otterscan/docs/custom-jsonrpc.md
2022-01-21 04:06:46 -03:00

338 lines
18 KiB
Markdown

# Otterscan JSON-RPC API extensions
The [standard Ethereum JSON-RPC APIs](https://ethereum.org/en/developers/docs/apis/json-rpc/) are very limited and in some cases non-performant for what you can do with an archive node.
There is plenty of useful data that can be extracted and we implemented some extra RPC methods for them.
They are all used by Otterscan, but we are documenting them here so others can try it, give feedback and eventually get it merged upstream if they are generalized enough.
We take an incremental approach when design the APIs, so there may be some methods very specific to Otterscan use cases, others that look more generic.
Please see the [install instructions](./install.md) if you want to run a patched Erigon with those customizations enabled.
## Quick FAQ
### Why don't you use _Some Product XXX_ for Otterscan? And why shouldn't I?
If you are happy using _Some Product XXX_, go ahead.
Otterscan pursues a minimalistic approach and at the same time it is very easy to modify Erigon for your own needs.
Most of the features we implemented are quite basic and it is unfortunate they are not part of the standard API.
> We believe most people end up using _Some Product XXX_ not because of its own unique features, but because the standard JSON-RPC API is quite limited even for basic features.
Implementing everything in-node allows you to plug a dapp directly to your node itself. No need to install any additional indexer middleware or SQL database, each of it own consuming extra disk space and CPU.
> Take Otterscan as an example, **ALL** you need is Otterscan itself (a SPA, can be served by any static provider) and our modified Erigon's rpcdaemon.
### But your API doesn't scale and it is slower than _Some Product XXX_!!!
Not everyone needs to serve thousands of requests per second. Go ahead and use _Some Product XXX_.
Some people just want to run standalone development tools and calculating some data on-the-fly works fine for single user local apps.
Even so, we may introduce custom indexes to speed up some operations in future if there is such demand, so you may opt-in for a better performance by spending more disk space.
### Wen PR upstream?
API design is hard and once it goes public you have to support it forever. For this reason we are primarily keeping it in our own fork and under a vendor specific namespace (`ots_`).
Also, the quality level of the current APIs differs, some are very generic, some are very Otterscan specific. Our API design has been driven mainly by Otterscan feature needs, which is a good thing (tm), so no useless features.
Having said that, we want to have people experimenting with our APIs, bringing other use cases, and driving the API evolution. If there are enough users vouching for a certain feature, we would gladly submit a PR to Erigon upstream repo.
The first step to achieving that is having this own page properly documenting our APIs so people don't have to look at our source code 😅.
Your feedback is important, please get in touch using our communication channels.
## How to use it?
They are all JSON-RPC methods, so your favorite web3 library _should_ have some way to custom call them.
For example, ethers.js wraps standard calls in nice, user-friendly classes and parses results into easy-to-use objects, but also allows you to do custom calls and get raw results while still taking advantage of their capabilities like automatic batching, network timeout handling, etc.
I'll use ethers.js as an example here because it is what I use in Otterscan, please check your web3 library docs for custom call support.
Let's call the `ots_getTransactionError` method to obtain the revert reason of a failed transaction. It accepts one string parameter containing the transaction hash and returns a byte blob that can be ABI-decoded:
```
const provider = ...; // Obtain a JsonRpcProvider object
const txHash = "..."; // Set the transaction hash
const result = (await provider.send("ots_getTransactionError", [txHash])) as string;
```
## Method summary
All methods are prefixed with the `ots_` namespace in order to make it clear it is vendor-specific and there is no name clash with other same-name implementations.
| Name | Description | Reasoning |
|-------------------|------------------|-----------|
| `ots_getApiLevel` | Totally Otterscan internal API, absolutely no reason for anything outside Otterscan to use it. | Used by Otterscan to check if it's connecting to a compatible patched Erigon node and display a friendly message if it is not. |
| `ots_getInternalOperations` | Return the internal ETH transfers inside a transaction. | For complex contract interactions, there may be internal calls that forward ETH between addresses. A very common example is someone swapping some token for ETH, in this case there is an ETH send to the sender address which is only unveiled by examining the internal calls. |
| `ots_hasCode` | Check if a certain address contains a deployed code. | A common way to check if an address is a contract or an EOA is calling `eth_getCode` to see if it has some code deployed. However this call is expensive regarding this purpose, as it returns the entire contract code over the network just for the client to check its presence. This call just returns a boolean. |
| `ots_getTransactionError` | Extract the transaction raw error output. | In order to get the error message or custom error from a failed transaction, you need to get its error output and decoded it. This info is not exposed through standard APIs. |
| `ots_traceTransaction` | Extract all variations of calls, contract creation and self-destructs and returns a call tree. | This is an optimized version of tracing; regular tracing returns lots of data, and custom tracing using a JS tracer could be slow. |
| `ots_getBlockDetails` | Tailor-made and expanded version of `eth_getBlock*` for block details page in Otterscan. | The standard `eth_getBlock*` is quite verbose and it doesn't bring all info we need. We explicitly remove the transaction list (unnecessary for that page and also this call doesn't scale well), log blooms and other unnecessary fields. We add issuance and block fees info and return all of this in just one call. |
| `ots_getBlockTransactions` | Get paginated transactions for a certain block. Also remove some verbose fields like logs. | As block size increases, getting all transactions from a block at once doesn't scale, so the first point here is to add pagination support. The second point is that receipts may have big, unnecessary information, like logs. So we cap all of them to save network bandwidth. |
| `ots_searchTransactionsBefore` and `ots_searchTransactionsAfter` | Gets paginated inbound/outbound transaction calls for a certain address. | There is no native support for any kind of transaction search in the standard JSON-RPC API. We don't want to introduce an additional indexer middleware in Otterscan, so we implemented in-node search. |
| `ots_getTransactionBySenderAndNonce` | Gets the transaction hash for a certain sender address, given its nonce. | There is no native support for this search in the standard JSON-RPC API. Otterscan needs it to allow user navigation between nonces from the same sender address. |
## Method details
> Some methods include a sample call so you call try it from cli. The examples use `curl` and assume you are running `rpcdaemon` at `http://127.0.0.1:8545`.
### `ots_getApiLevel`
Very simple API versioning scheme. Every time we add a new capability, the number is incremented. This allows for Otterscan to check if the Erigon node contains all API it needs.
Parameters:
`<none>`
Returns:
- `number` containing the API version.
### `ots_getInternalOperations`
Trace internal ETH transfers, contracts creation (CREATE/CREATE2) and self-destructs for a certain transaction.
Parameters:
1. `txhash` - The transaction hash.
Returns:
- `array` of operations, sorted by their occurrence inside the transaction.
The operation is an object with the following fields:
- `type` - transfer (`0`), self-destruct (`1`), create (`2`) or create2 (`3`).
- `from` - the ETH sender, contract creator or contract address being self-destructed.
- `to` - the ETH receiver, newly created contract address or the target ETH receiver resulting of the self-destruction.
- `value` - the amount of ETH transferred.
### `ots_hasCode`
Check if an ETH address contains a deployed code.
Parameters:
1. `address` - The ETH address to be checked.
2. `block` - The block number at which the code presence will be checked or "latest" to check the latest state.
Returns:
- `boolean` indicating if the address contains a bytecode or not.
Example 1: does Uniswap V1 Router address have a code deployed? (yes, it is a contract)
Request:
```
$ curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0", "id": 1, "method":"ots_hasCode","params":["0xc0a47dFe034B400B47bDaD5FecDa2621de6c4d95", "latest"]}' http://127.0.0.1:8545
```
Response:
```
{
"jsonrpc": "2.0",
"id": 1,
"result": true
}
```
Example 2: does Vitalik's public address have a code deployed? (no, it is an EOA)
Request:
```
$ curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0", "id": 1, "method":"ots_hasCode","params":["0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045", "latest"]}' http://127.0.0.1:8545
```
Response:
```
{
"jsonrpc": "2.0",
"id": 1,
"result": false
}
```
### `ots_traceTransaction`
Trace a transaction and generate a trace call tree.
Parameters:
1. `txhash` - The transaction hash.
Returns:
- `object` containing the trace tree.
### `ots_getTransactionError`
Given a transaction hash, returns its raw revert reason.
The returned byte blob should be ABI decoded in order to be presented to the user.
For instance, the most common error format is a `string` revert message; in this case, it should be decoded using the `Error(string)` method selector, which will allow you to extract the string message.
If it is not the case, it should probably be a solidity custom error, so you must have the custom error ABI in order to decoded it.
Parameters:
1. `txhash` - The transaction hash.
Returns:
- `string` containing the hexadecimal-formatted error blob or simply a "0x" if the transaction was sucessfully executed. It is returns "0x" if it failed with no revert reason or out of gas, make sure to analyze this return value together with the transaction success/fail result.
Example: get the revert reason of a random transaction spotted in the wild to Uniswap V3.
Request:
```
$ curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0", "id": 1, "method":"ots_getTransactionError","params":["0xcdb0e53c4f1b5f37ea7f0d2a8428b13a5bff47fb457d11ef9bc85ccdc489635b"]}' http://127.0.0.1:8545
```
Response:
```
{
"jsonrpc": "2.0",
"id": 1,
"result": "0x08c379a0000000000000000000000000000000000000000000000000000000000000002000000000000000000000000000000000000000000000000000000000000000135472616e73616374696f6e20746f6f206f6c6400000000000000000000000000"
}
```
> ABI-decoding this byte string against `Error(string)` should result in the "Transaction too old" error message.
### `ots_getBlockDetails`
Given a block number, return its data. Similar to the standard `eth_getBlockByNumber/Hash` method, but optimized.
Parameters:
1. `number` representing the desired block number.
Returns:
- `object` in a format _similar_ to the one returned by `eth_getBlockByNumber/Hash` (please refer to their docs), with some small differences:
- the block data comes nested inside a `block` attribute.
- the `transactions` attribute is not returned. The reason is that it doesn't scale, the standard methods return either the transaction hash list or the transaction list with their bodies. So we cap the transaction list entirely to avoid unnecessary network traffic.
- the transaction count is returned in a `transactionCount` attribute.
- the `logsBloom` attribute comes with `null`. It is a byte blob thas is rarely used, so we cap it to avoid unnecessary network traffic.
- an extra `issuance` attribute returns an `object` with the fields:
- `blockReward` - the miner reward.
- `uncleReward` - the total reward issued to uncle blocks.
- `issuance` - the total ETH issued in this block (miner + uncle rewards).
- an extra `totalFees` attribute containing the sum of fees paid by senders in this block. Note that due to EIP-1559 this is **NOT** the same amount earned by the miner as block fees since it contains the amount paid as base fee.
### `ots_getBlockTransactions`
Gets paginated transaction data for a certain block. Think of an optimized `eth_getBlockBy*` + `eth_getTransactionReceipt`.
The `transactions` field contains the transaction list with their bodies in a similar format of `eth_getBlockBy*` with transaction bodies, with a few differences:
- the `input` field returns only the 4 bytes method selector instead of the entire calldata byte blob.
The `receipts` attribute contains the transactions receipt list, in the same sort order as the block transactions. Returning it here avoid the caller of making N+1 calls (`eth_getBlockBy*` and `eth_getTransactionReceipt`).
For receipts, it contains some differences from the `eth_getTransactionReceipt` object format:
- `logs` attribute returns `null`.
- `logsBloom` attribute returns `null`.
### `ots_searchTransactionsBefore` and `ots_searchTransactionsAfter`
These are address history navigation methods. They are similar, the difference is `ots_searchTransactionsBefore` searches the history backwards and `ots_searchTransactionsAfter` searches forward a certain point in time.
They are paginated, you **MUST** inform the page size. Some addresses like exchange addresses or very popular DeFi contracts like Uniswap Router will return millions of results.
They return inbound (`to`), outbound (`from`) and "internal" transactions. By internal it means that if a transaction calls a contract and somewhere in the call stack it sends ETH to the address you are searching for or the address is a contract and it calls a method on it, the transaction is matched and returned in the search results.
Parameters:
1. `address` - The ETH address to be searched.
2. `blockNumber` - It searches for occurrences of `address` before/after `blockNumber`. A value of `0` means you want to search from the most recent block (`ots_searchTransactionsBefore`) or from the genesis (`ots_searchTransactionsAfter`).
3. `pageSize` - How many transactions it may return. See the detailed explanation about this parameter bellow.
Returns:
- `object` containing the following attributes:
- `txs` - An array of objects representing the transaction results. The results are returned sorted from the most recent to the older one (descending order).
- `receipts` - An array of objects containing the transaction receipts for the transactions returned in the `txs` attribute.
- `firstPage` - Boolean indicating this is the first page. It should be `true` when calling `ots_searchTransactionsBefore` with `blockNumber` == 0 (search from `latest`); because the results are in descending order, the search from the most recent block is the "first" one. It should also return `true` when calling `ots_searchTransactionsAfter` with a `blockNumber` which results in no more transactions after the returned ones because it searched forward up to the tip of the chain.
- `lastPage` - Boolean indicating this is the last page. It should be `true` when calling `ots_searchTransactionsAfter` with `blockNumber` == 0 (search from genesis); because the results are in descending order, the genesis page is the "last" one. It should also return `true` when calling `ots_searchTransactionsBefore` with a `blockNumber` which results in no more transactions before the returned ones because it searched backwards up to the genesis block.
There is a small gotcha regarding `pageSize`. If there are less results than `pageSize`, they are just returned as is.
But if there are more than `pageSize` results, they are capped by the last found block. For example, let's say you are searching for Uniswap Router address and it already found 24 matches; it then looks at the next block containing this addresses occurrences and there are 5 matches inside the block. They are all returned, so it returns 30 transaction results. The caller code should be aware of this.
Example: get the first 5 transactions that touched Uniswap V1 router (including the contract creation).
Request:
```
$ curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0", "id": 1, "method":"ots_searchTransactionsAfter","params":["0xc0a47dFe034B400B47bDaD5FecDa2621de6c4d95", 0, 5]}' http://127.0.0.1:8545
```
Response:
```
{
"jsonrpc": "2.0",
"id": 1,
"result": {
"txs": [
{
"blockHash": "0x06a77abe52c486f58696665eaebd707f17fbe97eb54480c6533db725769ce3b7",
"blockNumber": "0x652284",
"from": "0xd1c24f50d05946b3fabefbae3cd0a7e9938c63f2",
"gas": "0xf4240",
"gasPrice": "0x2cb417800",
"hash": "0x14455f1af43a52112d4ccf6043cb081fea4ea3a07d90dd57f2a9e1278114be94",
"input": "0x1648f38e000000000000000000000000e41d2489571d322189246dafa5ebde1f4699f498",
"nonce": "0x6",
"to": "0xc0a47dfe034b400b47bdad5fecda2621de6c4d95",
"transactionIndex": "0x71",
...
}
```
### `ots_getTransactionBySenderAndNonce`
Given a sender address and a nonce, returns the tx hash or `null` if not found. It returns only the tx hash on success, you can use the standard `eth_getTransactionByHash` after that to get the full transaction data.
Parameters:
1. `sender` - The sender ETH address.
2. `nonce` - The sender nonce.
Returns:
- `string` containing the corresponding transaction hash or `null` if it doesn't exist.
Example: get the 4th transaction sent by Vitalik's public address (nonce == 3).
Request:
```
$ curl -X POST -H "Content-Type: application/json" --data '{"jsonrpc":"2.0", "id": 1, "method":"ots_getTransactionBySenderAndNonce","params":["0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045", 3]}' http://127.0.0.1:8545
```
Response:
```
{
"jsonrpc": "2.0",
"id": 1,
"result": "0x021304206b2517c3f8f2df07014a55b79aac2ae097488fa807cc88eccd851a50"
}
```