# 运行这个库
import poocr
import pdfplumber
from PIL import Image

if __name__ == '__main__':
    # 1行代码，完成识别发票 + 导入Excel
    filename = r'C:\Users\xukai\PycharmProjects\pythonProject1\invoices\031001700211_86022771.pdf'
    filedir = r'C:\Users\xukai\PycharmProjects\pythonProject1\invoices'
    try:
        with pdfplumber.open(filename) as pdf:
            outjpgnam = r'output.jpg'
            page1 = pdf.pages[0]
            im = page1.to_image()
            im.save(outjpgnam)

            output = r'allinvoices.xlsx'
            config = r'poocr-config.toml'
            poocr.ocr2excel.VatInvoiceOCR2Excel(input_path=outjpgnam, output_excel=output, configPath=config)
    except Exception as e:
        print(e)
    finally:
        pass