spadille

Verifiable/deterministic fair tickets generation for lotteries, raffles and gambling games.

About

This is a library to generate random luck numbers in a fair way. Rather than relying on your process’ PRNG blackbox (even if it’s a cryptographic secure one), we can generate noise/randomness that is verifiable through a deterministic/replayable algorithm. Such kind of algorithms for casinos, raffles & promotions are called Provably fair algorithms.

It becomes even more important when we run in a cluster of processes and we must ensure that no matter which kind of process receives the request, it must deliver the same noise output. We prove that property by using HMAC-like PRNG as shown here. We can keep the random number generation secret by a moment by just keeping the “HMAC” key secret, and then open/revealing that for clients during gambling/raffle output/outcome verification.

In a broader context, it can be used for Secure Multi-party Computations too (mostly through Commitment Schemes). This library provides an API cross-compatible between servers (using Node.js engine with support to crypto OpenSSL’s bindings) and browsers (using modern browsers supporting the crypto.subtle API). A good verification flow, thus, would be to generate such noise number sequences on server-side and then verifying them on client-side.

Installation

If available on NPM, just type either npm i spadille or yarn add spadille. Otherwise, you can pin this project by npm link . or yarn link, and then linking externally with either npm link spadille or yarn link spadille. The release/stable front-end CDN is available on UNPKG once the library is available on NPM beforehand. Otherwise, you can just grab the front-end minified code (at dist/index.js).

Usage

To generate random sequences (paired on Brazillian lotteries if you want to run a raffle without any kind of audit person):

const Lottery = spadille.lottery.brazillian;
const megaSenaSequence = await Lottery.megaSena(secret, payload);
const federalNumbers = await Lottery.federal(secret, payload);

Here, secret is your “HMAC-signing”-key and payload is a user/session-derived content (possibly a session ID, request ID, raffle counter, whatever…). The megaSenaSequence is a Mega-Sena lottery sequence of unique and sorted numbers between 1 and 60, inclusively. Such sequence contains 6 numbers. The federalNumbers, on the other hand, is a string of 5 digits, each one between 0 and 9, and this sequence may contain repeated numbers (that is, a not unique sequence). Future plans include other famous Brazillian lotteries.

To generate arbitrary random sequences:

const arbitrarySequence = await spadille.prng.generate({
  secret,
  payload,
  minimum: minimumInclusiveValue,
  maximum: maximumInclusiveValue,
  amount: outputSequenceLength,
  distinct: booleanFlag,
});

Such sequence can be made of many elements as you wish (but keep the eye on hardware limits, e.g, the limits of 32-bits integer representation). The number of elements are configured by the amount parameter. The minimum and maximum are point parameters for an inclusive interval (closed on both sides). The distinct is a flag to compute the sequence of unique numbers (without repetitions).

If you want to generate a random number between 0 (closed interval) and 1 (open interval), there is the wrapper function spadille.prng.rand, inspired on the classic Random API as found in many programming languages in the wild. To use this function, just call:

const randomFraction = await spadille.prng.rand(secret, payload);

Given that we can generate arbitrary sequences, the random permutation algorithm becomes straightforward. This kind of permutation would just generate a random index sequence with minimum as 0, maximum as inputSequence.length - 1 and amount as inputSequence.length, where inputSequence is the list that we want to permute/shuffle. We then, in the end, use such random index sequence to map inputSequence entries into an output sequence by indexing with the random index sequence. This wrapper function is implemented as an API below:

const inputSequence = [ ... ] // an arbitrary list
const outputSequence = await spadille.prng.permute({
  secret, payload, inputSequence
})
/*
 outputSequence is a random permutation of inputSequence
 keep in mind that there are a still unlikely probability
 of random collision where the inputSequence order could
 be preserved for outputSequence, even if this is negligible
*/

Likewise, it’s possible to take only a randomly ordered sub-sequence from the original sequence. This wrapper function is called pick and the contract/typing follows:

const classes = [
  'warrior',
  'rogue',
  'mage',
  'priest',
  'hunter',
  'alchemist'
];
const partyClasses = await spadille.prng.pick({
  secret,
  payload,
  sequence: classes,
  distinct: true, // optional, default is `false` for pick()
  amount: 3, // optional, default is `1` for pick()
});

Note that {distinct: true} only filters duplications on array-index-level, not on array-value-level, it means that if your input sequence/array contains duplicated values, they aren’t deduplicated here. It also means if {distinct: false} and your input sequence/array contain just unique values, it is possible to generate duplicated values - it’s all because picking is implemented on array-index-level generation. The default behavior of pick is to retrieve just one random element from a given sequence, but the output/result is still a list, thus, you will likely use the following pattern in such cases:

const [randomElement] = await spadille.prng.pick({
  secret,
  payload,
  sequence
});

Note that pick will yield the same behavior of permute if you pass the same secret, payload, sequence for both calls, and {distinct: true} with {amount: sequenceLength} for pick. Therefore, pick is a generalisation/superset of permute, and the latter can contain the underlying implementation calling the former (actually this is not the case by now, but future refactor processes will end on that code deduplication).

There’s also a helper function provided to help you to generate fresh secrets. By using cryptograpically secure PRNGs for both Node (through crypto OpenSSL bindings) and browsers (through the crypto API), we ensure a good source of entropy for that noise string. The output string is under binary mode, but you can nevertheless convert to formats/encodings such as Base-64 and Hexadecimal. Just pass the amount of bytes to generate and be happy with that! :)

const amountOfBytes = 32;
const noiseSecret = await spadille.secret.generate(amountOfBytes);

Remember that once you generate such secret, you should store it somewhere to retrieve later to “sign” the random sequences. And in the end, you should also publish such secret in a commitment/opening style for public verification by your users/clients. To send/receive such secret while using HTTPS requests, for instance, you can use the browser-and-Node cross-compatible Base64 encoding provided here too:

// on server-side
// ...
const base64Secret = spadille.base64.encode(secret);
response.json({ base64Secret });
// ...

// on client-side
// ...
const response = await axios.get(endpoint, { headers });
const secret = spadille.base64.decode(response.data.base64Secret);
// the <secret> variable is ready to be used for
// raffle/promotion verification on client-side here
// ...

This helper Base64 submodule ensures that Node.js can encode a binary-content secret valid to be decoded on browsers and vice-versa. If now you are somehow confuse by this amount of functions/APIs described here, don’t worry, there’s a TypeScript typings file available in this library for easy IDE/Editor integration (such as auto-complete and parameters signature).

Remarks

Any doubts, enter in touch. Pull requests and issues are welcome! Have fun playing with this library! Happy hacking!