找出列表中的三个单词在给定文档中组合出现的频率

import re
from collections import defaultdict

def find_word_combinations(document, word_list):
    # 分词
    words = re.findall(r'\w+', document.lower())
    
    # 统计组合频率
    combinations = defaultdict(int)
    for i in range(len(words) - 2):
        if all(word in word_list for word in words[i:i+3]):
            combination = ' '.join(words[i:i+3])
            combinations[combination] += 1
    
    # 找出频率最高的三个组合
    top_combinations = sorted(combinations.items(), key=lambda x: x[1], reverse=True)[:3]
    
    return top_combinations

# 示例文档和单词列表
document = "This is a sample document. The document contains some sample words."
word_list = ["sample", "document", "words", "contains"]

# 调用函数并输出结果
result = find_word_combinations(document, word_list)
for combination, frequency in result:
    print(f"组合 '{combination}' 出现的频率为 {frequency} 次。")