Skip to main content
Transformers allow you to process and reshape data as it flows through your pipeline. This guide demonstrates creating custom transformers for common use cases.

Create a basic transformer

​
Adding Transformer

Extract transaction hashes from instructions.Use case: Transform raw Portal data before logging.
import { createTarget, createTransformer } from "@subsquid/pipes";
import {
  solanaPortalSource,
  type SolanaPortalData,
  SolanaQueryBuilder,
} from "@subsquid/pipes/solana";

const queryBuilder = new SolanaQueryBuilder()
  .addFields({
    block: { slot: true, hash: true },
    instruction: { programId: true, data: true, accounts: true, transactionHash: true },
  })
  .addInstruction({
    request: {
      programId: ["whirLbMiicVdio4qvUfM5KAg6Ct8VwpYzGff3uctyCc"],
      d8: ["0xf8c69e91e17587c8"],
    },
    range: { from: 200000000, to: 200000000 },
  });

const source = solanaPortalSource({
  portal: "https://portal.sqd.dev/datasets/solana-mainnet",
  query: queryBuilder,
});

const transformer = createTransformer({
  transform: async (data: SolanaPortalData<any>) => {
    return data.blocks.map((b) => b.instructions.map((i) => i.transactionHash));
  },
});

const target = createTarget({
  write: async ({ logger, read }) => {
    for await (const { data } of read()) {
      logger.info({ data }, "transaction hashes");
    }
  },
});

await source.pipe(transformer).pipeTo(target);
Keep transformers focused on a single responsibility for better testability and reusability.
Your transformer extracts transaction hashes from all instructions in the batch, making them easy to log or further process.

Create a query-aware transformer

​
Query from Transformer

Build query dynamically in transformer.Use case: Self-contained transformers that specify their own data requirements.
import { createTarget, createTransformer } from "@subsquid/pipes";
import {
  solanaPortalSource,
  type SolanaPortalData,
  SolanaQueryBuilder,
} from "@subsquid/pipes/solana";

// Start with empty query
const source = solanaPortalSource({
  portal: "https://portal.sqd.dev/datasets/solana-mainnet",
  query: new SolanaQueryBuilder(),
});

// Transformer adds to query
const transformer = createTransformer({
  query: ({ queryBuilder }) => {
    queryBuilder
      .addFields({
        block: { slot: true, hash: true },
        instruction: { programId: true, data: true, accounts: true, transactionHash: true },
      })
      .addInstruction({
        request: {
          programId: ["whirLbMiicVdio4qvUfM5KAg6Ct8VwpYzGff3uctyCc"],
          d8: ["0xf8c69e91e17587c8"],
        },
        range: { from: 200000000, to: 200000000 },
      });
  },
  transform: async (data: SolanaPortalData<any>) => {
    return data.blocks.map((b) => b.instructions.map((i) => i.transactionHash));
  },
});

const target = createTarget({
  write: async ({ logger, read }) => {
    for await (const { data } of read()) {
      logger.info({ data }, "data");
    }
  },
});

await source.pipe(transformer).pipeTo(target);
Key Distinction: Transformers can operate in two ways:
  • Pure transformation (first example): Only implements transform() — processes data after the Portal query is executed.
  • Query-aware transformation (this example): Implements both query() and transform() — contributes to building the Portal query and processes the resulting data.
When a transformer has a query() method, it augments the Portal query before data is fetched, allowing transformers to be self-contained with their data requirements.
Your query-aware transformer is now self-contained, specifying both what data it needs and how to process it.

​
Next Steps

For better type safety and developer experience, use solanaInstructionDecoder with ABIs generated by @subsquid/solana-typegen. See the Instruction Decoding guide for a typed approach.