正规式和有限自动机二（NFA转DFA）

思路（流程图）：

 

代码（基于正规式和有限自动机一）：

正规式和有限自动机一：https://blog.youkuaiyun.com/Willy__QI/article/details/102981220

from queue import Queue
from GrammarToNFA import GrammarToNFA

# 求集合关于终结符的闭包
# char = 'E' 求集合的ε闭包
def set_Closure(node_list,statusNum,char = 'E'):
    nfaNode_Closure = []
    if(char == 'E'): # 求集合的ε闭包
        visit = [False] * statusNum
        Q = Queue()
        nfaNode_Closure = node_list
        for node in node_list:
            if (not visit[node.stateNum]):
                Q.put(node)
            while (not Q.empty()):
                pre_node = Q.get()
                status = pre_node.stateNum
                if (not visit[status]):
                    visit[status] = True
                    for suf_node in pre_node.nextNodes:
                        if (suf_node.pathChar == char):
                            if suf_node not in nfaNode_Closure: nfaNode_Closure.append(suf_node)
                            if (not visit[suf_node.stateNum]): Q.put(suf_node)
    else:
        for pre_node in node_list:
            for suf_node in pre_node.nextNodes:
                if (suf_node.pathChar == char):
                    if suf_node not in nfaNode_Closure: nfaNode_Closure.append(suf_node)
    return nfaNode_Closure

# 判断当前dfa的划分中是否包含set集合
def exist(dfaNode_divide,status_set):
    exist_flag = False
    values = dfaNode_divide.values()
    # divide
    set_length = len(status_set)
    for key,value in dfaNode_divide.items():
        divide_length = len(value)
        if divide_length == set_length:
            i = 0
            while (i < divide_length):
                if value[i].stateNum not in status_set:
                    break
                i += 1
            if (i == divide_length):
                exist_flag = True
                return [exist_flag,key]
    return [exist_flag]

# 判断该划分中是否包含终态，为dfa最小化的终态组作准备
def in_finalstate_set(tail_state,dfaNode_set_char):
    status_set = []
    for node in dfaNode_set_char:
        status_set.append(node.stateNum)
    if tail_state in status_set:
        return True
    else:
        return False

def NfaToDfa(li): #li = [NFAStack.peek(),gnm.gsn.statusNum + 1,char_list]
    s = 0                   # 出现新状态自增
    final_state = []        # 终态组
    non_final_state = []    # 非终态组
    M = []                  # 状态转换矩阵
    tail_state = li[0].tailNode.stateNum
    # 划分
    start_set = set_Closure([li[0].headNode], li[1])
    if in_finalstate_set(tail_state,start_set): # 判断开始结点的闭包集是否属于终态组
        final_state.append(s)  # 终态组
    else:
        non_final_state.append(s)
    dfaNode_divide = {s:start_set}
    Q = Queue()
    Q.put(start_set)
    while(not Q.empty()):
        state_list = []         # 状态转换矩阵的一行
        dfaNode_set = Q.get()
        for char in li[2]:      # 字符集
            dfaNode_set_char = set_Closure(dfaNode_set, li[1], char) # 先求字符闭包
            dfaNode_set_char = set_Closure(dfaNode_set_char, li[1])  # 再求ε闭包
            if dfaNode_set_char:
                status_set = []  # 字符闭包结点的状态
                for node in dfaNode_set_char:
                    status_set.append(node.stateNum)
                exist_info = exist(dfaNode_divide, status_set)  # [exist_flag,key]
                if not exist_info[0]:
                    s += 1
                    new_item = {s:dfaNode_set_char}
                    dfaNode_divide.update(new_item)
                    state_list.append(s)
                    if tail_state in status_set:
                        final_state.append(s)  # 终态组
                    else:
                        non_final_state.append(s)
                    Q.put(dfaNode_set_char)
                else:
                    state_list.append(exist_info[1])
            else:
                state_list.append(-1)  # 表示不可达的状态
        M.append(state_list)
    return [dfaNode_divide,M,final_state,non_final_state]

def main():
    regExp = str(input("请输入正则式："))
    li = GrammarToNFA().ToNFA(regExp)
    dfaNode_divide_info = NfaToDfa(li)
    dfaNode_divide_dict = dfaNode_divide_info[0]
    for divide in dfaNode_divide_dict.values():
        print("{",end='')
        for node in divide:
            print(str(node.stateNum) + " ",end='')
        print("}\n")
if __name__ == '__main__':
   main()

测试：

使用集合中最小的状态代表整个集合，即：