python-生成随机包含要分析字符的字符串

File information

2010-3-14

磁针石：xurongzhong#gmail.com

博客：oychw.cublog.cn

这个脚本难在编码这块，尤其是设置默认编码。由于linux和windows和linux的默认编码，出现了程序在windows可以运行，但是linux无法运行的情况。
折腾了我几个小时。以下是脚本：

#!/usr/bin/env python
# -*- coding: gbk -*-
#用于指纹原文生成的脚本

import random  
import string  
import sys

#设定字符编码为GBK
reload(sys)
sys.setdefaultencoding('gbk')


def GBK1():  
    head = random.randint(0xA1, 0xA9)  
    tail = random.randint(0xA1, 0xFE)  
    val = ( head << 8 ) | tail  
    word = "%x" % val  
    return word.decode('hex')

def GBK2():  
    head = random.randint(0xB0, 0xF7)  
    tail = random.randint(0xA1, 0xFE)  
    val = ( head << 8 ) | tail  
    word = "%x" % val  
    return word.decode('hex')

def GBK3():  
    head = random.randint(0x81,0xA0)  
    tail = random.randint(0x40, 0xFE)  
    val = ( head << 8 ) | tail  
    word = "%x" % val  
    return word.decode('hex')

def GBK4():  
    head = random.randint(0xAA,0xFE)  
    tail = random.randint(0x40, 0xA0)  
    val = ( head << 8 ) | tail  
    word = "%x" % val  
    return word.decode('hex')

def GBK5():  
    head = random.randint(0xA8,0xA9)  
    tail = random.randint(0x40, 0xA0)  
    val = ( head << 8 ) | tail  
    word = "%x" % val  
    return word.decode('hex')

#产生大小写字母和数字
def ASCII():  
    if not random.randint(0,9):
        word = random.randint(48,57)
    else:   
        word = random.randint(65,90)
        if random.randint(0,1):
            word = word +32
    return chr(word)

#中要处理的GBK码
def GBK_NON():  
    l_gbk = [0xA1A2,0xA1A3,0xA1A4,0xA1AA,0xA1AB,0xA1AD,0xA1AE,0xA1AF,0xA1B0,0xA1B1,0xA1B2,0xA1B3,0xA1B4,0xA1B5,0xA1B6,0xA1B7,0xA1B8,0xA1B9,0xA1BA,0xA1BB,0xA1BC,0xA1BD,0xA1BE,0xA1BF,0xA3A1,0xA3A2,0xA3AC,0xA3A8,0xA3A9,0xA3BA,0xA3BB,0xA3BF]
    word = l_gbk[random.randint(0,31)]
    word = "%x" % word  
    return word.decode('hex')

#中要处理的ASCII码
def ASCII_NON():  
    l_ascii = [20,42]
    word = l_ascii[random.randint(0,1)]
    return chr(word)

#随机生成GBK码，其中一半来自常用汉字
def GBK():  
    char_type = random.randint(0,9)
    if char_type < 5:
        word = GBK2()
        return word
    if char_type == 6:
        word = GBK1()    
        return word    
    if char_type == 7:
        word = GBK3()    
        return word   
    if char_type == 8:
        word = GBK4()    
        return word  
    if char_type == 9:
        word = GBK5()    
        return word      

f = open("word.txt",'w')

for num in range(1000):
    length = random.randint(0,300)
    
   
    #定义生成的字符串，以汉字为主
    char_type = random.randint(0,9)
   
    #随机生成GBK中的汉字
    if char_type < 5:
        for i in range(length):
            word = GBK()
            if word:
                f.write(word)
           
    #随机生成GBK中的汉字,夹杂非法字符           
    if char_type == 5:
        for i in range(length):
            #每5个字串有一个特殊字符，80%的情况插入英文非法字符
            if not random.randint(0,5):
                if random.randint(0,1):
                    word = GBK_NON()
                else:
                    word = ASCII_NON()
            else:
                word = word = GBK()
            if word:
                f.write(word) 

    #随机生成英文,夹杂非法字符           
    if char_type >7 :
        for i in range(length):
            if not random.randint(0,5):
                if random.randint(0,1):
                    word = GBK_NON()
                else:
                    word = ASCII_NON()
            else:
                word = ASCII()
            if word:
                f.write(word)
           
    #随机生成GBK中的英文          
    if char_type == 7 :
        for i in range(length):
            word = ASCII()
            if word:
                f.write(word)         
       
 
    #随机生成GBK中的中英文,夹杂非法字符           
    if char_type == 5:
        for i in range(length):
            if not random.randint(0,5):
                if random.randint(0,1):
                    word = GBK_NON()
                else:
                    word = ASCII_NON()
            else:
                word = GBK()
            if not random.randint(0,5):
                word = ASCII()
            if word:
                f.write(word)
    f.write("\n")

f.close()

附上GBK的编码简表：
　　GBK字符集范围
　　分区                      高位     低位
　　----------------------------------------------
　　●GBK/1:GB2312非汉字符号: A1~A9 || A1~FE
　　●GBK/2:GB2312汉字      : B0~F7 || A1~FE
　　●GBK/3:扩充汉字        : 81~A0 || 40~FE
　　●GBK/4:扩充汉字        : AA~FE || 40~A0
　　●GBK/5:扩充非汉字      : A8~A9 || 40~A0

汉字在Unicode中的分布大致如下表：

	首字编码	尾字编码	个数
基本汉字	U4E00	U9FBF	20928
异性字	UF900	UFAFF	512
扩展A	U3400	U4D8F	512
扩展B	U20000	U2A6DF	42720
补充	U2F800	U2FA1F	544
其他

随机生成中文验证码的python代码

, 由admin创建.
来源：%E9%9A%8F%E6%9C%BA%E7%94%9F%E6%88%90%E4%B8%AD%E6%96%87%E9%AA%8C%E8%AF%81%E7%A0%81%E7%9A%84python%E4%BB%A3%E7%A0%81/

随机生成中文验证码的python代码
需要准备字体文件，或者通过fontPath参数指定。
默认是我用的字体wqy.ttc.
代码如下:

# -*- coding: utf-8 -*-
import Image,ImageDraw,ImageFont
import random
import math, string

class RandomChar():
  """用于随机生成汉字"""
  @staticmethod
  def Unicode():
    val = random.randint(0x4E00, 0x9FBF)
    return unichr(val)

  @staticmethod
  def GB2312():
    head = random.randint(0xB0, 0xCF)
    body = random.randint(0xA, 0xF)
    tail = random.randint(0, 0xF)
    val = ( head << 8 ) | (body << 4) | tail
    str = "%x" % val
    return str.decode('hex').decode('gb2312')

class ImageChar():
  def __init__(self, fontColor = (0, 0, 0),
                     size = (100, 40),
                     fontPath = 'wqy.ttc',
                     bgColor = (255, 255, 255),
                     fontSize = 20):
    self.size = size
    self.fontPath = fontPath
    self.bgColor = bgColor
    self.fontSize = fontSize
    self.fontColor = fontColor
    self.font = ImageFont.truetype(self.fontPath, self.fontSize)
    self.image = Image.new('RGB', size, bgColor)

  def rotate(self):
    self.image.rotate(random.randint(0, 30), expand=0)

  def drawText(self, pos, txt, fill):
    draw = ImageDraw.Draw(self.image)
    draw.text(pos, txt, font=self.font, fill=fill)
    del draw

  def randRGB(self):
    return (random.randint(0, 255),
           random.randint(0, 255),
           random.randint(0, 255))

  def randPoint(self):
    (width, height) = self.size
    return (random.randint(0, width), random.randint(0, height))

  def randLine(self, num):
    draw = ImageDraw.Draw(self.image)
    for i in range(0, num):
      draw.line([self.randPoint(), self.randPoint()], self.randRGB())
    del draw

  def randChinese(self, num):
    gap = 5
    start = 0
    for i in range(0, num):
      char = RandomChar().GB2312()
      x = start + self.fontSize * i + random.randint(0, gap) + gap * i
      self.drawText((x, random.randint(-5, 5)), RandomChar().GB2312(), self.randRGB())
      self.rotate()
    self.randLine(18)

  def save(self, path):
    self.image.save(path)

调用方法：

ic = ImageChar(fontColor=(100,211, 90))
ic.randChinese(4)
ic.save("1.jpeg")

效果图：

python 测试与应用:41357415
深圳IT招聘求职：105095215
武冈深圳高级群：66250781