Filters can come in many forms, but there are a couple primary use cases: wallet and contract addresses.

Filters also work twice per pipeline:

  1. They are applied for each network beat to determine if a beat should even be processed.
  2. After processing, the resulting records are filtered again using the pipeline’s filterKeys.

​
Example

Let’s say we want to filter vitalik.eth’s token transfers.

We’d first add a filter for his address, 0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045.

From there, we could test the following transformation against an Ethereum block containing a token transfer (e.g. block 22062886 with this tx):

// vitalik-example.js
function vitalikTokenTransfers(block) {
  const erc20Transfers = [];

  for (const tx of block.transactions) {
    for (const log of tx.receipt?.logs) {
      const decoded = utils.evmDecodeLog(log, [
        'event Transfer(address indexed from, address indexed to, uint256 value)',
      ]);

      if (decoded) {
        erc20Transfers.push({
          transaction_hash: tx.hash,
          log_index: log.logIndex,
          contract_address: log.address.toLowerCase(),
          from: decoded.from,
          to: decoded.to,
        });
      }
    }
  }

  return erc20Transfers;
}

NOTE: the above transformation does not have its own filter step. Instead, we can specify during testing and in the pipeline to filter by from, to, or both.

curl "https://app.indexing.co/dw/transformations/test?network=ethereum&beat=22062886&filter=vitalik&filterKeys\[0\]=to" -H 'x-api-key: <API KEY>' -F code=@./vitalik-example.js | jq

^ this will provide a filtered set of token transfers like this:

[
  {
    "transaction_hash": "0x9515ed82d894cdaf6d6a6b5952f8438cdaad948e17c9d842deb5c3f311207bc6",
    "log_index": 2,
    "contract_address": "0x7e1fb1be92693ec1da42ee8dfc072990bf3a3000",
    "from": "0x0000000000000000000000000000000000000000",
    "to": "0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045"
  }
]