python 实现 pdf表格转excel

#  pip install pdfplumber

import pdfplumber
from openpyxl import Workbook

file_name = '输入路径.pdf'
data_name = '输出路径.xlsx'

def analysis_table(pdf_file_path):
    # 打开表格
    workbook = Workbook()
    sheet = workbook.active

    # 打开pdf
    with pdfplumber.open(pdf_file_path) as pdf:
        # 遍历每页pdf 
        for page in pdf.pages:
            # 提取表格信息
            tables = page.extract_tables()
            
            # 遍历提取到的所有表格
            for table in tables:
                # 格式化表格数据
                for row in table:
                    sheet.append(row)
    
    workbook.save(filename=data_name)

analysis_table(file_name)