Recursive AI

Write EML messages to MBOX in C#

OneMail is a free, open-source email library for .NET from Recursive AI. MIT or Apache-2.0, no native dependencies.

Pack a.eml and b.eml from the zip into one mailbox.

Terminal
dotnet add package RecursiveAI.OneMail

Requires .NET 10 (net10.0); .NET 9 and earlier aren’t supported.

C#
using RecursiveAI.OneMail.Stores.Mbox;
using var output = File.Create("mail.mbox");
var writer = new MboxWriter(output);
writer.Append(File.ReadAllBytes("a.eml"));
writer.Append(File.ReadAllBytes("b.eml"));

Unzip the example, then dotnet run.

Writes or replaces mail.mbox with both messages.

The full program

The example contains two fictional EML files. a.eml contains:

Text
From: Ada <ada@example.test>
Date: Thu, 19 Nov 2026 09:00:00 +0000
Subject: Arrival
 
From the platform, turn left.

b.eml contains:

Text
From: Ben <ben@example.test>
Date: Thu, 19 Nov 2026 09:05:00 +0000
Subject: Re: Arrival
 
See you there.

The program sets fixed envelope senders and UTC dates, closes the output, then reopens it to check both subjects and the first message’s text.

C#
using RecursiveAI.OneMail;
using RecursiveAI.OneMail.Stores.Mbox;
 
using (var output = File.Create("mail.mbox"))
{
    var writer = new MboxWriter(output);
    var date = new DateTimeOffset(2026, 11, 19, 9, 0, 0, TimeSpan.Zero);
    writer.Append(File.ReadAllBytes("a.eml"), "ada@example.test", date);
    writer.Append(File.ReadAllBytes("b.eml"), "ben@example.test", date.AddMinutes(5));
}
 
using var mailbox = Mail.Open("mail.mbox");
var messages = mailbox.Enumerate().ToArray();
if (messages.Length != 2 || messages[0].Subject != "Arrival" || messages[1].Subject != "Re: Arrival")
    throw new InvalidDataException("Unexpected messages in the mailbox.");
 
const string expectedLine = "From the platform, turn left.";
var body = messages[0].TextBody?.ReadText();
if (body?.TrimEnd('\r', '\n') != expectedLine)
    throw new InvalidDataException("The From line did not round-trip.");
 
foreach (var message in messages)
    Console.WriteLine($"Subject: {message.Subject}");
Console.WriteLine($"Body: {expectedLine}");

dotnet run prints:

Text
Subject: Arrival
Subject: Re: Arrival
Body: From the platform, turn left.

If an input file is missing, the read fails after File.Create has opened the output. Use a new path if you need to keep an existing archive.

Variations

Append to an existing archive. Run the snippet at the top first to create mail.mbox. Then replace that entire snippet with the one below. Both use mboxrd, a variant that escapes body lines beginning with From so they aren’t mistaken for message separators.

C#
using RecursiveAI.OneMail.Stores.Mbox;
using var stream = File.Open("mail.mbox", FileMode.Open,
    FileAccess.ReadWrite);
var options = new MboxWriterOptions { AppendToExisting = true };
var append = new MboxWriter(stream, options);
append.Append(File.ReadAllBytes("a.eml"));

This adds a third record, another copy of a.eml. AppendToExisting lets the writer add a missing separator before the new record.

Read an MBOX file · Parse an EML file


The short answer

Create an output stream and call MboxWriter.Append once for each .eml file. Pass the entire file, including its headers and body, not just the body. The writer adds a mailbox separator and handles body lines that could be mistaken for separators. The input file order becomes the message order in the archive.

How do I convert multiple EML files into one MBOX file?

An EML file holds one email. MBOX puts several messages into a single stream and marks each one with an envelope line. If you concatenate two EML files without adding these lines, a reader has no MBOX record boundaries. The two Append calls in the snippet at the top do the packing; they don’t require you to parse or recreate each message first.

The writer’s envelope is a line like From ada@example.test Thu Nov 19 09:00:00 2026. It isn’t the From: header inside the email. The full program gives each call its own sender and UTC date so the envelope is repeatable. If you omit those arguments, the writer uses MAILER-DAEMON and the current time, not the sender and date in the message’s headers. See OneMail’s mailbox writing rules.

How do I create an MBOX file without splitting a message at a From line?

In mboxrd, the writer’s default, the first message’s From the platform, turn left. becomes >From the platform, turn left. on disk. Reading the archive restores the original line. If the original had already started with >From, the writer would add another >; the reader removes one, not all of them. That distinction lets the reader recover the original prefix. The variant table also explains the variants that use Content-Length instead of relying on quoting alone. The writer refuses headers larger than 1 MiB for those length-based variants.

Use the default for this example. If the program that will import the result requires a specific variant, check its requirements before writing or extending an archive. Set the writer to match that variant; the append option adjusts the separator at the end, not the format of the records already stored there.

How do I check an MBOX file after creating it?

Close the output first and reopen it as a mailbox, not as a single email. The full program checks for exactly two messages, compares each subject in order, and reads back the body line that starts with From. Checking only the file’s existence would miss both a broken boundary and a stray > showing up in the decoded text.

The archive can change how the message is stored, including adding a final line ending when needed. If you must keep an exact copy of each source file, keep the original EML files separately.

All onemail recipes