mmzk-typeid: A TypeID and UUIDv7 implementation for Haskell

[ data, library, mit, typeid, uuid, uuidv7 ] [ Propose Tags ]

TypeID is a type-safe, K-sortable, globally unique identifier inspired by Stripe IDs.

The specification is available at https://github.com/jetpack-io/typeid.

This library supports generating and parsing speç-conforming TypeIDs, with the following additional features:

  • Batch generating TypeIDs with the same UUIDv7 timestamp

  • Encode prefixes at type-level for better type-safety

It also serves as a (temporary) UUIDv7 implementation in Haskell, since there are no official ones yet.

Please enable the following extensions if working with KindID:

{-# LANGUAGE DataKinds #-}
{-# LANGUAGE PolyKinds #-}
{-# LANGUAGE TypeApplications #-}
{-# LANGUAGE TypeFamilies #-}

While the following is not required, it is quite convenient to have

{-# LANGUAGE OverloadedStrings #-}

[Skip to Readme]

Downloads

Maintainer's Corner

Package maintainers

For package maintainers and hackage trustees

Candidates

  • No Candidates
Versions [RSS] 0.1.0.0, 0.2.0.0, 0.3.0.0, 0.3.0.1, 0.3.1.0, 0.4.0.0, 0.4.0.1, 0.5.0.0, 0.5.0.1, 0.5.0.2, 0.6.0.0, 0.6.0.1, 0.6.2.0, 0.6.3.0
Change log CHANGELOG.md
Dependencies aeson (>=2.1 && <3), array (>=0.5 && <0.6), base (>=4.16 && <5), binary (>=0.8.5 && <0.9), bytestring (>=0.11 && <0.12), entropy (>=0.4 && <0.5), hashable (>=1.4 && <1.5), random (>=1.2 && <1.3), text (>=2.1 && <2.2), time (>=1.11 && <1.13), uuid (>=1.3 && <1.4), uuid-types (>=1.0 && <1.1) [details]
License MIT
Author Yitang Chen <mmzk1526@outlook.com>
Maintainer Yitang Chen <mmzk1526@outlook.com>
Category Data, UUID, UUIDv7, TypeID
Home page https://github.com/MMZK1526/mmzk-typeid
Bug tracker https://github.com/MMZK1526/mmzk-typeid/issues
Uploaded by MMZK1526 at 2024-06-01T08:56:11Z
Distributions
Reverse Dependencies 1 direct, 0 indirect [details]
Downloads 368 total (80 in the last 30 days)
Rating (no votes yet) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Docs uploaded by user
Build status unknown [no reports yet]

Readme for mmzk-typeid-0.6.3.0

[back to package description]

mmzk-typeid

Introduction

A TypeID implementation in Haskell. It is a "type-safe, K-sortable, globally unique identifier" extended on top of UUIDv7.

TypeIDs are canonically encoded as lowercase strings consisting of three parts:

  1. A type prefix (at most 63 characters in all lowercase snake_case ASCII [a-z_]);
  2. An underscore '_' separator;
  3. A 128-bit UUIDv7 encoded as a 26-character string using a modified base32 encoding.

For more information, please check out specification v0.3.0.

It also serves as a (temporary) UUIDv7 implementation in Haskell, since there are no official ones yet.

If you notice any issues or have any suggestions, please feel free to open an issue or contact me via email.

Highlights

In addition to the features provided by TypeID, this implementation also supports:

  1. Generating TypeIDs in a batch. They are guaranteed to have the same timestamp (up to the first 32768 ids) and of ascending order;
  2. Encoding the prefix in the type level, so that if you accidentally pass in an invalid prefix, the code won't compile, avoiding the need for runtime checks;
  3. Support TypeID with other UUID versions. Currently v7 (default), v1, v4, and v5 are supported.

Quick start

{-# LANGUAGE OverloadedStrings #-}

import           Control.Exception
import           Data.TypeID (TypeID)
import qualified Data.TypeID as TID

main :: IO ()
main = do

  -- Make a TypeID with prefix 'mmzk':
  typeID <- TID.genTypeID "mmzk"
  putStrLn $ TID.toString typeID

  -- Get components from the TypeID:
  let prefix = TID.getPrefix typeID -- "mmzk"
      uuid   = TID.getUUID typeID
      time   = TID.getTime typeID -- A 'Word64' representing the timestamp in milliseconds

  -- Make a TypeID without prefix:
  typeID' <- TID.genTypeID ""
  print typeID'

  -- Make 10 TypeIDs in a batch. They are guaranteed to have the same timestamp and of ascending order:
  typeIDs <- TID.genTypeIDs "mmzk" 10
  mapM_ print typeIDs

  -- Parse a TypeID from string:
  case TID.parseString "mmzk_01h455vb4pex5vsknk084sn02q" of
    Left err     -> throwIO err
    Right typeID -> print typeID

For a full list of functions on TypeID, see Data.TypeID.

More Usages

TypeID with other UUID Versions

We also support TypeID using some other versions of UUID, including v1 and v4, which loses the monoticity property. To use it, simply import Data.TypeID.V4 instead of Data.TypeID. The following is an example using v4:

{-# LANGUAGE OverloadedStrings #-}

import           Control.Exception
import           Data.TypeIDV4 (TypeIDV4)
import qualified Data.TypeIDV4 as TID

main :: IO ()
main = do

  -- Make a TypeID with prefix 'mmzk':
  typeID <- TID.genTypeID "mmzk"
  putStrLn $ TID.toString typeID

  -- Get components from the TypeID:
  let prefix = TID.getPrefix typeID -- "mmzk"
      uuid   = TID.getUUID typeID

  -- Make a TypeID without prefix:
  typeID' <- TID.genTypeID ""
  print typeID'

  -- Parse a TypeID from string:
  case TID.parseString "mmzk_5hjpeh96458fct8t49fnf9farw" of
    Left err     -> throwIO err
    Right typeID -> print typeID

Type-level TypeID (KindID)

When using TypeID, if we want to check if the type matches, we usually need to get the prefix of the TypeID and compare it with the desired prefix at runtime. However, with Haskell's type system, we can do this at compile time instead. We call this TypeID with compile-time prefix a KindID.

Of course, that would require the desired prefix to be known at compile time. This is actually quite common, especially when we are using one prefix for one table in the database.

For example, suppose we have a function that takes a KindID with the prefix "user", it may have a signature like this: f :: KindID "user" -> IO ().

Then if we try to pass in a KindID with the prefix "post", the compiler will complain, thus removing the runtime check and the associated overhead.

All the prefixes are type-checked at compile time, so if we try to pass in invalid prefixes, the compiler (again) will complain.

{-# LANGUAGE DataKinds #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE TypeApplications #-}

import           Control.Exception
import           Data.KindID (KindID)
import qualified Data.KindID as KID

main :: IO ()
main = do

  -- Make a KindID with prefix 'mmzk':
  kindID <- KID.genKindID @"mmzk" -- Has type `KindID "mmzk"`
  putStrLn $ KID.toString kindID

  -- Get components from the KindID:
  let prefix = KID.getPrefix kindID -- "mmzk"
      uuid   = KID.getUUID kindID
      time   = KID.getTime kindID -- A 'Word64' representing the timestamp in milliseconds

  -- Make a KindID without prefix:
  kindID' <- KID.genKindID @"" -- Has type `KindID ""`
  print kindID'

  -- Make 10 KindIDs in a batch. They are guaranteed to have the same timestamp and of ascending order:
  kindIDs <- KID.genKindIDs @"mmzk" 10
  mapM_ print kindIDs

  -- Parse a KindID from string:
  case KID.parseString @"mmzk" "mmzk_01h455vb4pex5vsknk084sn02q" of
    Left err     -> throwIO err
    Right kindID -> print kindID

For a full list of functions on KindID, see Data.KindID.

Functions with More General Types

TypeID and KindID shares many functions with the same name and functionality. So far, we are using qualified imports to diffentiate them (e.g KID.fromString and TID.fromString). Alternatively, we can use the methods of IDConv to use the same functions for both TypeID and KindID.

{-# LANGUAGE DataKinds #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE TypeApplication #-}

import           Control.Exception
import           Data.KindID
import           Data.TypeID

main :: IO ()
main = do

  -- Make a TypeID with prefix 'mmzk':
  typeID <- genID @TypeID "mmzk"
  print typeID

  -- Make a KindID with prefix 'mmzk':
  kindID <- genID @(KindID "mmzk")
  print kindID

  -- Parse a TypeID from string:
  case string2ID "mmzk_01h455vb4pex5vsknk084sn02q" :: Maybe TypeID of
    Left err     -> throwIO err
    Right typeID -> print typeID

  -- Parse a KindID from string:
  case string2ID "mmzk_01h455vb4pex5vsknk084sn02q" :: Maybe (KindID "mmzk") of
    Left err     -> throwIO err
    Right kindID -> print kindID

  -- Parse a KindID from string (wrong prefix):
  case string2ID "mmzk_01h455vb4pex5vsknk084sn02q" :: Maybe (KindID "foo") of
    Left err     -> throwIO err -- Will throw here as the prefix matches not
    Right kindID -> print kindID

We no longer need to use qualified imports, but on the down side, we need to add explicit type annotations. Therefore it is a matter of preference.

Note that with the class methods, the type application with Symbol no longer works as the full type must be provided. For example, string2ID @"mmzk" "mmzk_01h455vb4pex5vsknk084sn02q" will not compile.

For a full list of these functions, see Data.TypeID.Class.

KindID with Data Kinds

Instead of using raw Symbols as KindID prefixes, we can also define our custom data type for better semantics.

For example, suppose we have three tables for users, posts, and comments, and each table has a unique prefix, we can design the structure as following:

{-# LANGUAGE DataKinds #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE PolyKinds #-}
{-# LANGUAGE TypeApplications #-}

import           Data.KindID
import           Data.KindID.Class

data Prefix = User | Post | Comment

instance ToPrefix 'User where
  type PrefixSymbol 'User = "user"

instance ToPrefix 'Post where
  type PrefixSymbol 'Post = "post"

instance ToPrefix 'Comment where
  type PrefixSymbol 'Comment = "comment"

Now we can use Prefix as a prefix for KindIDs, e.g.

main :: IO ()
main = do
  -- ...
  userID    <- genKindID @'User -- Same as genKindID @"user"
  postID    <- genKindID @'Post -- Same as genKindID @"post"
  commentID <- genKindID @'Comment -- Same as genKindID @"comment"
  -- ...

For more information, see Data.KindID.Class.

Note

Functions not explicitly exported are considered internal and are subjected to changes.