Generate secure multiword passwords/passphrases, inspired by XKCD
A flexible and scriptable password generator which generates strong passphrases, inspired by XKCD 936:
$ xkcdpass > correct horse battery staple
xkcdpass can be easily installed using pip:
pip install xkcdpass
python setup.py install
The latest development version can be found on github: https://github.com/redacted/XKCD-password-generator
Contributions welcome and gratefully appreciated!
Python 2 (version 2.7 or later), or Python 3 (version 3.2 or later). Running module unit tests on Python 2 requires mock to be installed.
xkcdpass can be called with no arguments:
$ xkcdpass > pinball previous deprive militancy bereaved numeric
which returns a single password, using the default dictionary and default settings. Or you can mix whatever arguments you want:
$ xkcdpass --count=5 --acrostic='chaos' --delimiter='|' --min=5 --max=6 --valid-chars='[a-z]' > collar|highly|asset|ovoid|sultan > caper|hangup|addle|oboist|scroll > couple|honcho|abbot|obtain|simple > cutler|hotly|aortae|outset|stool > cradle|helot|axial|ordure|shale
- --count=5 5 passwords to choose from
- --acrostic='chaos' the first letters of which spell ‘chaos’
- --delimiter='|' joined using ‘|’
- --min=5 --max=6 with words between 5 and 6 characters long
- --valid-chars='[a-z]' using only lower-case letters (via regex).
A concise overview of the available xkcdpass options can be accessed via:
xkcdpass --help Usage: xkcdpass [options] Options: -h, --help show this help message and exit -w WORDFILE, --wordfile=WORDFILE Specify that the file WORDFILE contains the list of valid words from which to generate passphrases. Provided wordfiles: eff-long (default), eff-short, eff-special, legacy, spa-mich (Spanish), fin-kotus (Finnish) ita-wiki (Italian), ger-anlx (German) --min=MIN_LENGTH Minimum length of words to make password --max=MAX_LENGTH Maximum length of words to make password -n NUMWORDS, --numwords=NUMWORDS Number of words to make password -i, --interactive Interactively select a password -v VALID_CHARS, --valid-chars=VALID_CHARS Valid chars, using regexp style (e.g. '[a-z]') -V, --verbose Report various metrics for given options, including word list entropy -a ACROSTIC, --acrostic=ACROSTIC Acrostic to constrain word choices -c COUNT, --count=COUNT number of passwords to generate -d DELIM, --delimiter=DELIM separator character between words -s SEP, --separator SEP Separate generated passphrases with SEP. -C CASE, --case CASE Choose the method for setting the case of each word in the passphrase. Choices: ['alternating', 'upper', 'lower', 'random', 'first'] (default: 'lower'). --allow-weak-rng Allow fallback to weak RNG if the system does not support cryptographically secure RNG. Only use this if you know what you are doing.
Several word lists are provided with the package. The default,
eff-long, was specifically designed by the EFF for passphrase generation and is licensed under CC BY 3.0. As it was originally intended for use with Diceware ensure that the number of words in your passphrase is at least six when using it. Two shorter variants of that list,
eff-special, are also included. Please refer to the EFF documentation linked above for more information.
The original word list from
xkcdpass versions earlier than 1.10.0 is also provided as a convenience, and is available under
legacy. This word list is derived mechanically from 12Dicts by Alan Beale. It is the understanding of the author of xkcdpass that purely mechanical transformation does not imbue copyright in the resulting work. The documentation for the 12Dicts project at
http://wordlist.aspell.net/12dicts/ contains the following dedication:
The 12dicts lists were compiled by Alan Beale. I explicitly release them to the public domain, but request acknowledgment of their use.
Note that the generator can be used with any word file of the correct format: a file containing one ‘word’ per line.
- Spanish: a modifed version of archive.umich.edu in the
/linguisticsdirectory. It includes ~80k words. Less than 5 char. and latin-like words were deleted using regex. This list is public domain, see here.
- Finnish: a modified version of the Institute for the Languages of Finland XML word list. Profanities and expressions containing spaces were removed using regex. The resulting list contains ~93k words. The list is published under GNU LGPL, EUPL 1.1 and CC-BY 3.0 licenses.
- Italian: generated from dumps of the Italian-language Wikipedia, which is released under the Creative Commons Attribution-Share-Alike 3.0 licence.
- German: based on this GPL v3 list. Single and double character words have been removed.
Additional language word lists are always welcome!
Using xkcdpass as an imported module
The built-in functionality of xkcdpass can be extended by importing the module into python scripts. An example of this usage is provided in example_import.py, which randomly capitalises the letters in a generated password.
A simple use of import:
from xkcdpass import xkcd_password as xp # create a wordlist from the default wordfile # use words between 5 and 8 letters long wordfile = xp.locate_wordfile() mywords = xp.generate_wordlist(wordfile=wordfile, min_length=5, max_length=8) # create a password with the acrostic "face" print(xp.generate_xkcdpassword(mywords, acrostic="face"))
When used as an imported module,
generate_wordlist() takes the following args (defaults shown):
wordfile=None, min_length=5, max_length=9, valid_chars='.'
wordlist, numwords=6, interactive=False, acrostic=False, delimiter=" "
Insecure random number generators
xkcdpass uses crytographically strong random number generators where possible (provided by
random.SystemRandom() on most modern operating systems). From version 1.7.0 falling back to an insecure RNG must be explicitly enabled, either by using a new command line variable before running the script:
or setting the appropriate environment variable:
- 1.16.5 Adds title case option for
- 1.16.4 Improve unit tests, fixes broken test on python 2
- 1.16.3 Correct links for German worldist, updated docs to include the list
- 1.16.2 Fix exception on UTF8 open with python 2.x
- 1.16.1 Fix encoding issue on Windows
- 1.16.0 Case of words in passphrase can now be set using
- 1.15.1 Added more information about supported languages
- 1.15.0 Added
--separatorargument, German wordlist (GPL 3.0, thanks to @anlx-sw)
- 1.14.3 Refactor password generator, fixes for hardcoded python version in test
- 1.14.2 Improve unit test discovery, remove deprecation warnings
- 1.14.1 Fix wordlist order in
- 1.14.0 Added Finnish and Italian language support (thanks to Jussi Tiira and Lorenzo Mureu respectively)
- 1.13.0 Added Spanish language wordfile (thanks to Javier Meija)
- 1.12.0 Handle maximum word length < minimum case by setting max = min
- 1.11.1 Fix bug in entropy calc
- 1.11.0 Rewrite verbose report to take acrostics etc into account
- 1.10.0 Switch to EFF wordlist as default (note: decrease in entropy of default length passwords to 77 bits, still at EFF recommendations)
This is free software: you may copy, modify, and/or distribute this work under the terms of the BSD 3-Clause license. See the file LICENSE.BSD for details.