AppStd/DocumentService
8
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Enhance PDF attachment detection and counting

Browse Source 
Refactored `DevExpressPdfProcessor` to improve attachment detection:
- Changed `DetectEmbeddedFiles` return type to a tuple for better
  handling of attachment presence and count.
- Enhanced logic to parse `/Names` arrays and count object references
  for accurate attachment detection.
- Implemented robust search for `/EmbeddedFiles` to handle multiple
  occurrences and ensure proper context validation.

Updated PHASENPLAN.md and ROADMAP.md to reflect these changes, including
the addition of fixes for attachment detection and counting logic.

Added new tests in `DevExpressPdfProcessorTests`:
- Verified detection of multiple attachments and accurate counts.
- Ensured no crashes when processing PDFs with `/EmbeddedFiles`.

Included a new test resource (`pdfWithMoreThanOneAttachment.pdf`) for
validating multiple attachment scenarios.
This commit is contained in:
OlgunR
2026-06-25 17:17:32 +02:00
parent 1b38d5a729
commit 18e956c2cf
6 changed files with 82 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files

73
DocumentOperator.Infrastructure/Services/PdfProcessing/DevExpressPdfProcessor.cs
View File

@@ -43,10 +43,8 @@ public class DevExpressPdfProcessor : IPdfProcessor
// Attachments (embedded files)
// DevExpress PdfDocument API doesn't expose EmbeddedFiles directly.
// We use a simple PDF raw data scan for "/EmbeddedFiles" keyword.
// This is a pragmatic approach until Feature 2 (ExtractAttachments) is implemented.
bool hasAttachments = DetectEmbeddedFiles(pdfBytes);
int attachmentCount = hasAttachments ? -1 : 0; // -1 = "has attachments, count unknown"
// We scan PDF raw data for "/EmbeddedFiles" and parse the name tree to get count.
var (hasAttachments, attachmentCount) = DetectEmbeddedFiles(pdfBytes);
// 4. Create and return PdfMetadata Value Object (fully qualified name!)
return new DocumentOperator.Domain.Models.ValueObjects.PdfMetadata(
@@ -67,27 +65,70 @@ public class DevExpressPdfProcessor : IPdfProcessor
}
/// <summary>
/// Detects embedded files in PDF by scanning raw PDF data for /EmbeddedFiles keyword.
/// Detects embedded files in PDF by scanning raw PDF data for /EmbeddedFiles keyword
/// and parsing the name tree to count attachments.
/// This is a pragmatic approach as DevExpress PdfDocument API doesn't expose EmbeddedFiles directly.
/// </summary>
/// <param name="pdfBytes">PDF raw bytes</param>
/// <returns>True if PDF contains /EmbeddedFiles keyword in proper context, false otherwise</returns>
private static bool DetectEmbeddedFiles(byte[] pdfBytes)
/// <returns>Tuple: (hasAttachments, attachmentCount)</returns>
private static (bool hasAttachments, int attachmentCount) DetectEmbeddedFiles(byte[] pdfBytes)
{
// PDF embedded files are declared in the document catalog:
// /Names << /EmbeddedFiles << /Names [...] >> >>
// We search for the pattern "/Names" followed by "/EmbeddedFiles"
// The /Names array contains pairs: [name1, filespec1, name2, filespec2, ...]
string pdfText = System.Text.Encoding.ASCII.GetString(pdfBytes);
// Look for the specific PDF dictionary pattern: /Names and /EmbeddedFiles
// This is more precise than just searching for /EmbeddedFiles alone
int namesIndex = pdfText.IndexOf("/Names", StringComparison.Ordinal);
if (namesIndex == -1)
return false;
// Search for /EmbeddedFiles in the context of /Names dictionary
// Must appear after a /Names keyword to be valid
int searchStart = 0;
// Check if /EmbeddedFiles appears after /Names within reasonable distance (< 1000 chars)
int embeddedFilesIndex = pdfText.IndexOf("/EmbeddedFiles", namesIndex, Math.Min(1000, pdfText.Length - namesIndex), StringComparison.Ordinal);
return embeddedFilesIndex > namesIndex;
while (true)
{
// Find next occurrence of /EmbeddedFiles
int embeddedFilesIndex = pdfText.IndexOf("/EmbeddedFiles", searchStart, StringComparison.Ordinal);
if (embeddedFilesIndex == -1)
return (false, 0); // Not found
// Check if there's a /Names keyword BEFORE this /EmbeddedFiles
// within a reasonable distance (e.g., within the same PDF object, max 5000 chars back)
int contextStart = Math.Max(0, embeddedFilesIndex - 5000);
string contextBefore = pdfText.Substring(contextStart, embeddedFilesIndex - contextStart);
// Look for /Names in the context before /EmbeddedFiles
int lastNamesIndex = contextBefore.LastIndexOf("/Names", StringComparison.Ordinal);
if (lastNamesIndex != -1)
{
// Found /Names before /EmbeddedFiles - this is likely a valid embedded files declaration
// Now try to parse the /Names array
int namesArrayStart = pdfText.IndexOf("/Names", embeddedFilesIndex, StringComparison.Ordinal);
if (namesArrayStart == -1)
return (true, 0); // Has EmbeddedFiles but can't count
int arrayStart = pdfText.IndexOf('[', namesArrayStart);
if (arrayStart == -1)
return (true, 0); // Has EmbeddedFiles but can't count
int arrayEnd = pdfText.IndexOf(']', arrayStart);
if (arrayEnd == -1)
return (true, 0); // Has EmbeddedFiles but can't count
// Extract array content and count entries
string arrayContent = pdfText.Substring(arrayStart + 1, arrayEnd - arrayStart - 1);
// Count object references in array
// The /Names array contains pairs: (filename) objectReference (filename) objectReference ...
// Each object reference (pattern: "123 0 R") points to one embedded file
// So the number of object references = number of attachments
int objectCount = System.Text.RegularExpressions.Regex.Matches(arrayContent, @"\d+ \d+ R").Count;
return (true, Math.Max(1, objectCount)); // At least 1 if EmbeddedFiles found
}
// This /EmbeddedFiles was not in the right context, search for next occurrence
searchStart = embeddedFilesIndex + 1;
}
}
}