fix: Use stream-based file handling and serve actual test images

[Copilot]

Addresses memory inefficiency when attaching files with explicit ContentType and improves test accuracy for URL-based image attachments.

Changes

Stream-based file attachment handling

• Replaced File.ReadAllBytesAsync with File.OpenRead streams when creating attachments/linked resources with explicit ContentType
• Construct MimePart directly with MimeContent(stream) to avoid loading entire files into memory
• Added try-catch blocks to dispose streams if MimePart construction fails

Test server improvements

• Download server now serves actual image files (TestImage.png) from TestData directory for linked resource tests
• Maintains backward compatibility by serving synthetic "TestDownload" content for text file tests
• Implemented directory traversal search to locate TestData path robustly across test execution contexts

Code quality

• Extracted magic numbers to constants (MaxDirectoryTraversalLevels)
• Simplified image extension detection using HashSet<string> with case-insensitive comparison
• Fixed null handling for Path.GetDirectoryName() when reaching filesystem root

Example

Before (loads entire file into memory):

var data = await File.ReadAllBytesAsync(attachment.FilePath);
var linkedResource = bodyBuilder.LinkedResources.Add(fileName, data, contentType);

After (streams file content):

var fileStream = File.OpenRead(attachment.FilePath);
try {
    var linkedPart = new MimePart(contentType) {
        Content = new MimeContent(fileStream),
        // ... other properties
    };
    bodyBuilder.LinkedResources.Add(linkedPart);
} catch {
    fileStream?.Dispose();
    throw;
}

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[Copilot]

Pull request overview

This pull request improves memory efficiency when attaching files with explicit ContentType by using stream-based file handling, and enhances test accuracy by serving actual image files from the TestData directory instead of synthetic content.

Key changes:

• Replaced File.ReadAllBytesAsync with File.OpenRead streams for file-based attachments and linked resources with explicit ContentType
• Enhanced test download server to locate and serve actual image files from TestData directory
• Extracted magic numbers to named constants for improved code maintainability

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

File	Description
src/AspNetCore.MailKitMailer/Data/MailClient.cs	Implements stream-based file handling for attachments and linked resources with explicit ContentType to avoid loading entire files into memory
tests/AspNetCore.MailKitMailerIntegrationTests/Abstracts/MailTestAbstracts.cs	Adds directory traversal logic to locate and serve actual test images, with constants for configuration and case-insensitive image extension detection

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The manual stream disposal in the catch block is problematic because MimeContent takes ownership of the stream and will dispose it. If the MimePart constructor or property setters succeed but an exception occurs later (e.g., when adding to Attachments), the stream will be disposed twice - once by MimeContent and once in the catch block. Use a using statement instead to ensure proper disposal only when MimePart construction fails, or rely on MimeContent to dispose the stream in all cases.

[Copilot]

The ContentId assignment has changed behavior. The original code assigned attachment.ContentId directly (which is guaranteed to be non-null based on the isLinkedResource check at line 293). The new code adds a null-coalescing operator with a Guid fallback. This is inconsistent with how FileBytes-based linked resources are handled (line 303) and may mask bugs where ContentId is unexpectedly null. Either remove the null-coalescing operator to maintain consistency, or apply the same pattern to line 303.

Suggested change

	ContentId = attachment.ContentId ?? Guid.NewGuid().ToString()
	ContentId = attachment.ContentId

[Copilot]

The fallback behavior creates a text file with "TestDownload" content even when an image extension is requested but the actual image file is not found. This could cause tests to pass with incorrect data (serving text content with an image extension). Consider either returning a 404 status code when an expected image file is not found, or logging a warning to make debugging easier when the TestData directory or image file cannot be located.

[Copilot]

The manual stream disposal in the catch block is problematic because MimeContent takes ownership of the stream and will dispose it. If the MimePart constructor or property setters succeed but an exception occurs later (e.g., when adding to LinkedResources), the stream will be disposed twice - once by MimeContent and once in the catch block. Use a using statement instead to ensure proper disposal only when MimePart construction fails, or rely on MimeContent to dispose the stream in all cases.