嵌套字典和正则表达式，用于将剧本分类为对话行

Question

我正在使用 Python 3.6.3 获取 TV/Play/whatever 脚本并将其分类到字典中，该字典将字符及其对话行配对。

我已经能够得到我想要的结果，它为每个 Character: 嵌套了 {Line#:Line} 对，但我想知道是否有更有效的方法来达到这一点。特别是我最初拆分文本的方式，我首先获取对话中各个单词的列表，然后通过遍历字典的副本来加入这些列表。

import re

text = """
Steve: Is that his chart? 
Phil: Yes. 
Steve: Mm-hmm. I'll see him in a few moments. 
Phil: All right. Thank you, Doctor. 
P.A.: Dr. Braun, Dr. Miller, and Dr. Sullivan, emergency. 
Steve: How is she, Jessie? 
Jessie: Still fighting everybody and everything. She wants to live in the 
dark and never see her face again. That's about what she was doing when I went 
in. She had the blinds all drawn, towel over the mirror. """
## general hospital!    

dialog = {}
count = 0
cast = []
for word in text.split():
    if re.match(".*\:", word):
        character = word[:-1]
        count += 1      
        if character not in dialog:
            cast.append(character)
            dialog[character] = {}
            dialog[character][count] = []
        else:
            dialog[character][count] = []
    else:
        dialog[character][count].append(word)


fullLines = {}
for k,v in dialog.items(): 
    fullLines[k] = {}
    for k1,v1 in v.items():
        v1 = ' '.join(v1)
        fullLines[k][k1] = v1

有没有一种方法可以使用正则表达式拆分文本以识别对话提示 - “字符：”并用它拆分文本？我试着放置 re.compile(r".*\:") 变成 split() 就像这样

match = re.compile(".*\:")
for word in text.split(match): 

并得到错误TypeError: must be str or None, not _sre.SRE_Pattern。所以我基本上明白为什么这不起作用。我还在学习python，所以还在积累方法和pythonic的习惯。

Answer 1

你可以使用下面的表达式

^                             # start of the line
(?P<actor>[A-Z][^:\n\r]+):\s* # a potential actor
(?P<text>[\s\S]+?)            # the following text
(?=^[A-Z]|\Z)                 # lookaheads for the text

结合嵌套的defaultdict 作为容器。剩下的唯一事情是通过计算换行符到该点来计算行号（由match 对象定义）。

查看regex101.com 上的表达式演示。

---

在Python 中，这可能是：

import re
from collections import defaultdict

string = """Steve: Is that his chart? 
Phil: Yes. 
Steve: Mm-hmm. I'll see him in a few moments. 
Phil: All right. Thank you, Doctor. 
P.A.: Dr. Braun, Dr. Miller, and Dr. Sullivan, emergency. 
Steve: How is she, Jessie? 
Jessie: Still fighting everybody and everything. She wants to live in the 
dark and never see her face again. That's about what she was doing when I went 
in. She had the blinds all drawn, towel over the mirror."""

rx = re.compile(r'''
    ^
    (?P<actor>[A-Z][^:\n\r]+):\s*
    (?P<text>[\s\S]+?)
    (?=^[A-Z]|\Z)
    ''', re.MULTILINE | re.VERBOSE)

# create the nested defaultdict
result = defaultdict(lambda : defaultdict(int))

for m in rx.finditer(string):
    start = m.start()
    line = string.count('\n', 0, start) + 1
    result[m.group('actor')][line] = m.group('text').strip()

print(result)

这会产生

defaultdict(<function <lambda> at 0x10ffe0e18>, {'Steve': defaultdict(<class 'int'>, {1: 'Is that his chart?', 3: "Mm-hmm. I'll see him in a few moments.", 6: 'How is she, Jessie?'}), 'Phil': defaultdict(<class 'int'>, {2: 'Yes.', 4: 'All right. Thank you, Doctor.'}), 'P.A.': defaultdict(<class 'int'>, {5: 'Dr. Braun, Dr. Miller, and Dr. Sullivan, emergency.'}), 'Jessie': defaultdict(<class 'int'>, {7: "Still fighting everybody and everything. She wants to live in the \ndark and never see her face again. That's about what she was doing when I went \nin. She had the blinds all drawn, towel over the mirror."})})

Answer 2

你可以使用正则表达式和itertools.groupby:

import re
import itertools
text = """
Steve: Is that his chart? 
Phil: Yes. 
Steve: Mm-hmm. I'll see him in a few moments. 
Phil: All right. Thank you, Doctor. 
P.A.: Dr. Braun, Dr. Miller, and Dr. Sullivan, emergency. 
Steve: How is she, Jessie? 
Jessie: Still fighting everybody and everything. She wants to live in the dark and never see her face again. That's about what she was doing when I went in. She had the blinds all drawn, towel over the mirror. """
parts = [[a, i, b] for i, [a, b] in enumerate([re.split(':\s*', i) for i in filter(None, text.split('\n'))], start = 1)]
final_parts = {a:dict([i[1:] for i in b]) for a, b in itertools.groupby(sorted(parts, key=lambda x:x[0]), key=lambda x:x[0])}

输出：

{'Steve': {1: 'Is that his chart? ', 3: "Mm-hmm. I'll see him in a few moments. ", 6: 'How is she, Jessie? '}, 'Jessie': {7: "Still fighting everybody and everything. She wants to live in the dark and never see her face again. That's about what she was doing when I went in. She had the blinds all drawn, towel over the mirror. "}, 'Phil': {2: 'Yes. ', 4: 'All right. Thank you, Doctor. '}, 'P.A.': {5: 'Dr. Braun, Dr. Miller, and Dr. Sullivan, emergency. '}}