Important

The default branch of this repository is 5-stable (Sisimai 5) since 2nd February 2024. If you want to clone the old version, see the 4-stable¹ branch instead. We have moved away from using both the main and master branches in our development process.

Caution

Sisimai versions 4.25.14p11 and earlier contain a regular expression vulnerability ReDoS: CVE-2022-4891. If you are using one of these versions, please upgrade to v4.25.14p12 or later.

Warning

Sisimai 5 requires Ruby 2.4 or later. Check the version of Ruby in your system before installing/upgrading by ruby -v command.

Note

Sisimai is a Ruby Gem or Perl module, but it can be used in any environment where JSON can be read, such as PHP, Python, Go, and Rust. By obtaining the analysis results, it is very useful for understanding the bounce occurrence status.

README-JA(日本語)
What is Sisimai
- The key features of sisimai
- Command line demo
Setting Up Sisimai
- System requirements
- Install
  - From RubyGems
  - From GitHub
Usage
Differences between Sisimai 4 and Sisimai 5
Contributing
- Bug report
- Emails could not be decoded
Other Information
- Related sites
- See also
Author
Copyright
License

What is Sisimai

Sisimai is a library that decodes complex and diverse bounce emails and outputs the results of the delivery failure, such as the reason for the bounce and the recipient email address, in structured data. It is also possible to output in JSON format. The Ruby version of Sisimai is ported from the Perl version of Sisimai at github.com/sisimai/p5-sisimai.

The key features of Sisimai

Decode email bounces to structured data
- Sisimai provides detailed insights into bounce emails by extracting 24 key data points.²
  - Essential information: timestamp, origin
  - Sender information: addresser, senderdomain,
  - Recipient information: recipient, destination, alias
  - Delivery information: action, replycode,action, replycode, deliverystatus
  - Bounce details: reason, diagnosticcode, diagnostictype, feedbacktype, hardbounce
  - Message details: subject, messageid, listid,
  - Additional information: smtpagent, timezoneoffset, lhost, rhost, token, catch
- Output formats
  - Ruby (Hash, Array)
  - JSON
    - (by using oj gem at CRuby)
    - (by using jrjackson gem at JRuby)
  - YAML (yaml gem required)
- Easy to Install, Use.
  - gem install
  - git clone && make
- High Precision of Analysis
  - Support 73 MTAs/MDAs/ESPs
  - Support Feedback Loop Message(ARF)
  - Can detect 34 bounce reasons

Command line demo

The following screen shows a demonstration of dump method of Sisimai 5 at the command line using Ruby(rb-sisimai) and jq command.

Setting Up Sisimai

System requirements

More details about system requirements are available at Sisimai | Getting Started page.

Install

From RubyGems

$ sudo gem install sisimai
Fetching: sisimai-5.1.0.gem (100%)
Successfully installed sisimai-5.1.0
Parsing documentation for sisimai-5.1.0
Installing ri documentation for sisimai-5.1.0
Done installing documentation for sisimai after 6 seconds
1 gem installed

From GitHub

Warning

Sisimai 5 requires Ruby 2.4 or later. Check the version of Ruby in your system before installing/upgrading by ruby -v command.

% ruby -v
ruby 3.3.0 (2023-12-25 revision 5124f9ac75) [x86_64-darwin21]

$ cd /usr/local/src
$ git clone https://github.com/sisimai/rb-sisimai.git

$ cd ./rb-sisimai
$ sudo make depend install-from-local
gem install bundle rake minitest
...
3 gems installed
if [ -d "/usr/local/jr" ]; then \
		PATH="/usr/local/jr/bin:$PATH" /usr/local/jr/bin/gem install bundle rake minitest; \
	fi
...
3 gems installed
/opt/local/bin/rake install
sisimai 5.1.0 built to pkg/sisimai-5.1.0.gem.
sisimai (5.1.0) installed.
if [ -d "/usr/local/jr" ]; then \
		PATH="/usr/local/jr/bin:$PATH" /usr/local/jr/bin/rake install; \
	fi
sisimai 5.1.0 built to pkg/sisimai-5.1.0-java.gem.
sisimai (5.1.0) installed.

Usage

Basic usage

Sisimai->rise() method provides the feature for getting decoded data as Ruby Hash from bounced email messages as the following. Beginning with v4.25.6, new accessor origin which keeps the path to email file as a data source is available.

#! /usr/bin/env ruby
require 'sisimai'
v = Sisimai.rise('/path/to/mbox')       # or path to Maildir/

# In v4.23.0, the rise() and dump() methods of the Sisimai class can now read the entire bounce
# email as a string, in addition to the PATH to the email file or mailbox.
f = File.open('/path/to/mbox', 'r');    # or path to Maildir/
v = Sisimai.rise(f.read)

# If you also need analysis results that are "delivered" (successfully delivered), please specify
# the "delivered" option to the rise() method as shown below.
v = Sisimai.rise('/path/to/mbox', delivered: true)

# From v5.0.0, Sisimai no longer returns analysis results with a bounce reason of "vacation" by
# default. If you also need analysis results that show a "vacation" reason, please specify the
# "vacation" option to the rise() method as shown in the following code.
v = Sisimai.rise('/path/to/mbox', vacation: true );

if v.is_a? Array
  v.each do |e|
    puts e.class                # Sisimai::Fact
    puts e.recipient.class      # Sisimai::Address
    puts e.timestamp.class      # Sisimai::Time

    puts e.addresser.address    # "michitsuna@example.org" # From
    puts e.recipient.address    # "kijitora@example.jp"    # To
    puts e.recipient.host       # "example.jp"
    puts e.deliverystatus       # "5.1.1"
    puts e.replycode            # "550"
    puts e.reason               # "userunknown"
    puts e.origin               # "/var/spool/bounce/Maildir/new/1740074341.eml"
    puts e.hardbounce           # true

    h = e.damn                  # Convert to HASH
    j = e.dump('json')          # Convert to JSON string
    puts e.dump('json')         # JSON formatted bounce data
  end
end

Convert to JSON

Sisimai.dump() method provides the feature for getting decoded data as JSON string from bounced email messages like the following code:

# Get JSON string from path of a mailbox or a Maildir/
puts Sisimai.dump('/path/to/mbox')  # or path to Maildir/

# dump() method also accepts "delivered" and "vacation" option like the following code:
puts Sisimai.dump('/path/to/mbox', delivered: true, vacation: true)

Callback feature

:c___ (c and three _s, looks like a fishhook) argument of Sisimai.rise and Sisimai.dump is an Array and is a parameter to receive Proc objects for callback feature. The first element of :c___ argument is called at Sisimai::Message.sift for dealing email headers and entire message body. The second element of :c___ argument is called at the end of each email file parsing. The result generated by the callback method is accessible via Sisimai::Fact.catch.

[0] For email headers and the body

Callback method set in the first element of :c___ is called at Sisimai::Message.sift().

#! /usr/bin/env ruby
require 'sisimai'
code = lambda do |args|
  head = args['headers']    # (*Hash)  Email headers
  body = args['message']    # (String) Message body
  adds = { 'x-mailer' => '', 'queue-id' => '' }

  if cv = body.match(/^X-Postfix-Queue-ID:\s*(.+)$/)
    adds['queue-id'] = cv[1]
  end
  r['x-mailer'] = head['x-mailer'] || ''
  return adds
end

data = Sisimai.rise('/path/to/mbox', c___: [code, nil])
json = Sisimai.dump('/path/to/mbox', c___: [code, nil])

puts data[0].catch['x-mailer']  # "Apple Mail (2.1283)"
puts data[0].catch['queue-id']  # "43f4KX6WR7z1xcMG"

For each email file

Callback method set in the second element of :c___ is called at Sisimai.rise method for dealing each email file.

path = '/path/to/maildir'
code = lambda do |args|
  kind = args['kind']   # (String) Sisimai::Mail.kind
  mail = args['mail']   # (String) Entire email message
  path = args['path']   # (String) Sisimai::Mail.path
  fact = args['fact']   # (Array)  List of Sisimai::Fact

  fact.each do |e|
    # Store custom information in the "catch" accessor
    e.catch ||= {}
    e.catch['size'] = mail.size
    e.catch['kind'] = kind.capitalize

    if cv = mail.match(/^Return-Path: (.+)$/)
      # Return-Path: <MAILER-DAEMON>
      e.catch['return-path'] = cv[1]
    end
    e.catch['parsedat'] = Time.new.localtime.to_s

    # Save the original email with an additional "X-Sisimai-Parsed:" header to a different PATH.
    a = sprintf("X-Sisimai-Parsed: %d", fact.size)
    p = sprintf("/path/to/another/directory/sisimai-%s.eml", e.token)
    v = mail.sub(/^(From:.+?)$/, '\1' + "\n" + a)
    f = File.open(p, 'w:UTF-8')
    f.write(v)
    f.close

    # Remove the email file in Maildir/ after decoding
    File.delete(path) if kind == 'maildir'

    # Need to not return a value
  end
end

list = Sisimai.rise(path, c___: [nil, code])

puts list[0].catch['size']          # 2202
puts list[0].catch['kind']          # "Maildir"
puts list[0].catch['return-path']   # "<MAILER-DAEMON>"

More information about the callback feature is available at Sisimai | How To Parse - Callback Page.

One-Liner

% ruby -rsisimai -e 'puts Sisimai.dump($*.shift)' /path/to/mbox

Output example

[
  {
    "destination": "google.example.com",
    "lhost": "gmail-smtp-in.l.google.com",
    "hardbounce": 0,
    "reason": "authfailure",
    "catch": null,
    "addresser": "michitsuna@example.jp",
    "alias": "nekochan@example.co.jp",
    "smtpagent": "Postfix",
    "smtpcommand": "DATA",
    "senderdomain": "example.jp",
    "listid": "",
    "action": "failed",
    "feedbacktype": "",
    "messageid": "hwK7pzjzJtz0RF9Y@relay3.example.com",
    "origin": "./gmail-5.7.26.eml",
    "recipient": "kijitora@google.example.com",
    "rhost": "gmail-smtp-in.l.google.com",
    "subject": "Nyaan",
    "timezoneoffset": "+0900",
    "replycode": 550,
    "token": "84656774898baa90660be3e12fe0526e108d4473",
    "diagnostictype": "SMTP",
    "timestamp": 1650119685,
    "diagnosticcode": "host gmail-smtp-in.l.google.com[64.233.187.27] said: This mail has been blocked because the sender is unauthenticated. Gmail requires all senders to authenticate with either SPF or DKIM. Authentication results: DKIM = did not pass SPF [relay3.example.com] with ip: [192.0.2.22] = did not pass For instructions on setting up authentication, go to https://support.google.com/mail/answer/81126#authentication c2-202200202020202020222222cat.127 - gsmtp (in reply to end of DATA command)",
    "deliverystatus": "5.7.26"
  }
]

Differences between Sisimai 4 and Sisimai 5

The following table show the differences between Sisimai 4.25.16p1 and Sisimai 5. More information about differences are available at Sisimai | Differences page.

Features

Beginning with v5.0.0, Sisimai requires Ruby 2.4.0 or later.

Features	Sisimai 4	Sisimai 5
System requirements (CRuby)	2.1 - 3.3.0	2.4 or later
System requirements (JRuby)	9.0.4.0 - 9.1.17.0	9.2 or later
Callback feature for the original email file	N/A	Available³
The number of MTA/ESP modules	68	73
The number of detectable bounce reasons	29	34
Dependencies (Except Ruby Standard Gems)	1 gem	1 gem
Source lines of code	10,300 lines	11,660 lines
Test frameworks	rspec	minitest
The number of tests in spec/ or test/ directory	311,000 tests	414,000 tests
The number of bounce emails decoded/sec (CRuby)⁴	231 emails	305 emails
License	2 Clause BSD	2 Caluse BSD
Commercial support	Available	Available

Decoding Method

Some decoding method names, class names, parameter names have been changed at Sisimai 5. The details of the decoded data are available at LIBSISIMAI.ORG/EN/DATA

Decoding Method	Sisimai 4	Sisimai 5
Decoding method name	`Sisimai.make`	`Sisimai.rise`
Dumping method name	`Sisimai.dump`	`Sisimai.dump`
Class name of decoded object	`Sisimai::Data`	`Sisimai::Fact`
Parameter name of the callback	`hook`	`:c___`⁵
Method name for checking the hard/soft bounce	`softbounce`	`hardbounce`
Decode a vacation message by default	Yes	No
Sisimai::Message returns an object	Yes	No
MIME decoding class	`Sisimai::MIME`	`Sisimai::RFC2045`
Decoding transcript of SMTP session	No	Yes⁶

MTA/ESP Module Names

Three ESP module names have been changed at Sisimai 5. The list of the all MTA/ESP modules is available at LIBSISIMAI.ORG/EN/ENGINE

`Sisimai::`	Sisimai 4	Sisimai 5
Apple iCloud Mail (added at v5.1.0)	None	`Rhost::Apple`
Microsoft Exchange Online	`Rhost::ExchangeOnline`	`Rhost::Microsoft`
Google Workspace	`Rhost::GoogleApps`	`Rhost::Google`
Tencent	`Rhost::TencentQQ`	`Rhost::Tencent`
Yahoo Mail (added at v5.1.0)	None	`Rhost::YahooInc`
DragonFly Mail Agent (added at v5.1.0)	None	`Lhost::DragonFly`

Bounce Reasons

Five bounce reasons have been added at Sisimai 5. The list of the all bounce reasons sisimai can detect is available at LIBSISIMAI.ORG/EN/REASON

Rejected due to	Sisimai 4	Sisimai 5
sender domain authentication(SPF,DKIM,DMARC)	`SecurityError`	`AuthFailure`
low/bad reputation of the sender hostname/IP addr.	`Blocked`	`BadReputation`
missing PTR/having invalid PTR	`Blocked`	`RequirePTR`
non-compliance with RFC⁷	`SecurityError`	`NotCompliantRFC`
exceeding a rate limit or sending too fast	`SecurityError`	`Speeding`

Contributing

Bug report

Please use the issue tracker to report any bugs.

Emails could not be decoded

Bounce mails which could not be decoded by Sisimai are saved in the repository set-of-emails/to-be-debugged-because/sisimai-cannot-parse-yet. If you have found any bounce email cannot be decoded using Sisimai, please add the email into the directory and send Pull-Request to this repository.

Other Information

Related Sites

@libsisimai | Sisimai on Twitter (@libsisimai)
LIBSISIMAI.ORG | SISIMAI | MAIL ANALYZING INTERFACE | DECODING BOUNCES, BETTER AND FASTER.
Sisimai Blog | blog.libsisimai.org
Facebook Page | facebook.com/libsisimai
GitHub | github.com/sisimai/rb-sisimai
RubyGems.org | rubygems.org/gems/sisimai
Perl verson | Perl version of Sisimai
Fixtures | set-of-emails - Sample emails for "make test"

Author

@azumakuniyuki

Copyright

License

This software is distributed under The BSD 2-Clause License.

Specify -b 4-stable when you clone Sisimai 4 for example, git clone -b 4-stable https://github.com/sisimai/rb-sisimai.git ↩
The callback function allows you to add your own data under the catch accessor. ↩
The 2nd argument of :c___ parameter at Sisimai.rise method ↩
macOS Monterey/1.6GHz Dual-Core Intel Core i5/16GB-RAM/Ruby 3.3.0 ↩
:c___ looks like a fishhook ↩
Sisimai::SMTP::Transcript.rise Method provides the feature ↩
RFC5322 and related RFCs ↩

sisimai/rb-sisimai