Spacechem Random Level Generator

Built in Python 3.7.2

Usage

python level_generator.py

Default creates a Research level, printing the importable level code to cmd line. See Args for options.

Args

--view : Auto-open Cearn's mission-viewing website with the generated code (http://coranac.com/spacechem/mission-viewer?mode=editor).

--print : Pretty-print the level's molecules in addition to the level code.

--production : Create a production level instead of a research level.

--difficulty [int] : Int from 0 to 3 approximately controlling the level complexity (basically just molecule sizes, at the moment). Default random from 0 to 2.

--lanky : Set the difficulty to the maximum value.

--inputs : # of input zones to use; 1-2 for research, 1-3 for production.

--outputs : # of output zones to use; 1-2 for research, 1-3 for production.

--elements : Set of elements to exclusively select atoms from. Can be specified by atomic symbol or number - Greek elements can only be specified by number (200-203).

--basic : Select atoms exclusively from the set of 'basic' bond count atoms; equivalent to --elements H O B C N S Cl Os.

--large_output: Make a large output (8x4) research level.

--fusion: Add a fusion laser to the level and adjust random generation accordingly.

--fission: Add a fission laser to the level and adjust random generation accordingly.

--nuclear: Shortcut for combining --fission and --fusion.

--symmetric: Use symmetric molecules

--polymer: Use polymer molecules

e.g. python level_generator.py --view -d 0 --inputs 1 --elements C O H --fission

TBA:

--random (random inputs)

--sorting

--waste (imbalanced reaction that requires storing waste atoms in the reactor)

Behaviour

The process for generating a level is roughly as follows:

• Randomly generate level features (e.g. fuser) and # of output zones (if unspecified in args)

• Randomly generate output molecules which are either symmetrical (rotationally, reflexively, etc) or polymers (mostly composed of repeating segments).

• Divide each output molecule by its 'GCD', and combine all the resulting atoms. This sum creates the total formula of the outputs in the 'balanced' reaction equation.

• Randomly perform inverse nuclear transformations on the balanced output formula

• Calculate a set of input formulas of minimum total size, such that they span the balanced reaction formula. Ensure these formulas are valid based on element bond counts and input zone borders

• Construct each input molecule from the generated formulas (currently ugly construction algorithm)

• To reduce bias from the input -> output process, randomly swap inputs with outputs with 50% probability, if input/output specifications allow this.

• Randomly add level features (bonders, recycler, etc). Production levels have a 1/50 chance to use assemblers/disassemblers instead of regular reactors.

• Display the mission code (and prettyprint molecules or open cearn's mission viewer website, as specified).