Please complete the following:
- Set up a subquery indexer (e.g. Goldsky Subgraph)
- Follow the docs here: https://docs.goldsky.com/guides/create-a-no-code-subgraph
- General Steps
- create an account at app.goldsky.com
- deploy a subgraph or migrate an existing subgraph - https://docs.goldsky.com/subgraphs/introduction
- Use the slugs
linea-testnet
andlinea
when deploying the config
- Prepare Subquery query code according to the Data Requirement section below.
- Submit your response as a Pull Request to: https://github.com/delta-hq/l2-lxp-liquidity-reward
- With path being
/<your_protocol_handle>
- With path being
- Submit your contract addresses through this Form
- Create a function like below:
export const getUserTVLByBlock = async (blocks: BlockData) => {
const { blockNumber, blockTimestamp } = blocks
// Retrieve data using block number and timestamp
// YOUR LOGIC HERE
return csvRows
};
- Interface for input Block Data is, in below blockTimestamp is in epoch format.
interface BlockData {
blockNumber: number;
blockTimestamp: number;
}
- Output "csvRow" is a list.
const csvRows: OutputDataSchemaRow[] = [];
type OutputDataSchemaRow = {
block_number: number;
timestamp: number;
user_address: string;
token_address: string;
token_balance: bigint;
token_symbol: string; //token symbol should be empty string if it is not available
usd_price: number; //assign 0 if not available
};
- Make sure you add relevant package.json and tsconfig.json
- You can check the index.ts in Gravita project to refer this in use.
Goal: Hourly snapshot of TVL by User by Asset
For each protocol, we are looking for the following:
- Query that fetches all relevant events required to calculate User TVL in the Protocol at hourly level.
- Code that uses the above query, fetches all the data and converts it to csv file in below given format.
- Token amount should be raw token amount. Please do not divide by decimals.
Teams can refer to the example we have in there to write the code required.
Data Field | Notes |
---|---|
block_number | |
timestamp | Block timestamp |
user_address | |
token_address | |
token_symbol (optional) | Symbol of token |
token_balance | Balance of token (If the token was borrowed, this balance should be negative) |
usd_price (from oracle) | Price of token (optional) |
Sample output row will look like this:
blocknumber | timestamp | user_address | token_address | token_balance | token_symbol (optional) | usd_price(optional) |
---|---|---|---|---|---|---|
2940306 | 2024-03-16 19:19:57 | 0x4874459FE…d689 | 0xc02aaa39b223fe8d0a0e5c4f27ead9083c756cc2 | 100 | WETH | 0 |
Note: Expect multiple entries per user if the protocols has more than one token asset
Below is the query being used in the example we have in the repo link. For querying like this, please create a subgraph that has this data for your respective protocol. This data should be further transformed to get the data as per required Output schema.
{
positions( block: {number: 4004302} orderBy: transaction__timestamp, first:1000,skip:0) {
id
liquidity
owner
pool {
sqrtPrice
tick
id
}
tickLower{
tickIdx
}
tickUpper{
tickIdx
}
token0 {
id
decimals
derivedUSD
name
symbol
}
token1 {
id
decimals
derivedUSD
name
symbol
}
},
_meta{
block{
number
}
}
}
{
"id": "3",
"liquidity": "0",
"owner": "0x8ad255ee352420d3e257aa87a5811bd09f72d251",
"pool": {
"sqrtPrice": "4158475459167119976298502",
"tick": -197109,
"id": "0xf2e9c024f1c0b7a2a4ea11243c2d86a7b38dd72f"
},
"tickLower": {
"tickIdx": -198790
},
"tickUpper": {
"tickIdx": -198770
},
"token0": {
"id": "0x4200000000000000000000000000000000000006",
"decimals": "18",
"derivedUSD": "2754.920801587090063443331457265368",
"name": "Wrapped Ether",
"symbol": "WETH"
},
"token1": {
"id": "0xd988097fb8612cc24eec14542bc03424c656005f",
"decimals": "6",
"derivedUSD": "1",
"name": "USD Coin",
"symbol": "USDC"
}
}
On this scope, the code must read a CSV file with headers named hourly_blocks.csv
that contains the following columns:
number
- block numbertimestamp
- block timestamp
And output a CSV file named outputData.csv
with headers with the following columns:
block_number
- block numbertimestamp
- block timestampuser_address
- user addresstoken_address
- token addresstoken_symbol
- token symboltoken_balance
- token balanceusd_price
- USD price e.g.adapters/renzo/src/index.ts
For testing the adapter code for a single hourly block, use the following hourly_blocks.csv
file:
number,timestamp
4243360,1714773599
Please submit your Contract Addresses and Pool Addresses through this Form.
In this repo, there is an adapter example. This adapter aims to get data positions from the subrgaph and calculate the TVL by users. The main scripts is generating a output as CSV file.
- Please make sure to have a "compile" script in package.json file. So, we are able to compile the typescript files into
dist/index.js
file.
npm install // install all packages
npm run watch //other terminal tab
npm run start // other terminal tab
By this, we'll be able to generate the output csv file.