einvoice-cli 1.0.0

package/README.md

@@ -0,0 +1,139 @@
+# einvoice-cli
+
+电子发票识别 Node.js CLI 工具
+
+支持 PDF 和 OFD 两种电子发票格式识别，输出 JSON 格式的发票信息。
+
+## 📦 安装
+
+```bash
+npm install
+```
+
+## 🛠️ 使用
+
+### 命令行方式
+
+```bash
+node bin/cli.js <pdf_or_ofd_file>
+
+# 或安装全局后
+npm install -g .
+einvoice /path/to/invoice.pdf
+```
+
+### 输出示例
+
+```json
+{
+  "title": "电子发票（增值税专用发票）",
+  "machineNumber": null,
+  "code": "35070124",
+  "number": "0000600419",
+  "date": "2024-12-26",
+  "checksum": "EAJfXh",
+  "buyerName": "福州猿力信息科技有限公司",
+  "buyerCode": null,
+  "buyerAddress": null,
+  "buyerAccount": null,
+  "sellerName": "福建省软件行业协会",
+  "sellerCode": null,
+  "sellerAddress": null,
+  "sellerAccount": null,
+  "amount": "1000.00",
+  "taxAmount": "0",
+  "totalAmount": "1000.00",
+  "totalAmountString": "壹仟元整",
+  "payee": null,
+  "reviewer": null,
+  "drawer": "陈榕",
+  "password": null,
+  "type": "financial",
+  "details": [
+    {
+      "name": "单位会员费",
+      "model": null,
+      "unit": "元",
+      "count": "1",
+      "price": "1000.00",
+      "amount": "1000.00",
+      "taxRate": 0,
+      "taxAmount": "0"
+    }
+  ]
+}
+```
+
+## 📋 支持的发票类型
+
+### 1. 增值税专用发票
+- **识别特征**: 包含"电子发票（增值税专用发票）"字样
+- **关键字段**: `buyerCode`(购方纳税人识别号), `sellerCode`(销方纳税人识别号), `amount`(金额), `taxAmount`(税额), `totalAmount`(价税合计)
+- **示例**: `test4.pdf`
+
+### 2. 普通电子发票
+- **识别特征**: 包含"电子发票（普通发票）"字样
+- **关键字段**: 与增值税专用发票类似，税率通常为1%、3%、6%、9%、13%等
+- **特殊格式**: 支持`*餐饮服务*餐费`等包含星号的项目名称
+- **示例**: `test6.pdf`
+
+### 3. 福建省财政票据
+- **识别特征**: 包含"福建省社会团体会员费统一收据"字样
+- **关键字段**: `code`(票据代码), `number`(票据号码), `checksum`(校验码), `amount`(金额), `totalAmountString`(大写金额)
+- **特点**: 无税额(`taxAmount`为0)，明细格式特殊
+- **示例**: `test8.pdf`
+
+### 4. 其他发票类型
+- **通行费发票**: 包含"通行费"和"车牌号"字样
+- **OFD格式发票**: 支持OFD文件格式的电子发票
+
+## ✨ 特性
+
+- ✅ 多类型发票识别（增值税专用发票、普通发票、财政票据等）
+- ✅ PDF 和 OFD 双格式支持
+- ✅ JSON 格式输出，便于程序处理
+- ✅ 字段验证和自动矫正
+- ✅ 坐标定位精确提取
+- ✅ 多策略降级确保高成功率
+- ✅ 无界面，轻量级依赖
+
+## 📚 API 使用
+
+```javascript
+const { extractPdf, extractOfd } = require('./lib/extractor');
+
+// 提取 PDF 发票
+const invoice = await extractPdf('./invoice.pdf');
+console.log(JSON.stringify(invoice, null, 2));
+
+// 提取 OFD 发票
+const invoice = await extractOfd('./invoice.ofd');
+console.log(JSON.stringify(invoice, null, 2));
+```
+
+## 🏗️ 项目结构
+
+```
+einvoice-nodejs/
+├── bin/
+│   └── cli.js              # 命令行入口
+├── lib/
+│   ├── extractor.js        # 主提取器
+│   ├── Invoice.js          # 数据模型
+│   ├── RegexPatterns.js    # 正则优化库
+│   ├── PDFTextPositionAnalyzer.js # 坐标定位引擎
+│   ├── InvoiceValidator.js # 验证框架
+│   ├── PdfRegularInvoiceService.js # 普通发票服务
+│   ├── PdfFullElectronicInvoiceService.js # 全电发票服务
+│   ├── PdfFinancialInvoiceService.js # 财政票据服务
+│   ├── PdfInvoiceExtractor.js # PDF 加载器
+│   ├── OfdInvoiceExtractor.js # OFD 提取器
+│   └── StringUtils.js      # 字符串工具
+├── package.json
+├── README.md               # 项目文档
+└── .gitignore
+```
+
+## 📄 许可证
+
+MIT License

package/bin/cli.js

@@ -0,0 +1,52 @@
+#!/usr/bin/env node
+
+const { program } = require('commander');
+const fs = require('fs');
+const path = require('path');
+const { extract } = require('../lib/extractor');
+
+const packageJson = require('../package.json');
+
+program.version(packageJson.version).description(packageJson.description);
+
+program
+  .argument('<file>', 'PDF or OFD invoice file path')
+  .action(async (filePath) => {
+    try {
+      // 检查文件是否存在
+      if (!fs.existsSync(filePath)) {
+        console.error(`Error: File not found: ${filePath}`);
+        process.exit(1);
+      }
+
+      // 检查文件格式
+      const ext = path.extname(filePath).toLowerCase();
+      if (!['.pdf', '.ofd'].includes(ext)) {
+        console.error('Error: Unsupported file format. Only PDF and OFD are supported.');
+        process.exit(1);
+      }
+
+
+      // 提取发票信息
+      const invoice = await extract(filePath);
+
+      // 输出 JSON
+      const jsonOutput = invoice.toJSON();
+      // 如果发票有错误信息，也输出
+      if (invoice.extractionError) {
+        jsonOutput.extractionError = invoice.extractionError;
+      }
+      if (invoice.error) {
+        jsonOutput.error = invoice.error;
+      }
+      if (invoice.errorStack) {
+        jsonOutput.errorStack = invoice.errorStack;
+      }
+      console.log(JSON.stringify(jsonOutput, null, 2));
+    } catch (error) {
+      console.error('Error:', error.message);
+      process.exit(1);
+    }
+  });
+
+program.parse();

package/index.js

@@ -0,0 +1,23 @@
+/**
+ * index.js - 主入口模块
+ * 可以直接在 Node.js 中导入使用
+ */
+
+const { extract, extractPdf, extractOfd } = require('./lib/extractor');
+const { Invoice, Detail } = require('./lib/Invoice');
+
+module.exports = {
+  // 主要 API
+  extract,           // 自动识别文件格式
+  extractPdf,        // 专门处理 PDF
+  extractOfd,        // 专门处理 OFD
+  
+  // 数据模型（如果需要扩展）
+  Invoice,
+  Detail,
+};
+
+// 使用示例
+// const einvoice = require('./index');
+// const invoice = await einvoice.extract('./invoice.pdf');
+// console.log(invoice.toJSON());

package/lib/BaseInvoiceService.js

@@ -0,0 +1,158 @@
+const { Invoice, Detail } = require('./Invoice');
+const RegexPatterns = require('./RegexPatterns');
+const InvoiceValidator = require('./InvoiceValidator');
+const ErrorHandler = require('./ErrorHandler');
+
+/**
+ * 发票服务基类
+ * 提取三个发票服务类的公共方法和逻辑
+ */
+class BaseInvoiceService {
+  /**
+   * 提取基础字段 - 通用实现
+   * @param {Invoice} invoice - 发票对象
+   * @param {string} allText - 规范化后的文本
+   * @param {Object} options - 配置选项
+   */
+  static extractBasicFields(invoice, allText, options = {}) {
+    const patterns = options.patterns || RegexPatterns.BASIC_FIELDS;
+
+    for (const [key, pattern] of Object.entries(patterns)) {
+      const result = RegexPatterns.tryPatterns(allText, [pattern]);
+      if (result) {
+        invoice[key] = result.match[1] || result.match[0];
+      }
+    }
+  }
+
+  /**
+   * 提取人员信息（签单人、收款人、复核人）
+   * @param {Invoice} invoice - 发票对象
+   * @param {string} allText - 规范化后的文本
+   */
+  static extractPersonInfo(invoice, allText) {
+    // 开票人 - 匹配直到遇到关键词或标点符号
+    const drawerMatch = allText.match(/开票人[:：]\s*([^销售方复核收款人,，。;；\n\r]+?)(?=[销售方复核收款人,，。;；\n\r]|$)/);
+    if (drawerMatch) {
+      invoice.drawer = drawerMatch[1].trim();
+    }
+
+    // 收款人/收款人
+    const collectorMatch = allText.match(/收款人[:：]\s*([^开票人复核销售方,，。;；\n\r]+?)(?=[开票人复核销售方,，。;；\n\r]|$)/);
+    if (collectorMatch) {
+      invoice.collector = collectorMatch[1].trim();
+      // 同时设置 payee 字段
+      invoice.payee = collectorMatch[1].trim();
+    }
+
+    // 复核/复核人
+    const reviewerMatch = allText.match(/复核人?[:：]\s*([^开票人收款人销售方,，。;；\n\r]+?)(?=[开票人收款人销售方,，。;；\n\r]|$)/);
+    if (reviewerMatch) {
+      invoice.reviewer = reviewerMatch[1].trim();
+    }
+  }
+
+  /**
+   * 验证发票数据
+   * @param {Invoice} invoice - 发票对象
+   */
+  static validateInvoice(invoice) {
+    const validation = InvoiceValidator.validate(invoice);
+    invoice.validationResult = {
+      valid: validation.valid,
+      warnings: validation.warnings,
+      errors: validation.errors,
+      suggestions: validation.suggestions
+    };
+
+    // 修正常见错误
+    if (validation.suggestions && validation.suggestions.length > 0) {
+      InvoiceValidator.correctCommonErrors(invoice);
+    }
+  }
+
+  /**
+   * 安全执行提取操作
+   * @param {Function} extractFn - 提取函数
+   * @param {Array} args - 函数参数
+   * @returns {Invoice} 发票对象
+   */
+  static safeExtract(extractFn, ...args) {
+    return ErrorHandler.safeExtract(extractFn, args, 'invoice-service');
+  }
+
+  /**
+   * 清理文本中的控制字符和特殊字符
+   * @param {string} text - 原始文本
+   * @returns {string} 清理后的文本
+   */
+  static cleanText(text) {
+    if (!text) return '';
+
+    // 移除控制字符
+    let cleaned = text.replace(/[\x00-\x1F\x7F]/g, '');
+
+    // 移除多余的空格和换行
+    cleaned = cleaned.replace(/\s+/g, ' ').trim();
+
+    return cleaned;
+  }
+
+  /**
+   * 提取日期信息
+   * @param {string} text - 文本
+   * @returns {string|null} 日期字符串
+   */
+  static extractDate(text) {
+    const datePatterns = [
+      RegexPatterns.BASIC_FIELDS.date,
+      RegexPatterns.BASIC_FIELDS.dateDash,
+      RegexPatterns.BASIC_FIELDS.dateCompact,
+    ];
+
+    for (const pattern of datePatterns) {
+      const match = text.match(pattern);
+      if (match) {
+        return match[1];
+      }
+    }
+
+    return null;
+  }
+
+  /**
+   * 提取纳税人识别号
+   * @param {string} text - 文本
+   * @returns {Array} 识别号数组
+   */
+  static extractTaxIds(text) {
+    const matches = RegexPatterns.getAllMatches(
+      text,
+      /纳税人识别号[:：\s]*([\dA-Z]{14,20})/g
+    );
+    return matches.map(match => match[1]);
+  }
+
+  /**
+   * 提取购销方名称
+   * @param {string} text - 文本
+   * @returns {Object} 包含buyerName和sellerName的对象
+   */
+  static extractPartyNames(text) {
+    const result = { buyerName: null, sellerName: null };
+
+    const buyerMatch = text.match(/购.*?名称[:：\s]*([^\n\r]+?)(?=\s*销|$)/);
+    if (buyerMatch) {
+      result.buyerName = buyerMatch[1].trim().replace(/\s+/g, ' ');
+    }
+
+    const sellerMatch = text.match(/销.*?名称[:：\s]*([^\n\r]+?)(?=\s*(?:买售方方信|统一社会信用代码|纳税人识别号|项目名称|$))/);
+    if (sellerMatch) {
+      result.sellerName = sellerMatch[1].trim().replace(/\s+/g, ' ');
+    }
+
+    return result;
+  }
+}
+
+module.exports = BaseInvoiceService;

package/lib/ErrorHandler.js

@@ -0,0 +1,98 @@
+const { Invoice } = require('./Invoice');
+
+/**
+ * 统一错误处理工具类
+ */
+class ErrorHandler {
+  /**
+   * 创建错误发票对象
+   * @param {Error|string} error - 错误对象或错误消息
+   * @param {string} source - 错误来源（如 'pdf', 'ofd', 'extractor'）
+   * @returns {Invoice} 包含错误信息的发票对象
+   */
+  static createErrorInvoice(error, source = 'unknown') {
+    const errorInvoice = new Invoice();
+    errorInvoice.title = 'error';
+    errorInvoice.type = 'error';
+    errorInvoice.extractionError = typeof error === 'string' ? error : error.message;
+    errorInvoice.errorSource = source;
+
+    // 保留堆栈信息用于调试
+    if (error instanceof Error && error.stack) {
+      errorInvoice.errorStack = error.stack;
+    }
+
+    // 设置验证结果为无效
+    errorInvoice.validationResult = {
+      valid: false,
+      warnings: [],
+      errors: [`提取失败: ${errorInvoice.extractionError}`],
+      suggestions: ['请检查文件格式是否正确']
+    };
+
+    return errorInvoice;
+  }
+
+  /**
+   * 安全执行提取操作
+   * @param {Function} extractFn - 提取函数
+   * @param {Array} args - 函数参数
+   * @param {string} source - 错误来源
+   * @returns {Invoice} 发票对象（成功或错误）
+   */
+  static safeExtract(extractFn, args = [], source = 'extractor') {
+    try {
+      return extractFn(...args);
+    } catch (error) {
+      console.error(`${source} 提取异常:`, error.message);
+      return this.createErrorInvoice(error, source);
+    }
+  }
+
+  /**
+   * 检查是否为错误发票
+   * @param {Invoice} invoice - 发票对象
+   * @returns {boolean} 是否为错误发票
+   */
+  static isErrorInvoice(invoice) {
+    return invoice && (
+      invoice.title === 'error' ||
+      invoice.type === 'error' ||
+      !!invoice.extractionError ||
+      !!invoice.error
+    );
+  }
+
+  /**
+   * 获取错误信息
+   * @param {Invoice} invoice - 发票对象
+   * @returns {string|null} 错误信息
+   */
+  static getErrorMessage(invoice) {
+    if (!this.isErrorInvoice(invoice)) {
+      return null;
+    }
+
+    return invoice.extractionError || invoice.error || '未知错误';
+  }
+
+  /**
+   * 统一格式化错误输出
+   * @param {Invoice} invoice - 发票对象
+   * @returns {Object} 格式化的错误信息
+   */
+  static formatError(invoice) {
+    if (!this.isErrorInvoice(invoice)) {
+      return null;
+    }
+
+    return {
+      success: false,
+      error: this.getErrorMessage(invoice),
+      source: invoice.errorSource || 'unknown',
+      timestamp: new Date().toISOString()
+    };
+  }
+}
+
+module.exports = ErrorHandler;

package/lib/Invoice.js

@@ -0,0 +1,108 @@
+/**
+ * 发票类
+ */
+class Invoice {
+  constructor() {
+    // 基础信息
+    this.title = null;
+    this.type = null;
+
+    // 发票编号相关
+    this.machineNumber = null;
+    this.code = null;
+    this.number = null;
+    this.date = null;
+    this.checksum = null;
+
+    // 购买方信息
+    this.buyerName = null;
+    this.buyerCode = null;
+    this.buyerAddress = null;
+    this.buyerAccount = null;
+
+    // 销售方信息
+    this.sellerName = null;
+    this.sellerCode = null;
+    this.sellerAddress = null;
+    this.sellerAccount = null;
+
+    // 金额相关
+    this.amount = null;
+    this.taxAmount = null;
+    this.totalAmount = null;
+    this.totalAmountString = null;
+
+    // 签章信息
+    this.payee = null;
+    this.reviewer = null;
+    this.drawer = null;
+    this.password = null;
+
+    // 明细
+    this.details = [];
+    
+    // 错误信息
+    this.error = null;
+  }
+
+  toJSON() {
+    return {
+      title: this.title,
+      machineNumber: this.machineNumber,
+      code: this.code,
+      number: this.number,
+      date: this.date,
+      checksum: this.checksum,
+      buyerName: this.buyerName,
+      buyerCode: this.buyerCode,
+      buyerAddress: this.buyerAddress,
+      buyerAccount: this.buyerAccount,
+      sellerName: this.sellerName,
+      sellerCode: this.sellerCode,
+      sellerAddress: this.sellerAddress,
+      sellerAccount: this.sellerAccount,
+      amount: this.amount,
+      taxAmount: this.taxAmount,
+      totalAmount: this.totalAmount,
+      totalAmountString: this.totalAmountString,
+      payee: this.payee,
+      reviewer: this.reviewer,
+      drawer: this.drawer,
+      password: this.password,
+      type: this.type,
+      details: this.details,
+      ...(this.error && { error: this.error }),
+    };
+  }
+}
+
+/**
+ * 发票明细类
+ */
+class Detail {
+  constructor() {
+    this.name = '';
+    this.model = null;
+    this.unit = null;
+    this.count = null;
+    this.price = null;
+    this.amount = null;
+    this.taxRate = null;
+    this.taxAmount = null;
+  }
+
+  toJSON() {
+    return {
+      name: this.name,
+      model: this.model,
+      unit: this.unit,
+      count: this.count,
+      price: this.price,
+      amount: this.amount,
+      taxRate: this.taxRate,
+      taxAmount: this.taxAmount,
+    };
+  }
+}
+
+module.exports = { Invoice, Detail };