Python re.split() 将分隔符的一部分保留为第一个字符串的一部分，将其他部分保留为第二个字符串的一部分，等等

Question

我有一种情况，我想将一长段文本拆分成句子。我有一段工作代码可以根据需要拆分字符串，但是它会删除分隔符（我知道会这样）。现在，我希望能够将这些分隔符保留为输出字符串的一部分（适当地重新分配）。

我的例子是这样的：

import re

strings = ['UT Arlington 1st - Berthiaume reached on a fielding error by ss (0-0). O. Salinas fouled out to 1b (2-1 KBB). Q. Rohrbaugh flied out to cf (2-0 BB). B. Cox fouled out to lf (2-2 KBBKF)',
'Southeast Mo. State 1st - EZELL, T. lined out to 2b (2-2 FBBKFFF). HOLST, D. flied out to lf (0-2 FK). GAGAN, T. struck out swinging (1-2 BKKS).',
'UT Arlington 3rd - J. Minjarez hit by pitch (0-0); RJ Williams advanced to second. Berthiaume popped up to 1b (0-2 KF). O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.']

for s in strings:
        header = re.split(r'[ ][-][ ]', s)
        print(header[0])
        text = re.split(r'([a-z][.][ ][A-Z]|[)][.][ ][A-Z])', header[-1])
        print(text)

当前输出：

UT Arlington 1st
['Berthiaume reached on a fielding error by ss (0-0', '). O', '. Salinas fouled out to 1b (2-1 KBB', '). Q', '. Rohrbaugh flied out to cf (2-0 BB', '). B', '. Cox fouled out to lf (2-2 KBBKF)']
Southeast Mo. State 1st
['EZELL, T. lined out to 2b (2-2 FBBKFFF', '). H', 'OLST, D. flied out to lf (0-2 FK', '). G', 'AGAN, T. struck out swinging (1-2 BKKS).']
UT Arlington 3rd
['J. Minjarez hit by pitch (0-0); RJ Williams advanced to secon', 'd. B', 'erthiaume popped up to 1b (0-2 KF', '). O', '. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.']

我想要的输出：

UT Arlington 1st
['Berthiaume reached on a fielding error by ss (0-0)', 'O. Salinas fouled out to 1b (2-1 KBB)', 'Q. Rohrbaugh flied out to cf (2-0 BB)', 'B. Cox fouled out to lf (2-2 KBBKF)']
Southeast Mo. State 1st
['EZELL, T. lined out to 2b (2-2 FBBKFFF)', 'HOLST, D. flied out to lf (0-2 FK)', 'GAGAN, T. struck out swinging (1-2 BKKS).']
UT Arlington 3rd
['J. Minjarez hit by pitch (0-0); RJ Williams advanced to second', 'Berthiaume popped up to 1b (0-2 KF)', 'O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.']

Answer 1

您可能想看看nltk，而不是使用正则表达式：

from nltk import sent_tokenize

strings = ['UT Arlington 1st - Berthiaume reached on a fielding error by ss (0-0). O. Salinas fouled out to 1b (2-1 KBB). Q. Rohrbaugh flied out to cf (2-0 BB). B. Cox fouled out to lf (2-2 KBBKF)',
'Southeast Mo. State 1st - EZELL, T. lined out to 2b (2-2 FBBKFFF). HOLST, D. flied out to lf (0-2 FK). GAGAN, T. struck out swinging (1-2 BKKS).',
'UT Arlington 3rd - J. Minjarez hit by pitch (0-0); RJ Williams advanced to second. Berthiaume popped up to 1b (0-2 KF). O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.']

needle = " - "
for string in strings:
    pos = string.find(needle)
    header = string[:pos]
    text = string[pos + len(needle):]
    print(header)   
    print(sent_tokenize(text))

产量：

UT Arlington 1st
['Berthiaume reached on a fielding error by ss (0-0).', 'O. Salinas fouled out to 1b (2-1 KBB).', 'Q. Rohrbaugh flied out to cf (2-0 BB).', 'B. Cox fouled out to lf (2-2 KBBKF)']
Southeast Mo. State 1st
['EZELL, T. lined out to 2b (2-2 FBBKFFF).', 'HOLST, D. flied out to lf (0-2 FK).', 'GAGAN, T. struck out swinging (1-2 BKKS).']
UT Arlington 3rd
['J. Minjarez hit by pitch (0-0); RJ Williams advanced to second.', 'Berthiaume popped up to 1b (0-2 KF).', 'O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.']

通过字符串函数（.find()）提取标题，然后通过sent_tokenize()分析句子。

Answer 2

回答

简介

好的，这适用于您提出的所有用例，但绝不是完美的。句子中间的句号. 会引起并发症。这使得它变得复杂，因为它们不再是普通的句子终止符，而是代表其他东西，例如首字母。

代码

You can see this code in use here

\h*+(.{2,}?(?:\.|$))(?=(?:\h+[A-Z])|$)

---

结果

输入 1

J. Minjarez hit by pitch (0-0); RJ Williams advanced to second. Berthiaume popped up to 1b (0-2 KF). O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.

输出 1

J. Minjarez hit by pitch (0-0); RJ Williams advanced to second.

Berthiaume popped up to 1b (0-2 KF).

O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.

---

输入 2

EZELL, T. lined out to 2b (2-2 FBBKFFF). HOLST, D. flied out to lf (0-2 FK). GAGAN, T. struck out swinging (1-2 BKKS).

输出 2

EZELL, T. lined out to 2b (2-2 FBBKFFF).

HOLST, D. flied out to lf (0-2 FK).

GAGAN, T. 三振出局 (1-2 BKKS)。

---

输入 3

Berthiaume reached on a fielding error by ss (0-0). O. Salinas fouled out to 1b (2-1 KBB). Q. Rohrbaugh flied out to cf (2-0 BB). B. Cox fouled out to lf (2-2 KBBKF)

输出 3

Berthiaume reached on a fielding error by ss (0-0).

O. Salinas fouled out to 1b (2-1 KBB).

Q. Rohrbaugh flied out to cf (2-0 BB).

---

说明

正则表达式的工作方式如下：

• 在零和无限水平空白\h 字符之间尽可能多次匹配而不返回
• 捕获 2 到无限制的任意字符（换行符除外），但尽可能少，后跟 . 或字符串结尾 $
• 确保前一个后跟以下之一
  • 一个到无限的水平空白字符，后跟一个大写字母[A-Z]
  • 字符串结尾$

我使用 .{2,}? 的原因是指定我们要匹配至少 2 个字符（首字母在 . 之前只有 1 个字符，因此在例如 @987654339 的情况下这些将被忽略为句子@)。它使用惰性量词，以便在匹配下一个标记时停止（点\. [或字符串结尾$]）

---

编辑

由于python的re模块不支持所有格量​​词（并且根据regex101似乎也不支持\h作为水平空白字符，所以我稍微编辑了正则表达式，如下所示。

See this code in use here

\s*(\S.{1,}?(?:\.|$))(?=(?:\s+[A-Z])|$)

Answer 3

由于每个句子都以当前的球数和击球数结束，当句号在 i 后面有 ) 时，您可以拆分 - 或 .。此外，正则表达式检查句点前的最后一个字母是否为小写，后面的数据是一个空格，然后是一个大写字母（表示常规句子的结尾和新句子的开头）：

import re

strings = ['UT Arlington 1st - Berthiaume reached on a fielding error  by ss (0-0). O. Salinas fouled out to 1b (2-1 KBB). Q. Rohrbaugh flied out to cf (2-0 BB). B. Cox fouled out to lf (2-2 KBBKF)', 'Southeast Mo. State 1st - EZELL, T. lined out to 2b (2-2 FBBKFFF). HOLST, D. flied out to lf (0-2 FK). GAGAN, T. struck out swinging (1-2 BKKS).', 'UT Arlington 3rd - J. Minjarez hit by pitch (0-0); RJ Williams advanced to second. Berthiaume popped up to 1b (0-2 KF). O. Salinas flied out to cf to right center (2-1 KBB); RJ Williams advanced to third.']

new_data = [re.split("(?<!\d)-(?!\d)|(?<=\))\.|(?<=[a-z])\.(?=\s[A-Z])", i) for i in strings]

for plays in new_data:
    print new_data