pip install poocr

import poocr

# 单张发票识别并导出 Excel
poocr.ocr2excel.VatInvoiceOCR2Excel(
    input_path='发票文件路径.pdf',
    output_path='输出目录',
    id='你的SecretId',
    key='你的SecretKey'
)

# 批量识别文件夹中的发票
poocr.ocr2excel.VatInvoiceOCR2Excel(
    input_path='发票文件夹路径',
    output_path='输出目录',
    id='你的SecretId',
    key='你的SecretKey'
)

import poocr
import os

class InvoiceOCR:
    def __init__(self, secret_id, secret_key):
        self.SecretId = secret_id
        self.SecretKey = secret_key
    
    def recognize_invoice(self, input_path, output_path):
        """识别发票并导出 Excel"""
        if not os.path.exists(input_path):
            raise FileNotFoundError(f"文件不存在: {input_path}")
        
        poocr.ocr2excel.VatInvoiceOCR2Excel(
            input_path=input_path,
            output_path=output_path,
            id=self.SecretId,
            key=self.SecretKey
        )
        
        print(f"发票识别完成，结果已保存到: {output_path}")

# 使用示例
if __name__ == "__main__":
    ocr = InvoiceOCR(
        secret_id="你的SecretId",
        secret_key="你的SecretKey"
    )
    
    ocr.recognize_invoice(
        input_path="../test_files/VatInvoiceOCR",
        output_path="../test_files/VatInvoiceOCR"
    )