如何从Pandas DataFrame文本列提取目标词左右各3个词？

问题根源 #

你写的正则表达式直接把df['word']作为字符串字面量塞进了正则里，而不是动态替换成每行对应的目标词，这是导致全NaN的核心原因。另外原正则也没处理目标词多次出现、位于文本首尾或不存在的情况。

解决思路 #

用apply逐行处理文本，因为每行的目标词不同，需要针对每行单独计算保留范围：

• 把文本拆成单词列表，定位所有目标词的位置
• 计算需要保留的单词区间：从最左目标词的前3个词（不能小于0），到最右目标词的后3个词（不能超过单词列表长度）
• 目标词不存在时返回空字符串

代码实现 #

import pandas as pd

# 构造示例DataFrame
data = {
    'id': [1,2,3,4],
    'text': [
        'i am working with john he is my colleague',
        'i watched the bond movie and the bond actor was amazing',
        'mary is my friend and we work together',
        'hello world hello python'
    ],
    'word': ['john', 'bond', 'mary', 'peter']
}
df = pd.DataFrame(data)

def get_retain_text(text, target_word):
    words = text.split()
    # 找到所有目标词的索引
    target_indices = [i for i, word in enumerate(words) if word == target_word]
    if not target_indices:
        return ''
    # 计算保留的起始和结束索引
    start_idx = max(0, min(target_indices) - 3)
    end_idx = min(len(words), max(target_indices) + 3 + 1)  # +1因为切片是左闭右开
    return ' '.join(words[start_idx:end_idx])

# 新增retain列
df['retain'] = df.apply(lambda row: get_retain_text(row['text'], row['word']), axis=1)

print(df)

输出结果 #

id                                                text    word                                              retain
0   1                i am working with john he is my colleague    john                  am working with john he is my
1   2  i watched the bond movie and the bond actor was amazing    bond  i watched the bond movie and the bond actor was amazing
2   3                   mary is my friend and we work together    mary                              mary is my friend
3   4                             hello world hello python   peter                                                     

内容的提问来源于stack exchange，提问作者MG Fern