tail-eth-events

Overview

Small utility to tail events in the Ethereum log from the current block onward, in human readable(-ish) format. For example:

[Block 15454857] [Contract 0xDe07f45688cb6CfAaC398c1485860e186D55996D]  Transfer(indexed address sender=0x353d601aae0fc29908bb46a26cac562ad1a5bff2,indexed address receiver=0x0000000000000000000000000000000000000000,uint256 value=17268575233016576578)
[Block 15454857] [Contract 0x8eF11c51a666C53Aeeec504f120cd1435E451342]  Withdraw(indexed address owner=0x353d601aae0fc29908bb46a26cac562ad1a5bff2,uint256 shares=17268575233016576578,uint256 amount=18257840581989549385)
[Block 15454857] [Contract 0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2]  Transfer(indexed address src=0x7685cd3ddd862b8745b1082a6acb19e14eaa74f3,indexed address dst=0xbeefbabeea323f07c59926295205d3b7a17e8638,uint256 wad=610600133740867388)
[Block 15454857] [Contract 0xba5BDe662c17e2aDFF1075610382B9B691296350, ProxiedTo 0x31acaaea0529894e7c3a5c70d3f9ee6d7804684f] Transfer(indexed address from=0xbeefbabeea323f07c59926295205d3b7a17e8638,indexed address to=0x7685cd3ddd862b8745b1082a6acb19e14eaa74f3,uint256 value=4724999999999999737856)
[Block 15454857] [Contract 0x7685cD3ddD862b8745B1082A6aCB19E14EAA74F3]  Anonymous/Unknown()
[Block 15454857] [Contract 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48]  Anonymous/Unknown()

It uses web3.py to get all log data in the current block. Then, for each log entry, it attempts to find the contract ABI, including for proxied contracts, using the Etherscan API. If it can find the ABI it will decode the event arguments using the ABI.

Short polling is used to check when a new block is mined.

Usage

Set the following environment variables:

ETH_NODE_WEBSOCKET= # wss url of your eth node provider
ETHERSCAN_API_KEY=

And run the script with:

python3 src/tail.py

Unit tests can we run with:

nosetest

Future Work

• topic0 is a repository of event signatures, which could partially replace the current contract ABI lookup.
• Contract ABIs could/should be persisted in some long-term storage. A lightweight approach is to use sqlitedict
• Similarly, failed proxy lookups should also be persisted, so we don't do it each time we miss the event ABI.
• Since log entries are independent, event processing should be done via multithreading.
• Better testing, error handling, etc...