fork(3) download
copy 
  1.  
  2. import sre_parse
  3. import pprint
  4. import random
  5.  
  6.  
  7. class RandomRegexStringGenerator:
  8.     def __init__(self):
  9.         self._words = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_'
  10.         self._digits = '0123456789'
  11.         self._puncts = '!"#$%&()*+,-./:;<=>?@[\]^_`{|}~'
  12.         self._whitespaces = ' \t'  #  ' \t\n\r\v\f' 에서 조정
  13.         self._max_repeat = 5 # 무한 수량자가 나왔을때 최대 반복 리밋
  14.         self._result = ''
  15.  
  16.     def _gen_char(self, value):
  17.         return chr(value)
  18.  
  19.     def _gen_range(self, min_value, max_value):
  20.         # random 선택
  21.         idx = random.randint(min_value, max_value)
  22.         return chr(idx)
  23.  
  24.     def _gen_in_word(self):
  25.         return self._words[random.randint(0, len(self._words)-1)]
  26.  
  27.     def _gen_in_digit(self):
  28.         return self._digits[random.randint(0, len(self._digits)-1)]
  29.  
  30.     def _gen_in_whitespace(self):
  31.         return self._whitespaces[random.randint(0, len(self._whitespaces)-1)]
  32.  
  33.     def _gen_in_punct(self):
  34.         return self._puncts[random.randint(0, len(self._puncts)-1)]
  35.  
  36.     def _gen_class(self, cls_itm):
  37.         if cls_itm[0] == sre_parse.LITERAL:
  38.             self._result += self._gen_char(cls_itm[1])
  39.  
  40.         elif cls_itm[0] == sre_parse.RANGE:
  41.             self._result += self._gen_range(cls_itm[1][0], cls_itm[1][1])
  42.  
  43.         elif cls_itm[0] == sre_parse.CATEGORY:
  44.             if cls_itm[1] == sre_parse.CATEGORY_DIGIT:
  45.                 self._result += self._gen_in_digit()
  46.  
  47.             elif cls_itm[1] == sre_parse.CATEGORY_WORD:
  48.                 self._result += self._gen_in_word()
  49.  
  50.             elif cls_itm[1] == sre_parse.CATEGORY_SPACE:
  51.                 self._result += self._gen_in_whitespace()
  52.  
  53.             elif cls_itm[1] == sre_parse.CATEGORY_NOT_DIGIT:
  54.                 self._result += self._gen_in_punct()  # NOT-DIGIT => PUNCT
  55.  
  56.             elif cls_itm[1] == sre_parse.CATEGORY_NOT_WORD:
  57.                 self._result += self._gen_in_punct() # NOT-WORD => PUNCT
  58.  
  59.             elif cls_itm[1] == sre_parse.CATEGORY_NOT_SPACE:
  60.                 self._result += self._gen_in_word()  # NOT-SPACE => WORD
  61.  
  62.     def _gen_negate_class(self, body):
  63.         # 문자클래스에 속하지 않는 문자 랜덤 지정
  64.         # 완전 랜덤은 아니며 min, max  근방의 값으로 함
  65.         min_value = 9999999
  66.         max_value = 32
  67.         for cls_itm in body:
  68.             if cls_itm[0] == sre_parse.LITERAL:
  69.                 min_value = min(cls_itm[1], min_value)
  70.                 max_value = max(cls_itm[1], max_value)
  71.             elif cls_itm[0] == sre_parse.RANGE:
  72.                 min_value = min(cls_itm[1][0], min_value)
  73.                 max_value = max(cls_itm[1][1], max_value)
  74.  
  75.         if min_value-1 >= 32:
  76.             # random 선택
  77.             self._result += chr( random.randint(32, min_value-1) )
  78.         else:
  79.             if max_value + 1 <= 126:
  80.                 self._result += chr( random.randint(max_value + 1, 126) )
  81.             else:
  82.                 # ascii 범위 벗어남, 그냥 표정
  83.                 self._result += chr( random.randint(0x1F600, 0x1F609) )
  84.  
  85.  
  86.  
  87.  
  88.     def _proc_cmd(self, itm):
  89.         # 단문자
  90.         if itm[0] == sre_parse.LITERAL:
  91.             self._result += self._gen_char(itm[1])
  92.  
  93.         # 문자클래스
  94.         elif itm[0] == sre_parse.IN:
  95.             start_num = 0
  96.             body = itm[1]
  97.             if body[0][0] == sre_parse.NEGATE: # 제외조건 TODO:
  98.                 start_num = 1
  99.                 self._gen_negate_class(body[1:])
  100.             else:
  101.                 # random 선택
  102.                 idx = random.randint(start_num, len(body)-1)
  103.                 cls_itm = body[idx]
  104.                 self._gen_class(cls_itm)
  105.  
  106.         # 수량자
  107.         elif itm[0] in (sre_parse.MAX_REPEAT, sre_parse.MIN_REPEAT):
  108.             # 패턴 => XXX_REPEAT, (min, max, p)                 
  109.             min_repeat = itm[1][0] 
  110.             max_repeat = itm[1][1]
  111.             # set limit
  112.             if itm[1][1] == sre_parse.MAXREPEAT:
  113.                 max_repeat = self._max_repeat  # 지정한 max값으로 한정
  114.  
  115.             # random 선택
  116.             rand_repeat = random.randint(min_repeat, max_repeat)
  117.  
  118.             # random 만큼 반복
  119.             for i in range(rand_repeat):
  120.                 self._read_body(itm[1][2])
  121.  
  122.         # Union
  123.         elif itm[0] == sre_parse.BRANCH:
  124.             body = itm[1][1]
  125.             # random 선택
  126.             idx = random.randint(0, len(body)-1)
  127.             self._read_body(body[idx])
  128.  
  129.         else:            
  130.             if isinstance(itm[1], tuple):
  131.                 for titm in itm[1]:
  132.                     if isinstance(titm, sre_parse.SubPattern):
  133.                         #print('CMD: %s in body '%(itm[0]))
  134.                         self._read_body(titm)
  135.  
  136.     # [ (cmd), (cmd), ...] 이런 형태에 대해 cmd 마다 처리
  137.     def _read_body(self, body):
  138.         for itm in body:
  139.             self._proc_cmd(itm)
  140.  
  141.  
  142.     def generate(self, regex_pattern):
  143.  
  144.         p = sre_parse.parse( regex_pattern )
  145.         #pprint.pprint(p)
  146.  
  147.         self._result = ''
  148.         self._read_body( p )
  149.  
  150.         return self._result
  151.  
  152.  
  153. pats = [
  154.     r'01[069]-\d{3}\d?-\d{4}', # 전화번호
  155.     r'안녕하세요 제 이름은 [김이박최황안][가-힣][가-힣] (입니다|일까요\?)', # 유니코드
  156.     r'aa(bcd|\w{7}\d{3}|[a-z0-9]+)cc\W{2}\S{2}\D{2}', # union 수량자
  157.     r'(void|char|int|double) [A-Za-z_]\w{8} \((int|char) [A-Za-z_]\w{3}(, (int|char) [A-Za-z_]\w{3}){1,2}\);', # function style
  158.     r'[^ -~a-z]+', # ascii printable 범위 제외 문자클래스 => 그냥 유니코드 표정
  159. ]
  160.  
  161.  
  162.  
  163. g = RandomRegexStringGenerator()
  164.  
  165. for pat in pats:
  166.     print('pattern=>', pat)
  167.     for i in range(3):
  168.         print( g.generate( pat ) )
  169.     print('--------')
  170.  
  171.  
Success #stdin #stdout 0.04s 12200KB
comments (?)
stdin
copy 
Standard input is empty
stdout
copy 
pattern=> 01[069]-\d{3}\d?-\d{4}
016-8808-1054
010-6852-5285
019-7930-0908
--------
pattern=> 안녕하세요 제 이름은 [김이박최황안][가-힣][가-힣] (입니다|일까요\?)
안녕하세요 제 이름은 박픲냠 일까요?
안녕하세요 제 이름은 안놺샯 입니다
안녕하세요 제 이름은 박꼶뿽 입니다
--------
pattern=> aa(bcd|\w{7}\d{3}|[a-z0-9]+)cc\W{2}\S{2}\D{2}
aak56cc|>_4%{
aabcdcc)}Zp+>
aabcdcc-`Ak")
--------
pattern=> (void|char|int|double) [A-Za-z_]\w{8} \((int|char) [A-Za-z_]\w{3}(, (int|char) [A-Za-z_]\w{3}){1,2}\);
void qF_b45c0n (char WuLy, int TQsu);
double tHFl3jNvK (int _84Q, int _Maz, int rniw);
double iDiqhZ7rQ (char wtRo, int x13y);
--------
pattern=> [^ -~a-z]+
😅😉
😈
😂
--------
https://ideone.com/jjYKST
language:
Python 3 (python  3.12)
created:
4 years ago
visibility:
secret

Share or Embed source code

Discover > Sphere Engine API

The brand new service which powers Ideone!

Discover > IDE Widget

Widget for compiling and running the source code in a web browser!