发票验真 + 识别，一套代码全搞定

大家好，我是正在实战各种 AI 项目的程序员晚枫。

🔍 识别 + 验真，缺一不可

只识别不验真的风险：

假发票混入
作废发票报销
红冲发票重复使用
税务风险

完整流程：

1	发票 → OCR 识别 → 提取信息 → 税务验真 → 结果存储

今天：教你一套代码搞定识别 + 验真。

🛠️ 完整代码

import poocr
import requests
import hashlib

# 配置
SECRET_ID = os.getenv('TENCENT_SECRET_ID')
SECRET_KEY = os.getenv('TENCENT_SECRET_KEY')
TAX_API_URL = "https://inv-veri.chinatax.gov.cn/api"

def recognize_invoice(image_path):
    """识别发票"""
    result = poocr.ocr2excel.VatInvoiceOCR2Excel(
        input_path=image_path,
        output_path='./temp',
        id=SECRET_ID,
        key=SECRET_KEY
    )
    return parse_result(result)

def verify_invoice(invoice_data):
    """验真发票"""
    # 准备验真参数
    params = {
        'invoice_code': invoice_data['发票代码'],
        'invoice_number': invoice_data['发票号码'],
        'invoice_date': invoice_data['开票日期'],
        'amount': invoice_data['金额']
    }
    
    # 调用税务验真 API
    response = requests.post(
        f"{TAX_API_URL}/verify",
        json=params,
        headers={'Authorization': get_auth_token()}
    )
    
    result = response.json()
    
    if result['status'] == 'valid':
        return True, "验真通过"
    elif result['status'] == 'invalid':
        return False, "发票不存在或信息不符"
    elif result['status'] == 'cancelled':
        return False, "发票已作废"
    else:
        return None, "验真失败"

def process_invoice_with_verify(image_path):
    """识别 + 验真完整流程"""
    print(f"📄 处理发票：{image_path}")
    
    # 1. OCR 识别
    print("🔍 正在识别...")
    invoice_data = recognize_invoice(image_path)
    print(f"✅ 识别完成：{invoice_data['发票号码']}")
    
    # 2. 税务验真
    print("🔐 正在验真...")
    is_valid, message = verify_invoice(invoice_data)
    
    if is_valid:
        print(f"✅ 验真通过：{message}")
        save_to_database(invoice_data, status='valid')
    elif is_valid is False:
        print(f"❌ 验真失败：{message}")
        save_to_database(invoice_data, status='invalid')
    else:
        print(f"⚠️ 验真异常：{message}")
        save_to_database(invoice_data, status='pending')
    
    return invoice_data, is_valid, message

# 批量处理
for invoice_file in os.listdir('./invoices'):
    process_invoice_with_verify(f'./invoices/{invoice_file}')

📊 验真结果处理

验真状态	含义	处理方式
valid	真票	正常入账
invalid	假票	退回，不报销
cancelled	已作废	退回，换发票
red_flushed	已红冲	退回，不报销
pending	验真失败	人工复核