/**
* Inspect request headers and register the MULTIPART request data hook
* if it contains a multipart/form-data body.
*
* @param connp
*/
int htp_ch_multipart_callback_request_headers(htp_connp_t *connp) {
// Check the request content type to see if it matches our MIME type
if ((connp->in_tx->request_content_type == NULL) || (bstr_cmp_c(connp->in_tx->request_content_type, HTP_MULTIPART_MIME_TYPE) != 0)) {
#ifdef HTP_DEBUG
fprintf(stderr, "htp_ch_multipart_callback_request_headers: Body not MULTIPART\n");
#endif
return HOOK_OK;
}
#ifdef HTP_DEBUG
fprintf(stderr, "htp_ch_multipart_callback_request_headers: Parsing MULTIPART body\n");
#endif
htp_header_t *ct = table_get_c(connp->in_tx->request_headers, "content-type");
if (ct == NULL) return HOOK_OK;
char *boundary = NULL;
int rc = htp_mpartp_extract_boundary(ct->value, &boundary);
if (rc != HTP_OK) {
// TODO Invalid boundary
return HOOK_OK;
}
// Create parser instance
connp->in_tx->request_mpartp = htp_mpartp_create(connp->cfg, boundary);
if (connp->in_tx->request_mpartp == NULL) {
free(boundary);
return HOOK_ERROR;
}
if (connp->cfg->extract_request_files) {
connp->in_tx->request_mpartp->extract_files = 1;
connp->in_tx->request_mpartp->extract_dir = connp->cfg->tmpdir;
}
free(boundary);
// Register request body data callbacks
htp_tx_register_request_body_data(connp->in_tx, htp_ch_multipart_callback_request_body_data);
return HOOK_OK;
}
/**
* Process part headers. In the current implementation, we only parse the
* Content-Disposition header if it is present.
*
* @param part
* @return Success indication
*/
int htp_mpart_part_process_headers(htp_mpart_part_t *part) {
// Find C-D header
htp_header_t *h = (htp_header_t *) table_get_c(part->headers, "content-disposition");
if (h == NULL) {
// TODO Error message
return 0;
}
if (bstr_index_of_c(h->value, "form-data") != 0) {
return -1;
}
// The parsing starts here
unsigned char *data = (unsigned char *) bstr_ptr(h->value);
size_t len = bstr_len(h->value);
size_t pos = 9; // Start after "form-data"
// Main parameter parsing loop (once per parameter)
while (pos < len) {
// Find semicolon and go over it
while ((pos < len) && ((data[pos] == '\t') || (data[pos] == ' '))) pos++;
if (pos == len) return -2;
// Semicolon
if (data[pos] != ';') return -3;
pos++;
// Go over the whitespace before parameter name
while ((pos < len) && ((data[pos] == '\t') || (data[pos] == ' '))) pos++;
if (pos == len) return -4;
// Found starting position (name)
size_t start = pos;
// Look for ending position
while ((pos < len) && (data[pos] != '\t') && (data[pos] != ' ') && (data[pos] != '=')) pos++;
if (pos == len) return -5;
// Ending position is in "pos" now
// Is it a parameter we are interested in?
int param_type = htp_mpartp_cd_param_type(data, start, pos);
// Ignore whitespace
while ((pos < len) && ((data[pos] == '\t') || (data[pos] == ' '))) pos++;
if (pos == len) return -6;
// Equals
if (data[pos] != '=') return -7;
pos++;
// Go over the whitespace before value
while ((pos < len) && ((data[pos] == '\t') || (data[pos] == ' '))) pos++;
if (pos == len) return -8;
// Found starting point (value)
start = pos;
// Quoting char indicator
int qchar = -1;
// Different handling for quoted and bare strings
if (data[start] == '"') {
// Quoted string
qchar = data[start];
start = ++pos;
// Find the end of the value
while ((pos < len) && (data[pos] != qchar)) {
if (data[pos] == '\\') {
// Ignore invalid quoting pairs
if (pos + 1 < len) return -9;
// Go over the quoted character
pos++;
}
pos++;
}
} else {
// Bare string
while ((pos < len) && (!htp_is_token(data[pos]))) pos++;
}
switch (param_type) {
case PARAM_NAME:
// TODO Unquote quoted characters
part->name = bstr_dup_mem((char *) data + start, pos - start);
if (part->name == NULL) return -1;
break;
case PARAM_FILENAME:
// TODO Unquote quoted characters
part->file = calloc(1, sizeof (htp_file_t));
if (part->file == NULL) return -1;
part->file->filename = bstr_dup_mem((char *) data + start, pos - start);
if (part->file->filename == NULL) return -1;
part->file->source = HTP_FILE_MULTIPART;
break;
default:
// Ignore unknown parameter
// TODO Warn/log?
break;
}
// Skip over the quoting character
if (qchar != -1) {
pos++;
}
// Continue to parse the next parameter, if any
}
return 1;
}
/**
* Determines presence (and encoding) of a request body.
*
* @param connp
* @returns HTP_OK on state change, HTTP_ERROR on error, or HTP_DATA when more data is needed.
*/
int htp_connp_REQ_BODY_DETERMINE(htp_connp_t *connp) {
htp_header_t *cl = table_get_c(connp->in_tx->request_headers, "content-length");
htp_header_t *te = table_get_c(connp->in_tx->request_headers, "transfer-encoding");
// Check for the Transfer-Encoding header, which
// would indicate a chunked request body
if (te != NULL) {
// Make sure it contains "chunked" only
if (bstr_cmp_c(te->value, "chunked") != 0) {
// Invalid T-E header value
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0,
"Invalid T-E value in request");
}
// Chunked encoding is a HTTP/1.1 feature. Check
// that some other protocol is not used. The flag will
// also be set if the protocol could not be parsed.
//
// TODO IIS 7.0, for example, would ignore the T-E header when it
// it is used with a protocol below HTTP 1.1.
if (connp->in_tx->request_protocol_number < HTTP_1_1) {
connp->in_tx->flags |= HTP_INVALID_CHUNKING;
// TODO Log
}
// If the T-E header is present we are going to use it.
connp->in_tx->request_transfer_coding = CHUNKED;
// We are still going to check for the presence of C-L
if (cl != NULL) {
// This is a violation of the RFC
connp->in_tx->flags |= HTP_REQUEST_SMUGGLING;
// TODO Log
}
connp->in_state = htp_connp_REQ_BODY_CHUNKED_LENGTH;
connp->in_tx->progress = TX_PROGRESS_REQ_BODY;
} else
// Next check for the presence of the Content-Length header
if (cl != NULL) {
// It seems that we have a request body.
connp->in_tx->request_transfer_coding = IDENTITY;
// Check for a folded C-L header
if (cl->flags & HTP_FIELD_FOLDED) {
connp->in_tx->flags |= HTP_REQUEST_SMUGGLING;
// TODO Log
}
// Check for multiple C-L headers
if (cl->flags & HTP_FIELD_REPEATED) {
connp->in_tx->flags |= HTP_REQUEST_SMUGGLING;
// TODO Log
}
// Get body length
int i = htp_parse_content_length(cl->value);
if (i < 0) {
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0, "Invalid C-L field in request");
return HTP_ERROR;
} else {
connp->in_content_length = i;
connp->in_body_data_left = connp->in_content_length;
if (connp->in_content_length != 0) {
connp->in_state = htp_connp_REQ_BODY_IDENTITY;
connp->in_tx->progress = TX_PROGRESS_REQ_BODY;
} else {
connp->in_state = htp_connp_REQ_IDLE;
connp->in_tx->progress = TX_PROGRESS_WAIT;
}
}
} else {
// This request does not have a body, which
// means that we're done with it
connp->in_state = htp_connp_REQ_IDLE;
connp->in_tx->progress = TX_PROGRESS_WAIT;
}
// Check for PUT requests, which we need to treat as file uploads
if (connp->in_tx->request_method_number == M_PUT) {
if (connp->in_tx->connp->in_tx->request_transfer_coding != 0) {
// Prepare to treat PUT request body as a file
connp->put_file = calloc(1, sizeof (htp_file_t));
if (connp->put_file == NULL) return HTP_ERROR;
connp->put_file->source = HTP_FILE_PUT;
} else {
// TODO Warn about PUT request without a body
}
return HTP_OK;
}
// Host resolution
htp_header_t *h = table_get_c(connp->in_tx->request_headers, "host");
if (h == NULL) {
// No host information in the headers
// HTTP/1.1 requires host information in the headers
if (connp->in_tx->request_protocol_number >= HTTP_1_1) {
connp->in_tx->flags |= HTP_HOST_MISSING;
htp_log(connp, HTP_LOG_MARK, HTP_LOG_WARNING, 0,
"Host information in request headers required by HTTP/1.1");
}
} else {
// Host information available in the headers
// Is there host information in the URI?
if (connp->in_tx->parsed_uri->hostname == NULL) {
// There is no host information in the URI. Place the
// hostname from the headers into the parsed_uri structure.
htp_replace_hostname(connp, connp->in_tx->parsed_uri, h->value);
} else if (bstr_cmp_nocase(h->value, connp->in_tx->parsed_uri->hostname) != 0) {
// The host information is different in the
// headers and the URI. The HTTP RFC states that
// we should ignore the headers copy.
connp->in_tx->flags |= HTP_AMBIGUOUS_HOST;
htp_log(connp, HTP_LOG_MARK, HTP_LOG_WARNING, 0, "Host information ambiguous");
}
}
// Parse Content-Type
htp_header_t *ct = table_get_c(connp->in_tx->request_headers, "content-type");
if (ct != NULL) {
connp->in_tx->request_content_type = bstr_dup_lower(ct->value);
if (connp->in_tx->request_content_type == NULL) {
return HTP_ERROR;
}
// Ignore parameters
char *data = bstr_ptr(connp->in_tx->request_content_type);
size_t len = bstr_len(ct->value);
size_t newlen = 0;
while (newlen < len) {
// TODO Some platforms may do things differently here
if (htp_is_space(data[newlen]) || (data[newlen] == ';')) {
bstr_util_adjust_len(connp->in_tx->request_content_type, newlen);
break;
}
newlen++;
}
}
// Parse cookies
if (connp->cfg->parse_request_cookies) {
htp_parse_cookies_v0(connp);
}
// Parse authentication information
if (connp->cfg->parse_request_http_authentication) {
htp_parse_authorization(connp);
}
// Run hook REQUEST_HEADERS
int rc = hook_run_all(connp->cfg->hook_request_headers, connp);
if (rc != HOOK_OK) {
switch (rc) {
case HOOK_STOP:
return HTP_STOP;
case HOOK_ERROR:
case HOOK_DECLINED:
default:
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0,
"Request headers callback returned error (%d)", rc);
return HTP_ERROR;
}
}
return HTP_OK;
}
/**
* Determines presence (and encoding) of a response body.
*
* @param connp
* @returns HTP_OK on state change, HTTP_ERROR on error, or HTP_DATA when more data is needed.
*/
int htp_connp_RES_BODY_DETERMINE(htp_connp_t *connp) {
// If the request uses the CONNECT method, then not only are we
// to assume there's no body, but we need to ignore all
// subsequent data in the stream.
if (connp->out_tx->request_method_number == M_CONNECT) {
if ((connp->out_tx->response_status_number >= 200)
&& (connp->out_tx->response_status_number <= 299))
{
// This is a successful CONNECT stream, which means
// we need to switch into tunnelling mode.
connp->in_status = STREAM_STATE_TUNNEL;
connp->out_status = STREAM_STATE_TUNNEL;
connp->out_state = htp_connp_RES_IDLE;
connp->out_tx->progress = TX_PROGRESS_DONE;
return HTP_OK;
} else {
// This is a failed CONNECT stream, which means that
// we can unblock request parsing
connp->in_status = STREAM_STATE_DATA;
// We are going to continue processing this transaction,
// adding a note for ourselves to stop at the end (because
// we don't want to see the beginning of a new transaction).
connp->out_data_other_at_tx_end = 1;
}
}
// Check for an interim "100 Continue"
// response. Ignore it if found, and revert back to RES_FIRST_LINE.
if (connp->out_tx->response_status_number == 100) {
if (connp->out_tx->seen_100continue != 0) {
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0, "Already seen 100-Continue");
return HTP_ERROR;
}
// Ignore any response headers set
table_clear(connp->out_tx->response_headers);
connp->out_state = htp_connp_RES_LINE;
connp->out_tx->progress = TX_PROGRESS_RES_LINE;
connp->out_tx->seen_100continue++;
return HTP_OK;
}
// Check for compression
if (connp->cfg->response_decompression_enabled) {
htp_header_t *ce = table_get_c(connp->out_tx->response_headers, "content-encoding");
if (ce != NULL) {
if ((bstr_cmp_c(ce->value, "gzip") == 0) || (bstr_cmp_c(ce->value, "x-gzip") == 0)) {
connp->out_tx->response_content_encoding = COMPRESSION_GZIP;
} else if ((bstr_cmp_c(ce->value, "deflate") == 0) || (bstr_cmp_c(ce->value, "x-deflate") == 0)) {
connp->out_tx->response_content_encoding = COMPRESSION_DEFLATE;
}
if (connp->out_tx->response_content_encoding != COMPRESSION_NONE) {
connp->out_decompressor = (htp_decompressor_t *) htp_gzip_decompressor_create(connp,
connp->out_tx->response_content_encoding);
if (connp->out_decompressor != NULL) {
connp->out_decompressor->callback = htp_connp_RES_BODY_DECOMPRESSOR_CALLBACK;
} else {
// No need to do anything; the error will have already
// been reported by the failed decompressor.
}
}
}
}
// 1. Any response message which MUST NOT include a message-body
// (such as the 1xx, 204, and 304 responses and any response to a HEAD
// request) is always terminated by the first empty line after the
// header fields, regardless of the entity-header fields present in the
// message.
if (((connp->out_tx->response_status_number >= 100) && (connp->out_tx->response_status_number <= 199))
|| (connp->out_tx->response_status_number == 204) || (connp->out_tx->response_status_number == 304)
|| (connp->out_tx->request_method_number == M_HEAD)) {
// There's no response body
connp->out_state = htp_connp_RES_IDLE;
} else {
// We have a response body
htp_header_t *cl = table_get_c(connp->out_tx->response_headers, "content-length");
htp_header_t *te = table_get_c(connp->out_tx->response_headers, "transfer-encoding");
// 2. If a Transfer-Encoding header field (section 14.40) is present and
// indicates that the "chunked" transfer coding has been applied, then
// the length is defined by the chunked encoding (section 3.6).
if ((te != NULL) && (bstr_cmp_c(te->value, "chunked") == 0)) {
// If the T-E header is present we are going to use it.
connp->out_tx->response_transfer_coding = CHUNKED;
// We are still going to check for the presence of C-L
if (cl != NULL) {
// This is a violation of the RFC
connp->out_tx->flags |= HTP_REQUEST_SMUGGLING;
// TODO
}
connp->out_state = htp_connp_RES_BODY_CHUNKED_LENGTH;
connp->out_tx->progress = TX_PROGRESS_RES_BODY;
}// 3. If a Content-Length header field (section 14.14) is present, its
// value in bytes represents the length of the message-body.
else if (cl != NULL) {
// We know the exact length
connp->out_tx->response_transfer_coding = IDENTITY;
// Check for multiple C-L headers
if (cl->flags & HTP_FIELD_REPEATED) {
connp->out_tx->flags |= HTP_REQUEST_SMUGGLING;
// TODO Log
}
// Get body length
int i = htp_parse_content_length(cl->value);
if (i < 0) {
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0, "Invalid C-L field in response: %d", i);
return HTP_ERROR;
} else {
connp->out_content_length = i;
connp->out_body_data_left = connp->out_content_length;
if (connp->out_content_length != 0) {
connp->out_state = htp_connp_RES_BODY_IDENTITY;
connp->out_tx->progress = TX_PROGRESS_RES_BODY;
} else {
connp->out_state = htp_connp_RES_IDLE;
connp->out_tx->progress = TX_PROGRESS_DONE;
}
}
} else {
// 4. If the message uses the media type "multipart/byteranges", which is
// self-delimiting, then that defines the length. This media type MUST
// NOT be used unless the sender knows that the recipient can parse it;
// the presence in a request of a Range header with multiple byte-range
// specifiers implies that the client can parse multipart/byteranges
// responses.
htp_header_t *ct = table_get_c(connp->out_tx->response_headers, "content-type");
if (ct != NULL) {
// TODO Handle multipart/byteranges
if (bstr_index_of_c_nocase(ct->value, "multipart/byteranges") != -1) {
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0,
"C-T multipart/byteranges in responses not supported");
return HTP_ERROR;
}
}
// 5. By the server closing the connection. (Closing the connection
// cannot be used to indicate the end of a request body, since that
// would leave no possibility for the server to send back a response.)
connp->out_state = htp_connp_RES_BODY_IDENTITY;
connp->out_tx->progress = TX_PROGRESS_RES_BODY;
}
}
// NOTE We do not need to check for short-style HTTP/0.9 requests here because
// that is done earlier, before response line parsing begins
// Run hook RESPONSE_HEADERS_COMPLETE
int rc = hook_run_all(connp->cfg->hook_response_headers, connp);
if (rc != HOOK_OK) {
htp_log(connp, HTP_LOG_MARK, HTP_LOG_ERROR, 0,
"Response headers callback returned error (%d)", rc);
return HTP_ERROR;
}
return HTP_OK;
}
