@permaweb/aoconnect

ao Connect

The aoconnect library provides an abstraction for spawning, evaluating, and interacting with ao Processes.

This module will run in a browser or server environment.

• Read the result of an ao Message evaluation from a ao Compute Unit cu
• Send a Message targeting an ao Process to an ao Message Unit mu
• Spawn an ao Process, assigning it to an ao Scheduler Unit su

• Usage
  • ESM (Node & Browser) aka type: module
  • CJS (Node) type: commonjs
  • Usage with Webpack
  • API
    • result
    • results
    • message
    • spawn
    • connect
    • monitor
    • dryrun
    • assign
    • Environment Variables
    • createSigner
    • serializeCron
• Debug Logging
• Testing
• Project Structure

Usage

This module can be used on the server, as well as the browser:

ESM (Node & Browser) aka type: module

import { message, result, spawn } from "@permaweb/aoconnect";

CJS (Node) type: commonjs

const { spawn, message, result } = require("@permaweb/aoconnect");

The duration of this document will use ESM for examples

Usage with Webpack

AOConnect by default exposes a file designed to be used in the browser. If you're using Webpack to bundle an app for the server and are having issues, you may need to update import statements to point at an implementation for node like so.

import { createSigner, message } from "@permaweb/aoconnect/node";

API

result

Read the result of the message evaluation from an ao Compute Unit cu

import { result } from "@permaweb/aoconnect";

let { Messages, Spawns, Output, Error } = await result({
  message: "l3hbt-rIJ_dr9at-eQ3EVajHWMnxPNm9eBtXpzsFWZc",
  process: "5SGJUlPwlenkyuG9-xWh0Rcf0azm8XEd5RBTiutgWAg",
});

results

Read a batch of results from a process, this feature can be used as a polling mechanism looking for new results

import { results } from "@permaweb/aoconnect";

let results = await results({
  process: "5SGJUlPwlenkyuG9-xWh0Rcf0azm8XEd5RBTiutgWAg",
  from: cursor,
  sort: "ASC",
  limit: 25,
});

Parameters

Name	Description	Optional?
process	the process identifier	false
from	cursor starting point	true
to	cursor ending point	true
sort	list results in decending or ascending order, default will be ASC	true
limit	the number of results to return (default: 25)	true

message

send a message to an ao Message Unit mu targeting an ao process.

import { createSigner, message } from "@permaweb/aoconnect";

const messageId = await message({
  process,
  signer: createSigner(wallet),
  anchor,
  tags,
  data,
});

You can pass a 32 byte anchor to message which will be set on the DataItem

spawn

Spawn an ao process, assigning the ao Scheduler to schedule its messages

import { createSigner, spawn } from "@permaweb/aoconnect";

const processId = await spawn({
  module,
  scheduler,
  signer: createSigner(wallet),
  tags,
  data,
});

connect

If you would like the connect to use ao components other than the defaults, you can specify those components by providing their urls to connect. You can currently specify:

• GATEWAY_URL: (currently only used as the default host for GRAPHQL_URL)
• GRAPHQL_URL: the Arweave GraphQL gateway to use for querying data from Arweave. (defaults to ${GATEWAY_URL}/graphql)
• GRAPHQL_MAX_RETRIES: the number of times to retry querying the gateway, utilizing an exponential backoff (defaults to 0)
• GRAPHQL_RETRY_BACKOFF: the initial backoff, in milliseconds (moot if GRAPHQL_MAX_RETRIES is set to 0) (defaults to 300)
• MU_URL: The Messenger Unit URL
• CU_URL: The Compute Unit URL

import { connect } from "@permaweb/aoconnect";

const { spawn, message, result } = connect({
  GATEWAY_URL: "...",
  GRAPHQL_URL: "...",
  GRAPHQL_MAX_RETRIES: 2,
  GRAPHQL_RETRY_BACKOFF: 250,
  MU_URL: "...",
  CU_URL: "...",
});

If GATEWAY_URL is set but GRAPHQL_URL is not set, then the GATEWAY_URL provided MUST have a /graphql endpoint that serves the Arweave Gateway GraphQL Server. ie. https://arweave.net/graphql

If any value is not provided, a library default will be used. In this sense, invoking connect() with no parameters or an empty object is functionally equivalent to using the top-lvl exports of the library:

import {
 spawn,
 message,
 result
 connect
} from '@permaweb/aoconnect';

// These are functionally equivalent
connect() == { spawn, message, result }

monitor

When using cron messages, ao users need a way to start injesting the messages, using this monitor method, ao users can initiate the subscription service for cron messages.

import { createSigner, monitor } from "@permaweb/aoconnect";

const result = await monitor({
  process,
  signer: createSigner(wallet),
});

dryrun

DryRun is the process of sending a message object to a specific process and getting the Result object back, but the memory is not saved, it is perfect to create a read message to return the current value of memory. For example, a balance of a token, or a result of a transfer, etc.

import { createSigner, dryrun } from "@permaweb/aoconnect";

const result = await dryrun({
  process: 'PROCESSID',
  data: '',
  tags: [{name: 'Action', value: 'Balance'},
  anchor: '1234',
  ...rest are optional (Id, Owner, etc)
});

console.log(result.Messages[0]);

assign

Create an Assignment for an ao process

import { assign } from "@permaweb/aoconnect";

const processId = await assign({
  process: "process-id",
  message: "message-id",
  exclude: ["Data", "Tags", "etc."], // optional list of DataItem fields to exclude
});

Create a Assignment for an ao process with an L1 transaction

import { assign } from "@permaweb/aoconnect";

const processId = await assign({
  process: "process-id",
  message: "txid",
  baseLayer: true,
});

Environment Variables

The library also allows configuring ao components described above, using environment variables.

On NodeJS, you can use process.env to set these values.

In the browser, you can use globalThis to set these values.

In both cases, you should set environment variables prior to importing the module. If this is not possible, consider using connect and passing in values from the environment that way.

createSigner

createDataItemSigner is available for backwards compatibility, but is an alias for createSigner.

In mainnet mode all HTTP Messages and Data Items must be signed.

createSigner is a convenience api that, given a wallet, returns a function that can be passed as signer to connect in order to properly sign DataItems and HTTP Signed Messages.

The library provides a browser compatible and node compatible version that you can use OOTB.

The browser compatible versions expects an instance of window.arweaveWallet to be passed to it:

import { createSigner } from "@permaweb/aoconnect";

const signer = createSigner(globalThis.arweaveWallet);

The node compatible versions expects a JWK interface to be passed to it:

import fs from "node:fs";
import { createSigner } from "@permaweb/aoconnect";

const wallet = JSON.parse(fs.readFileSync(process.env.PATH_TO_WALLET));
const signer = createSigner(wallet);

You can also implement your own signer, as long as it satisfies the api. Here is what the API looks like in TypeScript:

type KeyPublicMeta = {
  publicKey: string;
  alg?: string;
  type?: number;
};

type CreateSignatureData = (key: KeyPublicMeta): Promise<Uint8Array>;

type SignatureDataType = "ans104" | "httpsig";

type SignatureResult = {
  signature: Uint8Array | string;
  address: string;
};

type RawSigner = (
  create: CreateSignatureData,
  type: SignatureDataType,
): Promise<SignatureResult>;

create will be a function that expects to receive, as its first arugments, an object that has the following fields:

• publicKey: the publicKey of the credentials, either as a Uint8Array, or base64 URL encoded string.
• alg (optional): the algorithm used for signing. If not provided, the default is rsa-pss-sha512 for HTTP Signed Messages and rsa-v1_5-sha256 for ANS-104 Data items (these are currently the only supported algorithms)
• type (optional): the canonical signature type. If not provided, the default is 1 which means arweave keys and signature type. (Currently only 1 is supported)

The second argument, type, in most cases can be ignored, but is provided in cases where different behavior is desired for signing the signature base of DataItems vs. HTTP Signed Messages.

In order to construct the data to sign, the signer MUST call create, providing it the information described above. create will return a Promise that resolves to a Uint8Array that will contain the data to be signed. Once the signer has signed this data, it should return a Promise that resolves to a SignatureResult with the following fields:

• signature: the raw bytes of the signature produced, or a base64 URL encoded string
• address: be the canonical address associated with the credentials.

Note that, in this signing flow, that the private credentials never need to be known by aoconnect. In this way, the actual signing operation may be an opaque operation -- a black box. Upon receiving the signature and address from the signer, aoconnect will handle encoding them, along with associated metadata such as a public key, alg and key type, in the proper context: as the signature in an ANS-104 Data Item, or as a Signature on an HTTP Signed Message.

For example, here is a custom signer where the credentials and signing are stored remotely in a hypothetical vault service:

import { connect } from 'aoconnect'

const myAwesomeSigner = async (create) => {
  /**
   * First retrieve the public key information,
   * and invoke create to produce the signature base
   */
  const dataToSign = await fetch('https://my.vault.io/.well-known/jwks.json')
    .then((res) => res.json())
    .then(async ([firstKey]) => {
      const publicKey = Buffer.from(firstKey.n, 'base64url')
      return create({ publicKey })
    })

  /**
   * Now sign the data
   */
  return fetch(
    'https://my.vault.io',
    {
      method: 'POST',
      body: dataToSign,
      ...
    }
  ).then(async (res) => ({
    signature: res.headers.get('signature'),
    address: res.headers.get('address')
  }))
}

/**
 * Inject the signer
 */
const { message } = connect({ signer: myAwesomeSigner })

serializeCron

Serialize a Cron Object into Cron Tags.

import { serializeCron } from "@permaweb/aoconnect";

const cron = {
  interval: "10-minutes",
  tags: [
    { name: "Foo", value: "bar" },
    { name: "Bar", value: "foo" },
  ],
};

const tags = serializeCron(cron);
/*
  tags = [
    { name: "Cron-Interval", value: "10-minutes" },
    { name: "Cron-Tag-Foo", value: "bar" },
    { name: "Cron-Tag-Bar", value: "foo" },
  ]
*/

Cron Interval must adhere to the AO Data Protocol

Debug Logging

You can enable verbose debug logging on the library. All logging is scoped under the name @permaweb/aoconnect*. You can use wildcards to enable a subset of logs ie. @permaweb/aoconnect/result*

For Node, set the DEBUG environment variable to the logs you're interested in.

For the Browser, set the localStorage.debug variable to the logs you're interested in.

Testing

Run npm test to run the tests.

Run npm run test:integration to run the integration tests.

Project Structure

The aoconnect project loosely implements the Ports and Adapters Architecture.

All business logic is in lib where each public api is implemented and tested.

dal.js contains the contracts for the driven adapters aka side-effects. Implementations for those contracts are injected into, then parsed and invoked by, the business logic. This is how we inject specific integrations for providers ie. Warp, Irys, or even platforms specific implementations like node and the browser while keeping them separated from the business logic -- the business logic simply consumes a black-box API -- easy to stub, and easy to unit test.

Because the contract wrapping is done by the business logic itself, it also ensures the stubs we use in our unit tests accurately implement the contract API. Thus our unit tests are simoultaneously contract tests.

client contains implementations, of the contracts in dal.js, for various platforms. The unit tests for the implementations in client also import contracts from dal.js to help ensure that the implementation properly satisfies the API.

Finally, the entrypoints (index.js for Node and index.browser.js for the Browser) orchestrate everything, choosing the appropriate implementations from client and injecting them into the business logic from lib.