GPU_GUARD_MONOREPO/packages/backend/skills/vehicle-invoice-ocr/scripts/index.cjs

#!/usr/bin/env node
'use strict';

const fs = require('fs');

function emit(stage, message, status = 'running', extra = {}) {
  process.stderr.write(JSON.stringify({
    type: 'process_event',
    stage,
    message,
    status,
    timestamp: new Date().toISOString(),
    ...extra,
  }) + '\n');
}

function readInput() {
  const raw = String(process.argv[2] || process.env.SKILL_INPUT || process.env.AIFLOW_SKILL_INPUT || fs.readFileSync(0, 'utf8')).trim();
  if (!raw) return {};
  return JSON.parse(raw);
}

function normalizeDate(value) {
  const text = String(value || '').trim();
  if (!text) return '';
  const compact = text.replace(/[.年/]/g, '-').replace(/月/g, '-').replace(/日/g, '').replace(/\s+/g, '');
  const match = compact.match(/^(\d{4})-?(\d{1,2})-?(\d{1,2})$/);
  if (!match) return text;
  const [, y, m, d] = match;
  return `${y}-${String(m).padStart(2, '0')}-${String(d).padStart(2, '0')}`;
}

function cleanAmount(value) {
  const text = String(value || '').trim();
  if (!text) return '';
  return text.replace(/[￥¥,\s]/g, '');
}

function cleanVin(value) {
  return String(value || '').trim().replace(/\s+/g, '').toUpperCase();
}

function parseJsonText(value) {
  if (!value) return null;
  if (typeof value === 'object') return value;
  const text = String(value).trim();
  if (!text) return null;
  try {
    return JSON.parse(text);
  } catch {
    const match = text.match(/\{[\s\S]*\}/);
    if (!match) return null;
    try {
      return JSON.parse(match[0]);
    } catch {
      return null;
    }
  }
}

function firstNonEmpty(...values) {
  for (const value of values) {
    if (value !== undefined && value !== null && value !== '') return value;
  }
  return '';
}

function readField(source, names) {
  if (!source || typeof source !== 'object') return '';
  for (const name of names) {
    const value = source[name];
    if (value === undefined || value === null || value === '') continue;
    if (typeof value === 'object') {
      return firstNonEmpty(value.words, value.word, value.value, value.text);
    }
    return String(value).trim();
  }
  return '';
}

function getCandidate(input) {
  return parseJsonText(input.modelResult)
    || parseJsonText(input.ocrResult)
    || parseJsonText(input.mockResult)
    || parseJsonText(input.result)
    || (input.fields ? { fields: input.fields } : null);
}

function normalizeVehicleInvoice(candidate) {
  const root = candidate && typeof candidate === 'object' ? candidate : {};
  const fields = root.fields && typeof root.fields === 'object' ? root.fields : root;
  const words = root.words_result && typeof root.words_result === 'object' ? root.words_result : fields;
  const normalizedFields = {
    invoiceCode: readField(words, ['invoiceCode', '发票代码']),
    invoiceNumber: readField(words, ['invoiceNumber', 'invoiceNo', '发票号码', '发票号']),
    issueDate: normalizeDate(readField(words, ['issueDate', '开票日期', '填发日期'])),
    buyerName: readField(words, ['buyerName', '购买方名称', '购货单位名称', '购买方', '买方名称']),
    buyerId: readField(words, ['buyerId', '购买方纳税人识别号', '购买方身份证号码/组织机构代码', '买方纳税人识别号']),
    buyerAddressPhone: readField(words, ['buyerAddressPhone', '购买方地址、电话', '购买方地址电话']),
    vehicleType: readField(words, ['vehicleType', '车辆类型']),
    brandModel: readField(words, ['brandModel', '厂牌型号', '品牌型号', '厂牌型号及配置']),
    origin: readField(words, ['origin', '产地']),
    certificateNo: readField(words, ['certificateNo', '合格证号']),
    importCertificateNo: readField(words, ['importCertificateNo', '进口证明书号']),
    inspectionNo: readField(words, ['inspectionNo', '商检单号']),
    engineNo: readField(words, ['engineNo', '发动机号码', '发动机号']),
    vin: cleanVin(readField(words, ['vin', 'VIN', '车辆识别代号/车架号码', '车辆识别代号', '车架号码', '车架号'])),
    taxpayerNo: readField(words, ['taxpayerNo', '纳税人识别号', '主管税务机关代码']),
    sellerName: readField(words, ['sellerName', '销货单位名称', '销售方名称', '销方名称']),
    sellerTaxpayerNo: readField(words, ['sellerTaxpayerNo', '销货单位纳税人识别号', '销售方纳税人识别号']),
    sellerAccount: readField(words, ['sellerAccount', '销货单位开户银行及账号', '销售方开户行及账号']),
    sellerAddressPhone: readField(words, ['sellerAddressPhone', '销货单位地址、电话', '销售方地址电话']),
    taxAuthorityCode: readField(words, ['taxAuthorityCode', '主管税务机关代码']),
    taxAuthorityName: readField(words, ['taxAuthorityName', '主管税务机关名称']),
    totalAmount: cleanAmount(readField(words, ['totalAmount', '价税合计', '价税合计小写', '金额合计'])),
    taxRate: readField(words, ['taxRate', '增值税税率或征收率', '税率']),
    taxAmount: cleanAmount(readField(words, ['taxAmount', '增值税税额', '税额'])),
    totalAmountUpper: readField(words, ['totalAmountUpper', '价税合计大写', '大写']),
    tonnage: readField(words, ['tonnage', '吨位']),
    limitedPassengerCount: readField(words, ['limitedPassengerCount', '限乘人数']),
  };
  return {
    success: true,
    fields: normalizedFields,
    raw: root,
  };
}

function buildPrompt(input) {
  return [
    '请使用多模态大模型识别机动车销售统一发票图片或PDF，只输出合法JSON，不输出解释。',
    '只根据图片/PDF可见内容提取字段，无法确认的字段输出空字符串，不要猜测。',
    '日期统一 YYYY-MM-DD；金额只保留数字和小数点；VIN 保留大写字符。',
    '输出schema：{"fields":{"invoiceCode":"","invoiceNumber":"","issueDate":"","buyerName":"","buyerId":"","buyerAddressPhone":"","vehicleType":"","brandModel":"","origin":"","certificateNo":"","importCertificateNo":"","inspectionNo":"","engineNo":"","vin":"","taxpayerNo":"","sellerName":"","sellerTaxpayerNo":"","sellerAccount":"","sellerAddressPhone":"","taxAuthorityCode":"","taxAuthorityName":"","totalAmount":"","taxRate":"","taxAmount":"","totalAmountUpper":"","tonnage":"","limitedPassengerCount":""}}',
    input.imageUrl ? `图片URL：${input.imageUrl}` : '',
    input.pdfUrl ? `PDF URL：${input.pdfUrl}` : '',
    input.imageBase64 ? '已提供 imageBase64，请直接看图识别。' : '',
    input.pdfBase64 ? '已提供 pdfBase64，请直接识别PDF。' : '',
  ].filter(Boolean).join('\n');
}

async function run(input) {
  const candidate = getCandidate(input);
  if (candidate) {
    emit('normalize', '归一化多模态机动车发票识别结果', 'completed');
    return normalizeVehicleInvoice(candidate);
  }

  return {
    success: false,
    needsModelVision: true,
    error: '未提供多模态模型识别结果。请先让多模态模型查看 imageUrl/imageBase64/pdfUrl/pdfBase64，并把模型返回JSON作为 modelResult 传入本skill。',
    prompt: buildPrompt(input),
    fields: {
      invoiceCode: '',
      invoiceNumber: '',
      issueDate: '',
      buyerName: '',
      buyerId: '',
      vehicleType: '',
      brandModel: '',
      engineNo: '',
      vin: '',
      totalAmount: '',
      sellerName: '',
    },
  };
}

(async () => {
  try {
    const input = readInput();
    process.stdout.write(JSON.stringify(await run(input)));
  } catch (err) {
    process.stdout.write(JSON.stringify({
      success: false,
      error: err && err.message ? err.message : String(err),
    }));
  }
})();