To help facilitate faster development of Cardano smart contracts, we present a collection of tried and tested modules and functions for implementing common design patterns.
Install the package with aiken
:
aiken package add anastasia-labs/aiken-design-patterns --version main
And you'll be able to import functions of various patterns:
use aiken_design_patterns/multi_utxo_indexer as multi_utxo_indexer
use aiken_design_patterns/singular_utxo_indexer as singular_utxo_indexer
use aiken_design_patterns/stake_validator as stake_validator
use aiken_design_patterns/tx_level_minter as tx_level_minter
Check out validators/
to see how the exposed functions can be used.
Here are the steps to compile and run the included tests:
1.Clone the repo and navigate inside:
git clone https://github.com/Anastasia-Labs/aiken-design-patterns
cd aiken-design-patterns
2.Run the build command, which both compiles all the functions/examples and also runs the included unit tests:
aiken build
This module offers two functions meant to be used within a multi-validator for implementing a "coupled" stake validator logic.
The primary application for this is the so-called "withdraw zero trick," which is most effective for validators that need to go over multiple inputs.
With a minimal spending logic (which is executed for each UTxO), and an arbitrary withdrawal logic (which is executed only once), a much more optimized script can be implemented.
spend
merely looks for the presence of a withdrawal (with arbitrary amount)
from its own reward address.
withdraw
takes a custom logic that requires 3 arguments:
- Redeemer (arbitrary
Data
) - Script's validator hash (
Hash<Blake2b_224, Script>
) - Transaction info (
Transaction
)
The primary purpose of this pattern is to offer a more optimized solution for a unique mapping between one input UTxO to one or many output UTxOs.
By specifying the redeemer type to be a pair of integers ((Int, Int)
), the
validator can efficiently pick the input UTxO, match its output reference to
make sure it's the one that's getting spent, and similarly pick the
corresponding output UTxO in order to perform an arbitrary validation between
the two.
The provided example validates that the two are identical, and each carries a single state token apart from Ada.
Here the validator looks for a set of outputs for the given input, through a
redeemer of type (Int, List<Int>)
(output indices are required to be in
ascending order to disallow duplicates). To make the abstraction as efficient
as possible, the provided higher-order function takes 3 validation logics:
- A function that validates the spending
Input
(single invocation). - A function that validates the input UTxO against a corresponding output UTxO. Note that this is executed for each associated output.
- A function that validates the collective outputs. This also runs only once. An example use-case could be checking for the number of outputs.
While the singular variant of this pattern is primarily meant for the spending
endpoint of a contract, a multi UTxO indexer utilizes the stake validator
provided by this package. And therefore the spending endpoint can be taken
directly from stake_validator
.
Subsequently, spend redeemers are irrelevant here. The redeemer of the withdrawal endpoint is expected to be a properly sorted list of pairs of indices (for the one-to-one case), or a list of one-to-many mappings of indices.
The distinction between one-to-one and one-to-many variants here is very similar to the singular case, so please refer to its section above for more details.
Very similar to the stake validator, this design pattern utilizes a multi-validator comprising of a spend and a minting endpoint.
The role of the spendig input is to ensure the minting endpoint executes. It does so by looking at the mint field and making sure a non-zero amount of its asset (where its policy is the same as the multi-validator's hash, and its name is specified as a parameter) are getting minted/burnt.
The arbitrary logic is passed to the minting policy so that it can be executed a single time for a given transaction.
The datatype that models validity range in Cardano currently allows for values that are either meaningless, or can have more than one representations. For example, since the values are integers, the inclusive flag for each end is redundant and can be omitted in favor of a predefined convention (e.g. a value should always be considered inclusive).
In this module we present a custom datatype that essentially reduces the value domain of the original validity range to a smaller one that eliminates meaningless instances and redundancies.
The datatype is defined as following:
pub type NormalizedTimeRange {
ClosedRange { lower: Int, upper: Int }
FromNegInf { upper: Int }
ToPosInf { lower: Int }
Always
}
The exposed function of the module (normalize_time_range
), takes a
ValidityRange
and returns this custom datatype.