ppgen

A passphrase generation module and command.

Usage

The simplest usage is:

ppgen <length>

Example:

% ppgen 6
solanum stokehold cowpea firmisternous cockbird dionymal

(We'll use the same words over again to emphasize the differences between the various invocations.)

Checks

One check is currently supported: that of least entropy.

Least entropy. To ensure that the generated passphrase contains at least a minimum bits of entropy:

-E<min-allowed-entropy>
--least-entropy=<min-allowed-entropy>

The length argument is nevertheless required. Ppgen will simply check that the entropy bound is satisfied, given the provided length and dictionary size. If the bound is not satisfied, ppgen will terminate with an error and will not output any passphrase.

Examples:

# Assuming a /usr/share/dict/words of 235886 words...
% wc /usr/share/dict/words
  235886  235886 2493109 /usr/share/dict/words

% ppgen -E90 5
Error: insufficient entropy (89.238590 < 90.000000): generate a longer passphrase or use a bigger dictionary

% ppgen -E90 6
solanum stokehold cowpea firmisternous cockbird dionymal

Transformations

Transformations alter the passphrase, deterministically or randomly.

Capitalize. To capitalize the first letter of the first word:

-C
--capitalize

Example:

% ppgen -C 6
Solanum stokehold cowpea firmisternous cockbird dionymal

Randomize. To randomize one character in a random position, specifying from which charset it is to be re-drawn:

-R<charset>
--randomize=<charset>

The <charset> argument is a combination of:

• one optional regex-like charset, e.g. [0-9A-F], surrounded by

• unions of the predefined charsets

  Charset	Tag
  decimal digits	d
  ASCII upper-case letters	u
  ASCII lower-case letters	l
  ASCII symbols	s

  expressed as concatenations of their tags, e.g. duls.

Examples:

Randomize with one printable, non-blank, ASCII character.

% ppgen -Rduls 6
solanum stokehold cowpea firmisternous cockbird dionymFl
                                                      ^

(The indicator is added for the sake of the reader.)

Randomize with two printable, non-blank, ASCII characters.

% ppgen -Rduls -Rduls 6
s5lanum stokehold cowpea firmisternous c2ckbird dionymal
 ^                                      ^

(They both happened to be digits.)

Randomize with a digit, an upper-case letter, and a symbol.

% ppgen -Rd -Ru -Rs 6
solanum stokehold cowpea firmiUternous cockbird di/ny9al
                              ^                   ^  ^

This is a typical password requirement.

Randomize with the characters ".,?!@_-#" (and only these).

% ppgen '-R[-.,?!@_#]' 6
solanum stokehold cowpea firmisternous cockbird !ionymal
                                                ^

Again, this is a typical requirement (if utterly idiotic in its arbitrary strictness).

Note two things. We single-quote the argument to make it opaque to the shell. Otherwise, we might get errors like:

-bash: !@_#]: event not found

Or worse, unexpected behaviour.

We also put the dash, -, first in the charset, to avoid confusion as to whether it carries the special meaning of range operator, as e.g. in [A-Z].

To be precise, the syntax for a regex-like charset is:

<enumeration> -> "[" (<range> | <character>) ... "]"
<range>       -> <character> "-" <character>
<character>   -> any character that the shell lets through

So <character> can also be a dash or a square bracket. How?

• Ppgen tries to match zero or more adjacent <range>s, left to right, optionally intermixed with <character>s.

  In general, a dash appearing where it cannot be interpreted as an operator stands for itself.

  So, for example, [----] represents, literally, the characters from - to -, plus the - character; while [-a-z] and [a-z-] both represent the character - plus the ASCII lower-case letters.

• Given a <charset> expression, <enumeration>s are matched greedily, so at most one of them can appear per expression (the upside being that they can contain square brackets without any escaping).

Translate. To apply deterministic substitutions:

-T<xs>:<ys>
--translate=<xs>:<ys>

Each character of <xs> will be replaced with the corresponding character of <ys>. Surplus characters, which do not have an image in <ys>, will be deleted.

Examples:

% ppgen -Tabc:XYZ 6
solXnum stokehold ZowpeX firmisternous ZoZkYird dionymXl
   ^              ^    ^               ^ ^ ^          ^

% ppgen -Talet:@137 6
so1@num s7ok3ho1d cowp3@ firmis73rnous cockbird dionym@1
  ^^     ^  ^  ^      ^^       ^^                     ^^

% ppgen -Thate:love -Twar: 6
solnum svokelold cope fimisvenous cockbid dionyml
   a    ^   ^      w a  r  ^ r          r       a

(Missing characters have been indicated under their old word-wise position.)

Shorten words. To take just the first <n> characters of each word:

-W<n>
--word-length=<n>

Example:

% ppgen -W4 6
sola stok cowp firm cock dion

Presentation

To use a given character (or string) <sep> as word separator:

-s<sep>
--separator=<sep>

Example:

% ppgen -s- 6
solanum-stokehold-cowpea-firmisternous-cockbird-dionymal

Dictionary

To choose a different dictionary file:

-f<dictionary>
--file=<dictionary>

The <dictionary> argument should point to a file containing one word per line. The file will be processed on-line, hence it can be a non-seekable stream, and have any finite length.

Installation

With pipx:

pipx install git+https://github.com/giucal/ppgen.py

With pip:

pip3 install --user git+https://github.com/giucal/ppgen.py