From cd87eb18e860f34c0cee99e240849519a036a8b3 Mon Sep 17 00:00:00 2001
From: Connor Yoh <connor@stirlingpdf.com>
Date: Thu, 31 Jul 2025 15:23:38 +0100
Subject: [PATCH] pdf-pdfa

---
 .../tools/convert/ConvertSettings.tsx         | 20 ++++++
 .../tools/convert/ConvertToPdfaSettings.tsx   | 60 +++++++++++++++++
 .../tools/convert/useConvertOperation.ts      |  8 ++-
 .../convert/useConvertParameters.test.ts      | 13 ++++
 .../tools/convert/useConvertParameters.ts     |  6 ++
 .../src/hooks/usePdfSignatureDetection.ts     | 66 +++++++++++++++++++
 .../ConvertSmartDetectionIntegration.test.tsx | 34 ++++++++++
 7 files changed, 206 insertions(+), 1 deletion(-)
 create mode 100644 frontend/src/components/tools/convert/ConvertToPdfaSettings.tsx
 create mode 100644 frontend/src/hooks/usePdfSignatureDetection.ts

diff --git a/frontend/src/components/tools/convert/ConvertSettings.tsx b/frontend/src/components/tools/convert/ConvertSettings.tsx
index 355ce042f..2d030a03f 100644
--- a/frontend/src/components/tools/convert/ConvertSettings.tsx
+++ b/frontend/src/components/tools/convert/ConvertSettings.tsx
@@ -12,6 +12,7 @@ import ConvertToImageSettings from "./ConvertToImageSettings";
 import ConvertFromImageSettings from "./ConvertFromImageSettings";
 import ConvertFromWebSettings from "./ConvertFromWebSettings";
 import ConvertFromEmailSettings from "./ConvertFromEmailSettings";
+import ConvertToPdfaSettings from "./ConvertToPdfaSettings";
 import { ConvertParameters } from "../../../hooks/tools/convert/useConvertParameters";
 import { 
   FROM_FORMAT_OPTIONS,
@@ -114,6 +115,9 @@ const ConvertSettings = ({
       downloadHtml: false,
       includeAllRecipients: false,
     });
+    onParameterChange('pdfaOptions', {
+      outputFormat: 'pdfa-1',
+    });
     // Disable smart detection when manually changing source format
     onParameterChange('isSmartDetection', false);
     onParameterChange('smartDetectionType', 'none');
@@ -161,6 +165,9 @@ const ConvertSettings = ({
       downloadHtml: false,
       includeAllRecipients: false,
     });
+    onParameterChange('pdfaOptions', {
+      outputFormat: 'pdfa-1',
+    });
   };
 
 
@@ -274,6 +281,19 @@ const ConvertSettings = ({
         </>
       )}
 
+      {/* PDF to PDF/A options */}
+      {parameters.fromExtension === 'pdf' && parameters.toExtension === 'pdfa' && (
+        <>
+          <Divider />
+          <ConvertToPdfaSettings
+            parameters={parameters}
+            onParameterChange={onParameterChange}
+            selectedFiles={selectedFiles}
+            disabled={disabled}
+          />
+        </>
+      )}
+
     </Stack>
   );
 };
diff --git a/frontend/src/components/tools/convert/ConvertToPdfaSettings.tsx b/frontend/src/components/tools/convert/ConvertToPdfaSettings.tsx
new file mode 100644
index 000000000..0422ee780
--- /dev/null
+++ b/frontend/src/components/tools/convert/ConvertToPdfaSettings.tsx
@@ -0,0 +1,60 @@
+import React from 'react';
+import { Stack, Text, Select, Alert } from '@mantine/core';
+import { useTranslation } from 'react-i18next';
+import { ConvertParameters } from '../../../hooks/tools/convert/useConvertParameters';
+import { usePdfSignatureDetection } from '../../../hooks/usePdfSignatureDetection';
+
+interface ConvertToPdfaSettingsProps {
+  parameters: ConvertParameters;
+  onParameterChange: (key: keyof ConvertParameters, value: any) => void;
+  selectedFiles: File[];
+  disabled?: boolean;
+}
+
+const ConvertToPdfaSettings = ({ 
+  parameters, 
+  onParameterChange,
+  selectedFiles,
+  disabled = false 
+}: ConvertToPdfaSettingsProps) => {
+  const { t } = useTranslation();
+  const { hasDigitalSignatures, isChecking } = usePdfSignatureDetection(selectedFiles);
+
+  const pdfaFormatOptions = [
+    { value: 'pdfa-1', label: 'PDF/A-1b' },
+    { value: 'pdfa', label: 'PDF/A-2b' }
+  ];
+
+  return (
+    <Stack gap="sm" data-testid="pdfa-settings">
+      <Text size="sm" fw={500}>{t("convert.pdfaOptions", "PDF/A Options")}:</Text>
+      
+      {hasDigitalSignatures && (
+        <Alert color="yellow" size="sm">
+          <Text size="sm">
+            {t("convert.pdfaDigitalSignatureWarning", "The PDF contains a digital signature. This will be removed in the next step.")}
+          </Text>
+        </Alert>
+      )}
+      
+      <Stack gap="xs">
+        <Text size="xs" fw={500}>{t("convert.outputFormat", "Output Format")}:</Text>
+        <Select
+          value={parameters.pdfaOptions.outputFormat}
+          onChange={(value) => onParameterChange('pdfaOptions', { 
+            ...parameters.pdfaOptions, 
+            outputFormat: value || 'pdfa-1' 
+          })}
+          data={pdfaFormatOptions}
+          disabled={disabled || isChecking}
+          data-testid="pdfa-output-format-select"
+        />
+        <Text size="xs" c="dimmed">
+          {t("convert.pdfaNote", "PDF/A-1b is more compatible, PDF/A-2b supports more features.")}
+        </Text>
+      </Stack>
+    </Stack>
+  );
+};
+
+export default ConvertToPdfaSettings;
\ No newline at end of file
diff --git a/frontend/src/hooks/tools/convert/useConvertOperation.ts b/frontend/src/hooks/tools/convert/useConvertOperation.ts
index cf01ee147..1d8d3f08e 100644
--- a/frontend/src/hooks/tools/convert/useConvertOperation.ts
+++ b/frontend/src/hooks/tools/convert/useConvertOperation.ts
@@ -46,6 +46,8 @@ const shouldProcessFilesSeparately = (
      parameters.toExtension === 'pdf' && !parameters.imageOptions.combineImages) ||
     // PDF to image conversions (each PDF should generate its own image file)
     (parameters.fromExtension === 'pdf' && isImageFormat(parameters.toExtension)) ||
+    // PDF to PDF/A conversions (each PDF should be processed separately)
+    (parameters.fromExtension === 'pdf' && parameters.toExtension === 'pdfa') ||
     // Web files to PDF conversions (each web file should generate its own PDF)
     ((isWebFormat(parameters.fromExtension) || parameters.fromExtension === 'web') && 
      parameters.toExtension === 'pdf') ||
@@ -143,7 +145,7 @@ export const useConvertOperation = (): ConvertOperationHook => {
       formData.append("fileInput", file);
     });
 
-    const { fromExtension, toExtension, imageOptions, htmlOptions, emailOptions } = parameters;
+    const { fromExtension, toExtension, imageOptions, htmlOptions, emailOptions, pdfaOptions } = parameters;
 
     // Add conversion-specific parameters
     if (isImageFormat(toExtension)) {
@@ -170,6 +172,9 @@ export const useConvertOperation = (): ConvertOperationHook => {
       formData.append("maxAttachmentSizeMB", emailOptions.maxAttachmentSizeMB.toString());
       formData.append("downloadHtml", emailOptions.downloadHtml.toString());
       formData.append("includeAllRecipients", emailOptions.includeAllRecipients.toString());
+    } else if (fromExtension === 'pdf' && toExtension === 'pdfa') {
+      // PDF to PDF/A conversion with output format
+      formData.append("outputFormat", pdfaOptions.outputFormat);
     } else if (fromExtension === 'pdf' && toExtension === 'csv') {
       // CSV extraction - always process all pages for simplified workflow
       formData.append("pageNumbers", "all");
@@ -199,6 +204,7 @@ export const useConvertOperation = (): ConvertOperationHook => {
           imageOptions: parameters.imageOptions,
           htmlOptions: parameters.htmlOptions,
           emailOptions: parameters.emailOptions,
+          pdfaOptions: parameters.pdfaOptions,
         },
         fileSize: selectedFiles[0].size
       }
diff --git a/frontend/src/hooks/tools/convert/useConvertParameters.test.ts b/frontend/src/hooks/tools/convert/useConvertParameters.test.ts
index 7ed63fcd9..25cd4f3ad 100644
--- a/frontend/src/hooks/tools/convert/useConvertParameters.test.ts
+++ b/frontend/src/hooks/tools/convert/useConvertParameters.test.ts
@@ -23,6 +23,7 @@ describe('useConvertParameters', () => {
       expect(result.current.parameters.emailOptions.maxAttachmentSizeMB).toBe(10);
       expect(result.current.parameters.emailOptions.downloadHtml).toBe(false);
       expect(result.current.parameters.emailOptions.includeAllRecipients).toBe(false);
+      expect(result.current.parameters.pdfaOptions.outputFormat).toBe('pdfa-1');
     });
 
     test('should update individual parameters', () => {
@@ -82,6 +83,18 @@ describe('useConvertParameters', () => {
       expect(result.current.parameters.emailOptions.includeAllRecipients).toBe(true);
     });
 
+    test('should update nested PDF/A options', () => {
+      const { result } = renderHook(() => useConvertParameters());
+      
+      act(() => {
+        result.current.updateParameter('pdfaOptions', {
+          outputFormat: 'pdfa'
+        });
+      });
+      
+      expect(result.current.parameters.pdfaOptions.outputFormat).toBe('pdfa');
+    });
+
     test('should reset parameters to defaults', () => {
       const { result } = renderHook(() => useConvertParameters());
       
diff --git a/frontend/src/hooks/tools/convert/useConvertParameters.ts b/frontend/src/hooks/tools/convert/useConvertParameters.ts
index 19c536d65..08fe4f762 100644
--- a/frontend/src/hooks/tools/convert/useConvertParameters.ts
+++ b/frontend/src/hooks/tools/convert/useConvertParameters.ts
@@ -32,6 +32,9 @@ export interface ConvertParameters {
     downloadHtml: boolean;
     includeAllRecipients: boolean;
   };
+  pdfaOptions: {
+    outputFormat: string;
+  };
   isSmartDetection: boolean;
   smartDetectionType: 'mixed' | 'images' | 'web' | 'none';
 }
@@ -67,6 +70,9 @@ const initialParameters: ConvertParameters = {
     downloadHtml: false,
     includeAllRecipients: false,
   },
+  pdfaOptions: {
+    outputFormat: 'pdfa-1',
+  },
   isSmartDetection: false,
   smartDetectionType: 'none',
 };
diff --git a/frontend/src/hooks/usePdfSignatureDetection.ts b/frontend/src/hooks/usePdfSignatureDetection.ts
new file mode 100644
index 000000000..96d80fb9d
--- /dev/null
+++ b/frontend/src/hooks/usePdfSignatureDetection.ts
@@ -0,0 +1,66 @@
+import { useState, useEffect } from 'react';
+import * as pdfjsLib from 'pdfjs-dist';
+
+export interface PdfSignatureDetectionResult {
+  hasDigitalSignatures: boolean;
+  isChecking: boolean;
+}
+
+export const usePdfSignatureDetection = (files: File[]): PdfSignatureDetectionResult => {
+  const [hasDigitalSignatures, setHasDigitalSignatures] = useState(false);
+  const [isChecking, setIsChecking] = useState(false);
+
+  useEffect(() => {
+    const checkForDigitalSignatures = async () => {
+      if (files.length === 0) {
+        setHasDigitalSignatures(false);
+        return;
+      }
+
+      setIsChecking(true);
+      let foundSignature = false;
+
+      try {
+        // Set up PDF.js worker
+        pdfjsLib.GlobalWorkerOptions.workerSrc = '/pdfjs-legacy/pdf.worker.mjs';
+
+        for (const file of files) {
+          const arrayBuffer = await file.arrayBuffer();
+          
+          try {
+            const pdf = await pdfjsLib.getDocument({ data: arrayBuffer }).promise;
+
+            for (let i = 1; i <= pdf.numPages; i++) {
+              const page = await pdf.getPage(i);
+              const annotations = await page.getAnnotations({ intent: 'display' });
+
+              annotations.forEach(annotation => {
+                if (annotation.subtype === 'Widget' && annotation.fieldType === 'Sig') {
+                  foundSignature = true;
+                }
+              });
+
+              if (foundSignature) break;
+            }
+          } catch (error) {
+            console.warn('Error analyzing PDF for signatures:', error);
+          }
+
+          if (foundSignature) break;
+        }
+      } catch (error) {
+        console.warn('Error checking for digital signatures:', error);
+      }
+
+      setHasDigitalSignatures(foundSignature);
+      setIsChecking(false);
+    };
+
+    checkForDigitalSignatures();
+  }, [files]);
+
+  return {
+    hasDigitalSignatures,
+    isChecking
+  };
+};
\ No newline at end of file
diff --git a/frontend/src/tests/convert/ConvertSmartDetectionIntegration.test.tsx b/frontend/src/tests/convert/ConvertSmartDetectionIntegration.test.tsx
index e08ae0462..f6be30407 100644
--- a/frontend/src/tests/convert/ConvertSmartDetectionIntegration.test.tsx
+++ b/frontend/src/tests/convert/ConvertSmartDetectionIntegration.test.tsx
@@ -322,6 +322,40 @@ describe('Convert Tool - Smart Detection Integration Tests', () => {
       expect(formData.get('downloadHtml')).toBe('true');
       expect(formData.get('includeAllRecipients')).toBe('true');
     });
+
+    test('should send correct PDF/A parameters for pdf-to-pdfa conversion', async () => {
+      const { result: paramsResult } = renderHook(() => useConvertParameters(), {
+        wrapper: TestWrapper
+      });
+      
+      const { result: operationResult } = renderHook(() => useConvertOperation(), {
+        wrapper: TestWrapper
+      });
+      
+      const pdfFile = new File(['pdf content'], 'document.pdf', { type: 'application/pdf' });
+      
+      // Set up PDF/A conversion parameters
+      act(() => {
+        paramsResult.current.updateParameter('fromExtension', 'pdf');
+        paramsResult.current.updateParameter('toExtension', 'pdfa');
+        paramsResult.current.updateParameter('pdfaOptions', {
+          outputFormat: 'pdfa'
+        });
+      });
+      
+      await act(async () => {
+        await operationResult.current.executeOperation(
+          paramsResult.current.parameters,
+          [pdfFile]
+        );
+      });
+      
+      const formData = mockedAxios.post.mock.calls[0][1] as FormData;
+      expect(formData.get('outputFormat')).toBe('pdfa');
+      expect(mockedAxios.post).toHaveBeenCalledWith('/api/v1/convert/pdf/pdfa', expect.any(FormData), {
+        responseType: 'blob'
+      });
+    });
   });
 
   describe('Image Conversion Options Integration', () => {