Python：用单词列表替换句子中的一个单词并将新句子放在熊猫的另一列中

Question

我有一个数据框，其中一些句子包含单词 'o'clock'，我想用我拥有的小时列表替换之前提到的时间，并将新句子放在另一列中，如下所示：

data= {"sentences":["I have a class at ten o'clock", "she is my friend", "she goes to school at eight o'clock"]}
my_list=['two', 'three','five','ten']

我希望看到的是一个额外的列，新句子如下所示，其中时间更改为列表中的所有时间：

输出：

     sentences                            new_sentences
0    I have a class at ten o'clock        I have a class at two o'clock, I have a class at three o'clock,...
1    she is my friend                     she is my friend
2    she goes to school at eight o'clock  she goes to school at two o'clock,....

new_sentences 列中的重复是可以的。我曾尝试使用 np.where:

np.where(data.str.contains('o\'clock', regex=False, case=False, na=False), data["sentence"].replace()... )

但我不知道如何替换'o'clock之前的单词

提前谢谢你

Answer 1

用途：

# STEP 1
df1 = data['sentences'].str.extract(
    r"(?i)(?P<before>.*)\s(?P<clock>\w+(?=\so'clock))\s(?P<after>.*)")

# STEP 2
df1['clock'] = df1['clock'].str.replace(
    r'\w+', ','.join(my_list)).str.split(',')

# STEP 3
data['new_sentences'] = df1.dropna().explode('clock').agg(
    ' '.join, 1).groupby(level=0).agg(', '.join)

# STEP 4
data['new_sentences'] = data['new_sentences'].fillna(data['sentences'])

说明/步骤：

步骤 1：使用Series.str.extract 和给定的正则表达式模式创建一个三列数据帧，其中第一列对应于时钟 e.g. 10 之前的句子，中间列对应于时钟本身，右列对应于时钟后的句子。

# df1
                  before  clock    after
0      I have a class at    ten  o'clock
1                    NaN    NaN      NaN
2  she goes to school at  eight  o'clock

步骤 2：使用Series.str.replace 将时钟列中的标记替换为my_list 中的所有项目。然后使用Series.str.split 将替换的标记拆分为分隔符,。

# df1
                  before                    clock    after
0      I have a class at  [two, three, five, ten]  o'clock
1                    NaN                      NaN      NaN
2  she goes to school at  [two, three, five, ten]  o'clock

第 3 步：Dataframe.explode 围绕列 clock 展开数据框 df1，使用 .agg 沿轴 1 连接列。然后在级别 0 上使用 groupby 进一步聚合此 datframe。

# data
                             sentences                                      new_sentences
0        I have a class at ten o'clock  I have a class at two o'clock, I have a class ...
1                     she is my friend                                                NaN
2  she goes to school at eight o'clock  she goes to school at two o'clock, she goes to...

第 4 步：最后使用Series.fillna 从对应的sentences 列中填充new_sentences 列中的缺失值。

# data
                             sentences                                      new_sentences
0        I have a class at ten o'clock  I have a class at two o'clock, I have a class ...
1                     she is my friend                                   she is my friend
2  she goes to school at eight o'clock  she goes to school at two o'clock, she goes to...

Answer 2

这符合您的预期吗？

import re
data= {"sentences":["I have a class at ten o'clock", "she is my friend", "she goes to school at eight o'clock"]}
my_list=['two', 'three','five','ten']

regex = re.compile(r"(\w+) (?=o'clock)", re.IGNORECASE)
new = []

for i in data["sentences"]:
    for j in my_list:
        new.append(re.sub(regex, j + ' ', i))

new = list(set(new))

print(new)

输出：

I have a class at two o'clock
I have a class at ten o'clock
she goes to school at two o'clock
she goes to school at five o'clock
I have a class at five o'clock
I have a class at three o'clock
she goes to school at ten o'clock
she goes to school at three o'clock
she is my friend

或等价物：

import re
data= {"sentences":["I have a class at ten o'clock", "she is my friend", "she goes to school at eight o'clock"]}
my_list=['two', 'three','five','ten']
regex = re.compile(r"(\w+) (?=o'clock)", re.IGNORECASE)
x = list(set([re.sub(regex, j + ' ', i) for j in my_list for i in data["sentences"]]))