Algorand Mapping

Mapping functions define how chain data is transformed into the optimised GraphQL entities that we have previously defined in the schema.graphql file.

• Mappings are defined in the src/mappings directory and are exported as a function.
• These mappings are also exported in src/index.ts.
• The mappings files are reference in project.yaml under the mapping handlers.

There are different classes of mappings functions for Algorand: Block handlers, and Transaction Handlers.

Block Handler

You can use block handlers to capture information each time a new block is attached to the chain, e.g. block number. To achieve this, a defined BlockHandler will be called once for every block.

import { AlgorandBlock } from "@subql/types-algorand";

export async function handleBlock(block: AlgorandBlock): Promise<void> {
  const record = BlockEntity.create({
    id: block.round.toString(),
    field1: block.round,
    field2: "block",
  });
  await record.save();
}

Transaction Handler

You can use transaction handlers to capture information about each of the transactions in a block. To achieve this, a defined TransactionHandler will be called once for every transaction. You should use Mapping Filters in your manifest to filter transactions to reduce the time it takes to index data and improve mapping performance.

import { AlgorandTransaction } from "@subql/types-algorand";

export async function handleTransaction(
  tx: AlgorandTransaction
): Promise<void> {
  const record = TransactionEntity.create({
    id: tx.id,
    field1: tx.roundTime,
    field2: "tx",
  });
  await record.save();
}

The Sandbox

SubQuery is deterministic by design, that means that each SubQuery project is guranteed to index the same data set. This is a critical factor that is required to decentralise SubQuery in the SubQuery Network. This limitation means that the indexer is by default run in a strict virtual machine, with access to a strict number of third party libraries.

You can bypass this limitation, allowing you to index and retrieve information from third party data sources like HTTP endpoints, non historical RPC calls, and more. In order to do to, you must run your project in unsafe-mode, you can read more about this in the references. An easy way to do this while developing (and running in Docker) is to add the following line to your docker-compose.yml:

subquery-node:
  image: onfinality/subql-node-algorand:latest
  ...
  command:
    - -f=/app
    - --db-schema=app
    - --unsafe
  ...

By default, the VM2 (opens new window) sandbox only allows the folling:

• only some certain built-in modules, e.g. assert, buffer, crypto,util and path
• third-party libraries written by CommonJS.
• hybrid libraries like @polkadot/* that uses ESM as default. However, if any other libraries depend on any modules in ESM format, the virtual machine will NOT compile and return an error.
• Historical/safe queries, see RPC Calls.
• Note HTTP and WebSocket connections are forbidden

Modules and Libraries

To improve SubQuery's data processing capabilities, we have allowed some of the NodeJS's built-in modules for running mapping functions in the sandbox, and have allowed users to call third-party libraries.

Please note this is an experimental feature and you may encounter bugs or issues that may negatively impact your mapping functions. Please report any bugs you find by creating an issue in GitHub (opens new window).

Built-in modules

Currently, we allow the following NodeJS modules: assert, buffer, crypto, util, and path.

Rather than importing the whole module, we recommend only importing the required method(s) that you need. Some methods in these modules may have dependencies that are unsupported and will fail on import.

import { hashMessage } from "ethers/lib/utils"; //Good way
import { utils } from "ethers"; //Bad way

export async function handleTransaction(
  tx: AlgorandTransaction
): Promise<void> {
  const record = new TransactionEntity(tx.id);
  record.field1 = hashMessage("Hello");
  await record.save();
}