Python - 替换文本中的缩写

Question

我的数据框有数千行。
它看起来像这样：

import pandas as pd
import numpy as np
text = ['please send us a dm...','…could you please dm me','dm me plz…','i dmed u yesterday…','dm me asap thx', 'i send a dm to u now', 'thx u r so nice dming u now', 'just sent u a dm']
df = pd.DataFrame({"text": text})

          text
0   please send us a dm...
1   …could you please dm me
2   dm me plz…
3   i dmed u yesterday…
4   dm me asap thx
5   i send a dm to u now
6   thx u r so nice dming u now
7   just sent u a dm

我编写了一个函数来替换“文本”列中的缩写。

def convert(dataframe, column):
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" dm ", " direct message "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" dming ", " direct message "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" dmed ", " direct message "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" plz ", " please "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" thx ", " thanks "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" u ", " you "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace(" asap ", " as soon as possible "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace("...", " "))
    dataframe[column] = dataframe[column].apply(lambda x: x.replace("…", " "))   

但是，我的代码无法正常工作，因此无法完全替换数据框中的所有缩写。

convert(df, 'text')

          text
0   please send us a dm
1   could you please direct message me
2   dm me plz
3   i direct message you yesterday
4   dm me as soon as possible thx
5   i send a direct message to you now
6   thx you r so nice direct message you now
7   just sent you a dm

所需的最终输出如下所示：

          text
0   please send us a direct message
1   could you please direct message me
2   direct message me plz
3   i direct message you yesterday
4   direct message me as soon as possible thanks
5   i send a direct message to you now
6   thanks you r so nice direct message you now
7   just sent you a direct message

我不明白为什么我的代码不起作用。

Answer 1

首先构建一个替换字典：

replacers = {'dm': 'direct message',
 'thx': 'thanks',
 'dming': 'direct messaging',
 'dmed': 'direct messaged',
 'plz': 'please',
 'u': 'you',
 'asap': 'as soon as possible',
 '...': '',
 '. . .': '',
 'r': 'are'}

然后使用应用功能用所需的单词替换缩写。最后将单词合并回一个完整的字符串。

(
    df.text.str.replace('[...…]','')
    .str.split()
    .apply(lambda x: ' '.join([replacers.get(e, e) for e in x]))
)

0                    please send us a direct message
1                 could you please direct message me
2                           direct message me please
3                    i direct messaged you yesterday
4       direct message me as soon as possible thanks
5                 i send a direct message to you now
6    thanks you are so nice direct messaging you now
7                     just sent you a direct message
Name: text, dtype: object