MailParser-MIT

MailParser-MIT is an asynchronous and non-blocking parser for node.js to parse mime encoded e-mail messages. Handles even large attachments with ease - attachments can be parsed in chunks and streamed if needed.

Community version

This module is a fork of Andris Reinman's MailParser module, which recently switched licenses during a large update. I will continue to update and maintain this fork and keep it MIT licensed.

For easier maintenance, I have also started to move some dependency code into this project from Andris's other projects. So far I have incorporated code from the latest MIT-licensed versions of libmime and libpq.

Because I've accepted the part of me that wants to fix things that aren't broken, I plan to:

• Convert to TypeScript
• Separate into a cascade of Stream processors, for more flexibility
• Add support for RTF bodies, including RTF-encoded HTML
• Add more test cases, especially for exotic encodings
• Continue support for iconv-lite and iconv

Now accepting and test cases and pull requests!

Do you have emails that MailParser-MIT fails to process? Please open an issue and I will take a look. If the test emails are sensitive, you can email them to me at [email protected], and I will try to generate a simple MIT-licensed test case for inclusion in this repository.

Overview

MailParser-MIT parses raw source of e-mail messages into a structured object.

No need to worry about charsets or decoding quoted-printable or base64 data, MailParser-MIT does all of it for you. All the textual output from MailParser-MIT (subject line, addressee names, message body) is always UTF-8.

For a 25MB e-mail it takes less than a second to parse if attachments are not streamed but buffered and about 3-4 seconds if they are streamed. Expect high RAM usage though if you do not stream the attachments.

If you want to send e-mail instead of parsing it, check out my other module Nodemailer.

ICONV NOTICE

Since v0.4 node-iconv is not included by default as a dependency. If you need to support encodings not covered by iconv-lite you should add iconv as a dependency to your own project so mailparser-mit could pick it up.

Installation

npm install mailparser-mit

Usage

Require MailParser module

var MailParser = require("mailparser-mit").MailParser;

Create a new MailParser object

var mailparser = new MailParser([options]);

Options parameter is an object with the following properties:

• debug - if set to true print all incoming lines to console
• streamAttachments - if set to true, stream attachments instead of including them
• unescapeSMTP - if set to true replace double dots in the beginning of the file
• defaultCharset - the default charset for text/plain and text/html content, if not set reverts to Latin-1
• showAttachmentLinks - if set to true, show inlined attachment links <a href="cid:...">filename</a>

MailParser object is a writable Stream - you can pipe directly files to it or you can send chunks with mailparser.write

When the headers have received, "headers" is emitted. The headers have not been pre-processed (except that mime words have been converted to UTF-8 text).

mailparser.on("headers", function(headers){
    console.log(headers.received);
});

When the parsing ends an 'end' event is emitted which has an object with parsed e-mail structure as a parameter.

mailparser.on("end", function(mail){
    mail; // object structure for parsed e-mail
});

Parsed mail object

• headers - unprocessed headers in the form of - {key: value} - if there were multiple fields with the same key then the value is an array
• from - an array of parsed From addresses - [{address:'[email protected]',name:'Sender Name'}] (should be only one though)
• to - an array of parsed To addresses
• cc - an array of parsed Cc addresses
• bcc - an array of parsed 'Bcc' addresses
• subject - the subject line
• references - an array of reference message id values (not set if no reference values present)
• inReplyTo - an array of In-Reply-To message id values (not set if no in-reply-to values present)
• priority - priority of the e-mail, always one of the following: normal (default), high, low
• text - text body
• html - html body
• date - date field as a Date() object. If date could not be resolved or is not found this field is not set. Check the original date string from headers.date
• attachments - an array of attachments

Decode a simple e-mail

This example decodes an e-mail from a string

var MailParser = require("mailparser-mit").MailParser;
var mailparser = new MailParser();

var email = "From: 'Sender Name' <[email protected]>\r\n"+
            "To: 'Receiver Name' <[email protected]>\r\n"+
            "Subject: Hello world!\r\n"+
            "\r\n"+
            "How are you today?";

// setup an event listener when the parsing finishes
mailparser.on("end", function(mail_object){
    console.log("From:", mail_object.from); //[{address:'[email protected]',name:'Sender Name'}]
    console.log("Subject:", mail_object.subject); // Hello world!
    console.log("Text body:", mail_object.text); // How are you today?
});

// send the email source to the parser
mailparser.write(email);
mailparser.end();

Pipe file to MailParser

This example pipes a readableStream file to MailParser

var MailParser = require("mailparser").MailParser;
var mailparser = new MailParser();
var fs = require("fs");

mailparser.on("end", function(mail_object){
    console.log("Subject:", mail_object.subject);
});

fs.createReadStream("email.eml").pipe(mailparser);

Attachments

By default any attachment found from the e-mail will be included fully in the final mail structure object as Buffer objects. With large files this might not be desirable so optionally it is possible to redirect the attachments to a Stream and keep only the metadata about the file in the mail structure.

mailparser.on("end", function(mail_object){
    mail_object.attachments.forEach(function(attachment){
        console.log(attachment.fileName);
    });
});

Default behavior

By default attachments will be included in the attachment objects as Buffers.

attachments = [{
    contentType: 'image/png',
    fileName: 'image.png',
    contentDisposition: 'attachment',
    contentId: '5.1321281380971@localhost',
    transferEncoding: 'base64',
    length: 126,
    generatedFileName: 'image.png',
    checksum: 'e4cef4c6e26037bcf8166905207ea09b',
    content: <Buffer ...>
}];

The property generatedFileName is usually the same as fileName but if several different attachments with the same name exist or there is no fileName set, an unique name is generated.

Property content is always a Buffer object (or SlowBuffer on some occasions)

Attachment streaming

Attachment streaming can be used when providing an optional options parameter to the MailParser constructor.

var mp = new MailParser({
    streamAttachments: true
}

This way there will be no content property on final attachment objects (but the other fields will remain).

To catch the streams you should listen for attachment events on the MailParser object. The parameter provided includes file information (contentType, fileName, contentId) and a readable Stream object stream.

var mp = new MailParser({
    streamAttachments: true
}

mp.on("attachment", function(attachment, mail){
    var output = fs.createWriteStream(attachment.generatedFileName);
    attachment.stream.pipe(output);
});

generatedFileName is unique for the parsed mail - if several attachments with the same name exist, generatedFileName is updated accordingly. Also there might not be fileName parameter at all, so it is better to rely on generatedFileName.

Testing attachment integrity

Attachment objects include length property which is the length of the attachment in bytes and checksum property which is a md5 hash of the file.

Running tests

Install MailParser-MIT with dev dependencies

npm install --dev mailparser-mit

And then run

npm test mailparser-mit

There aren't many tests yet but basics should be covered.

Issues

S/MIME

Currently it is not possible to verify signed content as the incoming text is split to lines when parsing and line ending characters are not preserved. One can assume it is always \r\n but this might not be always the case.

Seeking

Due to the line based parsing it is also not possible to explicitly state the beginning and ending bytes of the attachments for later source seeking. Node.js doesn't support the concept of seeking very well anyway.

License

MIT