Can we create a process that automatically transforms any integer field into a hashed string?

Question

Can we create a process that automatically transforms any integer field into a hashed string?

Is there a non-hacky way to hash all IDs before returning to the user? I have explored the documentation extensively but haven't found a solution that covers all scenarios.

I am working with Postgres and Prisma ORM, managing multiple models with relational fields. Ideally, I want to achieve this using Prisma.

I attempted Jordan's suggestion, resulting in the following code:

const getPrismaClient = () => {
  const client = new PrismaClient();
  client.$extends({
    result: {
      blog: {
        // Creating a virtual `hashedId` field
        hashedId: {
          needs: { id: true },
          compute(client) {
            return intToHashedString(client.id);
          },
        },
      },
    },
  });
  client.blog
    .findMany({
      select: { hashedId: true },
    })
    .then((r) => {
      console.log(r);
    });
  return client;
};

However, it resulted in the following error:

error - unhandledRejection: PrismaClientValidationError: 
Invalid `prisma.blog.findMany()` invocation:
{
  select: {
    hashedId: true,
    ~~~~~~~~
?   id?: true,
?   title?: true,
?   slug?: true,
?   image?: true,
?   description?: true,
?   views?: true,
?   authorId?: true,
  clientVersion: '4.11.0'
}

javascript postgresql express hash prisma

Answer 1

Answer №1

If you're looking to enhance your Prisma Client results, consider utilizing the Prisma Client results extension (Please note that this feature is still under a feature flag, so proceed with caution).

To get started, enable the feature by adding the following code to your schema.prisma file:

// schema.prisma
generator client {
  provider = "prisma-client-js"
  previewFeatures = ["clientExtensions"]
}

Once enabled, here's an example of how you can implement it:

import { PrismaClient } from '@prisma/client';
import { scrypt } from 'node:crypto';

const prisma = new PrismaClient();

const xprisma = prisma.$extends({
  result: {
    client: {
      // A virtual `hashedId` field is created here
      hashedId: {
        needs: { id: true },
        compute(client: { id: string }) {
          return new Promise<string>((resolve, reject) => {
            // In this example, I used scrypt
            scrypt(client.id, 'salt', 64, (err, hashed) => {
              if (err) return reject(err);
              return resolve(hashed.toString('hex'));
            })
          })
        }
      }
    }
  }
});

//                    ,--- Using xprisma instead of prisma
//                   v
const client = await xprisma.client.findFirst({
  where: { id: '092a5486-9ca2-4c4c-85e6-da7cedb281a5' },
  // Autocomplete for hashedId is available in fields
  select: { hashedId: true },
});

It's important to address performance concerns when implementing this solution, but it should fulfill your requirements.

UPDATE: Remember that prisma.$extends must return a new PrismaClient containing your extensions for successful compilation. This is why I followed the convention of using xprisma.

Answer 2

If you're looking to enhance your Prisma Client results, consider utilizing the Prisma Client results extension (Please note that this feature is still under a feature flag, so proceed with caution).

To get started, enable the feature by adding the following code to your schema.prisma file:

// schema.prisma
generator client {
  provider = "prisma-client-js"
  previewFeatures = ["clientExtensions"]
}

Once enabled, here's an example of how you can implement it:

import { PrismaClient } from '@prisma/client';
import { scrypt } from 'node:crypto';

const prisma = new PrismaClient();

const xprisma = prisma.$extends({
  result: {
    client: {
      // A virtual `hashedId` field is created here
      hashedId: {
        needs: { id: true },
        compute(client: { id: string }) {
          return new Promise<string>((resolve, reject) => {
            // In this example, I used scrypt
            scrypt(client.id, 'salt', 64, (err, hashed) => {
              if (err) return reject(err);
              return resolve(hashed.toString('hex'));
            })
          })
        }
      }
    }
  }
});

//                    ,--- Using xprisma instead of prisma
//                   v
const client = await xprisma.client.findFirst({
  where: { id: '092a5486-9ca2-4c4c-85e6-da7cedb281a5' },
  // Autocomplete for hashedId is available in fields
  select: { hashedId: true },
});

It's important to address performance concerns when implementing this solution, but it should fulfill your requirements.

UPDATE: Remember that prisma.$extends must return a new PrismaClient containing your extensions for successful compilation. This is why I followed the convention of using xprisma.

Answer 3

Answer №2

In the realm of Postgres database management, there exist hashing functions for every data type, though they are not officially documented. These functions are typically safe to use and serve various purposes such as hash partitioning and indexing. When it comes to integers, you can utilize the following functions:

hashint4(int) → int
hashint4extended(int, bigint) → bigint

The latter form, which was introduced in Postgres 12 or later, requires a seed as the second parameter. It generates a larger bigint hash in order to reduce the likelihood of collisions.

If you want to almost guarantee avoidance of hash collisions, consider using the trusty old md5() function, which yields strings consisting of 32 characters – similar to a UUID. For further information, check out:

What is the optimal data type for an MD5 field?

Note that md5() operates on text, so if your input is an integer, you'll need to perform a cast. See examples below:

SELECT hashint4(my_id)            AS my_id_int4_hash
     , hashint4extended(my_id, 0) AS my_id_int8_hash
     , hashint4(my_id)            AS my_id_md5_hash
FROM   tbl
LIMIT  10;

If you prefer working with textual results, ensure to cast numeric outcomes to text using ::text.

For additional insights, especially regarding the probability of hash collisions, explore:

Why doesn't my UNIQUE constraint trigger?

To streamline your workflow and conserve effort, contemplate creating a (MATERIALIZED) VIEW of your table with hashed values. This approach eliminates repetitive typing and computational tasks, even though the underlying functions exhibit exceptional performance.

Answer 4

In the realm of Postgres database management, there exist hashing functions for every data type, though they are not officially documented. These functions are typically safe to use and serve various purposes such as hash partitioning and indexing. When it comes to integers, you can utilize the following functions:

hashint4(int) → int
hashint4extended(int, bigint) → bigint

The latter form, which was introduced in Postgres 12 or later, requires a seed as the second parameter. It generates a larger bigint hash in order to reduce the likelihood of collisions.

If you want to almost guarantee avoidance of hash collisions, consider using the trusty old md5() function, which yields strings consisting of 32 characters – similar to a UUID. For further information, check out:

What is the optimal data type for an MD5 field?

Note that md5() operates on text, so if your input is an integer, you'll need to perform a cast. See examples below:

SELECT hashint4(my_id)            AS my_id_int4_hash
     , hashint4extended(my_id, 0) AS my_id_int8_hash
     , hashint4(my_id)            AS my_id_md5_hash
FROM   tbl
LIMIT  10;

If you prefer working with textual results, ensure to cast numeric outcomes to text using ::text.

For additional insights, especially regarding the probability of hash collisions, explore:

Why doesn't my UNIQUE constraint trigger?

To streamline your workflow and conserve effort, contemplate creating a (MATERIALIZED) VIEW of your table with hashed values. This approach eliminates repetitive typing and computational tasks, even though the underlying functions exhibit exceptional performance.

Can we create a process that automatically transforms any integer field into a hashed string?

Answer №1

Answer №2

Similar questions

Transforming Form Input Fields into a Nested JSON Format with the Power of JQuery

Tips for effectively managing relationships in a RESTful API

data is currently being uploaded to the storage, however, the grid is not displaying any

Packing third-party npm modules with Webpack for seamless integration

What is the best way to emphasize a div depending on a query outcome?

How can we determine the nL and nH values using an ESC/POS command?

Unable to modify the name of an element's class due to restrictions in JavaScript

What is the method for incorporating parameters into an array filter?

There seems to be an issue with the next function's functionality within a Nodejs middleware

Updating the row by substituting the content of two columns with the data retrieved from the action class

JavaScript Navigation Bar Error: The value of 'undefined' is not recognized as an object

Personalize the Loopback response following a successful save operation

Express.js seems to be having trouble identifying static files

executing a series of jQuery ajax calls iteratively

Encountered a problem when incorporating delay functions into redux-saga testing using the Redux Saga Test Plan library

Create a custom overlay for an image that is centered horizontally and does not have a fixed width

React does not allow _id to be used as a unique key

Invoking servlet using Ajax

How can you shut down the server in the Express.js library?

ajax is now blocking the use of Window.location