einvoice/ts/formats/validation/schematron.downloader.ts

import * as plugins from '../../plugins.js';
import * as path from 'path';
import { promises as fs } from 'fs';

/**
 * Schematron rule sources
 */
export interface SchematronSource {
  name: string;
  version: string;
  url: string;
  description: string;
  format: 'UBL' | 'CII' | 'BOTH';
}

/**
 * Official Schematron sources for e-invoicing standards
 */
export const SCHEMATRON_SOURCES: Record<string, SchematronSource[]> = {
  EN16931: [
    {
      name: 'EN16931-UBL',
      version: '1.3.14',
      url: 'https://github.com/ConnectingEurope/eInvoicing-EN16931/raw/master/ubl/schematron/EN16931-UBL-validation.sch',
      description: 'Official EN16931 validation rules for UBL format',
      format: 'UBL'
    },
    {
      name: 'EN16931-CII',
      version: '1.3.14',
      url: 'https://github.com/ConnectingEurope/eInvoicing-EN16931/raw/master/cii/schematron/EN16931-CII-validation.sch',
      description: 'Official EN16931 validation rules for CII format',
      format: 'CII'
    },
    {
      name: 'EN16931-EDIFACT',
      version: '1.3.14',
      url: 'https://github.com/ConnectingEurope/eInvoicing-EN16931/raw/master/edifact/schematron/EN16931-EDIFACT-validation.sch',
      description: 'Official EN16931 validation rules for EDIFACT format',
      format: 'CII'
    }
  ],
  XRECHNUNG: [
    {
      name: 'XRechnung-UBL',
      version: '3.0.2',
      url: 'https://github.com/itplr-kosit/xrechnung-schematron/raw/master/src/schematron/ubl-invoice/XRechnung-UBL-3.0.sch',
      description: 'XRechnung CIUS validation for UBL',
      format: 'UBL'
    },
    {
      name: 'XRechnung-CII',
      version: '3.0.2',
      url: 'https://github.com/itplr-kosit/xrechnung-schematron/raw/master/src/schematron/cii/XRechnung-CII-3.0.sch',
      description: 'XRechnung CIUS validation for CII',
      format: 'CII'
    }
  ],
  PEPPOL: [
    {
      name: 'PEPPOL-EN16931-UBL',
      version: '3.0.17',
      url: 'https://github.com/OpenPEPPOL/peppol-bis-invoice-3/raw/master/rules/sch/PEPPOL-EN16931-UBL.sch',
      description: 'PEPPOL BIS Billing 3.0 validation rules',
      format: 'UBL'
    },
    {
      name: 'PEPPOL-T10',
      version: '3.0.17',
      url: 'https://github.com/OpenPEPPOL/peppol-bis-invoice-3/raw/master/rules/sch/UBL-T10.sch',
      description: 'PEPPOL Transaction 10 (Invoice) validation',
      format: 'UBL'
    },
    {
      name: 'PEPPOL-T14',
      version: '3.0.17',
      url: 'https://github.com/OpenPEPPOL/peppol-bis-invoice-3/raw/master/rules/sch/UBL-T14.sch',
      description: 'PEPPOL Transaction 14 (Credit Note) validation',
      format: 'UBL'
    }
  ]
};

/**
 * Schematron downloader and cache manager
 */
export class SchematronDownloader {
  private cacheDir: string;
  private smartfile: any;
  
  constructor(cacheDir: string = 'assets/schematron') {
    this.cacheDir = cacheDir;
  }
  
  /**
   * Initialize the downloader
   */
  public async initialize(): Promise<void> {
    // Ensure cache directory exists
    this.smartfile = await import('@push.rocks/smartfile');
    await fs.mkdir(this.cacheDir, { recursive: true });
  }
  
  /**
   * Download a Schematron file
   */
  public async download(source: SchematronSource): Promise<string> {
    const fileName = `${source.name}-v${source.version}.sch`;
    const filePath = path.join(this.cacheDir, fileName);
    
    // Check if already cached
    if (await this.isCached(filePath)) {
      console.log(`Using cached Schematron: ${fileName}`);
      return filePath;
    }
    
    console.log(`Downloading Schematron: ${source.name} v${source.version}`);
    
    try {
      // Download the file
      const response = await fetch(source.url);
      if (!response.ok) {
        throw new Error(`HTTP ${response.status}: ${response.statusText}`);
      }
      
      const content = await response.text();
      
      // Validate it's actually Schematron
      if (!content.includes('schematron') && !content.includes('sch:schema')) {
        throw new Error('Downloaded file does not appear to be Schematron');
      }
      
      // Save to cache
      await fs.writeFile(filePath, content, 'utf-8');
      
      // Also save metadata
      const metaPath = filePath.replace('.sch', '.meta.json');
      await fs.writeFile(metaPath, JSON.stringify({
        source: source.name,
        version: source.version,
        url: source.url,
        format: source.format,
        downloadDate: new Date().toISOString()
      }, null, 2), 'utf-8');
      
      console.log(`Successfully downloaded: ${fileName}`);
      return filePath;
    } catch (error) {
      throw new Error(`Failed to download ${source.name}: ${error.message}`);
    }
  }
  
  /**
   * Download all Schematron files for a standard
   */
  public async downloadStandard(
    standard: 'EN16931' | 'XRECHNUNG' | 'PEPPOL'
  ): Promise<string[]> {
    const sources = SCHEMATRON_SOURCES[standard];
    if (!sources) {
      throw new Error(`Unknown standard: ${standard}`);
    }
    
    const paths: string[] = [];
    for (const source of sources) {
      try {
        const path = await this.download(source);
        paths.push(path);
      } catch (error) {
        console.warn(`Failed to download ${source.name}: ${error.message}`);
      }
    }
    
    return paths;
  }
  
  /**
   * Check if a file is cached
   */
  private async isCached(filePath: string): Promise<boolean> {
    try {
      await fs.access(filePath);
      
      // Check if file is not empty
      const stats = await fs.stat(filePath);
      return stats.size > 0;
    } catch {
      return false;
    }
  }
  
  /**
   * Get cached Schematron files
   */
  public async getCachedFiles(): Promise<Array<{
    path: string;
    metadata: any;
  }>> {
    const files: Array<{ path: string; metadata: any }> = [];
    
    try {
      const entries = await fs.readdir(this.cacheDir);
      
      for (const entry of entries) {
        if (entry.endsWith('.sch')) {
          const filePath = path.join(this.cacheDir, entry);
          const metaPath = filePath.replace('.sch', '.meta.json');
          
          try {
            const metadata = JSON.parse(await fs.readFile(metaPath, 'utf-8'));
            files.push({ path: filePath, metadata });
          } catch {
            // No metadata file
            files.push({ path: filePath, metadata: null });
          }
        }
      }
    } catch (error) {
      console.warn(`Failed to list cached files: ${error.message}`);
    }
    
    return files;
  }
  
  /**
   * Clear cache
   */
  public async clearCache(): Promise<void> {
    try {
      const entries = await fs.readdir(this.cacheDir);
      
      for (const entry of entries) {
        if (entry.endsWith('.sch') || entry.endsWith('.meta.json')) {
          await fs.unlink(path.join(this.cacheDir, entry));
        }
      }
      
      console.log('Schematron cache cleared');
    } catch (error) {
      console.warn(`Failed to clear cache: ${error.message}`);
    }
  }
  
  /**
   * Get the appropriate Schematron for a format
   */
  public async getSchematronForFormat(
    standard: 'EN16931' | 'XRECHNUNG' | 'PEPPOL',
    format: 'UBL' | 'CII'
  ): Promise<string | null> {
    const sources = SCHEMATRON_SOURCES[standard];
    if (!sources) return null;
    
    const source = sources.find(s => s.format === format || s.format === 'BOTH');
    if (!source) return null;
    
    return await this.download(source);
  }
  
  /**
   * Update all cached Schematron files
   */
  public async updateAll(): Promise<void> {
    console.log('Updating all Schematron files...');
    
    for (const standard of ['EN16931', 'XRECHNUNG', 'PEPPOL'] as const) {
      await this.downloadStandard(standard);
    }
    
    console.log('All Schematron files updated');
  }
}

/**
 * ISO Schematron skeleton URLs
 * These are needed to compile Schematron to XSLT
 */
export const ISO_SCHEMATRON_SKELETONS = {
  'iso_dsdl_include.xsl': 'https://github.com/Schematron/schematron/raw/master/trunk/schematron/code/iso_dsdl_include.xsl',
  'iso_abstract_expand.xsl': 'https://github.com/Schematron/schematron/raw/master/trunk/schematron/code/iso_abstract_expand.xsl',
  'iso_svrl_for_xslt2.xsl': 'https://github.com/Schematron/schematron/raw/master/trunk/schematron/code/iso_svrl_for_xslt2.xsl',
  'iso_schematron_skeleton_for_saxon.xsl': 'https://github.com/Schematron/schematron/raw/master/trunk/schematron/code/iso_schematron_skeleton_for_saxon.xsl'
};

/**
 * Download ISO Schematron skeleton files
 */
export async function downloadISOSkeletons(targetDir: string = 'assets/schematron/iso'): Promise<void> {
  await fs.mkdir(targetDir, { recursive: true });
  
  console.log('Downloading ISO Schematron skeleton files...');
  
  for (const [name, url] of Object.entries(ISO_SCHEMATRON_SKELETONS)) {
    try {
      const response = await fetch(url);
      if (!response.ok) {
        throw new Error(`HTTP ${response.status}`);
      }
      
      const content = await response.text();
      const filePath = path.join(targetDir, name);
      await fs.writeFile(filePath, content, 'utf-8');
      
      console.log(`Downloaded: ${name}`);
    } catch (error) {
      console.warn(`Failed to download ${name}: ${error.message}`);
    }
  }
  
  console.log('ISO Schematron skeleton download complete');
}