generate.py

# A python script used to generate password dictionary with custom settings
# Created by Junhan Duan, Aug. 18, 2020

from pypinyin import *
import itertools, argparse

customListNum = []
customListLet = []
customListRes = []
customListSym = []
extraList = []
resultList = []
symbolList = ['!', '@', '#', '$', '%', '&', '*', '_']

# Do cartesian product between %1(list) and %2(list), and write to %3(list)
def doCartesian(head, tail, dstList):
    for x in itertools.product(head, tail):
        temp = ''.join(list(x[0]))
        temp += ''.join(list(x[1]))
        dstList.append(temp)

# Determine whether %1(string) contains Chinese character
def containsChinese(str):
    for _char in str:
        if '\u4e00' <= _char <= '\u9fa5':
            return True
    return False

# Read content from %1(file) and write it to %2（list）
def readFile(filename, dstList):
    file = open(filename, 'r')
    for line in file:
        line = line.replace('\n', '').replace('\r', '')
        dstList.append(line.strip())

# Read the user-input custom file and do basic list initialization
def readCustom(filename):
    file = open(filename, encoding='utf-8')
    for line in file:
        line = line.replace('\n', '').replace('\r', '')
        # 英文字符自动加入Head，数字加入Tail
        if not containsChinese(line):
            if line.isnumeric():
                customListNum.append(line.strip())
            else:
                customListLet.append(line.strip())
        # 中文字符转字母
        else:
            # 中文字符的拼音整体
            customListLet.append(''.join(lazy_pinyin(line)))
            # 中文字符的拼音整体 - 首字母大写
            customListLet.append(''.join(lazy_pinyin(line)).capitalize())
            # 中文字符的拼音整体 - 全部大写
            customListLet.append(''.join(lazy_pinyin(line)).upper())
            # 中文字符的拼音首字母
            tempList = []
            combLetters = pinyin(line, style=Style.FIRST_LETTER)
            for _char in combLetters:
                tempList.append(''.join(_char))
            customListLet.append(''.join(tempList))
            # 中文字符的拼音首字母 - 首字母大写
            tempList = []
            combLetters = pinyin(line, style=Style.FIRST_LETTER)
            for _char in combLetters:
                tempList.append(''.join(_char))
            customListLet.append(''.join(tempList).capitalize())
            # 中文字符的拼音首字母 - 全部大写
            tempList = []
            combLetters = pinyin(line, style=Style.FIRST_LETTER)
            for _char in combLetters:
                tempList.append(''.join(_char).upper())
            customListLet.append(''.join(tempList))
            # 中文字符的前两个字
            if len(line) >= 2:
                firstTwo = line[:2]
                customListLet.append(''.join(lazy_pinyin(firstTwo)))
            # 中文字符的后两个字（通常为名字)
            if len(line) > 2:
                lastFirst = line[-1]
                lastSecond = line[-2]
                lastTwo = ''.join([lastSecond, lastFirst])
                customListLet.append(''.join(lazy_pinyin(lastTwo)))
            # 中文字符的前两个字 - 全部大写
            if len(line) > 2:
                firstTwo = line[:2]
                customListLet.append(''.join(lazy_pinyin(firstTwo)).upper())
            # 中文字符的后两个字（通常为名字) - 全部大写
            if len(line) > 2:
                lastFirst = line[-1]
                lastSecond = line[-2]
                lastTwo = ''.join([lastSecond, lastFirst])
                customListLet.append(''.join(lazy_pinyin(lastTwo)).upper())
    # 搭配自定字母 & 自定数字 & 符号
    # 字母 + 数字
    doCartesian(customListLet, customListNum, customListRes)
    # 字母 + 符号 + 数字
    doCartesian(customListLet, symbolList, customListSym)
    doCartesian(customListSym, customListNum, customListRes)
    # 字母 + 数字 + 符号
    tempList = []
    doCartesian(customListLet, customListNum, tempList)
    doCartesian(tempList, symbolList, customListRes)

# Strenghten custom password list by adding postfix from list
def strCustomL(strList):
    tempList = []
    # 字母 + file内容
    for x in itertools.product(customListLet, strList):
        temp = ''.join(list(x[0]))
        temp += ''.join(list(x[1]))
        customListRes.append(temp)
        tempList.append(temp)
    # 字母 + 符号 + file内容
    doCartesian(customListSym, strList, customListRes)
    # 字母 + file内容 + 符号
    doCartesian(tempList, symbolList, customListRes)

# Strenghten custom passwords generated by custom file by adding postfix from file input
def strCustomF(filename):
    fileList = []
    readFile(filename, fileList)
    strCustomL(fileList)

# Write list elements to file
def writeFile(filename, fromList):
    with open(filename, 'w') as file:
        for x in fromList:
            if len(x) <= 16:
                file.write("{}\n".format(x))

# Remove redundant password from a list
def removeRedundant(finalList):
    finalList = list(dict.fromkeys(finalList))

# 将部分纯数字密码加入Result
def addNumPswd(srcFile, dstList):
    srcLine = []
    readFile(srcFile, srcLine)
    for x in srcLine:
        if 6 <= len(x) <= 16:
            dstList.append(x)

# 构造经典密码词典 + 数字
def genClassic():
    numLine = []
    typicalLine = []
    readFile('Typical.txt', typicalLine)
    readFile('Numbers.txt', numLine)
    readFile('Typical.txt', extraList)
    doCartesian(typicalLine, numLine, extraList)

def cmd():
    args = argparse.ArgumentParser(description = 'Argument')
    args.add_argument('-c', '--complex', type = int, dest = 'complex', help = 'Complexity of password generated')
    args.add_argument('-f', '--file', type = str, dest = 'filename', help = 'The name of custom file to be input')
    args = args.parse_args()

    if args.complex is not None:
        if args.complex >= 1:
            genClassic()
            readFile('Top100.txt', extraList)
        if args.complex >= 2 and args.filename is not None:
            readCustom(args.filename)
            strCustomF('Numbers.txt')
            addNumPswd('Numbers.txt', customListRes)
        elif args.complex >= 2 and args.filename is None:
            print('No custom file detected. Please specify the filepath/filename with [-f]/[--file] option')
        if args.complex == 3:
            readFile('Top500.txt', extraList)
            readFile('Numbers.txt', extraList)
            readFile('Extra.txt', extraList)
    else:
        if args.filename is None:
            print('No [-c]/[--complex] defined, using default complexity level 1')
            genClassic()
            readFile('Top100.txt', extraList)
        else:
            print('No [-c]/[--complex] defined but file specified, using complexity level 2')
            genClassic()
            readFile('Top100.txt', extraList)
            readCustom(args.filename)
            strCustomF('Numbers.txt')
            addNumPswd('Numbers.txt', customListRes)
            
    resultList = extraList + customListRes
    removeRedundant(resultList)
    writeFile('Result.txt', resultList)
    print('Password has been written to Result.txt')

if __name__ == '__main__':
    cmd()