#unique-identifier #id-generator #unsigned-integer #unique-id #distributed-id #random-string #string-conversion

bin+lib block-id

Library for generating opaque, unique, and short string values from (unsigned) integers

5 releases

0.2.1 Apr 27, 2023
0.2.0 Apr 12, 2023
0.1.2 Apr 9, 2022
0.1.1 Apr 1, 2022
0.1.0 Apr 1, 2022

#4 in #distributed-id

Download history 532/week @ 2024-06-13 280/week @ 2024-06-20 62/week @ 2024-06-27 109/week @ 2024-07-04 95/week @ 2024-07-11 96/week @ 2024-07-18 114/week @ 2024-07-25 133/week @ 2024-08-01 69/week @ 2024-08-08 110/week @ 2024-08-15 91/week @ 2024-08-22 118/week @ 2024-08-29 40/week @ 2024-09-05 74/week @ 2024-09-12 65/week @ 2024-09-19 75/week @ 2024-09-26

265 downloads per month
Used in 2 crates

MIT/Apache

27KB
596 lines

block-id

GitHub Repo stars wokflow state crates.io docs.rs dependency status

block-id is a Rust library for generating opaque, unique, and short string values from (unsigned) integers.

tl;dr:

use block_id::{Alphabet, BlockId};

fn main() {
    // Random seed.
    let seed = 9876;
    
    // Code length.
    let length = 5;

    let generator = BlockId::new(Alphabet::alphanumeric(), seed, length);
    
    // Number to string.
    assert_eq!(Some("wjweA".to_string()), generator.encode_string(0));
    assert_eq!(Some("ZxJrE".to_string()), generator.encode_string(1));
    assert_eq!(Some("3e0IT".to_string()), generator.encode_string(2));

    // String to number.
    assert_eq!(Some(2), generator.decode_string("3e0IT"));
}

Introduction

Random-looking alphanumeric strings are often used in place of sequential numeric IDs for user-facing purposes. This has several advantages:

  • String identifiers are usually visually distinct, even if they were generated adjacently in sequence.
  • The higher information density of a larger alphabet allows for shorter codes.
  • Sequential identifiers reveal unnecessary information about ordering and object creation rate that you may not want to reveal.

block-id is the successor to tiny_id, which allows the creation of tightly-packed alphanumeric strings. tiny_id turned out to be difficult to use in a distributed environment because its state needs to be synchronized across every node that needs to generate IDs. Rather than building distributed functionality into a short ID generator, block-id provides a way of turning a sequential ID generator into a string ID generator by creating a one-to-one mapping between integers and random-looking short strings. That way, any system of generating sequential numeric IDs (for example, a database's sequence generator) can be turned into a system for generating random-looking string IDs.

use block_id::{Alphabet, BlockId};

fn main() {
    // The alphabet determines the set of valid characters in an ID.
    // For convenience, we include some common alphabets like `alphanumeric`. 
    let alphabet = Alphabet::alphanumeric();
    
    // The generator takes a u128 as a seed.
    let seed = 1234;

    // The length of a generated code. This is really a _minimum_ length; larger numbers
    // will be converted to longer codes since that's the only way to avoid collisions.
    let length = 4;

    // A small amount of pre-caching work happens when we create the BlockId instance,
    // so it's good to re-use the same generator where possible.
    let generator = BlockId::new(alphabet, seed, length);
    
    // Now that we have a generator, we can turn numbers into short IDs.
    assert_eq!(Some("In4R".to_string()), generator.encode_string(0));

    assert_eq!(Some("4A7N".to_string()), generator.encode_string(440));
    assert_eq!(Some("tSp9".to_string()), generator.encode_string(441));
    assert_eq!(Some("6z6y".to_string()), generator.encode_string(442));
    assert_eq!(Some("ft0M".to_string()), generator.encode_string(443));

    // When we've exhausted all 4-digit codes, we simply move on to 5-digit codes.
    assert_eq!(Some("YeyKs".to_string()), generator.encode_string(123456789));

    // ...and so on.
    assert_eq!(Some("pFbrRf".to_string()), generator.encode_string(1234567890));

    // Codes are reversible, assuming we have the seed they were generated with.
    assert_eq!(Some(1234567890), generator.decode_string("pFbrRf"));
}

How it works

block-id applies a pipeline of reversible transformations on a data in order to turn it into a string.

  • Base conversion turns the input integer into a base-N representation where N is the number of characters in the desired output alphabet.
  • Rounds consisting of:
    • Permutation applies an N-to-N map to every digit of the base-N representation. The permutation is generated from the random seed passed in the BlockId constructor.
    • Cascade applies a left-to-right cumulative sum, modulo N, to the base-N representation.
    • Rotate takes the first digit of the base-N representation and moves it to the last digit, shifting all of the other digits left by one.
  • Alphabetization translates every digit of the base-N representation to a “letter” in the alphabet provided at construction.

The number of rounds is the same as the number of digits in the base-N representation. This gives every digit a chance to influence every other digit.

Security

block-id is designed to make it easy for a human to distinguish between two sequential codes, not to make it impossible for an adversary to reverse. It should not be considered cryptographically secure.

Dependencies

~345KB