node *parse (const rules_char_type *start_,
const rules_char_type * const end_, const id_type id_,
const id_type user_id_, const id_type next_dfa_,
const id_type push_dfa_, const bool pop_dfa_,
const std::size_t flags_, id_type &nl_id_, const bool seen_bol_,
const bool macro_)
{
node *root_ = 0;
state state_ (start_, end_, id_, flags_, _locale, macro_);
token *lhs_token_ = 0;
std::auto_ptr<token> rhs_token_ (new token);
char action_ = 0;
_token_stack->push (static_cast<token *>(0));
_token_stack->top () = rhs_token_.release ();
rhs_token_.reset (new token);
tokeniser::next (_token_stack->top (), state_, rhs_token_.get ());
do
{
lhs_token_ = _token_stack->top ();
action_ = lhs_token_->precedence (rhs_token_->_type);
switch (action_)
{
case '<':
case '=':
_token_stack->push (static_cast<token *>(0));
_token_stack->top () = rhs_token_.release ();
rhs_token_.reset (new token);
tokeniser::next (_token_stack->top (), state_,
rhs_token_.get ());
break;
case '>':
reduce (state_);
break;
default:
{
std::ostringstream ss_;
ss_ << "A syntax error occurred: '" <<
lhs_token_->precedence_string () <<
"' against '" << rhs_token_->precedence_string () <<
"' preceding index " << state_.index () <<
" in rule id " <<
state_._id << '.';
throw runtime_error (ss_.str ());
break;
}
}
} while (!_token_stack->empty ());
if (_tree_node_stack.empty ())
{
std::ostringstream ss_;
ss_ << "Empty rules are not allowed in rule id " <<
state_._id << '.';
throw runtime_error (ss_.str ());
}
assert (_tree_node_stack.size () == 1);
node *lhs_node_ = _tree_node_stack.top ();
_tree_node_stack.pop ();
if (macro_)
{
// Macros have no end state...
root_ = lhs_node_;
}
else
{
_node_ptr_vector->push_back (static_cast<end_node *>(0));
node *rhs_node_ = new end_node (id_, user_id_, next_dfa_,
push_dfa_, pop_dfa_);
_node_ptr_vector->back () = rhs_node_;
_node_ptr_vector->push_back (static_cast<sequence_node *>(0));
_node_ptr_vector->back () = new sequence_node
(lhs_node_, rhs_node_);
root_ = _node_ptr_vector->back ();
}
if (seen_bol_)
{
fixup_bol (root_);
}
if (state_._nl_id != static_cast<id_type>(~0))
{
nl_id_ = state_._nl_id;
}
if ((flags_ & match_zero_len) == 0)
{
const typename node::node_vector &firstpos_ = root_->firstpos();
typename node::node_vector::const_iterator iter_ =
firstpos_.begin ();
typename node::node_vector::const_iterator end_ =
firstpos_.end ();
for (; iter_ != end_; ++iter_)
{
const node *node_ = *iter_;
if (node_->end_state ())
{
std::ostringstream ss_;
ss_ << "Rules that match zero characters are not allowed "
"as this can cause an infinite loop in user code. The "
"match_zero_len flag overrides this check. Rule id " <<
state_._id << '.';
throw runtime_error (ss_.str ());
}
}
}
return root_;
}
node *parse(const token_deque ®ex_, const id_type id_,
const id_type user_id_, const id_type next_dfa_,
const id_type push_dfa_, const bool pop_dfa_,
const std::size_t flags_, id_type &nl_id_, const bool seen_bol_)
{
typename token_deque::const_iterator iter_ = regex_.begin();
typename token_deque::const_iterator end_ = regex_.end();
node *root_ = 0;
token *lhs_token_ = 0;
// There cannot be less than 2 tokens
std::auto_ptr<token> rhs_token_(new token(*iter_++));
char action_ = 0;
_token_stack->push(static_cast<token *>(0));
_token_stack->top() = rhs_token_.release();
rhs_token_.reset(new token(*iter_));
if (iter_ + 1 != end_) ++iter_;
do
{
lhs_token_ = _token_stack->top();
action_ = lhs_token_->precedence(rhs_token_->_type);
switch (action_)
{
case '<':
case '=':
_token_stack->push(static_cast<token *>(0));
_token_stack->top() = rhs_token_.release();
rhs_token_.reset(new token(*iter_));
if (iter_ + 1 != end_) ++iter_;
break;
case '>':
reduce(nl_id_);
break;
default:
{
std::ostringstream ss_;
ss_ << "A syntax error occurred: '" <<
lhs_token_->precedence_string() <<
"' against '" << rhs_token_->precedence_string() <<
" in rule id " << id_ << '.';
throw runtime_error(ss_.str());
break;
}
}
} while (!_token_stack->empty());
if (_tree_node_stack.empty())
{
std::ostringstream ss_;
ss_ << "Empty rules are not allowed in rule id " <<
id_ << '.';
throw runtime_error(ss_.str());
}
assert(_tree_node_stack.size() == 1);
node *lhs_node_ = _tree_node_stack.top();
_tree_node_stack.pop();
_node_ptr_vector->push_back(static_cast<end_node *>(0));
node *rhs_node_ = new end_node(id_, user_id_, next_dfa_,
push_dfa_, pop_dfa_);
_node_ptr_vector->back() = rhs_node_;
_node_ptr_vector->push_back(static_cast<sequence_node *>(0));
_node_ptr_vector->back() = new sequence_node
(lhs_node_, rhs_node_);
root_ = _node_ptr_vector->back();
if (seen_bol_)
{
fixup_bol(root_);
}
if ((flags_ & match_zero_len) == 0)
{
const typename node::node_vector &firstpos_ = root_->firstpos();
typename node::node_vector::const_iterator iter_ =
firstpos_.begin();
typename node::node_vector::const_iterator end_ =
firstpos_.end();
for (; iter_ != end_; ++iter_)
{
const node *node_ = *iter_;
if (node_->end_state())
{
std::ostringstream ss_;
ss_ << "Rules that match zero characters are not allowed "
"as this can cause an infinite loop in user code. The "
"match_zero_len flag overrides this check. Rule id " <<
id_ << '.';
throw runtime_error(ss_.str());
}
}
}
return root_;
}
