正则表达式使用python（或nltk）提取两个特定单词之间的内容

Question

I'm trying to build a class and take each poem as an object, which has attributes of the title (followed by "POEM:"), author and content. I extracted title and author and put in a list. However, I don't know how to extract the content, and put into a list.

I have a txt file which includes many poems. Sample poems are:

POEM: lala AUTHOR: la
aaaaaaaaaaaaaa,
aaaaaaaaa,
akaaaaaaaa

POEM: alal AUTHOR: al
llllllllllll,
llllll.

llllllll,
lllllllllll

POEM: lal AUTHOR:as
sssssssss,
sssssss,
sssssss

This is what I did

import re
f=open('Poems.txt', 'r')
data=f.read().replace('\n','')
re.findall(r"^POEM:.*?(?=POEM)",data)

I want to get all the poems as separate strings in a list, but I can only get the first poem.

'POEM: lala AUTHOR: la, aaaaaaaaaaaaaa, aaaaaaaaa, akaaaaaaaa'

Answer 1

Much easier solution without using regular expressions, explained.

Explanation line by line

first you open the file

f=open('Poems.txt', 'r').read()

you will get your poems list with the expected output that you show in the last part of your question

poems_list = ["POEM" + s for s in f.split("POEM")]

we delete the first element because it is empty, due to the split function

poems_list.pop(0)

Up to here, poems_list would give us what the other user is posting in his question. But if you actually want to parse the data, which I guess it was your intention by using regex, you can just go ahead and to the following:

We go over each poem in the poem list to analyse the data that they contain

for poem in poems_list:

first we split it with the poem keyword, remember that you must leave a space between the semicolon and the poem name, or it wont work (without modifying the code)

    i1 = poem.split('POEM: ')

now we split it by the author, again leaving the trailing spaces as appropriate. We take i1 second element because the first one was the poem name, the rest of the contend is now stored in the second element of the list.

    i2 = i1[1].split(' AUTHOR: ')

again we will take the second element in the list to get the remaining part of the text. We split it by the new line because the poem begins after line break after stating its author

    i3 = i2[1].split('\n')

we save the values that we have obtained

    poem_name = i2[0]
    poem_author = i3[0]
    poem_content = i3[1]

And now its your turn to process the data how you wish. I recommend you to store it in a dictionary.

The full code

All the code without explanation (for copy-paste):

f=open('Poems.txt', 'r').read()
poems_list = ["POEM" + s for s in f.split("POEM")]
poems_list.pop(0)

for poem in poems_list:

    i1 = poem.split('POEM: ')
    i2 = i1[1].split(' AUTHOR: ')
    i3 = i2[1].split('\n')

    poem_name = i2[0]
    poem_author = i3[0]
    poem_content = i3[1]

Further thoughts

I do not recommend you to store your data like that in that file. It is very inefficient, and tiny modifications would cause great problems in the functioning of the code, which would require great modifications. Using databases, pandas, csv format or even pickle to store dictionaries is much more recommended, or at least format it a little bit better.

Answer 2

Code

See regex in use here

\s*(?=POEM:)

Note : The regex above is simply catching whitespace and asserting the position matches (with positive lookahead). See explanation for more details.

Usage

See code in use here

The basics

import re

s = "Your string here"
r = r"\s*(?=POEM:)"

print re.split(r, s)

In practice (with your sample string)

import re

s = """POEM: lala AUTHOR: la
aaaaaaaaaaaaaa,
aaaaaaaaa,
akaaaaaaaa

POEM: alal AUTHOR: al
llllllllllll,
llllll.

llllllll,
lllllllllll

POEM: lal AUTHOR:as
sssssssss,
sssssss,
sssssss"""

r = r"\s*(?=POEM:)"

print re.split(r, s)

---

Results

[
    'POEM: lala AUTHOR: la\naaaaaaaaaaaaaa,\naaaaaaaaa,\nakaaaaaaaa',
    'POEM: alal AUTHOR: al\nllllllllllll,\nllllll.\n\nllllllll,\nlllllllllll',
    'POEM: lal AUTHOR:as\nsssssssss,\nsssssss,\nsssssss'
]

---

Explanation

• \\s* Match any number of whitespace characters
• (?=POEM:) Positive lookahead ensuring what follows matches POEM: literally