einvoice/ts/formats/utils/format.detector.ts

import { InvoiceFormat } from '../../interfaces/common.js';
import { DOMParser, xpath } from '../../plugins.js';
import { CII_PROFILE_IDS, ZUGFERD_V1_NAMESPACES } from '../cii/cii.types.js';

/**
 * Utility class for detecting invoice formats
 */
export class FormatDetector {
  /**
   * Detects the format of an XML document
   * @param xml XML content to analyze
   * @returns Detected invoice format
   */
  public static detectFormat(xml: string): InvoiceFormat {
    try {
      // Quick check for empty or invalid XML
      if (!xml || typeof xml !== 'string' || xml.trim().length === 0) {
        return InvoiceFormat.UNKNOWN;
      }

      // Quick string-based pre-checks for performance
      const quickCheck = FormatDetector.quickFormatCheck(xml);
      if (quickCheck !== InvoiceFormat.UNKNOWN) {
        return quickCheck;
      }

      // More thorough parsing-based checks
      const doc = new DOMParser().parseFromString(xml, 'application/xml');
      const root = doc.documentElement;

      if (!root) {
        return InvoiceFormat.UNKNOWN;
      }

      // UBL detection (Invoice or CreditNote root element)
      if (FormatDetector.isUBLFormat(root)) {
        // Check for XRechnung customization
        if (FormatDetector.isXRechnungFormat(doc)) {
          return InvoiceFormat.XRECHNUNG;
        }
        return InvoiceFormat.UBL;
      }

      // Factur-X/ZUGFeRD detection (CrossIndustryInvoice root element)
      if (FormatDetector.isCIIFormat(root)) {
        return FormatDetector.detectCIIFormat(doc, xml);
      }

      // ZUGFeRD v1 detection (CrossIndustryDocument root element)
      if (FormatDetector.isZUGFeRDV1Format(root)) {
        return InvoiceFormat.ZUGFERD;
      }

      // FatturaPA detection
      if (FormatDetector.isFatturaPAFormat(root)) {
        return InvoiceFormat.FATTURAPA;
      }

      return InvoiceFormat.UNKNOWN;
    } catch (error) {
      console.error('Error detecting format:', error);
      return InvoiceFormat.UNKNOWN;
    }
  }

  /**
   * Performs a quick format check based on string content
   * This is faster than full XML parsing for obvious cases
   * @param xml XML string
   * @returns Detected format or UNKNOWN if more analysis is needed
   */
  private static quickFormatCheck(xml: string): InvoiceFormat {
    // Only scan a small prefix so large payloads do not create another full-size string copy.
    const sample = xml.slice(0, 65536);

    // Root-element checks avoid a DOM parse for the common invoice formats.
    if (/<(?:[A-Za-z_][\w.-]*:)?(?:Invoice|CreditNote)\b/.test(sample)) {
      const customizationIdMatch = sample.match(
        /<[^>]*CustomizationID[^>]*>\s*([^<]+?)\s*<\/[^>]*CustomizationID>/i,
      );
      const customizationId = customizationIdMatch?.[1] ?? '';

      if (/xrechnung/i.test(customizationId) || /urn:xoev-de:kosit:standard:xrechnung/i.test(customizationId)) {
        return InvoiceFormat.XRECHNUNG;
      }
      return InvoiceFormat.UBL;
    }

    if (/<(?:[A-Za-z_][\w.-]*:)?CrossIndustryInvoice\b/.test(sample)) {
      const guidelineIdMatch = sample.match(
        /<[^>]*GuidelineSpecifiedDocumentContextParameter[^>]*>[\s\S]*?<[^>]*ID[^>]*>\s*([^<]+?)\s*<\/[^>]*ID>/i,
      );
      const guidelineId = guidelineIdMatch?.[1] ?? '';

      if (/xrechnung/i.test(guidelineId)) {
        return InvoiceFormat.XRECHNUNG;
      }
      if (/factur-x/i.test(guidelineId) || /urn:cen\.eu:en16931:2017/i.test(guidelineId)) {
        return InvoiceFormat.FACTURX;
      }
      if (/zugferd/i.test(guidelineId) || /urn:ferd:/i.test(guidelineId) || /urn:zugferd/i.test(guidelineId)) {
        return InvoiceFormat.ZUGFERD;
      }
      return InvoiceFormat.CII;
    }

    if (/<(?:[A-Za-z_][\w.-]*:)?CrossIndustryDocument\b/.test(sample)) {
      return InvoiceFormat.ZUGFERD;
    }

    if (/<FatturaElettronica\b/.test(sample)) {
      return InvoiceFormat.FATTURAPA;
    }

    // Check for obvious Factur-X indicators
    if (
      /factur-x\.eu/i.test(sample) ||
      /factur-x\.xml/i.test(sample) ||
      /factur-x:/i.test(sample) ||
      (/urn:cen\.eu:en16931:2017/i.test(sample) && /factur-x/i.test(sample))
    ) {
      return InvoiceFormat.FACTURX;
    }

    // Check for obvious ZUGFeRD indicators
    if (
      /zugferd:/i.test(sample) ||
      /zugferd-invoice\.xml/i.test(sample) ||
      /urn:ferd:/i.test(sample) ||
      /urn:zugferd/i.test(sample)
    ) {
      return InvoiceFormat.ZUGFERD;
    }

    // Check for obvious XRechnung indicators
    if (
      /xrechnung/i.test(sample) ||
      /urn:xoev-de:kosit:standard:xrechnung/i.test(sample)
    ) {
      return InvoiceFormat.XRECHNUNG;
    }

    // Check for obvious FatturaPA indicators
    if (
      /fatturapa/i.test(sample) ||
      /fattura elettronica/i.test(sample) ||
      /fatturaelettronica/i.test(sample)
    ) {
      return InvoiceFormat.FATTURAPA;
    }

    // Need more analysis
    return InvoiceFormat.UNKNOWN;
  }

  /**
   * Checks if the document is a UBL format
   * @param root Root element
   * @returns True if it's a UBL format
   */
  private static isUBLFormat(root: Element): boolean {
    return (
      root.nodeName === 'Invoice' ||
      root.nodeName === 'CreditNote' ||
      root.nodeName === 'ubl:Invoice' ||
      root.nodeName === 'ubl:CreditNote' ||
      root.nodeName.endsWith(':Invoice') ||
      root.nodeName.endsWith(':CreditNote')
    );
  }

  /**
   * Checks if the document is an XRechnung format
   * @param doc XML document
   * @returns True if it's an XRechnung format
   */
  private static isXRechnungFormat(doc: Document): boolean {
    try {
      // Set up namespaces for XPath queries
      const namespaces = {
        'cbc': 'urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2',
        'ubl': 'urn:oasis:names:specification:ubl:schema:xsd:Invoice-2'
      };

      // Create XPath selector with namespaces
      const select = xpath.useNamespaces(namespaces);

      // Use getElementsByTagName directly for more reliable results
      const customizationNodes = doc.getElementsByTagName('cbc:CustomizationID');
      
      // Check if any CustomizationID node contains "xrechnung"
      for (let i = 0; i < customizationNodes.length; i++) {
        const node = customizationNodes[i];
        if (node.textContent && node.textContent.includes('xrechnung')) {
          return true;
        }
      }
      
      return false;
    } catch (error) {
      console.warn('Error checking for XRechnung format:', error);
      // If direct DOM access fails, try a string-based approach
      const xmlStr = new XMLSerializer().serializeToString(doc);
      return xmlStr.includes('xrechnung') || xmlStr.includes('XRechnung');
    }
  }

  /**
   * Checks if the document is a CII format (Factur-X/ZUGFeRD v2+)
   * @param root Root element
   * @returns True if it's a CII format
   */
  private static isCIIFormat(root: Element): boolean {
    return (
      root.nodeName === 'rsm:CrossIndustryInvoice' ||
      root.nodeName === 'CrossIndustryInvoice' ||
      root.nodeName.endsWith(':CrossIndustryInvoice')
    );
  }

  /**
   * Checks if the document is a ZUGFeRD v1 format
   * @param root Root element
   * @returns True if it's a ZUGFeRD v1 format
   */
  private static isZUGFeRDV1Format(root: Element): boolean {
    return (
      root.nodeName === 'rsm:CrossIndustryDocument' ||
      root.nodeName === 'CrossIndustryDocument' ||
      root.nodeName === 'ram:CrossIndustryDocument' ||
      root.nodeName.endsWith(':CrossIndustryDocument')
    );
  }

  /**
   * Checks if the document is a FatturaPA format
   * @param root Root element
   * @returns True if it's a FatturaPA format
   */
  private static isFatturaPAFormat(root: Element): boolean {
    const xmlns = root.getAttribute('xmlns') || '';
    return root.nodeName === 'FatturaElettronica' || xmlns.includes('fatturapa.gov.it');
  }

  /**
   * Detects the specific CII format (Factur-X vs ZUGFeRD)
   * @param doc XML document
   * @param xml Original XML string for fallback checks
   * @returns Detected format
   */
  private static detectCIIFormat(doc: Document, xml: string): InvoiceFormat {
    try {
      // Use direct DOM traversal instead of XPath for more reliable behavior
      const contextNodes = doc.getElementsByTagNameNS(
        'urn:un:unece:uncefact:data:standard:CrossIndustryInvoice:100',
        'ExchangedDocumentContext'
      );
      
      if (contextNodes.length === 0) {
        // Try without namespace
        const noNsContextNodes = doc.getElementsByTagName('ExchangedDocumentContext');
        if (noNsContextNodes.length === 0) {
          // Fallback to string-based detection
          return FormatDetector.detectCIIFormatFromString(xml);
        }
      }
      
      // Loop through all potential context nodes
      const allContextNodes = [...Array.from(contextNodes), ...Array.from(doc.getElementsByTagName('ExchangedDocumentContext'))];
      
      for (const contextNode of allContextNodes) {
        // Find guideline parameter
        const guidelineNodes = contextNode.getElementsByTagName('ram:GuidelineSpecifiedDocumentContextParameter');
        
        if (guidelineNodes.length === 0) {
          continue;
        }
        
        for (const guidelineNode of Array.from(guidelineNodes)) {
          // Find ID element
          const idNodes = guidelineNode.getElementsByTagName('ram:ID');
          
          if (idNodes.length === 0) {
            continue;
          }
          
          for (const idNode of Array.from(idNodes)) {
            const profileText = idNode.textContent || '';
            
            // Check for ZUGFeRD profiles (v1 and v2)
            if (
              profileText.includes('zugferd') ||
              profileText.includes('urn:ferd:') ||
              profileText === CII_PROFILE_IDS.ZUGFERD_BASIC ||
              profileText === CII_PROFILE_IDS.ZUGFERD_COMFORT ||
              profileText === CII_PROFILE_IDS.ZUGFERD_EXTENDED ||
              profileText === CII_PROFILE_IDS.ZUGFERD_V1_BASIC ||
              profileText === CII_PROFILE_IDS.ZUGFERD_V1_COMFORT ||
              profileText === CII_PROFILE_IDS.ZUGFERD_V1_EXTENDED
            ) {
              return InvoiceFormat.ZUGFERD;
            }

            // Check for Factur-X profiles
            if (
              profileText.includes('factur-x') ||
              profileText === CII_PROFILE_IDS.FACTURX_MINIMUM ||
              profileText === CII_PROFILE_IDS.FACTURX_BASIC ||
              profileText === CII_PROFILE_IDS.FACTURX_EN16931
            ) {
              return InvoiceFormat.FACTURX;
            }
          }
        }
      }
      
      // If we reach here, fall back to string checking
      return FormatDetector.detectCIIFormatFromString(xml);
    } catch (error) {
      console.warn('Error detecting CII format, falling back to generic CII:', error);
      return FormatDetector.detectCIIFormatFromString(xml);
    }
  }
  
  /**
   * Fallback method to detect CII format from string content
   * @param xml XML string
   * @returns Detected format
   */
  private static detectCIIFormatFromString(xml: string): InvoiceFormat {
    // Check for Factur-X indicators
    if (xml.includes('factur-x') || xml.includes('Factur-X')) {
      return InvoiceFormat.FACTURX;
    }
    
    // Check for ZUGFeRD indicators
    if (xml.includes('zugferd') || xml.includes('ZUGFeRD')) {
      return InvoiceFormat.ZUGFERD;
    }
    
    // Generic CII if we can't determine more specifically
    return InvoiceFormat.CII;
  }
}