/**
* Eudoxus first match callback function. Called when a match occurs.
*
* Always returns IA_EUDOXUS_CMD_STOP to stop matching (unless an
* error occurs). If capture is enabled the matched text will be stored in the
* capture variable.
*
* @param[in] engine Eudoxus engine.
* @param[in] output Output defined by automata.
* @param[in] output_length Length of output.
* @param[in] input Current location in the input (first character
* after the match).
* @param[in,out] cbdata Pointer to the ee_callback_data_t instance we are
* handling. This is needed for handling capture
* of the match.
* @return IA_EUDOXUS_CMD_ERROR on error, IA_EUDOXUS_CMD_STOP otherwise.
*/
static
ia_eudoxus_command_t ee_first_match_callback(ia_eudoxus_t* engine,
const char *output,
size_t output_length,
const uint8_t *input,
void *cbdata)
{
assert(cbdata != NULL);
assert(output != NULL);
ib_status_t rc;
uint32_t match_len;
const ee_callback_data_t *ee_cbdata = cbdata;
ib_tx_t *tx = ee_cbdata->tx;
ib_field_t *capture = ee_cbdata->capture;
ib_bytestr_t *bs;
ib_field_t *field;
const char *name;
assert(tx != NULL);
if (capture != NULL) {
if (output_length != sizeof(uint32_t)) {
return IA_EUDOXUS_CMD_ERROR;
}
match_len = *(uint32_t *)(output);
rc = ib_capture_clear(capture);
if (rc != IB_OK) {
ib_log_error_tx(tx, "Error clearing captures: %s",
ib_status_to_string(rc));
return IA_EUDOXUS_CMD_ERROR;
}
/* Create a byte-string representation */
rc = ib_bytestr_dup_mem(&bs,
tx->mp,
(input - match_len),
match_len);
if (rc != IB_OK) {
return IA_EUDOXUS_CMD_ERROR;
}
name = ib_capture_name(0);
rc = ib_field_create(&field, tx->mp, name, strlen(name),
IB_FTYPE_BYTESTR, ib_ftype_bytestr_in(bs));
if (rc != IB_OK) {
return IA_EUDOXUS_CMD_ERROR;
}
rc = ib_capture_set_item(capture, 0, tx->mp, field);
if (rc != IB_OK) {
return IA_EUDOXUS_CMD_ERROR;
}
}
return IA_EUDOXUS_CMD_STOP;
}
/**
* Set the matches into the given field name as .0, .1, .2 ... .9.
*
* @param[in] tx Current transaction.
* @param[in] capture Collection to capture to.
* @param[in] ovector The vector of integer pairs of matches from PCRE.
* @param[in] matches The number of matches.
* @param[in] subject The matched-against string data.
*
* @returns IB_OK or IB_EALLOC.
*/
static
ib_status_t pcre_set_matches(
const ib_tx_t *tx,
ib_field_t *capture,
int *ovector,
int matches,
const char *subject
)
{
assert(tx != NULL);
assert(tx->ib != NULL);
assert(capture != NULL);
assert(ovector != NULL);
ib_status_t rc;
int i;
rc = ib_capture_clear(capture);
if (rc != IB_OK) {
ib_log_error_tx(tx, "Error clearing captures: %s",
ib_status_to_string(rc));
}
/* We have a match! Now populate TX:0-9 in tx->data. */
ib_log_debug2_tx(tx, "REGEX populating %d matches", matches);
for (i = 0; i < matches; ++i)
{
/* The length of the match. */
size_t match_len;
/* The first character in the match. */
const char *match_start;
/* Field name */
const char *name;
/* Holder for a copy of the field value when creating a new field. */
ib_bytestr_t *bs;
/* Field holder. */
ib_field_t *field;
/* Readability. Mark the start and length of the string. */
match_start = subject+ovector[i*2];
match_len = ovector[i*2+1] - ovector[i*2];
/* Create a byte-string representation */
rc = ib_bytestr_dup_mem(&bs,
tx->mp,
(const uint8_t*)match_start,
match_len);
if (rc != IB_OK) {
return rc;
}
/* Create a field to hold the byte-string */
name = ib_capture_name(i);
rc = ib_field_create(&field, tx->mp, name, strlen(name),
IB_FTYPE_BYTESTR, ib_ftype_bytestr_in(bs));
if (rc != IB_OK) {
return rc;
}
/* Add it to the capture collection */
rc = ib_capture_set_item(capture, i, tx->mp, field);
if (rc != IB_OK) {
return rc;
}
}
return IB_OK;
}
/**
* @brief Execute the dfa operator
*
* @param[in] tx Current transaction.
* @param[in] instance_data Instance data needed for execution.
* @param[in] field The field to operate on.
* @param[in] capture If non-NULL, the collection to capture to.
* @param[out] result The result of the operator 1=true 0=false.
* @param[in] cbdata Callback data.
*
* @returns IB_OK most times. IB_EALLOC when a memory allocation error handles.
*/
static
ib_status_t dfa_operator_execute(
ib_tx_t *tx,
void *instance_data,
const ib_field_t *field,
ib_field_t *capture,
ib_num_t *result,
void *cbdata
)
{
assert(instance_data != NULL);
assert(tx != NULL);
int matches;
ib_status_t ib_rc;
const int ovecsize = 3 * MATCH_MAX;
modpcre_operator_data_t *operator_data =
(modpcre_operator_data_t *)instance_data;
int *ovector;
const char *subject;
size_t subject_len;
const ib_bytestr_t *bytestr;
dfa_workspace_t *dfa_workspace;
const char *id = operator_data->id;
int options; /* dfa exec options. */
int start_offset;
int match_count;
const ib_module_t *m = (const ib_module_t *)cbdata;
assert(m != NULL);
assert(operator_data->cpdata->is_dfa == true);
ovector = (int *)malloc(ovecsize*sizeof(*ovector));
if (ovector==NULL) {
return IB_EALLOC;
}
if (field->type == IB_FTYPE_NULSTR) {
ib_rc = ib_field_value(field, ib_ftype_nulstr_out(&subject));
if (ib_rc != IB_OK) {
free(ovector);
return ib_rc;
}
subject_len = strlen(subject);
}
else if (field->type == IB_FTYPE_BYTESTR) {
ib_rc = ib_field_value(field, ib_ftype_bytestr_out(&bytestr));
if (ib_rc != IB_OK) {
free(ovector);
return ib_rc;
}
subject_len = ib_bytestr_length(bytestr);
subject = (const char *) ib_bytestr_const_ptr(bytestr);
}
else {
free(ovector);
return IB_EINVAL;
}
/* Get the per-tx workspace data for this rule data id. */
ib_rc = get_dfa_tx_data(m, tx, id, &dfa_workspace);
if (ib_rc == IB_ENOENT) {
/* First time we are called, clear the captures. */
if (capture) {
ib_rc = ib_capture_clear(capture);
if (ib_rc != IB_OK) {
ib_log_error_tx(tx, "Error clearing captures: %s",
ib_status_to_string(ib_rc));
}
}
options = PCRE_PARTIAL_SOFT;
ib_rc = alloc_dfa_tx_data(m, tx, operator_data->cpdata, id, &dfa_workspace);
if (ib_rc != IB_OK) {
free(ovector);
return ib_rc;
}
}
else if (ib_rc == IB_OK) {
options = PCRE_PARTIAL_SOFT | PCRE_DFA_RESTART;
}
else {
free(ovector);
return ib_rc;
}
/* Perform the match.
* If capturing is specified, then find all matches.
*/
start_offset = 0;
match_count = 0;
do {
matches = pcre_dfa_exec(operator_data->cpdata->cpatt,
operator_data->cpdata->edata,
subject,
subject_len,
start_offset, /* Starting offset. */
options,
ovector,
ovecsize,
dfa_workspace->workspace,
dfa_workspace->wscount);
if (matches > 0) {
++match_count;
/* Use the longest match - the first in ovector -
* to set the offset in the subject for the next
* match.
*/
start_offset = ovector[1] + 1;
if (capture) {
pcre_dfa_set_match(tx, capture, ovector, 1, subject);
}
}
} while (capture && (matches > 0));
if (match_count > 0) {
ib_rc = IB_OK;
*result = 1;
}
else if ((matches == 0) || (matches == PCRE_ERROR_NOMATCH)) {
ib_rc = IB_OK;
*result = 0;
}
else if (matches == PCRE_ERROR_PARTIAL) {
ib_rc = IB_OK;
*result = 0;
}
else {
/* Some other error occurred. Set the status to false and
* return the error. */
ib_rc = IB_EUNKNOWN;
*result = 0;
}
free(ovector);
return ib_rc;
}
// FIXME: This needs to go away and be replaced with dynamic fields
static ib_status_t core_gen_placeholder_fields(ib_engine_t *ib,
ib_tx_t *tx,
ib_state_event_type_t event,
void *cbdata)
{
assert(ib != NULL);
assert(tx != NULL);
assert(tx->data != NULL);
assert(event == tx_started_event);
ib_status_t rc;
ib_field_t *tmp;
/* Core Request Fields */
rc = core_field_placeholder_bytestr(tx->data, "request_line");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_method");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_protocol");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_raw");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_scheme");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_username");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_password");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_host");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_host");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_port");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_path");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_path_raw");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_query");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_uri_fragment");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_content_type");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "request_filename");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "auth_type");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "auth_username");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "auth_password");
if (rc != IB_OK) {
return rc;
}
/* Core Request Collections */
rc = ib_data_add_list(tx->data, "request_headers", NULL);
if (rc != IB_OK) {
return rc;
}
rc = ib_data_add_list(tx->data, "request_cookies", NULL);
if (rc != IB_OK) {
return rc;
}
rc = ib_data_add_list(tx->data, "request_uri_params", NULL);
if (rc != IB_OK) {
return rc;
}
rc = ib_data_add_list(tx->data, "request_body_params", NULL);
if (rc != IB_OK) {
return rc;
}
/* ARGS collection */
rc = ib_data_get(tx->data, "ARGS", &tmp);
if (rc == IB_ENOENT) {
rc = ib_data_add_list(tx->data, "ARGS", NULL);
if (rc != IB_OK) {
return rc;
}
}
else if (rc != IB_OK) {
return rc;
}
/* Flags collection */
rc = ib_data_get(tx->data, "FLAGS", &tmp);
if (rc == IB_ENOENT) {
rc = ib_data_add_list(tx->data, "FLAGS", NULL);
if (rc != IB_OK) {
return rc;
}
}
else if (rc != IB_OK) {
return rc;
}
/* Initialize CAPTURE */
{
ib_field_t *capture;
rc = ib_capture_acquire(tx, NULL, &capture);
if (rc != IB_OK) {
return rc;
}
assert(capture != NULL);
rc = ib_capture_clear(capture);
if (rc != IB_OK) {
return rc;
}
}
/* Core Response Fields */
rc = core_field_placeholder_bytestr(tx->data, "response_line");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "response_protocol");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "response_status");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "response_message");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "response_content_type");
if (rc != IB_OK) {
return rc;
}
/* Core Response Collections */
rc = ib_data_add_list(tx->data, "response_headers", NULL);
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "FIELD_NAME");
if (rc != IB_OK) {
return rc;
}
rc = core_field_placeholder_bytestr(tx->data, "FIELD_NAME_FULL");
if (rc != IB_OK) {
return rc;
}
rc = ib_data_add_list(tx->data, "response_cookies", NULL);
return rc;
}
