum | "-" ) alphanum
* IPv4address = 1*digit "." 1*digit "." 1*digit "." 1*digit
* IPv6address = "[" ... "]"
* port = *digit
*/
/* IPv6 adress */
if (host->ptr[0] == '[') {
char *c = host->ptr + 1;
int colon_cnt = 0;
/* check the address inside [...] */
for (; *c && *c != ']'; c++) {
if (*c == ':') {
if (++colon_cnt > 7) {
return -1;
}
} else if (!light_isxdigit(*c) && '.' != *c) {
return -1;
}
}
/* missing ] */
if (!*c) {
return -1;
}
/* check port */
if (*(c+1) == ':') {
for (c += 2; *c; c++) {
if (!light_isdigit(*c)) {
return -1;
}
}
}
else if ('\0' != *(c+1)) {
/* only a port is allowed to follow [...] */
return -1;
}
return 0;
}
hostport_len = host_len = buffer_string_length(host);
if (NULL != (colon = memchr(host->ptr, ':', host_len))) {
char *c = colon + 1;
/* check portnumber */
for (; *c; c++) {
if (!light_isdigit(*c)) return -1;
}
/* remove the port from the host-len */
host_len = colon - host->ptr;
}
/* Host is empty */
if (host_len == 0) return -1;
/* if the hostname ends in a "." strip it */
if (host->ptr[host_len-1] == '.') {
/* shift port info one left */
if (NULL != colon) memmove(colon-1, colon, hostport_len - host_len);
buffer_string_set_length(host, --hostport_len);
if (--host_len == 0) return -1;
}
/* scan from the right and skip the \0 */
for (i = host_len; i-- > 0; ) {
const char c = host->ptr[i];
switch (stage) {
case TOPLABEL:
if (c == '.') {
/* only switch stage, if this is not the last character */
if (i != host_len - 1) {
if (label_len == 0) {
return -1;
}
/* check the first character at right of the dot */
if (is_ip == 0) {
if (!light_isalnum(host->ptr[i+1])) {
return -1;
}
} else if (!light_isdigit(host->ptr[i+1])) {
is_ip = 0;
} else if ('-' == host->ptr[i+1]) {
return -1;
} else {
/* just digits */
is_ip = 1;
}
stage = DOMAINLABEL;
label_len = 0;
level++;
} else if (i == 0) {
/* just a dot and nothing else is evil */
return -1;
}
} else if (i == 0) {
/* the first character of the hostname */
if (!light_isalnum(c)) {
return -1;
}
label_len++;
} else {
if (c != '-' && !light_isalnum(c)) {
return -1;
}
if (is_ip == -1) {
if (!light_isdigit(c)) is_ip = 0;
}
label_len++;
}
break;
case DOMAINLABEL:
if (is_ip == 1) {
if (c == '.') {
if (label_len == 0) {
return -1;
}
label_len = 0;
level++;
} else if (!light_isdigit(c)) {
return -1;
} else {
label_len++;
}
} else {
if (c == '.') {
if (label_len == 0) {
return -1;
}
/* c is either - or alphanum here */
if ('-' == host->ptr[i+1]) {
return -1;
}
label_len = 0;
level++;
} else if (i == 0) {
if (!light_isalnum(c)) {
return -1;
}
label_len++;
} else {
if (c != '-' && !light_isalnum(c)) {
return -1;
}
label_len++;
}
}
break;
}
}
/* a IP has to consist of 4 parts */
if (is_ip == 1 && level != 3) {
return -1;
}
if (label_len == 0) {
return -1;
}
return 0;
}
