/**
* Search patterns of the ac_tree matcher in the given buffer using a
* matching context. The matching context stores offsets used to process
* a search over multiple data segments. The function has option flags to
* specify to return where the first pattern is found, or after all the
* data is consumed, using a user specified callback and/or building a
* list of patterns matched
*
* @param ac_ctx pointer to the matching context
* @param data pointer to the buffer to search in
* @param len the length of the data
* @param flags options to use while matching
* @param mp memory pool to use
*
* @returns Status code
*/
ib_status_t ib_ac_consume(ib_ac_context_t *ac_ctx,
const char *data,
size_t len,
uint8_t flags,
ib_mpool_t *mp)
{
const char *end;
ib_ac_state_t *state = NULL;
ib_ac_state_t *fgoto = NULL;
int flag_match = 0;
ib_ac_t *ac_tree = ac_ctx->ac_tree;
ac_ctx->current_offset = 0;
if ((ac_ctx->ac_tree->flags & IB_AC_FLAG_PARSER_COMPILED) == 0)
{
ib_ac_build_links(ac_tree);
}
ac_tree = ac_ctx->ac_tree;
if (ac_ctx->current == NULL) {
ac_ctx->current = ac_tree->root;
}
state = ac_ctx->current;
end = data + len;
while (data < end) {
ib_ac_char_t letter = (unsigned char)*data++;
++ac_ctx->processed;
++ac_ctx->current_offset;
if (ac_tree->flags & IB_AC_FLAG_PARSER_NOCASE) {
letter = tolower(letter);
}
fgoto = NULL;
while (fgoto == NULL) {
fgoto = ib_ac_bintree_goto(state, letter);
if (fgoto != NULL) {
if (fgoto->flags & IB_AC_FLAG_STATE_OUTPUT) {
flag_match = 1;
++fgoto->match_cnt;
++ac_ctx->match_cnt;
ac_ctx->current = fgoto;
state = fgoto;
if (flags & IB_AC_FLAG_CONSUME_DOCALLBACK)
{
ib_ac_do_callback(ac_ctx, state);
}
if (flags & IB_AC_FLAG_CONSUME_DOLIST)
{
/* If list is not created yet, create it */
if (ac_ctx->match_list == NULL)
{
ib_status_t rc;
rc = ib_list_create(&ac_ctx->match_list, mp);
if (rc != IB_OK) {
return rc;
}
}
ib_ac_match_t *mt = NULL;
mt = (ib_ac_match_t *)ib_mpool_calloc(mp,
1, sizeof(ib_ac_match_t));
if (mt == NULL) {
return IB_EALLOC;
}
mt->pattern = state->pattern;
mt->data = state->data;
mt->pattern_len = state->level + 1;
mt->offset = ac_ctx->processed - (fgoto->level + 1);
mt->relative_offset = ac_ctx->current_offset -
(fgoto->level + 1);
ib_list_enqueue(ac_ctx->match_list, (void *) mt);
}
if ( !(flags & IB_AC_FLAG_CONSUME_MATCHALL))
{
return IB_OK;
}
ib_ac_state_t *outs = NULL;
for (outs = state->outputs;
outs != NULL;
outs = outs->outputs)
{
/* This are subpatterns of the current walked branch
* that are present as independent patterns as well
* in the tree */
++outs->match_cnt;
++ac_ctx->match_cnt;
if (flags & IB_AC_FLAG_CONSUME_DOCALLBACK)
{
ib_ac_do_callback(ac_ctx, outs);
}
if (flags & IB_AC_FLAG_CONSUME_DOLIST)
{
/* If list is not created yet, create it */
if (ac_ctx->match_list == NULL)
{
ib_status_t rc;
rc = ib_list_create(&ac_ctx->match_list, mp);
if (rc != IB_OK) {
return rc;
}
}
ib_ac_match_t *mt = NULL;
mt = (ib_ac_match_t *)ib_mpool_calloc(mp,
1, sizeof(ib_ac_match_t));
if (mt == NULL) {
return IB_EALLOC;
}
mt->pattern = outs->pattern;
mt->data = state->data;
mt->pattern_len = outs->level + 1;
mt->offset =
ac_ctx->processed - (outs->level + 1);
mt->relative_offset = ac_ctx->current_offset -
(outs->level + 1);
ib_list_enqueue(ac_ctx->match_list,
(void *) mt);
}
}
}
}
else {
/* if goto() failed, look at the fail states */
if (state != NULL && state->fail != NULL) {
if (state == ac_tree->root) {
break;
}
state = state->fail;
}
else {
state = ac_tree->root;
}
}
}
if (fgoto != NULL) {
state = fgoto;
}
ac_ctx->current = state;
}
ac_ctx->current = state;
/* If we have a match, return ok. Otherwise return IB_ENOENT */
if (flag_match == 1) {
return IB_OK;
}
return IB_ENOENT;
}
/**
* Push @a value to front of list.
*
* @param[in] value Value to push to back of list.
**/
void push_front(T value) const
{
Internal::throw_if_error(
ib_list_enqueue(m_ib, Internal::value_as_void(value))
);
}
/**
* Constructs fail links of branches (the failure transition function)
*
* @param ac_tree the ac tree matcher
*
* @return ib_status_t status of the operation
*/
static ib_status_t ib_ac_link_fail_states(ib_ac_t *ac_tree)
{
ib_status_t rc;
ib_ac_state_t *child = NULL;
ib_ac_state_t *state = NULL;
ib_ac_state_t *goto_state = NULL;
ib_list_t *iter_queue = NULL;
if (ac_tree->flags & IB_AC_FLAG_PARSER_COMPILED) {
return IB_OK;
}
ac_tree->root->pattern = 0;
rc = ib_list_create(&iter_queue, ac_tree->mp);
if (rc != IB_OK) {
return rc;
}
ac_tree->root->fail = ac_tree->root;
/* All first-level children will fail back to root state */
for (child = ac_tree->root->child;
child != NULL;
child = child->sibling)
{
child->fail = ac_tree->root;
rc = ib_list_enqueue(iter_queue, (void *) child);
if (rc != IB_OK) {
return rc;
}
}
while (ib_list_elements(iter_queue) > 0) {
rc = ib_list_dequeue(iter_queue, (void *) &state);
if (rc != IB_OK) {
return rc;
}
state->fail = ac_tree->root;
if (state->parent != ac_tree->root) {
goto_state = ib_ac_child_for_code(state->parent->fail,
state->letter);
if (goto_state != NULL) {
state->fail = goto_state;
}
}
for (child = state->child;
child != NULL;
child = child->sibling)
{
rc = ib_list_enqueue(iter_queue, (void *) child);
if (rc != IB_OK) {
return rc;
}
}
}
/* Link common outputs of subpatterns present in the branch*/
ib_ac_link_outputs(ac_tree, ac_tree->root);
/* Unlink invalid fail transitions. This guarantees that there will
* be at least one letter with transition in each fail state*/
ib_ac_unlink_unuseful(ac_tree, ac_tree->root);
if (ac_tree->root->child != NULL) {
ib_ac_build_bintree(ac_tree, ac_tree->root);
}
ac_tree->flags |= IB_AC_FLAG_PARSER_COMPILED;
return IB_OK;
}