void my_regex::NFA::create_nfa_node(const my_string ®)
{
// 第一遍遍历的时候主要处理\开头引出的特殊符号
std::vector<p_element> element_vec;
element_vec.reserve(reg.length());
p_element elem;
for (int i = 0; i < reg.length(); i++)
{
switch (reg[i])
{
case my_char('+'):
case my_char('?'):
case my_char('*'):
elem = p_element(new num_symbol(reg[i]));
element_vec.push_back(elem);
break;
case my_char('\\'):
elem = p_element(new num_symbol(reg[++i]));// 注意，这里现在是把所有跟在反斜杠后面的字符都当作普通字符来处理
element_vec.push_back(elem);
break;
case my_char('.'):
elem = p_element(new point_symbol);
element_vec.push_back(elem);
break;
case my_char('('):
elem = p_element(new open_parentheses);
element_vec.push_back(elem);
break;
case my_char(')'):
elem = p_element(new close_parentheses);
element_vec.push_back(elem);
break;
case my_char('|'):
elem = p_element(new or_symbol);
element_vec.push_back(elem);
break;
default:
elem = p_element(new character(reg[i]));
element_vec.push_back(elem);
break;
}
}
// 将基础元素转化为前缀表达式
element_vec_2_prefix_exp(element_vec);
//assert(element_vec.size() == 1);//最后只剩下一个大的前缀表达式//已经在函数里判断过了
p_nfa_node begin(new nfa_node), end(new nfa_node);
prefix_exp_2_nfa_node(element_vec, begin, end);
end->accept = true;
// todo:加入清理内存的操作
start = begin;
}
size_t my_regex::NFA::try_match(const my_string &str)
{
size_t farthest = 0;//记录匹配到的最远距离
std::vector<cp_nfa_node> current_pnodes, next_pnodes;
current_pnodes.push_back(start);
get_epson_closure(current_pnodes);
bool matched = start->accept;//当前字符匹配后是否满足正则表达式
for (auto &cur_node : current_pnodes)
{
matched |= cur_node->accept;
}
for (size_t i = 0; i < str.length(); i++)
{
matched = false;
next_pnodes.clear();
for (auto &cur_node : current_pnodes)
{
// 先沿着正常边走
for (auto &n_edg : cur_node->norm_edgs)
{
if (n_edg.c == str[i])
{
next_pnodes.push_back(n_edg.p_next_node);
}
}
// 小数点通配符
for (auto &p_node : cur_node->point_edgs)
{
next_pnodes.push_back(p_node.p_next_node);
}
}
if (next_pnodes.empty())
{
// 匹配不下去了，停止
return farthest;
}
else
{
current_pnodes.swap(next_pnodes);
get_epson_closure(current_pnodes);
for (auto &cur_node : current_pnodes)
{
matched |= cur_node->accept;
}
if (matched)
{
farthest = i+1;
}
}
}
return farthest;
}
my_string::my_string(my_string const &str) : data_(new char[str.length() + 1])
{
strcpy(data_, str.c_str());
}