From 34e38f19e5e1f74de4798578f9d66be8c5290071 Mon Sep 17 00:00:00 2001
From: TekH <h.tek@digitaldata.works>
Date: Tue, 21 Jul 2026 09:26:34 +0200
Subject: [PATCH] feat: Add PdfAttachmentController extract endpoints
 (multipart + Base64)

---
 .../Controllers/PdfAttachmentController.cs    | 87 +++++++++++++++++++
 1 file changed, 87 insertions(+)

diff --git a/DocumentOperator.API/Controllers/PdfAttachmentController.cs b/DocumentOperator.API/Controllers/PdfAttachmentController.cs
index a6d31af..0b03274 100644
--- a/DocumentOperator.API/Controllers/PdfAttachmentController.cs
+++ b/DocumentOperator.API/Controllers/PdfAttachmentController.cs
@@ -1,5 +1,6 @@
 using DocumentOperator.Application.CheckPdfAttachments.Queries;
 using DocumentOperator.Application.Common.DTOs;
+using DocumentOperator.Application.ExtractPdfAttachments;
 using DocumentOperator.Domain.Common.Exceptions;
 using MediatR;
 using Microsoft.AspNetCore.Http;
@@ -82,6 +83,80 @@ public class PdfAttachmentController(IMediator mediator) : ControllerBase
 
         return Ok(result);
     }
+
+    /// <summary>
+    /// Extracts all embedded files from a PDF and returns them as a ZIP archive.
+    /// Supports multipart/form-data file upload.
+    /// </summary>
+    /// <param name="file">The PDF file to extract attachments from</param>
+    /// <param name="cancellationToken">Cancellation token</param>
+    /// <returns>ZIP archive containing all extracted attachments</returns>
+    /// <response code="200">Attachments extracted successfully - returns ZIP file</response>
+    /// <response code="400">Invalid input (file missing, not a PDF, or corrupted)</response>
+    /// <response code="404">PDF contains no attachments</response>
+    /// <response code="500">Internal server error during PDF processing</response>
+    [HttpPost("extract")]
+    [Consumes("multipart/form-data")]
+    [ProducesResponseType(typeof(FileContentResult), StatusCodes.Status200OK)]
+    [ProducesResponseType(typeof(ProblemDetails), StatusCodes.Status400BadRequest)]
+    [ProducesResponseType(typeof(ProblemDetails), StatusCodes.Status404NotFound)]
+    [ProducesResponseType(typeof(ProblemDetails), StatusCodes.Status500InternalServerError)]
+    public async Task<IActionResult> ExtractAttachmentsFromFile(
+        IFormFile file,
+        CancellationToken cancellationToken)
+    {
+        // Use IFormFile stream directly (no intermediate byte[] conversion)
+        using var pdfStream = file.OpenReadStream();
+
+        // Send command to MediatR
+        var command = new ExtractPdfAttachmentsCommand { PdfStream = pdfStream };
+        byte[] zipBytes = await mediator.Send(command, cancellationToken);
+
+        // Return ZIP file
+        return File(zipBytes, "application/zip", "attachments.zip");
+    }
+
+    /// <summary>
+    /// Extracts all embedded files from a PDF and returns them as a ZIP archive.
+    /// Supports Base64-encoded PDF via JSON payload.
+    /// </summary>
+    /// <param name="request">Request containing Base64-encoded PDF</param>
+    /// <param name="cancellationToken">Cancellation token</param>
+    /// <returns>ZIP archive containing all extracted attachments</returns>
+    /// <response code="200">Attachments extracted successfully - returns ZIP file</response>
+    /// <response code="400">Invalid input (Base64 format error, not a PDF, or corrupted)</response>
+    /// <response code="404">PDF contains no attachments</response>
+    /// <response code="500">Internal server error during PDF processing</response>
+    [HttpPost("extract")]
+    [Consumes("application/json")]
+    [ProducesResponseType(typeof(FileContentResult), StatusCodes.Status200OK)]
+    [ProducesResponseType(typeof(ProblemDetails), StatusCodes.Status400BadRequest)]
+    [ProducesResponseType(typeof(ProblemDetails), StatusCodes.Status404NotFound)]
+    [ProducesResponseType(typeof(ProblemDetails), StatusCodes.Status500InternalServerError)]
+    public async Task<IActionResult> ExtractAttachmentsFromBase64(
+        [FromBody] ExtractPdfAttachmentsRequest request,
+        CancellationToken cancellationToken)
+    {
+        // Convert Base64 to stream (wrap in try-catch to throw BadRequestException)
+        byte[] pdfBytes;
+        try
+        {
+            pdfBytes = Convert.FromBase64String(request.Base64Pdf);
+        }
+        catch (FormatException ex)
+        {
+            throw new BadRequestException("Invalid Base64 format: " + ex.Message);
+        }
+        
+        using var pdfStream = new MemoryStream(pdfBytes);
+
+        // Send command to MediatR
+        var command = new ExtractPdfAttachmentsCommand { PdfStream = pdfStream };
+        byte[] zipBytes = await mediator.Send(command, cancellationToken);
+
+        // Return ZIP file
+        return File(zipBytes, "application/zip", "attachments.zip");
+    }
 }
 
 /// <summary>
@@ -95,3 +170,15 @@ public record CheckPdfAttachmentsRequest
     /// <example>JVBERi0xLjQKJeLjz9MKMSAwIG9iago8PC9UeXBlL0NhdGFsb2c...</example>
     public string Base64Pdf { get; init; } = string.Empty;
 }
+
+/// <summary>
+/// Request DTO for Base64-encoded PDF attachment extraction
+/// </summary>
+public record ExtractPdfAttachmentsRequest
+{
+    /// <summary>
+    /// PDF document encoded as Base64 string
+    /// </summary>
+    /// <example>JVBERi0xLjQKJeLjz9MKMSAwIG9iago8PC9UeXBlL0NhdGFsb2c...</example>
+    public required string Base64Pdf { get; init; }
+}