读取Word并遍历

# 读取Word文档
# 导入python-docx库
import docx
import csv

# 打开Word文档
doc = docx.Document('小样本税号.docx')

# 读取文档中的段落
my_set = set()
for para in doc.paragraphs:
    # print(para.text)
    # 使用空格分割字符串
    words = para.text.split()

    # 遍历分割后的单词列表
    for word in words:
        # print(word)
        if len(word) == 10:
            my_set.add(word)
    # print("-----------------")

print("Set size:", len(my_set))
print("Set content:", my_set)

# 打开CSV文件并写入set集合
with open('set_data.csv', 'w', newline='') as csvfile:
    writer = csv.writer(csvfile)

    # 将set集合的元素逐行写入CSV文件
    for item in my_set:
        writer.writerow([item])