einvoice/test/suite/einvoice_edge-cases/test.edge-08.mixed-formats.ts

import { tap } from '@git.zone/tstest/tapbundle';
import * as plugins from '../plugins.js';
import { EInvoice } from '../../../ts/index.js';
import { PerformanceTracker } from '../performance.tracker.js';

const performanceTracker = new PerformanceTracker('EDGE-08: Mixed Format Documents');

tap.test('EDGE-08: Mixed Format Documents - should handle documents with mixed or ambiguous formats', async (t) => {
  const einvoice = new EInvoice();

  // Test 1: Documents with elements from multiple standards
  const multiStandardElements = await performanceTracker.measureAsync(
    'multi-standard-elements',
    async () => {
      const mixedXML = `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns:ubl="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2"
         xmlns:cii="urn:un:unece:uncefact:data:standard:CrossIndustryInvoice:100"
         xmlns:cbc="urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2">
  <!-- UBL elements -->
  <ubl:ID>MIXED-001</ubl:ID>
  <cbc:IssueDate>2024-01-15</cbc:IssueDate>
  
  <!-- CII elements -->
  <cii:ExchangedDocument>
    <ram:ID>MIXED-001-CII</ram:ID>
  </cii:ExchangedDocument>
  
  <!-- Custom elements -->
  <CustomField>Custom Value</CustomField>
  
  <!-- Mix of both -->
  <LineItems>
    <ubl:InvoiceLine>
      <cbc:ID>1</cbc:ID>
    </ubl:InvoiceLine>
    <cii:SupplyChainTradeLineItem>
      <ram:AssociatedDocumentLineDocument>
        <ram:LineID>2</ram:LineID>
      </ram:AssociatedDocumentLineDocument>
    </cii:SupplyChainTradeLineItem>
  </LineItems>
</Invoice>`;
      
      try {
        const detection = await einvoice.detectFormat(mixedXML);
        const parsed = await einvoice.parseDocument(mixedXML);
        
        return {
          detected: true,
          primaryFormat: detection?.format,
          confidence: detection?.confidence,
          mixedElements: detection?.mixedElements || [],
          standardsFound: detection?.detectedStandards || [],
          parsed: !!parsed
        };
      } catch (error) {
        return {
          detected: false,
          error: error.message
        };
      }
    }
  );

  t.ok(multiStandardElements.detected, 'Multi-standard document was processed');
  t.ok(multiStandardElements.standardsFound?.length > 1, 'Multiple standards detected');

  // Test 2: Namespace confusion
  const namespaceConfusion = await performanceTracker.measureAsync(
    'namespace-confusion',
    async () => {
      const confusedNamespaces = [
        {
          name: 'wrong-namespace-binding',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
  <!-- Using CII elements in UBL namespace -->
  <ExchangedDocument>
    <ID>CONFUSED-001</ID>
  </ExchangedDocument>
</Invoice>`
        },
        {
          name: 'conflicting-default-namespaces',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<root>
  <Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
    <ID>UBL-001</ID>
  </Invoice>
  <Invoice xmlns="urn:un:unece:uncefact:data:standard:CrossIndustryInvoice:100">
    <ExchangedDocument>
      <ID>CII-001</ID>
    </ExchangedDocument>
  </Invoice>
</root>`
        },
        {
          name: 'namespace-switching',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
  <ID>START-UBL</ID>
  <Items xmlns="urn:un:unece:uncefact:data:standard:CrossIndustryInvoice:100">
    <SupplyChainTradeLineItem>
      <LineID xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">1</LineID>
    </SupplyChainTradeLineItem>
  </Items>
</Invoice>`
        }
      ];
      
      const results = [];
      
      for (const test of confusedNamespaces) {
        try {
          const detection = await einvoice.detectFormat(test.xml);
          const parsed = await einvoice.parseDocument(test.xml);
          const validation = await einvoice.validate(parsed);
          
          results.push({
            scenario: test.name,
            detected: true,
            format: detection?.format,
            hasNamespaceIssues: detection?.namespaceIssues || false,
            valid: validation?.isValid || false,
            warnings: validation?.warnings || []
          });
        } catch (error) {
          results.push({
            scenario: test.name,
            detected: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  namespaceConfusion.forEach(result => {
    t.ok(result.detected || result.error, `Namespace confusion ${result.scenario} was handled`);
    if (result.detected) {
      t.ok(result.hasNamespaceIssues || result.warnings.length > 0, 
        'Namespace issues should be detected');
    }
  });

  // Test 3: Hybrid PDF documents
  const hybridPDFDocuments = await performanceTracker.measureAsync(
    'hybrid-pdf-documents',
    async () => {
      const hybridScenarios = [
        {
          name: 'multiple-xml-attachments',
          description: 'PDF with both UBL and CII XML attachments'
        },
        {
          name: 'conflicting-metadata',
          description: 'PDF metadata says ZUGFeRD but contains Factur-X'
        },
        {
          name: 'mixed-version-attachments',
          description: 'PDF with ZUGFeRD v1 and v2 attachments'
        },
        {
          name: 'non-standard-attachment',
          description: 'PDF with standard XML plus custom format'
        }
      ];
      
      const results = [];
      
      for (const scenario of hybridScenarios) {
        // Create mock hybrid PDF
        const hybridPDF = createHybridPDF(scenario.name);
        
        try {
          const extraction = await einvoice.extractFromPDF(hybridPDF);
          
          results.push({
            scenario: scenario.name,
            extracted: true,
            attachmentCount: extraction?.attachments?.length || 0,
            formats: extraction?.detectedFormats || [],
            primaryFormat: extraction?.primaryFormat,
            hasConflicts: extraction?.hasFormatConflicts || false
          });
        } catch (error) {
          results.push({
            scenario: scenario.name,
            extracted: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  hybridPDFDocuments.forEach(result => {
    t.ok(result.extracted || result.error, 
      `Hybrid PDF ${result.scenario} was processed`);
  });

  // Test 4: Schema version mixing
  const schemaVersionMixing = await performanceTracker.measureAsync(
    'schema-version-mixing',
    async () => {
      const versionMixes = [
        {
          name: 'ubl-version-mix',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2"
         xmlns:cbc="urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2"
         xmlns:cbc1="urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-1">
  <cbc:UBLVersionID>2.1</cbc:UBLVersionID>
  <cbc1:ID>OLD-STYLE-ID</cbc1:ID>
  <cbc:IssueDate>2024-01-15</cbc:IssueDate>
</Invoice>`
        },
        {
          name: 'zugferd-version-mix',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<rsm:CrossIndustryDocument>
  <!-- ZUGFeRD 1.0 structure -->
  <rsm:SpecifiedExchangedDocumentContext>
    <ram:GuidelineSpecifiedDocumentContextParameter>
      <ram:ID>urn:ferd:CrossIndustryDocument:invoice:1p0</ram:ID>
    </ram:GuidelineSpecifiedDocumentContextParameter>
  </rsm:SpecifiedExchangedDocumentContext>
  
  <!-- ZUGFeRD 2.1 elements -->
  <rsm:ExchangedDocument>
    <ram:ID>MIXED-VERSION</ram:ID>
  </rsm:ExchangedDocument>
</rsm:CrossIndustryDocument>`
        }
      ];
      
      const results = [];
      
      for (const mix of versionMixes) {
        try {
          const detection = await einvoice.detectFormat(mix.xml);
          const parsed = await einvoice.parseDocument(mix.xml);
          
          results.push({
            scenario: mix.name,
            processed: true,
            detectedVersion: detection?.version,
            versionConflicts: detection?.versionConflicts || [],
            canMigrate: detection?.migrationPath !== undefined
          });
        } catch (error) {
          results.push({
            scenario: mix.name,
            processed: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  schemaVersionMixing.forEach(result => {
    t.ok(result.processed || result.error, 
      `Version mix ${result.scenario} was handled`);
  });

  // Test 5: Invalid format combinations
  const invalidFormatCombos = await performanceTracker.measureAsync(
    'invalid-format-combinations',
    async () => {
      const invalidCombos = [
        {
          name: 'ubl-with-cii-structure',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
  <!-- CII structure in UBL namespace -->
  <rsm:ExchangedDocumentContext>
    <ram:BusinessProcessSpecifiedDocumentContextParameter/>
  </rsm:ExchangedDocumentContext>
  <rsm:ExchangedDocument>
    <ram:ID>INVALID-001</ram:ID>
  </rsm:ExchangedDocument>
</Invoice>`
        },
        {
          name: 'html-invoice-hybrid',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<html>
  <body>
    <Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
      <ID>HTML-WRAPPED</ID>
    </Invoice>
  </body>
</html>`
        },
        {
          name: 'json-xml-mix',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice>
  <ID>JSON-MIX</ID>
  <JSONData>
    {"amount": 100, "currency": "EUR"}
  </JSONData>
  <XMLData>
    <Amount>100</Amount>
    <Currency>EUR</Currency>
  </XMLData>
</Invoice>`
        }
      ];
      
      const results = [];
      
      for (const combo of invalidCombos) {
        try {
          const detection = await einvoice.detectFormat(combo.xml);
          const parsed = await einvoice.parseDocument(combo.xml);
          const validation = await einvoice.validate(parsed);
          
          results.push({
            combo: combo.name,
            detected: !!detection,
            format: detection?.format || 'unknown',
            valid: validation?.isValid || false,
            recoverable: detection?.canRecover || false
          });
        } catch (error) {
          results.push({
            combo: combo.name,
            detected: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  invalidFormatCombos.forEach(result => {
    t.notOk(result.valid, `Invalid combo ${result.combo} should not validate`);
  });

  // Test 6: Partial format documents
  const partialFormatDocuments = await performanceTracker.measureAsync(
    'partial-format-documents',
    async () => {
      const partials = [
        {
          name: 'ubl-header-cii-body',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
  <cbc:ID>PARTIAL-001</cbc:ID>
  <cbc:IssueDate>2024-01-15</cbc:IssueDate>
  
  <!-- Switch to CII for line items -->
  <IncludedSupplyChainTradeLineItem>
    <ram:AssociatedDocumentLineDocument>
      <ram:LineID>1</ram:LineID>
    </ram:AssociatedDocumentLineDocument>
  </IncludedSupplyChainTradeLineItem>
</Invoice>`
        },
        {
          name: 'incomplete-migration',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice>
  <!-- Old format -->
  <InvoiceNumber>OLD-001</InvoiceNumber>
  
  <!-- New format -->
  <ID>NEW-001</ID>
  
  <!-- Mixed date formats -->
  <InvoiceDate>2024-01-15</InvoiceDate>
  <IssueDate>2024-01-15T00:00:00</IssueDate>
</Invoice>`
        }
      ];
      
      const results = [];
      
      for (const partial of partials) {
        try {
          const analysis = await einvoice.analyzeDocument(partial.xml);
          
          results.push({
            scenario: partial.name,
            analyzed: true,
            completeness: analysis?.completeness || 0,
            missingElements: analysis?.missingElements || [],
            formatConsistency: analysis?.formatConsistency || 0,
            migrationNeeded: analysis?.requiresMigration || false
          });
        } catch (error) {
          results.push({
            scenario: partial.name,
            analyzed: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  partialFormatDocuments.forEach(result => {
    t.ok(result.analyzed || result.error, 
      `Partial document ${result.scenario} was analyzed`);
    if (result.analyzed) {
      t.ok(result.formatConsistency < 100, 
        'Format inconsistency should be detected');
    }
  });

  // Test 7: Encoding format conflicts
  const encodingFormatConflicts = await performanceTracker.measureAsync(
    'encoding-format-conflicts',
    async () => {
      const encodingConflicts = [
        {
          name: 'utf8-with-utf16-content',
          declared: 'UTF-8',
          actual: 'UTF-16',
          content: Buffer.from('<?xml version="1.0" encoding="UTF-8"?><Invoice><ID>TEST</ID></Invoice>', 'utf16le')
        },
        {
          name: 'wrong-decimal-separator',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice>
  <!-- European format in US-style document -->
  <TotalAmount>1.234,56</TotalAmount>
  <TaxAmount>234.56</TaxAmount>
</Invoice>`
        },
        {
          name: 'date-format-mixing',
          xml: `<?xml version="1.0" encoding="UTF-8"?>
<Invoice>
  <Dates>
    <IssueDate>2024-01-15</IssueDate>
    <DueDate>15/01/2024</DueDate>
    <DeliveryDate>01-15-2024</DeliveryDate>
    <PaymentDate>20240115</PaymentDate>
  </Dates>
</Invoice>`
        }
      ];
      
      const results = [];
      
      for (const conflict of encodingConflicts) {
        try {
          let parseResult;
          
          if (conflict.content) {
            parseResult = await einvoice.parseDocument(conflict.content);
          } else {
            parseResult = await einvoice.parseDocument(conflict.xml);
          }
          
          const analysis = await einvoice.analyzeFormatConsistency(parseResult);
          
          results.push({
            scenario: conflict.name,
            handled: true,
            encodingIssues: analysis?.encodingIssues || [],
            formatIssues: analysis?.formatIssues || [],
            normalized: analysis?.normalized || false
          });
        } catch (error) {
          results.push({
            scenario: conflict.name,
            handled: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  encodingFormatConflicts.forEach(result => {
    t.ok(result.handled || result.error, 
      `Encoding conflict ${result.scenario} was handled`);
  });

  // Test 8: Format autodetection challenges
  const autodetectionChallenges = await performanceTracker.measureAsync(
    'format-autodetection-challenges',
    async () => {
      const challenges = [
        {
          name: 'minimal-structure',
          xml: '<Invoice><ID>123</ID></Invoice>'
        },
        {
          name: 'generic-xml',
          xml: `<?xml version="1.0"?>
<Document>
  <Header>
    <ID>DOC-001</ID>
    <Date>2024-01-15</Date>
  </Header>
  <Items>
    <Item>
      <Description>Product</Description>
      <Amount>100</Amount>
    </Item>
  </Items>
</Document>`
        },
        {
          name: 'custom-namespace',
          xml: `<?xml version="1.0"?>
<inv:Invoice xmlns:inv="http://custom.company.com/invoice">
  <inv:Number>INV-001</inv:Number>
  <inv:Total>1000</inv:Total>
</inv:Invoice>`
        }
      ];
      
      const results = [];
      
      for (const challenge of challenges) {
        const detectionResult = await einvoice.detectFormat(challenge.xml);
        
        results.push({
          scenario: challenge.name,
          format: detectionResult?.format || 'unknown',
          confidence: detectionResult?.confidence || 0,
          isGeneric: detectionResult?.isGeneric || false,
          suggestedFormats: detectionResult?.possibleFormats || []
        });
      }
      
      return results;
    }
  );

  autodetectionChallenges.forEach(result => {
    t.ok(result.confidence < 100 || result.isGeneric, 
      `Challenge ${result.scenario} should have detection uncertainty`);
  });

  // Test 9: Legacy format mixing
  const legacyFormatMixing = await performanceTracker.measureAsync(
    'legacy-format-mixing',
    async () => {
      const legacyMixes = [
        {
          name: 'edifact-xml-hybrid',
          content: `UNB+UNOC:3+SENDER+RECEIVER+240115:1200+1++INVOIC'
<?xml version="1.0"?>
<AdditionalData>
  <Invoice>
    <ID>HYBRID-001</ID>
  </Invoice>
</AdditionalData>
UNZ+1+1'`
        },
        {
          name: 'csv-xml-combination',
          content: `INVOICE_HEADER
ID,Date,Amount
INV-001,2024-01-15,1000.00

<?xml version="1.0"?>
<InvoiceDetails>
  <LineItems>
    <Item>Product A</Item>
  </LineItems>
</InvoiceDetails>`
        }
      ];
      
      const results = [];
      
      for (const mix of legacyMixes) {
        try {
          const detection = await einvoice.detectFormat(mix.content);
          const extraction = await einvoice.extractStructuredData(mix.content);
          
          results.push({
            scenario: mix.name,
            processed: true,
            formatsFound: detection?.multipleFormats || [],
            primaryFormat: detection?.primaryFormat,
            dataExtracted: !!extraction?.data
          });
        } catch (error) {
          results.push({
            scenario: mix.name,
            processed: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  legacyFormatMixing.forEach(result => {
    t.ok(result.processed || result.error, 
      `Legacy mix ${result.scenario} was handled`);
  });

  // Test 10: Format conversion conflicts
  const formatConversionConflicts = await performanceTracker.measureAsync(
    'format-conversion-conflicts',
    async () => {
      // Create invoice with format-specific features
      const sourceInvoice = `<?xml version="1.0" encoding="UTF-8"?>
<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2">
  <ID>CONVERT-001</ID>
  <!-- UBL-specific extension -->
  <UBLExtensions>
    <UBLExtension>
      <ExtensionContent>
        <CustomField>UBL-Only-Data</CustomField>
      </ExtensionContent>
    </UBLExtension>
  </UBLExtensions>
  
  <!-- Format-specific calculation -->
  <AllowanceCharge>
    <ChargeIndicator>false</ChargeIndicator>
    <AllowanceChargeReason>Discount</AllowanceChargeReason>
    <Amount currencyID="EUR">50.00</Amount>
  </AllowanceCharge>
</Invoice>`;
      
      const targetFormats = ['cii', 'xrechnung', 'fatturapa'];
      const results = [];
      
      for (const target of targetFormats) {
        try {
          const converted = await einvoice.convertFormat(sourceInvoice, target);
          const analysis = await einvoice.analyzeConversion(sourceInvoice, converted);
          
          results.push({
            targetFormat: target,
            converted: true,
            dataLoss: analysis?.dataLoss || [],
            unsupportedFeatures: analysis?.unsupportedFeatures || [],
            warnings: analysis?.warnings || []
          });
        } catch (error) {
          results.push({
            targetFormat: target,
            converted: false,
            error: error.message
          });
        }
      }
      
      return results;
    }
  );

  formatConversionConflicts.forEach(result => {
    t.ok(result.converted || result.error, 
      `Conversion to ${result.targetFormat} was attempted`);
    if (result.converted) {
      t.ok(result.dataLoss.length > 0 || result.warnings.length > 0, 
        'Format-specific features should cause warnings');
    }
  });

  // Print performance summary
  performanceTracker.printSummary();
});

// Helper function to create hybrid PDF
function createHybridPDF(scenario: string): Buffer {
  // Simplified mock - in reality would create actual PDF structure
  const mockStructure = {
    'multiple-xml-attachments': {
      attachments: [
        { name: 'invoice.ubl.xml', type: 'application/xml' },
        { name: 'invoice.cii.xml', type: 'application/xml' }
      ]
    },
    'conflicting-metadata': {
      metadata: { format: 'ZUGFeRD' },
      attachments: [{ name: 'facturx.xml', type: 'application/xml' }]
    },
    'mixed-version-attachments': {
      attachments: [
        { name: 'zugferd_v1.xml', version: '1.0' },
        { name: 'zugferd_v2.xml', version: '2.1' }
      ]
    },
    'non-standard-attachment': {
      attachments: [
        { name: 'invoice.xml', type: 'application/xml' },
        { name: 'custom.json', type: 'application/json' }
      ]
    }
  };
  
  return Buffer.from(JSON.stringify(mockStructure[scenario] || {}));
}

// Run the test
tap.start();