简体   繁体   English

如何使用马尔可夫链获得所有可能性?

[英]How can I get all possibilities using a markov chain?

I am trying to implement a linguistic passphrase cracker using markov chains. 我正在尝试使用markov链实现语言密码破解程序。

The idea behind this is to chose n-grams from a text, select a starting n-gram (usually a word that is at the beginning of a sentence) and represent it as a state, using the first n-1 characters. 其背后的思想是从文本中选择n-gram,选择起始n-gram(通常是句子开头的单词),并使用前n-1个字符将其表示为状态。 As an example, for "the" I will have "th". 例如,对于“ the”,我将拥有“ th”。 This will have a list of letters with their occurences, and will be represented as a dictionary. 这将具有出现字母的列表,并将其表示为字典。 dict["th"] = [('e', 120), ('a', 79)] etc. For each of these values I will try to create a markov chain that will satisfy either my password or my password length. dict["th"] = [('e', 120), ('a', 79)]等。对于这些值,我将尝试创建一个满足我的密码或密码长度的markov链。 What that means is that when the markov chain has the same length as my password that I am trying to find, I will stop the execution and check if the markov chain is the same with my password. 这意味着当markov链与我要查找的密码具有相同的长度时,我将停止执行并检查markov链与我的密码是否相同。 I am trying to implement this using a recursive function but for some reason I am getting stack overflow. 我试图使用递归函数来实现这一点,但是由于某种原因,我正在堆栈溢出。

def ceva(myTry, good_all, pwd, guess, level):
        save = myTry
        if len(pwd) == len(guess):
            if pwd == guess:
                return 1
        else:
            if myTry in good_all.keys():
                values = good_all[myTry]
                for i in range(0,len(values)):
                    #print(i, len(values))
                    letter = values[i][0]
                    #print("First",myTry, letter)

                    pwd += letter
                    if i != len(values)-1:

                        if len(pwd) == len(guess):
                            #print("In if", pwd, myTry)
                            if pwd == guess:
                                print("I found:", pwd)
                                return 1
                            else:

                                pwd = pwd[0:len(pwd)-1]
                        else:

                            myTry += letter
                            myTry = myTry[1:]
                            #print("In else: ",pwd, myTry)
                            return ceva(myTry, good_all, pwd, guess, level)
                    else:
                        if len(pwd) == len(guess):
                            #print("In if", pwd, myTry)
                            if pwd == guess:
                                print("I found:", pwd)
                                return 1

                        pwd = pwd[0:len(pwd)-1]


    for key, letterList in starter_follows.items():
        myTry = key.replace("_", "")

        # i will not treat the case when the starting phrase
        # is a single character
        if myTry == "i":
            pass
        else:
            for letter in letterList:

                if letter[0] not in "_.-\"!":
                    myTry += letter[0]
                    pwd = copy.copy(myTry)
                    #print("Starter:", pwd)
                    res=ceva(myTry, good_all, pwd, toGuess, 1)
                    myTry = myTry[0:len(myTry)-1]

With this algorithm i am reaching the maximum recursion depth. 通过这种算法,我达到了最大递归深度。 But I am trying to obtain all the markov chains until the passphrase is found. 但是我试图获取所有的马尔可夫链,直到找到密码短语为止。

EDIT 1: Now, with the updated code, the password is found but only because I am looping thorugh all the possible last letters. 编辑1:现在,使用更新的代码,找到了密码,但这仅是因为我正在循环所有可能的最后一个字母。

Eg: "indeed" 例如:“确实”

ind is already in my list of starters, and all the tri-grams I am finding have "e" as their most common next letter. ind已经在我的入门列表中,我发现的所有三字母都以“ e”作为最常见的下一个字母。 So e is added, then the next e, then a next e and now the password is "indeee", but i am slicing the last letter and going through the for again, and it ultimately finds "indeed", which is okay. 因此,先添加e,再添加下一个e,再添加下一个e,现在密码为“独立”,但是我将最后一个字母切成薄片并再次输入for,它最终找到“独立”,这是可以的。 The problem is that if I will give indedd it will not find my password, because the second "d" is never looped through. 问题是,如果我提供indedd ,它将找不到我的密码,因为第二个“ d”永远不会循环通过。 How can I go back in my iteration and loop through all possible letters at all levels? 如何返回迭代并遍历所有级别的所有可能字母?

I managed to pull it off eventually thanks to the answers given. 由于给出的答案,我最终设法完成了任务。 I am posting the working algorithm in hope that it will help someone someday. 我发布了有效的算法,希望有一天能对某人有所帮助。

def ceva(myTry, good_all, pwd, guess, flag):

if len(pwd) > len(guess):
    return 0
if len(pwd) == len(guess):
    if pwd == guess:
        flag = 1
        return 1
    else: 
        return 0
save = copy.copy(myTry)
#print("Start of functionn:", "[", pwd, ",", myTry, "]")

if flag == 1:
    return 1
else:   

    if myTry in good_all.keys():
        # get the list of letters for this specific trigram
        values = good_all[myTry]

        if len(pwd) <= len(guess):
            for i in range(0,len(values)):
                #myTry = copy.copy(save)
                # get the letter
                letter = values[i][0]

                # add the letter to the password
                pwd += letter


                # if i found the password, set flag to 1 and break the loop
                if pwd == guess:
                    flag = 1
                    print("I found:", pwd)
                    return 1

                # add the new letter to the trigram, and the get rid of the first
                # letter, in order to create the new trigram
                myTry += letter
                myTry = myTry[1:]
                #print("Pwd before cutting: [", pwd, "]")
                res = ceva(myTry, good_all, pwd, guess, flag)
                #print(res)
                if res == 0:
                    pwd = pwd[0:len(pwd)-1]
                    myTry = pwd[-3:]
                #   print("This is after stop: [", pwd,",", myTry, "]")
                else:
                    return 1


    if flag == 0:
        return 0
    else:
        return 1

flag = 0

for key, letterList in starter_follows.items():
    myTry = key.replace("_", "")
    # i will not treat the case when the starting phrase
    # is a single character
    if myTry == "i":
        pass

    else:
        #print("Aaaa")
        for letter in letterList:

            if letter[0] not in "_.-\"!":

                # add the letter to create a (n-1)-gram
                myTry += letter[0]
                # create a copy of the starting password
                pwd = copy.copy(myTry)

                # call the recursive function
                res=ceva(myTry, good_all, pwd, toGuess, flag)
                # go back to the begining, to try a new start
                # e.g.: "the" first, "tha" second
                myTry = myTry[0:len(myTry)-1]

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM