A library for parsing and evaluating dice notation

dice, dnd, mit, python, rpg
pip install dice==4.0.0



A Python library and command line tool for parsing and evaluating dice notation.

I consider this library "finished", and don't expect to add any more features to it. Bug and security fixes may still be released, especially if you find any bugs after all this time. If you're interested in other libraries in this space, dyce has a great comparison table of Python dice rolling libraries.



$ roll 3d6

The command line arguments are as follows:

  • -m --min Make all rolls the lowest possible result
  • -M --max Make all rolls the highest possible result
  • -h --help Show this help text
  • -v --verbose Show additional output
  • -V --version Show the package version

If your expression begins with a dash (-), then put a double dash (--) before it to prevent the parser from trying to process it as a command option. Example: roll -- -10d6. Alternatively, use parenthesis: roll (-10d6).

Python API

Invoking from python:

import dice

This returns an Element which is the result of the roll, which can be a list, int, or subclass thereof, depending on the top-level operator.



This library is available as dice on PyPI. Install it with your Python package or dependency manager of choice — if you're installing it as a command-line tool, I recommend pipx.

A recent version of Python 3 (3.8 or above) is required. You can probably run it or easily adapt it for older versions of Python, but I don't support any end-of-life Python versions. Beyond that, the only dependency is the pyparsing library.


The expression works like a simple equation parser with some extra operators.

The following operators are listed in order of precedence. Parentheses may be used to force an alternate order of evaluation.

The dice ([N]dS) operator takes an amount (N) and a number of sides (S), and returns a list of N random numbers between 1 and S. For example: 4d6 may return [6, 3, 2, 4]. Using a % as the second operand is shorthand for rolling a d100, and a using f is shorthand for ±1 fudge dice.

The fudge dice ([N]uS) operator is interchangeable with the dice operator, but makes the dice range from -S to S instead of 1 to S. This includes 0.

A wild dice ([N]wS) roll is special. The last roll in this set is called the "wild die". If this die's roll is the maximum value, the second-highest roll in the set is set to the maximum value. If its roll is the minimum, then both it and the highest roll in the set aer set to zero. Then another die is rolled. If this roll is the minimum value again, then ALL die are set to zero. If a single-sided wild die is rolled, the roll behaves like a normal one.

If N is not specified, it is assumed you want to roll a single die. d6 is equivalent to 1d6.

Rolls can be exploded with the x operator, which adds additional dice to the set for each roll above a given threshold. If a threshold isn't given, it defaults to the maximum possible roll. If the extra dice exceed this threshold, they "explode" again! Safeguards are in place to prevent this from crashing the parser with infinite explosions.

You can make the parser re-roll dice below a certain threshold with the r and rr operators. The single r variety allows the new roll to be below the threshold, whereas the double variety's roll changes the roll range to have a minimum of the threshold. The threshold defaults to the minimum roll.

The highest, middle or lowest rolls or list entries can be selected with (^ or h), (m or o), or (v or l) respectively. 6d6^3 will keep the highest 3 rolls, whereas 6d6v3 will select the lowest 3 rolls. If a number isn't specified, it defaults to keeping all but one for highest and lowest, and all but two for the middle. If a negative value is given as the operand for any of these operators, this operation will drop that many elements from the result. For example, 6d6^-2 will drop the two lowest values from the set, leaving the 4 highest. Zero has no effect.

A variant of the "explode" operator is the a ("again") operator. Instead of re-rolling values equal to or greater than the threshold (or max value), this operator doubles values equal to the provided threshold (or max value). When no right-side operand is specified, the left side must be a dice expression.

There are two operators for taking a set of rolls or numbers and counting the number of elements at or above a certain threshold, or "successes". Both require a right-hand operand for the threshold. The first, e, only counts successes. The second, f, counts successes minus failures, which are when a roll is the minimum possible value for the die element, or 1 for lists.

A list or set of rolls can be turned into an integer with the total (t) operator. 6d1t will return 6 instead of [1, 1, 1, 1, 1, 1]. Applying integer operations to a list of rolls will total them automatically.

A set of dice rolls can be sorted with the sort (s) operator. 4d6s will not change the return value, but the dice will be sorted from lowest to highest.

The +- operator is a special prefix for sets of rolls and lists. It negates odd roles within a list. Example: [1, 2, 3] -> [-1, 2, -3]. There is also a negate (-) operator, which works on either single elements, sets or rolls, or lists. There is also an identity + operator.

Values can be added or subtracted from each element of a list or set of rolls with the point-wise add (.+) and subtract (.-) operators. For example: 4d1 .+ 3 will return [4, 4, 4, 4].

Basic integer operations are also available: (16 / 8 * 4 - 2 + 1) % 4 -> 3.

Finally, there are two operators for building and extending lists. To build a list, use a comma to separate elements. If any comma-seperated item isn't a scalar (e.g. a roll), it is flattened into one by taking its total. The "extend" operator (|) is used to merge two lists into one, or append single elements to the beginning or end of a list.

Python API

The calls to dice.roll() above may be replaced with dice.roll_min() or dice.roll_max() to force ALL rolls to their highest or lowest values respectively. This might be useful to see what the minimum and maximum possible values for a given expression are. Beware that this causes wild dice rolls to act like normal ones, and rolls performed as explosions are not forced high or low.

The roll() function and variants take a boolean raw parameter which makes the library return the element instead of the result. Note that the evaluate_cached method is called as part of roll(), which populates element.result. Calling element.evaluate() will not reset this value.

To display a verbose breakdown of the element tree, the dice.utilities.verbose_print(element) function is available. If element.result has not yet been populated, the function calls evaluate_cached() first. Keep this in mind if you want to print the result of an evaluation with custom arguments. verbose_print() returns a str.

Most evaluation errors will raise DiceError or DiceFatalError, both of which are subclasses of DiceBaseError. These exceptions have a method named pretty_print, which will output a string indicating where the error happened::

>>> try:
...   dice.roll('1/0')
... except dice.DiceBaseException as e:
...   print(e.pretty_print())
  ^ Division by zero