正則表達式後退問題

Question

我在其他帖子中看到，在做(?<!X|Y|Z)時這是解決非固定寬度模式問題的正確方法，但不適用於我。

我嘗試以下

re.search(r'\b(?:(?<!Yummy)|(?<!Xoo))\bfoo\b', "Yummy foo", flags=re.UNICODE) is not None => True 
re.search(r'\b(?:(?<!Yummy)|(?<!Xoo))\bfoo\b', "Xoo foo", flags=re.UNICODE) is not None => True 
re.search(r'\b(?:(?<!Yummy)|(?<!Xoo))\bfoo\b', "other Yummy foo someone", flags=re.UNICODE) is not None => True 

它總是返回true，當它應該返回false。現在，如果我刪除或|，它工作正常。

re.search(r'\b(?:(?<!Yummy))\bfoo\b', "Yummy foo", flags=re.UNICODE) is not None => False 
re.search(r'\b(?<!Yummy)\bfoo\b', "Yummy foo", flags=re.UNICODE) is not None => False 
re.search(r'\b(?<!Yummy)\bfoo\b', " Yummy foo", flags=re.UNICODE) is not None => False 
re.search(r'\b(?<!Yummy)\bfoo\b', "other Yummy foo someone", flags=re.UNICODE) is not None => False 
re.search(r'\b(?<!Yummy)\bfoo\b', "foo someone", flags=re.UNICODE) is not None => True 
re.search(r'\b(?<!Yummy)\bfoo\b', "foo", flags=re.UNICODE) is not None => True 

有什麼建議嗎？

Answer 1

你不需要因爲lookbehinds都是零個寬度斷言使用的交替，你可以這樣寫：

re.search(r'\b(?<!\bYum)(?<!\bXooop)\s+foo\b', "Yum foo", flags=re.UNICODE) 
      ^^  ^
      | |   | 
      +-+---------+---These three assertions are tested at the same 
          position (i.e. immediatly before the \s+ match. 
          You can put them in any order you want, they are 
          only checks and don't eat characters.