我正在尝试从文本中提取包含指定单词的所有句子。
txt="I like to eat apple. Me too. Let's go buy some apples."
txt = "." + txt
re.findall(r"\."+".+"+"apple"+".+"+"\.", txt)
但它正在返回我：
[".I like to eat apple. Me too. Let's go buy some apples."]
而不是：
[".I like to eat apple., "Let's go buy some apples."]
有什么帮助吗？
【问题讨论】：
标签： python regex text-segmentation
你可以使用str.split，
>>> txt="I like to eat apple. Me too. Let's go buy some apples."
>>> txt.split('. ')
['I like to eat apple', 'Me too', "Let's go buy some apples."]
>>> [ t for t in txt.split('. ') if 'apple' in t]
['I like to eat apple', "Let's go buy some apples."]
【讨论】：
In [7]: import re
In [8]: txt=".I like to eat apple. Me too. Let's go buy some apples."
In [9]: re.findall(r'([^.]*apple[^.]*)', txt)
Out[9]: ['I like to eat apple', " Let's go buy some apples"]
但请注意@jamylak 的基于split 的解决方案更快：
In [10]: %timeit re.findall(r'([^.]*apple[^.]*)', txt)
1000000 loops, best of 3: 1.96 us per loop
In [11]: %timeit [s+ '.' for s in txt.split('.') if 'apple' in s]
1000000 loops, best of 3: 819 ns per loop
对于较大的字符串，速度差异较小，但仍然很重要：
In [24]: txt = txt*10000
In [25]: %timeit re.findall(r'([^.]*apple[^.]*)', txt)
100 loops, best of 3: 8.49 ms per loop
In [26]: %timeit [s+'.' for s in txt.split('.') if 'apple' in s]
100 loops, best of 3: 6.35 ms per loop
【讨论】：
不需要正则表达式：
>>> txt = "I like to eat apple. Me too. Let's go buy some apples."
>>> [sentence + '.' for sentence in txt.split('.') if 'apple' in sentence]
['I like to eat apple.', " Let's go buy some apples."]
【讨论】：
In [3]: re.findall(r"([^.]*?apple[^.]*\.)",txt)
Out[4]: ['I like to eat apple.', " Let's go buy some apples."]
【讨论】：
apple：['我喜欢吃苹果。']
r"\."+".+"+"apple"+".+"+"\."
这行有点奇怪；为什么要连接这么多单独的字符串？你可以只使用 r'..+apple.+.'.
无论如何，正则表达式的问题在于它的贪心。默认情况下，x+ 将尽可能频繁地匹配x。因此，您的 .+ 将匹配尽可能多的字符（any 个字符）；包括点和apples。
你想使用的是非贪婪表达式；您通常可以通过在末尾添加? 来做到这一点：.+?。
这将使您得到以下结果：
['.I like to eat apple. Me too.']
如您所见，您不再获得两个苹果句子，但仍然获得 Me too.。那是因为你在apple之后仍然匹配.，所以不可能不捕获下面的句子。
一个有效的正则表达式是这样的：r'\.[^.]*?apple[^.]*?\.'
在这里你看不到任何个字符，而只看到那些本身不是点的字符。我们还允许根本不匹配任何字符（因为在第一句中的apple 之后没有非点字符）。使用该表达式会导致：
['.I like to eat apple.', ". Let's go buy some apples."]
【讨论】：
显然，有问题的样本是extract sentence containing substring，而不是extract sentence containing word。如何通过python解决extract sentence containing word问题如下：
一个词可以在句子的开头|中间|结尾。不限于问题中的示例，我将提供一个在句子中搜索单词的通用功能：
def searchWordinSentence(word,sentence):
pattern = re.compile(' '+word+' |^'+word+' | '+word+' $')
if re.search(pattern,sentence):
return True
限于问题中的例子，我们可以这样解决：
txt="I like to eat apple. Me too. Let's go buy some apples."
word = "apple"
print [ t for t in txt.split('. ') if searchWordofSentence(word,t)]
对应的输出是：
['I like to eat apple']
【讨论】：
import nltk
search = "test"
text = "This is a test text! Best text ever. Cool"
contains = [s for s in nltk.sent_tokenize(text) if search in s]
【讨论】：